In 2010-2011 we enrolled a large group of older adults (n=814) who completed Pennsylvania’s Healthy Steps for Older Adults, 503 who took the program for the first time and another 311 who had completed the program in prior years. The first-time participants represent 12.5% of the total number of older adults (n=4040) in the state taking the program in 2010-11. (Pennsylvania only tracks first-time participants in the program.) A comparator group was recruited from the same senior center sites at the same time. These were people who attended the senior center but did not participate in the Healthy Steps program because they were not attending on the day of the program or declined to participate. Both groups completed an in-person telephone baseline interview after providing informed consent, and all were followed up to a year with monthly telephone interviews to track falls, as well as in-person telephone assessments at 6 and 12 months.

Healthy Steps is a half-day workshop open to anyone who wishes to attend and is offered at no cost to participants. Senior Center staff and lay volunteers conduct balance assessments and provide falls education and referrals. Participating county Area Agencies on Aging commit to offering the program to a pre-specified number of participants each year and are reimbursed by the state at $70 per participant to cover the expenses of the program. Importantly, Healthy Steps is a “walk-in” program. While some sites seek preregistration, in practice attendance mostly depends on who attends the senior center on a day the program is offered. The short interval between the time the program is announced and offered, as well as the difficulty of not offering (or delaying) the program in senior centers to establish a control condition, made random allocation difficult. The University of Pittsburgh Institutional Review Board approved the research.

The project included a data sharing agreement with the PA Department of Aging and involvement of PA Department of Aging staff as well as county health promotion coordinators (PrimeTime Health) at local sites. Stakeholders were engaged in the project from the start, both in the design phase and in follow-up through monthly conference calls. Results from the study will be used by program staff to refine the intervention and its outreach.

Pennsylvania’s Department of Aging has offered Healthy Steps for Older Adults (HSOA) on a statewide basis through its senior centers since 2007.[19] HSOA was developed under the auspices of Health Research for Action at UC-Berkeley.[20] Each year between 4000-7000 older adults aged 50 or older complete the program, and to date about 32,000 PA older adults have completed the program. Overall, 40 of 67 PA counties have hosted HSOA, which is funded though federal and state sources and administered by the Department of Aging’s health promotion unit, PrimeTime Health, which trains providers in county Area Agencies on Aging to offer the program. Senior centers and allied sites host the program, and older adults interested in the program may complete the program as part of their normal attendance at senior centers or specifically because of an interest in falls prevention. Thus, Healthy Steps represents a concerted use of a state’s aging services infrastructure to deliver a falls prevention program to older adults. In this effort, the PA Department of Aging works with county Area Agencies on Aging, who in turn deliver the program at local senior centers.

The program includes the following elements: physical performance assessments of balance and mobility conducted by staff or trained volunteers (Timed Up and Go, One-legged stand, Chair Stand); referrals for physician care and home safety for participants scoring below age- and gender-based norms on performance assessments; and a 2-hour falls prevention class involving recognition of home hazards and falls risk situations as well as demonstration of exercises designed to improve balance and mobility. Two of the performance tests, the Timed Up and Go and Chair Stand are validated assessment tools included in the CDC STEADI falls assessment tool kit (Stopping Elderly Accidents, Deaths, and Injuries).[21]

Program data are entered in a web-based system and were made available to the research team through a data sharing agreement. The PrimeTime Health office of the PA Department of Aging assures program fidelity by training staff at sites (over 350 sites yearly) to conduct balance and mobility assessments, provide referrals to physicians and home safety assessments when indicated, and follow the HSOA Manual guidelines for presenting information on falls risk, exercise, and home safety. PrimeTime Health also assures fidelity by monitoring data entry, conducting monthly conference calls with county Area Agencies on Aging, and surveying a random 10% of HSOA participants about program participation.

