Payment methods for healthcare providers working in outpatient healthcare settings

Liying Jia; Qingyue Meng; Anthony Scott; Beibei Yuan; Lu Zhang

doi:10.1002/14651858.CD011865.pub2

. 2021 Jan 20;2021(1):CD011865. doi: 10.1002/14651858.CD011865.pub2

Payment methods for healthcare providers working in outpatient healthcare settings

Liying Jia ^1,⁵, Qingyue Meng ², Anthony Scott ³, Beibei Yuan ^2,^✉, Lu Zhang ⁴

Editor: Cochrane Effective Practice and Organisation of Care Group

PMCID: PMC8094987 PMID: 33469932

Abstract

Background

Changes to the method of payment for healthcare providers, including pay‐for‐performance schemes, are increasingly being used by governments, health insurers, and employers to help align financial incentives with health system goals. In this review we focused on changes to the method and level of payment for all types of healthcare providers in outpatient healthcare settings. Outpatient healthcare settings, broadly defined as 'out of hospital' care including primary care, are important for health systems in reducing the use of more expensive hospital services.

Objectives

To assess the impact of different payment methods for healthcare providers working in outpatient healthcare settings on the quantity and quality of health service provision, patient outcomes, healthcare provider outcomes, cost of service provision, and adverse effects.

Search methods

We searched CENTRAL, MEDLINE, Embase (searched 5 March 2019), and several other databases. In addition, we searched clinical trials platforms, grey literature, screened reference lists of included studies, did a cited reference search for included studies, and contacted study authors to identify additional studies. We screened records from an updated search in August 2020, with any potentially relevant studies categorised as awaiting classification.

Selection criteria

Randomised trials, non‐randomised trials, controlled before‐after studies, interrupted time series, and repeated measures studies that compared different payment methods for healthcare providers working in outpatient care settings.

Data collection and analysis

We used standard methodological procedures expected by Cochrane. We conducted a structured synthesis. We first categorised the payment methods comparisons and outcomes, and then described the effects of different types of payment methods on different outcome categories. Where feasible, we used meta‐analysis to synthesise the effects of payment interventions under the same category. Where it was not possible to perform meta‐analysis, we have reported means/medians and full ranges of the available point estimates. We have reported the risk ratio (RR) for dichotomous outcomes and the relative difference (as per cent change or mean difference (MD)) for continuous outcomes.

Main results

We included 27 studies in the review: 12 randomised trials, 13 controlled before‐and‐after studies, one interrupted time series, and one repeated measure study. Most healthcare providers were primary care physicians. Most of the payment methods were implemented by health insurance schemes in high‐income countries, with only one study from a low‐ or middle‐income country. The included studies were categorised into four groups based on comparisons of different payment methods.

(1) Pay for performance (P4P) plus existing payment methods comparedwith existing payment methods for healthcare providers working in outpatient healthcare settings

P4P incentives probably improve child immunisation status (RR 1.27, 95% confidence interval (CI) 1.19 to 1.36; 3760 patients; moderate‐certainty evidence) and may slightly increase the number of patients who are asked more detailed questions on their disease by their pharmacist (MD 1.24, 95% CI 0.93 to 1.54; 454 patients; low‐certainty evidence). P4P may slightly improve primary care physicians' prescribing of guideline‐recommended antihypertensive medicines compared with an existing payment method (RR 1.07, 95% CI 1.02 to 1.12; 362 patients; low‐certainty evidence). We are uncertain about the effects of extra P4P incentives on mean blood pressure reduction for patients and costs for providing services compared with an existing payment method (very low‐certainty evidence). Outcomes related to workload or other health professional outcomes were not reported in the included studies. One randomised trial found that compared to the control group, the performance of incentivised professionals was not sustained after the P4P intervention had ended.

(2) Fee for service (FFS) comparedwith existing payment methods for healthcare providers working in outpatient healthcare settings

We are uncertain about the effect of FFS on the quantity of health services delivered (outpatient visits and hospitalisations), patient health outcomes, and total drugs cost compared to an existing payment method due to very low‐certainty evidence. The quality of service provision and health professional outcomes were not reported in the included studies. One randomised trial reported that physicians paid via FFS may see more well patients than salaried physicians (low‐certainty evidence), possibly implying that more unnecessary services were delivered through FFS.

(3) FFS mixed with existing payment methods compared with existing payment methods for healthcare providers working in outpatient healthcare settings

FFS mixed payment method may increase the quantity of health services provided compared with an existing payment method (RR 1.37, 95% CI 1.07 to 1.76; low‐certainty evidence). We are uncertain about the effect of FFS mixed payment on quality of services provided, patient health outcomes, and health professional outcomes compared with an existing payment method due to very low‐certainty evidence. Cost outcomes and adverse effects were not reported in the included studies.

(4) Enhanced FFS compared with FFS for healthcare providers working in outpatient healthcare settings

Enhanced FFS (higher FFS payment) probably increases child immunisation rates (RR 1.25, 95% CI 1.06 to 1.48; moderate‐certainty evidence). We are uncertain whether higher FFS payment results in more primary care visits and about the effect of enhanced FFS on the net expenditure per year on covered children with regular FFS (very low‐certainty evidence). Quality of service provision, patient outcomes, health professional outcomes, and adverse effects were not reported in the included studies.

Authors' conclusions

For healthcare providers working in outpatient healthcare settings, P4P or an increase in FFS payment level probably increases the quantity of health service provision (moderate‐certainty evidence), and P4P may slightly improve the quality of service provision for targeted conditions (low‐certainty evidence). The effects of changes in payment methods on health outcomes is uncertain due to very low‐certainty evidence. Information to explore the influence of specific payment method design features, such as the size of incentives and type of performance measures, was insufficient. Furthermore, due to limited and very low‐certainty evidence, it is uncertain if changing payment models without including additional funding for professionals would have similar effects.

There is a need for further well‐conducted research on payment methods for healthcare providers working in outpatient healthcare settings in low‐ and middle‐income countries; more studies comparing the impacts of different designs of the same payment method; and studies that consider the unintended consequences of payment interventions.

Plain language summary

Payment methods for healthcare providers in outpatient healthcare settings

What is the aim of this review?

The aim of this Cochrane Review was to assess the effect of different payment methods for healthcare providers working in outpatient healthcare settings. The review authors collected and analysed all relevant studies to answer this question and found 27 studies.

Key messages

This review suggests that different payment methods can affect healthcare provider behaviour in both positive and negative ways. For instance, whilst healthcare providers may be encouraged to provide more of specific services, they may also be encouraged to provide unnecessary services. Considerable gaps remain in the understanding of how payment of healthcare providers affects healthcare services, healthcare providers’ work morale and workload, and patient health.

What was studied in the review?

Healthcare providers may be paid in different ways. Different payment methods can encourage healthcare providers to give patients the treatment they need in the best and most cost‐efficient way, but they can also encourage healthcare providers to offer poor‐quality, expensive, and unnecessary care, and to avoid certain treatments or certain types of patients. Different payment methods can also influence healthcare providers’ work morale and workload. And they can cost more or less for the healthcare system.

The review authors searched for studies on the effects of different payment methods for healthcare providers working in outpatient care. Outpatient care is where patients get health care from healthcare providers outside of hospitals and where there is no need for a bed. Healthcare centres, family planning centres, and dental clinics are all examples of outpatient facilities.

The payment methods the review authors were interested in were as follows.

‐ Pay‐for‐performance: healthcare providers are paid for carrying out certain tasks or reaching certain targets.

‐ Fee‐for‐service: healthcare providers are paid for each service they provide to the patient.

‐ Salary: healthcare providers are paid based on the time they spend at work.

‐ Capitation: healthcare providers are paid according to how many patients they have.

‐ A mix of these different approaches.

What are the main results of the review?

The review authors found 27 relevant studies. Most of the studies looked at primary healthcare doctors in high‐income countries.

When pay‐for‐performance plus other payment methods (including capitation, salary, and fee‐for‐service) is compared to other payment methods: healthcare providers probably provide more of certain services, including immunisations. They may also provide better‐quality care, including how some medicines are used, but these improvements may be reduced when the pay‐for‐performance payments end. Effects on patient health may be mixed. We are uncertain about the effect on healthcare providers’ work morale or workload, or on cost, because the evidence is missing or of very low certainty.

When fee‐for‐service methods are compared to other payment methods (such as capitation or salary): healthcare providers paid by fee‐for‐service may provide more unnecessary services than those paid by salary. We are uncertain about the effect on the quality or quantity of care, patient health, healthcare providers’ work morale or workload, or cost because the evidence is missing or of very low certainty.

When fee‐for‐service mixed with other payment methods (including fee‐for‐service plus capitation and fee‐for‐service plus salary) are compared to other payment methods: healthcare providers may provide more of specific services. We are uncertain about the effect on the quality of care, patient health, healthcare providers’ work morale or workload, cost, or unintended effects because the evidence is missing or of very low certainty.

When fee‐for‐service methods using a higher fee are compared to fee‐for‐service methods using a lower fee: healthcare providers probably provide more of certain services, including immunisations. We are uncertain if the higher fee has an impact on cost because the evidence is of very low certainty. We are uncertain about the effect on the quality of care, patient health, healthcare providers’ work morale or workload, or unintended effects because the evidence is missing.

How up‐to‐date is this review?

The review authors searched for studies that had been published up to 5 March 2019.

Summary of findings

Background

Description of the condition

Health care is a labour‐intensive service industry, so healthcare provision is directly influenced by healthcare providers' behaviours in delivering services. Examining optimal payment methods for healthcare providers is a key issue in ensuring care is delivered in a cost‐effective way to patients. Previous Cochrane Reviews have focused largely on physician payment methods (Flodgren 2011; Giuffrida 1999; Gosden 2000; Scott 2011). This review focused on payment methods for all healthcare providers in stand‐alone outpatient healthcare settings. Outpatient healthcare settings can vary considerably across countries, and are defined in this review as settings outside of inpatient hospitals for care where a bed is not required. They could include the private or public offices or clinics of primary care physicians and other specialists and healthcare providers. These may be stand‐alone buildings or could be adjacent to hospitals. Examples are care provided in stand‐alone community healthcare centres, clinics, urgent care centres, family planning centres, dental clinics, and allied health care (e.g. physiotherapy), but would exclude surgical day‐only procedures and emergency departments. The defining characteristics of outpatient care are the inclusion of services that involve a consultation with patients by healthcare providers, where they provide advice, prescriptions, immunisations, simple diagnostic tests, and some simple minor procedures and treatments that can be given in an office setting. Outpatient care may be the first point of contact for patients, or patients may be referred by other healthcare providers to outpatient care for more specialised care. Outpatient can play an important role in gatekeeping and rationing access to more expensive hospital care, and so can help reduce health expenditures.

Description of the intervention

Individual healthcare providers' behaviours directly influence the type, quantity, quality, and access to health services provided to patients. Several studies have summarised the determinants of healthcare providers' behaviours in health services delivery (Rowe 2005; WHO 2006). The method by which healthcare providers are remunerated (the payment method) has been frequently used by governments, health insurers, and employers to influence the behaviours of healthcare providers to meet specific objectives (Fairbrother 1999; Langenbrunner 2004). Payment methods refer to the way in which funds are received by a healthcare provider and used a personal income. The focus is on payment methods that influence health professional's personal income, rather than funding models aimed at healthcare organisations, although the two may interact. The most commonly used payment systems to remunerate healthcare providers are salary, capitation, fee‐for‐service, pay for performance, and mixed or blended systems of payment.

