Automated telephone communication systems for preventive healthcare and management of long‐term conditions

Pawel Posadzki; Nikolaos Mastellos; Rebecca Ryan; Laura H Gunn; Lambert M Felix; Yannis Pappas; Marie‐Pierre Gagnon; Steven A Julious; Liming Xiang; Brian Oldenburg; Josip Car

doi:10.1002/14651858.CD009921.pub2

. 2016 Dec 14;2016(12):CD009921. doi: 10.1002/14651858.CD009921.pub2

Automated telephone communication systems for preventive healthcare and management of long‐term conditions

Pawel Posadzki ¹, Nikolaos Mastellos ², Rebecca Ryan ³, Laura H Gunn ⁴, Lambert M Felix ⁵, Yannis Pappas ⁶, Marie‐Pierre Gagnon ⁷, Steven A Julious ⁸, Liming Xiang ⁹, Brian Oldenburg ¹⁰, Josip Car ^1,^2,^11,^✉

Editor: Cochrane Consumers and Communication Group

PMCID: PMC6463821 PMID: 27960229

Abstract

Background

Automated telephone communication systems (ATCS) can deliver voice messages and collect health‐related information from patients using either their telephone's touch‐tone keypad or voice recognition software. ATCS can supplement or replace telephone contact between health professionals and patients. There are four different types of ATCS: unidirectional (one‐way, non‐interactive voice communication), interactive voice response (IVR) systems, ATCS with additional functions such as access to an expert to request advice (ATCS Plus) and multimodal ATCS, where the calls are delivered as part of a multicomponent intervention.

Objectives

To assess the effects of ATCS for preventing disease and managing long‐term conditions on behavioural change, clinical, process, cognitive, patient‐centred and adverse outcomes.

Search methods

We searched 10 electronic databases (the Cochrane Central Register of Controlled Trials; MEDLINE; Embase; PsycINFO; CINAHL; Global Health; WHOLIS; LILACS; Web of Science; and ASSIA); three grey literature sources (Dissertation Abstracts, Index to Theses, Australasian Digital Theses); and two trial registries (www.controlled‐trials.com; www.clinicaltrials.gov) for papers published between 1980 and June 2015.

Selection criteria

Randomised, cluster‐ and quasi‐randomised trials, interrupted time series and controlled before‐and‐after studies comparing ATCS interventions, with any control or another ATCS type were eligible for inclusion. Studies in all settings, for all consumers/carers, in any preventive healthcare or long term condition management role were eligible.

Data collection and analysis

We used standard Cochrane methods to select and extract data and to appraise eligible studies.

Main results

We included 132 trials (N = 4,669,689). Studies spanned across several clinical areas, assessing many comparisons based on evaluation of different ATCS types and variable comparison groups. Forty‐one studies evaluated ATCS for delivering preventive healthcare, 84 for managing long‐term conditions, and seven studies for appointment reminders. We downgraded our certainty in the evidence primarily because of the risk of bias for many outcomes. We judged the risk of bias arising from allocation processes to be low for just over half the studies and unclear for the remainder. We considered most studies to be at unclear risk of performance or detection bias due to blinding, while only 16% of studies were at low risk. We generally judged the risk of bias due to missing data and selective outcome reporting to be unclear.

For preventive healthcare, ATCS (ATCS Plus, IVR, unidirectional) probably increase immunisation uptake in children (risk ratio (RR) 1.25, 95% confidence interval (CI) 1.18 to 1.32; 5 studies, N = 10,454; moderate certainty) and to a lesser extent in adolescents (RR 1.06, 95% CI 1.02 to 1.11; 2 studies, N = 5725; moderate certainty). The effects of ATCS in adults are unclear (RR 2.18, 95% CI 0.53 to 9.02; 2 studies, N = 1743; very low certainty).

For screening, multimodal ATCS increase uptake of screening for breast cancer (RR 2.17, 95% CI 1.55 to 3.04; 2 studies, N = 462; high certainty) and colorectal cancer (CRC) (RR 2.19, 95% CI 1.88 to 2.55; 3 studies, N = 1013; high certainty) versus usual care. It may also increase osteoporosis screening. ATCS Plus interventions probably slightly increase cervical cancer screening (moderate certainty), but effects on osteoporosis screening are uncertain. IVR systems probably increase CRC screening at 6 months (RR 1.36, 95% CI 1.25 to 1.48; 2 studies, N = 16,915; moderate certainty) but not at 9 to 12 months, with probably little or no effect of IVR (RR 1.05, 95% CI 0.99, 1.11; 2 studies, 2599 participants; moderate certainty) or unidirectional ATCS on breast cancer screening.

Appointment reminders delivered through IVR or unidirectional ATCS may improve attendance rates compared with no calls (low certainty). For long‐term management, medication or laboratory test adherence provided the most general evidence across conditions (25 studies, data not combined). Multimodal ATCS versus usual care showed conflicting effects (positive and uncertain) on medication adherence. ATCS Plus probably slightly (versus control; moderate certainty) or probably (versus usual care; moderate certainty) improves medication adherence but may have little effect on adherence to tests (versus control). IVR probably slightly improves medication adherence versus control (moderate certainty). Compared with usual care, IVR probably improves test adherence and slightly increases medication adherence up to six months but has little or no effect at longer time points (moderate certainty). Unidirectional ATCS, compared with control, may have little effect or slightly improve medication adherence (low certainty). The evidence suggested little or no consistent effect of any ATCS type on clinical outcomes (blood pressure control, blood lipids, asthma control, therapeutic coverage) related to adherence, but only a small number of studies contributed clinical outcome data.

The above results focus on areas with the most general findings across conditions. In condition‐specific areas, the effects of ATCS varied, including by the type of ATCS intervention in use.

Multimodal ATCS probably decrease both cancer pain and chronic pain as well as depression (moderate certainty), but other ATCS types were less effective. Depending on the type of intervention, ATCS may have small effects on outcomes for physical activity, weight management, alcohol consumption, and diabetes mellitus. ATCS have little or no effect on outcomes related to heart failure, hypertension, mental health or smoking cessation, and there is insufficient evidence to determine their effects for preventing alcohol/substance misuse or managing illicit drug addiction, asthma, chronic obstructive pulmonary disease, HIV/AIDS, hypercholesterolaemia, obstructive sleep apnoea, spinal cord dysfunction or psychological stress in carers.

Only four trials (3%) reported adverse events, and it was unclear whether these were related to the interventions.

Authors' conclusions

ATCS interventions can change patients' health behaviours, improve clinical outcomes and increase healthcare uptake with positive effects in several important areas including immunisation, screening, appointment attendance, and adherence to medications or tests. The decision to integrate ATCS interventions in routine healthcare delivery should reflect variations in the certainty of the evidence available and the size of effects across different conditions, together with the varied nature of ATCS interventions assessed. Future research should investigate both the content of ATCS interventions and the mode of delivery; users' experiences, particularly with regard to acceptability; and clarify which ATCS types are most effective and cost‐effective.

Plain language summary

Automated telephone communication systems for preventing disease and managing long‐term conditions

Background

Automated telephone communication systems (ATCS) send voice messages and collect health information from people using their telephone's touch‐tone keypad or voice recognition software. This could replace or supplement telephone contact between health professionals and patients. There are several types of ATCS: one‐way voice messages to patients (unidirectional), interactive voice response (IVR) systems, those with added functions like referral to advice (ATCS Plus), or those where ATCS are part of a complex intervention (multimodal).

Review question

This review assessed the effectiveness of ATCS for preventing disease and managing long‐term conditions.

Results

We found 132 trials with over 4 million participants across preventive healthcare areas and for the management of long‐term conditions.

Studies compared ATCS types in many ways.

Some studies reported findings across diseases. For prevention, ATCS probably increase immunisation uptake in children, and slightly in adolescents, but in adults effects are uncertain. Also for prevention, multimodal ATCS increase numbers of people screened for breast or colorectal cancers, and may increase osteoporosis screening. ATCS Plus probably slightly increases attendance for cervical cancer screening, with uncertain effects on osteoporosis screening. IVR probably increases the numbers screened for colorectal cancer up to six months, with little effect on breast cancer screening.

ATCS (unidirectional or IVR) may improve appointment attendance, key to both preventing and managing disease.

For long‐term management, multimodal ATCS had inconsistent effects on medication adherence. ATCS Plus probably improves medication adherence versus usual care. Compared with control, ATCS Plus and IVR probably slightly improve adherence, while unidirectional ATCS may have little, or slightly positive, effects. No intervention consistently improved clinical outcomes. IVR probably improves test adherence, but ATCS Plus may have little effect.

ATCS were also used in specific conditions. Effects varied by condition and ATCS type. Multimodal ATCS, but not other ATCS types, probably decrease cancer pain and chronic pain. Outcomes may improve to a small degree when ATCS are applied to physical activity, weight management, alcohol use and diabetes.However, there is little or no effect in heart failure, hypertension, mental health or quitting smoking. In several areas (alcohol/substance misuse, addiction, asthma, chronic obstructive pulmonary disease, HIV/AIDS, high cholesterol, obstructive sleep apnoea, spinal cord dysfunction, carers' psychological stress), there is not enough evidence to tell what effects ATCS have.

Only four trials reported adverse events. Our certainty in the evidence varied (high to very low), and was often lowered because of study limitations, meaning that further research may change some findings.

Conclusion

ATCS may be promising for changing certain health behaviours, improving health outcomes and increasing healthcare uptake.

Summary of findings

Background

Description of the condition

The demand for information and communication technology applications in healthcare settings is increasing, driven by an interest in facilitating active participation of consumers in managing their own health as well as by the need to develop platforms that have greater reach and are also more cost‐effective than traditional approaches. Automated telephone communication systems (ATCS) are applications that have been used to deliver both preventive healthcare programmes as well as services to manage long‐term conditions.

The range of ATCS interventions included in this review encompasses the following.

Unidirectional ATCS. This is the non‐interactive form, which enables one‐way, non‐interactive voice communication.
Interactive ATCS. These are systems that enable two‐way real‐time communication. The most common form of this is the interactive voice response system or IVR, which might be used, for example, to provide automated tailored feedback based on the monitoring of an individual's progress.
ATCS Plus. These are also interactive ATCS systems, but they are more complex and include additional functions, such as access to an advisor to ask questions.

Additionally, this review includes several multimodal/complex ATCS interventions, defined as any type of ATCS (unidirectional, IVR or ATCS Plus) delivered as part of a complex, multimodal package, such as symptom monitoring by a health professional plus automated monitoring via IVR plus provision of medications.

Primary preventive healthcare

Primary preventive healthcare focuses on keeping people well, preventing disease and injury, and educating people about adopting healthier behaviours (Family Health Teams 2006). There are two types of primary prevention strategies: health promotion and disease prevention (Figure 1).

A major challenge for healthcare systems is to deliver preventive services that systematically target the factors that contribute to ill health (Gullotta 2014). In the prevention of diabetes mellitus, for example, a combination of cognitive, physiological, and behavioural factors (such as lack of knowledge around risk factors, lack of physical activity and unhealthy diet) may contribute to the development of the condition. An effective preventive strategy would therefore need to take an integrative approach and target each of the influencing factors (Figure 2).

Influencing factors and preventive strategies in type 2 diabetes

One possible method of communicating preventive activities to the population is via information and communication technology (ICT) (Baranowski 2012; Haluza 2015).

Management of long‐term conditions

Long‐term conditions such as cardiovascular diseases, cancer, diabetes and chronic lung diseases are the leading causes of death globally (O'Dowd 2014). People with long‐term conditions face challenges such as dealing with complex symptoms, medication regimens, disability, and lifestyle adjustments (Carolan 2014; Demain 2015).

Effective chronic disease management programmes bring together relevant information systems with continuous follow‐up and targeted management, incorporating ICT to provide accessible and convenient educational information as well as self‐management tools for people with long‐term conditions (Galdas 2015).

ICT for primary prevention and management of long‐term conditions

Consumers increasingly use ICT for health in a myriad of ways, such as accessing medical records through web portals; communicating online with others, one‐on‐one or in a virtual community (Sawmynaden 2012); surfing the Internet to find information about health and health services; and transmitting health data or communicate messages using the web or the telephone (Pappas 2011).

There is some evidence that tools such as ATCS can successfully deliver health information to consumers, which facilitates health promotion (Cohen‐Cline 2014; Oake 2009b), enables active participation of consumers in managing their own care, and facilitates epidemiological and public health research by using collected patient data (Hendren 2014; Maheu 2001).

ICT can also support the delivery and administration of disease management programmes. There is evidence that ATCS can successfully deliver health information to patients for the management of long‐term conditions (Derose 2009; Derose 2013).

Description of the intervention

ATCS incorporate a specialised computer technology platform to deliver voice messages and collect information from consumers using either touch‐tone telephone keypads or voice recognition software (Piette 2012c). There are three types of ATCS.

Unidirectional ATCS enable one‐way, non‐interactive voice communication. This might include interventions such as automated reminder calls to take medication or perform other actions.
Interactive ATCS (e.g. IVR systems) enable two‐way real‐time communication, for example asking questions and receiving responses and individualised interventions (Reidel 2008; Rose 2015). Different studies have tested interactive ATCS for managing diabetes (Katalenich 2015; Khanna 2014), heart failure (Chaudhry 2010; Krum 2013), coronary heart disease (Reid 2007), and asthma (Bender 2010). They have also been used in health promotion initiatives, focusing on dietary behaviour (Delichatsios 2001; Wright 2014), physical activity (David 2012; Pinto 2002), and substance use (Aharonovich 2012).
ATCS Plus interventions are also interactive systems but include additional functions.

- Advanced communicative functions including access to an advisor to request advice (e.g. 'ask the expert' function), scheduled contact with an advisor (e.g. telephone or face‐to‐face meetings), and peer‐to‐peer access (e.g. buddy systems).
- Supplementary functions including automated, non‐voice communication (e.g. email or short messaging service (SMS)) (Webb 2010).

In this review, we also include several multimodal/complex ATCS interventions. These are more complex packages of care than ATCS Plus interventions and can include any type of ATCS (unidirectional, IVR or ATCS Plus) delivered as part of a complex, multimodal package, such as symptom monitoring by a health professional plus automated IVR monitoring plus provision of medications.

How the intervention might work

ATCS is a mode of communication that can replace or supplement some of the human‐to‐human telephone communication with a computer‐to‐human communication (Lieberman 2012; McCorkle 2011).

There is recognition that ATCS – like all other health interventions – should be underpinned by appropriate theoretical models (Krupinski 2006; Puskin 2010). These include the transtheoretical model (Prochaska 1984); theory of planned behaviour (Ajzen 1985); the health belief model (Rosenstock 1974); social cognitive theory (Bandura 2001); and self‐regulation theory (Leventhal 1984). Self‐management or preventive skills can be developed using any of these models (Barlow 2002).

There is evidence to suggest that behaviour change interventions underpinned by a theory can significantly enhance health behaviours (Fisher 2007; Gourlan 2015; Michie 2009; Webb 2010). Figure 3 shows a conceptual framework on how theories can influence health behaviour and illustrates how ATCS are used in preventive healthcare.

Conceptual framework of ATCS in preventive healthcare

Social cognition models assume that any health outcome is the consequence of the complex interaction between social, environmental, economic, psychological and biomedical factors (Edelman 2000; Jekauc 2015; Kelly 2009). These models focus on key concepts, such as self‐efficacy and attitudes to influence behaviour, which in turn can lead to behaviour change (Hardeman 2005; Michie 2010; Vo 2015).

Healthcare interventions delivered through disease management programmes, such as those underpinned by the chronic care model, have produced improved consumer care and health outcomes (Gee 2015; Lee 2011; Piatt 2006; Schillinger 2009). According to the chronic care model, management of long‐term conditions requires an interaction between a prepared, proactive team of practitioners and an informed, engaged consumer (Gammon 2015; Wagner 2002). This can be achieved through the interplay between elements such as self‐management support, delivery system design, decision support, and clinical information systems (Webb 2006). Figure 4 describes a framework illustrating how ATCS might work in the management of long‐term conditions using the chronic care model, by educating, monitoring, and coaching patients.

Conceptual framework of ATCS in the management of long‐term conditions

The importance of verbal communication is a complex psycholinguistic, cognitive‐emotional, and educational process that involves the transfer of information between a source (or sender) and a destination (or receiver); it largely depends on the topic/perspective of communication, perceived efficacy of communication, a person's mastery in encoding and understanding the semantic meaning decoded in verbal messages, and communicative intentions. However, other variables such as accent, voice tone, speech rate, and background noise also need to be taken into account when evaluating ATCS (Krauss 2001).

Advantages of automated telephone communication systems

ATCS as a data collection tools have a number of advantages over traditional face‐to‐face consultation (Rosen 2015). These include convenience, simplicity, anonymity, and low cost (Lee 2003; Piette 2012c). ATCS can provide access to health care 24 hours a day, seven days a week, along with immediate feedback to the consumer (Hall 2000; Schroder 2009). Both patients and healthcare professionals using ATCS have reported a high degree of user satisfaction, indicating that it is user‐friendly and convenient (Abu‐Hasaballah 2007).

ATCS technology can facilitate access to difficult‐to‐reach populations (e.g. people from a lower socioeconomic background) as ATCS require access only to a telephone (private or public) (Schroder 2009). Different authors have found ATCS to be acceptable to low‐literacy populations (Glasgow 2004; Piette 2007; Piette 2012c), and others have confirmed these findings in frail elderly patients (Mundt 2001). Unlike face‐to‐face interaction, which can elicit socially desirable responses, leading to under‐reporting of stigmatising behaviours and over‐reporting of socially desirable behaviours, ATCS have been found to elicit better self‐reporting of sensitive issues (e.g. substance misuse, alcohol use and sexual history) and reduce self‐reporting bias (Schroder 2009). They also have the potential to reduce healthcare delivery costs (Phillips 2015; Szilagyi 2013).

Disadvantages of automated telephone communication systems

Programming of ATCS involves investment in software and hardware, for example to enable multiple simultaneous calls and the development of a voice script appropriate for the target population and the topic of investigation (Piette 2007; Schroder 2009). ATCS may also present difficulties with the provision of immediate participant support. Should questions arise during the interview (Schroder 2009), ATCS cannot capture, interpret, or respond to the users' non‐verbal responses (Williams 2001). Individuals with physical disabilities (e.g. severe loss of hearing or speech) may have difficulty with ATCS (Mundt 2001). Others may simply have a strong preference for interactions with humans rather than with automated voice messages (Mahoney 2003). In addition, for individuals targeted by several ATCS‐based interventions, ATCS could lead to information overload and outright rejection of the interventions. Finally, protection of individually identifiable health information could be a challenge.

Why it is important to do this review

Existing reviews found evidence of effectiveness of ATCS in preventive healthcare and management of long‐term conditions (Krishna 2002; Lieberman 2012; Oake 2009b). However, none of those was conclusive, nor did they explore the theoretical basis or the mechanism of action of the intervention. The present review fills this gap by investigating the effects of interventions based on theoretical constructs and by exploring the behaviour change techniques implemented in the intervention (Abraham 2008; Michie 2011).

In addition, it is not clear which types of ATCS are most effective for prevention or management of long‐term conditions, in what setting, or for which conditions. This review aims to explore different interfaces of ATCS programme design and layout that may be used for diverse population groups (considering factors such as age, socioeconomic status, preferred language, and literacy) (Car 2004; Pappas 2011). Numerous randomised controlled trials (RCTs) evaluating the effectiveness of ATCS have recently been published.

A new systematic review is thus needed to critically assess the available evidence and to guide the implementation of ATCS in preventive healthcare and management of long‐term conditions.

Objectives

To assess the effects of ATCS for preventing disease and managing long‐term conditions on behavioural change, clinical, process, cognitive, patient‐centred and adverse outcomes.

Specific secondary objectives include:

determining which type of ATCS is most effective for preventive healthcare and management of long‐term conditions;
exploring which components of the interventional design contribute to positive consumer behavioural change;
exploring the behaviour change techniques and theoretical models underpinning the ATCS interventions.

Methods

Criteria for considering studies for this review

Types of studies

We included RCTs, cluster RCTs, quasi‐RCTs, interrupted time series (ITS) and controlled before‐and‐after (CBA) studies.

We included CBA and ITS studies because they are often used to draw conclusions about 'promising interventions' ready for trial when RCTs may be too expensive or simply impractical or where there are insufficient RCTs on a particular type of intervention (Centre for Reviews and Dissemination 2008; Higgins 2011; Jackson 2005). Interrupted time series designs can address cyclical trends (i.e. the outcome may be increasing or decreasing over time such as seasonal or other cyclical observations). To be considered for inclusion, these studies must have met the criteria specified by the Cochrane Effective Practice and Organisation of Care Review Group (EPOC) (Ryan 2009). For CBA designs, the timing of data collection for the control and intervention groups had to have been the same, there must have been at least two intervention sites and two control sites, and both groups would have been comparable on key characteristics related to demographics and intervention context. For ITS designs, the studies had to use a clearly defined point in time when the intervention occurred and at least three data points before and three after the intervention.

Types of participants

We included consumers, including carers, who received ATCS for prevention or management of long‐term conditions, regardless of age, sex, education, marital status, employment status, or income.
For management of long‐term conditions, we included consumers who had one or more concurrent long‐term conditions (i.e. multimorbidity).
We included consumers in all settings.

Types of interventions

The ATCS interventions included in this review included the following.

Unidirectional ATCS: non‐interactive ATCS enabling one‐way voice communication.
Interactive ATCS: systems that enable two‐way, real‐time communication, such as interactive voice response systems or IVR.
ATCS Plus: interactive ATCS systems including additional functions.

The review also included several multimodal/complex ATCS interventions, defined as any type of ATCS (unidirectional, IVR or ATCS Plus) delivered as part of a complex, multimodal package.

We included studies that evaluated either unidirectional ATCS or interactive ATCS. We also included studies that compared ATCS interventions (e.g. unidirectional ATCS versus interactive ATCS and/or ATCS Plus) to compare the effects of different intervention designs on preventive healthcare or management of long‐term conditions.

Interactive ATCS had an automated function such as automated tailored feedback based on individual progress monitoring (e.g. comparison to norms or goals, reinforcing messages, coping messages, and automated follow‐up messages). Although our protocol (Cash‐Gibson 2012) indicated that we would include ATCS Plus interventions only if the study explicitly reported that the effects of the intervention could be attributed to the ATCS component, in the review we included all types of ATCS Plus interventions as, in a complex intervention such as this, it would be impossible to attribute the intervention effect to one of the intervention components. We also included studies that delivered any type of ATCS (unidirectional, IVR, or ATCS Plus) as part of a complex, multimodal (package) intervention.

