OUP user menu

Does general practice reduce health inequalities? Analysis of quality and outcomes framework data

Anna Dixon, Artak Khachatryan, Stuart Gilmour
DOI: http://dx.doi.org/10.1093/eurpub/ckq177 9-13 First published online: 8 December 2010

Abstract

Background: The government set a series of targets to reduce health inequalities in England by 2010. Primary care has an important role in reducing health inequalities. The aim of the study was to assess the impact of general practice on reducing area-based health inequalities. Methods: Analysis of differences in achievement on clinical indicators between practices in Spearhead and non-Spearhead Primary Care Trusts (PCTs) using data from the Quality and Outcomes Framework (QOF) for 2004/05 and 2005/06, practice characteristics and Spearhead status of PCTs. The study used data on 8339 primary care practices in England. Unweighted mean reported achievement on subset of 26 clinical indicators was calculated. The study analysed differences in achievement by Spearhead status and deprivation in both years and the change between years. Multiple regression analysis of relationship between Spearhead status, income deprivation, reported achievement and other factors also were carried out. Results: Practices in Spearhead PCTs performed worse than practices in non-Spearhead PCTs in both years but showed greater improvement. Among the most deprived practices, there were no differences in QOF achievement between Spearhead and non-Spearhead PCTs. Previous year’s achievement was the strongest predictor of performance. Conclusion: The narrowing in performance between practices in Spearhead and non-Spearhead PCTs may have indirectly contributed to a reduction in area-based health inequalities but the differences are small. The lack of difference between the most deprived practices in Spearhead and non-Spearhead PCTs suggest that area-based initiatives to tackle inequalities have not yet had an observable impact on deprived practices. Unobserved factors explain most of the variation in achievement.

Introduction

In 2001, the Labour government set a series of ambitious targets to reduce health inequalities in England by 2010. While significant progress has been made in improving life expectancy in absolute terms across all social groups, the government will miss its target to reduce by at least 10% the gap in life expectancy between the fifth of areas with the worst health and deprivation indicators (so-called Spearhead areas) and the population as a whole.1 Spearhead status was designated to Primary Care Trusts (PCTs) that were in the bottom fifth nationally for three or more of the following five indicators: male life expectancy at birth; female life expectancy at birth; cancer mortality rate in under 75s; cardiovascular disease mortality rate in under 75s; and the Index of Multiple Deprivation (IMD) 2004. The aim was to address geographical inequalities in life expectancy, cancer, heart disease, stroke and related diseases. There are 62 PCTs in England which are designated Spearhead PCTs [reduced from 88 following National Health Service (NHS) reconfiguration in 2006], which received additional funding. However, this funding was not ringfenced and, therefore, did not necessarily get spent on public health or on practices within these areas.

While the government has sought to reduce inequalities by tackling the wider determinants through a cross- departmental programme of activities,2 it has also recognized an important role for primary care in reducing health inequalities. Differential access to health services is recognized as an important potential cause of inequities in health although evidence of the impact of such differences is unclear.3 While equity of access to primary care and specialty services has not been fully achieved, the UK performs comparatively well on this dimension.4 This is particularly true in more deprived communities, where primary care has often played a key role more broadly in promoting health and tackling the wider social determinants of health in an effort to reduce health inequalities.

The new General Medical Services (GMS) contract introduced in April 2004 included for the first time a pay for performance component called the quality and outcomes framework (QOF) which rewarded practices financially for undertaking nationally specified clinical and organizational activities, including secondary prevention and better management of people with chronic conditions. The main objective was to drive up the quality of primary care, but more recently the government has made a reduction in health inequalities, another objective of QOF.5,6

Practices are also contracted under a Personal Medical Services (PMS) contract. This was introduced to provide general practitioners (GPs) with the option of being salaried.7 The majority of these practices also participate in QOF.

Studies published, to date, suggest that the differences in performance on QOF indicators between the least and most deprived practices are gradually narrowing.8,9 However, these differences do not necessarily translate into differences in quality of clinical care nor improved health outcomes.10 For example, the quality of chronic disease management for at least some conditions [Coronary Heart Disease (CHD), diabetes, hypertension—in terms of hypertension and cholesterol targets] in England was shown to be broadly equitable between socio-economic groups before the introduction of QOF, and remained so after its introduction.11

The main aim of this study is to assess the impact of general practice on reducing area-based health inequalities. In order for general practice to make a contribution to reducing health inequalities, as measured by current government targets, practices in Spearhead areas would need to outperform those in non-Spearhead areas. This study therefore examines the extent to which practices in Spearhead and non-Spearhead PCTs differ in their achievement on key clinical indicators in the first two years of the new contract.

