Family Practice Advance Access published online on November 29, 2007
Family Practice, doi:10.1093/fampra/cmm065
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
How representative of primary care are research active practices? Cross-sectional survey
Department of Primary Care and General Practice, Primary Care Clinical Sciences Building, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. Correspondence to FD Richard Hobbs; Email: f.d.r.hobbs{at}bham.ac.uk
Received 28 November 2006; Revised 20 August 2007; Accepted 22 September 2007.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Background. There has been a continued trend towards undertaking primary care-based research but the characteristics and generalizability of practices that participate in such research are not well known.
Objective. To compare research active and non-active practices in terms of practice demographics, disease prevalence and quality scores from the Quality and Outcomes Framework.
Design. Cross-sectional survey using publicly available data.
Setting. A total of 973 general practices from the West Midlands, UK.
Main outcome measures. Practice population characteristics, research status, disease prevalence, clinical and non-clinical quality scores.
Results. Of 973 practices, 298 (31%) were defined as research active. Research active practices had younger populations (% over 65: 15.2% versus 16.2%, z = 3.95, P < 0.0001) compared to non-research active practices, were larger [median list size 6123 (interquartile range, IQR, 3642–9691) versus 4059 (IQR 2675–7060) z = 6.96, P < 0.0001] and more likely to be in deprived areas [median Townsend quintile 5 (IQR 3–5) versus 4 (IQR 3–5), z = 3.23, P = 0.001]. Disease prevalence was similar in both research active and non-active practices but the former attained higher median quality scores for both clinical [research active 534/550 (IQR 508–546) versus non-research active 525/550 (IQR 483–542) z = 4.00, P < 0.0001] and non-clinical [310/320 (IQR 283–319) versus 296/320 (IQR 265–314), z = 5.76, P < 0.0001] areas.
Conclusion. General practices which participate in research are larger and located in more deprived areas than non-research active practices but disease prevalence is similar and research practices attain only modestly higher quality points. Research in research active practices is likely to be generalizable to the wider primary care community.
Keywords. Comparative study, family practice.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
In the UK, most patient contact and 90% of prescribing occur in primary care but until recently, much of the evidence to support this activity was generated in secondary or tertiary care.1 Differences between the patients (severity of disease, impact on quality of life, extent of comorbidity) seen in primary and secondary care have led to calls for more research in representative primary care populations.1,2 Networks of practices in the US and Europe aim to provide an appropriate environment for the execution of multi-centre primary care-based research.3–5 Nevertheless, there remains the possibility that research active practices are different from practices that do not participate in research and thus studies undertaken in this setting may be no more representative than those within secondary care. Information regarding the comparability of research active and research inactive practices is required if we wish to generalize research findings to primary care as a whole.
Few data exist that compare general practices which do or do not participate in research. Hammersley et al. in 20026 compared research practices from one area of England (Trent) with other practices in the same geographical area using data regarding practice characteristics, hospital admissions and cancer incidence and mortality. Although research activity was not well defined, they found no difference in the demography of registered patients, or in morbidity, mortality or use of secondary care despite some differences in practice structure including size. Tilyard et al.7 in New Zealand compared practices contributing to a computer research group and those not contributing and found differences in professional training, some process measures and patient populations but not in terms of morbidity although the latter analysis was limited. Harcourt et al.8 has compared the practice populations of Royal College of General Practitioners (RCGP) sentinel practices to the general population in terms of demographics but not GP performance.
The Midlands Research Practices Consortium (MidReC), based in the Department of Primary Care and General Practice at the University of Birmingham, UK, specializes in the design and execution of large primary care-based randomized controlled trials (RCTs) and epidemiological studies. MidReC has no formal membership criteria other than willingness to undertake research and to follow research governance procedures, and is one of the largest research networks in the UK.
