The impact of physical therapy on nursing home patient outcomes.

Key Words: Geriatrics, Long-term care, Nursing homes, Physical Therapy.

In many geriatric care settings, multidisciplinary care has been shown to be of benefit.(1-4) Recently, debate has centered on whether an individual intervention or the sum of all the multidisciplinary interventions is responsible for the benefits seen in function, cognition, and mortality. Previous research(5) has shown that discharge to nursing homes that include physical therapy as part of multidisciplinary care results in a higher likelihood of discharge to the community and of independent ambulation. Other than reports of beneficial outcomes of multidisciplinary geriatric interventions, there is little published information about physical therapy in the nursing home setting. Review of the English-language literature published since the 1960s yielded reports describing the role of physical therapists in nursing homes.(6,7) These reports, however, did not address the intensity or outcome of physical therapy administered to patients in nursing home settings. Because the provision of physical therapy to patients in nursing homes remains a controversial topic about which little data exist in the literature to date, this retrospective study was undertaken specifically to determine the contribution of physical therapy and other components of multidisciplinary care to successful outcome of nursing home care. The project had the following objectives: (1) identify individual components of interdisciplinary care within the nursing home and explore their effectiveness, (2) determine the extent of physical therapy offered in a teaching nursing home, (3) determine the efficacy of physical therapy and other program components in the study population, and (4) identify patient characteristics associated with receipt of physical therapy and with improvement in physical function. The information obtained will be useful in determining whether specific components, particularly physical therapy, merit clinical trials in the nursing home setting.

Method

Design and Site

This study was a retrospective review of consecutive admissions to the Extended Care Therapy Center from February through June 1988. The Extended Care Therapy Center is a 120bed teaching nursing home, which is part of the Audie L Murphy Memorial Veterans Affairs Medical Center in San Antonio, Tex.

Data Collection Of 94 patient admissions, 90 charts (96%) were available for review. Sources reviewed include medical records, physical therapy records, and the admission Patient Assessment Inventory (PAI), a case-mix measurement used for reimbursement.(8) The PAI contains observer-reported ratings of a patient's eating, toileting, transfer, and mobility skills. Two independent reviewers abstracted data from the first 4 months of each patient's stay using a standard form. The reviewers were physicians who are faculty geriatricians. Information abstracted included demographic data, medical diagnoses, number of medications used, cognitive function as indicated by admission Folstein Mini-Mental State examination score,(9) and activities-of-daily-living status. Each reviewer characterized the type and amount of care received by the patients according to the following program components: (1) physical therapy (ie, general conditioning exercises and functional skills training administered by a physical therapist); (2) speech therapy; (3) psychosocial therapy (ie, evaluation and management by a psychologist, psychiatrist, or social worker); (4) medication adjustment; and (5) other medical and nursing care. General conditioning training included activities directed toward achieving the strength, range of motion, motor control, and sensory integration that are prerequisite for successful performance of activities of daily living and achievement of independent mobility. Functional activity training complemented the general conditioning training and included teaching skills and adaptive techniques required for independence in mobility and activities of daily living. Both the general conditioning training and the functional skills training were administered by a physical therapist. There was no occupational therapist on staff during the majority of the study period.

The intensity of therapy received in each program area was rated as high, moderate, or minimal to none. For physical therapy and speech therapy, the intensity category was rated by number of sessions per week and duration of therapy. High-intensity therapy was defined as three or more sessions per week for at least 8 weeks. Moderate-intensity therapy consisted of one or two sessions per week for at least 8 weeks or three to five sessions per week for fewer than 4 weeks. If fewer than one session per week was received or duration was less than 4 weeks, intensity was rated as minimal to none. All psychosocial therapy and other medical and nursing care intensity ratings were based on the frequency and duration of the interventions received. For example, interventions requiring at least two visits or consultations per week for at least 4 weeks were ranked as high-intensity therapies. Those interventions requiring more than one visit per month but less than two visits per week were considered moderate-intensity therapies, and those requiring less than one visit per month were rated as minimal-intensity therapies. Medication adjustment intensity was based on the number of medication changes made. Scheduled prescription medications were defined as major medications, and non-prescription medications (eg, acetaminophen, psyllium) were defined as minor medications. Changes in five minor medications or three major medications were rated as high-intensity changes, changes in two to four minor medications or one or two major medications were rated as moderate-intensity changes, and changes in fewer than two minor medications were rated as minimal-intensity or no changes.