The interventions were delivered for one or more types of prevention or one or more types of management for long‐term conditions, as illustrated in Figure 1 and Figure 5, respectively.

We excluded studies in which interventions:

targeted health professionals or teachers for educational purposes;
were exclusively for the purpose of electronic history‐taking or data collection or risk assessment with no health promotion or interactive elements;
involved only a non‐ATCS component such as face‐to‐face communication or written communication;
were web‐based interventions that were accessed via a mobile phone.

Comparisons were made against various controls or standard or enhanced forms of usual care (i.e. no ATCS intervention). We also included comparisons of one type of ATCS against another, or the same type of ATCS that was delivered via different delivery modes (e.g. landline telephone versus mobile phone).

As part of this review, we piloted and applied the intervention Complexity Assessment Tool for Systematic Reviews version 1 (iCAT_SR) for assessing complex, multimodal interventions and reported results narratively/qualitatively (Lewin 2015).

Types of outcome measures

Primary outcomes

Primary outcomes consisted of health behaviour and clinical outcomes (defined below). For each study, we included all relevant primary outcomes, as these are likely to be most meaningful to clinicians, consumers, the general public, administrators and policymakers (Chandler 2013). Given the wide spread of the included studies and the fact that this review represents the first attempt to systematically assess all relevant evidence on broadly defined ATCS interventions, we felt that it was important to capture and report as much relevant information on outcomes and effects of interventions as possible, in order to assist with comprehensively mapping where the evidence lies and how it has been assessed. In future updates to this review, we may consider modifying this approach to focus on a smaller number and range of outcomes if this is likely to improve the clarity and meaningfulness of the collected data.

We reported the following outcomes in 'Summaries of findings' tables.

1. Health behaviour outcomes (category)

Changes in health‐enhancing behaviour (e.g. physical activity, adherence to medications/uptake of recommended laboratory or other testing)
Risk‐taking behaviour (e.g. tobacco consumption)

This outcome was either self‐reported or collected using a validated questionnaire that was either self‐administered or completed in an interview. In studies that measured the same outcome using both a self‐reported measure and an objective measure, we used the objective measure. For example, if a study on physical activity measured Metabolic Equivalent of Task (MET) scores using a self‐reported, seven‐day physical activity recall as well as a pedometer, we used the score obtained from the (objective) pedometer.

2. Clinical outcomes (category)

Physiological measures (e.g. blood pressure)
Blood biochemistry (e.g. glucose levels)

Secondary outcomes

For each study, we selected all relevant secondary outcomes as these were also meaningful for the various stakeholders.

1. Process outcomes (category)

Change in acceptability of service (e.g. consumer accessibility and usability of the interventions to apply information and support supplied through ATCS)
Satisfaction (e.g. patient and carer satisfaction with the intervention)
Cost‐effectiveness

2. Cognitive outcomes (category)

Changes in knowledge (i.e. accurate risk knowledge and perception)
Attitude and intention to change
Self‐efficacy (i.e. a person's belief in their capacity to carry out a specific action)

3. Patient‐centred outcomes (category)

Quality of life

4. Adverse outcomes

Unintended adverse events attributable to the intervention

Search methods for identification of studies

Electronic searches

We searched the following electronic databases.

The Cochrane Central Register of Controlled Trials (CENTRAL; 2015; Issue 5) in the Cochrane Library (searched 12 May 2015);
MEDLINE OvidSP (1980 to 12 May 2015);
Embase OvidSP (1980 to 12 May 2015);
PsycINFO OvidSP (1980 to 12 May 2015);
CINAHL EBSCOhost (Cumulative Index to Nursing and Allied Health Literature; 1980 to 12 June 2015);
Web of Science (1980 to 19 May 2015);
GlobalHealth EBSCOhost (1980 to 16 June 2015);
WHOLIS (1980 to 17 June 2015);
LILACS (1982 to 17 June 2015); and
ASSIA ProQuest (Applied Social Sciences Index and Abstracts; 1987 to 20 May 2015).

We detail the search strategies for each database in respective appendices: CENTRAL (Appendix 1), MEDLINE (Appendix 2), Embase (Appendix 3), PsycINFO (Appendix 4), CINAHL (Appendix 5), Web of Science (Appendix 6), Global Health (Appendix 7), and WHOLIS (Appendix 8). We also present the list of keywords used in trial registers (Appendix 9) and grey literature (Appendix 10).

We searched most databases from 1980 onwards because we expected that any prior evidence would have been incorporated into subsequent research, and because technology has advanced dramatically over the last thirty years, so integration of older research should be interpreted only in light of new findings. We tailored search strategies to each database and reported them in the review. There were no language restrictions.

Searching other resources

We searched grey literature (Dissertation Abstracts, Index to Theses, Australasian Digital Theses). We contacted experts in the field and authors of included studies for advice as to other relevant studies. We searched reference lists of relevant studies, including all included studies and previously published reviews. We also searched online trial registers (e.g. Current Controlled Trials, www.controlled‐trials.com; www.clinicaltrials.gov) for ongoing and recently completed studies.

Data collection and analysis

Selection of studies

We merged search results across databases using (EndNote 2015) reference management software and removed duplicate records. Following de‐duplication, two authors (PP, NM) independently examined titles and abstracts of records retrieved from the search. We retrieved the full text of the potentially relevant studies and assessed their eligibility according to the inclusion criteria. We linked multiple reports of the same study in order to determine whether the study was eligible for inclusion. Two authors independently performed both the initial screening and the full text screening. Authors corresponded with each other to make final decisions on study inclusion and resolved disagreement about study eligibility through discussion with a third review author (JC). We describe excluded studies, with reasons for exclusion, in Characteristics of excluded studies. We used an adapted PRISMA flow chart to describe the study selection process Figure 6 (Higgins 2011).

Data extraction and management

Two authors (PP, NM) independently extracted relevant characteristics related to participants, intervention, comparators, outcome measures, and results (effectiveness of the interventions) from all the included studies using a standard data collection form; any disagreements were resolved by discussion. We sought relevant missing information on the trial, particularly information required to judge the risk of bias, from the original author(s) of the article. One author (PP) transferred all the data from the extraction form into the Review Manager (RevMan) software while another author (NM) confirmed the accuracy of the transferred data (RevMan 2014).

Assessment of risk of bias in included studies

We included all studies meeting the inclusion criteria regardless of the outcome of the assessment of risk of bias. We assessed and reported on the methodological risk of bias of included studies in accordance with the Cochrane Handbook for Systematic Reviews of Interventions and Cochrane Consumers and Communication guidelines (Higgins 2011; Ryan 2011), which recommend explicitly reporting the following individual elements for RCTs: random sequence generation; allocation concealment; blinding (participants, personnel); blinding (outcome assessment); completeness of outcome data (attrition bias), selective outcome reporting (relevant outcomes reported); other sources of bias (baseline imbalances). For cluster RCTs, we also assessed and reported the risk of bias associated with an additional domain: selective recruitment of cluster participants. We referred to Cochrane Consumers and Communication guidelines to narratively describe the results of risk of bias for each domain for all included studies (Ryan 2011). We reported our assessment of risk of bias for each domain and included study, with a descriptive summary/synthesis of our judgments. In all cases, two authors (PP, NM) independently assessed the risk of bias of included studies, resolving any disagreements by discussion and consensus. We also contacted several study authors for additional information about the included study or for clarification of the study's methods. We incorporated the results of the risk of bias assessment into GRADE assessments and the review itself through standard tables together with systematic and descriptive summary, leading to an overall assessment of the risk of bias of included studies and a judgment about the internal validity of the review's results.

Measures of treatment effect

For dichotomous data, we reported risk ratios (RR), odds ratios (OR), or hazard ratios (HR), as well as their 95% confidence intervals (CI) and P values. For continuous data we reported mean values and standard deviations (SD) of the outcomes in each intervention group along with the number of participants and P values (Table 13).

1. Results.

Dichotomous outcomes
Primary outcome	Study ID	Timing of outcome assessment (months)	Intervention group		Comparator group		Between group difference			Notes
Primary outcome	Study ID	Timing of outcome assessment (months)	Observed (n)	Total (N)	Observed (n)	Total (N)	P value	Effect estimate (OR/RR/HR)	95% CI	Notes
IMMUNISATIONS
Immunisation uptake	Dini 2000	24	107	189	87	186	—	1.21	0.99 to 1.47	—
	Franzini 2000	*	270	314	273	429	—	—	—	—
	Hess 2013	3	146	5599	46	6383	< 0.001	3.69	2.64 to 5.15	Cluster RCT unadjusted for clustering. Approximate sample size calculations gave the following adjusted values: intervention 20/791; control 6/902; see Appendix 14 for calculations
	Lieu 1998	4	89	167	70	162	0.11	—	29.0 to 43.8	CIs for % values and for IVR alone; P value from Chi²
	Linkins 1994	1	1684	4636	955	3366	< 0.01	1.28	1.20 to 1.37	—
	LeBaron 2004	13	305	763	259	763	< 0.05	—	—	—
	Nassar 2014	2	3	26	3	24	—	—	—	—
	Stehr‐Green 1993	1	46	112	41	110	—	1.07	0.78 to 1.46	—
	Szilagyi 2006	18	928	1496	873	1510	0.02	—	—	—
	Szilagyi 2013	12	748	1423	651	1296	< 0.05	1.3	1.0 to 1.7	—
SCREENING
Screening rate	Baker 2014	6	191	225	90	225	< 0.001	—	—	—
	Cohen‐Cline 2014	6	801	8005	234	3005	0.0012	1.32	1.14 to 1.52
	Corkrey 2005	3	—	45,303	—	30,229	—	—	1.28 to 1.42^a 0.11 to 0.17^b	^aWomen aged 50‐69 years ^bWomen aged 20‐49 years
	DeFrank 2009	10 to 14	960	1355	574	847	0.014	1.32	1.06 to 1.64	—
	Fiscella 2011	12	55^a 47^b	134^a 163^b	23^a 16^b	137^a 160^b	—	3.44^a 3.70^b	1.91 to 6.19^a 1.93 to 7.09^b	^aBreast cancer screening; ^bColorectal cancer screening
	Fortuna 2014	12	36^a 24^b	158^a 157^b	28^a 19^b	157^a 156^b	> 0.05	1.4^a 1.3^b	0.8 to 2.4^a 0.7 to 2.5^b	^aBreast cancer screening; ^bColorectal cancer screening (both crude estimates)
	Hendren 2014	12	30^a 43^b	101^a 114^b	15^a 21^b	90^a 126^b	0.034^a 0.0002^b	1.96^a 3.22^b	0.87 to 4.39^a 1.65 to 6.30^b	^aBreast cancer screening; ^bColorectal cancer screening
	Heyworth 2014	12	385	1565	290	1558	< 0.001	—		—
	Mosen 2010	6	662	2943	474	2962	< 0.001	1.31	1.10 to 1.56	—
	Phillips 2015	9	19^a 33^b	90^a 198^b	17^a 27^b	88^a 199^b	—	—	—	^aBreast cancer screening; ^bColorectal cancer screening
	Simon 2010a	12	3192	10,432	3194	10,506	0.76	1.01	0.94 to 1.07	In the adjusted model
	Solomon 2007	10	144	997	97	976	0.006	1.52	1.13 to 2.05	In the adjusted model
APPOINTMENT REMINDERS
Reducing non‐attendance rates	Dini 1995	1	144	277	78	240	< 0.05	1.60	1.29 to 1.98	—
	Griffin 2011	1.5	333^a 169^b	794^a 411^b	324^a 164^b	790^a 409^b	> 0.05	—	−6 to 5^a −8 to 7^b	^aColonoscopy ^bFlexible sigmoidoscopy
	Maxwell 2001	2	347	700	322	670	> 0.05	—	—	—
	Parikh 2010	4	2662	3219	2576	3350	< 0.001	1.52	1.34 to 1.71	—
	Reekie 1998	1.5	473	500	453	500	< 0.001	3.41	1.87 to 6.20	—
	Tanke 1994	6	257	407	235	456	< 0.01	1.50	—	—
	Tanke 1997	3 days	652	701	617	701	< 0.05	1.71	—	—
ADHERENCE
Adherence to medications/laboratory tests	Bender 2010	2.5	16	25	12	25	0.003	—	—	—
	Derose 2009	3	453	2199	298	1550	0.31	1.09	0.92 to 1.28	At 8 weeks differences were not significant (P = 0.23)
	Feldstein 2006	25 days	177	267	53	237	< 0.001	4.1	3.0 to 5.6	—
	Friedman 1996	6	24	133	16	134	0.03	—	—	—
	Glanz 2012	12	47	157	42	155	> 0.05	—	—	—
	Green 2011	2 weeks	1180	4124	958	4182	< 0.001	—	—	—
	Ho 2014	12	109	122	88	119	0.003	—	—	—
	Lim 2013	5	29	38	34	42	0.233	—	—	After the mid‐study visit
	Migneault 2012	12	—	169	—	168	0.19	—	—	—
	Mu 2013	1	1096975	4153634	18395	84187	< 0.001	—	—	—
	Patel 2007	3 to 6	3362	6833	1865	4172		—	—	—
	Reynolds 2011	2 weeks	4318	15,356	3309	15,254	< 0.001	—	—	—
	Sherrard 2009	6	70	137	55	143	0.041	0.60	0.37 to 0.96	Primary composite outcome of adherence and adverse effects (emergency room visit and hospitalisation)
	Simon 2010b	12	—	600	—	600	—	0.93	0.71 to 1.22	—
	Stacy 2009	6	178	253	148	244	< 0.05	1.54	1.13 to 2.10	—
	Vollmer 2011	18	—	3171	—	3260	0.002	—	0.01 to 0.03	P value and CIs for Δ change
	Vollmer 2014	12	—	7247	—	7255	0.022	—	0.011 to 0.034	P value and CIs for Δ change
HEART FAILURE
Heart failure hospitalisation	Capomolla 2004	10 ± 6 (median 11)	17	67	58	66	< 0.05	—	—	—
Heart failure hospitalisation	Kurtz 2011	12	4	32	17	50	< 0.05	—	—	This was a cluster outcome: "cardiovascular deaths and hospitalisations‐ which ever event occurred first"
All‐cause mortality	Capomolla 2004	10 ± 6 (median 11 )	5	67	7	66	> 0.05	—	—	—
	Chaudhry 2010	6	92	826	94	827	0.86	0.97	0.73 to 1.30	Death or readmission
	Krum 2013	12	17	170	16	209	0.439	1.36	0.63 to 2.93	—
Cardiac mortality	Capomolla 2004	10 ± 6 (median 11)	2	67	6	66	> 0.05	—	—	—
Cardiac mortality	Kurtz 2011	12	3	32	5	50	> 0.05	—	—	Cluster outcome: "cardiovascular deaths and hospitalisations‐ which ever event occurred first"
SMOKING
Smoking abstinence	Brendryen 2008	12	74	197	48	199	0.02	1.91	1.12 to 3.26	—
	Ershoff 1999	9	20	120	25	111	—	—	—	—
	McNaughton 2013	12	12	23	14	21	0.33	—	—	—
	Regan 2011	3	105	361	95	364		1.13	0.90 to 1.41	—
	Reid 2007	12	23	50	17	49	0.25	1.60	0.71 to 3.60	—
	Rigotti 2014	6	51	198	30	199	0.009	1.71	1.14 to 2.56	—
	Velicer 2006	30	75	500	65	523	—	—	—	For 6 month prolonged abstinence
Continous outcomes
Primary outcome	study ID	Timing of outcome assessment (months)	Intervention group		Comparator group		Between‐group difference			Notes
Primary outcome	study ID	Timing of outcome assessment (months)	Mean	Standard deviation of change (SD) or SD	Mean	Standard deviation of change (SD) or SD	Change	Confidence intervals	P values	Notes
ALCOHOL CONSUMPTION
Drinks per drinking day	Hasin 2013	2	3.5	1.8	4.7	3.2	1.38	1.12 to 1.70	< 0.01	CIs are for the effect size
Drinks per drinking day	Rose 2015	2	4	0.4	4.3	0.4	—	—	0.45	—
CANCER
Symptom severity	Cleeland 2011	1	—	—	—	—	—	—	—	Effect size: intervention = 0.75; control = 0.68
	Mooney 2014	1.5	5.76 to 7.36 (range)	—	5.55 to 7.44 (range)	—	0.06	—	0.58	—
	Sikorskii 2007	2.5	20.73	—	20.80	—	—	—	> 0.05	Effect size: intervention = 0.59; control = 0.56
	Spoelstra 2013	2.5	11.6	—	11.0	—	—	—	0.02	—
DIABETES
Glycated haemoglobin (%)	Graziano 2009	3	7.87	1.09	7.82	1.14	—	—	0.89	—
	Katalenich 2015	6	8.10		7.90	—	—	—	> 0.05	Median values
	Khanna 2014	3	9.1	1.9	8.6	1.3	—	—	0.41	—
	Kim 2014	12	9.0	2	9.9	2.2	—	—	0.02	—
	Lorig 2008	6	7.0	1.4	7.3	1.5	—	—	0.04	—
	Piette 2001	12	8.1	1.15	8.2	1.18	—	—	0.3	—
	Schillinger 2009	12	8.7	1.9	9.0	2.2	0.1	0.5 to 0.4	0.8	—
	Williams 2012	6	7.9	1.2	8.7	1.8	0.91	0.86 to 0.93	0.002	—
Serum blood glucose (mg/dL)	Piette 2001	12	180	9	172	10	—	—	0.6	—
Serum blood glucose (mg/dL)	Homko 2012	26	107.4	12.9	109.7	16.5	—	—	0.44	—
Self‐monitoring of blood glucose	Graziano 2009	3	1.9	1.07	1.3	0.75	—	—	< 0.001	—
	Lorig 2008	6	0.05	0.387	0.08	0.365	—	—	0.457	—
	Piette 2001	12	4.6	0.1	4.4	0.1	—	—	0.05	—
	Schillinger 2009	12	4.3	2.6	3.3	2.9	0.8	0.1 to 1.5	0.03	—
Self‐monitoring of diabetic foot	Piette 2001	12	4.6	0.1	4.4	0.1	—	—	—	—
Self‐monitoring of diabetic foot	Schillinger 2009	12	5.1	1.4	4.6	1.7	0.6	0.2 to1.0	0.002	—
HYPERTENSION
Systolic Blood Pressure (mm Hg)	Dedier 2014	3	136.4	83.5	138.9	81.5	—	—	> 0.05	—
	Goulis 2004	6	123.8	14.2	128.6	19.4	—	—	> 0.05	—
	Friedman 1996	6	158	*	160.2	*	−1.8	—	0.20	—
	Harrison 2013	1	141.2	15.1	143.1	14.6		—	< 0.001	—
	Magid 2011	6	137.4	19.4	136.7	17.0	−0.7	—	0.006	—
	Piette 2012	6 weeks	142.5	2.3	143.6	2.4	−4.2	−9.1 to 0.7	0.09	—
Diastolic Blood Pressure (mm Hg)	Friedman 1996	6	80.9	—	83.2	—	—	—	0.02	—
	Goulis 2004	6	74.6	8.5	79.5	14.0	—	—	> 0.05	—
	Harrison 2013	1	80.3	12.6	81.3	12.5	—	—	< 0.001	—
	Magid 2011	6	82.9	12.9	81.1	11.7	−2.3	−4.9 to −0.2	0.07	—
OBSTRUCTIVE SLEEP APNOEA SYNDROME (OSAS)
Continuous positive airway pressure (CPAP) use	DeMolles 2004	2	4.4	—	2.9	—	—	—	0.076	—
Continuous positive airway pressure (CPAP) use	Sparrow 2010	12	—	—	—	—	—	1.18 to 2.48	0.004	—
WEIGHT MANAGEMENT
BMI in adults	Bennett 2012	18	36.54	2.01	36.84	1.90	−0.35	−0.75 to 0.06	—	—
	Bennett 2013	18	29.8	1.90	30.3	1.93	−0.6	−1.2 to −0.1	0.03	—
	Goulis 2004	6	33.7	5.2	37.2	8.7	—	—	0.06	—
BMI‐z scores in children	Estabrooks 2009	12	1.95	0.04	1.98	0.03	—	—	> 0.05	—
BMI‐z scores in children	Wright 2013	3	1.9	0.28	1.9	0.3	−0.03	—	0.48	—

