The Impact of Falls in an Inner-City Elderly African-American Population Jeane Ann Grisso MD, MSc,* Donald F. Schwarz, M D MPH, M B A , t Virginia Wolfson,t Marcia Polansky, ScD,$ and Karin LaPann* Objective: To describe the impact of falls in an elderly African-American urban community and to identify predictors of poor recovery from falls. Design: Prospective cohort study. Setting: Emergency departments of 11 hospitals in western Philadelphia. Subjects: Interviews were conducted with 197 African-American persons 65 years and older residing in West Philadelphia who were treated at an emergency department because of a fall. Second interviews were conducted a median of 7 months after the fall with a sample of 70 patients who had not recovered at the time of the first interview. Main Outcome Measures: Information abstracted from emergency department medical records and information on recovery obtained from two subsequent interviews. Results: A median of 8 weeks after the fall occurred, 43% of persons reported continued pain or restriction in their usual activities as a result of the fall. Predictors of poor recovery included the presence of grandchildren in the household, hearing impairment, severity of the injury, and injury to the
lower extremities. Having someone present at the time of the fall was associated with a lower risk of poor recovery. Fortyone percent of the 70 persons interviewed a second time reported continued pain or restriction in usual activities a median of 7 months after the fall occurred. However, only 7% and 39% had received the services of a home health aide or physical therapist, respectively, and only 14% reported that a physician or other health professional had been particularly helpful since the fall had occurred. Conclusions: A large proportion of elderly African-American persons treated at emergency departments for falls experience continued pain and restriction of activities after the fall. Many individuals have not recovered 7 months or longer after the fall and few persons report that a physician or other health professional has been particularly helpful since the fall occurred. We suggest that follow-up programs be developed for elderly persons in minority communities who come to emergency departments after a fall. J Am Geriatr SOC40:673678,1992
alls, a leading cause of death among the elderly, commonly occur among older persons living in the community.’ Although many believe that falls lead to disability and institutionalization, few prospective studies have been conducted to assess the impact of falls on subsequent function, and these studies have focused on white, often middle-class, No community-based studies of the sequelae of falls in inner-city minority populations have been reported in the literature to date. These were among the considerations that prompted us to establish the Philadelphia Injury Prevention Program to study injuries that occur in an indigent minority community. The program involves active community-based surveillance of all injuries that occur in a geographically defined population and are treated at emergency departments, an in-depth study of falls among the elderly, and the development of a comprehensive injury prevention program. In this report, we present the findings of the in-depth study of falls. This study evaluates the immediate sequelae of falls in an elderly minority population as well as recovery status a median of 2 months and 7 months after the fall occurred. Predictors of poor recovery are identified, and use of health services is described.
METHODS In this prospective cohort study, we collected data on elderly fallers at three points in time. First, medical record data at the time of the emergency department visit was abstracted as part of the ongoing surveillance system described below. Second, home interviews were conducted, on average, 8 weeks after the fall occurred. And finally, second interviews were conducted a median of 7 months after the fall with individuals who, at the time of the first interview, reported continued pain or restriction in their usual activities as a result of the fall. Each component of the cohort study is described below. Surveillance The Philadelphia Injury Prevention Program is a cooperative effort of the Philadelphia Department of Public Health and the University of Pennsylvania to evaluate fatal and non-fatal injuries in an urban African-American population of 68,103 people residing in 17 census tracts in western Philadelphia. This population is 97.2% African-American and poor, with a median family income in 1986 of $11,810. An active surveillance system employs trained medical abstractors to obtain information on injuries that occur among residents of the targeted community. Data are collected from 11 emergency departments, the Office of the Medical Examiner, and the Bureau of Vital Statistics. The methods for this ongoing, population-based active surveillance system have been described in detail e l ~ e w h e r eBriefly, .~ a multistage assessment was used to select emergency departments for surveillance and to evaluate the completeness of case ascertainment.
F
* From the Clinical Epidemiology Unit, Section of General Internal Medicine, University of Pennsylvania; tchildren’s Hospital of Philadelphia; and SHahnemann University, Philadelphia, Pennsylvania. Supported by a grant for the Division of Injury Epidemiology and Control. Centers for Disease Control. Correspondence to Jeane Ann Grisso, MD, MSc, Clinical Epidemiology Unit, 317R Nursing Education Building/6095, Philadelphia, PA 19104.
JAGS 40.673-678, 1992
0 1992 by the American Geriatrics Society
0002-8614/92/$3,50
674
GRISSO ET AL
This assessment involved (1) a 20-hospital evaluation of the utilization of inpatient services by census tract of residence; (2) a population-based door-to-door survey of 532 community residents to ascertain which hospitals are used for care of injuries; (3) a systematic evaluation of fire department, school, and emergency medical transport records; and (4) a survey of all private practitioners, clinics, and community health centers in West Philadelphia in which we learned that fewer than 5% of acute injuries were treated in non-emergency settings. On the basis of these results, the final sample of institutions for emergency room surveillance included 11 institutions: three regional trauma centers, three general hospitals, three regional burn centers, one children's hospital, and one eye hospital. Information on deaths was also collected from the vital statistics death certificate files and from the Office of the Medical Examiner. Case Identification and Recruitment Cases were identified from the surveillance system. All those aged 65 years and older who were seen because of a fall as defined by the International Classification of Diseases, Ninth Revision, External Causes of Injury Codes, E880E888, E817, E818, E824, E829 were included. During the period from September, 1987 to November, 1988, 299 community-dwelling African-American persons were identified. Of these, 11 denied that the injury had occurred, six could not be traced, and 67 refused to participate in the study (Figure 1). Those who refused to participate did not differ from participants in age, gender, fracture or hospitalization rate, or injury severity score (AIS). Interviews were completed with 200 persons; 175 interviews were with the injured person and 25 were with proxy respondents who were interviewed either because the injured person had died (5 cases) or be-
r---->
299 Fell
102 Excluded
VI
r---> r---> r--->
197 First Interviewed 113 Recovered
VI
84 Poor Recovery' 14 Second Interview Not Possible
VI
70 Second Interview 41 Recovered
VI
29 Poor Recovery •
FIGURE 1. Poor recovery (continued pain, restriction in usual activities, or both)" among elderly fallers at two time periods following the fall.
