Implementing an Evidence-Based Fall Prevention Intervention in Community Senior Centers

Abstract

Objectives. To evaluate the impact of implementing an evidence-based fall prevention intervention in community senior centers.

Methods. We used a single-group design to evaluate the Tai Ji Quan: Moving for Better Balance (TJQMBB) program’s adoption, population reach, implementation, effectiveness, and maintenance among 36 senior centers in 4 Oregon counties between 2012 and 2016. The primary outcome measure, as part of the effectiveness evaluation, was number of falls as ascertained by self-report. Trained TJQMBB instructors delivered the program to community-dwelling older adults for 48 weeks, with a 6-month postintervention follow-up.

Results. TJQMBB was adopted by 89% of the senior centers approached and reached 90% of the target population. The program resulted in a 49% reduction in the total number of falls and improved physical performance. Participation was well maintained after the program’s completion. The average cost-effectiveness ratio for the 48-week program implementation was $917 per fall prevented and $676 per fall prevented for multiple falls.

Conclusions. TJQMBB is an effective public health program that can be broadly implemented in community senior centers for primary prevention of falls among community-dwelling older adults.

There is now an array of proven exercise interventions that can be used by public health and clinical practice communities to reduce the risk and incidence of falls^1–3 and that have the potential for a positive return on investment.⁴ One type of intervention, tai ji quan,⁵ has been shown to be especially effective in improving strength and functional balance, thus reducing fear of falling and the risk of falls in populations of older adults and people with balance problems.^6–8 Although some fall prevention interventions³ have proven to be highly efficacious in rigidly controlled trials, it remains to be determined whether they can reach their target population of older adults in the diverse circumstances that make up public health practice.⁹

In this study, which built on prior developmental work^10,11 that showed the feasibility of our Tai Ji Quan: Moving for Better Balance (TJQMBB) program (formerly, Tai Chi: Moving for Better Balance), we assessed the implementation of TJQMBB in community senior centers. Specifically, our primary aim was to evaluate the program’s adoption and reach, with a secondary aim of focusing on the effectiveness and maintenance of the program’s implementation. Our tertiary aim was to assess the program’s cost-effectiveness in senior center settings.

METHODS

We employed a single-group, active intervention and postintervention follow-up design.¹² We framed the evaluation of the impact of the TJQMBB implementation within the RE-AIM (Reach, Evaluation, Adoption, Implementation, Maintenance) public health model,¹³ which allowed us to evaluate the program’s adoption and population reach. We set the intervention length at 48 weeks, which allowed us to evaluate the long-term effects of TJQMBB. We gathered data for evaluating maintenance of the TJQMBB intervention from a 6-month postintervention follow-up. The dissemination area covered 4 Oregon counties (Clackamas, Lane, Multnomah, and Washington). We collected data between 2012 and 2016.

Target Adopters and Population

Adopters.

The program’s disseminating partners were community senior centers (hereafter called “centers”) serving seniors and people with disabilities. A list compiled from combined sources (i.e., local Area Agencies on Aging, county and state Senior and Disability Services, and guides to directories of community resources for services for seniors) identified 46 centers serving the study area in 2011. On-site screening and inspection excluded 10 centers because they lacked sufficient physical space or were meal-only service sites, resulting in a total of 36 centers targeted for adoption in this study.

Population.

We set inclusion criteria with minimum restrictions so as to maximize the reach to the targeted older adult population. Specifically, participants were community-dwelling older adults who were (1) aged 65 years or older, (2) physically mobile (i.e., could walk 1 or 2 blocks with or without an assistive device), (3) without severe cognitive deficits defined by the Mini-Mental State Examination (≥ 19),¹⁴ and (4) able to provide medical clearance from a health care provider.

Procedures

Research staff initially contacted the 36 selected centers to solicit interest in participating in this study and provided the centers’ managers, program coordinators, or supervisors with a full description of the program. With help from the centers’ staffs, program promotion was carried out via the centers’ newsletters, Web sites, sign-up sheets, and postings and announcements at the participating centers.

Research staff initially screened prospective participants who responded to program promotions and announcements and then scheduled them for an in-person visit, either at our research facility or at a participating center. During the visit, a research assistant reviewed study procedures, obtained consent forms, administered a health survey, recorded the participant’s history of falls, and collected other baseline measures, including demographics, anthropometrics, and physical performance. Research staff instructed participants not to participate in other new physical activities for the duration of the program.

The Program and Its Delivery

We developed the TJQMBB program on the basis of early trials,^6,15 which were subsequently adapted in a community-based pilot translation and dissemination study.¹⁰ We further refined training techniques and protocols in a later study.¹¹ The background work related to TJQMBB is fully described elsewhere.¹⁶

Designed as a group-based intervention, the training protocol requires that each session begin with 5 to 10 minutes of preparatory exercises based on tai ji quan movements, followed by teaching and practicing a set of 8 single forms constituting the core routine, along with a set of therapeutic movements (40–45 minutes). Each session ends with a simple set of breathing exercises (3–5 minutes). Although TJQMBB contains major features of tai ji quan rooted in the Yang style,⁵ such as synchronized breathing and forms performed with weight shifting, unilateral weight bearing, head–shoulder–trunk alignment and rotation, and coordinated eye–head–hand movements, its practice is more therapeutically oriented, focusing on integrating self-initiated (proactive) and self-induced (reactive) movement perturbations involving breath-by-breath actions, such as lateral trunk rotation, ankle sways, heel-pivoting and push-off motions, gait initiation and termination, and secondary cognitive (verbal) tasks, executed while transitioning between stable and unstable positions. All exercises are designed to stimulate and integrate musculoskeletal, sensory, and cognitive systems.^11,17,18

Chair-supported progressions—from completely seated to sit-and-stand to chair-assisted—were also included to (1) meet the specific needs and performance capabilities of the participants and (2) train participants for functional activities involved in daily living. Four to 6 weeks into the program, a DVD providing samples of the program’s forms and movements was distributed by research staff to all participants, who were encouraged to use the material for additional 15-minute daily home practice.

After a center decided to adopt the program, project staff and the center’s managers and supervisors collaborated to develop an implementation plan that covered room size, number of participants, instructors, and class schedules. If a room rental was requested, centers were compensated at a rate of $15 to $25 per hour. