High-Deductible Insurance and Delay in Care

IN RESPONSE: Drs. Goulet and Goldstein are concerned that we excluded patients with MI and CVA from the analysis and that doing so could have detracted from the validity of [our] findings. However, we did not exclude such patients; rather, our numerator measures did not include MI and stroke for several reasons. Recent research has shown that delayed care may be a key factor causing adverse events among patients with high-deductible insurance (1). Our study thus aimed to examine care for macrovascular disease that patients could delay, such as visits for symptoms or testing. We hypothesized that having high-deductible insurance would either not change or accelerate time to MI and stroke at the population level. This hypothesis is based on previous evidence (24), and the inference that the delays in care for macrovascular complications that we detected could lead to unrecognized macrovascular disease and thus earlier complications. Mixing outcome measures that we hypothesize might be delayed (symptoms, diagnostic tests, and treatment) and accelerated (major cardiovascular events) would fail to disentangle causal pathways and be less informative to clinicians, health insurers, and policymakers. We did not include MI and stroke in our major symptoms measure because they are not symptoms or signs detected during the history and physical examination but rather clinical diagnoses that require intense work-up. However, our denominator included patients with diabetes with and without MI or stroke, and we captured their major symptoms, diagnostic testing, or treatment related to macrovascular disease during the study period. Moreover, MI and stroke are rare among patients with diabetes who are younger than 65 years. As such, including these conditions as major symptoms would have negligibly changed our results. We agree that determining the effects of high-deductible insurance on major cardiovascular events is crucial. Analyses examining these effects will require sample sizes larger than those in our current study, but we believe that our findings will help reveal mechanisms underlying the results of such research.