Are studies of geographical variation in medical practice useful in guiding decisions on disinvestment? Some pros and cons

¿Sirven los estudios de variabilidad geográfica de la práctica para informar la desinversión? Varias cautelas y algunas reflexiones Enrique Bernal-Delgado, Sandra García-Armesto, Carlos Campillo-Artero Gaceta Sanitaria . 01/2013; 27(1):7–11. DOI:10.1016/j.gaceta.2012.02.004

Disinvestment has been defined as the explicit process of cutting funding, either totally or partially, from health technologies deemed of low-value. Studies of geographic variation in medical practice have been suggested to be useful in guiding decisions on disinvestment, as they may identify unwarranted variations in procedure-rates at the population level. This study aimed to determine the utility of these studies.Methods
We performed an ecologic study of variations in standardized rates in four «low-value» interventions: proctologic surgery, arthroplasty revision, incisional hernia repair and tonsillectomy. Variation across 199 healthcare areas within the Spanish national health system between 2002 and 2007 was studied by using the extremal quotient (EQ), the empirical Bayes statistic (EB) and the standardized utilization ratio (SUR).ResultsA total of 168,363 proctologic interventions, 41,066 arthroplasty revisions, 222,427 incisional hernia repairs, and 72,724 tonsillectomies were studied. The EQ ranged from a 3-fold variation in proctologic surgery to a 6.5-fold variation in tonsillectomy. The EB figures varied from moderate to high systematic variation: 0.12 in hernia repair and proctology, 0.20 in arthroplasty revision, and 0.30 in tonsillectomy. Twenty-five percent of the healthcare areas showed SUR figures above 1.24 in proctologic interventions, 1.25 in arthroplasty revision, 1.32 in hernia repair and 1.35 in tonsillectomy.Conclusions
The interventions studied showed moderate to high systematic variation, supporting the usefulness of variation studies in guiding disinvestment policies. Nevertheless, caution should be exercised when evaluating interventions with an uncertain risk-benefit ratio.

Impact factor: 
1,12