Outcome Assessment

Outcomes were rated as improved (beneficial), unchanged, or worsened (detrimental) at 4 months post-admission. Outcome in each program area was ascertained by reviewing progress notes of the appropriate therapist as well as nursing and medical progress notes reflecting performance in the specific area. if the patient was discharged or died before 4 months postadmission, outcome at the time of discharge or death was used. (Agreement was reached on 85% of initial ratings of therapy intensity and outcome; all disagreements were adjudicated by discussion among the two raters until consensus was achieved.)

The physical therapy outcome rating was based on change in strength, endurance, or activities of daily living, as documented by the physical therapist, nurse, or physician. The speech therapy outcome rating was based on documented change in speech or swallowing.

Beneficial changes in medication adjustment included institution of medications that were indicated, discontinuation of medications that were not indicated, and documented benefits from medication changes. If no changes in medications occurred, outcome was rated as unchanged. Detrimental changes included institution of medications that were not indicated, discontinuation of medications that were indicated, and documented adverse effects from medications or medication changes.

The psychosocial therapy outcome rating was based on change in cognitive function, behavior, symptoms of depression or psychosis, or resolution of social barriers that facilitated discharge. The outcome for other medical and nursing care was rated on the basis of improvement or deterioration in the patient's medical condition and documented benefit or deterioration as a result of the evaluations or procedures performed.

Data Analysis

Descriptive statistics were used to describe patient characteristics, type of care received, and outcomes of care. In the analyses, outcomes were dichotomized as improved or not improved for each program area. All patients' data were included in the analysis. Patients receiving no therapy and those receiving minimal-intensity therapy were combined to form one group in the analyses. We intended to explore the following hypotheses: (1) The type and intensity of therapy would be associated with outcome; (2) the independent variables of age, cognitive function, baseline activities-of-daily-living status, and number of medical problems would correlate with outcome; and (3) patient characteristics (ie, age, Mini-Mental State examination score, and diagnosis) may be associated with type and intensity of therapy received. Age, cognitive function score, baseline activities-of-daily-living status, number of diagnoses, and therapy intensity were considered in univariate analyses in the groups improving and not improving in physical therapy and medication adjustment. Correlations were reported as odds ratios. Odds ratios are used in retrospective studies to measure strength of association. They consist of the odds of exposure to a factor (in this study, physical therapy) if a condition (in this study, improved physical function) is present divided by the odds of exposure if the condition is not present. Those factors found to have a significant association with outcome were then examined by stepwise multiple logistic regression analysis.

Results

Table 1 presents the demographic and medical characteristics of the patients in this study. The majority of the patients were elderly, unmarried, white men. Fifty percent of the patients were admitted after hospitalization, 24% were admitted from home, and 21% were admitted from community nursing homes. The patients' mean age was 69 years. Thirty-eight percent of the patients had a stroke, and 25% were demented. Serious medical conditions and functional dependencies were very frequent in this sample. During the 4-month observation period, 14% of patients died, 15.5% were discharged home, and 4% were discharged to another nursing home.

The proportion of patients receiving moderate- and high-intensity therapy in each program area is demonstrated in Figure 1. Physical therapy and medication adjustment were the most commonly received treatments. Thirty-seven percent of the patients received high-intensity physical therapy, and 29% received moderateintensity physical therapy. High- and moderate-intensity medication adjustment were received by 31% and 32% of the patients, respectively. Thirteen

Table 1. Baseline Patient
Characteristics
Characteristic
Mean age (y) ([+ or -]SD)          69.0 [+ or -]13.7
White (%)                          71
Mexican American                   21
Black and other  (%)                8
Married (%)                        48
Number of diagnoses                 5.7[+ or -]2.5
Stroke  (%)                         38
Dementia  (%)                       25
Terminal illness  (%)               10
Hip fracture  (%)                   13
Depression    (%)                   22
Number of scheduled
  medications                        5.0[+ or -]2.6
Urinary incontinence  (%)           50
Dependency in transfers  (%)        80
Dependency in feeding  (%)          45

percent and 37% of the patients received high and moderate intensities, respectively, of other medical and nursing care. Psychosocial therapy and speech therapy were received by fewer than 16% of patients.