Study ID	Study type^a	Study subtype^b	Country	Sample size	Mean age (years unless stated otherwise)	Male (%)	Female (%)	Ethnicity^c	Duration of condition	Comorbidities, medication	Incentives for participation	Incentives
Tucker 2012	P	Alcohol misuse	USA	187	45	63	37	White ‐ 54% Other ‐ 46%	—	—	Yes	Visa gift cards or checks (USD 50 per in‐person interview, USD 15 per phone interview). IG participants received USD 0.50 minimum for each daily call and USD 1.00 after seven consecutive calls
Franzini 2000	P	I	USA	1138	—	—	—	—	—	—	—	—
Hess 2013	P	I	USA	11,982	72	—	—	—	—	—	—	—
Dini 2000	P	I	USA	1227	2‐3 months	—	—	—	—	—	—	—
LeBaron 2004	P	I	USA	3050	9 months^d	49	51	Black ‐ 76% Hispanic ‐ 14% White ‐ 7% Other – 3%	—	—	—	—
Lieu 1998	P	I	USA	752	20 months	—	—	—	—	—	—	—
Linkins 1994	P	I	USA	8002	—	51	49	Black ‐ 50% White ‐ 45% Other – 5%	—	—	—	—
Nassar 2014	P	I	USA	50	24	0	100	Black ‐ 86% White ‐ 14%	—	—	Yes	USD 35 per month to help pay for cell phone service or a free, unlimited minutes cell phone until 6 weeks postpartum
Stehr‐Green 1993	P	I	USA	229	9 months	52	48	Black ‐ 90% Other – 7% Hispanic ‐ 3%	—	—	—	—
Szilagyi 2006	P	I	USA	3006	—	51	49	Other – 41% Black ‐ 35% White ‐ 17% Hispanic ‐ 7%	—	—	—	—
Szilagyi 2013	P	I	USA	4115	14	50	50	—	—	—	—	—
David 2012	P	Physical activity	USA	71	57	0	100	White ‐ 93% Other ‐ 7%	—	—	—	—
Dubbert 2002	P	Physical activity	USA	181	69	99	1	—	—	Mean (SD) number of comorbidities IG: 3.8 (1.5) CG: 3.9 (1.4)	Yes	USD 15 for completing each  visit to help defray expenses
Jarvis 1997	P	Physical activity	USA	85	67	24	76	Other ‐ 70% Black ‐ 30%	—	Mean co‐morbidities: 3	—	—
King 2007	P	Physical activity	USA	218	61	31	69	White ‐ 90% Other ‐ 10%	—	—	—	—
Morey 2009	P	Physical activity	USA	398	78	100	0	White ‐ 77% Black ‐ 23%	—	Mean (SD) number of diseases IG: 5.2 (2.5) CG: 5.5 (2.7)	—	—
Morey 2012	P	Physical activity	USA	302	67	97	3	White ‐ 70%	—	Mean (SD) number of comorbidities IG: 4.2 (2.4) CG: 3.9 (2.4)	—	—
Pinto 2002	P	Physical activity	USA	298	46	28	72	White ‐ 45% Black ‐ 45% Other ‐ 10%	—	—	—	—
Sparrow 2011	P	Physical activity	USA	103	71	69	31	—	—	Depression (unclear %)	—	—
Baker 2014	P	Screening	USA	450	60	28	72	Hispanic 87% Other‐ 13%	—	≥ 1 long‐term conditions ‐ 68%	No	—
Cohen‐Cline 2014	P	Screening	USA	11,010	61	54	46	White ‐ 86% Other ‐ 14%	—	—	—	—
Corkrey 2005	P	Screening	Australia	75,532	—	0	100	—	—	—	—	—
DeFrank 2009	P	Screening	USA	3547	—	0	100	White ‐ 88% Black ‐ 11% Asian or Other ‐ 1%	—	—	—	—
Durant 2014	P	Screening	USA	47,097	58	47	53	—	—	—	—	—
Fiscella 2011	P	Screening	USA	469	—	56 (for colorectal cancer)	44 (for colorectal cancer)	White ‐ 61% Black ‐ 28% Latinos ‐ 5% Asian ‐ 5%	—	—	—	—
Fortuna 2014	P	Screening	USA	1008	—	45	55	White ‐ 48% Black ‐ 37% Other ‐ 15%	—	—	No	—
Hendren 2014	P	Screening	USA	366	—	—	—	White ‐ 50% Black ‐ 41% Other (including Hispanic) ‐ 9%	—	—	—	—
Heyworth 2014	P	Screening	USA	4685	57	0	100	—	—	Anticonvulsants ‐ 6% Corticosteroids ‐ 4% COPD (unclear %) Oophorectomy ‐ 3%	—	—
Mosen 2010	P	Screening	USA	6000	60	50	50	White ‐ 93% Other ‐ 7%	—	Obesity ‐ 40%	—	—
Phillips 2015	P	Screening	USA	685	58	38	62	Non‐Hispanic white ‐ 78% Black ‐ 13% Other ‐ 9%	—	—	—	—
Simon 2010a	P	Screening	USA	20,936	57	47	53	White ‐ 86% Other ‐ 9% Black ‐ 5%	—	—	—	—
Solomon 2007	P	Screening	USA	1973	69	8	92	—	—	Use of oral glucocorticoids ‐ 22% Fractures ‐ 12%	—	—
Mahoney 2003	P	Stress management	USA	100 (caregiver)	63 (caregiver)	22 (caregiver)	78 (caregiver)	Black ‐ 64% Hispanic ‐ 21% White ‐ 15% Other ‐ 1%	—	—	—	—
Aharonovich 2012	P	Substance use	USA	33	46	76	24	White ‐ 64% White ‐ 15% Hispanic ‐ 21%	—	HIV medication ‐ 64% Hepatitis A, B, or C ‐ 49%	Yes	USD 20 gift certificates for each assessment
Bennett 2012	P	Weight management	USA	365	55	31	69	Black ‐ 71% Hispanic ‐ 13% White ‐ 4% Other ‐ 2%	—	Cholesterol medication ‐ 36% Diabetes medication ‐ 30% Mean BMI ‐ 37 kg/m²	Yes	USD 50 reimbursement at the first 3 follow‐up visits and USD 75 at 24 months
Bennett 2013	P	Weight management	USA	194	35	0	100	Black ‐ 100%	—	Hypertension ‐ 36% Metabolic syndrome ‐ 31% Depression ‐ 22% Diabetes mellitus ‐ 7%	Yes	Reimbursements of USD 50 each at baseline and at all follow‐up study visits
Estabrooks 2008	P	Weight management	USA	77	59	29	71	White ‐ 68% Hispanic ‐ 18% Other ‐ 7% Black ‐ 4% Asian ‐ 3%	—	—	—	—
Estabrooks 2009	P	Weight management	USA	220	11	54	46	White‐ 63% Hispanic ‐ 26% Other ‐ 11%	—	—	—	—
Goulis 2004	P	Weight management	Greece	122	44	12	88	—	—	Hypertension ‐ 13% Diabetes mellitus ‐ 2%	—	—
Vance 2011	P	Weight management	USA	140	—	—	—	—	—	—	—	—
Wright 2013	P	Weight management	USA	50 (child)	10 (child) 40 (parent)	58 (child) 4 (parent)	42 (child) 96 (parent)	Black ‐ 72% Other ‐ 22% White ‐ 6% (parent)	—	BMI (child) ‐ 25.7 kg/m² BMI (parent) ‐ 34 kg/m²	Yes	For completing assessments (USD 40 parents; USD 10 child)
Dini 1995	E	Appointment reminder	USA	517	—	—	—	—	—	—	—	—
Griffin 2011	E	Appointment reminder	USA	3610	63	95	5	White ‐ 83% Other ‐ 16% Hispanic ‐ 1%	—	—	—	—
Maxwell 2001	E	Appointment reminder	USA	2304	29	—	100	Hispanic ‐ 66% Black ‐ 19% White ‐ 13% Other ‐ 2%	—	—	—	—
Parikh 2010	E	Appointment reminder	USA	12,092	56	43	57	—	—	—	—	—
Reekie 1998	E	Appointment reminder	UK	1000	—	33	67	—	—	—	—	—
Tanke 1994	E	Appointment reminder	USA	2008	19^d	54	46	Spanish‐speaking ‐ 39% Vietnamese‐speaking ‐ 28% English‐speaking ‐14% Other – 13% Tagalog‐speaking Filipino – 6%	—	—	—	—
Tanke 1997	E	Appointment reminder	USA	701	< 12^e	45	55	English‐speaking ‐ 59% Spanish‐speaking ‐ 29% Vietnamese‐speaking ‐ 3% Other – 9%	—	—	—	—
Moore 2013	M	Illicit drugs addiction	USA	36	41	58	42	White ‐ 58% Black ‐ 28% Other – 14%	On methadone treatment mean = 21.7	—	Yes	USD 20 per week for completing weekly assessments and providing a urine sample
Andersson 2012	M	Alcohol consumption	Sweden	1423	—	—	—	—	—	—	—	—
Hasin 2013	M	Alcohol consumption	USA	254	46	78	22	Black ‐ 49% Hispanic ‐ 45% Other – 6%	12.8y	HIV/AIDS (unclear %)	Yes	USD 20; USD 40 at last 2 post‐treatment follow‐ups
Helzer 2008	M	Alcohol consumption	USA	338	46	64	36	White ‐ 97%	Currently dependent ‐ 67%	—	Yes	USD 30 for the  index, USD 25 for the 3‐month, and USD 60 for the 6‐month assessment
Litt 2009	M	Alcohol consumption	USA	110	49	58	42	White ‐ 86% Black ‐9%, Hispanic‐3% Other‐2%	Mean of 1.2 (SD 2.4)  treatments for alcohol dependence	—	Yes	The possible total incentive was USD 50.00 per week
Mundt 2006	M	Alcohol consumption	USA	60	42	55	45	White ‐ 95% Black ‐5%	52.3 heavy drinking days within past 3 months	—	Yes	Patients were paid USD 75 for the 30‐day follow‐up, USD 125 for the 90‐day follow‐up, and USD 200 for the 180‐day follow‐up
Rose 2015	M	Alcohol consumption	USA	158	49	53	47	—	Regular alcohol use mean = 17.94 years	—	Yes	USD 25 for each interview
Rubin 2012	M	Alcohol consumption	USA	47	57	60	40	Caucasian ‐ 83% African‐American ‐ 13%	—	—	—	—
Simpson 2005	M	Alcohol consumption	USA	98	46	91	9	White ‐ 45% Black ‐ 40% Native American ‐ 7% Other ‐ 6% Hispanic ‐ 2%	—	—	Yes	USD 25.00 each for the baseline and for the follow‐up assessments
Vollmer 2006	M	Asthma	USA	6948	52	35	65	White ‐ 92% Other ‐ 8%	—	Beta agonist ‐ 55% Oral steroids ‐ 46% COPD ‐ 33%	—	—
Xu 2010	M	Asthma	Australia	121	7	53	47	—	—	—	—	—
Cleeland 2011	M	Cancer	USA	79	60	53	47	White ‐ 85% Black ‐ 15%	—	—	—	—
Kroenke 2010	M	Cancer	USA	405	59	32	68	White ‐ 80% Black‐ 18% Other ‐ 2%	—	Depression ‐ 76% Pain ‐ 68%	—	—
Mooney 2014	M	Cancer	USA	250	55	24	76	White/Caucasian ‐ 91% Other ‐ 9%	—	—	—	—
Siegel 1992	M	Cancer	USA	239	58	50	50	White ‐ 89% Black‐ 6% Hispanic ‐ 4% Other 1%	Mean (SD) time since cancer diagnosis‐ IG: 36 months (35) CG: 26 months (32)	Mean (SD) symptoms (out of 13) IG: 3.4 (2.2) CG: 4.0 (2.4)	—	—
Sikorskii 2007	M	Cancer	USA	437	57	25	75	—	—	Mean comorbidities: 2	—	—
Spoelstra 2013	M	Cancer	USA	119	60	31	69	White ‐ 76% Asian ‐ 17 % Black ‐ 7%	—	Capecitabine ‐ 35% Erlotinib ‐ 24% Lapatinib ‐ 9% Imatinibf ‐ 8% Temozolomide ‐ 6% Sunitinib ‐ 5%  Sorafenib ‐ 2.5% Methotrexate ‐ 1.7%  Cyclophosphamide ‐ 0.8%	—	—
Yount 2014	M	Cancer	USA	253	61	49	51	White ‐ 58% Black ‐ 36% Other ‐ 6%	—	Planned single chemotherapy ‐ 9% Planned combination chemotherapy ‐ 90%	—	—
Kroenke 2014	M	Chronic pain	USA	250	55	83	17	White ‐ 77 %	≤ 5 years = 29% 6‐10 years = 19% > 10 years = 52%	Major depression‐ 24% Post‐traumatic stress disorder‐ 17%	—	—
Naylor 2008	M	Chronic pain	USA	55	46	14	86	White ‐ 96% Other ‐ 4%	—	—	—	—
Halpin 2009	M	COPD	UK	79	69	74	26	—	—	SABA ‐ 75% LAMA ‐ 43% LABA/ICS ‐ 43% ICS ‐ 33% SAMA ‐ 32% Oral steroids ‐ 25% LABA ‐ 18%	—	—
Adams 2014	M	Adherence	USA	475	5 (child) 35 (parent)	52 (child) 7 (parent)	48 (child) 93 (parent)	Black ‐ 67% (child) 47% (parent); Other – 33% (child) 53% (parent)	—	—	Yes	Gift cards
Bender 2010	M	Adherence	USA	50	42	36	64	White ‐ 58% Black ‐ 20% Hispanic ‐ 18% Asian ‐ 4%	—	—	Yes	USD 25 for each completed visit
Bender 2014	M	Adherence	USA	1187	—	—	—	—	—	—	—	—
Boland 2014	M	Adherence	USA	70	66	49	51	African American ‐ 58% European ‐ 32% Asian ‐ 6% Hispanic ‐ 3% Middle Eastern ‐ 1%	Median 5 years in IG; 4.5 years in CG	Bimatoprost ‐ 11.5% Travoprost ‐ 17.5% Latanoprost ‐ 71.5% Bilateral medication ‐ 70%	—	—
Cvietusa 2012	M	Adherence	USA	1393	—	—	—	—	—	—	—	—
Derose 2009	M	Adherence	USA	13,057	51	54	46	Other ‐ 48 % White ‐ 23% Hispanic ‐ 14% Black ‐ 10% Asian ‐ 5%	—	—	—	—
Derose 2013	M	Adherence	USA	5216	56	49	51	Hispanic ‐ 30% White ‐ 28% Unknown ‐ 23% Black ‐ 10% Asian and Pacific Islander ‐ 7.1% Other ‐ 1.7% Native American‐ 0.2%	—	Mean low‐density lipoproteins = 146 mg/dL	—	—
Feldstein 2006	M	Adherence	USA	961	59	47	53	—	—	Statins ‐ 32% Depression ‐ 11%	—	—
Friedman 1996	M	Adherence	USA	267	77	23	77	Other ‐ 89 % Black: 11%	—	Other – 81% Heart disease ‐ 32% Diabetes mellitus ‐ 18% Stroke ‐ 7%	—	—
Glanz 2012	M	Adherence	USA	312	63	62	38	Black ‐ 91% White ‐ 9%	—	—	Yes	USD 25 gift card
Green 2011	M	Adherence	USA	8306	—	—	—	—	—	—	—	—
Ho 2014	M	Adherence	USA	241	64	98	2	White ‐ 78%	—	Hypertension ‐ 91% Hyperlipidaemia ‐ 85% Diabetes mellitus ‐ 45% Chronic kidney disease ‐ 23% Chronic lung disease ‐ 20% Prior heart failure ‐ 12% Peripheral arterial disease ‐ 10% Cerebrovascular disease ‐ 7%	—	—
Leirer 1991	M	Adherence	USA	16	71	31	69	—	—	—	Yes	USD 25 for participating
Lim 2013	M	Adherence	USA	80	66	51	49	White ‐ 62% African‐American ‐ 10% Hispanic/Latino ‐ 9% Asian ‐ 9% East Indian ‐ 6%	Mean IG: 25.79 months; CG: 22.1 months	Number of medical problems: IG: 3.43 CG: 3.32	—	—
Migneault 2012	M	Adherence	USA	337	57	30	70	Black ‐ 100%	—	BMI ‐ 34.4 kg/m² Diabetes ‐ 38% Stroke ‐ 7.5%	—	—
Mu 2013	M	Adherence	USA	4,237,821	56	38.5	61.5	—	—	Correspondence with the author: "All participants were on maintenance medications"	—	—
Ownby 2012	M	Adherence	USA	27	80	—	—	—	—	Participants had cognitive (memory) impairment and were on donepezil, rivastigmine, or galantamine	—	—
Patel 2007	M	Adherence	USA	15,051	57	53	47	—	—	—	—	—
Reynolds 2011	M	Adherence	USA	30,610	—	—	—	—	—	—	—	—
Sherrard 2009	M	Adherence	Canada	331	63	—	—	—	—	—	—	—
Simon 2010b	M	Adherence	USA	1200	51	62	38	Other ‐ 95% Black ‐ 5%	—	Insulin ‐ 19.4% (participants)	—	—
Stacy 2009	M	Adherence	USA	497	54	38	62	—	—	Lipitor – 54% Zocor – 16% Other statin – 16%	—	—
Stuart 2003	M	Adherence	USA	647	—	—	—	—	—	—	—	—
Vollmer 2011	M	Adherence	USA	8517	54	34	66	White ‐ 50% Other – 26% Asian ‐ 11% Mixed – 7% Native Hawaiian/Pacific Islander ‐ 4% Black ‐ 2% American Indian/Alaskan Native – 1%	—	COPD ‐ 33%	—	—
Vollmer 2014	M	Adherence	USA	21,752	64	53	47	White ‐ 47% Asian ‐ 17% Black ‐15% Native Hawaiian/Pacific Islander ‐ 11% Black ‐ 2% American Indian/Alaskan Native – 1%	—	Diabetes mellitus ‐ 78% CVD ‐ 36% Statin only ‐ 40% ACEI/ARB only ‐ 25% Statin and ACEI/ARB ‐ 35% low‐density lipoproteins among statin users (mean) = 93.4 mg/dL	—	—
Graziano 2009	M	Diabetes mellitus	USA	119	62	55	45	White ‐ 77% Other – 23%	—	—	Yes	USD 25
Homko 2012	M	Diabetes mellitus	USA	80	30	0	100	White ‐ 41% Black ‐ 34% Hispanic ‐ 18% Asians or others ‐ 7%	—	BMI ‐ 34.1 kg/m²	—	—
Katalenich 2015	M	Diabetes mellitus	USA	98	59	40	60	Black ‐ 65% White ‐ 30% Other ‐ 3% Hispanic ‐ 1% Asian ‐ 1%	—	Other antidiabetic medications + insulin ‐ 80% Basal‐bolus regimen ‐ 40% Long‐acting insulin only ‐ 33% Mixed insulin only ‐ 17% Short‐acting insulin only ‐ 10%	—	—
Khanna 2014	M	Diabetes mellitus	USA	75	52	59	41	Hispanic ‐ 100%	—	—	Yes	USD 10 for initial visit and USD 20 incentive card at the follow‐up
Kim 2014	M	Diabetes mellitus	USA	100	—	—	—	—	—	Psychiatric illness ‐ 46% Had been hospitalised over the past year ‐ 28%	—	—
Lorig 2008	M	Diabetes mellitus	USA	417	53	38	62	Hispanic ‐ 100%	—	—	—	—
Piette 2000	M	Diabetes mellitus	USA	248	55	41	59	Hispanic ‐ 50% White ‐ 29% Other – 21%	—	BMI ‐ 33.7 kg/m² Mean comorbidities: 1	—	—
Piette 2001	M	Diabetes mellitus	USA	272	61	97	3	White ‐ 60% Black ‐ 18% Hispanic ‐ 12% Other ‐ 10%	—	BMI ‐ 31 kg/m² Mean comorbidities: 2	—	—
Schillinger 2009	M	Diabetes mellitus	USA	339	56	43	57	Hispanic ‐ 47% Asian ‐ 22% Black ‐ 20% White ‐ 8% Other ‐ 2%	—	BMI: 31 kg/m²	Yes	USD 15 and USD 25 for the baseline and 1‐year follow‐up visits, respectively
Williams 2012	M	Diabetes mellitus	Australia	120	57	63	37	—	—	Low depression – 73% Intermediate depression – 23% High depression – 4% Low anxiety – 89% Intermediate anxiety – 8% High anxiety – 3% BMI ‐ 33 kg/m² Insulin – 43%	—	—
Capomolla 2004	M	Heart failure	Italy	133	57	88	12	—	—	Diuretics ‐ 89% ACE inhibitors ‐ 84% Carvedilol ‐ 50% Nitrates ‐ 40% Digitalis ‐ 33% K + saver ‐ 21%	—	—
Chaudhry 2010	M	Heart failure	USA	1653	61^d	58	42	White ‐ 49% Black ‐ 39% Other – 12% (inclusive of Hispanic or Latino – 3%)	—	Beta‐blocker – 79% Loop diuretic – 78% Hypertension ‐ 77% ACE inhibitor or ARB – 70% Coronary artery disease ‐ 50% Diabetes mellitus ‐ 47% Chronic kidney disease ‐ 46% Aldosterone‐receptor antagonist – 33% Digoxin – 25% COPD ‐ 21%	No	—
Krum 2013	M	Heart failure	Australia	405	73	63	37	—	—	Diuretics ‐ 80% Heart failure specific beta‐blockers ‐ 61% Systolic heart failure ‐ 60%  Hypertension ‐ 60% ACE inhibitors ‐ 57.5%  Atrial fibrillation ‐ 37.5% Diabetes mellitus ‐ 30.5% Aldosterone antagonist ‐ 26% ARB ‐ 25% Diastolic heart  failure ‐ 18.5%  Pacemaker ‐ 12% Internal cardio defibrillator ‐ 4.5%	—	—
Kurtz 2011	M	Heart failure	France	138	68	79	21	—	—	Loop diuretic – 92% ACE/AT2– 79% Beta‐blocker – 79% Spironolactone – 29% Digoxin – 9% (mean values)	—	—
Shet 2014	M	HIV	India	631	—	57	43	Asian ‐100%	—	Zidovudine‐based antiretroviral treatment – 44% Tenofovir‐based antiretroviral treatment – 44% Stavudine‐based antiretroviral treatment – 12%	Yes	Mobile phone plus wage compensation
Hyman 1996	M	Hypercholestorolaemia	USA	115	48	25	75	White ‐ 87% Other – 13%	—	—	—	—
Hyman 1998	M	Hypercholesterolaemia	USA	123	57	25	75	Black ‐ 77% Other – 23%	—	Mean BMI ‐ 31 kg/m²	—	—
Bove 2013	M	Hypertension	USA	241	60	21	79	Black ‐ 81% White‐ 15% Hispanic ‐ 3% Other ‐ 1%	—	Hyperlipidaemia ‐ 46% Diabetes ‐ 32%	—	—
Dedier 2014	M	Hypertension	USA	253	58	27	73	Black ‐ 100%	—	—	—	—
Harrison 2013	M	Hypertension	USA	64,773	61	46	54	White ‐ 41%, Hispanic ‐ 25% Black ‐ 17% Other – 9% Asian ‐ 8%	—	Cardiovascular disease ‐ 38% Diabetes mellitus ‐ 27% Chronic kidney disease ‐ 10%	—	—
Magid 2011	M	Hypertension	USA	283	66	65	35	White ‐ 65% Other – 18% Hispanic ‐ 17%	—	Diabetes mellitus or chronic kidney disease – 55%	Yes	Clinically validated electronic blood pressure cuffs were provided at no cost to those who did not own one
Piette 2012	M	Hypertension	Honduras; Mexico	200	58	33	67	—	—	Blood pressure medication – 83% Diabetes mellitus ‐ 23% BMI ‐ 30.7 kg/m²	—	—
Farzanfar 2011	M	Mental health	USA	164	39	24	76	White ‐ 56% Black ‐ 32% Other – 12%	—	—	—	—
Greist 2002	M	Mental health	USA	218	39	58	42	White ‐ 93% Other – 7%	—	Social phobia ‐ 9%, Generalised anxiety disorder ‐8% Simple phobia – 6% Major depression – 2% Dysthymia – 2%	—	—
Zautra 2012	M	Mental health	USA	73	54	18	82	Other – 74% Hispanic – 26%	—	—	—	—
DeMolles 2004	M	OSAS	USA	30	46	—	—	—	—	BMI ‐ 38 kg/m²	—	—
Sparrow 2010	M	OSAS	USA	250	55^d	82	18	—	—	BMI ‐ 35.1 kg/m²	—	—
Brendryen 2008	M	Smoking	Norway	396	36	47	53	—	—	—	Yes	All participants in both groups received a sample packet of nicotine replacement therapy products.
Carlini 2012	M	Smoking	USA	521	36	36	64	White ‐ 81% Black ‐ 6% Other – 5% Hispanic/Latino ‐ 4% Native American – 3% Asian ‐ 1%	—	≥ 1 long‐term conditions ‐ 47%	No	—
Ershoff 1999	M	Smoking	USA	332	30	0	100	White ‐ 61% Black ‐ 16% Hispanic ‐ 15% Other – 8%	—	—	—	—
McNaughton 2013	M	Smoking	Canada	44	53	67	33	—	Mean cigarettes/d: intervention ‐ 18.5 control ‐ 17.3	—	—	—
Peng 2013	M	Smoking	Taiwan	116	20	92	8	Asian ‐ 100%	—	—	Yes	Equivalent of USD 6
Regan 2011	M	Smoking	USA	731	52	56	44	—	—	—	—	—
Reid 2007	M	Smoking	Canada	100	54	68	32	—	—	Acute coronary syndrome ‐ 83%	—	—
Reid 2011	M	Smoking	Canada	440	—	—	—	—	—	—	—	—
Rigotti 2014	M	Smoking	USA	397	53	48	52	White ‐ 81% Hispanic ‐ 6% Black ‐ 4% Other/unknown ‐ 4% Native American ‐ 3% Asian/Pacific Islander ‐2.5%	Mean cigarettes/d: IG = 17.1 CG = 16.3	Depressive symptoms (mean) IG = 9.3 CG = 10.3	Yes	USD 50 for a saliva sample
Velicer 2006	M	Smoking	USA	2054	51	77	23	White ‐ 89%, Black ‐ 5%, Other ‐ 4% Native American ‐ 2%	Mean cigarettes/d IG = 23.85 CG = 25.18	—	—	—
Houlihan 2013	M	Spinal cord dysfunction	USA	142	48	61	39	White ‐ 80% (inclusive of Hispanic or Latino – 7%) Black ‐ 11% Other ‐ 9%	11.7 y	Depression – 39% Pressure ulcers ‐ 7%	—	—

