Allocating census data to general practice populations: implications for study of prescribing variation at practice...

Area of Science:

• Public Health
• Biostatistics
• Geographic Information Systems

Background:

• Accurate demographic data is crucial for health service planning.
• Assigning census data to specific populations, like general practices, presents methodological challenges.
• Estimating the proportion of elderly patients (aged over 64) is important for resource allocation.

Purpose of the Study:

• To evaluate methods for allocating census data to general practice populations.
• To test the accuracy of different procedures in estimating the proportion of patients aged over 64 within these populations.

Main Methods:

• Utilized patient postcodes and a postcode-to-census district directory to geolocate patients.
• Employed four distinct allocation procedures with varying levels of census geography.
• Compared predicted proportions of patients aged over 64 with actual figures from practice registers.

Main Results:

• All four allocation procedures showed significant correlations (P < 0.0005) between predicted and actual proportions of patients aged over 64.
• However, the predicted ranges for the proportion of elderly patients were narrower than the actual observed ranges.
• Significant discrepancies were found between predicted and actual percentages for all tested methods.

Conclusions:

• Despite significant correlations, the allocation procedures demonstrated limitations in accurately predicting the proportion of elderly patients.
• The methods particularly struggled to predict values at the extreme ends of the distribution.
• These findings raise concerns about the validity of using similar techniques for allocating census variables to general practice populations.