The Society of Teachers of Family Medicine Presents its 2003 STFM Best Research Paper Award to David Mehr, MD, MS

At its annual conference in September 2003, the Society of Teachers of Family Medicine presented its 2003 Best Research Paper award to David R. Mehr, MD, MS, and his colleagues: Ellen Binder, MD; Robin Kruse, PhD; Steven Zweig, MD, MSPH; Richard Madsen, PhD; Lori Popejoy, MSN, RN; and Ralph D’Agostino, PhD. The paper, “Predicting Mortality in Nursing Home Residents with Lower Respiratory Tract Infection: The Missouri LRI Study,” was published in the Journal of the American Medical Association in November 2001.¹ Dr. Mehr, an associate professor at the University of Missouri-Columbia, presented the paper at the Research Award Winners session at the annual conference.

The Missouri LRI Study identified a new predictive model for 30-day mortality risk among nursing home residents with lower respiratory tract infections (LRIs). The 8 variables in the model are absolute lymphocyte count, level of independence in activities of daily living, body mass index, presence of mood deterioration within the previous 90 days, pulse, serum urea nitrogen level, sex, and white blood cell count. These variables, when assigned a point value and summed, accurately stratified nursing home residents with LRIs into quintiles of mortality risk. In the study’s validation sample, 30-day mortality ranged from 1.8% in the low-risk quintile of patients (with scores of 4 or lower) to 54.2% in the very high risk quintile (with scores of 11 or higher). If the rule is validated outside Missouri, it will be a particularly valuable tool for identifying relatively low risk nursing home residents who might not need hospitalization.

As with many important research findings, the motivation for the study arose from questions in clinical practice. Dr. Mehr practiced family medicine and geriatrics in a variety of community and nursing home settings through the 1980s, and he wondered about the best way to treat pneumonia in nursing home residents. “There was a major disconnect between the literature and clinical practice,” Dr. Mehr said. “Experts recommended hospitalizing all nursing home residents with pneumonia and treating them with parenteral antibiotics. Actual practice commonly included treatment in the nursing home with oral antibiotics.” As he studied the area in more depth as a geriatric medicine fellow in the late 1980s, he found that most studies were retrospective or not focused on the nursing home, whereas pneumonia and other LRIs are the leading cause of mortality. His time as a full-time clinician was invaluable in developing the Missouri LRI Study. “I had a clear focus on staying practical,” Dr. Mehr said. For example, the research team decided to expand the study beyond pneumonia to LRIs in general, because the clinical distinction among pneumonia, bronchitis, and tracheobronchitis often difficult.

Although the design focused on practicality, Dr. Mehr still reports he had many challenges in the study’s development. Because approximately two thirds of potential subjects would be cognitively impaired, informed consent was a potential barrier, “and those were the folks we most wanted to study,” Dr. Mehr added. He worked with his medical center’s institutional review board to develop a clinical protocol that was consistent with the standard of care, thus not requiring written consent of the patients. The challenge then became recruitment of nursing homes and physicians who would agree to the protocol. In the end, 36 nursing homes in central Missouri and St Louis became involved, and the project identified 1,044 individual residents with 1,406 LRI episodes.

The analysis posed its own challenges and surprises. The initial list of variables was huge, encompassing 25 categories of factors. A process of bivariable and multivariable analysis narrowed the model to the 8 specific criteria in the final model. “Creating a useful clinical model turned out to be a bigger challenge than I ever anticipated,” said Dr. Mehr. “I thought I knew a lot about modeling at the start, but I didn’t know anything!” He and his team were eventually able to identify the areas of overlap and eliminate some variables. Cognitive status, for example, is not included in the final model. “It unquestionably relates to ADL status, which is in the final model,” observed Dr. Mehr. He also said the importance of mood deterioration in the previous 90 days (as reported by nursing home staff) was surprising to the research team and might be a broader indicator of general decline. The practical decision to include other LRIs in addition to pneumonia led to another somewhat surprising finding—that the presence of pneumonia on a chest radiograph was not a significant predictor of mortality.

While individual clinical decisions involving LRIs in the nursing home depend on many factors, Dr. Mehr feels this study adds another tool that health care providers, their patients, and families can use to make a more informed decision. “Understanding what indicates high risk is important, and this helps us recognize that treatment in the nursing home with oral antibiotics is acceptable and reasonable in most cases,” which, he said, is more in line with what happens in other countries. His interest in international variations in nursing home LRI care led to a productive relationship with a group of Dutch physicians and researchers. He eventually spent a year in The Netherlands learning their approach to LRIs and advanced dementia. “The international collaboration that flowed out of this was hugely illuminating in terms of end-of-life care issues,” he said. “This whole experience reinforced how crucial it is to decide on goals of care ahead of time.” If patient and family preferences regarding hospitalization and aggressiveness of care are known before an LRI episode occurs, Dr. Mehr believes that the study’s prediction rule would make treatment decisions more straightforward. As a future research direction, he is interested in following up on another part of this study that identified communication barriers between nursing home staff and physicians; these barriers can account for unnecessary and often hazardous hospitalization. “Ultimately, we could probably treat at least three quarters of these cases in the nursing home,” he said.

Erik J. Lindbloom, MD, MSPH

University of Missouri-Columbia

Chair, STFM Research Committee

• © 2003 Annals of Family Medicine, Inc.