Service-level interventions to reduce waiting time in outpatient and community health settings may be sustained: a systematic review

Abstract

Aim

The primary aim of this systematic review of the literature was to determine whether interventions to reduce waiting time in outpatient and community health services can be sustained. The secondary aim was to describe associations between sustainability and features of waiting time interventions and the settings in which they have been implemented.

Methods

CINAHL, Medline, Embase and Psych Info databases were searched, combining the search concepts ‘waiting time or waiting lists’, ‘outpatient or community care’ and ‘sustainability’. Studies were included if they tested a service-level intervention that aimed to reduce waiting in an outpatient or community setting and reported data with a minimum 12-month follow-up period. Data were extracted and analysed using a descriptive synthesis. Methodological quality was evaluated using the mixed-methods appraisal tool (MMAT). Waiting interventions were rated as sustained, partially sustained or not sustained using predetermined criteria. The Grading of Recommendation, Assessment, Development and Evaluation was used to describe certainty of evidence for different intervention approaches.

Results

Screening of 7770 studies yielded 22 papers investigating the sustainability of waiting interventions for approximately 150 000 clients. Many were of lesser quality, with 14 not meeting more than 3 of 5 criteria on the MMAT checklist. Intervention types were categorised as referral entry, open access and substitution, used either alone or in combination. There was low certainty evidence that all interventions were associated with sustained reductions in waiting time, often with large effect sizes, but the findings are limited by low methodological quality of many studies and the risk of publication bias.

Conclusion

Reductions in wait times and waiting lists for health services can be achieved and sustained following interventions, but further high-quality research would better inform service providers about what interventions are most effective and provide the greatest return on investment.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Service-level interventions in community health and outpatient clinics can reduce waiting, but evaluations do not often report on the sustainability of outcomes.

WHAT THIS STUDY ADDS

• This review suggests interventions that embed changes to referral management processes and/or models of care can lead to sustained improvements in timely access. Existing evidence provides examples of strategies that have been successful in sustained reductions in waiting time in community and outpatient services that could provide initial guidance for service providers addressing long wait times.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• From a large body of evidence of over 150 000 patients/clients, there is low to very low certainty evidence that service-level interventions can lead to sustained reductions in waiting time. Further high-quality research would better inform service providers about what interventions are most effective and provide the greatest return on investment

Introduction

Timely access to health services is vital for efficient and effective delivery of care in outpatient and community settings. For patients, receiving the right service at the right time reduces the impacts of health conditions and improves the quality of life.¹ With increasing demand and inefficiencies in service delivery, however, many patients accessing healthcare in these settings experience long wait times. Factors increasing demand include an ageing population, increased chronic conditions and an increasing population, while supply is impacted by reduced workforce capacity and resource constraints.^{2 3} Research suggests that service-level strategies can reduce wait time for health services¹³,⁶ but are often measured over a short period, raising questions about long-term sustainability.^{45 7},¹⁰

Interventions to address waiting time for health services can involve changes at the policy level (such as funding incentives), service level (such as organisational policy or model of care changes) or patient/clinician behaviour (through choices regarding attendance, referrals or individual treatment). At the service level, a range of different strategies have been used to categorise waiting time interventions in outpatient and community settings. For example, a systematic review evaluating interventions to reduce wait times in paediatric outpatient and community services identified three broad categories of strategies to reduce wait times in health services, described as rapid access (streamlining access to an initial assessment appointment), process efficiencies (activities to improve workflow and reduce waste) or substitution (workforce innovations to use less specialised staff in advanced scope roles).⁹ Other authors have categorised waiting time interventions using terms, such as resource realignment, operational efficiency and process improvement,⁶ or according to stages of the patient journey, such as those focusing on referral management versus models of care.¹¹ Variation in the description of waiting list interventions may reflect diversity in the strategies that have been used across different healthcare settings. The low quality of the research in this field has been identified as a limitation in all these reviews, but all report consistent findings that different types of service-level interventions can reduce waitlists and wait times in outpatient and community settings.^{16 9 12},¹⁴ While two reviews reported on follow-up periods of the included studies,^{6 14} none addressed the issue of sustainability of reducing waiting time during analysis.

