Multimorbidity indices improve the prediction of consultation costs in primary care in the UK

INTRODUCTION: As the population ages, the number of people with multiple problems is likely to increase. There is growing recognition of the importance of multimorbidity as an entity in itself and a belief that management of patients with multiple problems may require resources over and above those needed to treat each condition individually. This has implications for budgetary management in primary care; however no standard measure of multimorbidity for this purpose has been established. A number of indices of multimorbidity have been derived with the main aims of describing prevalence and predicting outcomes. These have focused predominantly on secondary care in the United States. In this study we aim to investigate the relationship between multimorbidity and consultation costs in primary care in the UK. This is the first study we are aware of which investigates the relationship between multimorbidity and costs in this setting. METHODS: We used data on a stratified sample of 85,709 individuals aged over 18 years from 182 practices in the General Practice Research Database (GPRD). We used all historic patient diagnoses to measure patient-level multimorbidity using: QOF chronic disease count; Charlson Index score; count of Expanded Diagnostic Clusters (EDCs) identified by the John Hopkins ACG System; and average number of drugs prescribed per year over a 2-year period. We estimated patient-level cost of primary care consultations over the subsequent 12 months. We related cost to age, sex, deprivation and multimorbidity using Generalised Linear Models (GLMs), and assessed model performance using a variety of fit statistics including a deviance-based R-squared measure. RESULTS: The model including age, sex, deprivation and practice ID alone explained 11% of observed consultation costs. Inclusion of the number of prescribed drugs, count of EDCs, QOF disease count, or Charlson Index score increased this to 27%, 21%, 18%, and 14% respectively. All models suggest a reasonably linear relationship between consultation costs and the number of chronic conditions. CONCLUSIONS: Multimorbidity indices improve the prediction of future consultation costs. These indices can be constructed easily using routinely recorded General Practice data, and therefore use of these indices could help to improve budgetary management in primary care.

