Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Factors associated with falling in older Adelaide residents

Factors associated with falling in older Adelaide residents Gary R. Andrews Centrefor Ageing Studies, Flinders University o South Australia, Adelaide f Abstract: The aim of this study was to identify characteristics that predispose older residents of Adelaide to falling. Information collected in the baseline phase of the Australian Longitudinal Study of Ageing was used to draw cross-sectional comparisons between participants who reported having fallen on at least one occasion in the previous 12 months and those participants who reported not having fallen. The baseline cohort consisted of 1947 participants aged 70 years or more, of whom 550 (28 per cent) reported having fallen at least once in the previous year. Independent risk factors for falling were: age; having left school at an early age; a worsening of vision in recent years; and histories of Parkinson’s disease, fractured hip, glaucoma, stroke (including transient ischaemic attack), corns or bunions, or arthritis. The findings regarding medical histories suggest some possible opportunities for reducing the risk of falls in the elderly by managing the symptoms and risk factors of underlying conditions such as stroke and loss of vision. (Aust N ZJPublic Health 1997; 21: 462-8) LS are the leading cause of accidental death in Australians aged 65 years and older.’ In this age group, falls accounted for 42 per cent of all injury-related deaths registered in Australia in 1993; numbering 694, falls deaths were more than twice as frequent as transport or suicide deaths. Similarly, data from public hospitals in Victoria show that falls are the single most common cause of injury-related hospital admission for persons aged 65 years and over; over the period 1987 to 1993, falls accounted for 42 per cent of all injury-related admissions in Victoria (in the 65-and-over age group) and outnumbered transport accident admissions by a factor of nearly lo.* Fractures, particularly of the hip, are the main reason for hospital admission following a fall; many such fractures are associated with osteoporo~is.~,’ E“ health status. Other consequences of falls include the effects of any minor injuries that are sustained and the accompanying loss of confidence and incapacitating fear of falling which often follows in older The frequency of falling among older people and its serious sequelae have been recognised by health Additionally, older people suffer numerous falls that do not result in hospital admission or death. While accurate population-based numbers are not available, many Australian and overseas researchers have documented the frequency of falling among their study participants.”I3 The prevalence of having at least one fall in a year ranges from around 30 to 60 per cent of study participants. In general, the reported prevalence tends to be higher in prospective studies employing good means of fall ascertainment and in studies of older and frailer participants. It is evident, therefore, that a considerable proportion of the elderly population experiences falls. While most of the falls do not lead to serious injury or death, they are a cause for concern as likely markers of proneness to future falling, perhaps with worse outcomes, and as indicators of deteriorating Correspondence to Dr James Harrison, AIHW National Injury Surveillance Unit, Flinders University of South Australia, Bedford Park, SA 5042. Fax (08) 8374 0702. authorities in Australia, and national goals and targets have been set with the aim of reducing levels of falls-related mortality and hospital admission.” Achieving reductions in falls-related mortality and morbidity can be tackled by various approaches, including intervening to prevent falls occurring in the first place and providing better clinical management in cases where a fall has occurred and resulted in injury. Successful intervention, with efficacious targeting of at-risk persons, requires a better understanding of the causes of falls than is presently available. There exists a broad body of literature on risk factors for falling (for an extensive bibliography see the literature review by the Fall Prevention Research Group in New Zealandla). Suggested risk factors include poor state of health, limited mobility and fitness, hazards in the home environment, dizziness and disorders of balance, use of medications (especially psychotropic ones), and a history of falling.’a.lY However, there are many contradictory findings among studies over the predictive value of a wide range of putative medical and environmental risk factors. In part, this may be because studies of relatively frequent health problems in elderly people are prone to methodological difficulties, such as poor recall of events, low participation rates, and incomplete follow-up owing to illness or death.zoAnother reason for discrepant findings is the use of inadequate study designs to identify risk factors: common problems that have been recognised include not using a comparison group, and incorrectly ascribing NO. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1997 vot. 21 F FACTORS ASSOCIATED WITH FALLING causality to associations between variables whose temporal relation is not established.z&zz While several Australian studies have examined correlates of falling in small samples of institutionalised older s ~ b j e c t only two large-scale studies s ~ ~ ~ ~ ~ ~ ~ of community dwelling people have been undertaken, notably the Randwick Falls and Fractures StudyZ5sz6 and the Dubbo Osteoporosis Epidemiology Study." These two studies found that persons who experienced multiple falls tended to be identifiable by their relatively poor performance in sensorimotor and balance tests. Given the uncertain nature of the existing international literature and the relatively small number of Australian studies of community dwelling populations, a need exists for further epidemiological study of risk factors for falling within an Australian community setting. The Australian Longitudinal Study of Ageing being conducted by the Centre for Ageing Studies at the Flinders University of South Australia affords a sound opportunity to explore the correlates of falling in a sizeable cohort of older Adelaide residents, selected from both community and institutional residences.28 This paper examines factors associated with falling by drawing cross-sectional comparisons between fallers and nonfallers, using the baseline characteristics and experiences of the cohort. Method This study of factors associated with falling is part of the Australian Longitudinal Study of Ageing, details of which are available elsewhere.2sBriefly, its aim is to gain a better understanding of the determinants of healthy aging and the role of social, environmental and biomedical factors in age-related changes in the health of older people. The baseline cohort comprised 1947 participants aged at least 70 years, resident in Adelaide, South Australia (living either in the community or in an institution), and selected by age- and sex-stratified random sampling from the electoral roll. The level of participation at baseline represented a recruitment rate of 54 per cent. Baseline data collection started