Methods

Data from the QOF for more than 8000 primary care practices in England (N = 8339) were analysed. Of that, 343 practices with a list size less than 1000 patients were excluded (4% of the total).12 Only the first 2 years of data were included (2004/05 and 2005/06) because changes in the 2006/07 QOF indicators prevent direct comparison in subsequent years.

Information on practice characteristics was taken from the 2005 and 2006 General Medical Statistics dataset. Deprivation scores for each practice were obtained from an attribution data set which used the Indices of Multiple Deprivation (IMD 2004) to assign each practice a level of deprivation based on the postcodes of the practice population. Practices were then divided into quintiles of deprivation using the income domain.

In addition, the Spearhead status of PCTs in which the practice was located and the contract status of primary care practices (e.g. GMS or PMS) were included in the final data set.

QOF indicators

The main outcome variable is unweighted mean reported achievement for a subset of 26 clinical indicators within QOF. Reported achievement is calculated as the number of patients for which the indicator was achieved (numerator), divided by the total number of recorded patients for that particular indicator less than those exception reported (denominator). Patients may be excluded from the denominator because they meet at least one of the exception criteria outlined in the GMS ‘Statement of Financial Entitlements’.

Building on methodologies employed in other recent research,12–19 we selected a set of indicators which are believed to contribute to health gains. Selection was based on evidence for ‘lives saved’ and cost-effectiveness of QOF indicators,20 conditions and targets identified within the Health Inequalities Intervention Tool,21 and clinical validity. Indicators which simply involve record keeping or diagnosis were excluded. For a full set of the 26 clinical indicators employed, see Supplementary Data.

In order to ensure any variation in performance could not simply be explained by reporting behaviour, the analysis was repeated using population achievement (results not shown). Population achievement is the number of patients for whom the indicator was met as a proportion of the estimated number of patients with the condition, i.e. including exception reported patients in the denominator. No significant differences in the results were observed. In general, rates of exception reporting have been low, with little evidence of ‘gaming’.22 The distribution of exception reporting was roughly similar across practices by deprivation and Spearhead status.

Analysis

Differences in mean reported achievement were analysed using Analysis of Variance (ANOVA), for all practices and after stratification by Spearhead status in 2004/05 and 2005/06. The change in mean reported achievement between years was also analysed.

Multiple regression analysis was conducted to adjust the relationship between Spearhead status, income deprivation and mean reported QOF achievement for the following potential confounders: GP caseload [number of patients/Full Time Equivalents (FTEs)]; number of GPs per practice; the country of qualification of GPs (UK/non-UK); the type of contract held by the practice (PMS/GMS); the age distribution of the practice population (≤14 years old and ≥65 years old); and previous year performance (for 2005/06 only).

No model-building technique was employed because all included variables were of prior interest. Standard significance levels (P = 0.05) were assumed and 95% confidence intervals (CIs) are reported wherever possible. Overall, goodness-of-fit of the model was judged based on R2 statistics, which are reported in the text. Statistical analyses were conducted using Stata/IC 10.0.

Results

Descriptive analyses

Thirty-six per cent of the practices were located in Spearhead PCTs (for a description of practices, see table 1). Within non-Spearhead PCTs, there were 1411 practices (26%) drawn from more affluent areas and 518 practices (10%) whose population comes from the most deprived areas (subsequently referred to as least deprived or most deprived practices). Among Spearhead PCTs, there were 259 least deprived practices (9%) and 1116 most deprived practices (38%).

View this table:
Table 1

Mean reported achievement (%) on selected clinical indicators, by income quintile and Spearhead status, 2004/05 and 2005/06 and change between 2 years