Figure 2 shows therapy outcome in each program area. For physical therapy, 88%, 33%, and 13% of those patients who received high-, moderate-, and minimal-intensity therapy, respectively, improved. For medication adjustment, 93% of the high-intensity treatment group, 72% of the moderate-intensity treatment group, and 45% of the minimal-intensity treatment group improved. The mean number of scheduled medications at admission was 5.0 (SD=2.6). At 4 months postadmission, the mean number of scheduled medications was 3.6 (SD=1.8) (P=.0001 for the difference in number of medications at baseline versus at 4 months). Because physical therapy and medication adjustment were successful and were received by adequate numbers of patients to permit analysis, factors predictive of success were explored.

Previous reports in the literature(10,11) suggest that the outcome of nursing home stays is related to cognitive function and activities-of-daily-living status. Age, cognitive score, activities-of-daily-living score, number of diagnoses, and therapy intensity were compared in patients who improved and those who did not improve in physical function. Univariate analyses revealed significant associations between age and physical therapy outcome (odds ratio [OR]=0.96, P=.02) and between physical therapy intensity and outcome (OR=5.98, P[is not greater than].0001) (Tab. 2), with both younger age and higher intensity associated with improved outcome.

In an effort to assess possible selection biases, univariate analyses comparing age, cognitive function, and baseline activities-of-daily-living status in patients in each physical therapy intensity group were performed. Older age was associated with lower physical therapy intensity (P=.02). Activities-of-daily-living status and cognitive function were not associated with intensity of therapy. Stepwise multiple logistic regression analysis for the factors age, age-treatment intensity interaction, and physical therapy intensity revealed that only physical therapy intensity was significantly correlated with outcome (P[is not greater than].0001). There were no significant differences in mental status, activities-of-daily-living status, or number of diagnoses between those patients with improvement and those without improvement. A similar procedure was used for outcome of medication adjustment. Univariate analyses for the variables age, cognitive function, activities-of-daily-living status, number of diagnoses, and intensity of medication adjustment revealed that only therapy intensity was associated with improved outcome.

Discussion

This retrospective study was one of the first attempts to describe the contribution of physical therapy to successful outcome of nursing home care. Physical therapy was frequently used and was beneficial to the majority of patients who received it. Cognitive impairment, very advanced age, and severe functional dependency were perceived as decreasing the likelihood of benefit from physical therapy in this study. Surprisingly, even patients with moderate to severe cognitive and functional impairments improved in physical function. Age had no independent effect on physical therapy outcome. Medication adjustment was a frequent and effective intervention. In the majority of patients, the adjustments consisted of elimination of unnecessary or toxic medications. Our definition of improvement included reduction in medications with no deterioration in patient status as well as changes associated with documented improvement. The increased cost and the increased risk of adverse drug reactions with increasing numbers of medications justify considering medication reduction a desirable outcome.(12)

This study had a number of limitations. In any nonrandomized study, selection bias is a serious concern. Subtle selection criteria may have been used to select patients for physical therapy who were more likely to improve. Prognosis could have affected the intensity of therapy. Prognostic indicators, including baseline activities-of-daily-living status, cognitive function, and number of diagnoses, however, did not differ between those patients who improved and those who did not improve in physical function. The strength of the association between physical therapy intensity and outcome makes the association seem unlikely to be due to selection bias alone and is a very strong recommendation for physical therapy. Older patients were less likely to have improved physical function and less likely to have received high-intensity physical therapy. The multivariate analysis, however, revealed that only physical therapy intensity, and not age, was significantly correlated with outcome. Therefore, though selection bias cannot be definitively excluded, we have examined reasonable proxy measures for prognosis and have not found evidence that selection bias accounted for the substantial improvement associated with receiving physical therapy or the dose response effect between therapy intensity and outcome. There may have been biases in this study that we were unable to detect. In particular, positive relationships between therapy intensity and favorable outcomes could be due to recording bias on the part of the health care providers. The physical therapist and other care providers were not aware of the purpose of this study. Although reviewers were aware of the study objectives, documented change in physical function (for example, ambulation distance) was required to assign outcome category. Though we used standard definitions in determining outcomes, prospective use of independent assessments of therapy intensity and outcome would reduce the potential for abstractor bias and error. Several problems may be attributable to the use of chart diagnoses and data. Their completeness and accuracy may have been imperfect. Important components of therapy that may have been effective, especially characteristics of nursing care and psychological support, counseling, and exercise or activities-of-daily-living training provided by nursing staff, cannot be adequately assessed by traditional medical chart review. Thus, the importance of other medical and nursing care may be underestimated.

