Family Practice Vol. 21, No. 4, 355-363
Family Practice Vol. 21, No. 4 © Oxford University Press 2004, all rights reserved.
Adjusting for case mix and social class in examining variation in home visits between practices
a Public Health Policy Unit, School of Public Policy, University College London, London WC1H 9QU, b Medical Statistics Unit, Research and Development Directorate, University College London Hospitals NHS Trust, London WIP 9LL, c Department of Statistical Science, 1–19 Torrington Place, University College, London WC1E 6BT, UK and d Health Services Research and Development Center, Department of Health Policy and Management, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD 21205, USA
Correspondence to Caoimhe O Sullivan, Medical Statistics Unit, Research and Development Directorate, University College London Hospitals NHS Trust, Maple House, 149 Tottenham Court Road, London WIP 9LL, UK; E-mail: caoimhe.osullivan{at}uclh.nhs.uk
Received 23 October 2002; Revised 6 August 2003; Accepted 10 March 2004.
O Sullivan C, Omar RZ, Forrest CB and Majeed A. Adjusting for case mix and social class in examining variation in home visits between practices. Family Practice 2004; 21: 355–363.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAppendix 1. Home visits...Appendix 2. The Johns...References |
|---|
Objectives. The purpose of this study was to investigate whether adjusting for clinical case mix and social class explains more of the variation in home visits between general practices than adjusting for age and sex alone.
Methods. The setting was 60 general practices in England and Wales taking part in the 1 year Fourth National Morbidity Survey. The participants comprised 349 505 patients who were registered with one of the participating general practices for at least 180 days, and who had at least one consultation during the period. The outcome measure is whether or not a patient received a home visit in that year. A clinical case mix category (morbidity class) based on 1 year's diagnostic information was assigned to each patient using the Johns Hopkins Adjusted Clinical Groups (ACG) Case Mix System. The social class measure was derived from occupation and employment status and is similar to that of the 1991 UK census. Variations in home visits between practices were examined using multilevel logistic regression models. The variability between practices before and after adjusting for clinical case mix and social class was estimated using the intracluster correlation coefficient (ICC).
Results. The overall percentage of patients receiving a home visit over the 1 year study period was 17%, and this varied from 7 to 31% across the 60 practices. The percentage of the total variation in home visits attributable to differences between practices was 2.5% [95% confidence interval (CI) 1.4–3.2%] after adjusting for age and sex. This reduced to 1.6% (95% CI 1.1–2.4%) after taking into account morbidity class. The results were similar when social class was included instead of morbidity class. Morbidity and social class together reduced variation in home visits between practices to 1.5% (95% CI 1.1–2.2%).
Conclusions. Age, sex, social class and clinical case mix are strong determinants of home visits in the UK. Adjusting for morbidity and social class results in a small improvement in explaining the variability in home visits between practices compared with adjusting for age and sex alone. There is far more variation between patients within practices; however, it is not straightforward to examine the factors influencing this variation. In addition to morbidity and social class, there could also be other unmeasured factors such as varying patient demand for home visits, disability or differences in GP home visiting practice style that could influence the large within-practice variability observed in this study.
Keywords. Case mix, home visits, practice variation, primary care.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1. Home visits...Appendix 2. The Johns...References |
|---|
Primary care in England has undergone major changes over the last 15 years. A recent change is the creation of Primary Care Trusts (PCTs), which in April, 2002 replaced health authorities as the main purchasers of health services and as the body responsible for holding GPs' contracts. England's public health White Paper, Saving Lives: Our Healthier Nation, states that within the restructured NHS: "setting standards and measuring progress is now an integral part of the planning and delivery of services to patients in primary care".1 The NHS Plan also envisages greater monitoring of general practices' performance to ensure a high standard of care for patients and efficient use of services. Practice workload and performance have been shown to vary widely in areas such as consultation, referral and prescribing rates.2 Some of this variation can be attributable to differences in the demographic structure of practice populations.