JAGS-JULY 1992-VOL. 40, NO.7
cause the person was cognitively impaired (20 cases). Proxy respondents were obtained for 25 of the 40 persons eligible for proxy interview. One interview was incomplete and two others (both proxy interviews) did not provide adequate information to categorize the recovery of the injured persons. Thus, the results are based on 197 persons. Data from Emergency Department Medical Records Information indicating whether the person was hospitalized or discharged home was abstracted from the emergency department medical record. The presence and location of fractures was also recorded. Injury severity was classified according to the Abbreviated Injury Scale (AIS).6 AIS levels are based on location of injury and whether a fracture occurred. In our sample, an AIS score of 3 or 4 included those individuals who had major trauma, while a level of 2 included individuals who had long bone fractures (eg. femur, humerus), and level 1 included minor fracture (eg, finger, toe, etc., or soft tissue injures such as bruises.) First Interview Trained interviewers conducted the interview in the home. Hospitalized patients were interviewed after discharge either in the home or a nursing home. The questionnaire was a standardized, close-ended instrument that included questions about the circumstances of the fall, sociodemographic information, number of chronic illnesses, self-reproted visual and hearing impairment, pre-fall and post-fall functional status, household composition, whether the respondent was hospitalized, and the respondent's perception of the duration of pain and whether he or she had to "cut down" his or her usual activities. The prefall and post-fall assessments of activities of daily living included questions on whether the participant had difficulty with any of the following tasks: walking across a room, walking a block or more, going up or down stairs, standing for a long period, and bending and picking up a shoe from the floor. In the post-fall questions, if the participant indicated having difficulty with a task he or she was asked if it was due to the fall. Second Interview Second interviews were conducted with those who reported continued pain or limitation in function as a result of the fall at the time of the first interview. Follow-up interviews were carried out with 70 of 84 eligible respondents a median of 7.4 months after the fall occurred (Figure 1). Four persons had died in the interim period, four refused the second interview, and six could not be traced. At the time of the second interview, respondents were asked about their current level of function and pain, what health services had been used in the period since the first interview, and what assistance had been received from family or others. Recovery from the Fall The assessment of poor recovery was based on whether respondents reported continued pain, whether they reported that they continued to cut down on their usual activities, or both. When respondents reported either a limitation in general function or continued pain, they were asked whether this was because of the fall. Respondents were
]AGS-JLILY 1992-VOL.40, NO. 7
RECOVERY FROM FALLS IN AN URBAN COMMUNITY
classified as poorly recovered only if they attributed the limitation in function or continued pain to the fall. Statistical Analysis Frequency distributions of respondents who had and had not recovered were compared using chi-square tests and Fisher’s exact test for categorical variables and student’s t test for continuous measures. Models were constructed to assess predictors of poor recovery. Odds ratios with 95% confidence intervals were the primary measures of association. Multiple logistic regression models were constructed to assess the independent contribution of predictors of poor recovery while controlling for potential confound e r ~Predictor .~ variables with P values of 0.15 or less were initially retained. These variables were evaluated in subsequent logistic regression models and only those with P values of 0.05 or less were included in the final models. Models were constructed to identify predictors of poor recovery at the time of the first interview and on the subset who were interviewed a second time to identify predictors of those who still had pain or functional limitations as a result of the fall. All models included age and duration of time from the fall to the interview as possible confounding variables. Secondorder interactions were also assessed for predictor variables and duration of time from the fall to the interview. The study was approved by the Institutional Review Boards of each participating hospital. Written informed consent was obtained from each respondent.
RESULTS Circumstances of the Falls The most common time of day for falls was in the morning (40%), followed by afternoon (25%). Most falls occurred indoors (72%), usually in the respondent’s own home (65%). Of those falls that occurred indoors, the bedroom was the most common site (24%), followed by the stairs (23%). Most falls occurred at a standing height (49%), 29% involved falling from a greater than standing height (including falling down stairs), and 21% occurred while sitting, lying, or bending over. A tripping or slipping hazard was identified in 30% of falls, most commonly an irregular or wet surface or clutter, like clothes on the floor. Most persons (59%) got up within 5 minutes after the fall occurred; a few (6 persons) lay for 3 hours or longer (range, 3 to 13).