in September 1992 and was completed in March 1993. Participants completed a comprehensive face-to-face interview and underwent a series of home-based clinical assessments. The survey collected information on a range of sociodemographic, physiological and medical items. Basic anthropometric measures included height, weight, blood pressure, and assessment of gait. Data were collected by trained interviewers and collated by the Centre for Ageing Studies. Participants were asked to report on whether any falls had been experienced in the previous 12 months, and if so, to describe the circumstances that led to their occurrence. Using this information, we classified the cohort into comparison groups of fallers and nonfallers. The criterion for classification as a faller was one or more self-reported falls in the previous 12 months; those who reported no falls were classified as nonfallers. Information on medical histories was obtained by asking participants to identify from a list of 39 spec- ified conditions those that they had ever suffered from. For each identified condition, participants were asked for the year of diagnosis and whether they presently suffered from the condition. Additionally, there was opportunity for participants to report up to four other conditions not listed. Those conditions diagnosed within the 12 months prior to the baseline interview (0.2 per cent of all reported medical histories) were excluded from analyses because the onset of the condition might have been after the occurrence of a reported fall. Similarly, other potential risk factors were examined only if the exposure could not vary with time (for example, sex, country of birth) or occurred prior to the 12-month reporting period for falls. Consequently, a number of potentially important measures taken at the baseline stage of the study (including blood pressure, weight, gait analysis, ability to perform activities of daily living, medication usage, and psychological functioning) were not analysed because their values might have reflected the effects of a precedent fall and this would give rise to misleading associations. The role of such factors will be examined prospectively in subsequent stages of the study. Odds ratios and their 95 per cent confidence intervals were used as measures of the strength and significance of associations between falling and possible risk factors. Age-adjusted odds ratios were calculated from stratified analysis of unweighted data using Epi Info oftw ware.'^ Multivariate unconditional logistic regression analysis of unweighted data was undertaken in accordance with the hierarchical backward elimination procedure recommended by K l e i n b a ~ mthe~SPSS computer program was used ;~ for this p ~ r p o s e . ~ ' Results The baseline cohort consisted of 1947 participants, 1039 (53 per cent) men and 908 (47 per cent) women. The age distribution for the two sexes is shown in Table 1. Sixty-nine per cent of the cohort was born in Australia and most cohort members lived in the community, with only 8 per cent living in institutional accommodation. Two-thirds of the participants were known to be living with others at their usual place of residence. A total of 550 (28 per cent) participants reported having fallen on at least one occasion in the previous 12 months (Table 1). Falls were more frequent among women than men, with 31 per cent of women compared to 26 per cent of men reporting at least one fall. In both sexes the proportion of fallers increased as age increased. Age-adjusted analysis Medical histories associated with falling. Thirty-five medical conditions were examined for possible association with falling. No adjustment was made to the significance criterion for multiple comparisons, in accordance with recent views on this matter.3z,33 Medical histories associated with falling in ageadjusted analyses are presented in Table 2. The histories most strongly associated with falling in this VOL. AUSTRALIAN AND NEW ZEAIAND JOURNAL O PUBLIC HEALTH 1997 F 21 NO. 5 A63 Table 1: Falls experience of the study cohort of the Australian Longitudinal Study of Ageing, related to its age and sex composition Number of reported falls in the previous year Age group (years) Number of and sex participants Men o m /o 22 /o Missing o n /o %‘ 0 0 0 0 0 70-74 75-79 80-84 285 Total Women 70-74 75-79 80-84 285 Total Total Note: (a) Row percentages. cohort were Parkinson’s disease and hip fracture. Among conditions not associated with falling were: hypertension, heart attack, migraine, and chronic bronchitis. Physical factors and falling: The associations between falling and a number of physical and health-related factors are presented in Table 3. Increasing age was associated with an increased risk of falling; compared with participants aged 70 to 74, those aged 85 and over had twice the odds of falling. Women appeared to be at greater risk than men. The relative odds of falling also increased with the number of morbid conditions ever suffered and with worsening vision. A history of ever having smoked regularly appeared to decrease the risk of falling. Socioeconomic factors and falling: The age of leaving school was the only socioeconomic variable associated with falling (Table 4). While total income from all sources (including that of a spouse) was not Table 2: Medical histories associated with falling in the study cohort of the Australian Longitudinal Study of Ageing %of %of fallers nonfallers n=550 n= 139 1 statistically significant as a predictor, there was an inverse relationship between total income and the relative odds of falling. Multivariate analysis We specified an initial unconditional logistic regression model containing all the significant risk factors Table 3 Physical health factors and adds of falling for the study cohort of the Australian Longitudinal Study of Ageing ~~ Factor Age group (years) %of fallers nonfallers ~ 5 5 0 1 3 9 1 OR‘ ~ %of CI referent 0.89 to 1.59 1.05 to 1.90 1.59 to 2.82 referent 1.13 to 1.70 referent 70-74 75-79 8C-84 a85 Sex Male Female No. morbidities ever 0-2 3-5 6-8 29 1 .oo 1.19 1.41 2.1 1 1 .oo 1.39 1 .oo 1.35 1.73 1.92 Disease or medical condition Parkinson’s disease Fractured hip Anaemia Stroke or transient ischaemic attack Glaucoma Arthritis Corns or bunions O R CI 1.27 to 6.83 1.26 to 3.72 1.07 to 2.1 7 1.1 1 to 2.02 1.05 to 2.15 1.21 to 1.84 1.15 to 1.79 1 .OO to 1.84 1.25 to 2.38 1.31 to 2.82 referent Ever smoked regularly No Yes 1.oo 0.72 0.58 to 0.89 Vision worsening in past 5 years No Yes Notes: (a) O = ratio of odds of falling for those with a particular medical history R 1.OO 1.47 referent 1.19 to 1.82 (b) to odds of falling far those without that history, adiusted far age with the Mantel-Hoenszel technique. C = 95% confidence interval. I Note: (a) O = adds ratio adjusted far age with the Mantel-Haenszel technique. R (b) CI = 95% confidence interval. VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1997 F NO. FACTORS ASSOCIATED WITH FALLING Table 4 Socioeconomic factors and odds of falling for the study cohort of the Australian Longitudinal Study of Ageing %of %of fallers nonfallers Factor Marital status Married or de fact0 59 Widowed, separated or divorced 37 Never married 4 Table 5: Independent predictors of falling for the study cohort of the