Salary: healthcare providers are paid based on the time spent at work. Salaried healthcare providers are usually employees of healthcare organisations, with these organisations paid using other methods such as global budget, FFS, and capitation.
Capitation: healthcare providers are paid a prospective fixed payment per period of time for each individual enrolled with the health professional.
Fee for service (FFS): healthcare providers are reimbursed based on each specific service, procedure, or visit provided, such as consultations, x‐ray tests, or surgical operations.
Pay for performance (P4P): the payment is directly linked to the achievement of specific behaviours defined in terms of the performance of healthcare providers. P4P can be used to pay individuals, groups of people, or organisations by government or insurers. Pay‐for‐performance schemes vary widely in terms of the types of performance that are targeted, how performance is measured, when payments for performance are paid, and the proportion of total reimbursements that is paid for performance (Witter 2012).
Blended or mixed payment: more than one of the above payment methods may be used at the same time, with a proportion of the health professional's income coming from each type of method.

How the intervention might work

The mechanism through which payment methods influence behaviour is through their effect on the income of healthcare providers. Earning income is a key driver of individual's consumption and leisure activities, and economic theory assumes that individuals are motivated only by monetary extrinsic rewards. Generally and in practice, extensions of this theory recognise that payment methods may have less of an impact if there are other sources of motivation from working, such as improving the health of patients. Payment to healthcare providers happens when funds are transferred from employers, patients, or insurers to individual healthcare providers in exchange for the provision of healthcare services. Each payment method may be part of a formal contract or agreement between the payor and payee, which may also specify working conditions, other in‐kind benefits, and detail about what is required in providing the services. Under each payment method, the expected total income must first be sufficient to encourage the health professional to work and enter into a contract. Payment methods in these contracts can be changed, or healthcare providers can move to a different job (contract) with a different payment method, raising the issue of selection bias when examining the effects of changes to payment methods.

Assuming income is sufficiently important to healthcare providers, the payment method will influence behaviour depending on the change to the unit of payment or a change to the level of payment for each unit. The unit of payment for those on salaries and who receive an hourly wage is hours worked, and income can be increased only by working more hours. The unit of payment in capitation payment is the number of enrolled patients, and so income can only be increased by increasing the number of enrolled patients. For those on FFS, income can rise only by increasing the number of services provided to existing or new patients. For those paid by P4P, income can only be increased by improvements in performance. In blended payment, a number of different units of payment are used, and so a range of behaviours can increase income.

Moving from one payment method to another, or changing the mix of payment methods in blended payment, will depend on the change to the unit of payment. With four different payment methods, moving from one to another defines many different combinations of comparators and interventions, each of which may have different effects on behaviour depending on the comparator and intervention units of payment. Under a new unit of payment, healthcare providers are assumed to adjust their work behaviours in ways that maximise their utility (including utility from the consumption of goods and services derived from their income, and utility from non‐work activities). A new payment method will provide the health professional with a different way to increase their income (and utility) depending on the unit of payment. healthcare providers may also reduce the costs of providing care to maximise their income, such as changes in skill mix, how they use their time (e.g. changes to consultation length), changes in effort, and changes to other practice costs (e.g. use of administrative staff). A change in payment methods could also lead healthcare providers to change jobs, which can influence patient access to care. The main hypotheses regarding the expected changes in behaviour from each payment method are outlined below, although the precise hypotheses will also depend on the existing (comparator) payment method, as well as the new payment method.

Salaried payment

Salaried payment focuses on the amount of time at work, usually in terms of hours. For a given number of hours worked, salaries are fixed in the short run, and do not provide any financial incentives for increased effort or increased quality of care or performance. In the longer run, financial incentives can encourage effort beyond some minimum to avoid being sacked and can also provide financial incentives to meet some level of performance to keep the job if on fixed‐term contract. In the longer run, salaried payment is also accompanied by different levels of salary (wages) for different levels of seniority (salary increments and promotion). Financial incentives for increased performance can be built into the salary scale (increments and promotion), driven by additional income from advancement up the salary scale. Progression up the salary scale may be automatic depending only on years of service, for example, or it may be based on a subjective measure of performance assessed in an annual performance review. This provides 'career' financial incentives to improve performance to obtain promotion and advancement.

Capitation

Payments are made per patient, and so capitation is normally accompanied by a system with patients registering/enrolling with a specific provider to receive care. Capitation is a prospective‐based payment system where healthcare providers can predict their income from each enrolled patient in advance. Capitation payment encourages healthcare providers to attract more patients, and so compete with other healthcare providers, such as by improving quality and access to care and those aspects of care that are valued by patients. With a fixed payment for each patient, the difference between the costs of treatment and the fixed payment defines the profit or income going to the health professional. More complex patients lead to higher costs and less profit. Capitation therefore includes an inbuilt incentive to minimise costs by selecting only the healthiest patients to enrol and treat, which can have adverse consequences for access to health care for those with more complex and costly conditions. Capitation may provide incentives to reduce costs by changing skill mix or providing short consultations that could be related to increased prescribing, fewer treatments provided, and more referrals compared to FFS and P4P. Though cost‐consciousness is important, it can also lead to lower quality and poor access if these are not separately monitored and rewarded (e.g. in a blended payment scheme that combines capitation and P4P). Capitation payments are usually 'risk‐adjusted', where higher payments are provided for patients with more complex/costly conditions. This reduces the incentives for patient selection (or 'cream skimming') and cutting costs that might reduce quality. It is also argued that capitation payment may provide incentives for healthcare providers to provide more preventive activities to patients so that they do not return in the future. This reduces future costs and increases provider income in the longer term, as patients make fewer visits.

Fee for service

This is likely to increase the number of services provided by healthcare providers compared to other payment methods, either through providing more services to each patient or by attracting more patients. FFS could lead to increased and less predictable health expenditures compared to other forms of payment depending on how fees are determined, as well as the overprovision of unnecessary services, including overdiagnosis. FFS also provides incentives for healthcare providers to provide treatment themselves rather than refer to others, especially compared with salaried or capitation payment.

Pay for performance

P4P payment is directly linked to performance targets that are often related to the type, mix, and quality of care delivered. P4P designs are complex and vary depending on how performance is measured, the level of the performance target, and many other components. Performance is usually defined in terms of the quality of care provided, which might include whether certain activities were performed or not (e.g. taking blood pressure) or the outcome of that activity (e.g. whether the measure of blood pressure is within accepted clinical guidelines). Payment may be made for each extra activity (e.g. each patient who immunised), or as a lump‐sum bonus for achieving a prespecified target. Financial penalties may also be used if targets are not met. The performance target for payment may be absolute (e.g. 80% of a target population being immunised) or relative to other healthcare providers (e.g. whether the provider is in the top quartile of immunisation rates across all providers), or based on the absolute (or relative) change in performance from one period to the next (Ogundeji 2016; Van Herck 2010). In practice, P4P rarely exists on its own, and is usually part of a blended payment method.

Blended payments

Blended payments reduce the 'extremes' of single payment methods by providing a range of methods through which providers can increase their income, and can combine incentives to be cost‐conscious as well as maintain and improve quality (value‐based payment).

In addition, within each payment method the level of payment for each unit of payment (hours, patients, services, quality) may be decreased or increased. This may encourage more of the rewarded activity if the additional (marginal) income is greater than the additional (marginal) costs of increasing the activity, and also assuming that income is sufficiently important to the health professional compared to other sources of utility. The importance of income (monetary motivation) will likely vary across different types of healthcare providers leading to heterogenous effects of changes to payment methods.

Why it is important to do this review

Payment methods can change the behaviours of healthcare providers through financial incentives, and then influence number, mix, cost, and quality of care provided. The design of appropriate payment systems is a key issue for governments, health insurers, health organisation managers, and all relevant policymakers who expect the efficient use of limited funds in the health system.

This review focused on payment methods for individual healthcare providers. Other Cochrane Reviews examine payment methods for doctors across all types of setting (Flodgren 2011; Giuffrida 1999; Gosden 2000; Scott 2011), whilst this review includes all kinds of healthcare providers only in outpatient healthcare settings. Together with the Cochrane Reviews on payment methods for outpatient facilities (rather than individual healthcare providers) (Yuan 2017) and hospitals (Mathes 2014; Mathes 2019), this review contributes to evidence to encourage healthcare providers to provide high‐quality, efficient, and equitable health care to their patients.

Objectives

To assess the impact of different payment methods for healthcare providers working in outpatient healthcare settings on quantity and quality of health service provision, patient outcomes, healthcare provider outcomes, cost of service provision, and adverse effects.

Methods

Criteria for considering studies for this review

Types of studies

Randomised trials, including cluster‐randomised trials
Non‐randomised trials
Interrupted time series (ITS) and repeated measures studies with:
- a clearly defined point in time when the intervention occurred;
- at least three data points before and three data points after the intervention.
Controlled before‐after (CBA) studies with:
- contemporaneous data collection;
- a minimum of two intervention and two control sites.

Types of participants

Healthcare providers working in outpatient care facilities. Healthcare providers include, for example, primary care physicians and other non‐surgical specialists, dentists, midwives, nurses, or allied health (WHO 2006). Healthcare providers working in surgical day‐only procedures and emergency departments were excluded.

Types of interventions

The payment method is defined as the mechanism used to remunerate individual healthcare providers. Payment methods for healthcare providers include:

salary;
fee‐for‐service (FFS): healthcare providers are reimbursed based on specific items provided;
capitation: healthcare providers are paid a predetermined fixed rate in advance to provide a defined set of services for each enrolled individual for a fixed period;
pay for performance (P4P): the payment is directly linked to the performance of healthcare providers;
blended payments.

We included studies that evaluated changes from one type of payment method to another, changes to the design of a payment method, or changes to the level of payment. Any of these may change the level of provider income, or create different opportunities for the health professional to increase their income and change the care they provide.

We excluded studies of interventions that were primarily targeted at paying at practice or organisational level; in this review, we focused only on changes to payments made directly to healthcare providers. Another Cochrane Review has evaluated the payment at practice or organisational level (Yuan 2017). For example, the QOF (Quality and Outcomes Framework) applied in the UK is a pay‐for‐performance scheme delivering funding to general practices, not directly to general practitioners, and so studies evaluating the QOF were excluded from this review.

Types of outcome measures

Primary outcomes

The main behavior of healthcare providers is to provide health care to patients, so the primary outcomes included in this review are objective measures of health services provision and other measures which are closely related to the supply of health services in outpatient care facilities, including the following.

Service provision and process outcomes

- Quantity of health services provided, including those:
  - measured as risk ratio (RR), e.g. for patients getting an aspirin prescription, for referring smokers to a quit line, for women having any prenatal care;
  - measured as mean difference (MD) or standardised mean difference (SMD), e.g. rate of screening service per 100 prescriptions, average length of service time.
- Quality of health services provided, including those:
  - measured as RR, e.g. for physicians adhering to clinical guidelines;
  - measured as MD or SMD, e.g. the quality score for provision of certain health services.

Patient outcomes (the effect of changes in service provision or process outcomes on measures of length of life, quality of life, or clinical measures closely related to these outcomes)
- Patients' intermediate and final health outcomes, including those:
  - measured as RR, e.g. if a smoker has sustained abstinence from smoking, if blood pressure has been controlled;
  - measured as MD or SMD, e.g. blood pressure level of patients with hypertension, health‐related quality of life, mortality.
Healthcare provider outcomes (the outcomes related to consequences on individual providers after delivering services), including those:
- measured as RR, e.g. if individual professionals are satisfied with work or job turnover;
- measured as MD or SMD, e.g. hours worked or the income level of healthcare providers.
Costs of delivering services, including those:
- measured as MD or SMD, e.g. cost per service, administration costs, total cost for purchasers, changes in skill mix.
Unintended or adverse effects, including those:
- measured as RR, e.g. patient selection and poorer access to care for disadvantaged populations;
- measured as MD or SMD, e.g. the number of unnecessary services (overtreatment and overdiagnosis).

Secondary outcomes

Satisfaction of patients or other stakeholders; job mobility.

Search methods for identification of studies

Electronic searches

We searched the following databases.