675
Immediate Impact of Fall Of the 197 persons, 61 (31%) were hospitalized for injuries caused by the fall. Hospitalizations ranged from 1 to 192 days; the median duration was 8 days. Five persons were discharged to nursing homes. Fractures resulted from 58 falls (29%); the most common fracture sites were hip (16 cases), leg or ankle (14 cases), and arm or shoulder (8 cases). The remaining 20 fractures occurred in eight other sites. Seventeen persons lost consciousness as a result of the fall, and three others were unsure whether the loss of consciousness preceded or followed the fall. The distribution of AIS scores were as follows: 5% suffered no apparent injury, 61% sustained a minor injury (AIS = l), 23% sustained a moderate injury, usually a fracture of a femur (AIS = 2), and 11%suffered major trauma (AIS = 3 or 4). Recovery following the Fall at the Time of the First Interview The initial interviews were conducted a median of 8 weeks after the fall (range, 2.4 to 35). At the time of the first interview, 43% of participants reported that they had continued pain or limitation in their usual activities as a result of the fall. Table 1 shows the impact of the fall at the time of the first and second interviews. At the time of the first interview, the most common activities with which participants reported difficulty because of the fall were going up or down stairs (37%),walking a block or more (25%), walking across a room (22%), standing for long periods (22%), and bending over (20%). Fear of falling was reported by 46%; most (63 persons) said the fear arose as a result of the fall. Predictors of Poor Short-Term Recovery We evaluated predictors of poor recovery based on self reports of continued limitation in usual activities, continued pain, or both. Persons who had not recovered were less likely to be married and slightly more likely to be females. There were no differences in age, duration of education, or median monthly income. The models for predictors of poor recovery included sociodemographic factors (age, gender, income, education), medical factors, previous functional status, severity of trauma, and circumstances of the fall. All models controlled for age and duration of time from the fall to the first interview as potential confounding variables. The odds ratios and 95% confidence limits of risk
TABLE 1. IMPACT OF FALLS OF ELDERLY PERSONS AT TWO TIME PERIODS AFTER THE FALL OCCURRED First Interview* Second Interview** n = 70 n = 197 Continued pain Restricted usual activity Difficulty with: going up or down stairs walking one block or more walking across a room standing for long periods bending over Fear of falling
59 (30%) 39 (20%)
25 (36%) 24 (34%)
73 (37%) 49 (25%) 43 (22%) 43 (22%) 39 (20%) 91 (46%)
19 (27%) 16 (23%) 13 (19%) 15 (21%) 12 (17%) 44 (63%)
* The first interviews were conducted a median of 8 weeks after the fall occurred. ** The second interviews were restricted to a sample of 7 0 persons who reported continued pain and/or restricted activity at the time of the first interview. These interviews were conducted a median of 7.4 months after the fall occurred.
676
GRISSO ET AL
JAGS-JULY 1992-VOL.40, NO. 7
TABLE 2. PREDICTORS OF POOR RECOVERY FROM FALLS IN AN INNER-CITY ELDERLY AFRICAN-AMERICANPOPULATION* Risk Factors Odds Ratios (95% Confidence Limits)
Grandchildren living in household Hearing impairment Leg injury Severity of injury (AIS score) No injury AIS = 1 AIS = 2 AIS = 3 or 4
Previous difficulty in Activities of Daily Living** None One Two
Three or four H e b uresent at fall
3.3
1.1-9.5
2.6 2.4
1.3-5.5 1.O 1-5.7
1.0 (reference) 2.5 5.2 10.2
1.0 (reference) 0.3 0.5 0.3 0.5
0.3-25.6 0.5-50.4 0.9-113.5
0.1-0.8 0.2-1.4 0.1-0.6 0.3-1.0
* Logistic regression model based on factors listed above while controlling for age and duration from the time offall to the time of interview conducted a median of 8 weeks after the fall occurred. Poor recovery is defined as continued pain, limitation in usual activity, or both. **Activities of daily living include difficulty w i t h going up or down stairs, walking one block or more, walking across a room, standing for long periods, and bending over.
factors for poor recovery are shown in Table 2. The major risk factors for disability are the presence of grandchildren in the home, hearing impairment, leg injury and the severity of the injury. The presence of someone to help the respondent at the time of the fall was protective. Previous limitations in activities of daily living (ADL) also appeared protective. This may be due to respondents with previous disabilities being less likely to attribute limitations to the fall. When the same model was repeated except that persons were characterized as poorly recovered who reported continued limitations in function or current pain regardless of whether they attributed it to the fall, the effect of previous limitations in ADL reversed. In this model those with previous limitations in ADL were more likely to experience continued pain or limitations after the fall than were those without previous limitations in ADL. The effect of grandchildren living in the home was explored further. No relationship existed between previous disability or the number of chronic illnesses and the likelihood of living with grandchildren. We stratified respondents by sex to assess whether the presence of grandchildren was equally associated with poor recovery in men and in women. We found no association among men (P = 0.81) but a statistically significant effect in women (P = 0.05). Second Interviews: Predictors of Long-Term Recovery We conducted second interviews with 70 of the 84 persons who reported continued pain or restricted activity as a result of the fall at the time of the first interview. The median duration of time from the fall to the second interview was 7.4 months (range, 5.5-10.4). Respondents were asked about their current level of function and pain, what health services had been used in the period since the fall, and what assistance had been received from family or others. Forty-one percent of the 70 persons who had not
recovered at the time of the first interview reported continued pain or restrictions in usual activities as a result of the fall at the time of the second interview (Table 1). Thirty-six percent of respondents continued to have pain; 23% reported that their health remained worse or much worse than before the fall. Of the 24 persons (34%) who had to cut down their usual activities in the previous month, 14 (61%)reported that they had to cut down their usual activities every day. In response to the question as to whether they had to change the way they did things because of the fall, 39% of the 70 respondents answered "yes." The most common activities affected were housework, shopping, and walking. We assessed whether the factors that predicted poor recovery at the time of the first interview were predictive of continued poor recovery at the time of the second interview. Leg injuries (OR = 4.2, 95% C.I. 1.05, 16.7), the presence of grandchildren in the home (OR = 5.1, 95% C.I. 1.2, 21.6), and pre-fall hearing impairment (OR = 4.1, 95% C.I. 1.1, 14.9) were all associated with poor long-term recovery from falls in multivariate models controlling for age, sex, and duration since first interview. Three of the factors (AIS score, having help at the time of the fall, and pre-fall functional impairment) were not associated with longterm poor recovery as assessed at the time of the second interview. Use of Health Services and Non-Medical Services Of the 70 respondents who had continued pain or restricted activity at the time of the first interview, 26% had not seen a physician since the emergency department visit for the fall. Help from a physical therapist was obtained in 27 cases and from a visiting nurse in 19. Five persons received visits from a home health aide and three received service from Meals on Wheels. Fourteen persons (19%)had received help from family or friends, usually with meal prepa-
]AGS-]ULY 1992-VOL. 40,NO. 7
RECOVERY FROM FALLS IN AN URBAN COMMUNITY
ration. In response to the question, “Is there any person or place you feel has been particularly helpful in your recovery?” 27 persons (38%) reported a relative or friend, 10 persons (14%) reported a physician or other health professional, and 28 (39%)reported that no one had been particularly helpful.