Currently used methods of comparing practice performance, workload and resource utilization take into account the effects of age, sex and ecological measures of health and socio-economic status.3 These adjustments may be sufficient for larger populations such as those of PCTs but, because general practices are composed of much smaller populations, there are likely to be large differences among them in their clinical and socio-economic characteristics. Adjusting only for age and sex may mean that some GPs could be unfairly penalized for serving populations with a higher morbidity or ‘illness burden’.4 For example, a practice serving a sicker population will have a higher workload, which in turn may lead to higher prescribing, referral and hospital admission rates. Such adjustments may become increasingly important if the new GP contract has been implemented and the NHS is considering the publication of measures of general practice performance.5 If these measures are not adjusted appropriately for the case mix, erroneous conclusions may be drawn about the performance of individual general practices.
Home visits are an important aspect of GPs' workload, and home visiting patterns vary greatly from practice to practice.6 The overall practice morbidity, or clinical case mix of the practice, may be an important factor that should be allowed for in comparisons of home visits across practices. Similarly, a practice serving a population with a large proportion of patients from the lower socio-economic classes would be expected to have higher home visiting rates.
There are different methods of measuring clinical case mix available, and most of these depend on the most important or common morbidity of each patient.7,8 Unlike these, the Johns Hopkins ACG (Adjusted Clinical Groups) case mix system is a method of measuring clinical case mix based on patient morbidity.9–12 This population-based risk adjustment tool is well suited to the primary care setting and is widely used and tested in the USA. In this paper, we illustrate the use of the Johns Hopkins ACG case mix system in the UK primary care setting using data from the Fourth National Morbidity Survey, a 1 year prospective cohort study of 500 000 patients in 60 general practices in England and Wales.13 As a preliminary step in our study of case mix, we have focused on home visits, as the survey recorded 95% of home visits made. We investigated whether inclusion of clinical case mix and social class measures allowed us to explain more of the variation in home visits between practices than age and sex adjustment alone, which is the most common method of risk adjustment used in the UK at present. From a conceptual basis, the variability in home visits can be decomposed into practice and patient level variation. Because the focus of this paper is on the differences in home visiting patterns between practices, we only examine variation in home visits at the practice level.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1. Home visits...Appendix 2. The Johns...References |
|---|
Data source
The Fourth National Survey of Morbidity in General Practice was a 1 year prospective cohort study of
500 000 patients registered with 60 general practices in England and Wales.13 The main objective of the survey was to examine the workload and pattern of disease in general practice in relation to the age, sex and socio-economic status of patients. Patient diagnoses and types of consultations, including home visits, were collected for each patient contact, and these are linked to the socio-economic characteristics of each patient, making this a good data source for the purpose of our study.
Data recording and validating
Doctors and nursing staff from each practice attended three 2 day training sessions on the recording of morbidity data. Practices then collected data for 2–4 weeks before the start of the survey. The data collection software was designed so that all diagnoses were automatically coded using the Read classification system.14 The data were analysed and any errors and inconsistencies were reported to the practices and amended. Information on socio-economic status was obtained on 83% of patients in the survey by direct interview with specially trained interviewers. The interview method was tested successfully for feasibility and acceptability before the survey. The social class measure we used was derived from occupation and employment status. Social class was grouped in a similar way to the UK census groupings (Table 1). An International Classification of Diseases Ninth Revision (ICD-9) code was assigned at the Office of Population Censuses and Surveys.15 Data were well validated and the collection and validation process is described in detail in the MSGP4 publication.13
|
Study population
The study population for the Fourth National Morbidity Survey was a 1% sample (502 493 patients) of the population from 60 general practices in England and Wales. All patients in these practices who were on the NHS age/sex register were included. The sample was representative of the population of England and Wales for characteristics such as age, sex, social class and housing tenure, and under-representative of ethnic minority groups and people living alone.13
Exclusions
The following patients were excluded: 137 273 (27%) did not consult in the study period; 15 682 (4%) did consult but were registered at a practice for <6 months; and 33 patients had consulted but had no diagnosis information. After applying the exclusion criteria, the analysis data set comprised 349 505 patients.
The Johns Hopkins ACG system
The Johns Hopkins University ACG case mix system is a well-validated tool used to characterize the degree of overall morbidity in patients and populations.9 It is widely used in the USA for risk adjustment and performance profiling in primary care. Unlike other case mix systems, it was developed on the premise that clustering of morbidity is a better predictor of health service resource use than the presence of specific diseases. 12 The ACG grouping mechanism is openly available for scrutiny, and thus it is possible that we may adapt it to better suit the UK health care system.