IndicatorsNo. of practices2004/052005/06Change
Mean% (95% CI)Mean% (95% CI)Mean% (95% CI)
All practices
Spearhead statusP < 0.001P < 0.001P < 0.001
Yes297080.18 (79.86–80.49)84.87 (84.65–85.09)4.69 (4.48–4.91)
No535081.53 (81.32–81.73)85.47 (85.33–85.61)3.94 (3.79–4.09)
Income quintileP < 0.001P < 0.001P < 0.001
1 (least deprived)167081.88 (81.53–82.22)85.54 (85.28–85.79)3.66 (3.42–3.90)
2165981.40 (81.02–81.77)85.59 (85.35–85.82)4.19 (3.91–4.46)
3166081.26 (80.90–81.63)85.45 (85.19–85.71)4.18 (3.93–4.44)
4165580.73 (80.32–81.13)85.06 (84.79–85.33)4.33 (4.03–4.64)
5 (most deprived)163480.06 (79.64–80.49)84.59 (84.27–84.90)4.52 (4.24–4.81)
Practices in Spearhead PCTs by IMD 2004 income quintile
Income quintileP = 0.055P = 0.006P = 0.81
1 (least deprived)25979.81 (78.54–81.03)84.40 (83.48–85.31)4.59 (3.85–5.33)
231980.87 (80.00–81.74)85.71 (85.14–86.28)4.84 (4.19–5.50)
350180.98 (80.30–81.66)85.36 (84.86–85.87)4.38 (3.95–4.81)
475780.25 (79.69–80.80)84.86 (84.44–85.28)4.61 (4.20–5.02)
5 (most deprived)111679.80 (79.27–80.33)84.51 (84.13–84.89)4.71 (4.36–5.06)
Practices in non-Spearhead PCTs by IMD 2004 income quintile
Income quintileP < 0.001P = 0.003P = 0.02
1 (least deprived)141182.26 (81.92–82.60)85.75 (85.49–86.00)3.49 (3.24–3.74)
2134081.52 (81–11–81.94)85.56 (85.30–85.82)4.03 (3.73–4.34)
3115981.39 (80.95–81.82)85.49 (85.19–85.79)4.10 (3.79–4.41)
489881.13 (80.56–81.70)85.23 (84.88–85.58)4.10 (3.66–4.54)
5 (most deprived)51880.62 (79.91–81.34)84.75 (84.17–85.33)4.12 (3.63–4.62)

Mean reported achievement on selected indicators in 2004/05 was 81.0% (95% CI 80.9–81.2; median 82.6) across all practices, increasing in 2005/06 to 85.3% (95% CI 85.1–85.4; median 86.0). There was significant and strong correlation between reported achievement in 2004/05 and 2005/06 (Spearman rank correlation coefficient 0.70; P < 0.0001).

Differences in reported achievement between practices in Spearhead and non-Spearhead PCTs were small but statistically significant, with practices in non-Spearhead PCTs slightly outperforming practices in Spearhead PCTs (see table 1). Over 2004/05 and 2005/06, practices in Spearhead PCTs improved more than non-Spearhead PCTs. This change was small but statistically significant and was replicated amongst both the most and the least deprived practices.

However, after stratifying the analysis by Spearhead status, differences by income quintile in mean reported achievement, and change in achievement, became non-significant within Spearheads at baseline (2004/05). Income quintile in non-Spearhead practices remained significant, with the least deprived practices slightly outperforming the most deprived practices (see table 1).

We also compared the differences in mean reported achievement between Spearhead and non-Spearhead PCTs among the most and the least deprived practices only. There was no difference between the most deprived practices by Spearhead status. The difference became statistically significant when comparing means for the least deprived practices between non-Spearhead and Spearhead PCTs [82.3% (81.9–82.6) vs. 79.8% (78.5–81.0) in 2004/05 (t-statistic = 5.07; P < 0.001); 85.8% (85.5–86.0) vs. 84.4% (83.5–85.3) in 2005/06 (t-statistic = 3.75; P < 0.001)] (see table 1 and figure 1).

Figure 1

Mean QOF achievement by Spearhead status for the least deprived practices. The scale on Y-axis starts at 70%. Spearhead, least deprived (Open diamond); Spearhead, most deprived (open circle); Non-Spearhead, least deprived (filled square); Non-Spearhead, most deprived (dashed line)

Regression analyses

In a multiple regression model of baseline performance (2004/05), mean reported achievement was significantly associated with GP education in the UK, GP caseload and the contract status of primary care practices (table 2). The results for 2005/06 were broadly similar and are not presented here.