Cochrane Database of Systematic Reviews, part of the Cochrane Library (www.cochranelibrary.com/) (searched 5 March 2019)
Database of Abstracts of Reviews of Effects, part of the Cochrane Library (www.cochranelibrary.com/) (searched 15 July 2017)
Cochrane Central Register of Controlled Trials (CENTRAL2019, Issue 3, part of the Cochrane Library (www.cochranelibrary.com/) (searched 5 March 2019)
MEDLINE and Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations and Daily 1946 to March 04, 2019, Ovid (searched 5 March 2019)
Embase 1974 to 2019 March 04, Ovid (searched 5 March 2019)
Web of Science, Conference Proceedings Citation Index‐Science, 1990 to present (ISI Web of Knowledge) (searched 5 March 2019)
PubMed, NLM (searched 10 December 2018)
Dissertations and Theses Database, 1861 to present, ProQuest (searched 10 December 2018)
EconLit, 1969 to present, ProQuest (searched 10 December 2018)
China National Knowledge Infrastructure (CHKD‐CNKI), 1915 to present (searched 10 December 2018)
Chinese Medicine Premier (Wanfang Data), 1988 to present (searched 10 December 2018)
IDEAS (Research Papers in Economics), 1927 to present (searched 30 December 2017)
POPLINE (Population Information Online), 1970 to present, K4Health (searched 30 December 2017)

The EPOC Information Specialist (TSC) helped develop some of the search strategies in consultation with the review authors.

Search strategies are comprised of keywords and controlled vocabulary terms. We applied no language limits. We searched all databases from database start date to date of search.

Searching other resources

Grey literature

We conducted a grey literature search to identify studies not indexed in the databases listed above.

OpenGrey (System for Information on Grey Literature in Europe) (www.opengrey.eu/) (searched 10 December 2018)
World Health Organization (WHO) (www.who.int/) (searched 17 November 2018)
World Bank (www.who.int/) /searched 17 November 2018)

Trial registries

US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (clinicaltrials.gov/) (searched 27 June 2019)
World Health Organization International Clinical Trials Registry Platform (ICTRP) (www.who.int/clinical-trials-registry-platform) (searched 27 June 2019)

In addition, we:

searched reference lists of all relevant papers identified;
searched Web of Science Core collection; KCI‐Korean Journal Database; Russian Science Citation Index; SciELO Citation Index, Clarivate Analytics for papers that cited any of the included studies in this review (searched 8 February 2019);
searched PubMed for related citations to any studies to be included in the review;
contacted authors of relevant papers regarding any further published or unpublished work;

We re‐ran the search strategies in August 2020 and screened the identified records. Potentially relevant studies are awaiting classification and will be assessed at the next update.

All search strategies are provided in Appendix 1.

Data collection and analysis

Selection of studies

Two review authors scanned the titles and abstracts of all articles obtained from the search and retrieved the full text of articles deemed relevant. Two review authors independently assessed full texts of studies for inclusion. Any disagreements on inclusion were resolved by discussion with a third review author or EPOC editor.

The screening process and results are reported in a PRISMA flow diagram (Figure 1). All included studies are described in the Characteristics of included studies table, even those for which useable results for reanalysis or synthesis were not reported. Studies that initially appeared to meet the inclusion criteria but that were excluded are described in the Characteristics of excluded studies table.

Data extraction and management

Two review authors independently carried out data extraction using a data extraction form adopted from the Cochrane good practice data collection form (EPOC 2013a). We extracted the following information.

General information, including title, reference details, author contact details, and publication type.
Participants and setting.
Study method.
Intervention groups, including payment method description, duration of intervention, if patients can choose providers, how purchasers monitored the implementation of payment.
Outcomes, including outcome measures, time points measured, unit of measurement, and person measuring outcomes.
Results, including result reported by authors, analysis method, unintended effects, if analysis required and possibility.

Any disagreements were resolved by discussion with a third review author or the EPOC contact editor. For ITS studies that reported time series data that were not appropriately analysed, we extracted and reanalysed the data as described in the EPOC resources for review authors (EPOC 2013b).

Assessment of risk of bias in included studies

We used the EPOC suggested 'Risk of bias' criteria to assess the risk of bias for each outcome in all included studies (EPOC 2013c). For each criterion, two review authors independently described what was reported in the study, commented on the description, and judged the risk of bias. Any unresolved disagreements were discussed with a third review author and, if consensus could not be reached, with the EPOC contact editor. We summarised the overall risk of bias across criteria for the outcomes of the included studies. For randomised trials, non‐randomised trials, and controlled before‐after studies, we primarily considered four criteria: baseline outcome measurements; baseline characteristics measurements; incomplete outcome data addressed; and protection against contamination. If these four criteria were all scored 'low risk of bias' for the outcome in a given study, the summary assessment was low risk of bias; if one or more key criteria were scored 'unclear', the summary assessment was unclear risk of bias; and if one or more key criteria were scored 'high risk of bias', the summary assessment was high risk of bias. For ITS studies, we primarily considered the following criteria when summarising the overall risk of bias: intervention independence, intervention affecting data collection, and incomplete outcome data addressed.

Measures of treatment effect

For randomised trials, non‐randomised trials, and controlled before‐after studies, we recorded or calculated risk ratio (RR) with 95% confidence intervals (CI) for dichotomous outcomes. If adjusted analysis was done, we reported the effect estimates provided by the study authors, also converting them into RR if possible. For continuous outcomes, we recorded or calculated mean difference (MD) with 95% CI if the studies to be synthesised had the same outcome measures. We calculated the standardised mean difference (SMD) with 95% CI if the studies to be synthesised assessed slightly different outcome measures within the same broad outcome category.

Data were insufficient data to permit statistical pooling for most outcomes. In order to facilitate comparison of the effect sizes of the included studies, we did the following: if the baseline levels were available, we reported the point estimates of absolute change adjusted for baseline differences or the relative change adjusted for baseline differences for all included outcome measures. If the baseline levels were not available, we reported the point estimates of absolute change and relative change for all included outcome measures.

The absolute change adjusted for baseline differences is defined as 'baseline‐post difference', and the formula is: (intervention group post level − intervention group baseline level) − (control group post level − control group baseline level). The relative change adjusted for baseline differences is calculated by the following formula: ((intervention group post level − intervention group baseline level) − (control group post level − control group baseline level))/(control group post level − control group baseline level).

For ITS and repeated measures studies, we reported the difference between the predicted value based on the pre‐intervention trend and the estimated value based on the change in level and post‐intervention trend at relevant time points (including immediately after the intervention (change in level), one year, two years, and three years).

Unit of analysis issues

This review analysed the impact on performance on the level of the individual, so the allocation and analysis unit should be aggregated to physicians. We planned to reanalyse studies that allocated clusters (e.g. clinics in one district) but that did not account for clustering if we were able to extract the intracluster coefficient. All included studies reported having accounted for and adjusting for clustering in their analysis.

Dealing with missing data

If any data needed for meta‐analysis or reanalysis (e.g. standard deviations, numbers of events, and subgroup analyses) were missing, we attempted to calculate them based on available data on the same outcome (e.g. calculation of standard deviation from CI). If these data were not available, we contacted the study authors to request the missing data. However, we did not receive responses from the relevant study authors and therefore reported the data that were available. For these studies, we reported the point estimate of the effect measures without CIs. Where data on subgroup analyses were missing, we were not able to include these subgroups in our analysis.

Assessment of heterogeneity

We conducted meta‐analysis to synthesise the effect measures of included studies if they met the following criteria.

Similar intervention payment method and comparison payment, based on the payment method categories described above.
Similar participants, e.g. the targets of the payment method were all primary care physicians.
The outcome measures fell into the same outcome category, e.g. the outcome measures were all health service provision process measures, or patient outcomes measures.

When the included studies were sufficiently similar based on the above criteria, we used the Chi² test and I² statistic to assess statistical heterogeneity. When the P value from a Chi² test was smaller than 0.1, we interpreted this as an indication that the observed difference in results across studies was not compatible with chance alone. We used the I² statistic to quantify the level of statistical heterogeneity.

As there is considerable heterogeneity in the design of payment methods, we attempted to explore this through the prespecified subgroup analyses (see Subgroup analysis and investigation of heterogeneity). We used caution in interpreting results from meta‐analyses with high levels of heterogeneity.

Assessment of reporting biases

We planned to use a funnel plot to examine asymmetry and assess the potential of any asymmetry being due to publication bias. However, due to the limited number of studies with the same outcomes in each comparison, we decided not to undertake funnel plots.

Data synthesis

We conducted a structured synthesis as described in the EPOC resources for review authors (EPOC 2013d). We first categorised the comparisons and outcomes, and then described the effects of different kinds of payment methods on different categories of outcomes. As mentioned above, there were two kinds of payment change with mechanisms for changing behaviour: additional incentives leading to an increase in total funding, or changes in the payment model without changes to the total funding received. These two mechanisms were assessed in different comparisons.

We synthesised the effects of studies that used the same type of study design. We conducted meta‐analysis for randomised trials, but not for controlled before‐after and interrupted time series studies due to insufficient data. For the meta‐analyses, we initially used a fixed‐effect model to pool data within a study if the study included more than one outcome indicator under the same category of outcome measures. We then used a random‐effects model for meta‐analysis across studies. This was because the payment method designs usually included several components; the payment methods conducted by different purchasers or in different areas were rarely exactly the same; and there were differences in outcome measures falling in the same outcome category. For randomised trials where the available data did not permit meta‐analysis, and for all controlled before‐after and interrupted time series studies, we reported the medians and full ranges of the available point estimates of effect sizes. We analysed these firstly within, and then secondly across, the studies for the same category of outcome measures. It was not possible to report an interquartile range due to the very limited number of data points.

Subgroup analysis and investigation of heterogeneity

In the protocol for this review, we hypothesised a series of factors that might affect the size of effects of payment methods and planned to perform subgroup analysis based on these factors, such as size of fee, duration of follow‐up, targeted population, if there were multiple providers, if there was monitoring of delivery of services, the frequency of monitoring, the frequency of payment, and baseline performance level. However, these subgroup analyses could not be conducted due to the limited number of included studies with similar comparisons and outcomes. Furthermore, the included studies reported limited detail on the characteristics of payment methods. 'Duration of follow‐up' and 'targeted population' were reported more often in the included studies, and these factors are presented in Table 5 and Table 6.