DISCUSSION This prospective community-based study documents the extent of pain and functional limitation that results from falls treated in emergency rooms in an inner-city minority population. A large proportion of persons reported continued pain or restriction in very basic activities of daily living. Most falls were reported to have occurred in the home from a standing height. The circumstances of the falls were such that, if they had occurred in younger persons, in all likelihood few persons would have been injured and even fewer would be subsequently disabled. In this elderly cohort, however, rates of self-reported pain and disability remained high 5 to 10 months after the falls had taken place. Several limitations of this study should be noted. Although we believe that the emergency department surveillance system captured information on virtually all injuries that resulted in emergency department treatment in the community, those persons who did not seek emergency department care for a fall were not included in this study. Thus, the results of this study can only be generalized to those who seek emergency care for a fall. The level of disability was not independently assessed. No tests of neuromuscular performance were carried out. We do not believe that the elderly participants would have exaggerated their limitations; many elderly persons may be reluctant to admit to difficulties because of fear of in~titutionalization.~ We cannot be certain that the pain and debilitation reported in this population was caused entirely by the fall. Although we only included those responses in which the participants attributed their pain and functional limitations to the fall, it may be that other conditions contributed both to the risk for falling and to the subsequent pain and restricted activities. However, the number of chronic illnesses did not predict poor recovery, and it appeared that those with previous functional limitations were less likely to attribute subsequent limitations to the fall than those who had not been disabled prior to the fall. Several predictors of poor recovery at the time of the first and second interviews emerged. In this community, the presence of grandchildren in the home is associated with poor recovery. The reasons for this association are not clear. Those who lived with grandchildren did not have higher rates of prior disability or chronic illnesses than those who did not. The association of grandchildren and poor recovery was found in women, not in men. In poor urban African-American communities, many older women are assuming primary care for their children. It may be that women who are primary caretakers of their grandchildren recover less quickly from a fall. However, we do not
677
have information about whether these grandmothers were responsible for the care of their grandchildren. Although general measures of previous disability were not predictive of poor recovery from the fall, hearing impairment was associated with poor recovery. Hearing impairment may result in isolation and slower recovery, although the impact of hearing impairment was independent of household composition and frequency of social contacts. The severity of the injury, and leg injuries in particular, were also associated with poor recovery. We found the association with leg injuries to be independent of whether a fracture occurred. It may mean that the subsequent impairment in mobility due to lower extremity injuries is more devastating to elderly persons than restriction in activities related to upper extremity injury. The impact of falls in this elderly, inner-city, minority community is substantial. No only did a number of persons sustain fractures and require hospitalization, but even more importantly, the ability to carry out basic activities of daily living was restricted in a large proportion of cases. We cannot assess whether the sequelae of falls in this population are different from that in white, middle-class populations. The fracture rates in studies based on all self-reported falls (regardless of whether medical care is sought) are usually 5% to 6%, in contrast to the 28% fracture rate seen in our s t ~ d y .A ~ ,study ~ based on emergency department surveillance in Miami, FL reported that 27% of falls resulted in fracture.” The extent of limitations in function has not been documented in a similar population. In a study involving volunteers. Nevitt et a13 reported that 25% of those who fell reported having to restrict their activities after the fall occurred. In a prospective study of persons aged 75 ears and older living in the community, Tinetti et a1 found that 26% reported curtailment of their activities because of the fear of falling. Both of these studies included all self-reported falls and limited their assessment to the immediate period after the fall occurred. In our study, in which we conducted second interviews with persons who reported continued pain or restricted activity at the time of the first interview, less than half reported having received help from a visiting nurse or physical therapist, and very few (7%) had received visits from a home health aide in the 5 to 7month period after the fall occurred. Only 10 persons reported that a health professional had been particularly helpful in their recovery and 39% said that no person or place had been particularly helpful. Our impression is that many elderly persons in this community who fall, and who subsequently are in pain or disabled, do not receive much in the way of medical or rehabilitative services. We propose that follow-up programs be developed for elderly persons from urban minority communities who come to emergency rooms because of a fall and that these programs be investigated with well designed clinical trials.
Y
REFERENCES 1. Tinetti ME, Speechley M. Prevention of falls among the elderly. N Engl J Med 1989;320:1055.
