Barker WH, Borisute H, Cox C. A Study of the Impact of Influenza on the Functional Status of Frail Older People. Arch Intern Med. 1998;158(6):645–650. doi:10.1001/archinte.158.6.645
Excess hospitalization and death are well-known impacts of influenza on older people; however, little is known regarding the impact of influenza on functional status. We hypothesized that frail older people are at risk of functional decline as an outcome of influenza.
To measure the effect of acute influenza on the physical and mental status of older patients residing in nursing homes.
Our study was conducted in 6 nursing homes that participated in the Medicare Influenza Vaccine Demonstration and experienced laboratory-confirmed outbreaks of influenza in 1991 and 1992. A case-comparison design was used. One hundred sixteen of 131 residents who developed influenza-like illness and survived at least 4 months served as the case subjects; 127 of 132 residents without influenza-like illness who survived served as the comparison subjects. Measures of functional status 1 to 2 months before outbreak and 1 to 2 months and 3 to 4 months after outbreak were collected from medical records. Matched pairs analyses were conducted to ascertain changes in selected measures of functional status within each of the study groups. Wilcoxon signed rank tests for statistical significance were used.
Among surviving case subjects and comparison subjects, 25% and 15.7%, respectively, experienced decline in at least 1 major function (P=.04). Case subjects experienced significant decline in independence in bathing, dressing, and mobility while comparison subjects experienced decline in mental status.
Within the limitations of this study, influenza is observed to cause decline in major physical functions in more than 9% of survivors. Such disabling outcomes constitute an important new measure of impact of influenza on the frail elderly.
INFLUENZA IS an important cause of morbidity and mortality in the United States, with people 65 years or older experiencing the greatest impact. Extensive studies of influenza among older individuals living both in the community and in nursing homes have largely focused on mortality, hospitalization, respiratory symptoms, and cost as outcome measures.1- 8 The literature on the impact of influenza on the functional status of older people is scant and limited to studies of changes in level of care following hospitalization. Two recent studies9,10 of outcomes of hospitalization of individuals older than 65 years for pneumonia caused by influenza and other respiratory tract pathogens reported that 10% to 12% were discharged to higher levels of care compared with their residence before hospitalization.
Given the considerable physiological stress imposed by acute influenza virus infection,11 it is postulated that older people with preexisting chronic conditions and associated disabilities who experience acute influenza illness and survive are likely to experience temporary or permanent decline in one or more functional capacities as a consequence of the influenza illness. This study measures the effect of acute influenza on the physical and mental status of older patients residing in nursing homes. We compare the degree of functional change among nursing home residents who contracted influenza and survived with that occurring among fellow residents who did not contract influenza and survived during the same period.
Our study was conducted in Monroe County, New York, in cooperation with the county's participation in the 1989 through 1992 nationwide Medicare Influenza Vaccine Demonstration.12 The demonstration included communitywide, laboratory-based influenza surveillance involving selected medical practices, community hospitals, and all 33 nursing homes in the county. The nursing homes include approximately 5000 residents divided among 150 nursing units with 20 to 30 residents per unit.13
Nursing home surveillance, adapted from guidelines used by the Centers for Disease Control and Prevention and others,14 consisted of weekly telephone reporting from all nursing homes of all new influenza-like illnesses, defined as an acute onset of fever with a temperature higher than 37.7°C with respiratory symptoms. When a potential outbreak occurred, defined as 2 or more new case subjects with influenza-like illnesses in the same nursing unit within 24 to 48 hours, nasopharyngeal cultures were obtained from up to 5 early "sentinel" case subjects and processed by the clinical virology laboratory at the University of Rochester Medical Center, Rochester, NY. An outbreak was considered confirmed if 1 or more cultures were positive for influenza. In a confirmed outbreak, all new cases of influenza-like illnesses that occurred within 24 to 48 hours of a previous case of influenza-like illness were considered to be caused by influenza.
During the 1991 and 1992 national epidemic of influenza A/H3N2, confirmed outbreaks occurred in 14 of the 33 nursing homes. This study is based on 6 of the nursing homes, all of which experienced outbreaks of 10 or more cases of influenza-like illnesses among which influenza A virus was isolated from 3 or more of 5 sentinel case subjects. Attack rates in the outbreaks ranged from 16% to 30% of residents on affected units. A total of 131 residents who developed influenza-like illnesses during the confirmed outbreaks form the case subjects for the study. Comparison subjects were selected by simple random sampling among residents on the outbreak nursing units who did not develop acute respiratory tract illness during the outbreak period.
