Association for Public Policy Analysis & Management

Background: Hypertension, diabetes and high cholesterol are top contributors to premature death and ill-health in the United States, and lengthy gaps in diagnosis and treatment can lead to negative health consequences. However, about one-fifth of cases are undiagnosed nationally. Increasing diagnosis rates is a national health system priority, but it is not clear which newly screened patients will pursue care for screen-detected cases diabetes, hypertension or high cholesterol. 

Objective: To examine whether health screenings increase physician management of previously undetected cases of hyperlipidemia, hypertension and diabetes for a variety of patients, exploiting a randomized screening intervention with merged claims data and national recruitment.

Research Design: Using a commercial list of residential phone numbers, the REasons for Geographic And Racial Differences in Stroke (REGARDS) study offered free health screenings at random times to non-institutionalized, English speaking adults across the continental United States between 2003-2007. African Americans and individuals living in the stroke belt were over-sampled. Participants were asked about previously diagnosed conditions, and then had their fasting blood glucose, blood pressure, and lipid panel assessed in their home and were notified of their results. It is because of these notifications that REGARDS can be considered a screening intervention in addition to an epidemiological study.

We assessed the impact of screening on physician evaluation and management of hyperlipidemia, hypertension and diabetes cases using an interrupted time series regression and Medicare claims data from the 24 months before and 24 months after each individual was screened. We use the Chronic Conditions Warehouse classifications to identify whether each condition was evaluated and managed during a physician visit each six months. We could track this outcome over the four years of interest for 10,079 cases of hyperlipidemia, hypertension or diabetes (6,571 participants). We allowed the impact of screening to vary across previously undiagnosed and diagnosed conditions, and exploited the availability of biomarkers and random roll-out of the intervention to control for health, year, and aging effects.

Results: 15% of cases of hyperlipidemia, hypertension or diabetes were previously undetected in our sample. Physician management of screen-detected hypertension, diabetes and high cholesterol increased by 17 percentage points on average over the 24 months after screening. The largest impacts were shown among women, African Americans, individuals who reported low income and low health, and individuals who reported higher levels of stress prior to screening. The smallest impacts were shown among individuals with less than high school education or low scores on a cognitive screening test. Physician management of previously diagnosed conditions did not change as a result of screening.

Conclusion: Screen-detection of conditions increases relevant care for most patients, including many high-priority groups. However, many screen-detected cases remained unmanaged by a physician due to the patient’s preferences, price, lack of understanding of the screening results, or other factors.