Studies evaluating waiting time interventions often report relatively short follow-up periods, raising questions about long-term sustainability.^{45 7},^{10 15} Sustainability has been defined as ‘the ongoing delivery of health programmes, which may be measured by the longevity of independent projects, or how well programmes become institutionalised in organisations or health and social systems’ (Braithwaite, 2017, p. 1580). Evidence from several systematic reviews suggests that health service interventions, in general, are more likely to be sustained when supported by strong leadership, sufficient resources and adequate planning for sustainability, and when specific barriers, facilitators and necessary adaptations of the intervention to the service are considered.¹⁷,²⁰ Klaic et al recognised the level to which a new intervention is integrated into an organisation’s service setting and subsystems, described as penetration, and also has a strong impact on the sustainability of the intervention.²¹ An intervention with successful penetration no longer requires active monitoring, which supports the likelihood of sustainability once any initial supports and/or funding have ceased.^{20 21}

While several systematic reviews provide insights into factors associated with the sustainability of health service interventions, in general, no reviews to date have explored the sustainability of interventions to reduce waiting times. The primary aim of this systematic review was, therefore, to determine whether interventions implemented at the service level to reduce wait times in outpatient and community health services can be sustained. The secondary aim was to describe the features of wait time interventions and the settings in which they have been implemented that are associated with sustainability.

Method

This systematic review was registered prospectively with PROSPERO CRD 42022346200 (23/7/2022) and is reported consistent with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis.²²

Consumer involvement:

The topic for this systematic review has been informed by our previous qualitative work with consumers waiting for healthcare. Consumers told us that delays in access to services have an impact on health outcomes, as well as causing distress and anxiety, and that sustainable reduction of healthcare waiting lists is a high priority for health systems. Consumers were not directly involved in the design of the protocol or conduct of this review.

Search strategy

A search strategy was designed to identify studies that evaluated the sustainability of service-level interventions in healthcare organisations to reduce wait times. Electronic databases CINAHL, Medline, Embase and Psych Info were searched from the earliest available date to August 2022.

The search strategy combined the key concepts of outpatient health services, waiting and sustainability. Synonyms for each concept (eg, ambulatory, outpatient and community health; waiting list, waiting time and timely access; and sustainability, maintenance and continuation) were combined with ‘OR’ with results between concepts combined with ‘AND’. Full search strategies can be seen in online supplemental file 1. The citations and references list for included studies were manually checked to identify further studies that met the inclusion criteria that may have been missed in database searching.

Inclusion and exclusion criteria

Any intervention implemented at the service level for the purpose of reducing wait time from referral to initial appointment for assessment and commencement of treatment with a service or reducing the number of people on waiting lists for care was eligible for inclusion, including but not limited to process improvements, changes to models of care (such as use of telehealth or group interventions), workforce adjustments or referral processing systems. For the purpose of this review, sustainability was measured as maintenance of improvements for at least 12 months following the implementation of an intervention or service change. Any study design was, therefore, included, provided a quantitative waiting outcome (eg, mean/median waiting times or number of referrals on the waitlist) that was reported a minimum of three times (preintervention, postintervention and at a minimum of 12 months follow-up). This systematic review included any health service delivered in an outpatient or community setting, including but not limited to diagnostic services, allied health services, medical outpatient clinics, primary care or mental health clinics. Studies were excluded if they described interventions to reduce wait times for surgery (including organ transplant), emergency services (including ambulance arrival and hospital emergency departments), inpatient hospital care, school-based services or were not in English.

Article selection and data extraction

All potentially eligible studies were imported into review software, Covidence,²³ and duplicate references deleted. A review of the titles and abstracts was conducted by two reviewers independently to determine whether studies met the inclusion criteria. Studies that did not meet the inclusion criteria were excluded. Discrepancies were resolved through discussion between reviewers, and full-text manuscripts were obtained where doubt persisted.

Studies that met inclusion criteria based on title/abstract screening were subject to a full-text review by two reviewers, independently. When discrepancies arose, they were discussed between the reviewers. If the reviewers did not agree on an outcome, an additional reviewer was consulted.

Data were extracted from all studies using a predesigned data extraction table that included the study characteristics, research design, setting, details of intervention, outcomes measured, baseline measurement, postintervention measurement, follow-up measurement and the key sustainability findings. Data extraction was completed by two reviewers independently.

