Introduction
Inequalities in health and primary care quality are well documented internationally and are a vital issue for policymakers to tackle (Marmot, Reference Marmot2020, LaVeist et al., Reference Laveist, Perez-Stable, Richard, Anderson, Isaac, Santiago, Okoh, Breen, Farhat, Assenov and Gaskin2023, Karanikolos and Kentikelenis, Reference Karanikolos and Kentikelenis2016). General practitioner (GP) surgeries in poorer areas have fewer doctors per head of population and are more likely to receive poor quality ratings (The Health Foundation, 2020a, Salant et al., Reference Salant, Massou, Awan and Ford2024). Patients living in poorer areas are more likely to report worse experiences of access to general practice, and lower overall satisfaction with the services they receive (The Health Foundation, 2020b). Primary care inequalities are driven by an unequal distribution of the structural determinants of primary care; namely, funding, workforce, and population need (Salant et al., Reference Salant, Massou, Awan and Ford2024, McLean et al., Reference Mclean, Guthrie, Mercer and Watt2015, Kontopantelis et al., Reference Kontopantelis, Mamas, van Marwijk, Ryan, Bower, Guthrie and Doran2018).
While health inequalities can be narrowed by ambitious policies that target the unequal distribution of health’s social determinants (Barr et al., Reference Barr, Higgerson and Whitehead2017, Buck and Maguire, Reference Buck and Maguire2015, Vodden et al., Reference Vodden, Holdroyd, Bentley, Marshall, Barr, Massou and Ford2023, Holdroyd et al., Reference Holdroyd, Vodden, Srinivasan, Kuhn, Bambra and Ford2022), concurrent action must ensure that inequalities across healthcare systems are addressed. Addressing inequalities in funding is imperative given its impact on service provision, workforce, and primary care quality (Salant et al., Reference Salant, Massou, Awan and Ford2024, L’Esperance et al., Reference L’esperance, Sutton, Schofield, Round, Malik, White and Ashworth2017, McLean et al., Reference Mclean, Guthrie, Mercer and Watt2015). For example, in the UK, when funding across the National Health Service was increased and weighting for socioeconomic deprivation improved, this was associated with a reduction in mortality and mortality inequalities (Barr et al., Reference Barr, Bambra and Whitehead2014).
Primary care is predominantly financed by three groups of funding models (Tao et al., Reference Tao, Agerholm and Burström2016). Capitation funding pays primary care providers a set fee per capita, irrespective of the services provided. Given different care needs between patients, the set fee, or patient numbers are adjusted based on a range of factors (usually referred to as ‘weighted capitation’). These differ -between health systems, but can commonly include age, sex, morbidity, and socioeconomic status (Khezri et al., Reference Khezri, Mahboub-Ahari, Tabrizi and Nosratnejad2022). Pay-for-performance schemes reimburse primary care organizations for achieving pre-specified targets of activity or clinical outcomes, such as achieving 85% vaccination coverage or cervical screening. Fee-for-service models reimburse primary care organizations per item of clinical service delivered, such as a fee for each post-natal check completed. Some have argued that a fee-for-service model is beneficial due to its simplicity and feasibility (Ikegami, Reference Ikegami2015), whereas others suggest that pay-for-performance is a beneficial mechanism to ensure funding translates into outcomes (Roland and Olesen, Reference Roland and Olesen2016). Others prefer capitation because it allows better adjustment for population needs (James and Poulsen, Reference James and Poulsen2016). Primary care organizations are often funded by multiple funding models; this is known as blended financing.
There are widespread efforts internationally to reduce health inequalities, and much evidence that primary care has an important role to play in this (Gkiouleka et al., Reference Gkiouleka, Wong, Sowden, Bambra, Siersbaek, Manji, Moseley, Harmston, Kuhn and Ford2023, Roland and Everington, Reference Roland and Everington2016, McLean et al., Reference Mclean, Guthrie, Mercer and Watt2015). Understanding how structural features of primary care – for example, how health systems arrange funding for primary care services – impact health inequalities is an important step to maximizing primary care’s ability to reduce health inequalities.
Several systematic reviews have examined the effect of different models of primary care funding, with some investigating their impact on health inequalities. However, conclusions differ widely, and reviews have not been synthesized. This paper addresses this research gap by evaluating all published reviews that examine the effect of primary care funding models on inequalities in funding, access, clinical outcomes, or experience.