1. Outcome measures of included studies.

Study	Outcomes	Length of observation
Quantity of health services provided
Bilardi 2010	Women tested	12 months
Christensen 2000	Documentation rate Frequency of cognitive services (CS) Frequency of CS interventions Primary results of CS interventions CS intervention rates per 100 prescriptions Patients receiving CS, by problem type Pharmacists self‐reported time per CS	20 months
Clarkson 2008	Children with 1 or more sealant per dentist	18 months
Davidson 1992	Proportion of primary care physician visits Proportion of non‐primary care physician visits Proportion of hospitalisations	18 months
Fairbrother 1999	UTD immunisation coverage (% of patients) for diphtheria, tetanus, pertussis, and haemophilus influenzae type b; polio Percentage of visits in the 4 months prior to review that were missed opportunities to immunise Percentage of immunisations received outside the participating practice amongst children with 1 or more immunisations	12 months
Fairbrother 2001	UTD levels calculated from data from all sources for Diphtheria, Tetanus, Pertussis (DTP); Haemophilus Influenzae Type b (Hib); polio; and Measles, Mumps, and Rubella (MMR) vaccines; Average number of UTD coverage for DTP, Hib, polio vaccine, and MMR; Percentage of well visits in the 4 months before review that were missed opportunities to immunise Percentage of sick visits in the 4 months before review that were missed opportunities to immunise	12 months
Flierman 1992	Changes in numbers of services per week per 1000 registered patients performed	12 months
Gleeson 2017	The number of incentivised services provision (2 well care, 2 asthma, and 10 immunisation) The number of unincentivised measures (2 acute illness, 2 attention‐deficit/hyperactivity disorder, 2 immunisation, and a screening)	12 months
Gosden 2003	The proportion in which a prescription was given or a referral made Changes in cervical cytology, child immunisation, and pre‐school booster rates	18 months
Greene 2013	The number of diabetes test provision The number of cervical cancer screening provision	15 years
Hickson 1987	The number of visits in excess of the recommended	24 months
Kouides 1998	Influenza vaccination rate	12 months
Krasnik 1990	Number of face‐to‐face consultations per 1000 enlisted patients Number of consultations by telephone per 1000 enlisted patients Number of diagnostic services per 1000 enlisted patients Number of curative services per 1000 enlisted patients	9 months
Lee 2010	Number of essential exams/tests Number of diabetes‐related physician visits Number of diabetes‐related hospitalisations	24 months
Li 2013	Senior flu shots rate Toddler immunisation rate Pap smears rate Mammograms rate Colorectal cancer screenings rate	120 months
Lurie 1992	The number of referrals to specialists	24 months
Twardella 2007	The number of patients getting prescription of smoking cessation drugs	12 months
Yesalis 1980	Percentage of Medicaid prescriptions classified as multi‐source drug products Numbers of prescriptions involving changes in labeller Numbers of prescriptions involving changes in labeller on refills	36 months
Yesalis 1984	Average days' therapy per recipient‐month in control and experimental counties over time (institutional) Average days' therapy per recipient‐month in control and experimental counties over time (non‐institutional)	36 months
Young 2007	Mean rate of glycated haemoglobin check Mean rate of urinalysis check Mean rate of LDL check Mean rate of eye exams	‐
*Quality of health services provided*
Chung 2010	Mean rate of LDL check	24 months
Gosden 2003	Mean rate of eye exams	18 months
Petersen 2013	Use of guideline‐recommended antihypertensive medicines	12 months
Young 2007	4 diabetes performance measures: glycated haemoglobin check, urinalysis, LDL check, eye exam	‐
Young 2012	Performance scores of eye examination Performance scores of glycated haemoglobin tests Performance scores of lipoprotein density level screening Performance scores of nephropathy test	36 months
*Patient outcomes*
Gleeson 2017	Proportion of asthma, 12 to 18 years proportion of asthma, 5 to 11 years Proportion of pharyngitis Proportion of upper respiratory infection Proportion of influenza Proportion of rotavirus	‐
Houle 2016	Mean systolic blood pressure reduction Mean diastolic blood pressure reduction	45 months
Jensen 2014	Proportion of low birth weight Proportion of preterm birth Proportion of very preterm birth The average birth weight Mean fetal growth rate	21 months
Lurie 1992	General health status Index (4 to 16) Physical functioning index (0 to 9) Social contact index (0 to 120)	11 months
Petersen 2013	Proportion of blood pressure control or appropriate response to uncontrolled blood pressure of physicians' patients Proportion of use of guideline‐recommended antihypertensive medications of physicians’ patients	16 months
Singh 2015	Mean weight of the child Mean grade (as measured in the Anganwadi) of the child Mean z‐score of the child Malnourished status of the child according to WHO	12 months
Twardella 2007	Self‐reported smoking abstinence obtained at 12 months follow‐up and validated by serum cotinine	12 months
*Healthcare provider outcomes*
Gosden 2003	Working times	18 months
Gray 2015	Workload Annual incomes	72 months
*Costs*
Davidson 1992	Expenditures	18 months
Lee 2010	Expenses for diabetes‐related physician visits	24 months
Yesalis 1980	Cost saving from generic substitution	36 months
Yesalis 1984	Cost saving from generic substitution	36 months
*Adverse effects*
Hickson 1987	The number of well patients	24 months
Petersen 2013	Reduction in blood pressure control and appropriate response to uncontrolled blood pressure after end of the intervention	12 months

Target population	Payment methods	Outcomes
The elderly	Not reported in included studies	Not reported in included studies
The disabled	Not reported in included studies	Not reported in included studies
Minorities	Not reported in included studies	Not reported in included studies
People with low levels of education	Not reported in included studies	Not reported in included studies
Children
Clarkson 2008, cluster‐randomised trial	FFS remuneration	Children with 1 or more sealant per dentist
Davidson 1992, randomised trial	Capitation, FFS high rate compare with FFS (low rates)	Physician visits, hospitalisations
Jensen 2014, controlled before‐after study	Mixed system of capitation and FFS contracts	Birth weight, preterm birth, very preterm birth, fetal growth
Singh 2015, controlled before‐after study	Performance bonus	Weight, WHO malnourished status
Women
Bilardi 2010, cluster randomised trial	P4P (a small incentive payment per test)	Women being tested
People living in rural or remote areas
Yesalis 1980; Yesalis 1984, controlled before‐after study	Capitation compare with FFS	Rate of generic substitution per 100 prescriptions Percentage of Medicaid prescriptions classified as multi‐source drug products Numbers of prescriptions involving changes in labeller on refills (0 to 5 days)
Low‐income populations 2
Christensen 2000, randomised trial, Medicaid recipients	Financial incentive (P4P)	Patients receiving cognitive services
Gleeson 2017, controlled before‐after study, Medicaid	P4P plus existing FFS compare with FFS	Adolescent well care, inactivated polio vaccine

Comparison 1: P4P plus an existing payment method compared withan existing payment method
Study	Intervention	Comparison
Christensen 2000	P4P + existing payment method	Capitation
Chung 2010	P4P + existing payment method	Not described
Fairbrother 1999	P4P + existing payment method	FFS
Fairbrother 2001	P4P + existing payment method	FFS
Gleeson 2017	P4P + existing payment method	FFS
Greene 2013	P4P + existing payment method	FFS
Houle 2016	P4P + existing payment method	FFS
Kouides 1998	P4P + existing payment method	FFS
Lee 2010	P4P + existing payment method	FFS
Li 2013	P4P + existing payment method	FFS
Petersen 2013	P4P + existing payment method	Not described
Singh 2015	P4P + existing payment method	Salary
Young 2007	P4P + existing payment method	Salary
Young 2012	P4P + existing payment method	Salary
Comparison 2: FFS compared withexisting payment methods
Study	Intervention	Comparison
Davidson 1992	FFS	Capitation
Lurie 1992	FFS	Capitation
Hickson 1987	FFS	Salary
Yesalis 1984	FFS	Capitation
Yesalis 1980	FFS	Capitation
Comparison 3: FFS mixed with existing payment methods compared withexisting payment methods
Study	Intervention	Comparison
Bilardi 2010	FFS + existing payment method	Not explicitly described
Clarkson 2008	FFS + capitation	Capitation
Twardella 2007	FFS + existing payment method	Not explicitly described
Jensen 2014	FFS + capitation	Capitation
Gosden 2003	FFS + salary	Salary
Gray 2015	FFS + capitation	FFS
Flierman 1992	FFS + capitation	Capitation
Krasnik 1990	FFS + capitation	Capitation
Comparison 4: Enhanced FFS compared with FFS
Study	Intervention	Comparison
Davidson 1992	Increase in FFS per service payment rate	FFS
Fairbrother 1999	Increase in FFS per service payment rate	FFS
Fairbrother 2001	Increase in FFS per service payment rate	FFS

Bias/study design	Randomised controlled trial	Controlled before‐after study	Interrupted time series
Allocation (selection bias)	Unclear: Fairbrother 1999, Fairbrother 2001, Christensen 2000, Davidson 1992, Kouides 1998, Lurie 1992 High: Houle 2016, Twardella 2007 Low: Bilardi 2010, Hickson 1987, Petersen 2013, Clarkson 2008	High: Gosden 2003, Gray 2015, Chung 2010, Flierman 1992, Jensen 2014, Lee 2010, Li 2013, Singh 2015, Greene 2013, Yesalis 1980, Yesalis 1984, Krasnik 1990, Gleeson 2017	High: Young 2012, Young 2007
Blinding (performance bias and detection bias)	Unclear: Fairbrother 1999, Fairbrother 2001, Hickson 1987, Kouides 1998 High: Lurie 1992 Low: Bilardi 2010, Christensen 2000, Clarkson 2008, Davidson 1992, Houle 2016, Petersen 2013, Twardella 2007	Unclear: Li 2013, Singh 2015 High: Gosden 2003 Low: Gleeson 2017, Gray 2015, Greene 2013, Chung 2010, Flierman 1992, Jensen 2014, Lee 2010, Yesalis 1980, Yesalis 1984, Krasnik 1990	Low: Young 2007, Young 2012
Incomplete outcome data (attrition bias)	Unclear: Bilardi 2010, Clarkson 2008 High: Twardella 2007 Low: Christensen 2000, Fairbrother 1999, Fairbrother 2001, Hickson 1987, Houle 2016, Kouides 1998, Lurie 1992, Petersen 2013, Davidson 1992	Unclear: Chung 2010, Gosden 2003, Lee 2010 Flierman 1992, Jensen 2014, Greene 2013, Yesalis 1980, Yesalis 1984, Gleeson 2017 High: Gray 2015 Low: Li 2013, Singh 2015, Krasnik 1990	Low: Young 2007, Young 2012
Selective reporting (reporting bias)	Unclear: Clarkson 2008 Low: Bilardi 2010, Christensen 2000, Davidson 1992, Fairbrother 1999, Fairbrother 2001, Hickson 1987, Houle 2016, Kouides 1998, Lurie 1992, Petersen 2013, Twardella 2007	Unclear: Chung 2010, Lee 2010, Greene 2013 Low: Gosden 2003, Gray 2015, Flierman 1992, Jensen 2014, Li 2013, Singh 2015, Yesalis 1980, Yesalis 1984, Krasnik 1990, Gleeson 2017	Low: Young 2007, Young 2012