Risk of bias

Methodological quality was evaluated using the mixed-methods appraisal tool (MMAT).^{24 25} The MMAT provides checklists for researchers to critically review the methodological quality of qualitative research, randomised controlled trials, non-randomised studies, quantitative descriptive studies and mixed-methods studies. Adapted to each study design, five questions are asked to critique the methodological quality of each paper. Two researchers independently assessed the methodological quality for each included paper, with any discrepancies discussed until a consensus was achieved. A third researcher was available for consultation if consensus could not be achieved.

Analysis

A deductive narrative synthesis was conducted to group the included papers by intervention strategy. Three researchers (KM, KEH and AL) independently assigned interventions into categories. Reviewers then discussed their findings with any disagreements resolved by discussion until consensus was achieved.

The primary outcome was the sustainability of results following the implementation of service-level interventions to reduce wait times. Given the lack of guidance on a definition of sustainability for waiting time interventions in the literature, a classification system was developed for the purpose of this review based on a pragmatic judgement of clinical significance. The level of sustainability of each study was judged as ‘sustained’ (wait time/waitlist has reduced from baseline to postintervention and remained below postintervention levels at follow-up), ‘partially sustained’ (indicated by small increases in waiting time at follow-up of <10% or inconsistent results across subgroups or outcome measures) or ‘not sustained’ (wait times/waitlist increased by>10% at follow-up compared with postintervention levels).

We then applied the Grading of Recommendation, Assessment, Development and Evaluation (GRADE) to all studies within each of the intervention types to analyse the certainty of evidence, adapted for studies without a single estimate of effect.²⁴ For each outcome, the GRADE domains (methodological limitations of the studies, directness of the evidence, precision of estimates, consistency of the evidence and the likelihood of publication bias) were rated as not serious, borderline or serious (table 1). Overall assessment of the strength of the evidence for each outcome was judged to be high, moderate, low or very low. Strength of evidence from observational studies was initially assessed as being low, with potential for upgrade (due to magnitude of effect) or downgrade (due to at least one serious concern or two or more borderline concerns among the five GRADE domains).

Table 1. Criteria used to determine the GRADE category, after starting at ‘low’ for non-randomised studies.

GRADE domain	Description
Reason for downgrade
Methodological limitations	Not serious: all studies rated ‘Yes’ for four or five criteria of the MMATBorderline: at least 50% rated ‘Yes’ for four or five criteriaSerious: more than half rated ‘Yes’ for less than four criteria
Indirectness	Not serious: all provide direct evidence related to wait timesBorderline: data from less than half of studies can be indirect but not likely to cause serious risk of biasSerious: multiple studies provide indirect evidence causing serious concerns for risk of bias
Imprecision	Not serious: majority of studies report on population size, and total participants>400 OR narrow CIs reportedBorderline: total reported participants>400 but not reported by>half of studiesSerious: total reported participants<400
Inconsistency	Not serious: all studies consistent in direction and magnitude of effectBorderline: >75% consistent in direction of effect, variation in magnitudeSerious: substantial variation in direction and magnitude of effect across studies
Publication bias	Seriousness of concerns judged using funnel plots where sufficient data are available for this analysis.
Reason for upgrade
Magnitude of effect	Large effect size (>0.8) or>66% reduction in waiting time or number of people on the waiting list sustained in at least 66% of studies

GRADEGrading of Recommendation, Assessment, Development and EvaluationMMATmixed-methods appraisal tool

Results

Yield

The search of databases identified 7770 papers after removal of duplicates (figure 1). 82 papers were retained for full-text review. A further 60 papers were excluded during the full-text review with the most common reasons being wrong publication type (n=16), less than 12 month follow-up (n=13) and wrong outcomes (n=11), yielding 22 papers for inclusion.

Figure 1. Flowchart of studies identified as relevant for inclusion in review.

