Outdoor Falls Among Middle-Aged and Older Adults: A Neglected Public Health Problem

Wenjun Li, PhD; Theresa H.M. Keegan, PhD; Barbara Sternfeld, PhD; Stephen Sidney, MD, MPH; Charles P. Quesenberry Jr, PhD; Jennifer L. Kelsey, PhD


Am J Public Health. 2006;96(7):1192-1200. 

In This Article


The study population for our analyses was the control group of a large community-based case–control study that identified risk factors for fractures of the distal forearm, foot, proximal humerus, pelvis, and shaft of the tibia/fibula among older persons. Details about the study design and the study population have been published elsewhere.[23–26] Briefly, the control group was selected between October 1996 and May 2001 from 5 Northern California Kaiser Permanente Medical Centers (Hayward, Oakland, San Francisco, Santa Clara, and South San Francisco) with a stratified random sampling scheme. Every 3 months, people who were enrolled at the 5 Kaiser centers were stratified by gender and age group (45–49 years, 50–54 years, 55–59 years, 60–64 years, 65–69 years, 70–74 years, 75–79 years, 80–84 years, and 85 years) and were randomly ordered within each gender/age group; the first 34 women and 7 men within each group were then selected. All who belonged to a minority group or were of unknown race/ethnicity, 39% of White women, and 78% of White men within each age group were randomly chosen. Sixty-five percent of those persons selected for the control group participated; study participants who required proxy respondents because they were unable to answer the questionnaire themselves (n = 74) were excluded from our analyses. Persons in the control group who had a recent fracture or a previous fracture since age 45 years were included in our analyses.

Data about falls and possible risk factors for falls and fractures were obtained with a standardized structured questionnaire that was administered in either English or Spanish by trained interviewers. During the first 3 years of the study, most of the interviews were face-to-face; after November 15, 2000, most interviews were conducted by telephone to increase the response rate and the sample size. Mode of interview was controlled for in the multivariate analyses.

During the interview, all participants were asked how many times they had fallen during the past year. Those who had fallen at least once were asked for details about their most recent fall, including the place, circumstances, activity in which they were engaged, height from which they fell, direction in which they fell, type of surface on which they landed, whether they were wearing visual or hearing aids, and whether they took any medication or consumed alcohol before the fall.

Each respondent was classified as a nonfaller (did not fall during the past year), an indoor faller (most recent fall was indoors), or an outdoor faller (most recent fall was outdoors). An outdoor fall was defined as occurring outside a building or in a parking garage, and an indoor fall was defined as occurring inside any building other than a parking garage.

Potential risk factors for falls included demographic characteristics; weight and height, which were used to calculate body mass index (BMI); overall health status compared with others of similar age; history of practitioner-diagnosed medical conditions; self-reported foot problems; history of certain neuromuscular symptoms during the past year; history of using selected medications at least once a week for at least 1 year; recent use of medications for sleeping, calming nerves, or lifting mood; cigarette smoking; and alcohol consumption. To measure physical functioning, respondents were asked to report level of difficulty performing various tasks. Ability to perform activities of daily living during the past month was assessed with an approach similar to that used by Schwartz et al.[27]

Leisure-time physical activity was assessed with a modified Physical Activity History questionnaire,[28] which included questions about past-year frequency and duration of walking/hiking, gardening, exercise classes, swimming, bicycling, tennis, calisthenics/weight training, social dancing, jogging, bowling, golfing, stretching exercises/yoga, tai chi, and heavy housework. Each activity was assigned an appropriate metabolic equivalent value,[29] and a summary variable for total physical activity in metabolic equivalent hours of exercise per month was obtained by multiplying intensity by frequency by duration and then summing across all activities.

Approximately 10% of participants (n = 198) agreed to a slightly abbreviated interview that did not include questions about some or all of the following variables: physical activity, cigarette smoking, and part of the medication history. Because of the reduction in number of respondents for these variables, analyses that included those variables were based on slightly smaller numbers than analyses that did not include them.

Data were analyzed with Stata SE 9.0 software (Stata Corp, College Station, TX). Frequency and characteristics of falls were stratified by age and gender into the following categories: men aged 45 to 64 years (middle-aged men), men aged 65 years and older (older men), women aged 45 to 64 years (middle-aged women), and women aged 65 years and older (older women). Associations of indoor and outdoor falls with potential risk factors were assessed using unconditional multi nomial logistic regression; nonfallers were the referent category, and indoor fallers and outdoor fallers were treated as mutually exclusive categories. For both types of falls, the same predictors were included in the models, making it possible for us to examine whether the risk factor profiles differed between those who fell outdoors and those who fell indoors. A risk factor was included in the model if it was statistically significant (P < 0.10). Likelihood ratio tests evaluated the statistical significance of categorical risk factors. In all regression models, age in years, self-reported race/ethnicity (White, Native American, or other; Asian/ Pacific Islander; Black; Hispanic), and mode of interview were included to account for possible confounding from these variables. The analysis was conducted with and without probability sampling weights. The impact of weighting was negligible; therefore, the unweighted results are presented.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: