Date Published:
2010
Citation-Indexed Journal:
Health Policy and Planning. Volume 25, Issue 2, March 2010, Pages 162-169
Citation:
A medication-estimated health status measure for predicting primary care visits: The long-term therapeutic groups index.
Authors:
Dhabali, A.A.H.
Awang, R.
Awang, R.
Abstract:
Background:
Managed care is one of the means advocated for health care reforms. The Malaysian government has proposed managed care for its citizens. In the Malaysian private health care sector, managed care is practised on a small scale with crude risk adjustment. The main determinant of an individual's health service utilization is their health status (HS). HS is used as a risk adjuster for capitation payment. Prescribed medications represent a useful source for HS estimation. We aimed to develop and validate a medication-based HS estimate and to incorporate it in the Andersen model of health service utilization. This is a preparatory step in studying the feasibility of developing a model for risk assessment in the Malaysian context.
Methods:
Data were collected retrospectively from an academic year from computerized databases in University Sains Malaysia (USM) about users of USM primary care services. A user is a USM health scheme beneficiary who made at least one visit in the academic year to USM-assigned primary care providers. Socio-demographic variables, enrolment period, medications prescribed and number of visits were also collected. Chronic illness medications and some non-chronic illness medications were used to calculate the Long-Term Therapeutic Groups Index (LTTGI) which is an estimate of the HS of users. Using a random 50 of users, weighted least square methods were used to develop a model that predicts a user's number of visits. The other 50 were used for validation.
Results:
Socio-demographic variables explained 15 of variability in number of primary care visits among users. Adding the LTTGI improved the explanatory power of the model to 36 (P < 0.001). A similar contribution of the LTTGI was noted in the validation.
Conclusions:
The Long-Term Therapeutic Groups Index was successfully developed. Variability in number of primary care visits can be predicted by LTTGI-based models.
Managed care is one of the means advocated for health care reforms. The Malaysian government has proposed managed care for its citizens. In the Malaysian private health care sector, managed care is practised on a small scale with crude risk adjustment. The main determinant of an individual's health service utilization is their health status (HS). HS is used as a risk adjuster for capitation payment. Prescribed medications represent a useful source for HS estimation. We aimed to develop and validate a medication-based HS estimate and to incorporate it in the Andersen model of health service utilization. This is a preparatory step in studying the feasibility of developing a model for risk assessment in the Malaysian context.
Methods:
Data were collected retrospectively from an academic year from computerized databases in University Sains Malaysia (USM) about users of USM primary care services. A user is a USM health scheme beneficiary who made at least one visit in the academic year to USM-assigned primary care providers. Socio-demographic variables, enrolment period, medications prescribed and number of visits were also collected. Chronic illness medications and some non-chronic illness medications were used to calculate the Long-Term Therapeutic Groups Index (LTTGI) which is an estimate of the HS of users. Using a random 50 of users, weighted least square methods were used to develop a model that predicts a user's number of visits. The other 50 were used for validation.
Results:
Socio-demographic variables explained 15 of variability in number of primary care visits among users. Adding the LTTGI improved the explanatory power of the model to 36 (P < 0.001). A similar contribution of the LTTGI was noted in the validation.
Conclusions:
The Long-Term Therapeutic Groups Index was successfully developed. Variability in number of primary care visits can be predicted by LTTGI-based models.