Using standardized abstracting forms and instructions, data on clinical and functional status were collected retrospectively from medical records for case and comparison subjects. Clinical data collected on the case subjects included respiratory tract symptoms, malaise, highest daily temperature, occurrence of pneumonia, current influenza vaccination status, and treatment with antibiotics or amantadine hydrochloride. Underlying major chronic diseases known to predispose to influenza complications were abstracted from problem lists in medical records for case and comparison subjects; these belonged to one of the following broad categories: chronic pulmonary disease, cardiovascular disease, diabetes, cancer, or other.
Deaths and cause of death were ascertained from medical records for all case and comparison subjects during a 4-month period from time of onset of influenza outbreaks in their respective nursing homes. Functional status data were obtained from routine physical and mental status assessments completed by trained nurses at 30- to 60-day intervals using the standard Patient Review Instrument (PRI), which has been in use in all New York State nursing homes since 1985.15 For purposes of this study these data were ascertained from PRIs completed for each case subject and comparison subject at 3 points in time as follows: within 1 to 2 months before occurrence of the influenza outbreak, to establish baseline measures, and at 1- to 2-month and 3- to 4-month intervals following the end of the outbreak. Physical function and mental-behavioral status measures available from the PRI include mobility, transfer, dressing, bathing, alertness, and verbal aggression.
In addition to standard PRI measures of functional status, a general health status assessment was obtained on each surviving case subject and comparison subject prior to and following the outbreaks. This consisted of a review of resident medical and nursing records 1 month before outbreaks and during the 1 to 2 months and 3 to 4 months after outbreak intervals to ascertain evidence of intercurrent deterioration of health status that might not be reflected in standard PRI measures of functional status. Deterioration included documented worsening of underlying chronic medical conditions, mental status changes, or medically attended illness episodes such as infections and falls.
Although the overall design of the study involved the comparison of 2 independent groups, the primary question for analysis was whether individual subjects within each of the 2 groups changed with regard to functional status during the specified times of observation. This question requires the analysis of paired observations for individuals within each of the groups. Observations for each functional status category were made on 3-level ordinal scales ranging from better to worse.
The simplest way to display such paired data is to make contingency tables in which each subject is classified according to results of repeated measurements at different points in time; ie, baseline vs 1 to 2 months following outbreak, or baseline vs 3 to 4 months after outbreak. It is natural to use the earlier of the observations as the left margin of the table. Such a table clearly shows changes over time. Subjects in the diagonal cells of the table are those whose status was unchanged. Since the levels of function are ordered from better to worse, subjects in the portion of the table that is above the diagonal are those whose function was worse at the second observation. Similarly, subjects below the diagonal are those whose function improved. Using the functional status measurements at 3 different observation times, we constructed 2 different tables that compare individual subjects' functional levels, respectively, at 1 to 2 months and 3 to 4 months after outbreak vs the baseline levels before outbreak. To assess changes over time, we included percentages in each table, calculated by grouping subjects according to their initial status, ie, for subjects at a given functional level at the baseline observation, we calculated proportions who were improved, the same, or worse at the follow-up observations.
To answer the question of whether subjects in each group changed over time, Wilcoxon signed rank tests were performed on the differences for each of the 3 possible pairs of measurements. Although these tests do not make direct comparisons between the 2 groups, they address the question of whether there were changes in each group separately. Indirect comparisons of the experiences of the 2 groups can be made using the results of the 2 independent significance tests.
Clinical manifestations recorded among the 131 case subjects included cough in 90%, congestion in 76%, and malaise or myalgia in 25%. The mean duration of recorded symptoms was 6.2 days. The mean of the case subjects' highest recorded temperatures was 38.3°C. Pneumonia was clinically diagnosed in 8 case subjects (6%). The current year's influenza vaccine had been received by 108 case subjects (82%). Antibiotics and amantadine hydrochloride were administered to 25% and 18% of case subjects, respectively; time of initiation in the course of illnesses and amount of these drugs received by case subjects was not abstracted. Clinical findings, including pneumonia rates, were all similar among those case subjects with and without history of current vaccination or receipt of amantadine or antibiotics.