Comparative data from the vast majority of English general practices are now available from the Quality and Outcomes Framework (QOF) of the General Medical Services Contract and provide an opportunity to objectively compare practices taking part with those not taking part in research.9,10 The QOF was designed as a performance-related mechanism for paying GPs to provide satisfactory standards of care as judged (in the 2004–2005 financial year) by criteria around 11 clinical areas (550 points) along with a further group of ‘non-clinical areas’ including practice management, education and training and patient experience (320 points) with the remaining points to a maximum possible 1050 accrued from ‘direct enhanced services’ such as access targets. Points are obtained by practices depending on the proportion of eligible patients fulfilling process and outcome criteria, for example the proportion of people with coronary artery disease whose last cholesterol measurement was below 5 mmol/l.9 Data are collected automatically and anonymously from practice-computerized clinical records. Similar, albeit less widespread, systems of payment by performance are under development or early implementation elsewhere.11
Almost all practices in England are scored under the QOF, each of the 1050 available points being worth around £75 ($149, 
110) to an average-sized practice in the year 2004–2005.12 The aim of this study was to compare research active and inactive practices in terms of practice characteristics, disease prevalence and points attainment in the clinical and non-clinical domains of the QOF.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Population
For the purposes of this comparison, only those practices situated within the boundaries of the West Midlands Region of the UK were included. Research active practices were defined as those participating in at least one MidReC coordinated study in the three calendar years 2003–2005. Depending on the study design, practice involvement ranged from running computer searches and then sending invitations to potentially eligible populations for research that was delivered by the academic department, to research that was delivered within the primary care setting by the primary care team such as seeing patients and randomizing them in standard surgeries or bespoke clinics. Non-research practices were defined as those practices that had not undertaken a MidReC-sponsored study between 2003 and 2005.
Procedure
Data regarding disease prevalence and the points achieved in the 2004–2005 QOF were downloaded from the UK National Health Service (NHS) information centre.13 Additional practice data were obtained from a variety of NHS and UK Government Sources (see Table 1 for details). The data sources were the same whether or not a practice had undertaken research. Ethical approval was not required for this study which used publicly available data.
Analysis
Analyses were undertaken using Stata SE, version 9 (Stata Corp). Median practice values and their associated interquartile ranges (IQRs) are used to describe the distribution of values/scores within each group of practices in order to minimize the impact of outlying practices. The Mann–Whitney U test for non-parametric unpaired data was used to compare values from each group of practices. In view of the number of comparisons, a P-value of 0.01 was chosen as the cut-off for statistical significance.
Sensitivity analysis
In order to judge the effect of participation in more than one study and the effect of the type of study, sensitivity analyses were performed varying the number of studies participated from one or more to three or more and varying the type of study from one or more RCTs to three or more RCTs. In each case, the comparator was no research activity.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Exclusions
In total, 34 practices which did not have complete QOF and demographic data were excluded from the analyses.
Practice demographics
Out of a possible 973 practices within the West Midlands, 298 (31%) had participated in at least one MidReC coordinated research study within the 3-year time period with a median of one study (range 1–11 studies) and were defined as research active (Table 1). Research active practices were larger, had more GP principals, were more likely to be training practices and were more likely to be situated in the lowest quintile of deprivation (i.e. more deprived) for England and Wales than those not participating in research. Practice age/sex distribution was similar in research active and inactive practices but the former had more children aged under 15 years and fewer adults over 65 years. The age and sex distribution of all West Midlands Practices was comparable with practices in England and Wales as a whole (not shown, data available from the authors). Median population density in research active practices was twice that in those not undertaking research, reflecting an excess of research practices based in urban areas.
|
Disease prevalence
The median practice prevalence was within two cases per thousand population in 9 of the 11 diseases for practices participating and not participating in research (Table 2). Statistical testing showed that prevalence was significantly lower in research active practices for coronary heart disease, epilepsy and thyroid disease and higher for severe mental health problems.
|
Clinical domain quality scores
Research active practices gained a median of 534 of the available 550 clinical domain points compared with 525 points in non-research practices (z = 4.00, P < 0.0001). This difference reflects higher median scores in 7 of the 11 areas in this domain by research active practices. However, the absolute differences are small (within two points) in each area (Table 3).
|
Organizational, additional services and patient experience domain scores
For the three non-clinical domains combined, the median score for research active practices was 14 points higher than non-research practices (Table 4). The median scores were equivalent for five of the areas but research active practices scored more highly in the ‘Records and Information’ and ‘Medicines Management’ sections. Higher scores in the former were due to the proportion of clinical records that had been summarized (Table 5).
|
|
Sensitivity analyses
Practices that took part in three or more studies tended to have slightly larger list sizes [three or more studies 6639 patients (IQR 4115–10039) versus one or more studies 6123 (IQR 3642–9691)] and be more likely to have accreditation for postgraduate training of GPs (three or more studies 61.6% versus one or more studies 44.0%). Comparisons of crude disease prevalence and QOF clinical and non-clinical points between practices with one or more and three or more studies were similar for each individual area and domain included in the analysis, although overall non-clinical points were marginally higher in practices undertaking more studies (data available from the authors).