Finally, the fact that this study was performed in a Veterans Administration facility limits its generalizability. The Veterans Administration nursing home population is almost all male and is younger than the community nursing home population. Functional status in this study, however, measured as independence in feeding, transfers, and continence, was as severely impaired as that reported for patients in the National Nursing Home Survey.(13) Before attempting to generalize the results of this study to all patients in nursing homes, the efficacy of physical therapy should be studied in a community nursing home sample.

Summary and Conclusions

This study represents one of the first attempts to specifically characterize the relative benefits of individual therapies in a multidisciplinary program within a nursing home. We have found evidence that suggests that physical therapy is effective when given at least three times per week, even for patients with advanced age, marked functional dependency, and cognitive impairment. Rather than identifying groups of patients who were unlikely to benefit from physical rehabilitation, we found that a wider range of patients than expected experienced favorable outcomes. Because of limitations previously mentioned, these results must be considered preliminary. Important future directions should include prospective randomized trials to evaluate the benefits of physical therapy in patients in nursing home settings. Outcome measures that can validly assess physical and functional status changes should be used to ensure that the improvements are clinically significant. Finally, patients with a broad range of disabilities should be included in these trials to confirm that physical therapy is beneficial even for moderately to severely impaired individuals.

References

1 Schuman JE, Beattie EJ. Rehabilitative and geriatric teaching programs: clinical efficacy in a skilled nursing facility. Arch Phys Med Rebabil. 1980;7:310-315.

2 Wieland D, Rubenstein LZ. Organizing an academic nursing home: impacts on the institutionalized elderly.JAMA 1980;255: 2622-2627.

3 Jahnigen D, Kramer AM, Robbins LJ, et al. Academic affiliation with a nursing home: impact on patient outcome. J Am Geriatr Soc. 1985;33:472-478.

4 Rubenstein LZ, Josephson KR, Weiland GD, et al. A randomized clinical trial of a geriatric evaluation unit. N Engl J Med. 1984;311: 1664-1670.

5 Gerety MB, Soderholm-Diffatte N, Winograd CH. Impact of prospective payment and discharge location in the outcome of hip fracture. J Gen Intern Med 1989;4:388-391.

6 Skeist HD. Role of physical therapists in physical activity programs in nursing homes: a survey. J Am Geriatr Soc. 1980;28:124-129.

7 Kempinich MV. How are physical therapists assisting nursing homes in patient care? Phys Ther. 1969;49:187-190.

8 Cooney LM, Fries BE. Validation and use of resource utilization groups as a case-mix measure for long-term care. Med Care. 1985;23: 123-132.

9 Folstein MF, Folstein SE, McHugh P. MiniMental State: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12:189-198.

10 Keeler EB, Kane RL, Solomon DH. Short-and long-term residents of nursing homes. Med Care. 1981;19:363-369.

11 Liu K, Manton K. The characteristics and utilization patterns of an admission cohort of nursing home patients. Gerontologist 1984; 24:70-76.

12 May FE, Stewart RB, Cluff LE. Drug interactions and multiple drug administration. Clin Pharmacol Ther. 1977;22:322-328.

13 Hing E. Use of Nursing Homes by the Elderly Preliminary Results from the 1985 National Nursing Home Survey Advance data from Vital and Health Statistics.

LK Chiodo, MD, is Assistant Professor of Medicine, Division of General Medicine, The University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Dr, San Antonio, TX 78284-7879. Address correspondence to Dr Chiodo at Extended Care (18), Audie L Murphy Memorial Veterans Hospital, 7400 Merton Minter Blvd, San Antonio, TX 78284 (USA). MB Gerety, MD, is Assistant Professor of Medicine, Division of General Medicine, The University of Texas Health Science Center at San Antonio. CD Mulrow, MD, is Associate Professor of Medicine, Division of General Medicine, The University of Texas Health Science Center at San Antonio. MC Rhodes is a consultant in computer applications development. Her address is 411 Tammy, San Antonio, TX 78216. MR Tuley, PhD, is Statistician and Health Science Specialist, Geriatric Research, Education and Clinical Center, Audie L Murphy Memorial Veterans Hospital. This project was approved by the Institutional Review Board of the Audie L Murphy Memorial Veterans Hospital.

This article was submitted December 4, 1990, and was accepted August 28, 1991.