Study characteristics

The 22 studies included in this review were conducted in the UK (n=5), Canada (n=5), USA (n=5), Sweden (n=2), Australia (n=2), Spain (n=1), Netherlands (n=1) and New Zealand (n=1). Ten were set in outpatient specialist clinics,^{715 26},³³ four in mental health clinics³⁴,³⁷ and four allied health clinics,³⁸,⁴¹ three in MRI services⁴²,⁴⁴ and one in an adolescent medicine clinic.⁴⁵ Population numbers were not reported consistently with seven papers providing no indication of the sample size in their research.^{26 31 32 35 41 43 44} The population numbers reported across the 22 studies were approximately 200 000, although reporting was inconsistent. Characteristics of all included studies are summarised in table 2.

Table 2. Studies reporting healthcare interventions to address wait times and waitlists and their sustainability.

Author	Setting	Design	Key intervention components	Outcomes measured	Baseline measure	Outcome postintervention	Outcome at Follow-up	Key finding
Bhuva et al*42	Medical imaging clinic, UK	Single site, quarterly continuous observationsn=15	Reduced complexity ‘one-stop' service model (OA)	Median (IQR) wait times (days)	60 (34–71)	Year 1: 25 (estimate)	Year 2: 15 (4-45)	Sustained
Speed and Fyffe¶40	Occupational therapy clinic, Canada	Single site, interrupted time series, weekly observationsn=7524	Protected appointments for new patientsDirect booking to an appointment on referral (OA)	Mean wait times (days)	24.5 (stable over 3 years)	Year 1: 21 days (no change in slope)	Year 2: No change in slope, reduced variability	Not sustained
Speed§39	Physiotherapy clinic, Canada	Single site, interrupted time series, weekly observationsn=>10 000	Protected appointments for new patientsDirect booking to an appointment on referral (OA)	Wait times (days)	Increasing trend over 3 years prior	2 years post: decreasing trend (P<0.01)		Sustained
Strindhall and Henriks*41	Hospital and primary clinics, Sweden	Multiple sites (n=11), repeated measures at baseline, postintervention, follow-upn=not provided	Analysis of supply and demand.Redesigning the system to meet demand(OA)	Median wait times (days)	90	Year 1: 7	Year 7: 14	Sustained
van der Voort et al†32	Outpatient specialist clinic,Netherlands	Multiple sites (n=14), Repeated measures: baseline, postintervention, follow-upn=not provided	Introduction of the advanced access model49 (OA)	Mean wait times across all sites (days)Range of waiting times across sites	415–128	Year 1: 201–81 (all sites improved)	Year 3: 181–45 (10/14 sites sustained)	Partially sustained
Willis et al†33	Urology outpatients,UK	Multiple sites (n=0.18),Repeated measures (baseline, postintervention, follow-up) n>10 000	Analysis of supply and demand; Matching demand/supply; ‘One-stop’ clinic(OA)	Median wait time (days)	54	Year 1: 9	Year 2: 9	Sustained
Bhullaret al‡44	Medical imaging service,New Zealand	Single site, continuous weekly observationsn=not provided	More efficient triage system; waiting list segmentation and new scheduling process(RE)	Median wait times (days)Waitlist (number of patients)	96.41954	First 6 months:Not reported600	Year 2: 23.1413	Sustained
Hankinsonet al†26	Urology clinic, USA	Multiple sites, continuous quarterly observationsn=not provided	Introduction of the advanced access model49 (RE)	Mean wait times (days)	100.6	Year 1: 42	Year 4: 14.2	Sustained
Hazlewood et al‡27	Outpatient rheumatology clinic, Canada	Single site, annual observations n>10 000	Implementation of a centralised referral and triage system (RE)	Mean (SD) wait (days)Routine referralsModerate referralsUrgent referrals	155 (88)110 (5)729 (46)	Year 2:149 (65)78 (56)18 (23)	Year 7:25012025	Not sustained