P4P plus existing payment methods compared with existing payment methods for outpatient healthcare providers
Patient or population: healthcare providers working in outpatient healthcare settings Settings: Australia, Canada, India, Taiwan, the USA Intervention: P4P plus existing payment methods Comparison: existing payment methods (including capitation, FFS, and salary)
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	Results in words	No. of participants (studies)	Certainty of the evidence (GRADE)
Outcomes	Risk with existing payment methods	Risk with P4P plus existing payment methods	Relative effect (95% CI)	Results in words	No. of participants (studies)	Certainty of the evidence (GRADE)
Quantity of service provision (dichotomous) ‐ child immunisation status up‐to‐date	433 up‐to‐date per 1000 children	550 up‐to‐date per 1000 children (515 to 588 children)	RR 1.27 (1.19 to 1.36)	P4P added to existing payment methods probably increases the number of children with up‐to‐date immunisation status, compared with an existing payment method.	3760 children (2 RCTs)	⊕⊕⊕⊝ MODERATE¹
Quantity of service provision (continuous) ‐ immunisation rates for patients aged 65 years or older	The mean immunisation rate with P4P plus existing payment methods was 0.34 per cent points higher (0.2 per cent points lower to 0.87 points higher) compared to existing payment methods.		MD 0.34 (−0.20 to 0.87)	We are uncertain of the effect of P4P added to existing payment methods on immunisation rates for the elderly.	54 primary care practices (1 RCT)	⊕⊝⊝⊝ VERY LOW²
Quantity of service provision (continuous) ‐ number of detailed disease‐related consultation services per 100 prescriptions	The mean number of consultation services with P4P plus existing payment methods was 1.24 per cent higher per 100 prescriptions (0.93 per cent points higher to 1.54 points higher) compared to existing payment methods.		MD 1.24 (0.93 to 1.54)	P4P added to existing payment methods may slightly increase the number of patients who are asked more detailed questions on their disease by their pharmacist, compared with an existing payment method.	200 community pharmacies (1 RCT)	⊕⊕⊝⊝ LOW³
Quality of service provision ‐ physician prescribing practices	Insufficient data to calculate		RR 1.07 (1.02 to 1.12)	P4P added to existing payment methods may slightly improve providers' prescribing of guideline‐recommended antihypertensive medicines compared with an existing payment method.	362 people (1 RCT)	⊕⊕⊝⊝ LOW⁴
Patient outcomes ‐ reduction in blood pressure	The mean blood pressure reduction with P4P plus existing payment methods was 0.07 mmHG greater (2.22 less to 2.37 greater) compared to existing payment methods.		MD 0.07 (ranged from −2.22 to 2.37)	We are uncertain of the effect of P4P added to existing payment methods on mean blood pressure reduction compared with an existing payment method.	181 people (1 RCT)	⊕⊝⊝⊝ VERY LOW⁵
Quality of service provision and patient outcomes ‐ blood pressure management	Insufficient data to calculate		RR 1.13 (1.04 to 1.23)	P4P added to existing payment methods may improve blood pressure control or appropriate responses to patients with uncontrolled blood pressure, compared with an existing payment method.	362 people (1 RCT)	⊕⊕⊝⊝ LOW⁶
Healthcare provider outcomes (such as work morale or workload)	None of the included studies reported on healthcare provider outcomes.				‐	‐
Costs	We are uncertain of the costs of adding P4P to existing payment methods on expenditures for diabetes‐related services due to very low‐certainty evidence.				1 CBA	⊕⊝⊝⊝ VERY LOW⁷
Unintended or adverse effects	Insufficient data to calculate		RR 0.77 (0.71 to 0.82)	When the P4P intervention ended, there was an important reduction in performance (blood pressure control or appropriate responses to uncontrolled blood pressure) in the intervention group compared with the existing payment method.	362 people (1 RCT)	⊕⊕⊝⊝ LOW⁸
*The risk in the intervention group (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). CBA: controlled before‐after study; FFS: fee‐for‐service; MD: mean difference; P4P: pay for performance; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence High certainty: This research provides a very good indication of the likely effect. The likelihood that the effect will be substantially different is low. Moderate certainty: This research provides a good indication of the likely effect. The likelihood that the effect will be substantially different is moderate. Low certainty: This research provides some indication of the likely effect. However, the likelihood that it will be substantially different is high. Very low certainty: This research does not provide a reliable indication of the likely effect. The likelihood that the effect will be substantially different is very high. ** Substantially different = a large enough difference that it might affect a decision

FFS compared with existing payment methods for outpatient healthcare providers
Patient or population: healthcare providers working in outpatient healthcare settings Settings: the USA Intervention: FFS Comparison: existing payment method (input‐based payment ‐ capitation or salary)
Outcomes	Impact		No. of participants (studies)	Certainty of the evidence (GRADE)
	Relative effect (range)	Results in words
Quantity of service provision ‐ number of outpatients visits, specialist visits, or hospitalisations	Median change = 10.44% (range: −460% to +175.65%)	We are uncertain of the effect of FFS payments on the number of patient visits to health facilities, compared with input‐based payment methods, due to very low‐certainty evidence.	3 RCTs	⊕⊝⊝⊝ VERY LOW¹
Quality of service provision	None of the included studies reported on the quality of services provided.		‐	‐
Patient outcomes ‐ people with mental illness	Median change = −9.84% (range: −492% to +350%)	We are uncertain of the effect of FFS on patient outcomes for people with mental illness, compared with an input‐based payment method, due to very low‐certainty evidence.	1 RCT	⊕⊝⊝⊝ VERY LOW²
Healthcare provider outcomes (such as work morale or workload)	None of the included studies reported on healthcare provider outcomes.		‐	‐
Costs	We are uncertain of the effect of FFS on costs compared with an input‐based payment method (capitation) due to very low‐certainty evidence.		2 CBAs	⊕⊝⊝⊝ VERY LOW³
Unintended or adverse effects	Physicians who receive FFS payments may see more well patients (potentially an indicator of unnecessary service provision) compared with physicians paid by salary.		1 RCT	⊕⊕⊝⊝ LOW⁴
CBA: controlled before‐after study; FFS: fee‐for‐service; RCT: randomised controlled trial
GRADE Working Group grades of evidence High certainty: This research provides a very good indication of the likely effect. The likelihood that the effect will be substantially different is low. Moderate certainty: This research provides a good indication of the likely effect. The likelihood that the effect will be substantially different is moderate. Low certainty: This research provides some indication of the likely effect. However, the likelihood that it will be substantially different is high. Very low certainty: This research does not provide a reliable indication of the likely effect. The likelihood that the effect will be substantially different is very high. ** Substantially different = a large enough difference that it might affect a decision

FFS mixed with existing payment methods compared with existing payment method for outpatient healthcare providers
Patient or population: healthcare providers working in outpatient healthcare settings Settings: Australia, Canada, Denmark, Germany, the UK Intervention: FFS mixed with other payment methods Comparison: existing payment method (single payment method, including salary, capitation, and FFS)
Outcomes	*Anticipated absolute effects^ (95% CI)**		Impact		No. of participants (studies)	Certainty of the evidence (GRADE)
	Risk with existing payment	Risk with FFS mixed with other payment methods	Relative effect (95% CI)	Results in words
Quantity of service provision ‐ proportion of women/children receiving treatment	Insufficient data to calculate		RR 1.37 (1.07 to 1.76)	FFS mixed with other payment methods may increase the quantity of health services provided, compared with an existing payment method.	2 RCTs	⊕⊕⊝⊝ LOW¹
Quality of service provision	We are uncertain of the effect of FFS mixed with other payment methods on the quality of service provided compared with an existing payment method due to very low‐certainty evidence.				1 CBA	⊕⊝⊝⊝ VERY LOW²
Patient outcomes ‐ satisfaction with care	We are uncertain of the effect of FFS mixed with other payment methods on patient outcomes compared with an existing payment method due to very low‐certainty evidence.				1 RCT	⊕⊝⊝⊝ VERY LOW³
Healthcare provider outcomes ‐ working hours and income	We are uncertain of the effect of FFS mixed with other payment methods on healthcare provider outcomes compared with an existing payment method due to very low‐certainty evidence.				2 CBAs	⊕⊝⊝⊝ VERY LOW⁴
Costs	None of the included studies reported on costs.				‐	‐
Unintended or adverse effects	None of the included studies reported on adverse effects.				‐	‐
*The risk in the intervention group (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). CBA: controlled before‐after study; FFS: fee‐for‐service; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence High certainty: This research provides a very good indication of the likely effect. The likelihood that the effect will be substantially different is low. Moderate certainty: This research provides a good indication of the likely effect. The likelihood that the effect will be substantially different is moderate. Low certainty: This research provides some indication of the likely effect. However, the likelihood that it will be substantially different is high. Very low certainty: This research does not provide a reliable indication of the likely effect. The likelihood that the effect will be substantially different is very high. ** Substantially different = a large enough difference that it might affect a decision

Enhanced FFS compared with FFS for outpatient healthcare providers
Patient or population: healthcare providers working in outpatient healthcare settings Settings: Germany, the USA Intervention: enhanced FFS with higher unit payment levels Comparison: FFS with normal unit payment levels
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	Results in words	No. of participants (studies)	Certainty of the evidence (GRADE)
	Risk with FFS	Risk with Enhanced FFS
Quantity of service provision ‐ child immunisation status up‐to‐date	433 up‐to‐date per 1000 children	541 up‐to‐date per 1000 children (459 to 640 children)	RR 1.25 (1.06 to 1.48)	Paying higher fees to healthcare providers for immunisations delivered probably increases the proportion of children aged 3 to 35 months whose immunisation is up‐to‐date, compared with the normal level of fees.	2 RCTs	⊕⊕⊕⊝ MODERATE¹
Quantity of service provision ‐ primary healthcare visits by children	We are uncertain of the effect of paying higher fees for service on the number of primary care visits by children compared with the normal level of fees due to very low‐certainty evidence.				1 RCT	⊕⊝⊝⊝ VERY LOW²
Quality of service provision	None of the included studies reported on quality of service provision.				‐	‐
Patient outcomes	None of the included studies reported on patient outcomes.				‐	‐
Healthcare provider outcomes (such as work morale or workload)	None of the included studies reported on healthcare provider outcomes.				‐	‐
Costs	We are uncertain if paying higher fees to healthcare providers for immunisations delivered influences the net expenditure per year on eligible children compared with the normal level of fees because the certainty of the evidence is very low.				1 RCT	⊕⊝⊝⊝ VERY LOW³
Unintended or adverse effects	None of the included studies reported on adverse effects.
*The risk in the intervention group (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). FFS: fee‐for‐service; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence High certainty: This research provides a very good indication of the likely effect. The likelihood that the effect will be substantially different is low. Moderate certainty: This research provides a good indication of the likely effect. The likelihood that the effect will be substantially different is moderate. Low certainty: This research provides some indication of the likely effect. However, the likelihood that it will be substantially different is high. Very low certainty: This research does not provide a reliable indication of the likely effect. The likelihood that the effect will be substantially different is very high. ** Substantially different = a large enough difference that it might affect a decision

Comparison 1: P4P plus existing payment methods compared withexisting payment methods
Study	Outcome measures	Control/baseline level	Absolute change	Relative change
Quantity of health services provided
Fairbrother 1999	UTD immunisation rate (adjusted by baseline)	12.6%	12.7%	+140.77%
Fairbrother 2001	UTD immunisation rate (adjusted by baseline)	−2.5%	8.4%	−336%
Kouides 1998	Immunisation rate	4.7	6.3%	+134.04%
Christensen 2000	Documentation rate of CS per 100 Medicaid prescriptions	0.67%	0.92%	+137.31%
	CS intervention time per 100 Medicaid prescriptions	6.5%	1.4%	+21.54%
	Synthesised effects inside the study (median)	‐	1.16%	+79.43%
Synthesised effect across the above 4 studies (median)		‐	‐	+134.04%
Quality of health services provided
Petersen 2013	Use of guideline	0.47%	8.37%	+1780.85%
Patient outcomes
Petersen 2013	Blood pressure control or appropriate response	4.35%	4.72%	+108.51%
Houle 2016	Systolic BP level reduction	17.0	2.7	+15.88%
	Diastolic BP level reduction	8.2	−0.6	−7.32%
	Synthesised effects inside the study (median)	‐	1.05	+4.28%
Synthesised effect across the above 2 studies (median)		‐	‐	+15.88%
Comparison 2: FFS compared withexisting payment methods
Study	Outcome measures	Control/baseline level	Absolute change	Relative change
Quantity of health services
Davidson 1992	Primary care physician visits	−0.33	0.36	−61.02%
	Non‐primary visit	−0.05	0.23	−460%
	Clinic visits	−0.47	−0.05	+10.64%
	Hospitalisation	−0.042	−0.0276	+65.71%
	Synthesised effects inside the study (median)	‐	‐	−25.19%
Lurie 1992	Proportion of receiving outpatient	61%	10%	+16.39%
	Over average inpatient admissions	0.20	0.19	+95%
	Proportion of outpatient chemical depending treatment	4.5%	5.8%	+128.89%
	Proportion of inpatient chemical depending treatment	1.9%	3.3%	+173.68%
	Length of stays	1.56	2.74	+175.64%
	Proportion of being refused	17%	−5%	−29.41%
	Synthesised effects inside the study (median)	‐	‐	+111.95%
Hickson 1987	Average number of patient/physician	43.4	11.7	+26.96%
	Average number of visits attended/physician	111.6	−6.8	−6.09%
	% attended by primary physician	86.6%	−8.3%	−9.58%
	Emergency visit	0.12	0.1	+83.33%
	Synthesised effects inside the study (median)	‐	‐	+10.44%
Synthesised effect across the above 3 studies (median)		‐	‐	+10.44%
Patient outcomes
Lurie 1992	Self‐rated health	3.0%	−3.7%	+123.33%
	General health status	0.2	−0.2	+100.%
	Physical functioning index	0	0	0
	Social contact	−0.7	1.2	−171.43%
	Mean Global Assessment Scale for psychiatric status	−1	−0.3	−30%
	Depression	−0.8	2.9	−362.5%
	Anxiety	0.2	0.7	+350%
	Endogenous features	0.7	0.5	+71.43%
	Mania	0.2	−0.6	−3%
	Delusions	0.2	0	0
	Miscellaneous	0.1	0.4	+80%
	Arrested	0.3	1	−333.33%
	Jailed	0	−1.1	0
	Assaulted	1.3	−6.4	−492.31%
	Suicide	−1.2	0.2	−16.67%
	Living in sheltered setting	−1.9	−3.1	−147.62%
	With nights homeless during previous year	−0.3	1	−333.33%
	Working in sheltered setting	−0.2	0.1	−100%
Synthesised effects inside the study (median)		‐	0.05	−9.84%
Comparison 3: FFS mixed payment methods compared withsingle payment method
Study	Outcome measures	Control/baseline level	Absolute change	Relative change
Quantity of health services provided
Bilardi 2010	Proportion of patients tested	2.6%	−0.7%	−26.92%
Clarkson 2008	Percentage of children with at least 1 sealant treatment per dentist	26.3%	6.6%	+25.1%
Synthesised effect across the above 2 studies (median)		‐	‐	−0.91%
Patient outcomes
Twardella 2007	Self‐reported smoking abstinence	0.12%	0.03%	+25%
Comparison 4: Enhanced FFS compared with FFS
Study	Outcome measures	Control/baseline level	Absolute change	Relative change
Quantity of health services
Davidson 1992	Primary care physician visits	−0.59	0.56	−94.92%
	Non‐primary visit	0.19	−0.37	−194.74%
	Clinic visits	−0.34	−0.18	+52.94%
	Hospitalisation	−0.0312	−0.0384	+123.08%
	Synthesised effects inside the study (median)	‐	−0.1092	−20.99%
Fairbrother 1999	UTD immunisation rate	12.6%	−8.3	−65.87%
Fairbrother 2001	UTD immunisation rate	−2.5%	9.9%	−396%
Synthesised effects across the above 3 studies (median)		‐	‐	−65.87%