Study ID	ATCS^a	Content^b	Theory^c	BCTs^d	Received instructions?	Caller^e	Telephone keypad^f	Toll free	Study duration	Call duration^g	Frequency^h	Intensity ⁱ	Speaker   features	Security arrangement^j
Adams 2014	IVR	AF	—	1	—	P	Other	—	25 months	29.3 min	—	—	Synthetic speech	—
Aharonovich 2012	ATCS Plus	AF,SF	MI	9, 25, 40	Yes	E	Yes	Yes	2 months	1‐3 min	Daily	—	—	—
Andersson 2012	IVR	AF	—	25	—	—	—	—	1.5 months	—	—	< 500 words	—	—
Baker 2014	Unidirectional¹	AF	TCD	1, 42	No	H	—	—	6 months	—	2	92 words	—	—
Bender 2010	IVR	AF	HBM	20, 21, 42	—	H	Yes	Yes	2.5 months	< 5 min	2	—	—	—
Bender 2014	IVR	AF	—	20,21	—	—	—	—	24 months	—	—	—	—	—
Bennett 2012	ATCS Plus	AF, CF	SCT, TRA, TTM	3, 9, 12, 13, 17, 20, 21, 25, 33, 39, 40, 42	—	—	—	—	24 months	—	—	—	—	—
Bennett 2013	ATCS Plus¹	AF, CF	SCT, TRA, TTM, MI	3, 9, 12, 13, 17, 20, 21, 25, 33, 39, 40, 42	—	H	—	—	12 months	2–4 min	Weekly	—	—	—
Boland 2014	IVR	AF	—	29,42	—	H	Yes	—	3 months	—	Daily	51 words	—	Reminder information was sent securely to Memotext¹
Bove 2013	ATCS Plus¹	AF, CF, SF	—	9,10,42	Yes	P	Yes¹	Yes	6 months	—	Biweekly	—	—	Password and log‐in
Brendryen 2008	ATCS Plus¹	AF, SF	CBT, SCT, MI, SRT, SCL, RP	3,12, 20,27, 34, 39	Yes	P	Yes	—	24 months	—	Twice daily then biweekly¹ for 6 weeks	—	Personal pronouns and active voice	—
Capomolla 2004	ATCS Plus	AF, CF	—	20, 42	Yes	P	Yes	Yes	12 months	—	Daily	—	—	PIN²
Carlini 2012	ATCS Plus	AF, CF	—	3	—	H	Yes	Yes	4 months	—	—	—	—	—
Chaudhry 2010	ATCS Plus	AF, CF	—	20, 42	Yes	P	Yes	Yes	6 months	—	Daily	—	—	—
Cleeland 2011	ATCS Plus	AF, SF	—	21	Yes	H	Yes	—	1 month	4 weeks	Biweekly	—	—	—
Cohen‐Cline 2014	IVR	AF	—	1,42	—	H	—	—	12 months	5 min¹	1	—	—	—
Corkrey 2005	ATCS Plus	AF, CF	—	3, 27, 39	Yes	H	Yes	—	6 months	—	—	—	—	—
Cvietusa 2012	ATCS Plus	AF, CF	—	29,42	—	H	Other	—	12 months	—	≤ 3	—	—	—
David 2012	ATCS Plus	AF, CF	TTM, SCT, PST	3,10	Yes	E	—	—	3 months	15‐30 s²	Twice daily	—	—	—
Dedier 2014	IVR	AF	TTM, SCT	1,9	—	H	—	—	3 months	10 min	Weekly	—	Pre‐recorded human speech	—
DeFrank 2009	IVR	AF	HBM¹	27, 42	—	H	Yes	—	24 months	69 s	Average ‐ 3	224 words	Female voice	Verification³
DeMolles 2004	IVR	AF	—	3, 18, 20, 40, 42	Yes	P	Yes	—	2 months	—	3‐day call², weekly	—	—	Password protected⁴
Derose 2009	ATCS Plus	AF, SF	—	40, 42	—	H	Yes	Yes	6 months	40 s	1	100 words	—	PIN⁵
Derose 2013	ATCS Plus	AF, SF	—	40, 42	—	H	—	Yes	3 months	40 s	1	—	—	—
Dini 1995	Unidirectional	AF	—	40	No	H	—	—	1 months	—	1	—	—	—
Dini 2000	Unidirectional	AF	—	42	No	H	—	—	36 months	—	1³	—	—	—
Dubbert 2002	Unidirectional¹	AF	TTM	9,10,40	—	H	—	—	10 months	—	—	Approx 30 words	Nurse	—
Durant 2014	IVR	AF	—	42	—	H	—	—	2 weeks	—	1‐2 attempts	—	—	—
Ershoff 1999	IVR	AF	TTM	20, 39	Yes	P	Yes	Yes	34 weeks	5 min	—	—	Professional	Password protected⁴, PIN²
Estabrooks 2008	IVR	AF	—	3, 9	Yes	—	—	—	3 months	1‐10 min³	Weekly	—	—	—
Estabrooks 2009	ATCS Plus	AF, CF	GM	2, 3, 9, 15, 20	—	E	—	—	12 months	—	10	—	—	—
Farzanfar 2011	IVR	AF	—	3, 6, 20, 30, 39, 42	Yes	E	Yes	Yes	6 months	30‐90 min²	Monthly	—	Female voice actor	Password protected⁴
Feldstein 2006	IVR	AF	—	39, 42	—	—	—	—	25 days	—	—	—	—	PIN⁶
Fiscella 2011	ATCS Plus¹	AF, CF	—	20,42	—	H	—	—	26 weeks	—	Up to 4	—	—	—
Fortuna 2014	Unidirectional	AF	—	29,30	—	H	—	—	12 months	—	2	—	—	—
Franzini 2000	Unidirectional	AF	—	40,42	No	H	—	—	—	—	—	—	—	—
Friedman 1996	IVR	AF	SCT	21, 32, 35, 39	—	E	Yes	Yes	6 months	4 min	Weekly	—	—	Password protected⁴
Glanz 2012	IVR	AF	—	3,16	—	E	Other	—	9 months	—	12	—	—	—
Goulis 2004	IVR	AF	—	1, 10, 20, 21,	Yes	—	—	—	6 months	15 min	Weekly	—	—	—
Graziano 2009	IVR	AF	HBM	3, 7, 20, 21, 42	—	H	—	—	12 months	> 1 min	Daily	—	Trained actor	PIN⁷
Green 2011	IVR	AF	—	16	—	H	Yes	—	—	—	—	—	—	—
Greist 2002	ATCS Plus	AF, SF	BT	34	—	E	Yes	—	3 months	8.6 min¹	12	—	—	—
Griffin 2011	ATCS Plus	AF, CF	HBM, SMP	3, 7, 27, 30, 40	—	H	—	—	6W	—	1	—	—	—
Halpin 2009	ATCS Plus	AF, CF, SF	—	21, 40, 42	Yes	H	—	—	4 months	—	Daily, 4⁴	—	—	—
Harrison 2013	Unidirectional	AF	—	20,42	—	H	—	—	1 month	—	—	80 words	—	—
Hasin 2013	ATCS Plus	AF, CF, SF	MI	9,12	Yes	P	—	Yes	12 months	60 days	Daily	1‐3 min	—	—
Helzer 2008	ATCS Plus	AF,CF	—	20,34	Yes	P	Yes	Yes	6 months	2 min	Daily	—	—	—
Hendren 2014	ATCS Plus¹	AF, CF	—	20,42	—	H	—	—	25 weeks	25 s	Up to 4	—	—	—
Hess 2013	Unidirectional	AF	—	16	—	H	—	—	3 months	1 min	Monthly	2 30‐s scripts	—	—
Heyworth 2014	ATCS Plus	AF, CF	—	16	—	H	—	—	3 months	4‐5 min	1	—	—	—
Ho 2014	ATCS Plus¹	AF, CF	—	20,42	—	H	—	—	12 months	—	Monthly⁵	—	—	—
Homko 2012	IVR	AF	—	21	Yes	P	Yes	Yes	26 months	—	Weekly	45 s of speaking	—	PIN²
Houlihan 2013	IVR	AF	TTM, SCT	4,20,21	Yes	H	—	—	6 months	4.12 min	Weekly	12.4 calls¹	Digitised  speech	—
Hyman 1996	IVR	AF	—	20, 39, 42	—	E	Yes	—	6 months	—	Daily	—	—	—
Hyman 1998	ATCS Plus	AF, CF	SCT	20, 39, 42	—	E	Yes	—	6 months	2‐3 min	Biweekly	—	—	—
Jarvis 1997	IVR	AF	TTM	7, 9, 20, 39	Yes	P	Yes	—	3 months	—	Weekly	—	—	Password protected⁴
Katalenich 2015	ATCS Plus	AF,CF, SF	—	16,21,30	—	E	—	—	6 months	—	Daily	—	—	—
Khanna 2014	ATCS Plus	AF, SF	—	20	Yes	H	Yes	—	3 months	—	2.2/week	26 calls¹	Female voice	—
Kim 2014	ATCS Plus	AF, CF	—	20,29,30	—	H	—	—	12 months	≤ 10 min	Weekly	—	—	—
King 2007	IVR	AF	SCT, TTM	3, 9, 20, 39	Yes	H	Yes	—	12 months	10‐15 min	Weekly⁶	—	—	—
Kroenke 2010	ATCS Plus¹	AF, CF	TCM	39	—	E	—	—	12 months	—	Biweekly, weekly, bimonthly, monthly⁷	—	—	—
Kroenke 2014	ATCS Plus¹	AF, CF	—	21,39	—	H	—	—	12 months	—	Weekly⁸	—	—	—
Krum 2013	ATCS Plus	AF, CF, SF	—	20,21,42	—	E	Yes	—	12 months	—	Monthly	18 questions	—	—
Kurtz 2011	IVR	AF	—	39	No	P	Yes	—	24 months	48 s¹	Weekly	—	—	—
LeBaron 2004	ATCS Plus	AF, SF	—	40, 42	—	H	—	—	24 months	—	—	—	—	—
Leirer 1991	IVR	AF	—	20,42	Yes	H	Yes	—	2 weeks	—	—	3 segments	Personalised voice messages	—
Lieu 1998	IVR	AF	—	42	—	H	—	—	4 months	96.8 s	—	—	—	—
Lim 2013	Unidirectional	AF	—	42	—	H	—	—	5 months	—	Monthly	—	—	—
Linkins 1994	Unidirectional	AF	—	42	—	H	—	—	5 months	—	2/d⁹	—	—	—
Litt 2009	ATCS Plus	AF,SF	CBT	3,20,34,38	—	H	Yes	—	12 weeks	2.5 min	8/d¹⁰	—	—	—
Lorig 2008	ATCS Plus	AF, CF	—	21	—	—	—	—	15 months	90 s	Monthly	—	—	—
Magid 2011	ATCS Plus¹	AF, CF	—	20, 21, 25, 42	Yes	P	Yes	—	6 months	5‐10 min	Weekly	—	—	—
Mahoney 2003	ATCS Plus	AF, SF	PT, PM	3, 13, 25, 39, 42	Yes	P	Yes	Yes	18 months	18 min	—	22 h/d²	Professional radio announcer	Password protected⁴
Maxwell 2001	Unidirectional	AF	—	40	No	H	—	—	2 months	—	—	—	—	—
McNaughton 2013	IVR	AF	—	20, 21, 25, 34, 42	—	H	Other	—	24 months	3‐5 min	Biweekly	—	—	—
Migneault 2012	IVR	AF	SCT, TTM, MI	25, 39	Yes	—	Yes	Yes	8 months	—	Weekly	—	African – American voice professionals	—
Mooney 2014	ATCS Plus	AF, SF, CF	—	21	—	P	Yes	Yes	1.5 months	5.18 min¹	Daily	—	—	Personalised password
Moore 2013	ATCS Plus	AF, SF	CBT	17,34,38	Yes	H	Yes	—	1 month	9.3 min¹	Daily	—	—	—
Morey 2009	Unidirectional¹	AF	SCT, TTM	9, 10, 20, 25, 39	—	H	—	—	12 months	—	Monthly	Approx 60 words	Primary care provider	—
Morey 2012	Unidirectional¹	AF	SCT, TTM	9, 10, 20, 25, 39	—	H	—	—	12 months	—	Monthly	Approx 60 words	Primary care provider	—
Mosen 2010	IVR	AF	—	27, 42	—	H	Yes	Yes	6 months	1 min	3	—	—	—
Mu 2013	IVR	AF	—	25, 39	—	H	—	—	1 month	—	—	—	—	Correspondence with the author: "Upon answering a call, patients are required to authenticate with their date of birth."
Mundt 2006	ATCS Plus	AF, CF, SF	CBT	4,20,34,38	Yes	E	—	Yes	6 months	9.2 min	Daily	—	—	Valid ID number and a personally  selected 4‐digit pass code
Nassar 2014	Unidirectional	AF	—	29	Yes	H	—	—	2 months	—	At least every 3 days	Every hour for 2 consecutive hours	Community health worker	—
Naylor 2008	IVR	AF	CBT	3, 16, 20, 22, 25, 39, 42	Yes	P	—	Yes	4 months	3‐16 min	Daily	—	Experienced therapist	—
Ownby 2012	Unidirectional	AF	—	42	—	H	—	—	24 months	—	Daily	—	First author	—
Parikh 2010	IVR	AF	—	40	—	H	—	—	4 months	—	—	—	—	—
Patel 2007	IVR	AF	—	39, 42	—	—	—	—	6 months	—	3	—	—	—
Peng 2013	ATCS Plus	AF, SF	CBT, TTM, MI	9, 0, 12, 20, 21	—	H	Yes	—	2 months	18.9 min^^	Biweekly weeks 1‐3, weekly weeks 4‐6, no call week 7, weekly week 8‐9	8.61²,³	—	—
Phillips 2015	IVR	AF	—	29, 30	—	H	—	—	3.5 months	—	up to 5 times	—	—	—
Piette 2000	ATCS Plus	AF, CF	SCT	3, 18, 21, 25, 39	No	E	Yes	Yes	24 months	1‐8 min²	Biweekly¹¹	—	Human voice	PIN²
Piette 2001	ATCS Plus	AF, CF	SCT	3, 18, 21, 25, 39	—	H	Yes	—	12 months	1‐8 min²	Biweekly¹¹	—	—	—
Piette 2012	ATCS Plus	AF, CF	—	13, 20, 21, 39	Yes	—	Yes	Yes	1.5 months	≤ 9 min	Weekly	—	Native speaker	—
Pinto 2002	ATCS Plus	AF, SF	DMT, SCT, TTM	3, 19, 20, 32	Yes	P	Yes	—	6 months	10 min	Weekly (first 3 months), and at least biweekly thereafter	—	Digitised  human speech	—
Reekie 1998	Unidirectional	AF	—	39, 40	No	H	—	—	—	—	—	—	Receptionist	—
Regan 2011	ATCS Plus	AF, CF	—	20, 42	—	H	—	—	3 months	—	—	—	—	—
Reid 2007	ATCS Plus	AF, CF	—	3, 9, 13, 39, 40, 42	—	H	—	—	12 months	1‐20 min⁵	3	—	—	—
Reid 2011	ATCS Plus	AF, CF	—	20, 42	—	—	—	—	12 months	—	8	—	—	—
Reynolds 2011	IVR	AF	—	42	—	P	—	—	—	—	—	—	—	—
Rigotti 2014	ATCS Plus	AF, CF, SF	—	13, 34	—	H	—	Yes	6 months	—	5 times¹²	—	—	—
Rose 2015	ATCS Plus	AF, CF, SF	CBT	4, 20, 34, 38	Yes	P	—	—	4 months	—	Daily	—	—	—
Rubin 2012	IVR	AF	MI	9, 12	—	H	Other	—	6 months	—	≤ 26 calls over 13 weeks	—	—	—
Schillinger 2009	ATCS Plus	AF, CF	CCM, SCT	1, 3, 9, 10, 13, 20, 21, 35, 39	—	E	Yes	Yes	12 months	6‐12 min	Weekly	—	—	—
Sherrard 2009	ATCS Plus	AF, CF	—	27, 39, 42	—	H	—	—	6 months	—	11	—	—	Password protected⁴
Shet 2014	IVR¹	AF	TPB	16,20	—	H	—	—	24 months	—	Weekly	—	—	—
Siegel 1992	IVR	AF	—	17, 20, 42	—	H	Other	—	6 weeks	—	3 calls 6 weeks apart	12 questions; 397 words	Digitally  stored voice	—
Sikorskii 2007	IVR	AF	—	17, 20, 30, 39, 40	—	H	Yes	—	2 months	—	Weekly¹³	—	Female voice	—
Simon 2010a	IVR	AF	GMDBC, Others²	3, 27, 39	Yes	E	—	—	3 months	2‐6 min	—	—	—	Verification
Simon 2010b	ATCS Plus	AF, CF	—	30, 42	No	H	No	—	12 months	—	3¹⁴	—	Human voice	—
Simpson 2005	IVR	AF	—	20	Yes	P_H	Yes	Yes	1 months	—	Daily	—	—	—
Solomon 2007	ATCS Plus¹	AF, CF	—	16	—	H	—	Yes	—	—	—	—	Female voice	—
Sparrow 2010	IVR	AF	SCT, MI	3, 12, 17, 20, 25, 39, 40, 42	—	E	Yes	—	12 months	—	Weekly then monthly¹⁵	—	—	Password protected⁴
Sparrow 2011	IVR	AF	SCT	3, 12, 17, 20, 25, 39, 40, 42	Yes	E	Yes	—	12 months	—	Weekly then monthly¹⁵	—	—	Password protected⁴
Spoelstra 2013	ATCS Plus	AF, CF	CBT	2, 20	Yes	H	Yes	—	2 months	—	Weekly	—	—	—
Stacy 2009	ATCS Plus	AF, SF	HBM, CCM, SCT, TTM, MI, SRT, RL	2, 3, 20, 32, 39	—	H	—	Yes	6 months	—	3	—	—	—
Stehr‐Green 1993	Unidirectional	AF	—	42	—	H	—	—	1 month	—	1³	—	Human voice	—
Stuart 2003	IVR¹	AF	—	9, 10, 42	Yes	P	Yes	—	3 months	—	Daily/2 weeks; weekly/10 weeks	25 calls	Female voice	—
Szilagyi 2006	Unidirectional	AF	—	42	—	H	—	—	18 months	—	Weekly	—	—	—
Szilagyi 2013	Unidirectional	AF	—	42	—	H	—	—	12 months	—	—	—	—	—
Tanke 1994	Unidirectional	AF	HBM	7*, 39, 40, 42	No	H	—	—	—	—	1¹⁶	—	Female using native languages	—
Tanke 1997	Unidirectional	AF	HBM	39, 40, 42	No	H	—	—	—	—	1¹⁶	—	Female using native languages	—
Tucker 2012	IVR	AF	BET	9, 20, 25, 34	Yes	—	—	—	6 months	≤ 5 min	Daily	—	—	PIN⁶
Vance 2011	ATCS Plus	AF,SF	—	10	—	—	—	—	3 months	—	3/week	—	—	—
Velicer 2006	IVR¹	AF	TTM	4, 13, 20, 34, 38	—	E	Yes	—	6 months	15‐20 min	Weekly¹⁷	—	—	—
Vollmer 2006	ATCS Plus	AF, CF	—	39	—	H	Other	—	10 months	< 10 min	3¹⁸	—	—	—
Vollmer 2011	ATCS Plus	AF, CF	—	20, 42	—	H	Other	Yes	18 months	2‐3 min	—	—	—	—
Vollmer 2014	ATCS Plus	AF, CF, SF	—	20, 42	Yes	H	Other	Yes	12 months	2‐3 min	Monthly	—	—	—
Williams 2012	IVR	AF	—	3, 21, 39	Yes	P	Yes	Yes	6 months	5‐20 min	Weekly	—	—	PIN⁶
Wright 2013	IVR	AF	SCT	2, 9, 10	No	P	Other	—	3 months	—	Biweekly	—	Synthetic speech	—
Xu 2010	IVR	AF	—	21, 40	—	H	Yes	—	6 months	—	Biweekly	—	—	—
Yount 2014	ATCS Plus	AF, SF	—	21, 40	—	P	Yes	—	3 months	—	Weekly	—	—	PIN
Zautra 2012	Unidirectional	AF	SCT	4,13	—	—	—	—	—	—	—	—	—	—

