ELICITING THE EFFECTS OF LIPID PROFILE ON QUALITY OF LIFE FOR TYPE 2 DIABETES PATIENTS

Purpose:  The association between the health-related quality of life (HRQOL) and cardiovascular risk factors is often investigated by surveys which may involve logical inconsistencies and the risk of bias due to sampling. We elicit HRQOLs with varying lipid levels from the perspective of a central policy maker following Adult Treatment Panel (ATP III) guidelines and its variant (ATP III*) for Type 2 diabetes patients.

Method:  We considered patients who are diagnosed with the disease at age 40 and modeled the evolution of their total cholesterol (TC) and high density lipoprotein levels (HDL) as two independent Markov chains in four different categories as low, medium, high and very high. We defined the progression of other cardiovascular risk factors as a function of age and formulated a Markov decision process model in which decisions to initiate lipid treatment are revisited biannually so as to maximize the total expected QALYs before a first major cardiovascular event, i.e., heart attack or stroke. We employed an inverse-optimization framework to seek the values of HRQOLs in different lipid profiles (relative to the best possible lipid profile) that make the guidelines as close as possible to being cost-effective among the threshold-structured, implementable policies. We analyzed the sensitivity of the elicited values of HRQOLs with respect to disutility of cholesterol treatment.

Result: Under both guidelines, all ranges of TC except very high, were ineffective on HRQOLs irrespective of the range of HDL and HRQOLs in low HDL were sharply less than those in other HDL ranges. Under ATP III, elicited values of HRQOLs were higher for males than for females. In almost all lipid profiles, HRQOLs were higher for both genders under ATP III* compared to ATP III (see Table 1 for details). While HRQOLs showed a broader variation for females than for males under ATP III, they varied from 0.90 to 1 for males and from 0.95 to 1 for females with respect to increasing HDL levels under ATP III*. With their perceived HRQOLs neither of the guidelines were more than 0.3 % away from being cost-effective.

Conclusion: Our analyses showed that guidelines can be closest of being cost-effective under HRQOLs which vary more with HDL than TC but do not exhibit any fixed behavior with treatment disutility.