Abstract
Since the Stockholm County Council introduced a system of purchasers and providers there has been a quest for population-based resource allocation models to allocate monies to purchasers of health care. In contrast to models used in Britain, Swedish models have been based on individual level data. This paper presents recent developments in the model used in Stockholm for all care except primary care, testing new statistical methods for compression and clustering of the matrices used and the effect of introducing diagnostic information in addition to the demographic and socio-economic information used before. We also show the effect of using more current data sources by replacing existing census variables with data from annually updated registers. Since the aim is to use the resource allocation models for prospective budgeting we test and evaluate the predictive power of the models one to two years ahead. Moreover, two calibration methods are compared: Cross-sectional modelling, based on data for one year only, versus prospective modelling, using population characteristics for one year and registered health-care costs for a following year. While models including diagnostic information are deemed valuable, the prospective models yield little improvement. Further, although it takes a combination of new variables to replace the census based model, the resulting model now implemented by Stockholm County Council has fewer estimated parameters.
| Original language | English |
|---|---|
| Journal | Health Care Management Science |
| Volume | 3 |
| Issue number | 2 |
| Pages (from-to) | 141-9 |
| Number of pages | 9 |
| ISSN | 1386-9620 |
| Publication status | Published - 2000 |
Keywords
- Adolescent
- Adult
- Aged
- Aged, 80 and over
- Child
- Child, Preschool
- Cluster Analysis
- Community Health Planning
- Diagnosis-Related Groups
- Forecasting
- Health Care Rationing
- Hospital Costs
- Humans
- Infant
- Middle Aged
- Models, Econometric
- National Health Programs
- Needs Assessment
- Registries
- Reproducibility of Results
- Residence Characteristics
- Socioeconomic Factors
- Sweden