Practices undertaking RCTs (n = 184) as opposed to other types of study were also larger [median list size 7591 patients (IQR 4896–10475) versus 6123 patients (IQR 3642–9691)] and more likely to have training status (57.6% versus 44.0%). Practices randomizing patients to RCTs achieved similar QOF scores to practices doing other types of research in both clinical and non-clinical domains.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
This study shows that primary care practices participating in research in the UK in the period 2003–2005 have both similarities and differences to those that have not participated in research. Research active practices were larger, situated in more populous and deprived areas, more likely to undertake postgraduate training of GPs, have younger populations and score more highly in the QOF. However, the absolute difference between many of these parameters was small and unlikely to have any clinical impact: prevalence of disease in each of the clinical areas covered by the QOF was within two cases per thousand population, and there were no more than two points difference between the median quality score for research active and inactive practices in each clinical area.
Variation in size and location of practices between each group probably explains many of the other differences observed. Larger practices may be better able to develop the administrative infrastructure to enable engagement in activities peripheral to the delivery of a clinical service.14 This is likely to give an advantage in scoring highly in terms of QOF points, particularly, for example, for Records and Information where larger and training practices are more likely to have previously summarized clinical records making these scores easier to achieve.
The overrepresentation of research active practices in urban and deprived areas is likely to explain the slightly younger practice populations observed compared to research inactive practices. This may account for the marginally lower prevalence of coronary heart and thyroid disease observed as well as the increased numbers of patients appearing on practice severe mental health registers.
It is not clear from this study whether these differences in size and location are a result of volunteer bias or researcher choice. Large urban practices within easy reach of the research hub in Birmingham have historically been more likely to be contacted by research teams and at least one of our large studies has taken this approach for logistical reasons.15
The biggest strength of using QOF data in this comparison is that they have been collected in the same way from each practice, are almost universally available from all primary care practices in England and have been quality controlled by local Primary Care Trusts. Only 34 (3.4%) of the possible 1007 practices were excluded from the analyses because some of their QOF results were missing or could not be linked to demographic data and the exclusion of these practices is unlikely to have had any substantial effect. The clinical domains included in the 2004–2005 QOF were limited to 11 chronic disease areas (see Table 3) but did not include the full range of clinical activity taking place in practices.
Perhaps the biggest disadvantage with using the QOF data is that they were collected for the purpose of remunerating practices rather than research or audit. QOF data in the clinical domains may be subject to bias, if practices have limited the denominator populations within disease areas by exception reporting (which removes individuals from the denominator of part or all of a given area)10 but we have no reason to think that this might be more or less likely in practices that are engaged in research. It is difficult to adjust for exception reporting because rates were not routinely collected for practices. A further issue with the data is that many practices achieved close to the maximum points available (the average QOF points achieved by practices in England in 2004–2005 was 958.7 points, 91.3% of the available points),13 reducing the ability to differentiate between practices. Finally, the QOF prevalence and quality data were not adjusted to take account of the larger size, younger populations and higher rates of deprivation in research practices.
This comparison has used only practices with links to one research centre in one country. Apparently research inactive practices may have in fact led their own research or collaborated with other centres but incomplete recording of total research activity is likely to have minimized our ability to detect differences between research active and inactive practices, rather than lead to the identification of spurious differences. MidReC research is largely collaborative and multi-centre and may be different to other primary care research.16 However, in the time period evaluated, 20 studies ranging from large RCTs through cross-sectional surveys and qualitative studies were undertaken by the consortium and the methods used to recruit patients and practices are similar to those subsequently adopted by the new Primary Care Research Network (PCRN). UK research practices may differ from those elsewhere in Europe or America and while further research would be needed to make definitive comparisons internationally, motivations of service practitioners to take part in research may well be similar across national boundaries.
This study confirms some of the results from earlier investigations. Previous research reported that research active GPs tended to be involved in undergraduate teaching and based in larger practices.17 Nevertheless, studies in the UK and New Zealand have found few important differences in terms of clinical parameters between practices taking part in research and those that do not, other than for some prescribing parameters and immunization rates.6,7
The current study uses recent data to allow for the effect of recent NHS reforms and uses practice-based process and outcome data and thus might be expected to better reflect current primary care practice than previous work. Although some statistically significant differences have been found, these are small in absolute terms and reassuring in that research active practices are not substantially different in any clinically relevant aspects to those not participating in research. The slightly higher QOF scores achieved by research practices might even aid recruitment to research. Policy makers should be encouraged by the results from this study which suggest that evidence gathered from research active practices, at least in a big PCRN like MidReC, is likely to be generalizable to primary care as a whole.
|
Declaration |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
Funding: Department of Health via National Primary Care Post Doctoral Fellowship and Career Scientist Award to R.M. and S.W. NHS Budget 1 transitional funding to MidReC. This work was performed independently of the funders and does not necessarily reflect the views of the Department of Health.