Miller et al¶45	Adolescent medicine clinical trial units, USA	Multiple sites,eight sites, time-series analysis, quarterly observationsn=2142	Structural changes to referral processes and use of specialists (RE)	Average wait times (days)% patients linked within 42 days	Range 150–30062%	Year 1: Over 10075%	Year 5: 4487%	Sustained
Rey-Aldana et al¶15	Cardiology service,Spain	Single site, interrupted time series, monthly observationsn = >10 000	Initial e-consultation (RE)	Mean wait times (days)	~ 95	Year 1: ~ 15	Year 7: ~ 10	Sustained
Young and Wachter¶37	Psychiatric outpatients clinic,USA	Single site, interrupted time series, monthly observations over 3.5 yearsn=423	Toyota production system50(RE)	Median wait time (days)	33.1	Year 1: 22.8	Year 2: 21.7	Sustained
Creenet al¶38	Paediatric outpatient clinic, Australia	Single site, repeated measures (baseline, postintervention, follow-up)n=225	Advanced allied health practitioners’ completing initial assessment prior to medical practitioner (S)	Mean rank of wait times (days) over a 6-month period	169	First 6 months120	18 months to 2 years post: 48	Sustained
Krishnanet al†28	Trans ischaemic attack clinic,UK	Single site, continuous monthly observations n>2765	Reallocation of resources and increased consultation (S)	Mean wait times (days)	9	Year 1:3	End of Year 2:2.5	Sustained
Haggarty et al¶34	Mental health specialist clinic, Canada	Controlled (1 × intervention, 4 × comparison sites), annual observationsn=2476	Colocation of primary care provider with mental health services (RE, OA)	Mean wait times (days)	97.6 (73.0)Comparison site: 97.6 (73)	Year 1:45.3 (34.4)Comparison site: 85 (66.4)	Year 3: 42.2 (48.3)Comparison site: 84.4 (69.3)	Sustained
Erikssonet al†7	Two outpatient clinics in Sweden: Rheumatology Dermatology	Single site, continuous monthly observations n>10 000	Analysis of demand and capacity variations; more efficient referral processes to reduce demand (RE, OA)	% patients first visit<3 months Number of patients on list	55%3000	Year 1:80%Year 1: 2100	Year 6:100%Year 4: 400	Sustained
Keller†35	Outpatient mental health service,USA	Single site, continuous monthly observationsNot provided	New mental health roles. Preserved capacity for new appointments and expansion of groups(RE, OA)	Mean wait times (days)	21	Year 1:12	Year 2:6	Sustained
Kinnanet al‡36	Federally qualified psychiatric service, USA	Single site, continuous monthly observations n>10 000	New staff, optimising consult clinic and e-consults (RE, OA)	Waitlist (number of patients)	350	Year 1: 252	Year 3: 1	Sustained
Patterson et al†29	Neurology clinic,Northern Ireland, UK	Single site, annual observationsn=120	Email triage system(RE, OA)	Mean annual wait times (weeks)	72	Year 1: 52	Year 5: 4	Sustained
Poyah and Quraishi§30	Renal clinic,Canada	Single site, continuous quarterly observationsn=5620	New triage and booking system(RE, OA)	Number on waitlistMean wait times (days)	1019Urgent: 8Semiurgent: 32	Year 1: 873Year 1: 10Year 1: 26	Year 5: 578Year 4: 5Year 4: 34	Partially sustained
Schoch and Adair‡31	Orthopaedic outpatients,Australia	Single site, continuous monthly observationsn=not provided	Reallocation of resources, patient-focused bookings (RE, OA, S)	Waitlist (number of patients)Mean wait (months)	1100+23	After 3.5 years~350Not reported	Year 5: ~370Year 5: <3	Sustained
Greaveset al†43	Nuclear medicine clinic, UK	Single site, annual observationsn=unclear	Improvement to referral process and scheduling of appointments, implementing advanced practice roles and patient advice line and sharing of resources (RE, OA, S)	Mean wait times (weeks)	42	Year 2: 13	Year 5: 2	Sustained

