AmericanJocrmalofChpyghtO1991 b y l l w J o h n s H o p k s t s U r a v e r r r t y S c h o d o f ~ a n d ~ ~
Vd. 133. M. 1 1 Printed in U S A .
AQnghtsrEelVEd
Risk Factors Associated with Falls and Injuries among Elderly Institutionalized Persons
Ann H. Myers,' Susan P. Baker,' Mark L. Van Natta,' Helen Abbey,3 and Elizabeth G. Robinson'
A case-amtrd study among 184 matched pairs of patients 65 years of age and dder was undertaken to identrfy risk factors assodated with falls and injuries in a long-term care facility in BaJtimore, Maryland, in 1984-1985. Patients were matched on length of stay. Variables of interest included sociodemography, functional status, medications, and diagnoses. For all levels of care combined, the following factors were associated ( p 5 0.01) with increased falls: being abte to walk (relative odds (RO) = 4.0), age 90 years and dder (RO = 3.8), a history of falling (RO = 5.0), and taking a vasdilator (RO = 3.0). Among the 184 fallers, the diagnosis of dementia (RO = 7.5) or taking a diuretic (RO = 7.2) was positively assodated with injury ( p 5 0.01). In each of the analyses, medications were associated with falls or injuries, suggesting a feasible intervention. The combination of a history of falling, being aMe to walk, and being 90 years of age or older increased the relative odds to 51.9 and could alert clinicians to identify and monitor high-risk elderly persons in need of preventive measures. Am J Epidemiol1991;I 331179-90.
accidental falls; aged; fractures; long term w e ; nursing homes; risk factors; wounds and injuries
Among all types of injuries, falls are the leading cause of death among the elderly (1). Moreover, the morbidity associated with an injury such as a fractured femur poses a serious problem to the patient, family members, health professionals, and the public. In
1980, there were approximately 200,000 persons with hip fractures, with a median age of 79 years. If the incidence rate for hip fractures remains the same, by using US Census Bureau population projections for the year 2000, the annual number of persons sustaining hip fractures will have increased to almost 300,000 (2). Thus, these demoRecewed for putkatkm June 5, 1990, and in final form December 4,1990 graphic changes will increase demands on A b b r e v l a b K=DSCM, Ninth Revision of the Interthe health care system. natimal Uass~frcatmof D ~ s e s eClinical , Mod~ficafion; RO, rehtive odds. The great majority of fractures occur as a ' Deparbnent of Heatth Pdicy and Management, The result of a fall impact. An understanding of Johns Hopkins Unwersity Schod of Hygiene and Publc the risk factors associated with elderly perIimM,Baltimwe, MD. D e m e n t of Epldemrology, The Johns l-bpim M i- sons falling and being injured is needed in versity School of H y g m and Publlc Hem, EhHmwe, order to develop preventive strategies for MD. high-risk groups (3). D e p r h m t of Wtatistics, The Johns Hopktns University Schod of l-lypm and Rrblc H&, EMtimre. In an earlier paper (4), we reported an MD. analysis of medications and diagnoses in Reprint requests to Dr. Ann H. Myers, tnjury Prevention relation to falls in a long-term w e setting. Center, The Johns l-b$uns Universrty School of I-iyglwa In that case-control study report, we cited and Putkc He&, 624 N Broadway, Balhwre, hrFD 21205. This research was s u w by the National lnsbMe the crude (unadjusted) odds ratios for falling, on Agmg (grant 5801 AG04059) and in part by the Centers which were significantly greater for patients fcx Disease Control (grant R49 CCR302486) t o The Johns taking antidepressants, sedative/hypnotics, H q h m s Univwdy Injury Prevenh Center.