Study ID	Type^a	Subtype	Participant age (years)	Sex	Ethnicity^b	Primary outcome measures	Effect^c
Tucker 2012	P	Alcohol misuse	41‐70	> 50% M	W	Drinking practices Spending on alcohol	2 2 (median effect = 2)
Franzini 2000	P	Immunisation	—	—	—	Immunisation status Cost‐effectiveness	1 1 (median effect = 1)
Hess 2013	P	Immunisation	≥ 71	—	—	Herpes zoster immunisations	1
Dini 2000	P	Immunisation	0‐21	—	—	Immunisation status	2
LeBaron 2004	P	Immunisation	0‐21	M = F (± 3%)	W,B,H	Completion by the age of 24 months of the 4‐3‐1‐3 immunisations series	2
Lieu 1998	P	Immunisation	0‐21	—	—	Immunisation status	2
Linkins 1994	P	Immunisation	—	M = F (± 3%)	W,B	Immunisation status	1
Nassar 2014	P	Immunisation	22‐40	> 50% F	W,B	Immunisation rate	2
Stehr‐Green 1993	P	Immunisation	0‐21	M = F (± 3%)	B,H	Immunisation status	2
Szilagyi 2006	P	Immunisation	0‐21	M = F (± 3%)	W,B,H	Immunisation status	2
Szilagyi 2013	P	Immunisation	0‐21	M = F (± 3%)	—	Immunisation status Preventive visit rate	1 1
David 2012	P	Physical activity	41‐70	> 50% F	W	1‐mile walk after the intervention	5
Dubbert 2002	P	Physical activity	41‐70	> 50% M	—	Self‐reported (diary) walking adherence	1
Jarvis 1997	P	Physical activity	41‐70	> 50% F	B	Minutes walked per week	1
King 2007	P	Physical activity	41‐70	> 50% F	W	Minutes of moderate to vigorous physical activity	1
Morey 2009	P	Physical activity	≥ 71	> 50% M	W,B	Gait speed (usual and rapid) Self‐reported physical activity Function and disability	2, 1 1 2 (median effect = 2)
Morey 2012	P	Physical activity	41‐70	> 50% M	W	Homeostasis model assessment of insulin resistance	2
Pinto 2002	P	Physical activity	41‐70	> 50% F	W,B	Energy expenditure in moderate‐intensity‐physical activity % meeting recommendations for moderate‐intensity‐physical activity Motivational readiness for physical activity	2 2 2 (median effect = 2)
Sparrow 2011	P	Physical activity	≥ 71	> 50% M	—	Muscle strength Balance Walk distance Mood	1 1 2 1 (median effect = 1)
Baker 2014	P	Screening	41‐70	> 50% F	H	Colorectal cancer screening adherence (faecal occult blood testing)	1
Cohen‐Cline 2014	P	Screening	41‐70	> 50% M	W	Receipt of any recommended colorectal cancer screening	1
Corkrey 2005	P	Screening	—	> 50% F	—	Screening rate	2
DeFrank 2009	P	Screening	—	> 50% F	W,B,A	Mammography adherence	1
Durant 2014	P	Screening	41‐70	M = F (± 3%)	—	Receipt of colorectal cancer screening after 3 months	—
Fiscella 2011	P	Screening	—	> 50% F	W,B,H,A	Chart documentation of breast cancer screening, colorectal cancer screening, or both	1
Fortuna 2014	P	Screening	—	> 50% F	W,B	Breast cancer and colorectal cancer screening	2
Hendren 2014	P	Screening	—	> 50% F	W,B,H	Documentation of breast cancer screening, colorectal cancer screening, or both	1
Heyworth 2014	P	Screening	41‐70	> 50% F	—	Bone mineral density screening after 12 months	1
Mosen 2010	P	Screening	41‐70	M = F (± 3%)	W	Completion of faecal occult blood testing at 6 months	1
Phillips 2015	P	Screening	41‐70	> 50% F	W,B	Completed mammogram or colorectal cancer screening within 36 weeks of randomisation	2
Simon 2010a	P	Screening	41‐70	M = F (± 3%)	W	Colorectal cancer screening	2
Solomon 2007	P	Screening	41‐70	> 50% F	—	Bone mineral density testing and/or filling a prescription for a bone active medication	1
Mahoney 2003	P	Stress management among caregivers	41‐70	> 50% F	W,B	Caregiver's appraisal of the bothersome nature of caregiving Anxiety Depression	2 2 2 (median effect = 2)
Aharonovich 2012	P	Substance use	41‐70	> 50% M	W,H	Days using primary drug	2
Bennett 2012	P	Weight management	41‐70	> 50% F	B,H	Change in body weight and BMI	1
Bennett 2013	P	Weight management	22‐40	> 50% F	B	Change in body weight and BMI	1
Estabrooks 2008	P	Weight management	41‐70	> 50% F	W,B,H,A	Physical activity Dietary habits Weight loss	2 2 2 (median effect = 2)
Estabrooks 2009	P	Weight management	0‐21	> 50% M	W,H	BMI z‐score Physical activity and sedentary behaviour Dietary habits Eating disorder symptoms	2 2 2 2 (median effect = 2)
Goulis 2004	P	Weight management	41‐70	> 50% F	—	Body weight BMI Systolic blood pressure Diastolic blood pressure Plasma glucose Serum triglycerides Serum serum high‐density lipoprotein‐cholesterol Total serum cholesterol SF‐36 EQ‐5D Obesity Assessment Survey	1 2 2 2 2 1 1 2 2 2 2 (median effect = 2)
Vance 2011	P	Weight management	—	—	—	Weight change	2
Wright 2013	P	Weight management	0‐21	> 50% M	W,B	BMI Calorie intake Fat intake Fruit intake Vegetable intake Television‐viewing time	2 2 2 2 2 2 (median effect = 2)
Dini 1995	E	Appointment reminder	—	—	—	Appointment adherence	1
Griffin 2011	E	Appointment reminder	41‐70	> 50% M	W	Appointment non‐attendance Preparation non‐adherence	2 2 (median effect = 2)
Maxwell 2001	E	Appointment reminder	22‐40	> 50% F	W,B,H	Attendance rate	2
Parikh 2010	E	Appointment reminder	—	M = F (± 3%)	—	Appointment adherence	1
Reekie 1998	E	Appointment reminder	—	—	—	Appointment adherence	2
Tanke 1994	E	Appointment reminder	0‐21	> 50% M	H	Appointment adherence	1
Tanke 1997	E	Appointment reminder	0‐21	> 50% F	W,H	Appointment adherence (3‐day interval)	1
Moore 2013	M	Illicit drugs addiction	41‐70	> 50% M	W,B	Patient interest Perceived efficacy Ease of use Treatment satisfaction Retention rate Drug consumption Methadone counselling Coping	5 5 5 4 4 2 2 2 (median effect = 4)
Andersson 2012	M	Alcohol consumption	—	—	—	AUDIT score	1
Hasin 2013	M	Alcohol consumption	41‐70	> 50% M	B,H	Number of drinks per drinking day	1
Helzer 2008	M	Alcohol consumption	41‐70	> 50% M	W	Weekly alcohol consumption	3
Litt 2009	M	Alcohol consumption	41‐70	> 50% M	W,B,H	Proportion of days abstinent Proportion of heavy drinking days Continuous abstinence Drinking problems Coping problems	1 2 2 2 2 (median effect = 2)
Mundt 2006	M	Alcohol consumption	41‐70	> 50% M	W,B	Self‐reported drinking patterns Blood alcohol content Work and social adjustment scale Obsessive–compulsive drinking scale SF‐36 health survey Drinker inventory of consequences	2 2 2 2 2 2 (median effect = 2)
Rose 2015	M	Alcohol consumption	41‐70	M = F (± 3%)	—	Alcohol consumption	2
Rubin 2012	M	Alcohol consumption	41‐70	> 50% M	W,B	Number of heavy drinking days per month % days abstinent per month Drinks per drinking day	2 2 2 (median effect = 2)
Simpson 2005	M	Alcohol consumption	41‐70	> 50% M	W,B	Drinking habits Alcohol craving Post‐traumatic stress disorder symptoms	2 2 2 (median effect = 2)
Vollmer 2006	M	Asthma	41‐70	> 50% F	W	Healthcare utilisation Medication use QoL	2 2 2 (median effect = 2)
Xu 2010	M	Asthma	0‐21	M = F (± 3%)	—	Healthcare utilisation	4
Cleeland 2011	M	Cancer	41‐70	> 50% M	W	Number of symptom threshold events Cumulative distribution of symptom threshold events Differences in mean symptom severity between discharge and follow‐up	1 2 2 (median effect = 2)
Kroenke 2010	M	Cancer	41‐70	> 50% F	W,B	Depression severity Pain severity	1 1 (median effect = 1)
Mooney 2014	M	Cancer	41‐70	> 50% F	W	Symptom presence, severity, and distress data	2
Siegel 1992	M	Cancer	41‐70	M = F (± 3%)	W,B,H	Prevalence of unmet needs	2
Sikorskii 2007	M	Cancer	41‐70	> 50% F	—	Symptom severity	2
Spoelstra 2013	M	Cancer	41‐70	> 50% F	W,B,A	Adherence to medications Symptom severity	2 1 (median effect = 2)
Yount 2014	M	Cancer	41‐70	M = F (± 3%)	W,B,H	Symptom burden	2
Naylor 2008	M	Chronic Pain	41‐70	> 50% F	W	Pain Function/disability Coping	1 1 1 (median effect = 1)
Kroenke 2014	M	Chronic Pain	41‐70	> 50% M	W	Pain intensity	1
Halpin 2009	M	Chronic obstructive pulmonary disease	41‐70	> 50% M	—	Frequency of exacerbations Proportion of patients experiencing 1 or more exacerbations	2 2 (median effect = 2)
Adams 2014	M	Adherence	0‐21	> 50% M	B	Comprehensiveness of screening and counselling	1
Bender 2010	M	Adherence	41‐70	> 50% F	W,B,H,A	Medication adherence	1
Bender 2014	M	Adherence	0‐21	—	—	Medication adherence	1
Boland 2014	M	Adherence	41‐70	M = F (± 3%)	W,B,H,A	Medication adherence	1
Cvietusa 2012	M	Adherence	0‐21	—	—	Medication adherence	1
Derose 2009	M	Adherence	41‐70	> 50% M	W,B,H,A	Adherence (completion of all 3 recommended laboratory tests)	2
Derose 2013	M	Adherence	41‐70	M = F (± 3%)	W,B,H,A	Medication adherence	1
Feldstein 2006	M	Adherence	41‐70	M = F (± 3%)	—	Completion of all recommended laboratory tests	1
Friedman 1996	M	Adherence	≥ 71	> 50% F	B	Medication adherence Systolic blood pressure Diastolic blood pressure	1 2 1 (median effect = 1)
Glanz 2012	M	Adherence	41‐70	> 50% M	W,B	Medication adherence Refill adherence Appointment adherence	2 2 2 (median effect = 2)
Green 2011	M	Adherence	—	—	—	Medication refill rate	1
Ho 2014	M	Adherence	41‐70	> 50% M	W	Medication adherence	1
Leirer 1991	M	Adherence	≥ 71	> 50% F	—	Medication non‐adherence Cognitive assessment	1 2 (median effect = 2)
Lim 2013	M	Adherence	41‐70	M = F (± 3%)	W,B,H,A	Adherence rate Therapeutic coverage	2 2 (median effect = 2)
Migneault 2012	M	Adherence	41‐70	> 50% F	B	Medication adherence Diet Moderate or greater intensity physical activity	2 1 4 (median effect = 2)
Mu 2013	M	Adherence	—	—	—	Medication adherence	1
Ownby 2012	M	Adherence	≥ 71	—	—	Medication adherence	1
Patel 2007	M	Adherence	41‐70	M = F (± 3%)	—	HRQLAdherence to statins	1
Reynolds 2011	M	Adherence	—	—	—	Medication adherence (refill rate)	1
Sherrard 2009	M	Adherence	41‐70	—	—	Adherence and adverse events	1
Simon 2010b	M	Adherence	41‐70	> 50% M	B	Retinopathy examination	4
Stacy 2009	M	Adherence	41‐70	> 50% F	—	6‐month point prevalence	1
Stuart 2003	M	Adherence	—	—	—	Adherence to medications	2
Vollmer 2011	M	Adherence	41‐70	> 50% F	W,B,A	Medication adherence	1
Vollmer 2014	M	Adherence	41‐70	M = F (± 3%)	W,B,A	Medication adherence	1
Graziano 2009	M	Diabetes mellitus	41‐70	> 50% M	W	Glycated haemoglobin	2
Homko 2012	M	Diabetes mellitus	41‐70	> 50% F	W,B,H	Maternal blood glucose level Infant birth weight	2 2 (median effect = 2)
Katalenich 2015	M	Diabetes mellitus	41‐70	> 50% F	W,B,H,A	Glycated haemoglobin Medication adherence Quality of life Cost‐effectiveness	2 2 2 1 (median effect = 2)
Khanna 2014	M	Diabetes mellitus	41‐70	> 50% M	H	Glycated haemoglobin	4
Kim 2014	M	Diabetes mellitus	—	—	—	Glycated haemoglobin	1
Lorig 2008	M	Diabetes mellitus	41‐70	> 50% F	H	Glycated haemoglobin Health distress Global health Hypoglycaemia Hyperglycaemia Activity limitation Fatigue Physical activity levels Communication with physician Glucose monitoring Self‐efficacy Healthcare utilisation	4 4 4 2 4 4 4 2 4 1 4 2 (median effect = 4)
Piette 2000	M	Diabetes mellitus	41‐70	> 50% F	W,H	Depression  Anxiety  Self‐efficacy  Days in bed because of illness Days cut down on activities because of illness Diabetes‐specific health‐related quality of life Satisfaction with care (English speakers only) General health‐related quality of life (English speakers only)	1 2 1 1 2 6 1 1 (median effect = 1)
Piette 2001	M	Diabetes mellitus	41‐70	> 50% M	W,B,H	Glucose monitoring Foot inspection Weight monitoring Medication adherence Glycated haemoglobin Serum glucose levels Diabetic symptoms (all) Hyperglycaemic symptoms Hypoglycaemic symptoms Vascular symptoms Other symptoms Satisfaction with care (summary score)	1 1 2 4 2 2 1 2 2 2 1 1 (median effect = 2)
Schillinger 2009	M	Diabetes mellitus	41‐70	> 50% F	W,B,H,A	Change in self management behaviour (self‐monitoring of blood glucose and self‐monitoring of diabetic foot)	1
Williams 2012	M	Diabetes mellitus	41‐70	> 50% M	—	Glycated haemoglobin Health‐related quality of life (mental) Health‐related quality of life (physical)	1 1 2 (median effect = 1)
Capomolla 2004	M	Heart failure	41‐70	> 50% M	—	All‐cause mortality Re‐hospitalisation Emergency room use (composite outcome)	1
Chaudhry 2010	M	Heart failure	41‐70	> 50% M	W,B	Readmission for any reason or death from any cause	4
Krum 2013	M	Heart failure	≥ 71	> 50% M	—	Packer clinical composite score: death, hospital admission for heart failure, withdrawal from study due to worsening heart failure, 7‐point global health assessment questionnaire	2
Kurtz 2011	M	Heart failure	41‐70	> 50% M	—	Cardiovascular deaths and hospitalisations (outcomes in isolation included cardiovascular deaths, hospitalisations for heart failure, and time to primary endpoint)	1
Shet 2014	M	HIV	—	> 50% M	A	Time to virological failure (viral load > 400 copies/mL on 2 consecutive measurements)	2
Hyman 1996	M	Hypercholestorolemia	41‐70	> 50% F	W	Total cholesterol reduction	2
Hyman 1998	M	Hypercholesterolemia	41‐70	> 50% F	B	Total cholesterol reduction	2
Bove 2013	M	Hypertension	41‐70	> 50% F	W,B,H	Blood pressure control at 6 months	2
Dedier 2014	M	Hypertension	41‐70	> 50% F	H	Change in minutes of moderate or greater physical activity Change in systolic blood pressure	1 2 (median effect = 2)
Harrison 2013	M	Hypertension	—	—	—	Blood pressure	1
Magid 2011	M	Hypertension	41‐70	> 50% M	W,H	Proportion to achieve guideline‐recommended blood pressure goals Systolic blood pressure Diastolic blood pressure	2 1 2 (median effect = 2)
Piette 2012	M	Hypertension	41‐70	> 50% F	—	Systolic blood pressure	2
Farzanfar 2011	M	Mental health	22‐40	> 50% F	W,B	Quality of life (physical scale score and mental scale score) Depression Stress levels Total well‐being	2, 2 2 2 2 (median effect = 2)
Greist 2002	M	Mental health	22‐40	> 50% M	W	Yale‐Brown obsessive compulsive scale (YBOCS) score	1
Zautra 2012	M	Mental health	—	—	—	Emotional health Physical health Stress	1 1 1 (median effect = 1)
DeMolles 2004	M	Obstructive sleep apnoea syndrome	41‐70	—	—	Continuous positive airway pressure use	2
Sparrow 2010	M	Obstructive sleep apnoea syndrome	41‐70	> 50% M	—	Continuous positive airway pressure use	1
Brendryen 2008	M	Smoking	22‐40	M = F (± 3%)	—	Repeated point abstinence	1
Carlini 2012	M	Smoking	22‐40	> 50% F	W,B,H,A	Re‐enrollment into quit line support	1
Ershoff 1999	M	Smoking	22‐40	> 50% F	W,B	Smoking abstinence	2
McNaughton 2013	M	Smoking	41‐70	> 50% M	—	Self‐reported abstinence Biochemically confirmed smoking abstinence	2 4 (median effect = 3)
Peng 2013	M	Smoking	0‐21	> 50% M	A	Stage of change Self‐efficacy Decisional balance	5 5 5 (median effect = 5)
Regan 2011	M	Smoking	41‐70	> 50% M	—	Smoking abstinence	4
Reid 2007	M	Smoking	41‐70	> 50% M	—	Smoking abstinence	2
Reid 2011	M	Smoking	—	—	—	Smoking abstinence	2
Rigotti 2014	M	Smoking	41‐70	M = F (± 3%)	W,B,H,A	Biochemically confirmed tobacco abstinence at 6 months	1
Velicer 2006	M	Smoking	41‐70	> 50% M	W,B,A	24‐h point prevalence 7‐d point prevalence 6‐month prolonged abstinence	2 2 2 (median effect = 2)
Houlihan 2013	M	Spinal cord dysfunction	41‐70	> 50% M	W,B,H	Prevalence of pressure ulcers Depression severity Healthcare utilisation	2 1 2 (median effect = 2)

ATCS versus control on immunisation rates
Patient or population: participants at risk of under‐immunisation (children, adolescents and adults)  Settings: primary care  Intervention: ATCS (ATCS+, IVR, unidirectional) Comparison: no intervention, usual care or health information (letter)
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	No of participants  (studies)	Quality of the evidence  (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	ATCS
Behavioural outcome: immunisation rate ATCS Plus, IVR, unidirectional versus no calls, letters, usual care at median follow‐up of 4 months	Study population^a: children Comparator: no intervention		RR 1.25 (1.18 to 1.32)	10,454  (5 studies)	⊕⊕⊕⊝  Moderate^c	Franzini 2000 (N = 1138) reported that compared with controls (no calls), unidirectional ATCS (autodialer) may increase immunisation rates in children (86% versus 64%, low certainty).^d
	308 per 1000	385 per 1000   (363 to 406)
	Moderate^b
	373 per 1000	466 per 1000   (440 to 492)
Behavioural outcome: immunisation rate Unidirectional ATCS versus usual care at median follow‐up of 15 months	Study population^a: adolescents Comparator: usual care		RR 1.06   (1.02 to 1.11)	5725  (2 studies)	⊕⊕⊕⊝  Moderate^e	Szilagyi 2013 (N = 4115) also reported that unidirectional ATCS probably slightly improves the uptake of preventive care visits, compared with usual care (63% ATCS versus 59% usual care; moderate certainty evidence^f).
	543 per 1000	576 per 1000   (554 to 603)
	Moderate^b
	540 per 1000	572 per 1000   (551 to 599)
Behavioural outcome: immunisation rate Unidirectional ATCS versus no calls or health information at median follow‐up of 2.5 months	Study population^a: adults Comparator: no calls or health information		RR 2.18 (0.53 to 9.02)	1743  (2 studies)	⊕⊝⊝⊝  Very low^g,h	—
	10 per 1000	21 per 1000   (5 to 88)
	Moderate^b
	66 per 1000	144 per 1000   (35 to 595)
Adverse outcome: unintended adverse events attributable to the intervention ATCS+, IVR, unidirectional versus various controls	No studies reported adverse events.
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ATCS Plus: automated telephone communication systems with additional functions; ATCS: automated telephone communication systems; CI: confidence interval; IVR: interactive voice response; RR: risk ratio; unidirectional ATCS enable non‐interactive voice communication and use one‐way transmission of information or reminders.
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS versus control on physical activity levels
Patient or population: participants at risk of developing long‐term conditions Settings: various settings Intervention: ATCS (multimodal/complex intervention, ATCS+, IVR) Comparison: no intervention, usual care, or IVR
Outcomes	Effect of intervention ^a	No of participants  (studies)	Quality of the evidence  (GRADE)
Behavioural outcome: physical activity Multimodal/complex intervention^b versus no calls	The intervention may slightly improve the frequency of walks.	181 (1 study)	⊕⊕⊝⊝ Low^c
Behavioural outcome: physical activity, 12 months Multimodal/complex intervention^d versus usual care	The intervention probably has mixed effects on gait speeds, little effect on functional outcomes (moderate certainty^e) and may slightly increase physical activity levels (low certainty^f).	700 (2 studies)	—
Behavioural outcome: physical activity ATCS Plus versus IVR control	2 studies reported that ATCS Plus intervention may have little or no effect on different indices of physical activity.	369 (2 studies)	⊕⊕⊝⊝ Low^c
Behavioural outcome: physical activity IVR versus usual care, control or health education	3 studies reported that IVR interventions may slightly improve several indices of physical activity (muscle strength, balance, moderate to vigorous physical activity) but may have little or no effect on others (physical activity levels, walking distance).	216 (3 studies)	⊕⊕⊝⊝ Low^g
Clinical outcome: metabolic markers, 12 months Multimodal/complex intervention^d versus usual care	The intervention may have little or no effect on glycated haemoglobin, fasting insulin and glucose levels.	302 (1 study)	⊕⊕⊝⊝ Low^f
Clinical outcome: body weight measures Multimodal/complex intervention^d ATCS Plus versus usual care or control	ATCS Plus intervention may have little or no effect on BMI, weight, waist or waist‐hip ratio, compared with control (71 participants; low certainty evidence^c). Multimodal/complex intervention may have little or no effect on BMI, waist circumference or physical function, compared with usual care (302 participants; low certainty evidence^f).	373 (2 studies)	⊕⊕⊝⊝ Low
Adverse outcome: unintended adverse events attributable to the intervention Multimodal/complex intervention, ATCS Plus, IVR versus various controls	No studies reported adverse events.	—	—
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.
ATCS: automated telephone communication systems; ATCS Plus: automated telephone communication systems with additional functions; IVR: interactive voice response.