Comparison 1: P4P plus existing payment methods compared withexisting payment methods
Study	Immediate change in level		Change in trend		Other effects results reported by authors
Study	Estimate	Confidence interval	Estimate	Confidence interval	Other effects results reported by authors
Young 2007, RM
Adherence rate of glycated haemoglobin tests	0.03	0.01 to 0.05	0.07	0.03 to 0.10	‐
Adherence rate of urinalysis	0.03	0.002 to 0.06	0.09	0.06 to 0.11	‐
Adherence rate of lipoprotein density level screening	0.05	0.02 to 0.08	0.2	0.18 to 0.24	‐
Adherence rate of eye examination	0.07	0.04 to 0.09	0.14	0.11 to 0.17	‐
Young 2012, RM
Performance scores of eye examination	0.09	0.01 to 0.17	0.11	0.01 to 0.21	‐
Performance scores of glycated haemoglobin tests	0.04	0.01 to 0.07	0.03	0.02 to 0.04	‐
Performance scores of lipoprotein density level screening	0.03	0.02 to 0.04	0.04	0.01 to 0.07	‐
Performance scores of nephropathy test	0.02	0.01 to 0.03	0.03	0.01 to 0.05	‐

Review	Focus	Key findings
Giuffrida 1999	Focuses on “target payments” ‐ linking payment to a specific level of activity or quality	Linking payment to physicians' target behaviours was associated the increased immunisation rates.
Gosden 2000	Focuses on the effects of different payment methods on the clinical behaviours of primary care physicians	Fee‐for‐service resulted in a higher number of service provision process outcomes compared with fixed payment.
Scott 2011	Includes all kinds of payment changes, but focuses on outcomes related to quality of health care	Linking financial incentives directly to quality of primary care physicians had moderate positive effects.
Witter 2012	Focuses on paying‐for‐performance schemes targeting different levels of healthcare providers in low‐ and middle‐income countries	There were mixed results on the effects of performance‐based funding on service provision and health outcome, and the effects of paying for performance depended on intervention design.
Brocklehurst 2013	Focuses on participants and outcomes related to clinical activities undertaken by primary care dentists	Limited included studies provided low‐/very low‐evidence showing financial incentives within remuneration systems may produce changes to clinical activity undertaken by primary care dentists.
Flodgren 2011	An overview of reviews evaluating the effectiveness of all kinds of payments at both the individual and organisation level on changing healthcare provider behaviour and patient outcomes	Financial incentives may be effective in changing healthcare provider practice. The evidence has serious methodological limitations and is also very limited in its completeness and generalisability.

ID	Search
#1	MeSH descriptor: [Physician Incentive Plans] this term only
#2	MeSH descriptor: [Group Practice, Prepaid] this term only
#3	MeSH descriptor: [Reimbursement Mechanisms] this term only
#4	MeSH descriptor: [Reimbursement, Incentive] this term only
#5	MeSH descriptor: [Prospective Payment System] this term only
#6	MeSH descriptor: [Single‐Payer System] this term only
#7	MeSH descriptor: [Prepaid Health Plans] this term only
#8	MeSH descriptor: [Capitation Fee] this term only
#9	MeSH descriptor: [Fees and Charges] this term only
#10	MeSH descriptor: [Fee‐for‐Service Plans] this term only
#11	MeSH descriptor: [Salaries and Fringe Benefits] this term only
#12	MeSH descriptor: [Remuneration] this term only
#13	MeSH descriptor: [Value‐Based Purchasing] this term only
#14	{or #1‐#13}
#15	(reimbursement or economic next incentive* or financial next incentive* or monetary next incentive* or economic next reward* or financial next reward* or monetary next reward* or incentive next payment* or payment next incentive* or performance next bonus* or bonus next payment* or "bonus for practice" or target next payment* or conditional next payment* or performance payment* or "pay for procedure" or "paid for procedure" or "paying for procedure" or "pay for performance" or "paid for performance" or "paying for performance" or p4p or performance next based next payment* or "performance based subsidy" or "performance based subsidies" or "performance based financing" or result* next based next financing or result* next based next funding or result* next based next payment* or "input based financing" or "input based funding" or "output based financing" or "output based funding" or remunerate or remuneration or capitation or "capitated financing" or "fee for service" or "fee for services" or salary or salaries or salaried next contract* or prepaid next plan* or prepaid next health next plan* or prepaid next healthcare next plan* or prepaid next health next care next plan* or pre next paid next plan* or pre next paid next health next plan* or pre next paid next healthcare next plan* or pre next paid next health next care next plan* or prepaid next service* or prepaid next health next service* or prepaid next healthcare next service* or prepaid next health next care next service* or pre next paid next service* or pre next paid next health next service* or pre next paid next healthcare next service* or pre next paid next health next care next service* or "prepaid care" or "pre paid care" or "prepaid healthcare" or "pre paid healthcare" or "prepaid health care" or "pre paid health care" or prospective next payment* or retrospective next payment* or payment near/2 method* or payment near/2 mechanism* or payment near/2 system*):ti,ab
#16	MeSH descriptor: [Physicians, Primary Care] this term only
#17	MeSH descriptor: [Physicians, Family] this term only
#18	MeSH descriptor: [General Practitioners] this term only
#19	MeSH descriptor: [Family Nurse Practitioners] this term only
#20	MeSH descriptor: [Nurses, Community Health] this term only
#21	MeSH descriptor: [Primary Health Care] this term only
#22	MeSH descriptor: [Ambulatory Care] this term only
#23	MeSH descriptor: [Family Practice] this term only
#24	MeSH descriptor: [Family Nursing] this term only
#25	MeSH descriptor: [General Practice] this term only
#26	MeSH descriptor: [General Practice, Dental] this term only
#27	MeSH descriptor: [Private Practice] this term only
#28	MeSH descriptor: [Group Practice] this term only
#29	MeSH descriptor: [Group Practice, Dental] this term only
#30	MeSH descriptor: [Office Visits] this term only
#31	MeSH descriptor: [Ambulatory Care Facilities] this term only
#32	MeSH descriptor: [Community Health Centers] this term only
#33	MeSH descriptor: [Community Mental Health Centers] this term only
#34	MeSH descriptor: [Outpatient Clinics, Hospital] this term only
#35	MeSH descriptor: [Community Health Services] this term only
#36	MeSH descriptor: [Community Health Nursing] this term only
#37	MeSH descriptor: [Community Mental Health Services] this term only
#38	MeSH descriptor: [Community Pharmacy Services] this term only
#39	MeSH descriptor: [Dental Health Services] this term only

ID	Search	Hits
#1	MeSH descriptor: [Physician Incentive Plans] this term only	14
#2	MeSH descriptor: [Group Practice, Prepaid] this term only	9
#3	MeSH descriptor: [Reimbursement Mechanisms] this term only	47
#4	MeSH descriptor: [Reimbursement, Incentive] this term only	73
#5	MeSH descriptor: [Prospective Payment System] this term only	2
#6	MeSH descriptor: [Single‐Payer System] this term only	2
#7	MeSH descriptor: [Prepaid Health Plans] this term only	6
#8	MeSH descriptor: [Capitation Fee] this term only	32
#9	MeSH descriptor: [Fees and Charges] this term only	55
#10	MeSH descriptor: [Fee‐for‐Service Plans] this term only	36
#11	MeSH descriptor: [Salaries and Fringe Benefits] this term only	45
#12	MeSH descriptor: [Remuneration] this term only	10
#13	MeSH descriptor: [Value‐Based Purchasing] this term only	1
#14	{or #1‐#13}	285
#15	(reimbursement or economic next incentive* or financial next incentive* or monetary next incentive* or economic next reward* or financial next reward* or monetary next reward* or incentive next payment* or payment next incentive* or performance next bonus* or bonus next payment* or "bonus for practice" or target next payment* or conditional next payment* or performance payment* or "pay for procedure" or "paid for procedure" or "paying for procedure" or "pay for performance" or "paid for performance" or "paying for performance" or p4p or performance next based next payment* or "performance based subsidy" or "performance based subsidies" or "performance based financing" or result* next based next financing or result* next based next funding or result* next based next payment* or "input based financing" or "input based funding" or "output based financing" or "output based funding" or remunerate or remuneration or capitation or "capitated financing" or "fee for service" or "fee for services" or salary or salaries or salaried next contract* or prepaid next plan* or prepaid next health next plan* or prepaid next healthcare next plan* or prepaid next health next care next plan* or pre next paid next plan* or pre next paid next health next plan* or pre next paid next healthcare next plan* or pre next paid next health next care next plan* or prepaid next service* or prepaid next health next service* or prepaid next healthcare next service* or prepaid next health next care next service* or pre next paid next service* or pre next paid next health next service* or pre next paid next healthcare next service* or pre next paid next health next care next service* or "prepaid care" or "pre paid care" or "prepaid healthcare" or "pre paid healthcare" or "prepaid health care" or "pre paid health care" or prospective next payment* or retrospective next payment* or payment near/2 method* or payment near/2 mechanism* or payment near/2 system*):ti,ab	2344
#16	MeSH descriptor: [Physicians, Primary Care] this term only	140
#17	MeSH descriptor: [Physicians, Family] this term only	443
#18	MeSH descriptor: [General Practitioners] this term only	230
#19	MeSH descriptor: [Family Nurse Practitioners] this term only	1
#20	MeSH descriptor: [Nurses, Community Health] this term only	12
#21	MeSH descriptor: [Primary Health Care] this term only	3719
#22	MeSH descriptor: [Ambulatory Care] this term only	3079
#23	MeSH descriptor: [Family Practice] this term only	1963
#24	MeSH descriptor: [Family Nursing] this term only	35
#25	MeSH descriptor: [General Practice] this term only	422
#26	MeSH descriptor: [General Practice, Dental] this term only	52
#27	MeSH descriptor: [Private Practice] this term only	70
#28	MeSH descriptor: [Group Practice] this term only	35
#29	MeSH descriptor: [Group Practice, Dental] this term only	0
#30	MeSH descriptor: [Office Visits] this term only	430
#31	MeSH descriptor: [Ambulatory Care Facilities] this term only	432
#32	MeSH descriptor: [Community Health Centers] this term only	193
#33	MeSH descriptor: [Community Mental Health Centers] this term only	109
#34	MeSH descriptor: [Outpatient Clinics, Hospital] this term only	572
#35	MeSH descriptor: [Community Health Services] this term only	934
#36	MeSH descriptor: [Community Health Nursing] this term only	333
#37	MeSH descriptor: [Community Mental Health Services] this term only	680
#38	MeSH descriptor: [Community Pharmacy Services] this term only	236
#39	MeSH descriptor: [Dental Health Services] this term only	24