Number of stars (*,**,***,****,*****) correspond with the number of MMAT quality criteria met by each study.

^*indicates 1 of 5 quality criteria met on the MMAT checklist.

^†indicates 2 of 5 quality criteria met on the MMAT checklist.

^‡indicates 3 of 5 quality criteria met on the MMAT checklist.

^§indicates 4 of 5 quality criteria met on the MMAT checklist.

^¶indicates 5 of 5 quality criteria met on the MMAT checklist.

MMATmixed-methods appraisal toolOAopen accessREreferral entrySsubstitution

There were no randomised trials. All studies were historically controlled, providing data collected during periods before and after the intervention either using simple, uncontrolled pre–post designs, controlled before and after studies or interrupted time series. Five studies conducted time-series analyses.^{15 37 39 40 45} 17 studies implemented their intervention at a single site.^{715 27},^{31 35} Four studies introduced the intervention across multiple sites^{26 32 33 41} and one had a comparison site.³⁴ There were insufficient data across the studies to formally explore the likelihood of publication bias using funnel plots.

18 studies collected observations continuously over regular time periods.^{715 26},^{31 34} Three measured outcomes at weekly intervals,^{39 40 44} seven at monthly intervals,715 28 31 35,³⁷ four at quarterly intervals^{26 30 42 45} and four used annual observations.^{27 29 34 43} The remaining four studies used repeated measures at key time points,^{32 33 38 41} which typically included baseline, postintervention and follow-up.

The follow-up period varied from 18 months³⁸ to 7 years.^{15 27 41} The most common time measurement was 2 years postintervention (n=7 studies),^{28 33 35 39 40 42 44} followed by 5 years (n=5).²⁹,^{3143 45}

Risk of bias

Overall, six studies met all five MMAT criteria.^{15 34 37 38 40 45} Less than half of the studies described the target population. The outcome measures and the intervention were sufficiently described in 21 of the 22 studies.^{715 26},^{40 42} 13 studies were rated as having complete outcome data715 26 27 30 34 36,^{40 44 45} with insufficient information to assess this in 9 studies.^{2829 31},^{33 35 41} Confounders were accounted for in 8 of the studies^{15 28 34 37 38 40 43 45} and 16 of the studies reported enough information for the reviewers to ascertain if the intervention was implemented as intended.¹⁵²⁹,^{41 44 45} Full details of the MMAT ratings for all studies can be found in the online supplemental file 2.

Study findings

After conducting a deductive thematic analysis, the interventions in the included papers were grouped into three categories: referral entry, open access and substitution, used either as stand-alone interventions or in combination (figure 2). Referral entry interventions targeted the process from entry to acceptance of a referral within an organisation and encompassed interventions that developed or improved triage systems, such as the use of e-referrals and centralised intake systems. Open access interventions involved preserving space for new appointments based on demand, including systems, such as advanced access, the choice and partnership approach and the Specific Timely Assessment for Triage (STAT) model that are based on this principle. Substitution involved training less specialised workers to perform aspects of care usually conducted by those with a higher level of training, such as the use of nurse practitioners, allied health staff or more junior medical staff to complete tasks historically completed by medical specialists.

Figure 2. Service-level intervention types.

14 studies were considered to have used interventions within a single category, and 8 used a combination of intervention types. Most incorporated additional process efficiency measures, but these were integrated within these broader categories and were, therefore, not considered as a stand-alone category. While some studies accessed short-term resources dedicated to supporting the implementation of an intervention, all interventions were presented as ongoing changes to the service; no studies described time-limited interventions, such as temporary increase in supply to address backlog.

Referral entry interventions

There was low certainty evidence from six studies that reduced wait times achieved from referral entry interventions were sustained over time (table 3). Referral entry interventions were used in mental health services (n=1),³⁷ a service for HIV-infected youth (n=1),⁴⁵ medical resonance imaging clinic (n=1),⁴⁴ urology clinic (n=1),²⁶ cardiology service (n=1)¹⁵ and an outpatient rheumatology clinic (n=1).²⁷ Miller et al changed referral screening and intake processes in an HIV clinic, with sustained improvements in linkage and waiting times reduced by more than two-thirds.⁴⁵ Interventions introduced by Bhullar et al⁴⁴ and Hankinson et al²⁶ focused on grouping like services and reducing variability of intervention types, with wait times reduced by more than two-thirds and maintained at 2 and 4 years, respectively. Two studies reported implementing a centralised system for intake and triage, finding the quality of referrals improved and wait times were managed more effectively,^{15 27} although only one of these was able to achieve and maintain reduced waiting times.¹⁵ Overall, five of the six studies saw a reduction in wait times after the minimum 1-year post intervention follow-up of which four had large enough effect sizes to upgrade the rating of certainty of the evidence by one GRADE level.

Table 3. Summary of the certainty of evidence.