Australian Longitudinal Study of Ageing n=550 e l 3 9 1 OR" 1.oo CI reFerent Medical history' Parkinson's disease Fractured hip Glaucoma Stroke or transient ischaemic attack Corns or bunions Arthritis 1 .oo 0.97 to 1.54 0.46 to 1.46 referent 0.63 to 1.OO 1 .oo 1.05 to 6.32 1.10 to 3.40 1.13 to 2.37 1.04 to 1.98 1.12 to 1.81 1.08 to 1.69 referent Country of birth Australia Elsewhere Age group (years) Difficulty with English language No Yes 70-74 75-79 80-84 a85 Vision worsening in past 5 years No 1 .oo 1.34 1.02 to 1.95 1.13 to 2.17 1.35 to 2.63 referent 1 .oo 0.88 referent 0.46 to 1.60 Income (from all sources) Yes 1.06 to 1.68 referent Age of leaving school (years) >$30 000 7 $1 2 001 to $30 000 51 $s12 000 42 Age of leaving school 1.oo 1.06 1.24 1 .oo 1.36 1.24 1.54 referent 0.70 to 1.63 0.81 to 1.92 referent a17 15-16 14 <14d 1.oo 1.40 1.25 1.62 1.01 to 1.94 0.89 to 1.74 1.05 to 2.50 217 15-16 14 <14 0.94 to 1.98 0.87 to 1.81 1.02 to 2.34 Notes: (a) OR = odds ratio adjusted for age with the Mantel-Haenszel technique. (b) CI = 95% confidence interval. (c) Or 170 schooling Notes: (a) OR = odds ratio calculated from final logistic regression model, as described in Results, Multivariate analysis. Model-fitting statistics: -2 log likelihood = 1878.9, improvement x2 = 100.9, 19 df, P c 0.0001, (b) CI = 95% Confidence interval. (c) Referent category is no history of the particular medical condition. (d) Or no schooling shown in Tables 2 to 4 as main effects, the remaining factors in Tables 2 to 4 along with various other medical histories as possible confounders, and age and sex as effect modifiers (with corresponding firstorder interaction terms). A final model was obtained by reducing the initial model in accordance with Kleinbaum's hierarchical backward elimination procedure.30 The independent risk factors to emerge from the final model were: history of Parkinson's disease, fractured hip, glaucoma, stroke, corns or bunions, arthritis; increasing age; deteriorating eyesight; and lack of schooling or early age of leaving school (Table 5). Other variables retained in the final model but not independently associated with falling were sex, history of asthma, history of ever having smoked regularly and marital status. The final model also included a statistically significant first-order interaction term for sex by history of X asthma (improvement ' = 4.9, 1 df, P < 0.03). For men, a history of asthma appeared to increase the odds of falling (adjusted odds ratio (OR) 1.56, 95 per cent confidence interval (CI) 0.86 to 2.84), whereas for women it appeared to decrease the odds of falling (OR 0.60, CI 0.33 to 1.08). These sex-specific odds ratios did not achieve statistical significance although the corresponding interaction term made a statistically significant improvement to the logistic regression model. Discussion As this work was part of a larger collaborative project which set out to examine the broad determinants of healthy aging, the nature of information and the level of detail that could be obtained specifically on falls was limited. When considering the results of this study, it is important to recognise the limitations of the methods that were used. The main concerns are: self-report of information on exposure and outcome variables; modest response rate; relevance of antecedent exposures; and aggregation of different types of falls into a single outcome measure. The exposure measures reported on in this paper were all obtained by self-report, and this may have resulted in some misclassification bias. To reduce misclassification bias from underreporting of potentially sensitive topics such as income and medical histories, prompt cards were used during the personal interview. Past exposures may also have been underreported simply because they were not recalled by the elderly participants; this would have produced nondifferential misclassification bias and diminution of effects unless the pattern of recall was associated with falling, in which case spurious associations might have been produced. The potential for misclassification of exposure was perhaps greatest for medical histories. In addition to possible problems with recall such as those already discussed, elderly people might not have been familiar with medical nomenclature or with their actual AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1997 VOL. 21 NO. 5 F DOLlNlS E I AL. diagnoses, particularly if they involved complex patterns of disease or comorbidities. A further possibility is that some medical conditions might have been overreported by those who thought the condition to be a cause for their falling. While this possibility cannot be ruled out, it is expected to have had at most a small effect, since the questionnaire asked for medical histories before it asked about falls experience, and participants did not know in advance that they would be asked questions about falls. Although self-reports have clear limitations, they potentially could be used within a community or primary care setting as a relatively efficient first means of identifying persons at increased risk of falling, with subsequent clinical follow-up to confirm medical diagnoses. Classification of participants into groups of fallers and nonfallers was based on self-reported falls experience, and is therefore subject to recall bias. Studies have shown that while the number of reported falls over a 12-month period is not particularly accurate, recall of having experienced any falls versus zero falls is reasonably reliable.z0,34 Misclassification occurs predominantly in the direction of forgetting falls as opposed to reporting falls that did not occur. The usual effect of this type of misclassification is to bias measures of effect towards the null value. In the current study, the important outcome was whether a fall had occurred rather than the exact number of falls over the 12 months, and hence it is considered unlikely that there was a significant misclassification of outcome. The recruitment rate at baseline (54 per cent) was similar to that achieved in the Randwick Falls and Fractures Studyz6and in the Dubbo Osteoporosis Epidemiology Studyz’ but lower than in other studies of risk factors for falls among the elderly. In part, the lower rate of participation may have been because of the relatively high level of ongoing involvement required of participants over the course of the main study. In view of the participation rate, findings from this study may not be generalisable to all older Australians. A difficulty faced by all researchers who make use of cross-sectional data or current data on exposures combined with historical data on falling is to establish whether the temporal relationship between these measures is consistent with a causal path. In this paper, examination of risk factors was restricted to stable exposures such as the age of leaving school and the country of birth, and exposures that were known to be antecedent to the falls experience about which the participants were questioned (for example, disease history, marital status). Even so, further research will be needed to elucidate the mechanisms by which such past exposures might operate to bring about an increased risk of falling. Different types of falls (for example, those resulting