ATCS versus control on screening rates
Patient or population: participants at risk for breast, colorectal or cervical cancer; or osteoporosis Settings: primary, secondary and tertiary care Intervention: ATCS (multimodal/complex intervention, ATCS Plus, IVR, unidirectional) Comparison: usual care, enhanced usual care or no intervention
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	No of participants  (studies)	Quality of the evidence  (GRADE)	Comments
	Assumed risk	Corresponding risk
	Usual care or enhanced usual care or no intervention	ATCS
Behavioural outcome: breast cancer screening Multimodal/complex intervention versus usual care at 12 months follow‐up	Study population^a		RR 2.17   (1.55 to 3.04)	462  (2 studies)	⊕⊕⊕⊕  High	—
	167 per 1000	363 per 1000   (259 to 508)
	Moderate^b
	167 per 1000	363 per 1000   (259 to 508)
Behavioural outcome: breast cancer screening IVR versus enhanced usual care at median follow‐up of 12 months	Study population^a		RR 1.05   (0.99 to 1.11)	2599  (2 studies)	⊕⊕⊕⊝  Moderate^c	Unidirectional ATCS versus letter 1 further study (Fortuna 2014) (N = 1008) found that unidirectional ATCS (plus letter) probably has little or no effect on breast cancer screening rates at 12 months, adjusted OR 1.3 (95% CI 0.7 to 2.4; moderate certainty^d).
	585 per 1000	614 per 1000   (579 to 649)
	Moderate^b
	432 per 1000	454 per 1000   (428 to 480)
Behavioural outcome: colorectal cancer screening Multimodal/complex intervention versus usual care at median follow‐up of 12 months	Study population^a		RR 2.19   (1.88 to 2.55)	1013  (3 studies)	⊕⊕⊕⊕  High	—
	249 per 1000	545 per 1000   (468 to 635)
	Moderate^b
	167 per 1000	366 per 1000   (314 to 426)
Behavioural outcome: colorectal cancer screening IVR versus usual care at 6‐month follow‐up	Study population^a		RR 1.36   (1.25 to 1.48)	16915  (2 studies)	⊕⊕⊕⊝  Moderate^e	IVR versus control 1 other study (Durant 2014) (N = 47,097) reported that IVR probably increases screening, with 1773 participants from the IVR group and 100 from the no‐call control group completing colorectal cancer screening within 3 months (moderate certainty^f). IVR versus usual care 1 study (Mosen 2010) (N = 6000) also reported that IVR probably increases completion of any colorectal cancer screening (moderate certainty^g).
	119 per 1000	161 per 1000   (148 to 176)
	Moderate^b
	119 per 1000	162 per 1000   (149 to 176)
Behavioural outcome: colorectal cancer screening IVR, unidirectional ATCS versus usual care or letter at longer (9‐12 months) follow‐up	Study population^a		RR 1.01   (0.97 to 1.05)	21,335  (2 studies)	⊕⊕⊕⊝  Moderate^h	IVR versus usual care 1 study (Simon 2010a) (N = 20,000) also reported that IVR probably increases slightly colorectal cancer screening via colonoscopy (moderate certaintyⁱ). Unidirectional ATCS versus letter 1 further study (Fortuna 2014) (N = 1008) at 12 months found that unidirectional ATCS (plus letter) has probably little or no effect on colorectal cancer screening rates at 12 months (15.3% versus 12.2%; adjusted OR 1.2; 95% CI 0.6 to 2.4; moderate certainty^d).
	302 per 1000	305 per 1000   (293 to 317)
	Moderate^b
	245 per 1000	247 per 1000   (238 to 257)
Behavioural outcome: cervical cancer screening ATCS Plus versus control (no calls) at 3 month follow‐up	See comment	See comment	Not estimable	75,532 (1 study)	⊕⊕⊕⊝  Moderate^j	Corkrey 2005 found that ATCS Plus intervention probably slightly improves cervical cancer screening rates at 3 months.
Adverse outcome: unintended adverse events attributable to the intervention Multimodal/complex intervention, ATCS Plus, IVR, unidirectional versus various controls	No studies reported adverse events.
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  ATCS: automated telephone communication systems; ATCS Plus: automated telephone communication systems with additional functions; BMD: bone mineral density; CI: confidence interval; HR: hazard ratio; IVR: interactive voice response; OR: odds ratio; RR: risk ratio.
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS versus control for body weight
Patient or population: overweight or obese individuals (both children and adults)  Settings: various settings  Intervention: ATCS (multimodal/complex intervention, ATCS Plus, IVR) Comparison: usual care, no intervention or control
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	No of participants  (studies)	Quality of the evidence  (GRADE)	Comments^a
	Assumed risk	Corresponding risk
	Controls	ATCS
Clinical and behavioural outcome: BMI score in adults Multimodal/complex intervention, ATCS Plus or IVR versus usual care at median follow‐up of 18 months	The mean BMI in the control groups was 34.7 kg/m²	The mean BMI of adults in the intervention groups was 0.64 kg/m²lower   (1.38 lower to 0.11 higher)	Not estimable	672  (3 studies)	⊕⊕⊝⊝  Low^b	ATCS Plus versus control Vance 2011 (N = 140) found that ATCS Plus may reduce slightly BMI (low certainty evidence^c).
Clinical and behavioural outcome: body weight in adults, 12 weeks	See comment	See comment	Not estimable	See comment	See comment	ATCS Plus versus control Vance 2011 (N = 140) found that ATCS Plus may reduce slightly body weight and waist circumference (low certainty evidence^c). IVR versus control Estabrooks 2008 (N = 77) reported that IVR may have little or no effect on body weight (percent lost or change in) (low certainty evidence^d).
Clinical and behavioural outcome: body weight in adults, at median follow‐up of 18 months	See comment	See comment	Not estimable	See comment	See comment	ATCS (multimodal/complex intervention, ATCS Plus, IVR) versus usual care Bennett 2012 (N = 365) found that ATCS Plus probably slightly reduces body weight at 18 months (moderate certainty evidence).^eBennett 2013 (N = 194) found that multimodal/complex intervention may reduce body weight at 18 months (low certainty evidence).^f IVR versus usual care Goulis 2004 (N = 122) found that IVR probably reduces slightly body weight but probably has little or no effect on obesity assessment scores at 6 months (moderate certainty evidence).^f
Clinical and behavioural outcome: blood pressure, blood glucose, cholesterol levels	See comment	See comment	Not estimable	See comment	See comment	ATCS (ATCS Plus, IVR) versus usual care/control Bennett 2012 (N = 365) found that ATCS Plus probably has little or no effect on systolic or diastolic blood pressure at 18 months (moderate certainty evidence^e). ATCS Plus versus control Vance 2011 found that ATCS Plus may slightly improve slightly systolic blood pressure and blood glucose levels at 12 weeks (low certainty evidence^c). IVR versus usual care Goulis 2004 (N = 122) found that IVR probably has little or no effect on systolic or diastolic blood pressure, plasma glucose levels, or high‐density lipoprotein cholesterol, but it probably slightly reduces total cholesterol and triglyceride levels at 6 months (moderate certainty evidence).^e
Clinical outcome: BMI z‐score in children at median follow‐up of 7.5 months	See comment	See comment	Not estimable	See comment	⊕⊕⊕⊝  Moderate^e	ATCS Plus versus control Estabrooks 2009 (N = 220) found that ATCS Plus has probably little or no effect on BMI z‐scores in children at 12 months. IVR versus control Wright 2013 (N = 100) found that IVR has probably little or no effect on BMI z‐scores in children at 3 months.
Behavioural outcome: physical activity, dietary habits in children at median follow‐up of 7.5 months	See comment	See comment	Not estimable	See comment	⊕⊕⊕⊝  Moderate⁴	ATCS Plus versus control Estabrooks 2009 (N = 220) found that ATCS Plus has probably little or no effect on self‐reported physical activity, sedentary behaviours or dietary habits at 12 months. IVR versus control (no calls) Wright 2013 (N = 100) found that IVR has probably little or no effect on total caloric intake, fruit intake, or sedentary behaviours at 3 months.
Adverse outcome: unintended adverse events attributable to the intervention IVR versus usual care	See comment	See comment	See comment	559 (2 studies)	See comment	Bennett 2012 (N = 365) reported 1 serious musculoskeletal injury in the intervention group and 3 events (1 cardiovascular and 2 cases of gallbladder disease) in the usual care group (moderate certainty evidence).^e,g Bennett 2013 (N = 194) reported 6 serious adverse events in the intervention arm, including gynaecological surgery in 2 participants and knee replacement, breast abscess, musculoskeletal injury, and cancer diagnosis in 1 participant each; all participants except the one with the cancer diagnosis required hospitalisation (low certainty evidence).^f,g
The corresponding risk* (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ATCS: automated telephone communication systems; ATCS Plus: automated telephone communication systems with additional functions; BMI: body Mass Index; CI: confidence interval; IVR: interactive voice response; SMD: Standardised mean difference.
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS versus control as appointment reminders (reducing non‐attendance rates)
Patient or population: patients/healthcare consumers  Settings: various settings  Intervention: ATCS (ATCS Plus, IVR, unidirectional) Comparison: no intervention (calls) or nurse‐delivered calls
Outcomes	Effect of intervention^a	No of participants (studies)	Quality of the evidence  (GRADE)
Health behaviour: attendance rates, 6 weeks ATCS Plus versus nurse‐delivered calls	ATCS Plus calls delivered 3 or 7 days prior to flexible sigmoidoscopy or/and colonoscopy examinations probably have little or no effect on appointment non‐attendance or preparation non‐adherence.	3610 (1 study)	⊕⊕⊕⊝  Moderate^b
Health behaviour: attendance rates, 4 months IVR versus no calls	IVR improves attendance rates: OR 1.52 (95% CI 1.34 to 1.71).	12,092 (1 study)	⊕⊕⊕⊕ High
Health behaviour: return tuberculin test rate, 3 days Unidirectional ATCS versus no calls	Unidirectional ATCS may improve test return rates.	701 (1 study)	⊕⊕⊝⊝  Low^c
Health behaviour: attendance rates, 1 month Unidirectional ATCS versus no calls	Undirectional ATCS may improve attendance rates RR 1.60 (95% CI 1.29 to 1.98).	517 (1 study)	⊕⊕⊝⊝  Low^c
Health behaviour: attendance rates, 6‐8 weeks Unidirectional ATCS versus no calls	2 studies reported conflicting results: Reekie 1998 (N = 1000) reported that unidirectional ATCS probably decrease non‐attendance rates at 6 weeks; while Maxwell 2001 (N = 2304) reported the interventions probably have little or no effect at 2 months.	3304 (2 studies)	⊕⊕⊕⊝  Moderate^d
Health behaviour: attendance rates, 6 months Unidirectional ATCS versus no calls	Unidirectional ATCS may improve attendance: OR 1.50 (P < 0.01).	2008 (1 study)	⊕⊕⊝⊝  Low^e
Adverse outcome: unintended adverse events attributable to the intervention ATCS Plus, IVR, unidirectional ATCS versus various controls	No studies reported adverse events.
ATCS: automated telephone communication systems; ATCS Plus: automated telephone communication systems with additional functions; CI: confidence interval; IVR: interactive voice response; OR: odds ratio; RR: risk ratio.
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS versus control for adherence to medication or laboratory tests
Patient or population: patients with various conditions or at risk of low adherence to medication or laboratory tests Settings: various settings Intervention: ATCS (multimodal/complex intervention, ATCS Plus, IVR, unidirectional ATCS) Comparison: usual care, no calls, controls (other ATCS)
Outcomes	Effect of interventions^a	No of participants  (studies)	Quality of the evidence  (GRADE)	Comments
Behavioural outcome: adherence to medication Multimodal/complex interventions^b versus usual care or control	The effects of multimodal/complex interventions are inconclusive.	888 (2 studies)	See comment	Ho 2014 (N = 241) reported that the multimodal/complex intervention probably improves adherence to cardioprotective medications at 12 months (moderate certainty^c). Stuart 2003 (N = 647) found uncertain effects of the intervention on adherence to antidepressant medications (very low certainty^c,d).
Behavioural outcome: adherence to medication ATCS Plus versus control or single IVR call	Results suggest that ATCS Plus probably slightly improve measures of adherence.	2340 (2 studies)	See comment	Cvietusa 2012 (N = 1393) reported that ATCS Plus, compared with control, probably improves time to first inhaled corticosteroid refill and probably slightly improves the proportion of days with medication on hand in children (moderate certainty^e). Stacy 2009 (N = 947) reported that ATCS Plus probably slightly improves statin adherence at 6 months, compared with a single IVR call (moderate certainty^f).
Behavioural outcome: adherence to laboratory tests ATCS Plus or IVR versus no intervention or usual care	Results suggest that ATCS Plus probably has little or no effect on adherence to testing, while IVR probably improves test completion.	15,218 (3 studies)	See comment	ATCS Plus versus no intervention Derose 2009 (N = 13,057) found that ATCS Plus probably has little or no effect on adherence to testing (completion of all 3 recommended laboratory tests for diabetes patients) at 12 weeks (moderate certainty^g). Simon 2010b (N = 1200) found that these interventions probably have little or no effect on retinopathy examination rates or tests for glycaemia, hyperlipidaemia or nephropathy in diabetic patients at 12 months (moderate certainty^h). IVR versus usual care Feldstein 2006 (N = 961) found that IVR probably improves patients' completion of all recommended laboratory tests at 25 days follow‐up (moderate certaintyⁱ).
Behavioural outcome: adherence to medication or composite outcome (medication adherence and rate of adverse events) ATCS Plus versus usual care	Results indicate that ATCS Plus probably improves medication adherence and may slightly improve a composite measure.	35,816 (4 studies)	See comment	2 studies (Derose 2013 (N = 5216) and Vollmer 2014 (N = 21,752)) reported that ATCS Plus probably improves adherence to statins to some extent. Vollmer 2011 (N = 8517) found that ATCS Plus probably slightly improves adherence to inhaled corticosteroids (moderate certainty^j). Sherrard 2009 (N = 331) found that ATCS Plus may slightly improve a composite measure of medication adherence and adverse events at 6 months follow‐up (low certainty^c,k).
Behavioural outcome: adherence to medication or laboratory tests IVR versus control	Results suggest that IVR probably improves slightly medication adherence.	4,238,362 (4 studies)	See comment	Adams 2014 (N = 475) found that IVR may slightly improve comprehensiveness of screening and counselling (low certainty^c,l). Bender 2010 (N = 50) reported that IVR may improve adherence to anti‐asthmatic medications at 2.5 months follow‐up (low certainty^c,e). Leirer 1991 (N = 16) reported that IVR may slightly reduce medication non‐adherence (low certainty^m). Mu 2013 (N = 4,237,821) found that IVR probably slightly improves medication refill rates at 1 month (moderate certaintyⁿ).
Behavioural outcome: adherence to medication IVR versus usual care	Results indicate that IVR probably slightly improves some measures of medication adherence.	56,140 (8 studies)	See comment	2 studies (Boland 2014 (N = 70); Friedman 1996 (N = 267)) reported that IVR probably slightly improves adherence to glaucoma and anti‐hypertensive medications at 3 and 6 months respectively (moderate certainty).^o 2 further studies (Glanz 2012 (N = 312); Migneault 2012 (N = 337)) reported that IVR has probably little or no effect on medication adherence at 8 and 12 months, respectively (moderate certainty).^p 2 studies (Green 2011 (N = 8306); Reynolds 2011 (N = 30,610)) assessed adherence via refill rates, reporting that IVR probably slightly improves medication refill rates at 2 weeks (moderate certainty).^q 2 further studies reported medication adherence assessed by medication possession ratio (MPR) at different time points. Patel 2007 (N = 15,051) found that IVR probably slightly improves MPR at 3 to 6 months, while Bender 2014 (N = 1187) reported that IVR probably improves MPR at 24 months (both studies of moderate certainty^r).
Behavioural outcome: adherence to medication Unidirectional ATCS versus control	Results suggest that unidirectional ATCS may have little effect, or improve medication adherence to a small degree.	107 (2 studies)	See comment	2 studies (Lim 2013 (N = 80); Ownby 2012 (N = 27)) reported that the intervention may have little effect or slightly improve medication adherence (low certainty^s).
Clinical outcome: blood pressure Multimodal/complex, ATCS Plus, IVR versus usual care	Results suggest that ATCS Plus probably slightly reduces blood pressure, while multimodal/complex or IVR interventions probably have little or no effect on blood pressure.	22,597 (3 studies)	See comment	Multimodal/complex intervention versus usual care Ho 2014 (N = 241) reported that multimodal intervention probably has little or no effect on achieving reduced blood pressure targets (moderate certainty^c). ATCS Plus versus usual care Vollmer 2014 (N = 21,752) reported that ATCS Plus probably slightly reduces systolic blood pressure (moderate certainty^t). IVR versus usual care Migneault 2012 (N = 337) reported that IVR probably has little or no effect on systolic or diastolic blood pressure (moderate certainty^c), while Friedman 1996 (N = 267) found that IVR may have little or no effect on systolic blood pressure but may slightly decrease diastolic blood pressure (low certainty^c,f).
Adverse outcome: unintended adverse events attributable to the intervention Multimodal/complex intervention, ATCS Plus, IVR, unidirectional versus various controls	No studies reported adverse events.
ATCS Plus: automated telephone communication systems with additional functions; CI: confidence interval; HR: hazard ratio; IVR: interactive voice response; MPR: medication possession ratio; OR: odds ratio; RR: risk ratio; SD: standard deviation
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS versus control on alcohol consumption
Patient or population: participants addicted to alcohol Settings: various settings Intervention: ATCS (ATCS Plus, IVR) Comparison: no intervention, usual care, advice/education or packaged CBT
Outcomes	Effect of intervention^a	No of participants  (studies)	Quality of the evidence  (GRADE)
Behavioural outcomes: number of drinks per drinking day ATCS Plus, IVR versus usual care, (various) controls at median follow‐up of 2 months	ATCS Plus versus usual care Rose 2015 (N = 158) reported that ATCS Plus may have little or no effect on the number of drinks per drinking day at 2 months (low certainty^b,c). ATCS Plus versus control (advice/education) Hasin 2013 (N = 254) found that ATCS Plus may reduce the number of drinks per drinking day in the last 30 days at 2 months (low certainty^b,c), but it may have little effect at 12 months. IVR versus control (information) Rubin 2012 (N = 47) reported that IVR may slightly reduce the number of drinks per drinking day at 6 months (low certainty^c,e).	459 (3 studies)	⊕⊕⊝⊝  Low
Behavioural outcomes: drinking days, heavy drinking days, or total number of drinks consumed ATCS Plus, IVR versus (various) controls	ATCS Plus versus no intervention Mundt 2006 (N = 60) found that ATCS Plus may have little or no effect on drinking days, heavy drinking days, or total number of drinks consumed (low certainty^c,f). ATCS Plus versus control (packaged CBT) Litt 2009 (N = 110) found that ATCS Plus may have little or no effect on the number of heavy drinking days at 12 weeks posttreatment (low certainty^c,g). IVR versus control (information) Rubin 2012 (N = 47) reported that IVR may slightly reduce the number of heavy drinking days per month at 6 months (low certainty^c,e).	217 (3 studies)	⊕⊕⊝⊝  Low
Behavioural outcomes: proportion of days abstinent, other alcohol consumption indices, 12 weeks ATCS Plus versus control (packaged CBT)	ATCS Plus may slightly reduce the proportion of days abstinent but have little or no effect on coping or drinking problems or continuity of abstinence (Litt 2009).	110 (1 study)	⊕⊕⊝⊝  Low^c,g
Behavioural outcomes: weekly alcohol consumption, 6 months ATCS Plus versus usual care	ATCS Plus may have little or no effect on weekly alcohol consumption (Helzer 2008).	338 (1 study)	⊕⊕⊝⊝  Low^c,h
Behavioural outcomes: AUDIT score, 6 weeks IVR versus control (no intervention)	IVR probably improve slightly AUDIT scores (Andersson 2012).	1423 (1 study)	⊕⊕⊕⊝  Moderateⁱ
Behavioural outcomes: other alcohol consumption indices, 4 weeks IVR versus control (no intervention)	IVR may have little or no effect on drinking habits, alcohol craving, or PTSD symptoms (Simpson 2005).	98 (1 study)	⊕⊕⊝⊝  Low^c,h
Adverse outcome: unintended adverse events attributable to the intervention ATCS Plus, IVR versus various controls	No studies reported adverse events.
ATCS Plus: automated telephone communication systems with additional functions; AUDIT: Alcohol Use Disorders Identification Test; CBT: cognitive behavioural therapy; IVR: interactive voice response; PTSD: post‐traumatic stress disorder.
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