Transcript

1. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Exploration of the relationship between multimorbidity and use of primary care resources in the UK Sam Brilleman1, Sandra Hollinghurst1, Sarah Purdy1, Hugh Gravelle3, Frank Windmeijer2, Chris Salisbury1 1 School of Social and Community Based Medicine, University of Bristol on behalf of the National School for Primary Care Research 2 Department of Economics, University of Bristol 3 University of York, and linked to Manchester via NPCRDC
2. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Background • Increasing number of individuals living with multimorbidity • Long-term chronic illness managed predominantly in primary care setting • May be useful in budget setting • Previous work: USA/Secondary care
3. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Aim • To explore the relationship between multimorbidity and the use of primary care resources in the UK • Objectives: – How do different measures compare when used to predict primary care costs? – What is the nature of the relationship between multimorbidity and cost?
4. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Data • General Practice Research Database (GPRD) – 86,100 patients from 174 practices – Diagnoses up until 31st March 2007 • Used to measure multimorbidity – Resource use data for one year beginning on 1st April 2007 • Used to estimate cost
5. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Diagnoses recorded by GP Diagnoses grouped into specific chronic diseases Simple count of 17 chronic diseases included in QOF 17 diseases weighted (by association with mortality) Sum of disease weights All diagnoses input to ACG software Diagnoses grouped based on disease area (EDCs) Simple count of 114 (of 264) disease groups (EDCs) defined as chronic Diagnoses grouped based on expected resource need (ADGs) Individuals allocated to mutually exclusive categories (ACGs) based on age, gender and combination of ADGs QOF disease count Charlson Index score EDC count Mutually exclusive ACG categories
6. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Diagnoses recorded by GP Diagnoses grouped into specific chronic diseases Simple count of 17 chronic diseases included in QOF 17 diseases weighted (by association with mortality) Sum of disease weights All diagnoses input to ACG software Diagnoses grouped based on disease area (EDCs) Simple count of 114 (of 264) disease groups (EDCs) defined as chronic Diagnoses grouped based on expected resource need (ADGs) Individuals allocated to mutually exclusive categories (ACGs) based on age, gender and combination of ADGs QOF disease count Charlson Index score EDC count Mutually exclusive ACG categories
7. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Diagnoses recorded by GP Diagnoses grouped into specific chronic diseases Simple count of 17 chronic diseases included in QOF 17 diseases weighted (by association with mortality) Sum of disease weights All diagnoses input to ACG software Diagnoses grouped based on disease area (EDCs) Simple count of 114 (of 264) disease groups (EDCs) defined as chronic Diagnoses grouped based on expected resource need (ADGs) Individuals allocated to mutually exclusive categories (ACGs) based on age, gender and combination of ADGs QOF disease count Charlson Index score EDC count Mutually exclusive ACG categories
8. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Diagnoses recorded by GP Diagnoses grouped into specific chronic diseases Simple count of 17 chronic diseases included in QOF 17 diseases weighted (by association with mortality) Sum of disease weights All diagnoses input to ACG software Diagnoses grouped based on disease area (EDCs) Simple count of 114 (of 264) disease groups (EDCs) defined as chronic Diagnoses grouped based on expected resource need (ADGs) Individuals allocated to mutually exclusive categories (ACGs) based on age, gender and combination of ADGs QOF disease count Charlson Index score EDC count Mutually exclusive ACG categories
9. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Diagnoses recorded by GP Diagnoses grouped into specific chronic diseases Simple count of 17 chronic diseases included in QOF 17 diseases weighted (by association with mortality) Sum of disease weights All diagnoses input to ACG software Diagnoses grouped based on disease area (EDCs) Simple count of 114 (of 264) disease groups (EDCs) defined as chronic Diagnoses grouped based on expected resource need (ADGs) Individuals allocated to mutually exclusive categories (ACGs) based on age, gender and combination of ADGs QOF disease count Charlson Index score EDC count Mutually exclusive ACG categories
10. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Diagnoses recorded by GP Diagnoses grouped into specific chronic diseases Simple count of 17 chronic diseases included in QOF 17 diseases weighted (by association with mortality) Sum of disease weights All diagnoses input to ACG software Diagnoses grouped based on disease area (EDCs) Simple count of 114 (of 264) disease groups (EDCs) defined as chronic Diagnoses grouped based on expected resource need (ADGs) Individuals allocated to mutually exclusive categories (ACGs) based on age, gender and combination of ADGs QOF disease count Charlson Index score EDC count Mutually exclusive ACG categories
11. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research 0 10 20 30 40 50 60 0 1 2 3 4 5 6+ QOF chronic disease count 0 10 20 30 40 50 60 70 Percentage of sample (%) 0 1 2 3 4 5 6 7+ Charlson Index score 0 5 10 15 20 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1718+ Expanded Diagnostic Cluster (EDC) count 0 2 4 6 8 10 12 Percentage of sample (%) 1731 5110 5200 1000 600 1600 1100 500 400 300 1300 1711 2400 1712 700 2800 1800 4000 2100 3900 2200 1721 1732 1722 3400 2500 3200 1500 3300 1751 1752 4710 1742 4310 1741 1200 4320 3500 4720 800 900 4810 3700 4820 2300 1762 4100 4330 1761 4410 2600 1400 2700 3600 4830 4730 1772 1771 4420 4910 4430 5040 4920 4930 4940 5050 5060 5070 Adjusted Clinical Group (ACG) Distribution of each measure of multimorbidity
12. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Costing Resources Total annual £ per patient Consultations Prescription drugs Tests and investigations
13. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Costing Resources Consultations Combination of… -Consultation type (surgery consultation, home visit, clinic, out of hours, telephone) and… - Primary staff role (GP, practice nurse, district nurse, physiotherapist, counsellor, etc.) Additional cost for administrative activities performed by a receptionist, administrator, or secretary
14. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Costing Resources Prescription drugs - Matched to cost using drug name, strength and formulation - Cost based on total quantity prescribed (eg. number of tablets, number of millilitres for liquids)
15. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Costing Resources Tests and investigations In-practice (during consultation): cost for consumables only (e.g. pregnancy tests or urine dipstick) Laboratory tests: cost by pathology discipline (source: NHS Reference Costs) Hospital based investigations: costed individually (source: NHS Reference Costs)
16. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Costing Resources Total annual £ per patient Consultations Prescription drugs Tests and investigations
17. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Patient cost, by age category and deprivation decile 0 200 400 600 800 20-29 30-39 40-49 50-59 60-69 70-79 80-89 90+ Age category 0 200 400 600 800 Mean annual patient cost (£) 1 2 3 4 5 6 7 8 9 10 Deprivation decile (10 most deprived)
18. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research How well does each multimorbidity measure predict patient cost? Model Deviance-based R-squared Age, gender, deprivation, and practice + Charlson Index score + QOF disease count + ACG categories + EDC count
19. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research How well does each multimorbidity measure predict patient cost? Model Deviance-based R-squared Age, gender, deprivation, and practice 14% + Charlson Index score 21% + QOF disease count 27% + ACG categories 29% + EDC count 31%
20. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research How well does each multimorbidity measure predict patient cost? Model Deviance-based R-squared Age, gender, deprivation, and practice 14% + Charlson Index score 21% + QOF disease count 27% + ACG categories 29% + EDC count 31%
21. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Predicted mean annual patient cost + 0 500 1000 1500 2000 2500 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Predicted mean annual patient cost (£) EDC count
22. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Conclusions • Including measures of multimorbidity improved model fit • A count of Expanded Diagnostic Clusters (EDCs) performed best in a model predicting patient cost in primary care • Increases in patient cost were roughly proportional to the number of EDCs
23. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research Thank you This project was funded by the NIHR School for Primary Care Research
24. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research
25. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research References (1) • QOF/EDC count “chronic” classifications: – Salisbury, C., Johnson, L., Purdy, S., Valderas, J. M., & Montgomery, A. A. 2011, "Epidemiology and impact of multimorbidity in primary care: a retrospective cohort study", British Journal of General Practice, vol. 61, no. 582, pp. 18-24. • Charlson Index score: – Khan, N. F., Perera, R., Harper, S., & Rose, P. W. 2010, "Adaptation and validation of the Charlson Index for Read/OXMIS coded databases", BMC Family Practice, vol. 11. – Deyo, R. A., Cherkin, D. C., & Ciol, M. A. 1992, "Adapting A Clinical Comorbidity Index for Use with Icd-9-Cm Administrative Databases", Journal of Clinical Epidemiology, vol. 45, no. 6, pp. 613-619. – Charlson, M. E., Pompei, P., Ales, K. L., & Mackenzie, C. R. 1987, "A New Method of Classifying Prognostic Co-Morbidity in Longitudinal- Studies - Development and Validation", Journal of Chronic Diseases, vol. 40, no. 5, pp. 373-383.
26. School for Primary Care Research Increasing the evidence base for primary care practice The School for Primary Care Research is a partnership between the Universities of Birmingham, Bristol, Keele, Manchester, Nottingham, Oxford, Southampton and UCL and is part of the National Institute for Health Research. School for Primary Care Research References (2) • John Hopkins ACG System: – Johns Hopkins Bloomberg School of Public Health. The Johns Hopkins ACG® Case-mix System Version 8.2. Baltimore, 2008. • Deviance-based R-squared: – Cameron, A. C. & Windmeijer, F. A. G. 1997, "An R-squared measure of goodness of fit for some common nonlinear regression models", Journal of Econometrics, vol. 77, no. 2, pp. 329-342.