from slips, trips, sudden loss of consciousness) were not distinguished in the analyses, as the level of detail provided by participants was judged to be insufficient to make useful distinctions. It is quite probable that different types of falls have different sets of risk factors, and the lack of risk specificity produced by grouping all falls together is likely to have diminished some associations or possibly even masked them a l t ~ g e t h e r . * ~number of classificaA ,~~ tions for falls have been proposed and with further evaluation may prove useful in revealing aetiological risk factors hitherto obscured.l2 The results of this study indicated that histories of Parkinson’s disease, fractured hip, glaucoma, stroke (including transient ischaemic attack), corns or bunions and arthritis are independent risk factors for falling. A history of anaemia was associated with an increased risk of falling in the age-adjusted analysis but was displaced as a predictor when other factors were controlled for in the multivariate analysis. In age-adjusted analysis, there was a trend for risk of falling to increase with the number of conditions that had ever been suffered but this was not an independent predictor. Other work has suggested that the number of conditions and disabilities a person reports may be a marker of poor health generally and that such a state is associated with an increased risk of falling.I5 Our finding that Parkinson’s disease increased the risk of falling in older people accords with the finding of Nevitt et al.7 The risk may be related to the familiar symptoms of Parkinsonism, which include tremor, muscular rigidity and difficulty in walking. A history of hip fracture approximately doubled the odds of falling. Given that past falling is a predictor of future falling,’*and that falling is the usual mechanism underlying hip fractures, this association was not unexpected. Arthritis has previously been identified as a possible risk factor for falling7 A causal relationship is plausible, in view of the degradation of lower limb function caused by the deterioration of joints in the lower limbs. Likewise, corns and bunions may increase the risk of falling by interfering with normal gait. An appropriate mix of drug, physical and occupational therapies is important to maintain normal functioning in persons affected by these conditions. Problems with vision are known to increase the chances of falling.7~z6 results of this study conThe firm the importance of vision problems in relation to falling. A history of glaucoma and self-identified worsening of vision in the last five years were identified as predictors of falling in the multivariate analysis. Early detection and treatment of glaucoma before it causes loss of vision should be promoted as an important public health issue particularly as the benefits are likely to extend beyond healthy vision to other areas such as reduced injury from falls. The independent association of stroke with the occurrence of falls is consistent with the findings of several other The symptoms of this paralysing disorder include vertigo, difficulty in walking, and visual disturbances, all of which could result in a fall. The risk factors for stroke are essentially the same as those for other cardiovascular diseases, and include hypertension, high cholesterol, smoking and obesity. All of these risk factors have received and are continuing to receive much attenVOL. AUSTRALIAN AND N W ZEALAND JOURNAL O PUBLIC HEALTH 1997 E F NO. FACTORS ASSOCIATED WITH FALLING tion in health promotion and disease prevention efforts. Success in reducing these risk factors may well lead to reductions in injuries from strokerelated falls. Age was an independent predictor of falling in this study and has been found to be so in a number of other Also in agreement with other work was the finding that sex was not independently associated with falling.'0,'' Smoking appeared to be associated with a decreased likelihood of falling in the age-adjusted analysis, but this association was not present when other factors were controlled for in the multivariate analysis. In view of the established association between smoking and cardiovascular diseases including stroke, it would appear unlikely that a history of smoking would confer any protection against falling. Early age at leaving school was the only socioeconomic variable that was an independent risk factor for falling. Total income was not significantly associated with falling although there was an inverse relationship between it and risk of falling in the age-adjusted analysis. It may be that educational attainment (as measured by age at leaving school) is a more sensitive indicator of lifetime economic disadvantage than the current total income of elderly persons.36 Given that socioeconomic factors are important determinants of the distribution of many diseases, the results of this study tentatively suggest that economic factors might play a role in the occurrence of falling. There were no apparent associations between marital status, country of birth or English language ability and the occurrence of falls. The final logistic regression model included a statistically significant interaction term suggesting that sex might be an effect modifier in the relationship between asthma and falling. Men reporting a history of asthma had a statistically nonsignificant higher risk of falling (OR 1.56, CI 0.86 to 2.84) while women reporting a history of asthma had a statistically nonsignificant lower risk (OR 0.60, CI 0.33 to 1.08). Although different falls risk factors for men and women have previously been doc~rnented,'~ our result must be treated with caution because of the nonsignificance of the sex-specific odds ratios. Few epidemiological studies in Australia have addressed the causes of falls among older people living in the community. This study provides information about risk factors for falling which should be of use to researchers and to those who are designing and running trials of falls prevention programs in Australia. The results of this study appear plausible, are generally consistent with previous knowledge, and add to the evidence regarding risk factors for falls among community dwelling older people. A number of disease conditions were identified as likely risk factors, and in the case of stroke and conditions that affect vision, a public health approach to their prevention may also lead to fewer falls among the elderly. Our cross-sectional baseline findings need further confirmation and the longitudinal phases of the Australian Longitudinal Study of Ageing will provide an opportunity to extend this work. Acknowledgments We gratefully acknowledge the work of the staff at the Centre for Ageing Studies, the Flinders University of South Australia, who carried out the Australian Longitudinal Study of Ageing and provided the data for this paper. The Australian Longitudinal Study of Ageing was funded in part by the South Australian Health Commission, the Australian Rotary Health Research Fund, and by a grant from the US National Institutes of Health (grant no. AG 0852302). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

Factors associated with falling in older Adelaide residents

Download PDF
7 pages

Loading next page...