Methods
This umbrella review was conducted in accordance with the established methodology (Higgins et al., Reference Higgins, Thomas, Chandler, Cumpston, Li, Page and Welch2019) and reported in line with the Preferred Reporting Items for Systematic reviews and Meta-Analyses statement (Page et al., Reference Page, Mckenzie, Bossuyt, Boutron, Hoffmann, Mulrow, Shamseer, Tetzlaff, Akl, Brennan, Chou, Glanville, Grimshaw, Hróbjartsson, Lalu, Li, Loder, Mayo-Wilson, Mcdonald, Mcguinness, Stewart, Thomas, Tricco, Welch, Whiting and Moher2021). The review was preregistered on PROSPERO (CRD42024501203).
Search strategy and selection criteria
Three electronic databases (Ovid Medline, OVID Embase, and Cochrane Reviews) were systematically searched, with no time restrictions, on the 14th of January 2024. Search terms were based on existing search terms for identifying reviews (Gkiouleka et al., Reference Gkiouleka, Wong, Sowden, Bambra, Siersbaek, Manji, Moseley, Harmston, Kuhn and Ford2023), primary care (Gkiouleka et al., Reference Gkiouleka, Wong, Sowden, Bambra, Siersbaek, Manji, Moseley, Harmston, Kuhn and Ford2023), and inequalities (Prady et al., Reference Prady, Uphoff, Power and Golder2018). Search terms regarding funding were based on a range of systematic reviews, forming the most comprehensive search available. The appendix presents the search terms and full search results. Following the manual removal of duplicates, all abstracts and titles were independently screened by two researchers according to inclusion and exclusion criteria using the software Rayyan with conflicts resolved through discussion.
The inclusion criteria were:
• Systematic reviews that used a structured search and included primary studies.
• Reviews in any language.
• Reviews assessing the impact of primary care funding.
• Reviews assessing outcomes of inequalities in funding, access, experience, or clinical outcomes.
Reviews in high-income countries as defined by the world bank.
Exclusion criteria were:
-
• Primary research studies.
-
• Reviews not investigating inequalities.
-
• Reviews reporting non-health inequalities.
Further papers were identified from the Health Equity Evidence Centre’s Living Evidence Maps; a machine learning resource to identify and collate studies that focus on what works to address inequalities in primary care (Health Equity Evidence Centre, 2024). Additionally, all citations of included reviews were screened.
The full text of all identified articles were reviewed by two researchers. For all reviews that met inclusion criteria, the following information was extracted: first author, year of publication, aim, number of studies, time period of analysis, population, primary care funding model(s) being investigated, health inequalities measured, and main findings. Outcomes of interest were any change or difference in inequality occurring due to a change in primary care funding or difference in primary care funding. Quality assessment was performed using the AMSTAR2 tool (Shea et al., Reference Shea, Reeves, Wells, Thuku, Hamel, Moran, Moher, Tugwell, Welch, Kristjansson and Henry2017). Data extraction and quality assessment were performed independently by two authors before any disagreements were resolved with discussion with a third author.
Due to the small number of reviews, a large amount of data heterogenicity, and lack of data in systematic reviews, it was inappropriate to perform meta-analysis. Instead, the review’s findings were narratively synthesized. When reviews differed in their conclusions about the effects of the same intervention, efforts were made to understand the reasons behind these differences and reach a consensus. This involved assessing several factors including whether different definitions were used to frame outcomes, the primary studies identified in each review, the weight assigned to individual studies’ findings, unique considerations specific to each review, and the overall quality of the review. Any significant findings that helped explain the differences in conclusions were incorporated into the narrative synthesis.
Results
After the removal of duplicates, the search identified 2480 unique articles. Sixty-four were reviewed in full text, and 14 were included in the final review. Figure 1 shows a flowchart of included reviews.
Figure 1. PRISMA diagram of study selection criteria.
