Working Paper: NBER ID: w26363
Authors: Mattan Alalouf; Sarah Miller; Laura R. Wherry
Abstract: Over the past 30 years, the criteria used to diagnose many illnesses have been relaxed, resulting in millions more relatively healthy individuals receiving treatment. This paper explores the impact of receiving a diagnosis of a common disease among such “marginally ill” patients. We apply a regression discontinuity design to the cutoff in blood sugar levels used to classify patients as having diabetes. We find that a marginally diagnosed patient with diabetes spends $1,097 more on drugs and diabetes-related care annually after diagnosis, but find no corresponding changes in self-reported health or healthy behaviors. These increases in spending persist over the 6-year period we observe the patients. These marginally diagnosed patients experience improved blood sugar after the first year of diagnosis, but this improvement does not persist in subsequent years. Other clinical measures of health, such as BMI, blood pressure, cholesterol, and mortality show no improvement. The diagnosis rates for preventable disease-related conditions such as diabetic retinopathy, neuropathy, and kidney disease increase following a diagnosis, likely due to more intensive screening. Our results imply that a small relaxation in the diagnosis cutoff would increase total spending on diabetes-related care by about $2.4 billion annually and minimally impact patient health.
Keywords: Diabetes; Diagnosis; Health Care Spending; Regression Discontinuity
JEL Codes: I11; I12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| A1C levels just above 6.5% (I10) | diabetes diagnosis on health insurance claim (I13) |
| diabetes diagnosis (C52) | increase in spending on diabetes-related care (H51) |
| diabetes diagnosis (C52) | increase in spending on endocrinology specialty care (H51) |
| diabetes diagnosis (C52) | increase in spending on office-based care and laboratory tests (I11) |
| diabetes diagnosis (C52) | initial improvement in blood sugar levels (Y60) |
| initial improvement in blood sugar levels (Y60) | improvement does not persist beyond the first year (C41) |
| diabetes diagnosis (C52) | increased rates of complications related to diabetes (I12) |
| increased rates of complications related to diabetes (I12) | complications include diabetic retinopathy, neuropathy, and kidney disease (I12) |
| diabetes diagnosis (C52) | no evidence of improvement in self-reported health or health behaviors (I12) |
| relaxation of diagnosis cutoff (C22) | increase in total spending on diabetes-related care (H51) |
| relaxation of diagnosis cutoff (C22) | minimal impact on patient health outcomes (I14) |