The primary outcome was falls incidence over up to 12 months of follow-up, which we measured as fall-months, months in which participants reported a fall, per 100 person-months of follow-up. A secondary outcome was fall-months per 100 person-months of follow-up adjusted for how active participants were. In both cases we tracked the occurrence of any fall, not the number of falls in a month, because subsequent interviews with fallers showed that 89% of fall-months involved a single fall. We were also concerned that the reported number of falls in a month may be less reliable than reports of any fall. We also examined the number of fallers in each study arm using the same denominators. Falls were defined as any occasion when a person ends up on the floor or ground without being able to stop or prevent it.

The activity-based secondary outcome adjusts the denominator for follow-up time according to self-reports of activity. It is important to make this adjustment for activity over follow-up because older people with balance problems or mobility limitation may reduce their daily activity to minimize risk of falling.[22-24] For this adjustment, participants reported in each month of follow-up the number of days in the past week in which they were physically active. We defined “physically active” as moderate or vigorous activity for at least 30 min in the day. Using these reports of activity we constructed an “active month” equivalent (for example, a participant reporting 4 of 7 days of activity in the past week would have 16 days of activity in the month and contribute 0.53 months of follow-up rather than 1 month). To calculate incidence density, we summed the number of months in which respondents reported a fall (fall months) and both follow-up months and activity-adjusted follow-up month equivalents.

Falls and activity were elicited in a monthly telephone call using an interactive voice response (IVR) system. [25] We tracked all falls, not just injurious falls. Participants who signed consent were registered in a web-based system that generated the monthly telephone calls. Participants were scheduled for follow-up each month, every 30 days, beginning 30 days after their baseline interview. As the scheduled date approached, the automated system generated two calls each day (one morning, one evening) for up to 8 days around the scheduled day until the person answered and the monthly interview was completed. If a respondent did not complete the follow-up interview in the 8-day window, the follow-up was considered missing and a call was attempted in the next month.

The automated call elicited whether the person had fallen, weekly activity, hospitalization, and emergency department use in the prior 30 days. A final question asked if participants would like a telephone call from the research team. Respondents answered questions by pushing buttons on the telephone. For falls, respondents were asked, “Think about the last 30 days. Did you fall in the last 30 days, that is, end up on the floor or ground because you were unable to stop yourself? Press 1 for yes, 2 for no.” When participants reported a fall, the system generated an email message to the research team, who followed up with an in-person telephone interview to collect information about the fall, including when the fall occurred (during the day/evening or at night after getting up from bed) and the location (at home or outside). Given the limitations of the automated telephone interview, we relied on a simple global measure of activity. Respondents were asked to think about the last 7 days and to push 0 if they had no days in which they were active at least 30 min, 1 if they were active one day, 2 for two days, 3 for three days, and so on up to 7. Active days were defined as days when participants “walked, or did exercises, or did a hobby or volunteer work that involved being on your feet for at least 30 minutes.”

The monthly calls took a mean of 2.5 min to complete. Most calls were completed in the first day of the IVR-generated monthly call. Compliance was reasonably high, with about 20% opting out at baseline. These respondents received personal telephone calls. In-person telephone calls were also reserved for people with hearing impairment, Spanish speakers, and people lacking touchtone telephones. Participants could opt out of the automated IVR system at any point over follow-up. To assess the reliability of IVR-reported falls, we compared falls reports from a subset of people (n=65) who completed an IVR assessment and an in-person telephone assessment in the same month. Reports of falls agreed in 95.3% of cases.

