State practice variation in the use of tube feeding for nursing home residents with severe cognitive impairment.

OBJECTIVE To describe the differences in prevalence of tube feeding among states and to examine possible factors that could explain practice patterns. DESIGN Analysis of random samples from an interstate data bank comprised of the Minimum Data Set (MDS), a standardized, federally mandated assessment instrument for nursing home residents. SETTING Nursing homes in four states participating in a federal demonstration project of case mix payment plus five others with existing MDS data systems. PARTICIPANTS Individuals 65 years of age and older (N = 57,029), who had very severe cognitive impairment, including total dependence in eating, and who resided in nursing homes during 1994, the most recent year for which uniform data were available. MEASUREMENTS State-by-state differences in prevalence of tube feeding, controlling for demographic and clinical variables. RESULTS The prevalence of tube feeding ranged from 7.5% in Maine to 40.1% in Mississippi. Each state had a significantly elevated prevalence of tube feeding compared with Maine, with odds ratios (ORs) ranging from 1.50 to 5.83, P < .001. Specific directives not to provide tube feeding (OR 0.41, P < .001), and white race (OR 0.45, P < .001) were strongly and negatively associated with tube feeding. CONCLUSIONS Wide regional variations exist in the use of tube feeding of nursing home residents with equivalent impairments. Sociodemographic factors could be important, but more study is needed to determine whether physician characteristics, such as race, attitudes, or knowledge, have an impact and to clarify medical standards for the use of tube feeding in this population.