 
/lp/wiley/factors-associated-with-falling-in-older-adelaide-residents-j60p5vgIzs

References (39)

Publisher
Wiley
Copyright
Copyright © 1997 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1467-842X.1997.tb01736.x
Publisher site
See Article on Publisher Site

Abstract

Gary R. Andrews Centrefor Ageing Studies, Flinders University o South Australia, Adelaide f Abstract: The aim of this study was to identify characteristics that predispose older residents of Adelaide to falling. Information collected in the baseline phase of the Australian Longitudinal Study of Ageing was used to draw cross-sectional comparisons between participants who reported having fallen on at least one occasion in the previous 12 months and those participants who reported not having fallen. The baseline cohort consisted of 1947 participants aged 70 years or more, of whom 550 (28 per cent) reported having fallen at least once in the previous year. Independent risk factors for falling were: age; having left school at an early age; a worsening of vision in recent years; and histories of Parkinson’s disease, fractured hip, glaucoma, stroke (including transient ischaemic attack), corns or bunions, or arthritis. The findings regarding medical histories suggest some possible opportunities for reducing the risk of falls in the elderly by managing the symptoms and risk factors of underlying conditions such as stroke and loss of vision. (Aust N ZJPublic Health 1997; 21: 462-8) LS are the leading cause of accidental death in Australians aged 65 years and older.’ In this age group, falls accounted for 42 per cent of all injury-related deaths registered in Australia in 1993; numbering 694, falls deaths were more than twice as frequent as transport or suicide deaths. Similarly, data from public hospitals in Victoria show that falls are the single most common cause of injury-related hospital admission for persons aged 65 years and over; over the period 1987 to 1993, falls accounted for 42 per cent of all injury-related admissions in Victoria (in the 65-and-over age group) and outnumbered transport accident admissions by a factor of nearly lo.* Fractures, particularly of the hip, are the main reason for hospital admission following a fall; many such fractures are associated with osteoporo~is.~,’ E“ health status. Other consequences of falls include the effects of any minor injuries that are sustained and the accompanying loss of confidence and incapacitating fear of falling which often follows in older The frequency of falling among older people and its serious sequelae have been recognised by health Additionally, older people suffer numerous falls that do not result in hospital admission or death. While accurate population-based numbers are not available, many Australian and overseas researchers have documented the frequency of falling among their study participants.”I3 The prevalence of having at least one fall in a year ranges from around 30 to 60 per cent of study participants. In general, the reported prevalence tends to be higher in prospective studies employing good means of fall ascertainment and in studies of older and frailer participants. It is evident, therefore, that a considerable proportion of the elderly population experiences falls. While most of the falls do not lead to serious injury or death, they are a cause for concern as likely markers of proneness to future falling, perhaps with worse outcomes, and as indicators of deteriorating Correspondence to Dr James Harrison, AIHW National Injury Surveillance Unit, Flinders University of South Australia, Bedford Park, SA 5042. Fax (08) 8374 0702. authorities in Australia, and national goals and targets have been set with the aim of reducing levels of falls-related mortality and hospital admission.” Achieving reductions in falls-related mortality and morbidity can be tackled by various approaches, including intervening to prevent falls occurring in the first place and providing better clinical management in cases where a fall has occurred and resulted in injury. Successful intervention, with efficacious targeting of at-risk persons, requires a better understanding of the causes of falls than is presently available. There exists a broad body of literature on risk factors for falling (for an extensive bibliography see the literature review by the Fall Prevention Research Group in New Zealandla). Suggested risk factors include poor state of health, limited mobility and fitness, hazards in the home environment, dizziness and disorders of balance, use of medications (especially psychotropic ones), and a history of falling.’a.lY However, there are many contradictory findings among studies over the predictive value of a wide range of putative medical and environmental risk factors. In part, this may be because studies of relatively frequent health problems in elderly people are prone to methodological difficulties, such as poor recall of events, low participation rates, and incomplete follow-up owing to illness or death.zoAnother reason for discrepant findings is the use of inadequate study designs to identify risk factors: common problems that have been recognised include not using a comparison group, and incorrectly ascribing NO. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1997 vot. 21 F FACTORS ASSOCIATED WITH FALLING causality to associations between variables whose temporal relation is not established.z&zz While several Australian studies have examined correlates of falling in small samples of institutionalised older s ~ b j e c t only two large-scale studies s ~ ~ ~ ~ ~ ~ ~ of community dwelling people have been undertaken, notably the Randwick Falls and Fractures StudyZ5sz6 and the Dubbo Osteoporosis Epidemiology Study." These two studies found that persons who experienced multiple falls tended to be identifiable by their relatively poor performance in sensorimotor and balance tests. Given the uncertain nature of the existing international literature and the relatively small number of Australian studies of community dwelling populations, a need exists for further epidemiological study of risk factors for falling within an Australian community setting. The Australian Longitudinal Study of Ageing being conducted by the Centre for Ageing Studies at the Flinders University of South Australia affords a sound opportunity to explore the correlates of falling in a sizeable cohort of older Adelaide residents, selected from both community and institutional residences.28 This paper examines factors associated with falling by drawing cross-sectional comparisons between fallers and nonfallers, using the baseline characteristics and experiences of the cohort. Method This study of factors associated with falling is part of the Australian Longitudinal Study of Ageing, details of which are available elsewhere.2sBriefly, its aim is to gain a better understanding of the determinants of healthy aging and the role of social, environmental and biomedical factors in age-related changes in the health of older people. The baseline cohort comprised 1947 participants aged at least 70 years, resident in Adelaide, South Australia (living either in the community or in an institution), and selected by age- and sex-stratified random sampling from the electoral roll. The level of participation at baseline