Regional Variations in Diagnostic Practices

Dartmouth Institute for Health Policy and Clinical Practice, Hanover, NH, USA.
New England Journal of Medicine (Impact Factor: 55.87). 06/2010; 363(1):45-53. DOI: 10.1056/NEJMsa0910881
Source: PubMed


Current methods of risk adjustment rely on diagnoses recorded in clinical and administrative records. Differences among providers in diagnostic practices could lead to bias.
We used Medicare claims data from 1999 through 2006 to measure trends in diagnostic practices for Medicare beneficiaries. Regions were grouped into five quintiles according to the intensity of hospital and physician services that beneficiaries in the region received. We compared trends with respect to diagnoses, laboratory testing, imaging, and the assignment of Hierarchical Condition Categories (HCCs) among beneficiaries who moved to regions with a higher or lower intensity of practice.
Beneficiaries within each quintile who moved during the study period to regions with a higher or lower intensity of practice had similar numbers of diagnoses and similar HCC risk scores (as derived from HCC coding algorithms) before their move. The number of diagnoses and the HCC measures increased as the cohort aged, but they increased to a greater extent among beneficiaries who moved to regions with a higher intensity of practice than among those who moved to regions with the same or lower intensity of practice. For example, among beneficiaries who lived initially in regions in the lowest quintile, there was a greater increase in the average number of diagnoses among those who moved to regions in a higher quintile than among those who moved to regions within the lowest quintile (increase of 100.8%; 95% confidence interval [CI], 89.6 to 112.1; vs. increase of 61.7%; 95% CI, 55.8 to 67.4). Moving to each higher quintile of intensity was associated with an additional 5.9% increase (95% CI, 5.2 to 6.7) in HCC scores, and results were similar with respect to laboratory testing and imaging.
Substantial differences in diagnostic practices that are unlikely to be related to patient characteristics are observed across U.S. regions. The use of clinical or claims-based diagnoses in risk adjustment may introduce important biases in comparative-effectiveness studies, public reporting, and payment reforms.

Download full-text


Available from: John E Wennberg
  • Source
    • "In a sensitivity analysis , length of stay was included in the logistic model because hospitals with shorter stays may be more likely to discharge patients alive ( Drye et al . 2012 ) . In another analysis , recent evidence on regional variation in diag - nostic intensity ( Song et al . 2010 ; Welch et al . 2011 ) was addressed by adding the average level of predicted mortality risk within a hospital referral region ( HRR ) to the model , and interacting predicted patient mortality risk with the HRR average ( Romley et al . 2013 ) . Next , we linked each discharge record to community - level characteristics from the 2000 U "

    Full-text · Dataset · Oct 2015
  • Source
    • "At the same time, it is comprehensive and can be assumed to be highly accurate. This variable may nevertheless not be independent of the behaviour of the service providers if diagnostic styles vary from region to region (Song et al., 2010 andWelch et al., 2011). Even in Germany, there are indications of regional differences in the coding quality (Ozegowski, 2013). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Health care expenditure in Germany shows clear regional differences. Such geographic variations are often seen as an indicator for inefficiency. With its homogeneous health care system, low co-payments and uniform prices, Germany is a particularly suited example to analyse regional variations. We use data for the year 2011 on expenditure, utilization of health services and state of health in Germany's statutory health insurance system. This data, which originate from a variety of administrative sources and cover about 90% of the population, are enriched with a wealth of socio-economic variables, data on pollutants, prices and individual preferences. State of health and demography explains 55% of the differences as measured by the standard deviation while all control variables account for a total of 72% of the differences at county level. With other measures of variation, we can account for an even greater proportion. A higher proportion of variation than usually supposed can thus be explained. Whilst this study cannot quantify inefficiencies, our results contradict the thesis that regional variations reflect inefficiency. Copyright © 2015 John Wiley & Sons, Ltd. Copyright © 2015 John Wiley & Sons, Ltd.
    Full-text · Article · May 2015 · Health Economics
  • Source
    • "If a risk-adjusted measure is used to track improvement , it should be graphed along with its components—observed and expected mortality . The expected mortality is affected not just by changes in case mix, but also by coding (Jacques, Fotheringham, Campbell, & Nicholl, 2013; Song et al., 2010; Welch, Sharp, Gottlieb , Skinner, & Wennberg, 2011; Wennberg et al., 2013). Changes in coding over time alter expected mortality and produce a reduction in adjusted mortality that may not reflect improvement in care. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Risk-adjusted hospital-wide mortality has been proposed as a key indicator of system-level quality. Several risk-adjusted measures are available, and one—the hospital standardized mortality ratio (HSMR)—is publicly reported in a number of countries, but not in the United States. This paper reviews potential uses of such measures. We conclude that available methods are not suitable for interhospital comparisons or rankings and should not be used for pay-for-performance or value-based purchasing/payment. Hospital-wide mortality is a relatively imprecise, crude measure of quality, but disaggregation into condition- and service-line–specific mortality can facilitate targeted improvement efforts. If tracked over time, both observed and expected mortality rates should be monitored to ensure that apparent improvement is not due to increasing expected mortality, which could reflect changes in case mix or coding. Risk-adjusted mortality can be used as an initial signal that a hospital's mortality rate is significantly higher than statistically expected, prompting further inquiry.
    Full-text · Article · Aug 2014 · Journal for Healthcare Quality
Show more