S27	s26
S26	s10 and s25
S25	S11 or S12 or S13 or S14 or S15 or S16 or S17 or S18 or S19 or S20 or S21 or S22 or S23 or S24
S24	AB "time series" or TI "time series"
S23	AB ("pre test" or pretest or "post test" or posttest or preintervention or postintervention) or TI ("pre test" or pretest or "post test" or posttest or preintervention or postintervention)
S22	TI (singl* or doubl* or tripl* or trebl) and TI (blind or mask*)
S21	AB (singl* or doubl* or tripl* or trebl) and AB (blind or mask*)
S20	AB (random* or trial or placebo* or assign* or allocat* or volunteer* or factorial* or experiment* or control* or compar* or intervention* or chang* or evaluat* or impact* or effect?) or TI (random* or trial or placebo* or assign* or allocat* or volunteer* or factorial* or experiment* or control* or compar* or intervention* or chang* or evaluat* or impact* or effect?)
S19	PT Clinical Trial
S18	MH Quasi‐Experimental Studies+
S17	MH Quantitative Studies
S16	MH Placebos
S15	MH Crossover Design
S14	MH Comparative Studies
S13	MH Random Assignment
S12	MH Experimental Studies+
S11	"randomi?ed controlled trial" or PT randomized controlled trial
S10	s1 or s2 or s3 or s4 or s5 or s6 or s7 or s8 or s9
S9	(telephon* or phone*) N3 (system or technology)
S8	((automat* or computer) and (intervention or counsel* or advice* or advis* or educat* or remind* or messag* or service* or support or appointment)) and MW (phone or telephon* or teleconsultation* or hotline* or answering‐service*)
S7	answering N1 (service* or machine*)
S6	automat* N1 (call* or answer*)
S5	(prerecorded or pre‐recorded) and (telephon* or phone* or voice* or hotline* or "hot line" or call or calls or message)
S4	computer* N2 (telephon* or phone*)
S3	"speech recognition"
S2	voice N1 (response or recognition or messag* or system* or technolog*)
S1	(automat* or interactive) N5 (telephon or phone* or voice* or hotline* or "hot line*")

# 11	#9 and #10 Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH
# 10	TS=(health* or medic or patient* or clinic* or hospital* or illness* or disease* or disorder* or therap or physician* or doctor* or psychiatr* or treatment* or counsel) Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH*
# 9	#1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH
# 8	TS=(touch‐tone) Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH
# 7	TS=(answering near/1 (service* or machine)) Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH*
# 6	TS=(automat* near/1 (call* or answer)) Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH*
# 5	TS=((prerecorded or pre‐recorded) and (telephon* or phone* or voice* or hotline* or "hot line" or call or calls or message)) Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH
# 4	TS=("speech recognition" and (software or automat* or interactive)) Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH*
# 3	TS=(computer* near/2 (telephon* or phone)) Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH*
# 2	TS=(voice near/1 (response or recognition or messag* or system* or technolog)) Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH*
# 1	TS=((automat* or interactive) near/5 (telephon or phone* or voice* or hotline* or "hot line")) Indexes=SCI‐EXPANDED, SSCI, A&HCI, CPCI‐S, CPCI‐SSH*

Study	Hess 2013
intervention	no. patients (n_i)	no. vaccin (Z_i)	Z_i*(n_i‐Z_i)/n_i
A	454	4	3.964758
B	1191	25	24.47523
C	1156	30	29.22145
D	351	10	9.7151
E	729	17	16.60357
F	392	17	16.26276
G	642	16	15.60125
H	684	27	25.93421
Total.I	5599	146	141.7783
Control
I	816	0	0
J	657	11	10.81583
K	1231	5	4.979691
L	435	3	2.97931
M	579	0	0
N	1152	3	2.992188
O	1157	21	20.61884
P	356	3	2.974719
Total.C	6383	46	45.36058
Overall Total (N)	11982	192	187.1389
no. clusters (k)	16
pi	0.016024036
ICC.Fleiss‐Cuzick	0.00812106
average cluster size (M)		748.875
design effect		7.073537484

original data	ATCS events: 146; total: 5599 and control events: 46; total: 6383
intervention	5599	146
control	6383	46
updated data	divided by the design effect
intervention	791.5417163	20.64030909
control	902.3773486	6.503111082

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Immunisation in children	5	10454	Risk Ratio (M‐H, Random, 95% CI)	1.25 [1.18, 1.32]
2 Immunisation in adolescents	2	5725	Risk Ratio (M‐H, Random, 95% CI)	1.06 [1.02, 1.11]
3 Immunisation in adults	2	1743	Risk Ratio (M‐H, Random, 95% CI)	2.18 [0.53, 9.02]

ATCS versus control on severity of cancer symptoms
Patient or population: cancer patients Settings: various settings Intervention: ATCS (multimodal/complex intervention, ATCS Plus, IVR) Comparison: usual care, control (other ATCS, nurse‐delivered calls)
Outcomes	Effects of intervention^a	No of participants  (studies)	Quality of the evidence  (GRADE)	Comments
Clinical outcomes: symptoms (severity or burden) ATCS Plus versus usual care (via ATCS) or control, 4‐12 weeks	Results suggest that ATCS Plus may have little or no effect on symptom severity, distress or burden.	701 (4 studies)	See comment	Cleeland 2011 (N = 79) found that ATCS Plus may slightly reduce symptom threshold events and cumulative distribution of symptom threshold events; and it may have little or no effect on mean symptom severity between discharge and 4 week follow‐up (low certainty^b,c). Mooney 2014 (N = 250) found that ATCS Plus probably has little or no effect on symptom severity scores at 6 week follow‐up (moderate certainty^c). Spoelstra 2013 (N = 119) found that ATCS Plus may have little or no effect on symptom severity at 10 week follow‐up (low certainty^c,d). Yount 2014 (N = 253) reported that ATCS Plus may have little or no effect on symptom burden at 12 weeks (low certainty^c,e).
Clinical outcomes: symptom severity, 10 weeks IVR versus nurse delivered calls	Results suggest that IVR may have little or no effect on symptom severity.	437 (1 study)	⊕⊕⊝⊝  Low^c,f	—
Clinical outcomes: pain Multimodal/complex intervention^g versus usual care	Results indicate that multimodal intervention probably reduces pain at 3 months and probably slightly reduces pain at 12 months.	405 (1 study)	⊕⊕⊕⊝  Moderate^c	—
Clinical outcomes: depression Multimodal/complex intervention^g versus usual care	Results indicate that multimodal intervention probably slightly reduces depression at 3 and 12 months.	405 (1 study)	⊕⊕⊕⊝  Moderate^c	—
Clinical outcomes:distress, 6 weeks ATCS Plus versus usual care (via IVR)	Results indicate that ATCS Plus probably has little or no effect on distress.	250 (1 study)	⊕⊕⊕⊝  Moderate^c	—
Behavioural outcome: medication adherence ATCS Plus versus usual care	Results indicate that ATCS Plus may have little or no effect on medication non‐adherence.	119 (1 study)	⊕⊕⊝⊝  Low^c,d	—
Adverse outcome: unintended adverse events attributable to the intervention Multimodal/complex intervention, ATCS Plus, IVR versus various controls	No studies reported adverse events.
ATCS: automated telephone communication systems; ATCS Plus: automated telephone communication systems with additional functions; IVR: interactive voice response.
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS versus usual care for managing diabetes mellitus
Patient or population: patients with diabetes mellitus  Settings: various settings  Intervention: ATCS (ATCS Plus, IVR) Comparison: usual care
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	No of Participants  (studies)	Quality of the evidence  (GRADE)	Comments^a
	Assumed risk	Corresponding risk
	Usual care	ATCS
Clinical outcome: glycated haemoglobin or blood glucose ATCS Plus, IVR versus usual care at median follow‐up of 6 months	The mean glycated haemoglobin in the control groups was 8.41%	The mean glycated haemoglobin in the intervention groups was  0.26% lower   (0.50 to 0.01 lower)	Not estimable	1216  (7 studies)	⊕⊕⊝⊝  Low^b	ATCS Plus versus usual care 1 further study, Katalenich 2015 (N = 98), found that ATCS Plus may have little or no effect on median glycated haemoglobin levels compared with usual care at 6 months follow‐up (low certainty^c). IVR versus usual care 1 additional study, Homko 2012 (N = 80), found that IVR may have little or no effect on fasting blood glucose levels in pregnancy or infant birth weight at 26 months (low certainty^c).
Behavioural outcome: self‐monitoring of diabetic foot (various scales) ATCS Plus versus usual care at 12 months follow‐up	The mean self‐monitoring of diabetic foot in the control groups was 4.5 (range from 0 to 7, with higher scores indicating better foot care)	The mean self‐monitoring of diabetic foot in the intervention groups was  0.40 points higher^d   (0.10 to 0.71 points higher)	Not estimable	498  (2 studies)	⊕⊕⊕⊝  Moderate^e	—
Behavioural outcome: self‐monitoring of blood glucose ATCS Plus, IVR versus usual care, 6‐12 months	See comment	See comment	Not estimable	See comment	See comment	ATCS Plus versus usual care Lorig 2008 (N = 417) found that ATCS Plus may have little no effect on self‐monitoring of blood glucose at 6 months (low certainty evidence^f). At 12 months, 2 studies (Piette 2001 (N = 272); Schillinger 2009 (N = 339)) reported that ATCS Plus probably slightly improves self‐monitoring of blood glucose (moderate certainty^e). IVR versus usual care Graziano 2009 (N = 112) found that IVR probably slightly increases the mean change in frequency of self‐monitoring of blood glucose (moderate certainty evidence^g).
Behavioural outcome: medication adherence or use ATCS Plus versus usual care, 6‐12 months	See comment	See comment	Not estimable	370 (2 studies)	See comment	Katalenich 2015 (N = 98) reported that ATCS Plus may have little or no effect on adherence rates at 6 months (low certainty^c), and Piette 2001 (N = 272) found that ATCS Plus has probably little or no effect on medication use at 12 months (moderate certainty^g.
Behavioural outcome: physical activity, diet, weight monitoring ATCS Plus versus usual care, 6‐12 months	See comment	See comment	Not estimable	1028 (3 studies)	See comment	Lorig 2008 (N = 417) found that ATCS Plus may have little or no effect on aerobic exercise at 6 months (low certainty^f). Schillinger 2009 (N = 339) found that ATCS Plus may slightly improve diet and exercise and moderate intensity physical activity levels, but it may have little or no effect on vigorous intensity physical activity levels at 12 months (low certainty^c). Piette 2001 (N = 272) reported that ATCS Plus probably has little or no effect on weight monitoring (moderate certainty^g).
Adverse outcome: unintended adverse events attributable to the intervention ATCS Plus, IVR versus usual care	No studies were found that reported adverse events.
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ATCS Plus: automated telephone communication systems with additional functions; CI: confidence interval; HRQoL: health‐related quality of life; IVR: interactive voice response; SMD: standardised mean difference.
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS versus usual care for patients with heart failure
Patient or population: patients with heart failure  Settings: various settings  Intervention: ATCS (multimodal/complex intervention, ATCS Plus, IVR) Comparison: usual care or usual community care
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	No of participants  (studies)	Quality of the evidence  (GRADE)	Comments^a
	Assumed risk	Corresponding risk
	Usual care or usual community care	ATCS
Clinical outcome: cardiac mortality ATCS Plus, IVR versus usual care or usual community care at median follow‐up of 11.5 months	Study population^b		RR 0.60   (0.21 to 1.67)	215  (2 studies)	⊕⊝⊝⊝  Very low^d,e	—
	95 per 1000	57 per 1000   (20 to 158)
	Moderate^c
	96 per 1000	58 per 1000   (20 to 160)
Clinical outcome: all‐cause mortality ATCS Plus versus usual care or usual community care at median follow‐up of 11 months	Study population^b		RR 1 (0.79 to 1.28)	2165  (3 studies)	⊕⊕⊕⊝  Moderate^f	—
	106 per 1000	106 per 1000   (84 to 136)
	Moderate^c
	106 per 1000	106 per 1000   (84 to 136)
Clinical outcome: heart failure hospitalisation ATCS Plus, IVR versus usual care or usual community care at median follow‐up of 11.5 months	See comment	See comment	Not estimable	2329 (4 studies)	See comment	ATCS Plus versus usual care or usual community care Chaudhry 2010 (N = 1653) found that the intervention had little or no effect on hospitalisation for heart failure (high certainty). Krum 2013 (N = 405) also reported that there was probably little or no effect of the intervention for this same outcome (moderate certainty^g), while Capomolla 2004 (N = 133) reported that ATCS Plus may decrease hospitalisation rates for heart failure (low certainty^h). IVR versus usual care Kurtz 2011 (N = 138) reported that IVR intervention has uncertain effects on hospitalisation for heart failure (very low certaintyⁱ).
Clinical outcome: all‐cause hospitalisation ATCS Plus versus usual care or usual community care	See comment	See comment	Not estimable	2191 participants (3 studies)	See comment	ATCS Plus versus usual care Capomolla 2004 (N = 133) found that ATCS Plus may reduce all‐cause hospitalisation (for chronic heart failure, cardiac cause and other cause; low certainty^h), and Krum 2013 (N = 405) similarly reported that the intervention probably slightly decreased all‐cause hospitalisation (moderate certainty^g).^fChaudhry 2010 (N = 1653) found that ATCS Plus has little or no effect on readmission for any reason (high certainty).
Clinical outcome: global health (well‐being) rating (7‐item questionnaire) ATCS Plus versus usual care 12 months	See comment	See comment	Not estimable	405 participants (1 study)	⊕⊕⊕⊝  Moderate^g	Krum 2013 (N = 405) reported that ATCS Plus probably increases slightly the proportion of patients with improved global health questionnaire ratings at 12 months.
Clinical outcome: emergency room and other health service use outcomes ATCS Plus versus usual care or usual community care	See comment	See comment	Not estimable	1786 participants (2 studies)	See comment	Emergency room use Capomolla 2004 (N = 133) found that ATCS Plus may reduce emergency room use at (median) 11 months (low certainty^h). Other service use Chaudhry 2010 (N = 1653) found that ATCS Plus had little or no effect on number of days in hospital or number of hospitalisations (readmissions)(high certainty).
Adverse outcome: unintended adverse events attributable to the intervention ATCS Plus, IVR versus usual care	See comment	See comment	See comment	1791 (2 studies)	See comment	ATCS Plus versus usual care Chaudhry 2010 (N = 1653) reported that no adverse events had occurred during the study (high certainty). IVR versus usual care Kurtz 2011 (N = 138) classified adverse events as cardiac mortality plus rehospitalisation for heart failure, reporting uncertain effects upon this composite outcome (very low certaintyⁱ).
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ATCS Plus: automated telephone communication systems with additional functions; CI: confidence interval; CI: confidence interval; RR: risk ratio.
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS versus usual care for management of hypertension
Patient or population: patients with hypertension  Settings: various settings  Intervention: ATCS (multimodal/complex intervention, ATCS Plus, IVR, unidirectional) Comparison: usual care, with and without education
Outcomes	*Illustrative comparative risks (95% CI)**		No of Participants   (comparisons)	Quality of the evidence  (GRADE)	Comments^a
	Assumed risk	Corresponding risk
	Usual care	ATCS
Clinical outcome: systolic blood pressure (automated sphygmomanometer or electronic pressure monitor) ATCS Plus or IVR versus usual care at median follow‐up of 6 weeks	The mean systolic blood pressure in the control group was 141.1 mmHg	The mean systolic blood pressure in the intervention groups was  1.89 mmHg lower (2.12 to 1.66 lower)	65,256  (3 studies)	⊕⊕⊕⊝  Moderate^b	1 additional study (Dedier 2014) (N = 253) reported that compared with usual care plus education, IVR may have little or no effect on systolic blood pressure at 3 months (low certainty^c).
Clinical outcome: diastolic blood pressure (automated sphygmomanometer and electronic cuff) ATCS Plus, unidirectional versus usual care at median follow‐up of 14 weeks	The mean diastolic blood pressure in the control group was 81.2 mmHg	The mean diastolic blood pressure in the intervention groups was 0.02 mmHg higher (2.62 lower to 2.66 higher)	65,056  (2 studies)	⊕⊕⊝⊝  Low^d,e	—
Clinical outcome: blood pressure control, 26 weeks Multimodal/complex intervention^f versus usual care	See comment	See comment	166 (1 study)	⊕⊕⊕⊝  Moderate^g	Bove 2013 (N = 241) found that a multimodal/complex intervention probably has little or no effect on blood pressure control.
Clinical outcome: Health status^h, depressionⁱ, 6 weeks ATCS Plus versus enhanced usual care (plus information)	See comment	See comment	200 (1 study)	⊕⊕⊝⊝  Low^j	Piette 2012 (N = 200) found that ATCS Plus may slightly improve health status and may decrease depressive symptoms.
Behavioural outcome: medication use Multimodal/complex^k, ATCS Plus versus usual care or enhanced usual care (plus information)	See comment	See comment	483 (2 studies)	⊕⊕⊝⊝  Low	Multimodal/complex versus usual care Magid 2011 (N = 283) found that multimodal/complex intervention may have little or no effect on medication adherence assessed by Medication Possession Ratio or proportion adherent (low certainty^l). ATCS Plus versus enhanced usual care Piette 2012 (N = 200) found that ATCS Plus may reduce the number of medication‐related problems^m (low certainty^j).
Behavioural outcome: physical activity levels, 12 weeks IVR versus enhanced usual care	See comment	See comment	253 (1 study)	⊕⊕⊝⊝  Low^c	IVR versus enhanced usual care Dedier 2014 (N = 253) reported that IVR may slightly increase physical activity levels.
Adverse outcome: unintended adverse events attributable to the intervention Multimodal/complex intervention, ATCS Plus, IVR, unidirectional ATCS versus various controls	No studies reported adverse events.
ATCS Plus: automated telephone communication systems with additional functions; CI: confidence interval; IVR: interactive voice response; MD: mean difference; SD: standard deviation.
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS versus control for smoking cessation
Patient or population: patients with tobacco dependence  Settings: various settings  Intervention: ATCS (multimodal/complex intervention, ATCS Plus, IVR) Comparison: usual care, control (no calls, 'placebo' (inactive) ATCS, self‐help intervention, stage‐matched manuals)
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	No of participants  (studies)	Quality of the evidence  (GRADE)	Comments^a
	Assumed risk	Corresponding risk
	Control	ATCS
Behavioural outcome: smoking abstinence Multimodal/complex intervention, ATCS Plus, IVR versus (various) controls or usual care at median follow‐up of 12 months	Study population^b		RR 1.2   (0.98 to 1.46)	2915  (7 studies)	⊕⊕⊝⊝  Low^d,e	ATCS Plus versus usual care 1 further study, Reid 2011 (N = 440), reported that ATCS Plus may improve smoking abstinence rates at 26 weeks, and this may be maintained at 52 weeks (low certainty evidence^f).
	201 per 1000	241 per 1000   (197 to 293)
	Moderate^c
	241 per 1000	289 per 1000   (236 to 352)
Behavioural outcome: medication use Multimodal/complex, ATCS Plus versus control (inactive IVR or self‐help booklet)	See comment	See comment	See comment	1127 (2 studies)	⊕⊕⊕⊝ Moderate^g	Multimodal/complex intervention versus control (self‐help booklet) Brendryen 2008 (N = 396) found that a multimodal/complex intervention probably has little or no effect on adherence to NRT (moderate certainty evidence). ATCS Plus versus control (inactive IVR) Regan 2011 (N = 731) found that ATCS Plus probably has little or no effect on medication use (moderate certainty evidence).
Behavioural outcome: support programme enrolment ATCS Plus versus control (inactive IVR)	See comment	See comment	See comment	521 (1 study)	⊕⊕⊝⊝  Low^h	Carlini 2012 found that ATCS Plus may improve re‐enrolment into a quit line support programme.
Adverse outcome: unintended adverse events attributable to the intervention Multimodal/complex intervention, ATCS Plus, IVR versus various controls	No studies were found that reported adverse events.
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ATCS Plus: automated telephone communication systems with additional functions; CI: confidence interval; IVR: interactive voice response; NRT: nicotine replacement therapy; OR: odds ratio; RR: risk ratio;
GRADE Working Group grades of evidence  High quality: further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: we are very uncertain about the estimate.

ATCS	automated telephone communication system
AUDIT	Alcohol Use Disorders Identification Test
BMI	body mass index
CBA	control before and after
HR	hazard ratio
HR‐QoL	health related‐quality of life
ITS	interrupted time series
IVR	interactive voice response
MD	mean difference
NS	not specified
OR	odds ratio
PE	percent effect
PIN	personal identification number
QoL	quality of life
r	Pearson's correlations coefficient
RCT	randomised controlled trial
RD	risk difference
RR	relative risk
SAMA	short‐acting muscarinic antagonist
SD	standard deviation
SEM	standard error of the mean
SF‐12/36	12/36 Item Short Form Survey
SMBG	self monitoring of blood glucose
SMD	standardised mean difference
TLC	telephone‐linked computer

Level (quality) of evidence	Important benefit or harm	Less important benefit or harm	No important benefit/harm or null effect
High	improves*	improves slightly	little or no difference in [outcome]
Moderate	probably improves	probably improves slightly	probably little or no difference in [outcome]
Low	may improve	may improve slightly	may have little or no difference in [outcome]**
Very low	We are uncertain whether [intervention] improves [outcome]
No events or rare events	Use comments in SoF table in a plainer language or summarise the results
No studies	No studies were found that looked at [outcome]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Breast cancer screening	4		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 Multimodal/complex interventions	2	462	Risk Ratio (M‐H, Random, 95% CI)	2.17 [1.55, 3.04]
1.2 IVR	2	2599	Risk Ratio (M‐H, Random, 95% CI)	1.05 [0.99, 1.11]
2 Colorectal cancer screening	7		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Multimodal/complex intervention	3	1013	Risk Ratio (M‐H, Random, 95% CI)	2.19 [1.88, 2.55]
2.2 IVR (shorter follow‐up)	2	16915	Risk Ratio (M‐H, Random, 95% CI)	1.36 [1.25, 1.48]
2.3 IVR (longer follow‐up)	2	21335	Risk Ratio (M‐H, Random, 95% CI)	1.01 [0.97, 1.05]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Glycated haemoglobin	7	1216	Mean Difference (IV, Random, 95% CI)	‐0.26 [‐0.50, ‐0.01]
2 Self‐monitoring of diabetic foot	2	498	Std. Mean Difference (IV, Random, 95% CI)	0.24 [0.06, 0.42]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Cardiac mortality	2	215	Risk Ratio (M‐H, Random, 95% CI)	0.60 [0.21, 1.67]
2 All‐cause mortality	3	2165	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.79, 1.28]

PERMALINK

Automated telephone communication systems for preventive healthcare and management of long‐term conditions

Pawel Posadzki

Nikolaos Mastellos

Rebecca Ryan

Laura H Gunn

Lambert M Felix

Yannis Pappas

Marie‐Pierre Gagnon

Steven A Julious

Liming Xiang

Brian Oldenburg

Josip Car

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Primary preventive healthcare

1.