represented a recruitment rate of 54 per cent. Baseline data collection started in September 1992 and was completed in March 1993. Participants completed a comprehensive face-to-face interview and underwent a series of home-based clinical assessments. The survey collected information on a range of sociodemographic, physiological and medical items. Basic anthropometric measures included height, weight, blood pressure, and assessment of gait. Data were collected by trained interviewers and collated by the Centre for Ageing Studies. Participants were asked to report on whether any falls had been experienced in the previous 12 months, and if so, to describe the circumstances that led to their occurrence. Using this information, we classified the cohort into comparison groups of fallers and nonfallers. The criterion for classification as a faller was one or more self-reported falls in the previous 12 months; those who reported no falls were classified as nonfallers. Information on medical histories was obtained by asking participants to identify from a list of 39 spec- ified conditions those that they had ever suffered from. For each identified condition, participants were asked for the year of diagnosis and whether they presently suffered from the condition. Additionally, there was opportunity for participants to report up to four other conditions not listed. Those conditions diagnosed within the 12 months prior to the baseline interview (0.2 per cent of all reported medical histories) were excluded from analyses because the onset of the condition might have been after the occurrence of a reported fall. Similarly, other potential risk factors were examined only if the exposure could not vary with time (for example, sex, country of birth) or occurred prior to the 12-month reporting period for falls. Consequently, a number of potentially important measures taken at the baseline stage of the study (including blood pressure, weight, gait analysis, ability to perform activities of daily living, medication usage, and psychological functioning) were not analysed because their values might have reflected the effects of a precedent fall and this would give rise to misleading associations. The role of such factors will be examined prospectively in subsequent stages of the study. Odds ratios and their 95 per cent confidence intervals were used as measures of the strength and significance of associations between falling and possible risk factors. Age-adjusted odds ratios were calculated from stratified analysis of unweighted data using Epi Info oftw ware.'^ Multivariate unconditional logistic regression analysis of unweighted data was undertaken in accordance with the hierarchical backward elimination procedure recommended by K l e i n b a ~ mthe~SPSS computer program was used ;~ for this p ~ r p o s e . ~ ' Results The baseline cohort consisted of 1947 participants, 1039 (53 per cent) men and 908 (47 per cent) women. The age distribution for the two sexes is shown in Table 1. Sixty-nine per cent of the cohort was born in Australia and most cohort members lived in the community, with only 8 per cent living in institutional accommodation. Two-thirds of the participants were known to be living with others at their usual place of residence. A total of 550 (28 per cent) participants reported having fallen on at least one occasion in the previous 12 months (Table 1). Falls were more frequent among women than men, with 31 per cent of women compared to 26 per cent of men reporting at least one fall. In both sexes the proportion of fallers increased as age increased. Age-adjusted analysis Medical histories associated with falling. Thirty-five medical conditions were examined for possible association with falling. No adjustment was made to the significance criterion for multiple comparisons, in accordance with recent views on this matter.3z,33 Medical histories associated with falling in ageadjusted analyses are presented in Table 2. The histories most strongly associated with falling in this VOL. AUSTRALIAN AND NEW ZEAIAND JOURNAL O PUBLIC HEALTH 1997 F 21 NO. 5 A63 Table 1: Falls experience of the study cohort of the Australian Longitudinal Study of Ageing, related to its age and sex composition Number of reported falls in the previous year Age group (years) Number of and sex participants Men o m /o 22 /o Missing o n /o %‘ 0 0 0 0 0 70-74 75-79 80-84 285 Total Women 70-74 75-79 80-84 285 Total Total Note: (a) Row percentages. cohort were Parkinson’s disease and hip fracture. Among conditions not associated with falling were: hypertension, heart attack, migraine, and chronic bronchitis. Physical factors and falling: The associations between falling and a number of physical and health-related factors are presented in Table 3. Increasing age was associated with an increased risk of falling; compared with participants aged 70 to 74, those aged 85 and over had twice the odds of falling. Women appeared to be at greater risk than men. The relative odds of falling also increased with the number of morbid conditions ever suffered and with worsening vision. A history of ever having smoked regularly appeared to decrease the risk of falling. Socioeconomic factors and falling: The age of leaving school was the only socioeconomic variable associated with falling (Table 4). While total income from all sources (including that of a spouse) was not Table 2: Medical histories associated with falling in the study cohort of the Australian Longitudinal Study of Ageing %of %of fallers nonfallers n=550 n= 139 1 statistically significant as a predictor, there was an inverse relationship between total income and the relative odds of falling. Multivariate analysis We specified an initial unconditional logistic regression model containing all the significant risk factors Table 3 Physical health factors and adds of falling for the study cohort of the Australian Longitudinal Study of Ageing ~~ Factor Age group (years) %of fallers nonfallers ~ 5 5 0 1 3 9 1 OR‘ ~ %of CI referent 0.89 to 1.59 1.05 to 1.90 1.59 to 2.82 referent 1.13 to 1.70 referent 70-74 75-79 8C-84 a85 Sex Male Female No. morbidities ever 0-2 3-5 6-8 29 1 .oo 1.19 1.41 2.1 1 1 .oo 1.39 1 .oo 1.35 1.73 1.92 Disease or medical condition Parkinson’s disease Fractured hip Anaemia Stroke or transient ischaemic attack Glaucoma Arthritis Corns or bunions O R CI 1.27 to 6.83 1.26 to 3.72 1.07 to 2.1 7 1.1 1 to 2.02 1.05 to 2.15 1.21 to 1.84 1.15 to 1.79 1 .OO to 1.84 1.25 to 2.38 1.31 to 2.82 referent Ever smoked regularly No Yes 1.oo 0.72 0.58 to 0.89 Vision worsening in past 5 years No Yes Notes: (a) O = ratio of odds of falling for those with a particular medical history R 1.OO 1.47 referent 1.19 to 1.82 (b) to odds of falling far those without that history, adiusted far age with the Mantel-Hoenszel technique. C = 95% confidence interval. I Note: (a) O = adds ratio adjusted far age with the Mantel-Haenszel technique. R (b) CI = 95% confidence interval. VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1997 F NO. FACTORS ASSOCIATED WITH FALLING Table 4 Socioeconomic factors and odds of falling for the study cohort of the Australian Longitudinal Study of Ageing %of %of fallers nonfallers Factor Marital status Married or de fact0 59 Widowed, separated or divorced 37 Never married 4 Table 