The 14 reviews were published between 2009 and 2021 (Table 1). Only one compared funding models, comparing capitation reimbursement to fee-for-service (Tao et al., Reference Tao, Agerholm and Burström2016). Twelve reviews investigated the effects of pay-for-performance reimbursement (Ahmed et al., Reference Ahmed, Hashim, Khankhara, Said, Shandakumar, Zaman and Veiga2021, Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010, Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Forbes et al., Reference Forbes, Marchand and Peckham2016, Gillam et al., Reference Gillam, Siriwardena and Steel2012, Gupta and Ayles, Reference Gupta and Ayles2020, Lin et al., Reference Lin, Yin, Huang and Du2016, Mandavia et al., Reference Mandavia, Mehta, Schilder and Mossialos2017, Steel and Willems, Reference Steel and Willems2010, Tao et al., Reference Tao, Agerholm and Burström2016, Van Herck et al., Reference Van Herck, de Smedt, Annemans, Remmen, Rosenthal and Sermeus2010, Annemans et al., Reference Annemans, Boeckxstaens, Borgermans, De Smedt, Duchesnes, Heyrman, Remmen, Sermeus, van den broeke, Van Herck, Vanmeerbeek, Willems and De Gauquier2009). One review investigated the effects of Swedish reform to allow privately funded primary care and increased patient choice of provider (Burstrom et al., Reference Burstrom, Burstrom, Nilsson, Tomson, Whitehead and Winblad2017). One review investigated the effects of fee-for-services and the effect of targeting funding to minority groups (Gibson and Segal, Reference Gibson and Segal2015). Results, and quality assessments, are summarized in Table 2. Full quality assessment results can be found in the appendix.
Table 1. Characteristics of all included reviews
Table 2. Review findings, sorted by aim and quality assessment result
‘Medical processes’: a process done at the GP surgery, for example, recording smoking status, recording blood pressure; ‘prescribing’: patients being offered a drug they are eligible for; ‘outcomes’, for example, Hb1Ac, blood pressure; ‘quality of care’: total QOF points; ‘COPD’: combined obstructive pulmonary disease.
Results comparing different funding models
One moderate quality review of six studies explored inequalities in access, continuity and quality of care, and patient satisfaction of patients in capitation-based plans compared to fee-for-service (Tao et al., Reference Tao, Agerholm and Burström2016). Two US studies and one Canadian study found that capitation-based plans had reduced inequalities in access to care between ethnic groups. Both US studies also found reduced inequalities in continuity of care and ambulatory care sensitive condition admissions between ethnic groups in capitation-based plans compared to fee-for-service. In contrast, one study assessing the quality of physician’s communication skills found reduced ethnic inequalities in two of twelve measures of patient satisfaction in a fee-for-service care plan compared to capitation. One study found no effect on ethnic or educational inequalities in mortality.
Pay-for-performance: the quality and outcomes framework
The 12 reviews that examined the equity impact of pay-for-performance reimbursement focused primarily on the effect of the introduction of the Quality and Outcomes Framework (QOF) in the UK. All reviews included primary studies investigating the QOF, and seven concentrated on the QOF alone (Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Forbes et al., Reference Forbes, Marchand and Peckham2016, Gillam et al., Reference Gillam, Siriwardena and Steel2012, Mandavia et al., Reference Mandavia, Mehta, Schilder and Mossialos2017, Steel and Willems, Reference Steel and Willems2010, Tao et al., Reference Tao, Agerholm and Burström2016, Annemans et al., Reference Annemans, Boeckxstaens, Borgermans, De Smedt, Duchesnes, Heyrman, Remmen, Sermeus, van den broeke, Van Herck, Vanmeerbeek, Willems and De Gauquier2009). This scheme was introduced in 2004 with general practices receiving payment for achieving a range of standards. These standards include (1) medical processes (e.g. taking blood pressure readings, offering smoking cessation advice), (2) prescribing relevant medication, and (3) improving patient outcomes (e.g. blood pressure, Hba1c).
The reviews included studies that used a range of before-and-after, serial cross-sectional, and cross-sectional analyses and examined various outcomes such as clinical processes, prescribing, and outcomes. Before-and-after analysis was facilitated by comparing pre- and post-QOF quality of care. Pre-QOF quality of care was measured by practice performance on a range of non-incentivized quality indicators (including medical processes, prescribing, and clinical outcomes) similar or identical to QOF indicators. The most commonly assessed conditions were cardiovascular disease, diabetes, coronary heart disease, hypertension, asthma, combined obstructive pulmonary disease, stroke, and preventative health. The review’s conclusions varied significantly.
