An action taken to protect patient privacy – removing claims related to substance abuse from Medicare and Medicaid databases – inadvertently caused an immediate and marked suppression of vital data pertaining to related disorders such as HIV, depression, anxiety, and hepatitis C, according to data published online March 15 in JAMA.
©Katarzyna Bialasiewicz/thinkstockphotos.com
In 2007, the Centers for Medicare & Medicaid Services implemented a federal regulation that prohibits third-party payers from releasing information from federally funded substance abuse treatment programs. To comply, the CMS had to change its longstanding practice of making all claims data available to researchers, instead removing from its database all claims containing a diagnostic or procedure code related to substance abuse, said Kathryn Rough of the division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital, Boston, and her associates.
To gauge the effect this had on the information available to researchers, the investigators compared de-identified Medicaid data from before the policy change (2000-2006) against that from afterward (2007-2010). They focused on diagnoses for six disorders that frequently occur in tandem with substance abuse (HIV, tobacco use, depression, anxiety, hepatitis C, and cirrhosis) and four that are unrelated to substance abuse (type 2 diabetes, stroke, hypertension, and kidney disease).
The study period included 63 million inpatient and 13.6 billion outpatient claims. Compared with data available to researchers before the policy change, afterward there was an immediate and substantial reduction. Inpatient diagnosis rates declined 24% for HIV, 51% for tobacco use, 38% for depression, 27% for anxiety, 57% for hepatitis C, and 49% for cirrhosis. Declines in outpatient diagnosis rates were less marked and reached statistical significance only for anxiety, which dropped 6.3%.
In contrast, diagnosis rates for disorders unrelated to substance abuse did not change appreciably after the new policy was implemented, Ms. Rough and her associates said (JAMA. 2016;315:1164-6). “Underestimation of diagnoses has the potential to bias health services research studies and epidemiological analyses” and could lead to “spurious conclusions.” For example, “a hospital that regularly admits substance abusers will [show] artificially low rates of readmission, giving a false appearance of better performance,” they noted.