5: Independent predictors of falling for the study cohort of the Australian Longitudinal Study of Ageing n=550 e l 3 9 1 OR" 1.oo CI reFerent Medical history' Parkinson's disease Fractured hip Glaucoma Stroke or transient ischaemic attack Corns or bunions Arthritis 1 .oo 0.97 to 1.54 0.46 to 1.46 referent 0.63 to 1.OO 1 .oo 1.05 to 6.32 1.10 to 3.40 1.13 to 2.37 1.04 to 1.98 1.12 to 1.81 1.08 to 1.69 referent Country of birth Australia Elsewhere Age group (years) Difficulty with English language No Yes 70-74 75-79 80-84 a85 Vision worsening in past 5 years No 1 .oo 1.34 1.02 to 1.95 1.13 to 2.17 1.35 to 2.63 referent 1 .oo 0.88 referent 0.46 to 1.60 Income (from all sources) Yes 1.06 to 1.68 referent Age of leaving school (years) >$30 000 7 $1 2 001 to $30 000 51 $s12 000 42 Age of leaving school 1.oo 1.06 1.24 1 .oo 1.36 1.24 1.54 referent 0.70 to 1.63 0.81 to 1.92 referent a17 15-16 14 <14d 1.oo 1.40 1.25 1.62 1.01 to 1.94 0.89 to 1.74 1.05 to 2.50 217 15-16 14 <14 0.94 to 1.98 0.87 to 1.81 1.02 to 2.34 Notes: (a) OR = odds ratio adjusted for age with the Mantel-Haenszel technique. (b) CI = 95% confidence interval. (c) Or 170 schooling Notes: (a) OR = odds ratio calculated from final logistic regression model, as described in Results, Multivariate analysis. Model-fitting statistics: -2 log likelihood = 1878.9, improvement x2 = 100.9, 19 df, P c 0.0001, (b) CI = 95% Confidence interval. (c) Referent category is no history of the particular medical condition. (d) Or no schooling shown in Tables 2 to 4 as main effects, the remaining factors in Tables 2 to 4 along with various other medical histories as possible confounders, and age and sex as effect modifiers (with corresponding firstorder interaction terms). A final model was obtained by reducing the initial model in accordance with Kleinbaum's hierarchical backward elimination procedure.30 The independent risk factors to emerge from the final model were: history of Parkinson's disease, fractured hip, glaucoma, stroke, corns or bunions, arthritis; increasing age; deteriorating eyesight; and lack of schooling or early age of leaving school (Table 5). Other variables retained in the final model but not independently associated with falling were sex, history of asthma, history of ever having smoked regularly and marital status. The final model also included a statistically significant first-order interaction term for sex by history of X asthma (improvement ' = 4.9, 1 df, P < 0.03). For men, a history of asthma appeared to increase the odds of falling (adjusted odds ratio (OR) 1.56, 95 per cent confidence interval (CI) 0.86 to 2.84), whereas for women it appeared to decrease the odds of falling (OR 0.60, CI 0.33 to 1.08). These sex-specific odds ratios did not achieve statistical significance although the corresponding interaction term made a statistically significant improvement to the logistic regression model. Discussion As this work was part of a larger collaborative project which set out to examine the broad determinants of healthy aging, the nature of information and the level of detail that could be obtained specifically on falls was limited. When considering the results of this study, it is important to recognise the limitations of the methods that were used. The main concerns are: self-report of information on exposure and outcome variables; modest response rate; relevance of antecedent exposures; and aggregation of different types of falls into a single outcome measure. The exposure measures reported on in this paper were all obtained by self-report, and this may have resulted in some misclassification bias. To reduce misclassification bias from underreporting of potentially sensitive topics such as income and medical histories, prompt cards were used during the personal interview. Past exposures may also have been underreported simply because they were not recalled by the elderly participants; this would have produced nondifferential misclassification bias and diminution of effects unless the pattern of recall was associated with falling, in which case spurious associations might have been produced. The potential for misclassification of exposure was perhaps greatest for medical histories. In addition to possible problems with recall such as those already discussed, elderly people might not have been familiar with medical nomenclature or with their actual AUSTRALIAN AND NEW ZEALAND JOURNAL O PUBLIC HEALTH 1997 VOL. 21 NO. 5 F DOLlNlS E I AL. diagnoses, particularly if they involved complex patterns of disease or comorbidities. A further possibility is that some medical conditions might have been overreported by those who thought the condition to be a cause for their falling. While this possibility cannot be ruled out, it is expected to have had at most a small effect, since the questionnaire asked for medical histories before it asked about falls experience, and participants did not know in advance that they would be asked questions about falls. Although self-reports have clear limitations, they potentially could be used within a community or primary care setting as a relatively efficient first means of identifying persons at increased risk of falling, with subsequent clinical follow-up to confirm medical diagnoses. Classification of participants into groups of fallers and nonfallers was based on self-reported falls experience, and is therefore subject to recall bias. Studies have shown that while the number of reported falls over a 12-month period is not particularly accurate, recall of having experienced any falls versus zero falls is reasonably reliable.z0,34 Misclassification occurs predominantly in the direction of forgetting falls as opposed to reporting falls that did not occur. The usual effect of this type of misclassification is to bias measures of effect towards the null value. In the current study, the important outcome was whether a fall had occurred rather than the exact number of falls over the 12 months, and hence it is considered unlikely that there was a significant misclassification of outcome. The recruitment rate at baseline (54 per cent) was similar to that achieved in the Randwick Falls and Fractures Studyz6and in the Dubbo Osteoporosis Epidemiology Studyz’ but lower than in other studies of risk factors for falls among the elderly. In part, the lower rate of participation may have been because of the relatively high level of ongoing involvement required of participants over the course of the main study. In view of the participation rate, findings from this study may not be generalisable to all older Australians. A difficulty faced by all researchers who make use of cross-sectional data or current data on exposures combined with historical data on falling is to establish whether the temporal relationship between these measures is consistent with a causal path. In this paper, examination of risk factors was restricted to stable exposures such as the age of leaving school and the country of birth, and exposures that were known to be antecedent to the falls experience about which the participants were questioned (for example, disease history, marital status). Even so, further research will be needed to elucidate the mechanisms by which such past exposures might operate to bring about an increased risk of falling. Different