© 2003 American Thoracic Society
Estimated frequency of nursing facility–acquired pneumonia?To the Editor::I read with interest the recent article by Kaplan and colleagues regarding community-acquired pneumonia in the elderly (1). The article is a valuable contribution to the understanding of the epidemiology of this problem. However, I believe that the authors may have made a significant methodological error that may invalidate some of their results. They state that only 4.3% of the Medicare patients hospitalized with pneumonia came from a nursing facility. This contrasts with previous multihospital cohort studies in which patients from nursing facilities have comprised between 23 and 30% of all elderly Medicare patients admitted to the hospital with pneumonia (2, 3). Fine and colleagues found that nursing home–acquired pneumonia represents nearly 10% of all pneumonia hospitalizations (4). I believe that the authors' low reported prevalence of nursing facility–acquired pneumonia may be explained by their misinterpreting the "admission source" data field from the Medicare claims data. Unfortunately, this data field cannot be used to determine where a patient resided before entering the hospital, because the emergency department is counted as an admission source. Thus, the large number of nursing home patients who enter the hospital via the emergency department will not be coded as having come from a nursing home. Since most of the of nursing home patients in the authors' cohort were probably misidentified as coming from the community, the comparisons they make between nursing home and non-nursing home–acquired pneumonia are likely invalid. Furthermore, since nursing home residence was a strong predictor of mortality, the multivariate analyses results regarding outcomes may also be inaccurate.
University of Connecticut School of Medicine Farmington, Connecticut REFERENCES
From the Authors:Dr. Metersky brings up an excellent point in his letter regarding our manuscript (1). Space limitations precluded our discussion of this issue in the original paper but we are happy to address this issue now. Dr. Metersky is correct that the admission field is difficult to interpret because one of the choices is "emergency department." Thus, many patients who came from home, another hospital, or a nursing facility via the emergency department may be coded as being admitted from the "emergency department," rather than from their "true" source. However, those patients whose records do state they were admitted from a nursing home were admitted from a nursing home. The issue, therefore, as it pertains to our study is one of undercounting the total number of nursing home admissions—which exerts selective effects on our analyses. Faced with the opportunity to either completely ignore a limited but important piece of information or make the best use of what was available, we chose the latter.Specifically, although our cohort of nursing home admissions is likely an underestimate, we still found that this cohort was at higher risk of death and differed in other clinical characteristics from the remaining population. We stand by these findings and point out that, if anything, a more accurate count of nursing home admissions would only have further magnified these differences (because the nursing home patients currently counted in the "remainder" population obscure the magnitude of differences between the identified nursing home population and the remainder). In other words, we concur with Dr. Metersky's points regarding the coding accuracy of the admission source field and agree that our estimate of nursing home admissions is an underestimate. However, our findings that nursing home admissions differ in their clinical presentation and risk of death from other cases is still valid and, indeed, is an underestimate if anything.
a University of Pittsburgh Pittsburgh, Pennsylvania REFERENCES
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