ADGs, ACGs and morbidity class
Each ICD-9 diagnosis code for each patient is mapped to one of 32 diagnosis groups known as ADGs (aggregated diagnosis groups). A small proportion (1%) of ICD-9 codes could not be assigned an ADG.16 Diagnoses within the same ADG are of similar severity and expected need for health care resources over time. These diagnosis groups are clustered according to the following clinical characteristics: (i) duration of the condition (acute, recurrent or chronic); (ii) severity of the condition (e.g. minor and stable versus major and unstable); (iii) diagnostic certainty (symptoms versus documented disease); (iv) aetiology of the condition (infectious, injury or other); and (v) likelihood of specialty care involvement (medical, surgical, obstetric, haematology, etc.).
Patients are assigned to an ADG if they have one or more of the ADGs constituent diagnoses and, hence, each patient may have between zero and 32 ADGs. Each individual can also be assigned a single mutually exclusive ACG, which is derived from a combination of age, sex, presence of specific ADGs, number of major ADGs and total number of ADGs. The ACG groupings contain individuals with similar needs for health care resources based on overall expenditures. Patients with similar predicted (or expected) overall utilization may be assigned different ACGs if they have different epidemiologic patterns of morbidity. For analysis purposes, the ACGs are collapsed to one of eight mutually exclusive morbidity classes, known as resource utilization bands (RUBs), higher numbers indicating higher morbidity. 10
Statistical methods
Home visit frequency distributions and odds ratios (ORs) were calculated for each of the age, sex, morbidity and social class groupings. Home visit frequency distributions were also computed by ADG.
Multilevel logistic regression models (each with a random intercept) were used to investigate important determinants of home visits.17 Four separate models with different sets of predictors were examined, the outcome of interest being whether or not a patient had a home visit in the 1 year study period. The sets of predictors included were: (i) age group and sex; (ii) morbidity class; (ii) social class; and (iv) morbidity class and social class. Model diagnostics included plotting standardized residuals against their normal scores for each model to ensure that the assumption of normality was satisfied, and checking for overdispersion. Adjusted ORs were computed from the results of each of the models. An intracluster correlation coefficient (ICC) was estimated from each of the models and used to assess how the variation in home visits between practices was altered when comparing models with different sets of predictors.18 The total variation in home visits is due to different characteristics influencing home visits both between patients within practices (within-practice variance) and from practice to practice (between-practice variance). The ICC quantifies the proportion of this total variance in home visits that is attributable to differences between practices.19 The Turner formula for estimating ICC relates the overall proportion of home visits with an estimate of the between-practice variance and does not require a direct estimate of the within-practice variance. Thus, comparisons of ICCs resulting from the four models will not be affected by different within-practice variance estimates.18 Confidence intervals (CIs) for the ICCs were constructed using resampling methods.17,20
Using the results obtained on fitting the model with age and sex as predictors, the probability (and 95% interval) of a patient having at least one home visit was estimated and plotted for each age group for both males and females. Similarly, after fitting the model with morbidity as a predictor, the probability of home visits (and 95% range) was estimated and plotted for each morbidity class.
Descriptive statistics were computed using Stata 7.0; graphs were drawn in MS Excel; and multilevel modelling was carried out with MlwiN v1.10.21,22
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1. Home visits...Appendix 2. The Johns...References |
|---|
Demographics
The total number of patients included in the analysis was 349 505 (55% female, 45% male). Table 1 presents the number of patients by age, sex, morbidity group and social class, the number of home visits and the ORs for each of these groups. Of all patients included in this study, 17% had at least one home visit over the study period. The crude percentage of patients receiving a home visit ranged from 7 to 31% across practices, with a median of 18%. Home visits showed a bimodal distribution and were lowest in the 16–44 age group, with peaks at 0–4 years and for those aged 65 years and over. The median (range) percentage receiving home visits according to age in the children (0–15 years), adult (16–64 years) and elderly (65 years onwards) populations were 20% (4–43%), 11% (4–21%) and 38% (22–56%), respectively. The odds of home visits are greatest for those aged
85 years [OR = 8.1 (7.57–8.66)], while the odds of home visits are lowest for those aged between 16 and 64 years. The proportion of patients in each ADG with at least one home visit over the study period are presented in Appendix 1. Appendix 2 contains a paragraph on the grouping mechanism used by the Johns Hopkins ACG case mix system.