2. Lucht U. A prospective study of accidental falls and resulting injuries in the home among elderly people. Acta Socio-Med Scand 1971;3:105. 3. Nevitt MC, Cummings SR, Kidd S et al. Risk factors for recurrent nonsyncopal falls. JAMA 1980;261:2663. 4. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319:1701. 5. Grisso JA, Schwarz DF, Wishner AR et al. Injuries in an elderly inner-city population. J Am Geriatr SOC1990;38:1326. 6. Committee on Injury Scaling, The Abbreviated Injury Scale. Morton
Grove, I L American Association of Automotive Medicine, 1980. 7. Kleinbaurn DG, Kupper lL, Morgenstem H. Epidemiologic Research Principles and Quantitative Methods. Belmont, C A Wadsworth, Inc., 1982. 8. Lavizzo-Mourey R, Day SC, Diserens D, Grisso JA. Practicing Prevention for the Elderly. Functional Status: An Approach to Tertiary Prevention. Philadelphia: Hanley & Belfus, Inc., 1989, p 141. 9. Sattin RW, Lambert Huber DA, DeVito CA et al. The incidence of fall injury events among the elderly in a defined population. Am J Epidemiol 1990;131:1028.
lower extremities. Having someone present at the time of the fall was associated with a lower risk of poor recovery. Fortyone percent of the 70 persons interviewed a second time reported continued pain or restriction in usual activities a median of 7 months after the fall occurred. However, only 7% and 39% had received the services of a home health aide or physical therapist, respectively, and only 14% reported that a physician or other health professional had been particularly helpful since the fall had occurred. Conclusions: A large proportion of elderly African-American persons treated at emergency departments for falls experience continued pain and restriction of activities after the fall. Many individuals have not recovered 7 months or longer after the fall and few persons report that a physician or other health professional has been particularly helpful since the fall occurred. We suggest that follow-up programs be developed for elderly persons in minority communities who come to emergency departments after a fall. J Am Geriatr SOC40:673678,1992
alls, a leading cause of death among the elderly, commonly occur among older persons living in the community.’ Although many believe that falls lead to disability and institutionalization, few prospective studies have been conducted to assess the impact of falls on subsequent function, and these studies have focused on white, often middle-class, No community-based studies of the sequelae of falls in inner-city minority populations have been reported in the literature to date. These were among the considerations that prompted us to establish the Philadelphia Injury Prevention Program to study injuries that occur in an indigent minority community. The program involves active community-based surveillance of all injuries that occur in a geographically defined population and are treated at emergency departments, an in-depth study of falls among the elderly, and the development of a comprehensive injury prevention program. In this report, we present the findings of the in-depth study of falls. This study evaluates the immediate sequelae of falls in an elderly minority population as well as recovery status a median of 2 months and 7 months after the fall occurred. Predictors of poor recovery are identified, and use of health services is described.
METHODS In this prospective cohort study, we collected data on elderly fallers at three points in time. First, medical record data at the time of the emergency department visit was abstracted as part of the ongoing surveillance system described below. Second, home interviews were conducted, on average, 8 weeks after the fall occurred. And finally, second interviews were conducted a median of 7 months after the fall with individuals who, at the time of the first interview, reported continued pain or restriction in their usual activities as a result of the fall. Each component of the cohort study is described below. Surveillance The Philadelphia Injury Prevention Program is a cooperative effort of the Philadelphia Department of Public Health and the University of Pennsylvania to evaluate fatal and non-fatal injuries in an urban African-American population of 68,103 people residing in 17 census tracts in western Philadelphia. This population is 97.2% African-American and poor, with a median family income in 1986 of $11,810. An active surveillance system employs trained medical abstractors to obtain information on injuries that occur among residents of the targeted community. Data are collected from 11 emergency departments, the Office of the Medical Examiner, and the Bureau of Vital Statistics. The methods for this ongoing, population-based active surveillance system have been described in detail e l ~ e w h e r eBriefly, .~ a multistage assessment was used to select emergency departments for surveillance and to evaluate the completeness of case ascertainment.
F
* From the Clinical Epidemiology Unit, Section of General Internal Medicine, University of Pennsylvania; tchildren’s Hospital of Philadelphia; and SHahnemann University, Philadelphia, Pennsylvania. Supported by a grant for the Division of Injury Epidemiology and Control. Centers for Disease Control. Correspondence to Jeane Ann Grisso, MD, MSc, Clinical Epidemiology Unit, 317R Nursing Education Building/6095, Philadelphia, PA 19104.
JAGS 40.673-678, 1992
0 1992 by the American Geriatrics Society
0002-8614/92/$3,50
674
GRISSO ET AL
This assessment involved (1) a 20-hospital evaluation of the utilization of inpatient services by census tract of residence; (2) a population-based door-to-door survey of 532 community residents to ascertain which hospitals are used for care of injuries; (3) a systematic evaluation of fire department, school, and emergency medical transport records; and (4) a survey of all private practitioners, clinics, and community health centers in West Philadelphia in which we learned that fewer than 5% of acute injuries were treated in non-emergency settings. On the basis of these results, the final sample of institutions for emergency room surveillance included 11 institutions: three regional trauma centers, three general hospitals, three regional burn centers, one children's hospital, and one eye hospital. Information on deaths was also collected from the vital statistics death certificate files and from the Office of the Medical Examiner. Case Identification and Recruitment Cases were identified from the surveillance system. All those aged 65 years and older who were seen because of a fall as defined by the International Classification of Diseases, Ninth Revision, External Causes of Injury Codes, E880E888, E817, E818, E824, E829 were included. During the period from September, 1987 to November, 1988, 299 community-dwelling African-American persons were identified. Of these, 11 denied that the injury had occurred, six could not be traced, and 67 refused to participate in the study (Figure 1). Those who refused to participate did not differ from participants in age, gender, fracture or hospitalization rate, or injury severity score (AIS). Interviews were completed with 200 persons; 175 interviews were with the injured person and 25 were with proxy respondents who were interviewed either because the injured person had died (5 cases) or be-
r---->
299 Fell
102 Excluded
VI
r---> r---> r--->
197 First Interviewed 113 Recovered
VI
84 Poor Recovery' 14 Second Interview Not Possible
VI
70 Second Interview 41 Recovered
VI
29 Poor Recovery •
FIGURE 1. Poor recovery (continued pain, restriction in usual activities, or both)" among elderly fallers at two time periods following the fall.