Outcome	Effect	Number of participants	Certainty of evidence
Sustainability of referral entry interventions	Six studies implemented referral entry interventions with five sustaining reduced wait times for 2 years. One did not achieve sustainability.	>100 000	LowDowngraded one level due to methodological limitationsUpgraded one level due to the magnitude of effect
Sustainability of open access interventions	Six studies implemented open access interventions. Four reported sustained effect. One did not attain sustainability after 2 years39 and one partially attained sustainability after 3 years15.	>20 000 reported across three studies	LowDowngraded one level due to serious risk of bias related to methodological limitations and borderline concerns about imprecision and consistency.Upgraded one level due to the magnitude of effect
Sustainability of substitution intervention types	Two studies implemented substitution intervention type. Sustainability was attained over an 18-month to 2-year period for one and at the end of 2.5 years for the second.	2990(n=225 and n=2765)	LowDowngraded one level due to serious concerns in methodological limitations in one of the two studies.Upgraded one level due to the magnitude of effect
Sustainability of combined intervention types, including referral entry and open access	Eight studies included both referral entry and open access intervention types. Seven attained sustainability with one partially sustained.	>50 000	Very lowDowngraded one level due to serious concerns in methodology and imprecision

Open access interventions

There was low certainty evidence of sustained reductions in wait times from six studies following the introduction of open access interventions (table 3). These studies were implemented in allied health clinics (n=3), an outpatient specialist clinic (n=1), a urology clinic (n=1) and an MRI clinic (n=1). All studies clearly reported implementing open/advanced access strategies^{3233 39},⁴² with three stating the relevance of understanding the supply and demand for each service as important to their intervention.^{32 33 40} Four studies were classified as demonstrating sustained improvements, with three describing wait reductions of more than two-thirds^{33 41 42} and one demonstrating a significantly reducing trend in waiting time on an interrupted time-series analysis.⁴⁰ Speed and Fyfe⁴⁰ were able to reduce waiting variability but not waiting time over the 2 years of their trial in an occupational therapy service, and one open access trial in medical outpatient clinics was judged as being partially sustained due to improvements being maintained at 10 of 14 trial sites.³²

Substitution

There was low certainty of evidence from two studies that substitution strategies sustained reduced wait times (table 3). Of these, one study achieved a high methodological quality meeting all criteria on the MMAT³⁸ and the second had low methodological quality meeting only two criteria.²⁸

Both studies reported that a substitution-only intervention achieved sustained reductions to wait times. One study described a mean wait time reduction from 169 to 48 days in an allied health clinic after implementing a strategy in which an advanced allied health practitioner assessed new referrals.³⁸ The remaining study reported a reduction in wait times at an outpatient specialist clinic from 9 days to 2.5 after upskilling support workers to complete tasks that were, otherwise, completed by the specialist nurses.²⁸

Combined interventions

There was very low certainty evidence that combined interventions sustained wait times (table 3). Eight studies introduced combined interventions that included elements from both referral entry and open access⁷²⁹,^{31 34} and two studies combined all three intervention types.^{31 43} The combination of these intervention types typically aimed to create new processes for ensuring that appropriate referrals were accepted into the service, triaged and provided appointments to the most appropriate service in a timely manner. Seven of the eight studies were considered to have sustained reduced wait times,^{7 29 31 34 35 43} often with very large effect sizes; all reduced and maintained waiting times or the number of people on the waiting list at less than 50% of preintervention levels, and three reduced wait times by more than 90% over 5 years.^{29 31 43} One study was considered partially sustained due to maintained improvements in the number of people on the waiting list, but reported limited effect for waiting time.³⁰

Discussion

This review of 22 studies investigated the sustainability of service-level interventions delivered to over 150 000 people to reduce wait times in outpatient and community health settings. Overall, we found low to very low certainty evidence that referral entry, open access and substitution interventions implemented either alone or in combination were effective in sustaining reductions in waiting in outpatient and community health services. Most studies introduced a range of general process efficiencies alongside other interventions to support implementation and sustainability. 18 of the 22 studies reported sustained wait time and/or waitlist reductions at a minimum of 12-months postintervention, many with very large effect sizes. Two studies partially sustained reduced wait times and two studies were unable to achieve sustainability. No single type of intervention appeared to be more commonly associated with sustained reductions in wait time than others. Our review adds to the literature by demonstrating that implementing service-level interventions may lead to sustained reductions in wait times across diverse community and outpatient health services, despite well-described challenges in maintaining the outcomes of service improvement interventions.^{13 17 18 20}