Regarding socioeconomic inequalities, two reviews reported that pay-for-performance schemes reduced inequalities (Ahmed et al., Reference Ahmed, Hashim, Khankhara, Said, Shandakumar, Zaman and Veiga2021, Gillam et al., Reference Gillam, Siriwardena and Steel2012), whereas another concluded that it had no effect (Tao et al., Reference Tao, Agerholm and Burström2016). Other papers, with generally higher quality assessment scores, and a larger number of included papers, offered more nuanced findings (Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010, Annemans et al., Reference Annemans, Boeckxstaens, Borgermans, De Smedt, Duchesnes, Heyrman, Remmen, Sermeus, van den broeke, Van Herck, Vanmeerbeek, Willems and De Gauquier2009, Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Steel and Willems, Reference Steel and Willems2010). Primary studies typically investigated socioeconomic inequalities by comparing patients living in more or less affluent areas, as measured by deprivation scores. A minority of studies focused on the occupation of patients. Synthesized results of these reviews suggested that in the first year following the introduction of QOF, there were temporarily reduced scores in less affluent areas, representing a temporary widening of inequality. This difference attenuated in the second and third years of QOF, indicating that inequalities narrowed (Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010, Annemans et al., Reference Annemans, Boeckxstaens, Borgermans, De Smedt, Duchesnes, Heyrman, Remmen, Sermeus, van den broeke, Van Herck, Vanmeerbeek, Willems and De Gauquier2009, Boeckxstaens et al.). Only one review investigated differences across the UK, finding that this narrowing effect was not seen in Scotland, where there was less improvement in quality indicators in the most deprived areas, causing a widening of inequality. (Steel and Willems, Reference Steel and Willems2010). For some indicators, large inequalities still remain (Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Steel and Willems, Reference Steel and Willems2010).
Five reviews synthesized findings on inequalities of age. One review reported that QOF widened inequalities, in favour of older patients who had better quality of care (Mandavia et al., Reference Mandavia, Mehta, Schilder and Mossialos2017). However, this review, scored as low quality, only identified two studies, with outcomes that were better in older patients at the time of the QOF. The remaining four reviews, which had a wider literature base, reported that before QOF, older patients generally had worse quality of care compared to younger patients when measured by a range of medical processes, prescribing, and patient outcomes (Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010, Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Steel and Willems, Reference Steel and Willems2010, Annemans et al., Reference Annemans, Boeckxstaens, Borgermans, De Smedt, Duchesnes, Heyrman, Remmen, Sermeus, van den broeke, Van Herck, Vanmeerbeek, Willems and De Gauquier2009). One reported that inequalities persisted in a range of medical processes following the introduction of QOF (Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010), while three reported a narrowing of inequalities (Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Steel and Willems, Reference Steel and Willems2010, Annemans et al., Reference Annemans, Boeckxstaens, Borgermans, De Smedt, Duchesnes, Heyrman, Remmen, Sermeus, van den broeke, Van Herck, Vanmeerbeek, Willems and De Gauquier2009). The former review, reporting the persistence of inequalities, did indeed report findings of primary studies that generally showed a narrowing of inequalities however chose to assess QOF’s success based on whether inequalities disappeared, rather than narrowed. Therefore overall, QOF’s introduction coincided with reduced inequalities in age, favouring older patients who had a worse quality of care before the QOF.
Five reviews reported synthesized findings regarding inequalities between sexes. At the time of QOF’s implementation, men generally had higher quality of care measured by clinical process, prescription, and clinical outcomes (Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010, Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Steel and Willems, Reference Steel and Willems2010). One study found limited evidence of improved inequalities, (Annemans et al., Reference Annemans, Boeckxstaens, Borgermans, De Smedt, Duchesnes, Heyrman, Remmen, Sermeus, van den broeke, Van Herck, Vanmeerbeek, Willems and De Gauquier2009) and the other four reviews concluded that QOF’s implementation corresponded with a widening of inequalities (Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010, Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Gillam et al., Reference Gillam, Siriwardena and Steel2012, Steel and Willems, Reference Steel and Willems2010). Pro-male pre-QOF differences persisted or widened. For some inequalities that did not have a difference between sexes, a pro-male difference emerged in the years following QOF’s implementation.
