Methodology masking true variations in Medicare spend

By Caroline Price, Senior medwireNews Reporter

Variations in regional mortality and spending rates in the US Medicare system are being blurred by inadequate risk adjustment methodology, researchers suggest.

Their study indicates that current estimates are open to "substantial bias" that skews the perception of the true underlying burden of illness, because they adjust for illness based only on recorded diagnoses and fail to incorporate the intensity of patient observation - here assessed by the frequency of physician visits.

"Adjusting without correction for regional variation in visit rates tends to make regions with high rates of visits seem to have lower mortality and lower costs, and vice versa," report John Wennberg (The Audrey and Theodor Geisel School of Medicine at Dartmouth, Hanover, New Hampshire) and team in the BMJ.

Indeed, the team found that after risk adjustment using Medicare's usual method, regions in the highest fifth for visits by physicians had a mean mortality rate that was about 20% lower than that for regions in the lowest fifth.

Correcting for the frequency of physicians visits in the year prior to a patient's death was more efficient than the standard method - but still accounted for less than 25% of geographic variation in age, gender, and race adjusted mortality among fee-for-service Medicare beneficiaries, the team reports.

Previous studies have already "raised the question of the extent to which variation in resource use and healthcare outcomes, after the conventional adjustments in the United States for age, sex, and race, are justified because of differences in risk," Wennberg and team explain. Here, the researchers examined the extent of bias introduced by intensity of patient observation based on the frequency of physician visits.

Analysis of over 5 million beneficiaries enrolled in parts A and B of Medicare, aged 65-99 years, in 2007 showed that the mean number of physician visits in the last 6 months of life among patients who died varied from 10 to 59. Whereas physician visits showed no correlation with age, gender, and race adjusted mortality, they correlated strongly with the number of diagnoses observed in claims data.

Depending on the index for comorbid conditions used, comorbidity explained around twice as much of the regional variation in age, gender, and race adjusted mortality when using diagnoses data adjusted for visits as when based on diagnoses alone (21-24% versus 10-12%).

Further analysis revealed that visit-corrected and age, gender, and race adjusted mortality rates were similar in hospital referral regions with the highest and lowest fifths of visits, which varied 2.4-fold in physician visit rates. However, after adjustment for comorbidity using the standard method, mortality was 18% lower in the highest fifth, at 46.4 deaths per 1000, than in the lowest, at 56.3 deaths per 1000.

"Our study points to the importance of developing risk adjustment methods that better explain variation in age, sex, and race mortality rates and suggests that these will be found by using data that are clearly independent of the effects of supply," conclude the authors.

Licensed from medwireNews with permission from Springer Healthcare Ltd. ©Springer Healthcare Ltd. All rights reserved. Neither of these parties endorse or recommend any commercial products, services, or equipment.