1180
Myers etal.
or vasodilators or for patients with osteoarthritis or depression. In this paper, we report the findings of a matched pairs analysis of 184 cases and controls in the same facility and then focus on identifying risk factors for the largest subset, the nursing home population. Having matched on length of stay, we examined sociodemographic and functional variables, medications, and diagnoses to identify risk factors associated with falling. Among fallers, risk factors for injury were also identified. MATERIALS AND METHODS
The study was conducted in a 283-bed long-term care facility in metropolitan Baltimore, Maryland. The facility categorizes and places the patient based on need by care level. There are three levels of care at the facility: nursing home, chronic hospital, and rehabilitation. Nursing home patients had chronic medical conditions requiring nursing and medical therapies. Typically, chronic hospital patients had an acute phase of a chronic illness. The chronic hospital level may be a unique categorization in the State of Maryland. Rehabilitation patients had a higher functional level or rehabilitation potential than either nursing home or chronic hospital patients. Falls were defined as events in which the staff filed an incident report citing a "fall." These included a variety of circumstances in which there was displacement of the resident's body to the floor. They included any fall such as out of bed, out of chair, or while walking or transferring. The outcome of injury was defined to include burn/scald, contusion, laceration, hematoma, sprain/strain, fracture, or dislocation. Each long-term care resident aged 65 years and over who fell between January 1, 1984, and October 31, 1984, was matched solely on the basis of length of stay with a single control who did not fall during the same 10month period. A total of 184 cases were identified from the incident reports completed by the staff at the time of the fall. For cases, the date of the first fall in 1984 was
selected as the date for which pertinent data were abstracted from the medical records. For controls, data were abstracted for a date representing a similar number of days after admission. Data abstracted for both cases and controls included age, sex, visual and ambulatory status, bowel and bladder function, diagnoses, and drugs taken within the preceding 24-hour period. Visual status was a dichotomous variable based on medical record information on corrected vision. "Low vision" was assigned to any subject having bilateral visual acuity of 20/70 or less. Persons who could only count fingers or had only light perception or no light perception were categorized as having low vision status. "Normal vision" status was assigned to subjects whose records indicated no vision problems. Those for whom insufficient data were available were assigned an unknown status (43 percent); this was the case among patients for whom eye examinations had not yet been performed after admission. The "unknown" records were subsequently recoded to normal vision for analysis purposes. Ambulatory status was a dichotomous variable. "Ambulatory" was assigned if the patient was able to walk independently or with staff and/or assistive devices. "Nonambulatory" was assigned if the patient was not able to walk with assistance and was chair or bed bound. Diagnoses for each person, up to a maximum of 10, were obtained from the problem list in the medical record and coded according to the Ninth Revision of the International Classification of Diseases, Clinical Modification (ICD-9-CM) (5). The diagnoses examined in this paper were the nine most commonly found diagnoses among the study group. In order of frequency they were arteriosclerotic cardiovascular disease (ICD9-CM code 429.2), dementia (ICD-9-CM codes 290 and 310), osteoarthritis (ICD-9CM codes 715 and 716), congestive heart failure (ICD-9-CM code 428), cardiac arrhythmias (ICD-9-CM codes 426 and 427), depression (ICD-9-CM code 311), other
Risk Factors for Falls and Injuries among the Elderly
neurotic disorders (ICD-9-CM codes 300309), neoplasms (ICD-9-CM codes 320349), and all central nervous system diagnoses (ICD-9-CM codes 320-349). History of falling in the previous 3 months was determined from incident reports and the medical record. Drugs were grouped into classes according to Drug Information 85 (6), with the exception of aspirin. To gain a clearer picture of the role of the other nonsteroidal antiinflammatory drugs, aspirin was excluded from drug analyses. The eight classes of drugs analyzed, in order of frequency, were diuretics, cardiac drugs (glycosides and antiarrhythmics), vasodilators, sedative/hypnotics (including anxiolytics), tranquilizers, antidepressants, hypotensives, and nonsteroidal antiinflammatory drugs. Data were abstracted for drugs administered in the 24-hour period prior to the fall. For controls, data on drugs were abstracted along with other variables as described above. Sedative/hypnotics were further categorized by long elimination half-life (>24 hours) or short elimination half-life (
Vd. 133. M. 1 1 Printed in U S A .