Results from models
The ORs in Table 1 show that home visits are more common in the oldest and youngest age groups, and less common in adults. Age and sex show a strong association with home visits. Home visits are less common for males than females [OR = 0.75 (0.74–0.79)]. The odds of having a home visit increase steadily with increasing morbidity. The odds of home visits are lowest for the highest social classes.
The adjusted ORs from fitting the four models (Table 2) are similar to those of Table 1, although the effect of gender is not as large [OR = 0.87 (0.85–0.88)]. In examining which of the factors explain most of the variability, the ICCs in Table 2 show that most variation in home visits is occurring between patients within the same practices rather than from practice to practice. Model 1 shows that 2.5% (95% CI 1.4–3.2%) of the total variation in home visits is attributable to practices after taking into account age and sex.
|
We then included morbidity class as a predictor for home visits to see if this could improve on the age and sex model in explaining variation in home visits across practices. There is a highly statistically significant association between morbidity class and home visits (P < 0.001). The level of need for home visits increases steadily with increasing morbidity, as indicated by the ORs in Table 2. After adjustment for morbidity, the odds of having a home visit are almost 11 times greater in the sickest morbidity class compared wih the healthiest morbidity class (OR = 10.8, 95% CI 10.3–11.2). Estimates of the log(odds), the between-practice variance and the percentage ICCs and associated CIs from each of the models are included at the end of Table 2. The ICC for model 2, which fitted only morbidity class to the home visits, was 1.6% (95% CI 1.1–2.4%). (A model with age group, sex and morbidity class as explanatory variables did not reduce the variability any further.) The ICC for model 3 fitting only social class was similar to that from model 2: 1.6% (95% CI 1.1–2.8%). The decrease in ICC from a model including age and sex only is not likely to be statistically significant for models 2 and 3 as the CIs overlap considerably. Model 4 with explanatory variables morbidity and social class resulted in the greatest reduction in ICC to 1.5% (95% CI 1.1–2.2%) and, again, the CIs overlap.
The probability of home visits is highest for children and the elderly compared with adults, and the 95% intervals demonstrate that the variability of these estimates for children and elderly patients is also higher (Fig. 1). Home visit probabilities were highest for the sickest patients and, again, more variability is apparent in these estimates for the sickest patients (Fig. 2). The variability among the morbidity groups in Figure 1 does not appear to be noticeably lower than that for the age groups in Figure 2, and this is reflected in the small non-significant decrease in ICC after adjusting for morbidity compared with adjusting for age and sex.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1. Home visits...Appendix 2. The Johns...References |
|---|
The lack of appropriate adjustment for case mix is an important limitation of many previous studies because case mix has been shown to have a major impact on GPs' workload and performance.4 We used a case mix adjustment system developed in the USA in an attempt to explain the variation in one important area of GPs' workload in the UK. We found that the Johns Hopkins ACG Case Mix System is a strong determinant of home visits for patients in the UK.
Figures 1 and 2 show that patterns of association are very similar for both intermediate age groups and intermediate morbidity groups, while the elderly, children and the sickest patients are more likely to have had a home visit and the variability in home visits for these groups is also highest. Crude home visits varied from 7 to 31% across practices. However, this variation is composed of variability arising from both practice and patient levels. Adjusting for clinical case mix resulted in a small improvement in explaining variability in home visits between practices compared with adjusting for age and sex only. However, this improvement is not likely to be statistically significant. Adjusting for social class resulted in a similar improvement, while adjusting for both clinical case mix and social class explained slightly more of the variation. Most of the variation is occurring between patients within the same practices rather than between practices.
Strengths and weaknesses of the study
A major strength of this study is that it is population based as general practices in the UK register patients from all sections of society. Hence, unlike studies of US health maintenance organizations, no socio-economic group was excluded. We also used data collected prospectively for 1 year as part of a morbidity survey and recorded to a high standard. Individual level socio-economic information was collected during the survey, and we were thus able to compare the effect of adjustment for clinical case mix with adjustment for social class. Both of these had a similar effect in reducing variation between practices. One weakness of the study is that only data from primary care were used to generate the measures of case mix. Using data from secondary care in addition to primary care data may have resulted in more accurate measures of case mix. Another limitation is that the method of assigning case mix was developed in the USA and may need some modification to maximize its utility in the UK. Our outcome measure was whether or not patients had a home visit, and it is possible that outcomes in the form of cost data could give a greater insight into the variation in home visits across practices. Finally, there may be an underlying tendency for certain practices to see their patients as more sick and therefore code them that way. This is a potential source of confounding that cannot be evaluated with these data.