Commentary

The benefits of physical therapy in any setting have only recently been tested against patient health outcomes such as long-term nursing home placement or functional status. By functional status, I mean the measured ability of the patient to function physically, psychologically, and socially in everyday life. Those studies that have evaluated patient outcomes have usually focused on hospitalized patients. Chiodo and colleagues have, in this article, issued a challenge to the field of physical therapy. Can therapists document that they make a significant difference to patients? It is critical to the future of physical therapy in the nursing home setting, and to the health of patients in nursing homes, if we believe therapy to be effective, that this question be answered; yet, the literature on this subject is sparse.(1) Physical therapy is expensive and is viewed as a more discretionary portion of the nursing home bill than basic nursing care, food, or medications. As discharges to nursing homes, and pressures to control nursing home costs, increase, the quantity and quality of physical therapy will come under increasing scrutiny. As has been true for other aspects of medical practice, not all outcomes are likely to be favorable.(2) Some patients will benefit more than others, and some types of therapy will be better than others. Current patient health outcome measures, such as activities-of-daily-living (ADL) assessments and other functional status measures, may require modification to fully reflect meaningful clinical improvement. Yet, the result of outcome-oriented studies, even when negative, will be to define, refine, and ultimately strengthen the field. This article provides us with a preliminary "yes" to the question: "Does the intensity of physical therapy, independently or in combination with other measures of the intensity of medical care, improve patient outcomes?" The study design had important strengths in setting about answering this question. Although the study was retrospective, a cohort of all admissions to the nursing home during a specified time frame were identified and followed forward through time albeit via medical records). Selection bias was thus minimized. Another important feature of the design was the identification of baseline patient characteristics and an analysis of how these characteristics might have affected the intensity of physical therapy. This is important because it could be that physical therapy intensity is a proxy measure for how sick the patient is. The better outcomes of the patients who received more therapy could be due to the better health of these patients. By examining the relationships between physical therapy intensity and age, cognitive score, and baseline ADL status, the authors have done about as well as a retrospective study could do to help us evaluate this potential problem. Age, which was associated with lower physical therapy intensity, was then tested in a regression analysis that included the age-intensity interaction as well as physical therapy intensity to predict outcomes, with the result that physical therapy intensity alone was significantly associated with improvement.

Can we then conclude that physical therapy improves outcomes? The authors of this article would agree with me that the answer is not yet available. They acknowledge that their study took place at a single site, that the design was retrospective, and that their measures of patient prognosis and outcome were imperfect. The major objective of this study was to determine whether the intensity of physical therapy, when controlling for other types of therapeutic intensity and for patient characteristics, was associated with improved patient outcomes.

To further understand the study's limitations, and to set an agenda for future research, we must carefully review the components of this objective. Can the study design support the study objectives? Were patient characteristics assessed adequately? Was the intensity of physical therapy measured well (reliably and validly)? Were the intensities of other types of care measured well, and were the most important types of care assessed? Was the outcome assessment used reliable and valid?

The research design used in this study can, although with low generalizability because of the study's single site, support the study objective of determining whether there is an association between physical therapy and positive patient outcomes. (There was one.) Association, however, is not causation. The most likely alternative hypothesis to the desired one-which is that physical therapy causes improved outcomes-is that the provision of physical therapy is a better measure of patient prognosis than are the measures used by the authors. As the authors used standard measures for three of the prognostic variables they measured, we can assume that baseline cognitive status, ADL status, and age were measured reliably and validly. The number of medical problems has not been well correlated with outcomes in the literature, nor was such a correlation demonstrated in this study. What other prognostic variables could be important? Other researchers, who may be conducting prospective studies with fewer limitations on what information can be obtained, could, for example, measure the physician's or physical therapist's reasons for providing submaximal therapy. The baseline patient assessment could include other factors we can hypothesize will affect the patient's frequency and duration of therapy, such as the patient's level of motivation, pain, conditioning, social support, and severity of illness. The intensity and duration of physical therapy were measured about as well as a record review could measure them, and the fact that the measures showed an association with outcome supports their validity. Future studies, however, will need to address intensity in more detail-What exactly was done during these sessions? Likewise, the intensity of speech therapy and of psychosocial therapy were measured by frequency of services. Medical and nursing care intensity were also addressed by study measures, but it is difficult to evaluate the reliability and validity of these measures, which were more interpretive. More detail about nursing care intensity may be available in future studies through the nursing home uniform data sets now being mandated by the Health Care Financing Administration. Of the measures used in this study, the outcome measures may be the most problematic. It is a positive feature of this study that outcomes were rated by two reviewers, that the level of agreement between them is reported, and that disagreements were adjudicated. In future studies, however, interrater reliabilities (eg, Kappa statistics) should be reported, using more reviewers. Reviewers should be blinded to the purpose of the study, and explicit criteria for exactly what constitutes a given outcome level should be defined. Improvement should be defined by comparing baseline measures of those aspects of patient health status that are to be improved by therapy (eg, functional status) and subsequent patient health status, preferably at some predefined time after nursing home admission (eg, 3, 6, or 12 months). Other patient characteristics that might affect the likelihood of a good outcome, such as the patient's overall sickness level, should be controlled for.

