The Dartmouth Atlas Reports on geographic variation in the cost and quality of health care have sparked a long-running controversy over the actual existence of such variation for various types of insurance and if it does exist, what the causes might be. The latest report in the series looks at variation in pediatric care. (Dartmouth Report) The report’s value is limited by analyzing only data from Vermont, New Hampshire and Maine, which are hardly representative of the entire United States and represent only a small fraction of the health care delivered in the country. But it does give some indication of what might be expected to be found if a similar analysis were done for the entire country. The authors analyze variation from their now familiar categories of effective care, preference-sensitive care and supply-sensitive care. They find that all three factors appear to be involved in geographic variation in pediatric care. Using data from 2007 to 2010 in the Medicaid and all-payer claims databases from the three states, they found a wide variation in annual office visit rates, from 1.5 in some areas to 3.5 in others. At least part of this variation appears to be due to the number and location of physician offices in relation to the child. A similar variation is found in emergency room visits, with a similar higher rate of use when ERs are nearer to the patient. Socioeconomic factors and source of insurance do not appear to explain much of the variation. There does appear to be a strong substitution effect between office visit rates and ER use, with higher ER use where there are lower rates of office visits, suggesting that ERs are substituting for effective primary care. There was significant variation in performance on common pediatric care quality measures, in some cases as much as twofold, but overall, performance was fairly good. Rates of hospital use also varied extensively, which is somewhat surprising because there are few indications for pediatric hospitalization and the guidelines are fairly well agreed to and disseminated. The use of ear tubes to control ear infections varied by four times across the area, but this is an area where this is not clear consensus about appropriate use. Tonsillectomies showed a two-fold variation, and here there likely is a high rate of inappropriate use, since research no suggests little benefit in most cases. Diagnostic imaging again displayed large variation, and concerns about children’s exposure to radiation should counsel for conservatism in this area. Finally, drug use was highly variable, particularly for psychiatric medications, which are widely perceived to be overused in children. The overall picture is one of high variation, with at least a suggestion that in some high-utilizing areas there may be a fair amount of inappropriate care delivery for children.