The interventions identified in these studies (referral entry, service access and substitution) shared common elements with previous reviews of wait time interventions but also had differences. Previous research considered referral entry and open access as a single intervention type,⁹ but they presented as two distinct categories in this group of studies. Substitution was classified in line with previous research but was only used in two studies as the primary intervention. Process efficiencies were not identified as a stand-alone intervention in this group of studies but were reported as being implemented alongside other primary interventions. The finding that no single category of intervention was superior in achieving sustained reductions in wait times is consistent with previous literature.^{9 12}

Explicit use of additional resources was reported in eight of the studies,^{728 31 33},^{36 38} and it is possible that this addition of resources to a service (presumably increasing supply) is primarily responsible for the reduction in waiting times. While there was no clear pattern, all studies that specified additional resources did sustain reduced wait times, although there were also many examples of interventions that resulted in sustained wait time reductions without additional resources. Resources can be used to directly increase supply, but previous research has suggested that, in isolation, this approach is unlikely to result in sustained change.^{8 46} Consistent with this finding, none of the studies in this review reported using additional resources to temporarily increase supply within the existing model of care. Rather, any additional resources were used to support new models of care, such as adding an allied health practitioner³⁸ or a patient navigator³⁶ to the team. The additional resources were, therefore, not only used to supplement the supply available but also to disrupt the existing practices and act as an enabler to ongoing change.

One factor that is likely to have contributed to the sustained changes reported in these studies is that all studies used strategies that were implemented as ongoing changes to the service. For example, changes to procedures and documentation for triaging referrals to clarify eligibility and identify appropriate services^{1527 29},^{31 38 39} or appoint an allied health practitioner to reduce the load on more specialised health practitioners.³⁸ Other studies used an analysis of supply and demand to inform the implementation of protected appointments for new patients or changes to the appointment types offered.^{7 26 30 33 43} All these interventions were implemented as new and ongoing processes that were not limited to a specific time period. Process efficiencies were frequently described and implemented alongside other interventions, often removing wasteful processes, consistent with ‘Lean’ methodology.^{37 47} These process changes may have supported change in behaviour that may have helped to achieve penetration of the intervention and sustained improvements.

The quality of reporting for these studies was frequently low, contributing to a serious risk of bias and low certainty in the findings. The lack of clear descriptions of the interventions and reporting of fidelity in these studies results in difficulty for others to replicate the research process. These findings are aligned with research into the sustainability of service-level interventions in the literature,³ as well as findings from other reviews of waiting time interventions.^{6 9 14} With more comprehensive reporting, key components contributing to the sustainability of a service-level intervention would be better understood.⁴⁸ We would then be better positioned to draw conclusions about what works and to confidently build on the previous research.

Limitations of this review include the possibility of publication bias influencing the evidence in this field; however, this could not be formally assessed due to limitations of the available data. Only two studies reported on interventions that were not sustained at follow-up. Most studies in this review were framed as quality improvement activities or retrospectively described the impact of interventions that were intended to improve services, rather than being prospectively planned as research trials. This may increase the likelihood that those leading these types of innovations are more motivated to share a success story in which an intervention worked and achieved sustained change than those that have been unsuccessful. Defining sustainability as improvement maintained for 12-months post service-level intervention could be viewed as a limitation, as it may not consider the broader facets of sustainability, such as the extent to which changes are embedded in an organisation. However, this criterion enabled clear rules in relation to inclusion and exclusion. Additionally, the use of pragmatic thresholds for interventions to be considered sustained and/or have a large magnitude of effect could be considered a limitation, although the selection of conservative thresholds mitigates the likelihood of overinterpreting the findings as a result of this approach.

Conclusion

This review suggests that reductions in wait times and waiting lists for health services may be achieved and often sustained following interventions at the health service level, but certainty of the evidence is low due to research designs with a high risk of bias and limitations in reporting. Further high-quality research would better inform service providers about what interventions are most effective and provide the greatest return on investment. Ongoing changes to the structure of referral and booking processes and/or changes to models of care that influence the flow of patients through healthcare systems appear to be a common feature in studies reporting sustained improvements in the delivery of timely care.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.