JAGS-JULY 1992-VOL. 40, NO.7
cause the person was cognitively impaired (20 cases). Proxy respondents were obtained for 25 of the 40 persons eligible for proxy interview. One interview was incomplete and two others (both proxy interviews) did not provide adequate information to categorize the recovery of the injured persons. Thus, the results are based on 197 persons. Data from Emergency Department Medical Records Information indicating whether the person was hospitalized or discharged home was abstracted from the emergency department medical record. The presence and location of fractures was also recorded. Injury severity was classified according to the Abbreviated Injury Scale (AIS).6 AIS levels are based on location of injury and whether a fracture occurred. In our sample, an AIS score of 3 or 4 included those individuals who had major trauma, while a level of 2 included individuals who had long bone fractures (eg. femur, humerus), and level 1 included minor fracture (eg, finger, toe, etc., or soft tissue injures such as bruises.) First Interview Trained interviewers conducted the interview in the home. Hospitalized patients were interviewed after discharge either in the home or a nursing home. The questionnaire was a standardized, close-ended instrument that included questions about the circumstances of the fall, sociodemographic information, number of chronic illnesses, self-reproted visual and hearing impairment, pre-fall and post-fall functional status, household composition, whether the respondent was hospitalized, and the respondent's perception of the duration of pain and whether he or she had to "cut down" his or her usual activities. The prefall and post-fall assessments of activities of daily living included questions on whether the participant had difficulty with any of the following tasks: walking across a room, walking a block or more, going up or down stairs, standing for a long period, and bending and picking up a shoe from the floor. In the post-fall questions, if the participant indicated having difficulty with a task he or she was asked if it was due to the fall. Second Interview Second interviews were conducted with those who reported continued pain or limitation in function as a result of the fall at the time of the first interview. Follow-up interviews were carried out with 70 of 84 eligible respondents a median of 7.4 months after the fall occurred (Figure 1). Four persons had died in the interim period, four refused the second interview, and six could not be traced. At the time of the second interview, respondents were asked about their current level of function and pain, what health services had been used in the period since the first interview, and what assistance had been received from family or others. Recovery from the Fall The assessment of poor recovery was based on whether respondents reported continued pain, whether they reported that they continued to cut down on their usual activities, or both. When respondents reported either a limitation in general function or continued pain, they were asked whether this was because of the fall. Respondents were
]AGS-JLILY 1992-VOL.40, NO. 7
RECOVERY FROM FALLS IN AN URBAN COMMUNITY
classified as poorly recovered only if they attributed the limitation in function or continued pain to the fall. Statistical Analysis Frequency distributions of respondents who had and had not recovered were compared using chi-square tests and Fisher’s exact test for categorical variables and student’s t test for continuous measures. Models were constructed to assess predictors of poor recovery. Odds ratios with 95% confidence intervals were the primary measures of association. Multiple logistic regression models were constructed to assess the independent contribution of predictors of poor recovery while controlling for potential confound e r ~Predictor .~ variables with P values of 0.15 or less were initially retained. These variables were evaluated in subsequent logistic regression models and only those with P values of 0.05 or less were included in the final models. Models were constructed to identify predictors of poor recovery at the time of the first interview and on the subset who were interviewed a second time to identify predictors of those who still had pain or functional limitations as a result of the fall. All models included age and duration of time from the fall to the interview as possible confounding variables. Secondorder interactions were also assessed for predictor variables and duration of time from the fall to the interview. The study was approved by the Institutional Review Boards of each participating hospital. Written informed consent was obtained from each respondent.
RESULTS Circumstances of the Falls The most common time of day for falls was in the morning (40%), followed by afternoon (25%). Most falls occurred indoors (72%), usually in the respondent’s own home (65%). Of those falls that occurred indoors, the bedroom was the most common site (24%), followed by the stairs (23%). Most falls occurred at a standing height (49%), 29% involved falling from a greater than standing height (including falling down stairs), and 21% occurred while sitting, lying, or bending over. A tripping or slipping hazard was identified in 30% of falls, most commonly an irregular or wet surface or clutter, like clothes on the floor. Most persons (59%) got up within 5 minutes after the fall occurred; a few (6 persons) lay for 3 hours or longer (range, 3 to 13).
675
Immediate Impact of Fall Of the 197 persons, 61 (31%) were hospitalized for injuries caused by the fall. Hospitalizations ranged from 1 to 192 days; the median duration was 8 days. Five persons were discharged to nursing homes. Fractures resulted from 58 falls (29%); the most common fracture sites were hip (16 cases), leg or ankle (14 cases), and arm or shoulder (8 cases). The remaining 20 fractures occurred in eight other sites. Seventeen persons lost consciousness as a result of the fall, and three others were unsure whether the loss of consciousness preceded or followed the fall. The distribution of AIS scores were as follows: 5% suffered no apparent injury, 61% sustained a minor injury (AIS = l), 23% sustained a moderate injury, usually a fracture of a femur (AIS = 2), and 11%suffered major trauma (AIS = 3 or 4). Recovery following the Fall at the Time of the First Interview The initial interviews were conducted a median of 8 weeks after the fall (range, 2.4 to 35). At the time of the first interview, 43% of participants reported that they had continued pain or limitation in their usual activities as a result of the fall. Table 1 shows the impact of the fall at the time of the first and second interviews. At the time of the first interview, the most common activities with which participants reported difficulty because of the fall were going up or down stairs (37%),walking a block or more (25%), walking across a room (22%), standing for long periods (22%), and bending over (20%). Fear of falling was reported by 46%; most (63 persons) said the fear arose as a result of the fall. Predictors of Poor Short-Term Recovery We evaluated predictors of poor recovery based on self reports of continued limitation in usual activities, continued pain, or both. Persons who had not recovered were less likely to be married and slightly more likely to be females. There were no differences in age, duration of education, or median monthly income. The models for predictors of poor recovery included sociodemographic factors (age, gender, income, education), medical factors, previous functional status, severity of trauma, and circumstances of the fall. All models controlled for age and duration of time from the fall to the first interview as potential confounding variables. The odds ratios and 95% confidence limits of risk
TABLE 1. IMPACT OF FALLS OF ELDERLY PERSONS AT TWO TIME PERIODS AFTER THE FALL OCCURRED First Interview* Second Interview** n = 70 n = 197 Continued pain Restricted usual activity Difficulty with: going up or down stairs walking one block or more walking across a room standing for long periods bending over Fear of falling
59 (30%) 39 (20%)
25 (36%) 24 (34%)
73 (37%) 49 (25%) 43 (22%) 43 (22%) 39 (20%) 91 (46%)
19 (27%) 16 (23%) 13 (19%) 15 (21%) 12 (17%) 44 (63%)
* The first interviews were conducted a median of 8 weeks after the fall occurred. ** The second interviews were restricted to a sample of 7 0 persons who reported continued pain and/or restricted activity at the time of the first interview. These interviews were conducted a median of 7.4 months after the fall occurred.