View this table:
Table 2

Relationship between mean reported QOF achievement on selected clinical indicators and practice-level indicators, 2004/05

VariableRegression coefficientaP-value95% CI
Area deprivation (IMD 2004 income quintile)0.0289b
Quintile 1 (least quintile)
Quintile 2−0.3560.183−0.880 to 0.167
Quintile 3−0.2280.398−0.757 to 0.301
Quintile 4−0.5500.045−1.089 to –0.012
Quintile 5 (most deprived)−0.9020.002−1.472 to –0.332
Spearhead status of PCTs−0.3780.055−0.764 to 0.008
GP education in the UK (percentage)0.029<0.0010.024 to 0.034
GP caseload (number of patients/FTEs)−0.00030.005−0.001 to –0.0001
Proportion of older age group (≥65 years)0.0050.757−0.029 to 0.040
PMS status of practices−0.642<0.001−0.996 to –0.289
Number of GPs per practice0.056b
1 GP
2–4 GPs0.5800.0170.105 to 1.056
≥5 GPs0.4410.118−0.112 to 0.995
  • a: The regression coefficient for the variable is the expected change in response (outcome) per unit change in the variable, all other things being equal

  • b: Overall P-value

Improvement in mean reported achievement between 2004/05 and 2005/06 was associated with GP education in the UK, the contract status of practices and number of GPs per practice (table 3). Deprivation remained significantly associated with mean reported achievement at baseline (table 2) and in 2005/06, but was not associated with change over time (table 3). Spearhead status of PCTs was not significantly associated with mean reported achievement or changes over time, or was only of borderline significance after adjusting for practice-level factors. A modified analysis in 2005/06 found that previous year’s performance was strongly associated with mean reported achievement, as well as change over time, consistent with the finding of previous research that there is a strong correlation between years.9

View this table:
Table 3

Relationship between difference in mean reported QOF achievement on selected clinical indicators between 2004/05 and 2005/06 and practice-level indicators

VariableRegression coefficientaP-value95% CI
Area deprivation (IMD 2004 income quintile)0.142b
Quintile 1 (least quintile)
Quintile 20.5000.0090.123 to 0.877
Quintile 30.2830.144−0.097 to 0.700
Quintile 40.3110.116−0.077 to 0.700
Quintile 5 (most deprived)0.2830.176−0.127 to 0.694
Spearhead status of PCTs0.2800.0480.002 to 0.559
GP education in the UK (percentage)−0.012<0.001−0.016 to –0.009
GP caseload (number of patients/FTEs)0.00020.1150.00004 to 0.0003
Proportion of older age group (≥65 years old)0.0090.4930.016 to 0.034
PMS status of practices0.475<0.0010.223 to 0.727
Number of GPs per practice<0.001b
1 GP
2–4 GPs−0.749<0.001−1.130 to –0.368
≥5 GPs−1.230<0.001−1.666 to –0.795
  • a: The regression coefficient for the variable is the expected change in response (outcome) per unit change in the variable, all other things being equal

  • b: Overall P-value

The distribution of practice characteristics within Spearhead and non-Spearhead PCTs is shown in Supplementary table S1). Practices in non-Spearhead PCTs had a significantly higher number of GPs per practice, lower GP caseload and higher proportion of GPs who qualified in the UK—factors associated with lower performance on QOF. Although there were no significant differences between the most- and least-deprived practices within Spearhead PCTs, there were within non-Spearhead PCTs. The most deprived practices within non-Spearheads had a higher GP caseload, less GPs per practice and lower proportion of GPs qualified in the UK, i.e. the practices in the most deprived areas of non-Spearhead PCTs bore a closer resemblance to those in Spearhead PCTs.

The final multiple regression models explained only between 1% (2005/06) and 4% (2004/05) of the variance in reported achievement. However, after adjusting for previous year’s performance in the model for 2005/06, the R2 statistic of the model increased dramatically to ∼49–52%. This indicates that, while there were statistically significant associations between some of the examined practice-level characteristics and mean reported achievement, other unobserved factors (e.g. related to the clinical practice of the GPs or population characteristics) may play a much greater role in explaining what little variation in QOF achievement exists between practices.

Discussion

Our findings are broadly consistent with previous research on QOF achievement by general practice. However, linking QOF data to whether practices are located in Spearhead or non-Spearhead PCTs has allowed additional analysis of area-based inequalities in the performance of general practice.

Practices in Spearhead PCTs performed worse in both years but showed greater improvement. In general, the more deprived practices, regardless of Spearhead status, performed worse but improved more. The least deprived practices in Spearhead PCTs performed significantly worse than similar practices in non-Spearhead PCTs. Within Spearhead PCTs, however, there were no significant differences in mean reported achievement by income quintile.

After adjusting for practice-level factors, Spearhead status did not explain performance. Previous year’s achievement was the strongest predictor of performance and suggests that other unobserved factors relating to practice or population characteristics explain most of the variation in achievement. However, practices in non-Spearhead PCTs had a significantly higher number of GPs per practice, lower GP caseload and a higher proportion of GPs who qualified in the UK in both years of the study, all factors associated with high levels of achievement. Previous international research23 has found that practices in deprived urban areas are not sufficiently rewarded for their higher workload. Despite recent investments in new primary care facilities in England and the narrowing in QOF achievement, more may need to be done to compensate practices in deprived areas.