types of falls (for example, those resulting from slips, trips, sudden loss of consciousness) were not distinguished in the analyses, as the level of detail provided by participants was judged to be insufficient to make useful distinctions. It is quite probable that different types of falls have different sets of risk factors, and the lack of risk specificity produced by grouping all falls together is likely to have diminished some associations or possibly even masked them a l t ~ g e t h e r . * ~number of classificaA ,~~ tions for falls have been proposed and with further evaluation may prove useful in revealing aetiological risk factors hitherto obscured.l2 The results of this study indicated that histories of Parkinson’s disease, fractured hip, glaucoma, stroke (including transient ischaemic attack), corns or bunions and arthritis are independent risk factors for falling. A history of anaemia was associated with an increased risk of falling in the age-adjusted analysis but was displaced as a predictor when other factors were controlled for in the multivariate analysis. In age-adjusted analysis, there was a trend for risk of falling to increase with the number of conditions that had ever been suffered but this was not an independent predictor. Other work has suggested that the number of conditions and disabilities a person reports may be a marker of poor health generally and that such a state is associated with an increased risk of falling.I5 Our finding that Parkinson’s disease increased the risk of falling in older people accords with the finding of Nevitt et al.7 The risk may be related to the familiar symptoms of Parkinsonism, which include tremor, muscular rigidity and difficulty in walking. A history of hip fracture approximately doubled the odds of falling. Given that past falling is a predictor of future falling,’*and that falling is the usual mechanism underlying hip fractures, this association was not unexpected. Arthritis has previously been identified as a possible risk factor for falling7 A causal relationship is plausible, in view of the degradation of lower limb function caused by the deterioration of joints in the lower limbs. Likewise, corns and bunions may increase the risk of falling by interfering with normal gait. An appropriate mix of drug, physical and occupational therapies is important to maintain normal functioning in persons affected by these conditions. Problems with vision are known to increase the chances of falling.7~z6 results of this study conThe firm the importance of vision problems in relation to falling. A history of glaucoma and self-identified worsening of vision in the last five years were identified as predictors of falling in the multivariate analysis. Early detection and treatment of glaucoma before it causes loss of vision should be promoted as an important public health issue particularly as the benefits are likely to extend beyond healthy vision to other areas such as reduced injury from falls. The independent association of stroke with the occurrence of falls is consistent with the findings of several other The symptoms of this paralysing disorder include vertigo, difficulty in walking, and visual disturbances, all of which could result in a fall. The risk factors for stroke are essentially the same as those for other cardiovascular diseases, and include hypertension, high cholesterol, smoking and obesity. All of these risk factors have received and are continuing to receive much attenVOL. AUSTRALIAN AND N W ZEALAND JOURNAL O PUBLIC HEALTH 1997 E F NO. FACTORS ASSOCIATED WITH FALLING tion in health promotion and disease prevention efforts. Success in reducing these risk factors may well lead to reductions in injuries from strokerelated falls. Age was an independent predictor of falling in this study and has been found to be so in a number of other Also in agreement with other work was the finding that sex was not independently associated with falling.'0,'' Smoking appeared to be associated with a decreased likelihood of falling in the age-adjusted analysis, but this association was not present when other factors were controlled for in the multivariate analysis. In view of the established association between smoking and cardiovascular diseases including stroke, it would appear unlikely that a history of smoking would confer any protection against falling. Early age at leaving school was the only socioeconomic variable that was an independent risk factor for falling. Total income was not significantly associated with falling although there was an inverse relationship between it and risk of falling in the age-adjusted analysis. It may be that educational attainment (as measured by age at leaving school) is a more sensitive indicator of lifetime economic disadvantage than the current total income of elderly persons.36 Given that socioeconomic factors are important determinants of the distribution of many diseases, the results of this study tentatively suggest that economic factors might play a role in the occurrence of falling. There were no apparent associations between marital status, country of birth or English language ability and the occurrence of falls. The final logistic regression model included a statistically significant interaction term suggesting that sex might be an effect modifier in the relationship between asthma and falling. Men reporting a history of asthma had a statistically nonsignificant higher risk of falling (OR 1.56, CI 0.86 to 2.84) while women reporting a history of asthma had a statistically nonsignificant lower risk (OR 0.60, CI 0.33 to 1.08). Although different falls risk factors for men and women have previously been doc~rnented,'~ our result must be treated with caution because of the nonsignificance of the sex-specific odds ratios. Few epidemiological studies in Australia have addressed the causes of falls among older people living in the community. This study provides information about risk factors for falling which should be of use to researchers and to those who are designing and running trials of falls prevention programs in Australia. The results of this study appear plausible, are generally consistent with previous knowledge, and add to the evidence regarding risk factors for falls among community dwelling older people. A number of disease conditions were identified as likely risk factors, and in the case of stroke and conditions that affect vision, a public health approach to their prevention may also lead to fewer falls among the elderly. Our cross-sectional baseline findings need further confirmation and the longitudinal phases of the Australian Longitudinal Study of Ageing will provide an opportunity to extend this work. Acknowledgments We gratefully acknowledge the work of the staff at the Centre for Ageing Studies, the Flinders University of South Australia, who carried out the Australian Longitudinal Study of Ageing and provided the data for this paper. The Australian Longitudinal Study of Ageing was funded in part by the South Australian Health Commission, the Australian Rotary Health Research Fund, and by a grant from the US National Institutes of Health (grant no. AG 0852302).

Journal

Australian and New Zealand Journal of Public Health – Wiley

Published: Aug 1, 1997

There are no references for this article.