AQnghtsrEelVEd
Risk Factors Associated with Falls and Injuries among Elderly Institutionalized Persons
Ann H. Myers,' Susan P. Baker,' Mark L. Van Natta,' Helen Abbey,3 and Elizabeth G. Robinson'
A case-amtrd study among 184 matched pairs of patients 65 years of age and dder was undertaken to identrfy risk factors assodated with falls and injuries in a long-term care facility in BaJtimore, Maryland, in 1984-1985. Patients were matched on length of stay. Variables of interest included sociodemography, functional status, medications, and diagnoses. For all levels of care combined, the following factors were associated ( p 5 0.01) with increased falls: being abte to walk (relative odds (RO) = 4.0), age 90 years and dder (RO = 3.8), a history of falling (RO = 5.0), and taking a vasdilator (RO = 3.0). Among the 184 fallers, the diagnosis of dementia (RO = 7.5) or taking a diuretic (RO = 7.2) was positively assodated with injury ( p 5 0.01). In each of the analyses, medications were associated with falls or injuries, suggesting a feasible intervention. The combination of a history of falling, being aMe to walk, and being 90 years of age or older increased the relative odds to 51.9 and could alert clinicians to identify and monitor high-risk elderly persons in need of preventive measures. Am J Epidemiol1991;I 331179-90.
accidental falls; aged; fractures; long term w e ; nursing homes; risk factors; wounds and injuries
Among all types of injuries, falls are the leading cause of death among the elderly (1). Moreover, the morbidity associated with an injury such as a fractured femur poses a serious problem to the patient, family members, health professionals, and the public. In
1980, there were approximately 200,000 persons with hip fractures, with a median age of 79 years. If the incidence rate for hip fractures remains the same, by using US Census Bureau population projections for the year 2000, the annual number of persons sustaining hip fractures will have increased to almost 300,000 (2). Thus, these demoRecewed for putkatkm June 5, 1990, and in final form December 4,1990 graphic changes will increase demands on A b b r e v l a b K=DSCM, Ninth Revision of the Interthe health care system. natimal Uass~frcatmof D ~ s e s eClinical , Mod~ficafion; RO, rehtive odds. The great majority of fractures occur as a ' Deparbnent of Heatth Pdicy and Management, The result of a fall impact. An understanding of Johns Hopkins Unwersity Schod of Hygiene and Publc the risk factors associated with elderly perIimM,Baltimwe, MD. D e m e n t of Epldemrology, The Johns l-bpim M i- sons falling and being injured is needed in versity School of H y g m and Publlc Hem, EhHmwe, order to develop preventive strategies for MD. high-risk groups (3). D e p r h m t of Wtatistics, The Johns Hopktns University Schod of l-lypm and Rrblc H&, EMtimre. In an earlier paper (4), we reported an MD. analysis of medications and diagnoses in Reprint requests to Dr. Ann H. Myers, tnjury Prevention relation to falls in a long-term w e setting. Center, The Johns l-b$uns Universrty School of I-iyglwa In that case-control study report, we cited and Putkc He&, 624 N Broadway, Balhwre, hrFD 21205. This research was s u w by the National lnsbMe the crude (unadjusted) odds ratios for falling, on Agmg (grant 5801 AG04059) and in part by the Centers which were significantly greater for patients fcx Disease Control (grant R49 CCR302486) t o The Johns taking antidepressants, sedative/hypnotics, H q h m s Univwdy Injury Prevenh Center.