Five reviews synthesized findings regarding ethnicity. Overall, there was no clear pattern of changes in inequalities following QOF’s implementation (Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010, Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Steel and Willems, Reference Steel and Willems2010, Tao et al., Reference Tao, Agerholm and Burström2016, Annemans et al., Reference Annemans, Boeckxstaens, Borgermans, De Smedt, Duchesnes, Heyrman, Remmen, Sermeus, van den broeke, Van Herck, Vanmeerbeek, Willems and De Gauquier2009).
Multiple reviews noted the existence of gaming of the QOF; several highlighted exception reporting as a limitation. Exception reporting permits practices to exclude certain patients from target calculations based on patient compliance, disease status, medication suitability, availability of services, or patients being new to the practice. This is designed to factor in patient autonomy and prevent practices from losing potential funding due to factors outside of their control. However, findings indicated that it is possible for GPs to game this process and that there are unintended consequences: patients who are older, less affluent, and multi-morbid are slightly more likely to be exception reported, and these patients are more likely to die in the following year (Forbes et al., Reference Forbes, Marchand and Peckham2016, Gillam et al., Reference Gillam, Siriwardena and Steel2012, Annemans et al., Reference Annemans, Boeckxstaens, Borgermans, De Smedt, Duchesnes, Heyrman, Remmen, Sermeus, van den broeke, Van Herck, Vanmeerbeek, Willems and De Gauquier2009).
Reviews investigating other pay-for-performance schemes
Five reviews also investigated the effects of other pay-for-performance schemes (Ahmed et al., Reference Ahmed, Hashim, Khankhara, Said, Shandakumar, Zaman and Veiga2021, Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010, Gupta and Ayles, Reference Gupta and Ayles2020, Lin et al., Reference Lin, Yin, Huang and Du2016, Van Herck et al., Reference Van Herck, de Smedt, Annemans, Remmen, Rosenthal and Sermeus2010). Findings were limited by a number of factors, most notably a lack of primary studies that were not focusing on the QOF. Additionally, review quality limited findings: one review’s quality was scored as critically low, and three were scored as low. In the two reviews which had a notable number of studies not concerning the QOF, conclusions were limited by a lack of qualitative description of the type of schemes and their outcomes, given the wide heterogenicity of these factors (Lin et al., Reference Lin, Yin, Huang and Du2016, Van Herck et al., Reference Van Herck, de Smedt, Annemans, Remmen, Rosenthal and Sermeus2010).
Reviews investigating other changes to primary care funding
One review explored changes in inequity occurring after Swedish health reform in 2008–10 where privately funded primary healthcare practices could be founded and citizens could choose amongst primary care providers (Burstrom et al., Reference Burstrom, Burstrom, Nilsson, Tomson, Whitehead and Winblad2017). Overall, patterns suggested increased inequality due to patients with lower health needs benefiting more from these changes. Private practices were most likely to be found in areas with populations with lower health needs. Additionally, while access increased for all groups it did so more for those with lower health needs. Reforms acted as a barrier to integrated care for those with greater health needs.
One review found evidence that increasing funding specifically targeted at minority groups, with poorer health, was effective in narrowing inequalities (Gibson and Segal, Reference Gibson and Segal2015). Three primary studies identified Indigenous groups in the USA, Australia, and New Zealand, who had worse type 2 diabetes clinical outcomes compared to the national average. Of the two studies that were successful in improving health outcomes (measured by improved HbA1c), both involved the unconditional transfer of financial resources to facilitate improved care. The third, which utilized a fee-for-service model, where providers were paid per diabetic care plan, failed to show a narrowing in inequalities.
Discussion
What this research means
Capitation models were more equitable than fee-for-service plans for patient access, continuity, and quality of care. This however is not to say that all capitation systems are necessarily more equitable. Capitation models differ globally, and research did not identify what factors make capitation models more equitable. For example, in the UK, the capitation formula is over 20 years old. Many argue that it is outdated and does not sufficiently account for clinical needs (Roland and Everington, Reference Roland and Everington2016, Kontopantelis et al., Reference Kontopantelis, Mamas, van Marwijk, Ryan, Bower, Guthrie and Doran2018). It has been suggested that – by failing to weigh the additional health needs of people who live in poorer areas – this formula compounds funding inequalities: areas with higher deprivation scores receive less funding relative to their population’s higher health needs (Kontopantelis et al., Reference Kontopantelis, Mamas, van Marwijk, Ryan, Bower, Guthrie and Doran2018, McLean et al., Reference Mclean, Guthrie, Mercer and Watt2015, Roland and Everington, Reference Roland and Everington2016). Recent evidence underscores this point: adjusting capitation funding in England by deprivation data would improve funding equity (Holdroyd et al., Reference Holdroyd, Appel, Massou and Ford2024). The use of capitation models alone, therefore, is not sufficient to reduce health inequalities- individual capitation models must be well designed to ensure that such an outcome occurs.