Among the 131 case subjects with influenza and the 132 comparison subjects (who did not experience influenza) there were 15 (11.5%) and 5 (3.8%) deaths, respectively, during the period up to 4 months following the influenza outbreak (P=.19). Deaths among case subjects were caused primarily by pneumonia or cardiovascular disease, including several myocardial infarctions and 1 case of stroke, while those among comparison subjects were attributed to a variety of causes.
The 116 case subjects and 127 comparison subjects who survived at least 4 months after the outbreak were used to assess impact of acute influenza on functional status. Baseline age, sex, vaccination status, prevalence of underlying major chronic diseases, and measures of functional status were similar between these 2 groups. While more than 75% of subjects in both groups were women, there was a larger proportion of men in the comparison group; however, there was no association between level of functional dependency and sex (Table 1).
Table 2 and Table 3 summarize the findings from analysis of paired observations to identify documented changes in status for selected functions occurring in individual case and comparison subjects at 1 to 2 months and 3 to 4 months after the outbreak, respectively. In these tables, for each function studied, the number of individuals classified by each level (eg, for physical functions: independent, partial assistance, or complete dependence; for mental-behavioral problematic function: none, partial, or severe) at the baseline before outbreak is shown in the left-hand column. The respective changes, if any, in levels from baseline measures for case subjects and for comparison subjects are shown as numbers and percentages in the rows in the tables. As an example, in Table 2 under "Bathing," of 34 case subjects who were independent at baseline, 30 (88.2%) were still independent, 1 (2.9%) had become partially dependent, and 3 (8.8%) had become completely dependent at 1 to 2 months after the outbreak; among 51 comparison subjects who were independent at baseline, at 1 to 2 months 50 (98%) were still independent, 1 (2.0%) was partially dependent, and none completely dependent. Similar data are shown in the respective rows for case and comparison subjects who were partially or completely dependent in bathing at baseline. Statistical significance of observed changes in levels of function for case and comparison subjects are based on 2-sided Wilcoxon signed rank tests.
At 1 to 2 months after the outbreak there was a tendency toward declines in the level of independence for bathing, dressing, and mobility among case subjects and a statistically significant decline in mental alertness among comparison subjects (P=.03) (Table 2). At 3 to 4 months after the outbreak, statistically significant declines in functional levels were documented among case subjects for bathing, dressing, and mobility, with marginally significant decline in independence in transfer. With exception of the persisting significant worsening in mental function, there was no significant decline in any functions among comparison subjects (Table 3).
In aggregate comparison of the 2 groups of subjects, 29 (25%) of the 116 surviving case subjects and 20 (15.7%) of the 127 surviving comparison subjects experienced decline in 1 or more functions at 3- to 4-month follow-up. This finding translates to a 9.3% excess of decline among case subjects over what would have been expected in the nursing home setting. Furthermore, case subjects with functional decline were more likely than comparison subjects to experience decline in multiple functions (Table 4).
With respect to general health status, which was a measure of occurrence of intercurrent medically attended events, it is notable that the 14 case subjects who experienced such events during the 1- to 2-month interval after outbreak represented a significant increase compared with occurrences among case subjects before outbreak (P=.004); however, no significant increase in occurrence of such events was observed among comparison subjects. The various manifestations of intercurrent deterioration of general health among 14 case and 8 comparison subjects in the 1- to 2-month interval after outbreak are shown in Table 5.
This study has compared functional status decline among a series of nursing home residents who experienced acute influenza with that occurring among a series of residents of the same nursing homes who did not experience influenza-like illnesses. Decline in 1 or more functions was observed among 25% of the former and 16% of the latter, yielding an estimated net attributable rate of functional decline of 9 per 100 residents with influenza in this population. Specifically at 3 to 4 months after outbreak, statistically significant declines in bathing, dressing, and mobility were observed among case subjects, while, with the exception of mental alertness, there was no such evidence of decline among comparison subjects. A significant increase in frequency of intercurrent manifestations of deteriorating health status within 1 to 2 months after the outbreak was also documented among case subjects but not among comparison subjects.