Comparisons with previous research
Previous primary care research in the UK has not often adjusted for case mix measures that are based on patient morbidity. Research from North America suggests that adjusting for clinical case mix is an important factor when comparing practice profiles, e.g. when comparing referral rates among primary care physicians.4 In a recent study, the ACG system was used to control for case mix in a comparison of referral rates among primary care physicians in the USA and UK. 23
Implications
We found that clinical case mix is a strong determinant of general practice home visits. As such, it holds promise as a targeting tool to predict which patients or patient groups are most likely to receive home visit services. Both clinical case mix and social class explain a similar proportion of the between-practice variation. Therefore, it may be useful to adjust for social class in studies comparing general practice home visits, when diagnostic information is not available. We have shown that adjusting for clinical case mix and social class results in a small improvement in explaining variability in home visits between practices compared with adjusting for age and sex. There is far more variation between patients within practices; however, it is not straightforward to examine the factors influencing this variation. In addition to clinical case mix and social class, there could also be other unmeasured factors such as varying patient demand for home visits, disability or differences in GP home visiting practice style that could influence the within-practice variability observed in this study. Care should be taken when comparing crude rates of performance indicators between practices. Appropriate consideration should be given to the different levels of variation resulting from the clustered nature of general practice data.
In the future, general practices' performance will come under much greater scrutiny. Experience from the USA shows that it is important to take into account case mix when assessing practice performance and the use of resources to avoid ‘good apples being labelled as bad’.4 Clinical case mix measurement may have a role in the evaluation of other outcomes in primary care.
Conclusions
Our findings suggest that case mix measurement techniques developed in the USA can be applied to primary care populations in the UK. We have shown that clinical case mix is a strong determinant of home visits in the UK. Adjusting for social class may be useful when comparing home visits between practices in situations where diagnostic information is not available. Morbidity and social class adjustment is a small improvement in explaining variability in home visits between practices compared with adjusting for age and sex. However, there is far more variation between patients within practices and it is not straightforward to examine the factors influencing this variation. In addition to clinical case mix and social class, there could also be other unmeasured factors, such as varying patient demand for home visits, disability or differences in GP home visiting practice style. Case mix may also be an important factor in studies of other aspects of between-practice variation. The use of clinical case mix measurement systems in the UK should be explored further, particularly for producing case mix adjusted measures of practice performance and to help in determining budgetary allocations for general practices.
|
Appendix 1. Home visits by ADG |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1. Home visits...Appendix 2. The Johns...References |
|---|
|
|
Appendix 2. The Johns Hopkins Adjusted Clinical Group (ACG) case mix system |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1. Home visits...Appendix 2. The Johns...References |
|---|
The John Hopkins Adjusted Clinical Group (ACG) case mix system is a measure of a patient's health status that uses their medical history to place them in one of
100 different groups, through a two-stage process. The first step is to assign each diagnosis in the patient's past medical history to one of 32 ADG (aggregated diagnosis group) morbidity clusters. Diagnoses are clustered based on a number of criteria including clinical similarity, the likelihood that the condition will persist or recur, and the likelihood that the patient will return to their physician for treatment or will need a referral to a specialist. Patients are assigned to an ADG if they have one of more of the ADG's constituent diagnoses and, hence, can have between zero and 32 different ADGs. The combination of ADGs for each patient is then used along with information on age and sex to assign the patient to one of 100 mutually exclusive ACGs. The methods thus have some similarity to those used to assign hospital patients to Diagnostic Related Groups (DRGs) in the USA and Health Related Groups (HRGs) in the UK, but make use of all the diagnoses in the patient's medical history during a specified period of time and not just the diagnoses from a single episode of care. Further details are available from http://www.acg.jhsph.edu
|
Acknowledgments |
|---|
We thank Kevin Carroll for his help in extracting the data for this study. COS holds an NHS National Primary Care Researcher Development Award and AM holds an NHS National Primary Care Career Scientist Award, funded by the Department of Health. The Fourth National Survey of Morbidity in General Practice was funded by the Department of Health.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1. Home visits...Appendix 2. The Johns...References |
|---|
1 Department of Health. Saving Lives: Our Healthier Nation. (Cm 4386) London: Stationery Office; 1999.
2 Majeed A, Bindman AB, Weiner JP. Use of risk adjustment in setting budgets and measuring performance in primary care I: how it works. Br Med J 2001; 323: 604–607.