Sufficient assessment of pay-for-performance schemes was limited to that of the QOF. Overall, conclusions varied widely – likely due to variation in the number of studies identified in reviews (2–32 concerning equity) and varying levels of review quality (critically low to moderate). In the first year of QOF’s introduction, there appeared to be an initial widening of inequality when comparing the quality of primary care (measured by QOF indicators) in practices in more and less deprived areas. Differences in performance by levels of deprivation narrowed and attenuated over the following years in England, but not in Scotland. Inequalities in age narrowed, while inequalities by sex persisted or widened. No evidence indicated that these changes were due to the QOF itself, rather than other wider societal factors. Reviews and primary studies focused on the early impact of QOF’s introduction and not its ongoing effect on health equity.
It was notable that socioeconomic inequalities narrowed in England, but not in Scotland following QOF’s introduction. At this time, there was a National Health Inequalities Strategy in England, which was conducted between 1999 and 2010. This was successful in narrowing socioeconomic health inequalities (Barr et al., Reference Barr, Higgerson and Whitehead2017, Buck and Maguire, Reference Buck and Maguire2015, Vodden et al., Reference Vodden, Holdroyd, Bentley, Marshall, Barr, Massou and Ford2023, Holdroyd et al., Reference Holdroyd, Vodden, Srinivasan, Kuhn, Bambra and Ford2022). Given that this strategy only applied to England, but not devolved nations, this may explain why narrowed socioeconomic inequalities were only seen in England. Such evidence indicates the ability of the central government to reduce health inequalities and strengthens calls for future strategies internationally.
Evidence that directing funding to specific disadvantaged communities can improve their health and has implications for a range of countries internationally for whom large health inequalities exist between Indigenous populations and the rest of the country. This evidence suggested unconditional transfers (where money is given regardless of the outcome), rather than fee-for-service funding, were most effective.
In Sweden, allowing privately operated general practices to open without oversight or equity-promoting financial incentives resulted in some measures of increased inequality.
Strengths and limitations
This umbrella review is the first to collate and synthesize all reviews of the effect of different primary care funding models on health inequalities. This review’s main strength is in its robust and preregistered methodology, with an extensive search strategy followed by a robust screening, data extraction, and quality assessment process completed by two authors. We included a wide range of peer-reviewed articles and grey literature. Such an approach allows the synthesis of a wide literature base and consensus to be formed on the debated effects of a range of policy interventions, most notably, QOF. This approach allows not just consensus on these issues but identifies a range of issues currently underexplored in the literature.
The quality assessment identified several limitations within the reviews. AMSTAR criteria highlighted key collective weaknesses, most notably the lack of a republished methodology, insufficient explanation of study designs for inclusion, failure to consider funding sources, and inadequate accounting for the risk of bias. The overall quality of the evidence base was significantly weakened, as many reviews relied on primary studies that produced correlational rather than causal results. For example, the effects of QOF cannot be seen in isolation given that until 2010, the government was simultaneously committed to reducing socioeconomic health inequalities, with a range of ambitious policy interventions that improved social determinants of health, health behaviours, and other inequalities in funding (Barr et al., Reference Barr, Higgerson and Whitehead2017, Vodden et al., Reference Vodden, Holdroyd, Bentley, Marshall, Barr, Massou and Ford2023, Barr et al., Reference Barr, Bambra and Whitehead2014, Holdroyd et al., Reference Holdroyd, Vodden, Srinivasan, Kuhn, Bambra and Ford2022). Such conclusions are common when evaluating the effect of national policy. Research from more federal systems, with regional differences in policy, would strengthen the literature base.