Analysis of deprivation ‘within’ and ‘between’ Spearhead and non-Spearhead PCTs means we have identified more specifically the way in which deprivation and area-based inequalities interact. Observed differences in mean reported achievement between Spearhead and non-Spearhead PCTs appear to be mainly driven by differences in performance between the least deprived practices. It appears that practices that serve wealthier populations but which are surrounded by deprivation perform worse.

There are well-documented limitations associated with the use of the QOF data set as a measure of practice performance: the data is linked directly to financial incentives; it only covers a limited number of conditions; few of the indicators relate to outcomes; and data are at the practice not the patient level.

There are a number of further limitations. First, differences in achievement are small and the distribution of performance is extremely concentrated at the upper end of the range. Most observed differences are very small but may be statistically significant due to a large sample size (number of practices). Second, the ability to exclude patients from the denominator (through exception reporting) means data coverage is not 100%. However, published research and our own analyses found that exception reporting is relatively small and rates do not systematically differ by Spearhead status or deprivation. We repeated analysis using population achievement (i.e. including exception reported cases in denominator) but this did not have a substantial impact on observed associations. Third, we used the mean achievement across a subset of clinical indicators. There is significant variation in performance on individual clinical indicators which may have been lost by looking at a composite measure of performance. Fourth, deprivation scores are derived from data on small geographical areas (super output areas) and therefore may be vulnerable to misattribution, i.e. not truly reflect the actual practice population. To minimize the effects of misattribution of income deprivation, we analysed the most and least deprived practices separately.

Finally, while there is a possibility of ‘regression to the mean’ (RTM), our study uses the same measurement instruments, and follow-up measurements were examined using the whole sample24 so the extent of this should be negligible.

By far, the strongest predictor of QOF achievement was previous year’s performance suggesting that the most important explanatory factors were not observed in this data set. Future research needs more detailed information on GP, practice and population characteristics in order to fully explain differences in QOF achievement and thereby identify the most relevant policy solutions.

By analysing only a subset of clinical indicators that contribute to health gain, we have been able to isolate the likely impact on population health. Thus, we observe differences not only in the quality of general practice, but also those which may have an impact on health inequalities.

The narrowing in performance between practices in Spearhead and non-Spearhead PCTs in the first 2 years may suggest that QOF has indirectly contributed to a reduction in area-based health inequalities. Analysis using other data sources such as outcomes data would be needed to confirm this, as would qualitative evidence of how practices in these areas have responded to these incentives.

The lack of a substantial difference in performance between the most deprived practices in Spearhead PCTs and non-Spearhead PCTs suggests that Spearhead status and its associated policies have not yet had an observable impact on the performance of deprived practices. Spearhead activities only really began in 2004, so there may be a time lag. Further research is needed to examine area-based differences in later years. Future efforts to tackle health inequalities may need to focus on smaller areas or target individuals regardless of the area in which they are located.

Funding

The National Institute for Health Research Service and Delivery Organisation Programme (ref no: PHA 202).

Conflicts of interest: None declared.

Key points

  • We already know that there are small and declining absolute differences in performance on QOF and that the differences between the least and most deprived practices are gradually narrowing.

  • The evidence is equivocal on whether these improvements are influenced by the incentives created by QOF, but it appears that on some clinical indicators improvements predate the introduction of QOF.

  • This study has shown that differences in performance are narrowing between practices in Spearhead and non-Spearhead PCTs but these appear to be mainly driven by differences in performance between the least deprived practices.

  • The lack of a substantial difference in performance between the most deprived practices in Spearhead PCTs and non-Spearhead PCTs suggests that Spearhead status and its associated policies have not yet had an observable impact on the performance of deprived practices.

  • Further research is needed to examine area-based differences in later years. This might also suggest that future efforts to tackle health inequalities should focus on deprived practices, regardless of the area in which they are located.

Supplementary data

Supplementary data are available at EURPUB online.

Acknowledgements

This report is independent research commissioned by the National Institute for Health Research Service and Delivery Organisation (SDO) Programme. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health. The authors acknowledge the advice of Dr Tammy Boyce on health inequalities policy and feedback on early findings from other members of the project team and Advisory Group.

References

View Abstract