Ethical approval: None.
Conflicts of interest: None.
|
Acknowledgments |
|---|
Martin Wilkinson and Carol Harper from West Midlands Deanery for provision of training status data. MidReC practices for continuing to support primary care research in West Midlands and beyond.
Contributors: RJM, SW and FDRH had the idea for the paper. RJM wrote the first draft and oversaw the analyses which were performed by RR and MJ. All authors contributed to subsequent drafts of the paper and have approved the final draft. FDRH will act as guarantor.
|
Notes |
|---|
McManus RJ, Ryan R, Jones M, Wilson S and Hobbs FDR. How representative of primary care are research active practices? Cross-sectional survey. Family Practice 2007; Pages 1–7 of 7.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDeclarationReferences |
|---|
1 Mant D. National Working Group on R&D in Primary Care: Final Report (1997) Bristol: NHS Executive South and West.
2 Mant J, McManus RJ, Hare R. Applicability to primary care of national clinical guidelines on blood pressure lowering for people with stroke: cross sectional study. Br Med J (2006) 332:635–637.
3 Thomas P, Kai J, O'Dwyer A, Griffiths F. Primary care groups and research networks: opportunities for R&D in context. Br J Gen Pract (2000) 50:91–92.[Web of Science][Medline]
4 Mold JW, Peterson KA. Primary care practice-based research networks: working at the interface between research and quality improvement. Ann Fam Med (2005) 3(suppl 1):S12–S120.
5 Vickers M, Hand L, Hand C. The MRC general practice research framework. In: Research Opportunities in Primary Care—Carter Y, Thomas C, eds. (1999) Oxford: Radcliffe Medical Press. 29–42.
6 Hammersley V, Hippisley-Cox J, Wilson A, Pringle M. A comparison of research general practices and their patients with other practices—a cross-sectional survey in Trent. Br J Gen Pract (2002) 52:463–468.[Web of Science][Medline]
7 Tilyard MW, Dovey SM, Spears GF. Biases in estimates from the RNZCGP computer research group. N Z Med J (1995) 108:118–121.[Web of Science][Medline]
8 Harcourt SE, Edwards DE, Fleming DM, Smith RL, Smith GE. How representative is the population covered by the RCGP spotter practice scheme? Using Geographical Information Systems to assess. J Public Health (Oxf) (2004) 26:88–94.[CrossRef][Medline]
9 General Medical Services Contract. http://www.dh.gov.uk/PolicyAndGuidance/HumanResourcesAndTraining/ModernisingPay/GPContracts/fs/en (accessed on October 7, 2006).
10 Doran T, Fullwood C, Gravelle H, et al. Pay-for-performance programs in family practices in the United Kingdom. N Engl J Med (2006) 355:375–384.
11 Epstein AM, Lee TH, Hamel MB. Paying physicians for high-quality care. N Engl J Med (2004) 350:406–410.
12 Department of Health. http://www.dh.gov.uk/assetRoot/04/07/86/66/04078666.pdf (accessed on October 7, 2006).
13 NHS Information Centre. http://www.ic.nhs.uk/services/qof (accessed on October 7, 2006).
14 Hippisley-Cox J, Pringle M, Coupland C, Hammersley V, Wilson A. Do single handed practices offer poorer care? cross sectional survey of processes and outcomes. Br Med J (2001) 323:320–323.
15 Mant JW, Richards SH, Hobbs FD, et al. Midlands Research Consortium of general practice. Protocol for Birmingham Atrial Fibrillation Treatment of the Aged study (BAFTA): a randomised controlled trial of warfarin versus aspirin for stroke prevention in the management of atrial fibrillation in an elderly primary care population [ISRCTN89345269]. BMC Cardiovasc Disord (2003) 3:9.[CrossRef][Medline]
16 Macfarlane F, Shaw S, Greenhalgh T, Carter YH. GP Practices as emergent research organisations—a qualitative study into organisational development. Fam Pract (2005) 22:298–304.
17 Jowett S, McLeod J, Wilson S, Hobbs R. How research active is primary care? Br J Gen Pract (2000) 50:387–389.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Ives, H. Draper, S. Damery, and S. Wilson Do family doctors have an obligation to facilitate research? Fam. Pract., December 1, 2009; 26(6): 543 - 548. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||