#	Searches	Results
1	Physician Incentive Plans/ec [Economics]	590
2	Group Practice, Prepaid/	551
3	or/1‐2 [PAYMENT & PRACTITIONERS/PRACTICE mixed]	1138
4	Physician Incentive Plans/ or Reimbursement Mechanisms/ or Reimbursement Incentive/ or Prospective Payment System/ or Single‐Payer System/ or Prepaid Health Plans/ or Capitation Fee/ or "Fees and Charges"/ or Fee‐for‐Service Plans/ or "Salaries and Fringe Benefits"/ or Remuneration/ or Value‐Based Purchasing/	52017
5	(reimbursement or economic incentive? or financial incentive? or monetary incentive? or economic reward? or financial reward? or monetary reward? or incentive payment? or payment incentive? or performance bonus* or bonus payment? or bonus for practice or target payment? or conditional payment? or performance payment? or pay for procedure or paid for procedure or paying for procedure or pay for performance or paid for performance or paying for performance or p4p or performance based payment? or performance based subsidy or performance based subsidies or performance based financing or result? based financing or result? based funding or result? based payment? or input based financing or input based funding or output based financing or output based funding or remunerate or remuneration or capitation or capitated financing or fee for service? or salary or salaries or salaried contract? or prepaid plan? or prepaid health plan? or prepaid healthcare plan? or prepaid health care plan? or pre paid plan? or pre paid health plan? or pre paid healthcare plan? or pre paid health care plan? or prepaid service? or prepaid health service? or prepaid healthcare service? or prepaid health care service? or pre paid service? or pre paid health service? or pre paid healthcare service? or pre paid health care service? or prepaid care or pre paid care or prepaid healthcare or pre paid healthcare or prepaid health care or pre paid health care or prospective payment? or retrospective payment? or (payment adj2 method?) or (payment adj2 mechanism?) or (payment adj2 system?)).ti,ab,kf.	48165
6	or/4‐5 [PAYMENT METHODS]	85413
7	Physicians, Primary Care/	2845
8	Physicians, Family/	15945
9	General Practitioners/	6711
10	Family Nurse Practitioners/	37
11	Nurses, Community Health/	715
12	or/7‐11 [OUTPATIENT HEALTH PROFESSIONALS]	26063
13	Primary Health Care/	70902
14	Ambulatory Care/	41074
15	Family Practice/	64286
16	Family Nursing/	1372
17	General Practice/	12042
18	General Practice, Dental/	4725
19	Private Practice/	8072
20	Group Practice/	7800
21	Group Practice, Dental/	612
22	Office Visits/	6560
23	or/13‐22 [OUTPATIENT CARE / PRACTICE]	200140
24	Ambulatory Care Facilities/	17510
25	Community Health Centers/	6788
26	Community Mental Health Centers/	2871
27	Outpatient Clinics, Hospital/	15358
28	or/24‐27 [OUTPATIENT FACILITIES]	41997
29	Community Health Services/	30386
30	Community Health Nursing/	19219
31	Community Mental Health Services/	18014
32	Community Pharmacy Services/	4126
33	Dental Health Services/	3990
34	or/29‐33 [OUTPATIENT SERVICES]	73883
35	Quality of Health Care/	68748
36	Quality Improvement/	19322
37	Quality Assurance, Health Care/	54666
38	Quality Indicators, Health Care/	14391
39	or/35‐38 [QUALITY]	147946

#	Searches	Results
1	medical fee/	13185
2	capitation fee/	3746
3	prospective payment/	8259
4	reimbursement/	52616
5	fee for service/	5851
6	(capitation or reimbursement or fee for service? or prospective payment? or retrospective payment? or target payment? or pay* for performance or p4p or performance payment? or performance based pay* or performance based financing or financial incentive? or economic incentive? or monetary incentive? or prepaid health* plan? or pre paid health* plan? or prepaied health care plan? or pre paied health care plan? or (payment? adj2 method?) or (payment? adj2 mechanism?) or (payment? adj2 system?)).ti,ab,od,ct,kw.	86582
7	or/1‐6 [PAYMENT METHIODS]	97015
8	ambulatory care/	33832
9	primary health care/	60691
10	primary medical care/	94258
11	outpatient care/	32813
12	outpatient department/	56842
13	general practitioner/	89987
14	general practice/	74378
15	community care/	51541
16	(family physician? or family practitioner? or family clinician? or family doctor? or general practitioner? or community physician? or family practice? or general practice? or primary care or primary healthcare or primary health care or ambulatory care or ambulatory patient? or outpatient? or (community adj3 service?) or (community adj3 care) or ambulatory facility or ambulatory care facility or ambulatory facilities or ambulatory care facilities or ambulatory clinic? or (community adj3 clinic?) or (community adj3 center?) or (community adj3 centre?) or dental clinic? or dental service? or pharmac* service? or community pharmac*).ti,ab,od,ct,kw.	674465
17	or/8‐16 [OUTPATIENT]	753505
18	7 and 17	17829
19	(performance based financing or (pay* and performance) or target payment? or financial incentive? or incentive payment? or capitation or capitated financing or capitated payment? or reimbursement or remuneration system? or fee for service? or (pay* and behaviour) or (pay* and behavior)).ti. [TERMS TAKEN FROM TITLES OF INCLUDED STUDIES]	10822
20	18 or 19	26880
21	Randomized Controlled Trial/	534657
22	Quasi Experimental Study/	5333
23	Pretest Posttest Control Group Design/	367
24	Time Series Analysis/	22359
25	(randomis* or randomiz* or randomly or controlled or control group? or (before adj5 after) or (pre adj5 post) or ((pretest or pre test) and (posttest or post test)) or quasiexperiment* or quasi experiment* or evaluat* or time series or time point? or repeated measur* or groups).ti,ab.	7648001
26	(trial or intervention? or effect? or impact?).ti.	2454297
27	or/21‐26	9085911
28	exp animals/ or exp invertebrate/ or animal experiment/ or animal model/ or animal tissue/ or animal cell/ or nonhuman/	25442079
29	human/ or normal human/ or human cell/	19376288
30	28 not (28 and 29)	6119288
31	(systematic review or literature review).ti.	150895
32	"cochrane database of systematic reviews".jn.	13058
33	27 not (30 or 31 or 32)	7133848
34	20 and 33	8335
35	("Health Policy, Economics and Management" or "Public Health, Social Medicine and Epidemiology").ec.	2719999
36	34 and 35	4112
37	limit 36 to embase	4100

*Study characteristics*
Methods	Cluster‐randomised trial
Participants	General practitioners
Interventions	Victoria, Australia Intervention: pay for performance (a small incentive payment per test): receive a AUD 5 payment per chlamydia test for testing 16‐ to 24‐year‐old women for chlamydia Control: receive no payment for testing 16‐ to 24‐year‐old women for chlamydia Year 2008 to 2009, intervention happened in 2008
Outcomes	Quantity of health services provided Women tested
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Using a predetermined randomisation sequence prepared by the trial statistician
Allocation concealment (selection bias)	Low risk	The allocation was not revealed to staff at either of the paired practices until representatives from both had completed the pre‐trial requirements (questionnaire, audit, education session). General practitioners in the practice were then contacted by letter to inform them of the allocation relevant to their practice and reminded of testing payment or non‐payment.
Baseline outcome measurements similar	Unclear risk	No comparison on the baseline outcome measurements was reported.
Baseline characteristics similar	Low risk	An imbalanced distribution of variables was also evident at trial commencement, with GPs in the control group more likely to be younger and to have a postgraduate qualification. But the adjusted analysis controlling the characteristics variables was conducted.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	No information
Knowledge of the allocated interventions adequately prevented	Low risk	The allocation was not revealed to staff at either of the paired practices until representatives from both had completed the pre‐trial requirements (questionnaire, audit, education session). GPs in the practice were then contacted by letter to inform them of the allocation relevant to their practice and reminded of testing payment or non‐payment.
Study adequately protected against contamination	Low risk	Mid‐trial, GPs in the intervention group received a letter to remind them of the incentive offered for chlamydia testing. They were not provided with any information about the number of tests performed to date nor the amount of money they had accrued through testing. Payment was made to GPs at the end of the trial period.
Selective reporting (reporting bias)	Low risk	No missing outcome data
Other bias	Low risk

*Study characteristics*
Methods	Randomised trial
Participants	Community pharmacies
Interventions	State of Washington, US Intervention: financial incentive (Pay‐for‐performance). All pharmacies agreed to document cognitive services (CS) for Medicaid recipients in exchange for a modest monthly fee (USD 40). Additionally, study group pharmacies were eligible to be reimbursed for each CS documented and billed to Medicaid. They were compensated at the rate of USD 4.00 for interventions up to 6 minutes in duration, and USD 6.00 for interventions of 6 minutes or longer. 110 community pharmacies Control: pharmacies agreed to document CS for Medicaid recipients in exchange for a modest monthly fee (USD 40). 90 community pharmacies First study month to 20th study month, intervention happened in 1990
Outcomes	Quantity of health services provided Documentation rate Frequency of cognitive services (CS) Frequency of CS interventions Primary results of CS interventions CS intervention rates per 100 prescriptions Patients receiving CS, by problem type Pharmacists self‐reported time per CS
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The randomised cluster sampling methodology is described in an earlier article.
Allocation concealment (selection bias)	Unclear risk	No more information in the text
Baseline outcome measurements similar	Unclear risk	No comparison on the baseline outcome measurements was reported.
Baseline characteristics similar	Unclear risk	No comparison on the baseline characteristics was reported.
Incomplete outcome data (attrition bias) All outcomes	Low risk	The missing data are lower than 0.1%.
Knowledge of the allocated interventions adequately prevented	Low risk	The outcomes are objective.
Study adequately protected against contamination	Low risk	Allocation was by county and country, and it is unlikely that the control group received the intervention.
Selective reporting (reporting bias)	Low risk	Assessment of the cost impact of these changes is reported separately.
Other bias	High risk	It is possible that pharmacists performed but did not document CS except in response to the financial incentive. And any event‐based reimbursement system should recognise the need to compensate pharmacists more for longer intervention times and for multiple interventions.