In addition to self-reported falls and activity over follow-up, we collected measures at baseline to assess comparability of the HSOA and comparator groups. For self-reported balance, we asked, “How would you rate your balance: Excellent, very good, good, fair, or poor?” This simple self-report, drawn from the NIH Claude D. Pepper Older Americans Independence Centers assessment battery, is highly correlated with falls risk. In this sample, the odds of reporting a fall in the prior 12 months were 2.86 (95% confidence interval, 2.2-3.7) in people reporting fair or poor balance at baseline compared to those reporting good, very good, or excellent balance. Other self-report measures included self-reported medical conditions, measures of function and symptoms (adapted from the EQ-5D to assess disability in the basic and instrumental activities of daily living, mobility, pain, and presence of symptoms of anxiety or depression [26]), physical performance (Community Healthy Activities Model Program for Seniors (CHAMPS) physical activity measure [27]), falls in the prior 12 months, self-rated balance, and memory performance (Memory Impairment Screen-Telephone, MIS [28, 29]). The CHAMPS questionnaire assesses weekly frequency and duration of 40 different activities typically undertaken by older adults. We summed the number of tasks performed in the prior week to develop a measure of total physical activity and dichotomized scores at the median (10 activities) for analyses. The MIS involves registration of four words along with a semantic category cue. After 3-4 minutes of distraction with other questions, respondents are asked to recall the four words. Scores range from 0-8. We dichotomized the measure at the median of the distribution (0-6 vs. 7-8) as an indicator of cognitive status. Given the geographic dispersion of the sample across 19 PA counties, all measures were obtained by telephone.

Descriptive statistics were calculated for the HSOA and comparator groups. Incidence rate ratios (IRR) and 95% confidence intervals were estimated using a Poisson regression model. We estimated a multivariate model to assess differences in falls incidence between the HSOA group and comparator adjusted for fall risk covariates. Fall risk covariates included age, gender, race, disability, memory performance, self-reported balance, fall in the prior year, and physical activity. Covariates were dichotomized to aid in interpretation, but model results were similar using continuous measures. Analyses were limited to follow-up data collected from study inception in October 2010 through June 2011 to assess the initial impact of the program.

Participants included 1833 older adults, aged 50+, from senior centers across 19 PA counties in 2010-2011. Of the 814 who completed HSOA, 9.5% completed baseline assessments before the HSOA program, 45.7% within 2 months of the program, and 44.8% 2-4 months after the program. Because the groups did not differ in baseline features or falls incidence over follow-up (results available upon request), they were combined and considered a single intervention arm for assessing the effectiveness of the program. Follow-up began after completion of the baseline interview. Another 1019 older adults who did not complete the program were recruited from the same sites during the same period.

Figure 1 presents the CONSORT diagram for the study. Participation was high (90% of participants providing contact information signed consent, and 83% of people providing consent completed the baseline assessment). Follow-up response was also excellent (97% had 1 or more months of follow-up). At a median of 7.5 months of follow-up, the cohort had provided 13,227 of 16,500 possible monthly follow-up assessments for a completion rate of 80%. Attrition in the HSOA and comparator groups was similar: 5.7% in the HSOA group and 5.8% in the comparator withdrew or were lost to follow-up. Deaths, inability to participate because of illness or cognitive impairment, and relocation outside the state were also similar between the groups.

Older adults who completed Healthy Steps for Older Adults, a statewide effort in primary prevention of falls, had a significantly lower incidence of falls relative to a comparator group ascertained at the same sites. In multivariate models, Healthy Steps was associated with a lower rate of falls (IRR = 0.83, a significant reduction of about 17%) and smaller proportion of people falling (6-8% lower, which, however, did not reach significance) in a median of 7.5 months follow-up after baseline assessment. This reduction in rate of falls is similar to estimates derived from a meta-analysis of randomized controlled trials for falls prevention conducted in 2000-2009 (RR = 0.85).[30]

Because Healthy Steps uses the distributed network of senior center sites operated by county Area Agencies on Aging, it can be offered to large numbers of older adults. Its use of simple performance measures that can be administered by lay volunteers keeps costs low. Standardized data collection allows appropriate program audit and management across sites and counties. In 2010-11, The PA Department reimbursed sites $70 per person for delivering the program and allocated $1.2M to the program as a whole. With this state commitment, Healthy Steps is a scalable, effective platform for mass screening of older adults for falls risk. It relies on existing aging services infrastructure, identification of older adults at high risk of falls, referral of high-risk older adults to personal physicians for falls assessment, local resources for home safety, and education. The program provides a booklet with exercises and demonstration of balance and strength exercises for falls prevention but does not involve exercise classes.

