Claims Data May Not Be Enough to Determine Nonpayment

Larry Hand

September 12, 2012

September 12, 2012 — Although epidemiological studies show that 59% to 86% of hospital-acquired urinary tract infections (UTIs) are catheter-associated, only 2.6% of all hospital-acquired UTIs were described as catheter-associated in claims data from Michigan hospitals in 2009, according to an article published in the September 4 issue of the Annals of Internal Medicine. In addition, nonpayment for hospital-acquired catheter-associated UTIs (CAUTIs) applied to just 0.003% (25/781,343) of hospitalizations.

In 2008, the Centers for Medicare & Medicaid Services (CMS) stopped paying hospitals for specific complications that are deemed "reasonably preventable" as a way to encourage hospitals to reduce Medicaid costs and improve patient safety. CAUTI was the first hospital-acquired diagnosis that triggered nonpayment. The policy was later adopted by other payers, including Blue Cross and Blue Shield of Michigan, whose foundation funded this study.

To assess the effect of this policy, Jennifer A. Meddings, MD, from the Department of Medicine, University of Michigan Health System, Ann Arbor, and colleagues analyzed the claims records of 96 nonfederal acute care Michigan hospitals for the years 2007 and 2009. The study population included adults older than 18 years whose hospital stays lasted 2 days or longer. It excluded obstetric patients, but did include all payers. The researchers excluded hospitals not affected by the CMS Hospital Acquired Conditions Initiative.

Dr. Meddings and colleagues compared CAUTIs and UTIs not associated with catheters (non-CAUTI) for both years to assess rates before and after the CMS designation. For 2009, they also compared rates for hospital-acquired and present-on-admission conditions and assessed the effect of nonpayments.

Overall, they reviewed 767,531 discharges for 2007 and 781,343 for 2009. The proportion of discharges in 2007 for a secondary diagnosis of non-CAUTI ranged from 5.2% to 17.1% (mean, 10.0%; 95% confidence interval [CI], 9.5% - 10.5%). For 2009, the proportion ranged from 5.0% to 20.2% (mean, 10.3%; 95% CI, 9.8% - 10.9%]). The proportion of discharges in 2007 for secondary-diagnosis CAUTI ranged from 0% to 1.10% (mean, 0.09%; 95% CI, 0.06% - 0.12%). That proportion for 2009 ranged from 0% to 0.95% (mean, 0.14%; 95% CI, 0.11% to 0.17%).

In 2009, the proportion of discharges that included hospital-acquired CAUTIs in claims data was 0.03% (95% CI, 0.02% - 0.04%). Moreover, of the 96 hospitals in the study, 45 (47%) coded zero Medicare hospitalizations including a diagnosis of hospital-acquired CAUTI. "Of all hospital-acquired UTIs, few (mean, 2.6% [CI, 1.6% to 3.6%]) were described as CAUTIs," the researchers write.

Nonpayment applied to just 25 (0.003%) of the 781,343 discharges in 2009. The researchers estimated that those 25 discharges, if they had been from the University of Michigan, would have reduced Medicare payments by $132,675, or 0.06% of total payments from the Acute Inpatient Prospective Payment System. This estimate was based on the average base Medicare payment of $8700 per hospital admission that was in place at the University of Michigan in 2009. Payments to community hospitals might be at lower bases, the researchers write, and in addition, these results might not be generalizable to other states.

"[W]e showed that the current hospital discharge data set rarely identifies CAUTIs," the researchers write. "The effect of a nonpayment policy based on these data is small. The accuracy of reporting from the data set is suspect. Moreover, we conclude that the current hospital discharge data set is not accurate or valid for comparing hospital-acquired CAUTI rates for the purpose of public reporting or imposing financial incentives or penalties."

In an accompanying editorial, Bernard Rosof, MD, from the North Shore–Long Island Jewish Health System, Huntington, New York, writes, "First and foremost, we must use superior data elements derived from electronic health records and registries and share that data system-wide to begin the process of ensuring data credibility.

"Second, institutions should share best practices to assist, encourage, and promote improvement," he continues.

"Last," Dr. Rosof writes, "there must be a cultural shift in all organizations to patient-centered care and patient safety."

He concludes, "Ensuring the accuracy of the data on which we base our efforts is among the first challenges we must meet."

The study was funded by a grant from the Blue Cross Blue Shield of Michigan Foundation. Other author support included grants from the Agency for Healthcare Research and Quality, the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute on Nursing Research, and a loan repayment program from the National Institutes of Health. One coauthor has reported receiving honoraria from the Veterans Health Administration, Institute for Healthcare Improvement. The authors and editorialist have disclosed no other relevant financial relationships.

Ann Intern Med. 2012;157:305-312, 379-380. Article abstract, Editorial extract