Comparison to previous research
One previous umbrella review aimed to explore the effects of pay-for-performance in health care, with one of six aims regarding inequalities (Eijkenaar et al., Reference Eijkenaar, Emmert, Scheppach and Schöffski2013). This review concluded that pay-for-performance schemes result generally in reduced rather than increased socioeconomic health inequalities, but not inequalities by age, sex, or ethnicity. These findings were drawn from a limited literature base, with only four reviews identified. While we find similar evidence regarding the narrowing of socioeconomic health inequalities following QOF’s implementation, we identify many limitations of the literature that prevent such generalized statements from being made regarding the impact of pay-for-performance. Additionally, our review’s findings indicate that the period following QOF’s implementation saw a widening of inequalities by sex and a narrowing of inequalities by age. A second review also included systematic reviews; however, it identified only one of the reviews included in this umbrella review (Forbes et al., Reference Forbes, Marchand and Peckham2016). This review had limited conclusions due to a lack of evidence concerning inequalities being identified. This review builds on these findings by identifying a larger literature base, exploring a wider range of changes in funding, and exploring reasons for differing results.
Implications for policy and research
Current research focuses predominantly on QOF; more research is needed on a wider range of funding models. For example, only one review examined the inequity effects associated with capitation models, despite their widespread adoption globally (Khezri et al., Reference Khezri, Mahboub-Ahari, Tabrizi and Nosratnejad2022). Despite targeted search terms, no research was identified for salaried payment models. This less commonly used model pays primary care providers per hour worked. It has less relevance in socialized healthcare systems as an individual is more likely to act within an organization that receives other forms of payment. Further research could address this by reviewing evidence of its effect on health inequalities. Additionally, research could investigate the impact of blended funding models.
Many factors could improve the strength of research. Addressing the reliance on cross-sectional, observational evidence necessitates a shift towards quasi-causal methodologies, such as the difference-in-difference approach (Cha and Escarce, Reference Cha and Escarce2022). Current research lacks granularity, exploring the overarching effects of funding changes without delving into their underlying components. Key areas of interest include the effects of different amounts of funding, the target recipients of funding (ranging from large institutions to individual practitioners), and the utilization of financial incentives versus penalties. Furthermore, reviews are predominantly dated, and an updated review would ensure that new primary studies are identified. Only one review compared different funding models. Further analysis understanding the differential effect of these funding models, especially on socioeconomic health inequalities, would further deepen understanding of their respective impacts.
Results regarding exception reporting widening inequality were concerning. To address these issues, exception reporting mechanisms should be updated to prevent GPs from measuring their performance against only patients who likely received better quality of care. Examples include reducing the time that patients can be excluded after joining the list or being less permissive of GPs not engaging with patients – currently, three letters through the post is sufficient for excepting.
Some evidence suggests that following a change in the funding model that requires action by primary care organizations, there may be an immediate short-term widening of inequality. This was seen in the first year after QOF’s introduction and also after a pay-for-performance scheme introduced in the UK in 1999 to incentivize cervical cancer screening targets (Alshamsan et al., Reference Alshamsan, Millett, Majeed and Khunti2010, Boeckxstaens et al., Reference Boeckxstaens, Smedt, Maeseneer, Annemans and Willems2011, Steel and Willems, Reference Steel and Willems2010). In both cases, these inequalities narrowed over the following years. Reasons suggested for this vary. It may be that in more affluent areas practices can better respond to change due to more capacity as a result of lower clinical need. Alternatively, the inverse equity hypothesis was suggested, whereby affluent groups in society preferentially benefit from any societal interventions, and less affluent groups only benefit once the maximum benefit has been extracted (Victora et al., Reference Victora, Vaughan, Barros, Silva and Tomasi2000). Regardless of the causal reason, the potential for short-term increases in inequality should be considered when interventions occur, and contingencies for this should be made.
Conclusion
This umbrella review highlights evidence indicating the effect of primary care funding on inequalities in healthcare access, experience, and clinical outcomes. Some evidence indicated the pro-equity benefits of capitation models over fee-for-service models. A range of previously contested evidence concerning the effect of the QOF was synthesized, finding that overall, socioeconomic inequalities in the quality of primary care narrowed over the years following implementation. A stronger and more diverse literature base would more easily allow policy decisions to be driven by strong evidence.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S146342362500012X.
Funding statement
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interests
None.
Open access