2.

Management of long‐term conditions

ICT for primary prevention and management of long‐term conditions

Description of the intervention

How the intervention might work

3.

4.

Advantages of automated telephone communication systems

Disadvantages of automated telephone communication systems

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

5.

Types of outcome measures

Primary outcomes

1. Health behaviour outcomes (category)

2. Clinical outcomes (category)

Secondary outcomes

1. Process outcomes (category)

2. Cognitive outcomes (category)

3. Patient‐centred outcomes (category)

4. Adverse outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

6.

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

1. Results.

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Summary of findings tables

Results

Description of studies

2. Participants.

3. Intervention.

4. Intervention Complexity Assessment Tool for Systematic Reviews.

Results of the search

Included studies

7.

5. Effectiveness of ATCS.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Systolic blood pressure	3	65256	Mean Difference (IV, Random, 95% CI)	‐1.89 [‐2.12, ‐1.66]
2 Diastolic blood pressure	2	65056	Mean Difference (IV, Random, 95% CI)	0.02 [‐2.62, 2.66]

Methods	Aim: to determine whether use of Personal Health Partner (PHP) was associated with significant differences in parental report of primary care visit content. Additional goals included evaluating the intervention effect on medication management, asthma care, and parent and clinician satisfaction Study design: RCT; recruitment: primary care (mail) Study duration: 25 months; study type: prevention; subtype: adherence to medication/laboratory tests
Participants	Inclusion criteria: children aged 4 months to 11 years (and their parents) who had a routine healthcare maintenance or well‐child visit. Parents and children had to speak English and could not be planning to move away from the Boston area within 3 months Sample size: 475; mean age: 5 years (child) 35 years (parent);sex: women ‐ 48% (child), 93% (parent); men ‐ 52% (child) 7% (parent); ethnicity: African‐American^a 67% (child); 47% (parent); other 33% (child); 53% (parent) Country: USA
Interventions	Personal Health Partner (PHP) tailors call content based on the participant's age and prescription of asthma medication. Call content was based on American Academy of Pediatrics Bright Futures topics reflected in the electronic health record (EHR) templates at the study site as well as Medicaid‐recommended health risk questions for routine healthcare maintenance (RHCM), asthma symptoms, and medication safety. When available, PHP scripts were based on validated tools. RHCM areas include general health supervision, developmental screening, diet and physical activity, tuberculosis risk assessment, smoking risk assessment, and maternal depression screening. Each call also addressed medication safety, examining what medications on the EHR medication list the child was actually taking, age‐appropriate medication use, and proper use of asthma controller and reliever medication if applicable. The day before each scheduled visit, PHP data were transferred to the EHR. PHP questions yielding actionable data generate an "Alert" displayed within the "Alerts" section of the "Patient Entered Data Review" form. Control group completed a single automated call, but the content was limited to the 18‐question Framingham Safety Survey. At the completion of the call, parents in the control group received tailored advice related to unsafe behaviours reported during the call. Because the Framingham Safety Survey was not part of routine primary care at Boston Medical Center, data from these calls were not shared with the EHR.
Outcomes	Comprehensiveness of screening and counselling (primary), assessment of medications and their management, and parent and clinician satisfaction (secondary)
Funding	Agency for Healthcare Research and Quality, grant R18HS017248
Declaration of conflict of interest	No potential conflicts of interest disclosed
Power calculations for sample size	No
Notes	The authors have been contacted for results from the Medication Adherence Scale with no response
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Children were randomly assigned to groups at the start of each call". Comment: insufficient information to judge whether random sequence generation was ensured
Allocation concealment (selection bias)	Unclear risk	Insufficient information
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "Study staff members were not aware of allocation group at the time of interviews"
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Insufficient information
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Insufficient information
Selective reporting (reporting bias)	Unclear risk	No results from the Medication Adherence Scale have been reported. Comment: insufficient information to judge whether this introduced bias
Other bias	Unclear risk	Insufficient information

Methods	Aims: to compare motivational interviewing (MI) HealthCall to MI‐only to reduce non‐injection drug use (NIDU) in urban HIV primary care patients Study design: RCT; recruitment: primary care (health professional referral) Study duration: 2 months; study type: prevention; subtype: substance abuse
Participants	Inclusion criteria: HIV‐positive, English‐ or Spanish‐speaking, aged 18 years, enrolled in a New York City hospital‐affiliated HIV primary care clinic, using drugs ≥ 4 days during the prior 30 days (including illicit non‐injection drugs or prescription drugs taken without prescription or more than prescribed) Sample size: 33; mean age: 46 years; sex: men ‐ 76%; women ‐ 24%; ethnicity: African American 64%, Hispanic 21%, Caucasian (understood to be white) 15% Country: USA
Interventions	MI + HealthCall: participants call HealthCall daily via a toll‐free number to report on the targeted health behaviour and potentially related moods, behaviours, and situations that occurred in the prior 24 h. HealthCall menu for NIDU included a short set of prerecorded questions in English or Spanish about the previous day covering use of primary drug, dollar amount spent on the drug used, use of other drugs, HIV medication adherence, and feelings of wellness, stress, and overall quality of the day. Participants responded by pressing numbers on the telephone keypad. After the practice call, counsellors helped participants identify an accessible telephone and convenient time for daily calls and set the watch alarm to this time as a reminder to call. Counselors were bilingual (English/Spanish) and from the same race/ethnic groups as most of the participants. HealthCall data were automatically uploaded to a database and used to provide personalised feedback to participants about their drug use in a single‐page form that included a computer‐generated graph of participants' drug use as called into the IVR and a set of summary statistics during the 30‐ and 60‐day visits. The personalised graph contained the participant's goal set in the baseline MI interview with the counsellor (NIDU Goal), with diamond‐shaped dots representing the dollar amount of drugs used on the days that the participant called HealthCall Participants in MI‐only arm (control) received a 20–25 min MI at baseline, using standard MI techniques, e.g. dialogue about health consequences of NIDU, exploring ambivalence, barriers to change, developing a change plan, including (for those who chose) a specific NIDU‐reduction goal (reflected in USD amounts) for the next 30 days. Participants then received a digital alarm watch which they were told they could use as a medication reminder. At 30 and 60 days, counsellor and participant met for 10–15 min to review overall drug use and set or re‐set a drug reduction goal for the next 30 days.
Outcomes	Days used primary drug in last 30 days (primary); patient satisfaction (secondary)
Funding	NA
Declaration of conflict of interest	NA
Power calculations for sample size	NA
Notes	—
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "The randomisation was done via 10‐block standard ABAB design"
Allocation concealment (selection bias)	Low risk	Quote: "Patients were blinded to their random assignment until after the MI session"
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Counsellors were not blinded to their random assignment
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Insufficient information
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Quote: "Treatment groups did not differ on attrition (p = 0.10) and thus attrition is not likely to be a source of bias in our results."
Selective reporting (reporting bias)	Unclear risk	Insufficient information
Other bias	Low risk	No significant baseline differences

Methods	Aims: to study if there is a difference in effect between automated interventions delivered by IVR and over the Internet Study design: RCT; recruitment: other ‐ university (web‐based survey) Study duration: 6 weeks; study type: management; subtype: alcohol consumption
Participants	Inclusion criteria: Swedish university students having an AUDIT score above cutoff (8 and 6) Sample size: 1423; mean age: ; sex: ; ethnicity: * Country: Sweden
Interventions	Single IVR call of less than 500 words, one week after the baseline assessment, consisting of feedback on the baseline assessment and instructions on how to obtain a recommended Blood Alcohol Concentration (BAC) below 0.6‰ (0.06 percentages) Single Internet‐delivered intervention given one week after baseline Repeated IVR call Repeated Internet‐delivered intervention given 1 and 2 weeks after intervention No intervention (controls)
Outcomes	Alcohol Use Disorders Identification Test (primary)
Funding	Swedish National Institute of Public Health and Edwin Berger Foundation
Declaration of conflict of interest	NA
Power calculations for sample size	NA
Notes	In the present review we report a comparison between single IVR call and no intervention. Information from abstract only
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information
Allocation concealment (selection bias)	Unclear risk	Insufficient information
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Insufficient information
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Insufficient information
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Insufficient information
Selective reporting (reporting bias)	Unclear risk	Insufficient information
Other bias	Unclear risk	Insufficient information

Methods	Aims: to determine whether a multifaceted intervention increases adherence to annual faecal occult blood testing compared with usual care Study design: RCT; recruitment: community centre (organisation referral) Study duration: 12 months; study type: prevention; subtype: screening
Participants	Inclusion criteria: age 51 to 75 years; preferred language listed as English or Spanish; and a negative faecal occult blood testing result obtained between 1 March 2011, and 28 February 2012 Sample size: 450; mean age: 60 years; sex: men ‐ 28%; women ‐ 72%; ethnicity: Latino ‐ 87%, other ‐ 13% Country: USA
Interventions	The multimodal intervention group received (1) a mailed reminder letter, a free faecal immunochemical test with low‐literacy instructions, and a postage‐paid return envelope; (2) an automated telephone and text message reminding them that they were due for screening and that a faecal immunochemical test was being mailed to them; (3) an automated telephone and text reminder 2 weeks later for those who did not return the faecal immunochemical test; and (4) personal telephone outreach by a colorectal cancer screening navigator after 3 months in addition to UC which included computerised reminders, standing orders for medical assistants to give participants home faecal immunochemical tests, and clinician feedback on colorectal cancer screening rates. Usual care (control group) at participating health centres included computerised reminders, standing orders for medical assistants to give participants home faecal immunochemical tests, and clinician feedback on colorectal cancer screening rates.
Outcomes	Completion of faecal occult blood testing within 6 months of the date the participant was due for annual screening (primary) Costs (secondary)
Funding	Grant P01 HS021141‐ the Agency for Healthcare Research and Quality
Declaration of conflict of interest	None reported
Power calculations for sample size	To detect a 10% difference (45% vs 35%) with 80% power (2‐tailed alpha = 0.05), we would need 752 participants (376 in each arm).This is less than the 800 participants that we estimated will be eligible for the study
Notes	The estimated cost of the outreach intervention was USD 34.59 per participant
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number generator
Allocation concealment (selection bias)	Unclear risk	Insufficient information
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Investigators were blinded to the outcomes in the control group
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Quote: "Using only EHR data for outcome assessment is conceptually similar to blinded outcome assessment"
Incomplete outcome data (attrition bias)   All outcomes	Low risk	All 450 participants were included in the analyses
Selective reporting (reporting bias)	Low risk	Outcomes of interest reported
Other bias	Low risk	Groups were comparable at baseline

Methods	Aim: to test the effectiveness of a theory‐based IVR intervention to improve adherence to controller medications among adults with asthma Study design: RCT; recruitment: community (advert in newspaper) Study duration: 10 weeks; study type: management; subtype: adherence to medication/laboratory tests
Participants	Inclusion criteria: no significant disease or disorder (chronic health disorders, current substance abuse or dependence, mental retardation, or psychiatric disorder); and lack of participation in any other asthma‐related research or clinical trial Sample size: 50; mean age:42 years;sex: women ‐ 59%; men ‐ 41% Ethnicity: white ‐ 58%, African American ‐ 20%, Hispanic ‐ 18%, Asian ‐ 4% Country: USA
Interventions	In the IVR group, each participant received ≥ 2 calls separated by 1 month. Calls were programmed to reach out at several time points throughout the day and evening until the participant answered. If an answering machine was reached, a toll‐free number was provided, which the participant could use to call back. When a call connection was completed, the IVR call identified itself as coming from the Denver Interactive Asthma Learning System program and verified that the correct person had been called. Content of the call then included an explanation of how the call works followed by 3 questions inquiring whether during the previous week the participant had been awakened at night, had limited their activities, or had used their rescue inhaler more than twice because of asthma symptoms (symptom module). Participants who responded affirmatively to any of the 3 questions were told that daily use of their controller medication should help prevent such symptoms and were advised to discuss the symptoms with their physician. All participants also listened to a short module about the benefits of their asthma medication and were asked about whether they were filling and using their medication, with IVR responses tailored to specific participant responses (refill module). Finally, participants were informed about the Lung Line, a free telephone service staffed by nurses capable of answering most questions about asthma, and about the Colorado Quit Line, offering free telephone based tobacco cessation intervention (resources module) Participants in the control group received no calls.
Outcomes	Medication adherence (primary); Asthma Control Test, Asthma Quality of Life Questionnaire, Beliefs about Medications Questionnaire (secondary)
Funding	Astra Zeneca
Declaration of conflict of interest	None declared
Power calculations for sample size	Power and sample size calculations indicated that 25 participants in each group would provide 75% power to detect a group difference of 36%
Notes	—
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A randomisation table generated before study initiation determined group assignment by order of entry into the study"
Allocation concealment (selection bias)	Unclear risk	Insufficient information
Blinding of participants and personnel (performance bias)   All outcomes	Low risk	Quote: "The investigators remained blind to treatment until the final data set was completed"
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Insufficient information
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Insufficient information
Selective reporting (reporting bias)	Unclear risk	Insufficient information
Other bias	Low risk	Participants were comparable at baseline

Methods	Aim: to improve adherence in paediatric asthma Study design: RCT; recruitment: secondary care () Study duration: 24 months; study type: management; subtype*: adherence to medication/laboratory tests
Participants	Inclusion criteria: children, ages 3‐12 years, treated for persistent asthma at Kaiser Permanente of Colorado Sample size: 1187; mean age: *sex: * ethnicity: * Country: USA
Interventions	Parents in the IVR group received a call reminding them that inhaled corticosteroid fill was overdue, and assisted with automated mail order refills or transfer to a Kaiser Permanente of Colorado pharmacy or asthma nurse specialist. Telephone calls in this group pulled information from the electronic health record (EHR) enabling the automated call to provide personalised participant and medication information. Parents in the control group received usual care
Outcomes	Medication adherence (primary); utilisation of care (secondary)
Funding	NA
Declaration of conflict of interest	NA
Power calculations for sample size	NA
Notes	Information from abstract only
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information
Allocation concealment (selection bias)	Unclear risk	Insufficient information
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Insufficient information
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Insufficient information
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Insufficient information
Selective reporting (reporting bias)	Unclear risk	Insufficient information
Other bias	Unclear risk	Insufficient information

Methods	Aims: to evaluate the effectiveness of a behavioural intervention that emphasised weight loss and hypertension medication adherence among primary care patients in the community health centre setting Study design: RCT; recruitment: community centre (telephone) Study duration: 24 months; study type: prevention; subtype: weight management
Participants	Inclusion criteria: BMI 30‐50 kg/m² (and weighing < 181.4 kg (400 pounds)), undergoing treatment for hypertension, aged ≥ 21 years, and enrolled participant at one of the participating community health centres (CHC). Additionally, participants had to read and speak English or Spanish, provide informed consent, and be willing to change diet, physical activity, and weight Sample size: 365; mean age: 55 years; sex: men ‐ 31%; women ‐ 69%; ethnicity: non‐Hispanic black ‐ 71%, Hispanic ‐ 13%, non‐Hispanic white ‐ 4%, other – 12% Country: USA
Interventions	Be Fit, Be Well: participants can choose to use either the Internet or print + IVR as a mode of delivery of the intervention. In print + IVR condition, participants track their behavioural goals daily on a paper log and then enter this information weekly using the telephone keypad during their IVR telephone call. The goals are divided into 3 categories: dietary, physical activity, and lifestyle goals. For the first 13 weeks, participants work on 3 goals; for the rest of the intervention period, they work on 4 goals simultaneously. Participants pick new behaviour change goals every 13 weeks. 2 goals ("Walk 10,000 steps per day" and "Take your blood pressure medicine the right way every day") remain constant throughout the intervention period. Skill training materials, in print, provide instruction in behavioural strategies to facilitate achieving their behavioural goals. The also provide additional dietary, physical activity, and lifestyle goals that may need additional contextualisation. Participants monitor their behavioural goals over the telephone using IVR. After entering data on their behaviour, participants receive immediate feedback on their progress compared to the previous entry. Participants receive social support via telephone coaches administered by community health educators (CHE) and group support sessions. CHE call the participants once a month in the 1st year and then bimonthly in the following year, during which they discuss progress, barriers, strategies to overcome barriers,self‐monitoring, and social support. Each call lasts for 15‐20 min. Group sessions include an interactive skill training and a physical activity component. The intervention materials include information on community resources such as public parks, local walking groups, and farmers' markets that can aid participants in their behaviour change efforts. All participants receive a walking kit that includes a pedometer and maps of the local community with associated step counts.Participants receive a personalised, tailored behaviour change "prescription" (generated from the baseline data) with the doctor's signature included electronically. This "prescription" presents recommendations for making changes in the targeted risk behaviours, and lets patients know that their doctor considers these recommendations to be important to their health Participants in the control group received usual care (self‐help booklet)
Outcomes	Change in body weight and BMI (primary); change in blood pressure; medication adherence; adverse‐events (secondary)
Funding	National Heart, Lung, and Blood Institute; National Cancer Institute
Declaration of conflict of interest	NA
Power calculations for sample size	The trial was designed to provide 80% power to detect a mean weight change in 24 months of 2.75 kg in the intervention arm, assuming no weight change in usual care
Notes	All participants are diagnosed with hypertension. In addition, 36% are diagnosed with hypercholesterolaemia, and 20% with type 2 diabetes mellitus
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated allocations were performed, blocked by clinic and sex
Allocation concealment (selection bias)	Unclear risk	Insufficient information
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "The trial design precluded blinding either patients or interventionists to treatment assignment."
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Insufficient information
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Quote: "All 365 participants are included in the primary outcomes analysis, including 15 participants (4.1%) who had only a baseline assessment."
Selective reporting (reporting bias)	Unclear risk	Insufficient information
Other bias	Low risk	Both the groups were balanced in all other characteristics at baseline

Methods	Aims: to compare changes in weight and cardiometabolic risk during a 12‐month period among black women randomised to a primary care–based behavioural weight gain prevention intervention or to usual care Study design: RCT; recruitment: community centre (mail) Study duration: 18 months; study type: prevention; subtype: weight management
Participants	Inclusion criteria: aged 25‐44 years, BMI of 25‐34.9 kg/m², ≥ 1 visit to a Piedmont Health Center in the previous 24 months, North Carolina residency, and self‐reported English fluency Sample size: 194; mean age: 35 years; sex: women ‐ 100%; ethnicity: black ‐ 100% Country: USA
Interventions	The multimodal intervention (the Shape Program) contained 5 components: obesogenic behaviour change goals; self‐monitoring via IVR phone calls; tailored skills training materials; 12 interpersonal counselling calls; and a 12‐month YMCA membership Participants in the control group received usual care: study staff made no attempts to influence the medical treatment provided to those in the usual care arm. Every 6 months, we sent usual‐care participants newsletters that covered general wellness topics but did not discuss weight, nutrition, or physical activity
Outcomes	Change in body weight and BMI (primary); maintenance of change at 18 months; adverse‐events (secondary)
Funding	R01DK078798 from the National Institute for Diabetes and Digestive and Kidney Diseases; and K05CA124415 from the National Cancer Institute
Declaration of conflict of interest	None declared
Power calculations for sample size	This trial was designed to have 80% power to detect significant BMI differences of 1.03 kg/m² between treatment groups 12 months after baseline
Notes	6 serious adverse events were reported among participants in the intervention arm, including gynaecological surgery in 2 participants and knee replacement, breast abscess, musculoskeletal injury, and cancer diagnosis in 1 participant each; all participants except the one with the cancer diagnosis required hospitalisation. The authors of the study could not conclusively determine whether reported events resulted from study participation.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "After completing baseline assessments, research staff initiated a computer‐generated randomisation algorithm to allocate participants equally (1:1) across the two treatment arms (intervention and usual care); those in the intervention arm were further randomised to one of two interventionists."
Allocation concealment (selection bias)	Unclear risk	Insufficient information
Blinding of participants and personnel (performance bias)   All outcomes	High risk	Quote: "The study design precluded blinding patients and interventionists to treatment assignment."
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Insufficient information
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Missing data balanced in numbers across groups (low attrition). ITT analysis was used to include all participants who received the intervention or usual care in the analysis. ITT analyses were based on the mean difference in weight and BMI between treatment arms at 12 months after adjustment for health centre.
Selective reporting (reporting bias)	Unclear risk	Insufficient information
Other bias	Low risk	The groups were well‐balanced at baseline.

Methods	Aims: to assess the ability of automated reminders to improve adherence with once‐daily glaucoma medications Study design: RCT; recruitment: primary care () Study duration: 6 months; study type: management; subtype*: adherence to medication/laboratory tests
Participants	Inclusion criteria: patients non‐adherent with their medications after 3 months of electronic monitoring (prospective cohort study phase) Sample size: 70; mean age: 66 years; sex: men ‐ 49%, women ‐ 51%; ethnicity: African American ‐ 58%, European ‐ 32%, Asian ‐ 6%, Hispanic ‐ 3%, Middle Eastern ‐ 1% Country: USA
Interventions	Automated reminders (by telephone or text message) informed each participant in the intervention group that it was time to take his or her medication. The IVR system also allowed participants to reset the reminder and receive it again in 1 hour: "Hello, this is your automated reminder to take your eye drop. Press 1 if you have or are about to take your drop. If you are not able to take your eye drop right now and would like a second reminder in 1 hour, please press 2 now." Participants in the control group received usual care
Outcomes	Medication adherence
Funding	Microsoft BeWell Fund
Declaration of conflict of interest	None reported
Power calculations for sample size	No
Notes	Communication with the author: "there was only one person (1.42% of the sample) who specified SMS (text) reminders in the study, however"
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Study participants were then assigned to a control or intervention group using assignments randomised equally in blocks of 10 and placed in envelopes."
Allocation concealment (selection bias)	Low risk	Quote: "Study participants were then assigned to a control or intervention group using assignments randomised equally in blocks of 10 and placed in envelopes."
Blinding of participants and personnel (performance bias)   All outcomes	Unclear risk	Insufficient information
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Insufficient information
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Quote: "Large percentage of participants lost to and unavailable for follow‐up, however ITT analysis was used in addition to real efficacy approach"
Selective reporting (reporting bias)	Low risk	Relevant outcomes were reported
Other bias	High risk	Quote: "At baseline, there were statistically significant differences between the two groups with regard to age, educational level, and Mini‐Mental State Examination (MMSE) score"