*Study characteristics*
Methods	Interrupted time series study
Participants	334 primary health physicians members of the Rochester Individual Practice Association (RIPA)
Interventions	Rochester, New York (USA) Interventions: pay for performance. In 2001, RIPA began the process of establishing a pay‐for‐performance programme as part of its contract with Excellus Health Plan to provide professional services for the approximately 450,000 individuals enrolled in Excellus’ Blue Choice Health Maintenance Organization (HMO) insurance product. Each RIPA primary care physician was eligible for reward payments of up to approximately USD 15,000 depending on his or her relative ranking on a composite performance measure for the diabetes‐related tests and screens. Control: no explicit information, but may be salary. RIPA served as an umbrella organisation for its member physicians with responsibilities for contracting with health plans on behalf of its member physicians and developing arrangements by which member physicians were reimbursed for their services. Years 1999 to 2004, intervention happed in 2002, 3 time points for pre‐/postintervention
Outcomes	Quality of health care provision 4 diabetes performance measures: glycated haemoglobin check, urinalysis, lipoprotein density level check, eye exam. Glycated haemoglobin check: increase in level without significance, but post‐trend did not differ from the pre‐trend Urinalysis: increase in level without significance, but post‐trend did not differ from the pre‐trend LDL check: increase in level without significance, but post‐trend did not differ from the pre‐trend Eye exams: increase in levels (7 percentage point increase after 1 year of intervention, but the increase did not persist), post‐trend did not differ from the pre‐trend
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Intervention independent of other changes (ITS)	Low risk	The increase in the eye examination score of intervention programme (Rochester (New York) Individual Practice Association) was contrary to the trends observed in the eye examination scores in Health Employer Data Information System nationally and statewide during this period, which were largely flat from 2000 to 2002. The observed increase thus did not appear to be attributable to a secular trend.
Shape of the intervention effect pre‐specified (ITS)	Low risk	The point of analysis is clear, the intervention start time: 2002.
Intervention unlikely to affect data collection (ITS)	Unclear risk	The data used for analysis in this paper were from patient electronic records, which were not influenced by payment reform. However, it is unclear if GPs' recording was changed with the start of the new payment policy.
Knowledge of the allocated interventions adequately prevented (ITS)	Low risk	Outcomes were objective.
Incomplete outcome data adequately addressed (ITS)	Low risk	No missing outcome data
Study free from selective outcome reporting (ITS)	Low risk	All outcomes mentioned in the methods section were reported in the results section.
Other risk of bias (ITS)	Low risk	The study included physicians who were members of the health plan for all 6 years, and so does not account for the potential selection bias if poorly performing physicians chose not to participate or withdraw.

*Study characteristics*
Methods	Repeated measures study
Participants	General practitioner practices
Interventions	Rochester, New York (USA) Interventions: pay for performance. In 2001, Rochester Individual Practice Association (RIPA) began the process of establishing a pay‐for‐performance programme as part of its contract with Excellus Health Plan to provide professional services for the approximately 450,000 individuals enrolled in Excellus’ Blue Choice Health Maintenance Organization (HMO) insurance product. Each RIPA primary care physician was eligible for reward payments of up to approximately USD 15,000 depending on his or her relative ranking on a composite performance measure for the diabetes‐related tests and screens. Control: no explicit information, but may be salary. RIPA served as an umbrella organisation for its member physicians with responsibilities for contracting with health plans on behalf of its member physicians and developing arrangements by which member physicians were reimbursed for their services. Years 1999 to 2004, intervention happened in 2002, 3 time points for pre‐/postintervention
Outcomes	Quality of health care provision Diabetes‐related composite measure performance score (4 diabetes performance measures include glycated haemoglobin testing, lipoprotein density level screening, nephropathy, eye examination): Repeated Measure regression model controlling all factors showed that P4P was related to 0.100 (Standard Error 0.032, P < 0.01) higher score.
Notes
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Intervention independent of other changes (ITS)	Low risk	As noted earlier, the study sample consisted of physicians who had received reports regarding their relative performance on the diabetes‐related measures both during the baseline period and after introduction of the incentive. As such, a strength of the study design is that the participants did not experience any changes in performance goals or performance‐related information during the study period. The key relevant change was the introduction of the financial incentive.
Shape of the intervention effect pre‐specified (ITS)	Low risk	The point of analysis is clear, the intervention start time: 2002.
Intervention unlikely to affect data collection (ITS)	Unclear risk	The data used for analysis in this paper were from patient electronic records, which were not influenced by payment reform. However, it is unclear if GPs' recording was changed with the start of the new payment policy.
Knowledge of the allocated interventions adequately prevented (ITS)	Low risk	Outcomes were objective.
Incomplete outcome data adequately addressed (ITS)	Low risk	No missing outcome data
Study free from selective outcome reporting (ITS)	Low risk	All outcomes mentioned in the methods section were reported in the results.

Study	Reason for exclusion
Abelsen 2015	This is a cross‐sectional study.
Agee 2014	Just a comparative study
Allard 2014	This study just used a model of physician treatment and referral decisions under endogenous payment form in the flavour of Allard and colleagues.
Allen 2016	Not a Controlled before‐after study (CBA) study design: although it uses difference‐in‐differences model, study lacks a control group
Alshamsan 2012	The provider is institution.
An 2008	Study participants are clinics.
Arrowsmith 2014	The provider is institution.
Baker 2005	This study is not an Randomised Controlled Trial (RCT), CBA, or Interrupted Time Series (ITS).
Barnato 2017	The target of the intervention in this study is the patient.
Basu 2016	The outcome index concerns the impact on financial outcome.
Boyden 2000	The provider is institution.
Broadway 2017	This study is not an RCT, CBA, or ITS; it uses a structure, discrete choice model.
Carey 1990	Not control study
Chen 2016	Not a CBA study design; control group was created by matching methods to conduct difference‐in‐differences analysis
Clancy 1988	Not control study
Coleman 2007	Not a CBA study, only a before‐and‐after study
Davies 1986	Provider is insurance plan.
Douven 2015	Not an ITS study design because it does not have a clearly defined point in time; only time trends are described
Engineer 2016	The participants of pay for performance in this study are primary care facilities.
Erickson 2016	In this study the participants are physicians in haemodialysis facility and their patients, but only patients are described and there is no more information about the physicians.
Erickson 2017	The intervention in this study is for inpatients.
Erus 2017	Not designed as a CBA study, no control group
Feng 2015	Not a CBA study design, but rather a before‐after comparative study without control group
Gallagher 2015	The provider is institution.
Giguere 2015	The intervention of this study is not related to payment methods.
Greene 2013	The interventions are for both general practitioners and general practices.
Hamilton 2016	Not a CBA study design, but rather a before‐after comparative study without control group
Han 2015	This study does not have at least 3 time points before and 3 after the intervention. An interrupted time‐series analysis allows separate analysis of immediate effects and monthly trends following policy implementation, but it is shown through 3 periods.
Hickey 2015	Not a CBA study design; only 1 site
Hysong 2017	The participants of the study are not ambulatory care professionals.
Jones 2015	Not a CBA study design; no control group
Kiran 2015	Not a CBA study design. It does not have a clearly defined point in time, but just describes the change trend of different physicians payment methods.
Kliner 2015	Not a CBA study design. The intervention was only performed in a regional hospital, which made it difficult to attribute any observed differences to the intervention rather than to other site‐specific variables.
Lagarde 2016	The participants in this study are not in outpatient care facilities.
Lee 2010	Participants of this study were patients. Although the physicians could enrol individual patients in the programme and receive regular reimbursement, there is no description of physicians, and not all the physicians come from outpatient facilities.
Lee 2015	The participants of this study are patients.
Lezzi 2014	Not an RCT, CBA, or ITS study design; a panel count data model is used
Lix 2016	This study is not designed as a CBA, but rather a retrospective case‐control study.
Maini 2014	The participants of this study are facilities, and the intervention objects are patients.
Merilind 2015	This study is not designed as an ITS. It compared childhood immunisation coverage rates of all Estonian family physicians in 2 groups, joined and not joined to the quality system, during the observation period of 2006 to 2012. Also, the family physicians had joined the quality system and were not assigned by investigators in 2006.
Michel 2015	Cross‐sectional study
Mullen 2009	Provider is not clear.
Murray 1992	Cross‐sectional study
Odesjo 2015	The participants of this study are practices.
Olivier 2015	Not a CBA study design; no control group
Petersen 2016	The participants in this study are in‐hospital, and the incentives are for both physicians and facility.
Ritchie 1992	Not an ITS study; it only measured 7 quarters after intervention
Robertson 2017	This study used quality measure, plan‐do‐study‐act, root‐cause analyse, lean six sigma.
Roski 2003	The participants of the study are clinics.
Rudasingwa 2017	The participants of this study are healthcare facilities.
Shelley 2012	Not about payment methods
Shen 2003	The provider is not clear.
Shen 2017a	The participants of the study are health centres.
Shen 2017b	The participants of the study are not ambulatory care professionals.
Sicsic 2015	Not a CBA study; only a before‐and‐after study
Simonsen 2017	The intervention in this study is not related to payment methods.
Simpson 2011	The participants of this study were patients. Physicians are only mentioned in the background (intervention can increase a doctor's income).
Stearns 1992	Not a CBA study; only a before‐and‐after study
Sun 2016	The participants of this study are healthcare facilities.
To 2015	Cross‐sectional study
van Dijk 2014	Not a CBA study design, just a simple before‐after comparative study
van Dijk 2015	Not a CBA study; no control group
Vats 2014	The participants of this study are primary care practices.
Wei 2015	The intervention in this study was just an assumption.
White 2006	The participant is not health care professionals in ambulatory care facilities.

Methods	Cluster randomised trial
Participants	Physicians
Interventions	Illinosisi, US Intervention 1: Enhanced Pay‐for‐performance. Providing maximum Pay‐for‐perfomrance bonuses larger than previous years by a mean of USD 3355 per physician, an approximately 32% increase in bonus size and an increase of USD 16 per patient. Intervention 2: Enhanced Pay‐for‐performance plus loss aversion. Prefunded incentives are put in a virtual health system bank account in the physician name. If at the end of intervention year physicians earned less pay‐for‐performance bonus than were placed in the virtual accounts, the physician were required to pay back funds. Intervention 3: Enhanced Pay‐for‐performance plus increased social pressure. Providing increased proportion of pay‐for‐performance bonus determined by group performance from 30% to 50%.
Outcomes	The proportion of 20 evidence‐based quality measures achieved at the patient level
Notes

Study name	Transforming primary care payment in Hawaii
Methods	Randomised trial
Participants	Primary care providers
Interventions	Intervention 1:Pay‐for‐performance (P4P). A quality incentive payment based upon attainment of 16 quality metrics, and a possible bonus payment for savings in total cost of care at the provider organisation level. Intervention 2:P4P + social comparisons. Providers will also receive weekly emails that will show comparisons of their own performance against their peers within the same provider organisation on specific quality measures and total cost of care. Intervention 3:P4P + social comparisons + Hemoglobin A1cmember/provider incentive. There is also a shared incentive between the member and the provider. The member incentive will be a payment made to diabetic patients with an A1C of greater than or equal to 9% who experience a reduction of at least 0.5%. Each participating member and primary care providers can receive up to USD 75 per quarter for A1C reduction. Control:FFS.
Outcomes	Quantity of health care provided Improvement in Hemoglobin A1c among poorly controlled diabetics Quality of health care provided Improvement in provider performance Cost Primary care spending
Starting date	April 2016
Contact information	ClinicalTrials.gov Identifier: NCT02731716
Notes

PERMALINK

Payment methods for healthcare providers working in outpatient healthcare settings

Liying Jia

Qingyue Meng

Anthony Scott

Beibei Yuan

Lu Zhang

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

Description of the intervention

How the intervention might work

Salaried payment

Capitation

Fee for service

Pay for performance

Blended payments

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

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Grey literature

Trial registries

Data collection and analysis

Selection of studies

1.

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

1. Outcome measures of included studies.

2. Payment methods for the vulnerable populations.

Sensitivity analysis

Summary of findings and assessment of the certainty of the evidence

Results

Description of studies

Results of the search

Included studies

Study design

Participants and setting

Interventions and comparisons

3. Interventions and comparisons in included studies.

1. P4P plus existing payment methods compared with existing payment methods

2. FFS compared with existing payment methods

3. FFS mixed with existing payment methods compared with existing payment methods

4. Enhanced FFS compared with FFS

Study outcomes

Primary outcomes

Service provision process outcomes

Patient outcomes

Healthcare provider outcomes

Costs

Adverse effects

Secondary outcomes

Excluded studies

Risk of bias in included studies

Allocation