676
GRISSO ET AL
JAGS-JULY 1992-VOL.40, NO. 7
TABLE 2. PREDICTORS OF POOR RECOVERY FROM FALLS IN AN INNER-CITY ELDERLY AFRICAN-AMERICANPOPULATION* Risk Factors Odds Ratios (95% Confidence Limits)
Grandchildren living in household Hearing impairment Leg injury Severity of injury (AIS score) No injury AIS = 1 AIS = 2 AIS = 3 or 4
Previous difficulty in Activities of Daily Living** None One Two
Three or four H e b uresent at fall
3.3
1.1-9.5
2.6 2.4
1.3-5.5 1.O 1-5.7
1.0 (reference) 2.5 5.2 10.2
1.0 (reference) 0.3 0.5 0.3 0.5
0.3-25.6 0.5-50.4 0.9-113.5
0.1-0.8 0.2-1.4 0.1-0.6 0.3-1.0
* Logistic regression model based on factors listed above while controlling for age and duration from the time offall to the time of interview conducted a median of 8 weeks after the fall occurred. Poor recovery is defined as continued pain, limitation in usual activity, or both. **Activities of daily living include difficulty w i t h going up or down stairs, walking one block or more, walking across a room, standing for long periods, and bending over.
factors for poor recovery are shown in Table 2. The major risk factors for disability are the presence of grandchildren in the home, hearing impairment, leg injury and the severity of the injury. The presence of someone to help the respondent at the time of the fall was protective. Previous limitations in activities of daily living (ADL) also appeared protective. This may be due to respondents with previous disabilities being less likely to attribute limitations to the fall. When the same model was repeated except that persons were characterized as poorly recovered who reported continued limitations in function or current pain regardless of whether they attributed it to the fall, the effect of previous limitations in ADL reversed. In this model those with previous limitations in ADL were more likely to experience continued pain or limitations after the fall than were those without previous limitations in ADL. The effect of grandchildren living in the home was explored further. No relationship existed between previous disability or the number of chronic illnesses and the likelihood of living with grandchildren. We stratified respondents by sex to assess whether the presence of grandchildren was equally associated with poor recovery in men and in women. We found no association among men (P = 0.81) but a statistically significant effect in women (P = 0.05). Second Interviews: Predictors of Long-Term Recovery We conducted second interviews with 70 of the 84 persons who reported continued pain or restricted activity as a result of the fall at the time of the first interview. The median duration of time from the fall to the second interview was 7.4 months (range, 5.5-10.4). Respondents were asked about their current level of function and pain, what health services had been used in the period since the fall, and what assistance had been received from family or others. Forty-one percent of the 70 persons who had not
recovered at the time of the first interview reported continued pain or restrictions in usual activities as a result of the fall at the time of the second interview (Table 1). Thirty-six percent of respondents continued to have pain; 23% reported that their health remained worse or much worse than before the fall. Of the 24 persons (34%) who had to cut down their usual activities in the previous month, 14 (61%)reported that they had to cut down their usual activities every day. In response to the question as to whether they had to change the way they did things because of the fall, 39% of the 70 respondents answered "yes." The most common activities affected were housework, shopping, and walking. We assessed whether the factors that predicted poor recovery at the time of the first interview were predictive of continued poor recovery at the time of the second interview. Leg injuries (OR = 4.2, 95% C.I. 1.05, 16.7), the presence of grandchildren in the home (OR = 5.1, 95% C.I. 1.2, 21.6), and pre-fall hearing impairment (OR = 4.1, 95% C.I. 1.1, 14.9) were all associated with poor long-term recovery from falls in multivariate models controlling for age, sex, and duration since first interview. Three of the factors (AIS score, having help at the time of the fall, and pre-fall functional impairment) were not associated with longterm poor recovery as assessed at the time of the second interview. Use of Health Services and Non-Medical Services Of the 70 respondents who had continued pain or restricted activity at the time of the first interview, 26% had not seen a physician since the emergency department visit for the fall. Help from a physical therapist was obtained in 27 cases and from a visiting nurse in 19. Five persons received visits from a home health aide and three received service from Meals on Wheels. Fourteen persons (19%)had received help from family or friends, usually with meal prepa-
]AGS-]ULY 1992-VOL. 40,NO. 7
RECOVERY FROM FALLS IN AN URBAN COMMUNITY
ration. In response to the question, “Is there any person or place you feel has been particularly helpful in your recovery?” 27 persons (38%) reported a relative or friend, 10 persons (14%) reported a physician or other health professional, and 28 (39%)reported that no one had been particularly helpful.