What aspects of the program are responsible for the reduction in falls requires further analysis. Only 21.5% of older adults informed they were at high risk of falls followed up with physicians. On the other hand, over three-quarters of Healthy Steps participants conducted home safety assessments, and a third reduced home hazards. It is possible that simply informing older adults of high risk status and heightening sensitivity to falls risk situations may also reduce falls. Some evidence for this interpretation can be seen in the greater benefit of the program for people with self-rated fair or poor balance. Further analysis of the experience of the cohort will be required to identify mechanisms responsible for reduction in falls incidence.

The 17% reduction in falls incidence places Healthy Steps between individual-tailored multicomponent interventions, such as the Yale component of FICSIT (Frailty and Injuries Cooperative Studies of Intervention Techniques), which led to a 35% reduction, [10] and mass provider outreach and education efforts, which have reduced falls incidence by 9%. [31] The program offered most benefit to people reporting fair or poor balance. In these older adults, Healthy Steps was associated with 4 fewer months with falls per 100 active-month equivalents, that is, 14 vs. 18 months with falls. This difference amounts to a reduction of 0.48 falls per person-year, or 1 less fall per person every 2 years. The potential for Healthy Steps to reduce falls in PA is high, since the population over age 50 is about 4.5 million. By the same token, the reach of the program is still small, with only 32,000 completing the program in its first 6 years of operation.

Healthy Steps is a hybrid program. It is a combination of falls education, referral for medical care, and referral to social service agencies for home safety assessments, all offered in the setting of ongoing senior center contact with participants. For these reasons, it is not strictly comparable to falls education programs or one-time clinical referral efforts, which have not been successful in falls prevention. Also, it is designed to promote falls prevention for all community-dwelling elderly, not just those at high risk, which again makes comparison to other efforts difficult. While results from this research support this approach to primary prevention of falls, a randomized controlled trial of Healthy Steps remains the definitive test. Further evaluation and dissemination efforts would also be valuable. Program enhancements to Healthy Steps, such as referrals for medication therapy management or notification of physicians when participants perform poorly on measures of mobility, may also be useful.

Results from this research should be interpreted in light of limitations in study design. Most centrally, assignment to Healthy Steps and the comparator was not random. The comparator group included people receiving services at the same senior centers who did not participate in the falls prevention program. However, all participants in the research received the same standardized baseline assessment and monthly follow-up, and the two groups were similar at baseline in falls risk and medical status. Interviewers conducting baseline assessments were blind to HSOA status, and high compliance with the automated monthly falls assessment reduced potential interviewer bias. Attrition was not differential across groups and was uniformly low. While baseline assessments were obtained after the intervention for the majority of Healthy Steps participants, timing of the baseline was not associated with baseline features or falls incidence over follow-up.

Because participants were drawn from across the state, we were limited to telephone contact and brief assessments of monthly physical activity. We were also unable to conduct clinical assessments or performance tests. We used self-reports and a telephone-based memory assessment as an alternative. The strong association between self-reported balance and falls incidence suggests that self-reports were a reasonable proxy for balance assessment.

Finally, ascertainment of falls was also based on self-report. We attempted to overcome biases in self-reports (such as telescoping or omission of minor falls) by ascertaining falls each month using an automated falls registration system. Compliance with monthly assessments was high, suggesting that a telephone-based interactive voice response system (supplemented by personal calls) is a reasonable way to ascertain falls or other recurrent endpoints. Finally, recognizing that older adults at high risk of falls may reduce activity in an effort to lower risk, we calculated incidence in two ways, both total follow-up months and an activity-adjusted month equivalent. The effectiveness of Healthy Steps in multivariate analyses was evident for both incidence measures.

In conclusion, primary prevention of falls using existing aging services infrastructure is feasible and is associated with significant reductions in the rate of falls, especially among older adults who report fair or poor balance.