1180
Myers etal.
or vasodilators or for patients with osteoarthritis or depression. In this paper, we report the findings of a matched pairs analysis of 184 cases and controls in the same facility and then focus on identifying risk factors for the largest subset, the nursing home population. Having matched on length of stay, we examined sociodemographic and functional variables, medications, and diagnoses to identify risk factors associated with falling. Among fallers, risk factors for injury were also identified. MATERIALS AND METHODS
The study was conducted in a 283-bed long-term care facility in metropolitan Baltimore, Maryland. The facility categorizes and places the patient based on need by care level. There are three levels of care at the facility: nursing home, chronic hospital, and rehabilitation. Nursing home patients had chronic medical conditions requiring nursing and medical therapies. Typically, chronic hospital patients had an acute phase of a chronic illness. The chronic hospital level may be a unique categorization in the State of Maryland. Rehabilitation patients had a higher functional level or rehabilitation potential than either nursing home or chronic hospital patients. Falls were defined as events in which the staff filed an incident report citing a "fall." These included a variety of circumstances in which there was displacement of the resident's body to the floor. They included any fall such as out of bed, out of chair, or while walking or transferring. The outcome of injury was defined to include burn/scald, contusion, laceration, hematoma, sprain/strain, fracture, or dislocation. Each long-term care resident aged 65 years and over who fell between January 1, 1984, and October 31, 1984, was matched solely on the basis of length of stay with a single control who did not fall during the same 10month period. A total of 184 cases were identified from the incident reports completed by the staff at the time of the fall. For cases, the date of the first fall in 1984 was
selected as the date for which pertinent data were abstracted from the medical records. For controls, data were abstracted for a date representing a similar number of days after admission. Data abstracted for both cases and controls included age, sex, visual and ambulatory status, bowel and bladder function, diagnoses, and drugs taken within the preceding 24-hour period. Visual status was a dichotomous variable based on medical record information on corrected vision. "Low vision" was assigned to any subject having bilateral visual acuity of 20/70 or less. Persons who could only count fingers or had only light perception or no light perception were categorized as having low vision status. "Normal vision" status was assigned to subjects whose records indicated no vision problems. Those for whom insufficient data were available were assigned an unknown status (43 percent); this was the case among patients for whom eye examinations had not yet been performed after admission. The "unknown" records were subsequently recoded to normal vision for analysis purposes. Ambulatory status was a dichotomous variable. "Ambulatory" was assigned if the patient was able to walk independently or with staff and/or assistive devices. "Nonambulatory" was assigned if the patient was not able to walk with assistance and was chair or bed bound. Diagnoses for each person, up to a maximum of 10, were obtained from the problem list in the medical record and coded according to the Ninth Revision of the International Classification of Diseases, Clinical Modification (ICD-9-CM) (5). The diagnoses examined in this paper were the nine most commonly found diagnoses among the study group. In order of frequency they were arteriosclerotic cardiovascular disease (ICD9-CM code 429.2), dementia (ICD-9-CM codes 290 and 310), osteoarthritis (ICD-9CM codes 715 and 716), congestive heart failure (ICD-9-CM code 428), cardiac arrhythmias (ICD-9-CM codes 426 and 427), depression (ICD-9-CM code 311), other
Risk Factors for Falls and Injuries among the Elderly
neurotic disorders (ICD-9-CM codes 300309), neoplasms (ICD-9-CM codes 320349), and all central nervous system diagnoses (ICD-9-CM codes 320-349). History of falling in the previous 3 months was determined from incident reports and the medical record. Drugs were grouped into classes according to Drug Information 85 (6), with the exception of aspirin. To gain a clearer picture of the role of the other nonsteroidal antiinflammatory drugs, aspirin was excluded from drug analyses. The eight classes of drugs analyzed, in order of frequency, were diuretics, cardiac drugs (glycosides and antiarrhythmics), vasodilators, sedative/hypnotics (including anxiolytics), tranquilizers, antidepressants, hypotensives, and nonsteroidal antiinflammatory drugs. Data were abstracted for drugs administered in the 24-hour period prior to the fall. For controls, data on drugs were abstracted along with other variables as described above. Sedative/hypnotics were further categorized by long elimination half-life (>24 hours) or short elimination half-life (
