Talk to the Veterans Crisis Line now
An official website of the United States government
Timbie JW (Ann Arbor COE), Hayward RA
(Ann Arbor COE), Vijan S
(Ann Arbor COE)
Objectives:
To assess the impact of variation in patients' responses to and tolerance of treatments prescribed to control diabetes risk factors on rates of goal attainment.
Methods:
We used Monte Carlo simulation to estimate the effectiveness of an aggressive treatment regimen designed to lower diabetes risk factors (LDL-cholesterol, hemoglobin A1c, and blood pressure) in pursuit of aggressive risk factor targets (LDL: 130 mg/dl, A1c: 7%, BP: 130/80 mmHg). Treatment regimens included 5 incremental doses of statins, 4 steps of antihyperglycemic therapy, and 8 steps of antihypertensive medications. We abstracted mean treatment efficacy parameters and discontinuation rates from meta-analyses of placebo-controlled trials and conducted our own meta-analysis to obtain estimates of variation in treatment efficacy. We adjusted the variance estimates for measurement error and non-compliance, both of which inflate variance. Baseline risk factor data were from the National Health and Nutrition Examination Survey.
Results:
In our meta-analysis we found that heterogeneity in treatment efficacy was high and varied across risk factors. Coefficients of variation, defined as the standard deviation in efficacy relative to the mean were 40% for statin therapy, 73-99% for the antihyperglycemic treatments, and 62%-117% for the antihypertensives. All-cause discontinuation rates varied from 5% to 27% across therapies. Both factors led to simulated goal attainment rates of 84% for LDL, 55% for A1c and 74% for blood pressure. Substituting parameters more likely to characterize a typical clinic population lowered attainment rates to 67%, 43%, and 60%, respectively. Over 50% of subjects ended the intensification regimen on five or more therapies.
Implications:
Natural variation between patients in their clinical response to and tolerance of treatments will cause a significant proportion of patients to fail to reach goals, despite the use of multiple medications. Intermediate outcomes account for neither of these factors in evaluating performance. Moreover, in seeking to avoid low quality ratings, these measures provide an implicit incentive to use non-standard therapies of dubious benefit with poorly established safety profiles after conventional therapies are exhausted.
Impacts:
Although commonly endorsed by national quality measurement organizations, intermediate outcome measures should be considered flawed performance measures. Low-risk populations are likely to receive little benefit from aggressive treatment but will be unnecessarily exposed to the harms of treatment. Alternative measures, such as process measures of timely intensification, which allow for exceptions, should be considered to enhance diabetes quality measurement.