3 Carr-Hill RA, Rice N, Roland M. Socio-economic determinants of rates of consultation in general practice based on the fourth national morbidity survey of general practices. Br Med J 1996; 312: 1008–1012.
4 Salem-Schatz S, Moore G, Rucker M, Pearson SD. The case for case-mix adjustment in practice profiling: when good apples look bad. J Am Med Assoc 1994; 272: 871–874.
5 http://www.bma.org.uk/ap.nsf/Content/__Hub+GPC+contract
6 Aylin P, Majeed FA, Cook DG. Home visiting by general practitioners in England and Wales. Br Med J 1996; 313: 207–210.
7 Ellis R, Pope G, Iezzoni L. Diagnosis-based risk adjustment for medical capitation payments. Health Care Financing Rev 1996; 17: 101–128 (www.dxcg.com).[Web of Science][Medline]
8 Averill RF, Goldfield NI, Eisenhandler J et al. Development and Evaluation of Clinical Risk Groups (CRGs). Wallingford (CT): 3M Health Information Systems; 1999.
9 The Johns Hopkins University ACG case-mix system. http://acg.jhsph.edu
10 Weiner JP, Abrams C. ACG Virtual Library: Version 4.5 (February 2000) ACG Software Documentation/Users Manual. Health Services Research & Development Centre, The Johns Hopkins University; 1991–2001.
11 Weiner JP, Starfield BH, Steinwachs DM, Mumford LM. Development and application of a population-oriented measure of ambulatory care case mix. Med Care 1991; 29: 453–472.
12 Starfield B, Weiner J, Mumford L, Steinwachs D. Ambulatory care groups: a categorization of diagnoses for research and management. Health Serv Res 1991; 26: 53–74.[Web of Science][Medline]
13 Royal College of General Practitioners, the Office of Population Censuses and Surveys, and the Department of Health. Morbidity Statistics from General Practice. Fourth National Study 1991–1992. HMSO: London; 1995.
14 Saint-Yves IF. The Read clinical classification. Health Bull 1992; 50: 422–427.
15 International Classification of Diseases, Ninth Revision, Clinical Modification. New York: McGraw-Hill; 1996.
16 Majeed A, Bindman AB, Weiner JP. Use of risk adjustment in setting budgets and measuring performance in primary care II: advantages, disadvantages, and practicalities. Br Med J 2001; 323: 607–610.
17 Goldstein H. Multilevel Statistical Models. London: Edward Arnold/New York: Wiley; 1995.
18 Turner RM, Omar RZ, Thompson SG. Bayesian methods of analysis for cluster randomised trials with binary outcome data. Stat Med 2001; 20: 453–472.[CrossRef][Web of Science][Medline]
19 Diez Roux AV. A glossary for multilevel analysis. J Epidemiol Community Health 2002; 56: 588–594.
20 Goldstein, H, Browne, WJ, Rasbash, J. Partitioning variation in multilevel models. Understanding Statistics 2002; 1: 223–232.
21 Rasbash J, Browne W, Goldstein H et al. A User's Guide to MlwiN. Multilevel Models Project, Institute of Education, University of London; 2000.
22 StataCorp. Stata Statistical Software: Release 7.0. College Station (TX): Stata Corporation; 2001.
23 Forrest CB, Majeed A, Weiner JP, Carroll K, Bindman AB. Comparisons of specialty referral rates in the United Kingdom and the United States: retrospective cohort analysis. Br Med J 2002; 325: 370–371.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. Z Omar, C. O'Sullivan, I. Petersen, A. Islam, and A. Majeed A model based on age, sex, and morbidity to explain variation in UK general practice prescribing: cohort study BMJ, July 14, 2008; 337(jul14_2): a238 - a238. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. L Campbell Provision of primary care in different countries BMJ, June 16, 2007; 334(7606): 1230 - 1231. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||