In summary, this article assesses the efficacy of an understudied technology-physical therapy in the nursing home. The authors are to be commended for taking as broad a view of their subject as possible within the constraints of their study design and for providing an excellent basis for future studies of this important topic.

Lisa V Rubenstein, MD Pilot Ambulatory Care and Education Program (OOP) Sepulveda VA Medical Center 16111 Plummer St Sepulveda, CA 91343

References

1 Keith RA. The comprehensive treatment team in rehabilitation. Arch Phys Med Rehabil. 1991;72:269-274.

2 Jette AM, Harris BA, Cleary PD, Campion EW. Functional recovery after hip fracture. Arch Phys Med Rehabil. 1987;68:735-740.

Author Response

We would like to thank Dr Rubenstein for her commentary on our article. We found her comments to be insightful, accurate, and useful in setting an agenda for future research on physical therapy in the nursing home setting. The study limitations noted by Dr Rubenstein fall into three categories. She correctly cautions against generalizing the results of a study conducted at a single site. She then makes several important comments about the potential for selection bias. Finally, she discusses the outcome measures we used and suggests how they could be improved in future studies. As Dr Rubenstein notes, it would be premature to conclude that physical therapy is beneficial to patients in nursing homes on the basis of the results of our study. Dual concerns arise from our particular study site. First, we do not know whether interventions that are effective for patients in Veterans Administration VA) nursing homes are likely to benefit other patients in nursing home settings. Second, we do not know whether the patients and programs in VA teaching nursing homes with multidisciplinary teams differ from those in nursing homes not affiliated with teaching programs. Though answers to these questions are not yet available, we do know something about the similarities and differences between patients in VA and community nursing home settings. Though the level of function in activities of daily living is similar in these patients, patients in VA nursing home programs are almost exclusively male, are younger, and have a higher prevalence of stroke and a lower prevalence of dementia than do patients in community nursing home settings. Each of these characteristics could influence the outcome of physical therapy. Clearly, our study is only a beginning. Much work needs to be done to determine which groups of patients in nursing homes will benefit from physical therapy.

We strongly agree with Dr Rubenstein that selection bias remains a concern in our study, as in any retrospective study. Though we did everything we could to determine whether there were important baseline differences in patients receiving different intensities of physical therapy, her theory that physical therapy intensity might be a proxy measure for prognosis is credible and important. The prognostic variables she lists-severity of illness, motivation, social support, and level of pain-could not be measured in a retrospective study and could significantly confound results. Performance of a prospective study is the only way to address this concern. Though we used predetermined criteria to determine outcome at the 4-month point, the use of validated, reliable instruments by blinded raters would greatly strengthen conclusions about the effectiveness of physical therapy.

After completing our data analysis, we had to decide which, if any, components of nursing home care should be studied further. We believed the evidence for benefit from physical therapy was strong enough to warrant a clinical trial. We are conducting a randomized, controlled trial of physical therapy in patients in community nursing home settings. The study design avoids the limitations of our initial retrospective study and incorporates many of the elements identified by Dr Rubenstein. Reliable, valid measures of physical status (ie, strength, range of motion, mobility, activities of daily living), cognition, affect, and social function are administered by research assistants blinded to treatment assignment. Reliable assessment protocols and standardized treatment application methods were designed by project physical therapists to ensure that the content of the intervention is consistent and reproducible. We hope that when this trial is completed, we will have a better understanding of who benefits from physical therapy in nursing homes and whether physical therapy affects one or more than one domain of function.

Laura K Chiodo, MD Meghan B Gerety, MD Cynthia D Mulrow, MD Mary C Rhodes Michael R Tuley, PhD