DISCUSSION This prospective community-based study documents the extent of pain and functional limitation that results from falls treated in emergency rooms in an inner-city minority population. A large proportion of persons reported continued pain or restriction in very basic activities of daily living. Most falls were reported to have occurred in the home from a standing height. The circumstances of the falls were such that, if they had occurred in younger persons, in all likelihood few persons would have been injured and even fewer would be subsequently disabled. In this elderly cohort, however, rates of self-reported pain and disability remained high 5 to 10 months after the falls had taken place. Several limitations of this study should be noted. Although we believe that the emergency department surveillance system captured information on virtually all injuries that resulted in emergency department treatment in the community, those persons who did not seek emergency department care for a fall were not included in this study. Thus, the results of this study can only be generalized to those who seek emergency care for a fall. The level of disability was not independently assessed. No tests of neuromuscular performance were carried out. We do not believe that the elderly participants would have exaggerated their limitations; many elderly persons may be reluctant to admit to difficulties because of fear of in~titutionalization.~ We cannot be certain that the pain and debilitation reported in this population was caused entirely by the fall. Although we only included those responses in which the participants attributed their pain and functional limitations to the fall, it may be that other conditions contributed both to the risk for falling and to the subsequent pain and restricted activities. However, the number of chronic illnesses did not predict poor recovery, and it appeared that those with previous functional limitations were less likely to attribute subsequent limitations to the fall than those who had not been disabled prior to the fall. Several predictors of poor recovery at the time of the first and second interviews emerged. In this community, the presence of grandchildren in the home is associated with poor recovery. The reasons for this association are not clear. Those who lived with grandchildren did not have higher rates of prior disability or chronic illnesses than those who did not. The association of grandchildren and poor recovery was found in women, not in men. In poor urban African-American communities, many older women are assuming primary care for their children. It may be that women who are primary caretakers of their grandchildren recover less quickly from a fall. However, we do not
677
have information about whether these grandmothers were responsible for the care of their grandchildren. Although general measures of previous disability were not predictive of poor recovery from the fall, hearing impairment was associated with poor recovery. Hearing impairment may result in isolation and slower recovery, although the impact of hearing impairment was independent of household composition and frequency of social contacts. The severity of the injury, and leg injuries in particular, were also associated with poor recovery. We found the association with leg injuries to be independent of whether a fracture occurred. It may mean that the subsequent impairment in mobility due to lower extremity injuries is more devastating to elderly persons than restriction in activities related to upper extremity injury. The impact of falls in this elderly, inner-city, minority community is substantial. No only did a number of persons sustain fractures and require hospitalization, but even more importantly, the ability to carry out basic activities of daily living was restricted in a large proportion of cases. We cannot assess whether the sequelae of falls in this population are different from that in white, middle-class populations. The fracture rates in studies based on all self-reported falls (regardless of whether medical care is sought) are usually 5% to 6%, in contrast to the 28% fracture rate seen in our s t ~ d y .A ~ ,study ~ based on emergency department surveillance in Miami, FL reported that 27% of falls resulted in fracture.” The extent of limitations in function has not been documented in a similar population. In a study involving volunteers. Nevitt et a13 reported that 25% of those who fell reported having to restrict their activities after the fall occurred. In a prospective study of persons aged 75 ears and older living in the community, Tinetti et a1 found that 26% reported curtailment of their activities because of the fear of falling. Both of these studies included all self-reported falls and limited their assessment to the immediate period after the fall occurred. In our study, in which we conducted second interviews with persons who reported continued pain or restricted activity at the time of the first interview, less than half reported having received help from a visiting nurse or physical therapist, and very few (7%) had received visits from a home health aide in the 5 to 7month period after the fall occurred. Only 10 persons reported that a health professional had been particularly helpful in their recovery and 39% said that no person or place had been particularly helpful. Our impression is that many elderly persons in this community who fall, and who subsequently are in pain or disabled, do not receive much in the way of medical or rehabilitative services. We propose that follow-up programs be developed for elderly persons from urban minority communities who come to emergency rooms because of a fall and that these programs be investigated with well designed clinical trials.
Y
REFERENCES 1. Tinetti ME, Speechley M. Prevention of falls among the elderly. N Engl J Med 1989;320:1055.
2. Lucht U. A prospective study of accidental falls and resulting injuries in the home among elderly people. Acta Socio-Med Scand 1971;3:105. 3. Nevitt MC, Cummings SR, Kidd S et al. Risk factors for recurrent nonsyncopal falls. JAMA 1980;261:2663. 4. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319:1701. 5. Grisso JA, Schwarz DF, Wishner AR et al. Injuries in an elderly inner-city population. J Am Geriatr SOC1990;38:1326. 6. Committee on Injury Scaling, The Abbreviated Injury Scale. Morton
Grove, I L American Association of Automotive Medicine, 1980. 7. Kleinbaurn DG, Kupper lL, Morgenstem H. Epidemiologic Research Principles and Quantitative Methods. Belmont, C A Wadsworth, Inc., 1982. 8. Lavizzo-Mourey R, Day SC, Diserens D, Grisso JA. Practicing Prevention for the Elderly. Functional Status: An Approach to Tertiary Prevention. Philadelphia: Hanley & Belfus, Inc., 1989, p 141. 9. Sattin RW, Lambert Huber DA, DeVito CA et al. The incidence of fall injury events among the elderly in a defined population. Am J Epidemiol 1990;131:1028.