The validity and significance of our findings should be interpreted in light of the study's strengths and limitations. Case subjects were all involved in focal outbreaks from which influenza A (H3N2) virus was isolated from multiple sentinel case subjects, which occurred concurrent with well-documented epidemic influenza A (H3N2) in the surrounding community.13 Under these circumstances it is reasonable to assume that virtually all the case subjects were affected by influenza. The involved nursing homes were participants in influenza surveillance of all nursing homes in the county, constituting an unselected experience for assessing impact of influenza on functional status of nursing home residents. Baseline demographic, medical, and functional status of case and comparison groups was similar, thus removing the likelihood that the difference in functional decline could be attributed to differences in underlying health status. Measures of functional status at several points in time were extracted from routinely collected administrative data in which measurement criteria are consistently applied by trained staff. Measurement of intercurrent deterioration in general health status was based on episodes documented in medical records. While it is possible that such episodes were more likely to be recorded by staff following recent flulike illness, hence giving a spurious finding of increased frequency among these residents, this seems unlikely given the clinically significant nature of the events that were noted (Table 5). Since more than 80% of residents in the involved nursing homes were vaccinated for influenza, generalizability of our findings would be limited to highly immunized nursing home populations.
Functional status decline as observed in this study following acute influenza illness may be a consequence of either specific or nonspecific pathogenic mechanisms leading to loss of strength. Myalgia, malaise, and fatigue are common manifestations of acute influenza, indicative of the systemic nature of the disease and an apparent propensity for muscle tissue. Severe myopathy and isolation of influenza virus from muscle tissue have been described in case reports involving older patients.16 While such cases are rare, it is reasonable to surmise that a mild form of direct muscle damage may be common and impact on physical function among frail older individuals with limited muscle reserves. A second fully plausible explanation would attribute decline in function to deconditioning, including loss of muscle tissue, as a consequence of being at bed rest or chair bound during the acute and convalescent stages of a bout of influenza.17,18 Because of the limited data available in our retrospective medical record study, we are unable to examine the possible contribution of bed rest and chair boundedness to the observed excess decline in functional status among case subjects with influenza.
For several reasons it is likely that this study does not capture the full impact of acute influenza on functional status among frail older people. First, and most important, is the possibility of instrument bias. Because our subjects were already substantially compromised in many functions measured by the PRI (Table 1), there may have been limited room for further measurable decline to occur, ie, a "floor effect."19 The somewhat broad measures of functional status of the PRI, designed for administrative and reimbursement purposes to measure burden of care needs (eg, partial vs complete assistance), are likely to miss subtle but significant changes in functional well-being within care levels. Such changes might be perceived by the subject but not by the observer. As an alternative to traditional administrative functional status measures, use of the nursing home–adapted sickness impact profile, which directly assesses resident perception of change in health status, might more sensitively detect impact of acute influenza on functional well-being.20,21
Second, the extent of major functional decline among subjects who experienced influenza-like illnesses might have been partially attenuated by the high level of influenza vaccination among our subjects. Under these circumstances, while vaccination may not completely protect against the acute febrile respiratory illness of influenza, it may limit duration and severity of incapacitating systemic effects of influenza infection. This would be analogous to earlier observations of vaccination attenuation of severity of influenza-associated pneumonia among both community- and nursing home–dwelling older individuals.3,5,22 While no such effect was noted in the present study, the sample of nonvaccinated case subjects for comparison was small in the highly immunized nursing home populations whom we studied.
In conclusion, further understanding of the impact of acute influenza on functional well-being of frail older individuals would be well served; first, by observing subjects prospectively and using more discriminating measures of function than are afforded by administrative instruments such as that used in the present retrospective study; and second, if possible, by observing large enough numbers of unvaccinated as well as vaccinated subjects to determine whether influenza vaccine protects against the impact of influenza on functional status.
Accepted for publication August 17, 1997.
This work was supported in part by grant H53/CCH2033941 from the Public Health Service, Washington, DC.
A copy of the PRI is available on request from Dr Barker.
We acknowledge the helpful cooperation of the administrative, nursing, and medical records staff of the participating nursing homes. Laboratory support was provided by the virology diagnostic laboratory at the University of Rochester Medical Center, Rochester, NY, under the supervision of Marilyn Menegus, PhD; Charlene Freundlich assisted with data collection and coordination; and Matthew Carlson and Christine Brower assisted with statistical analysis. Ann Falsey, MD, University of Rochester, and Nancy Arden, RN, MPH, Centers for Disease Control and Prevention, Atlanta, Ga, provided valuable consultation on preparation of the manuscript.
Reprints: William H. Barker, MD, Department of Community and Preventive Medicine, Box 644, University of Rochester Medical Center, Rochester, NY 14642.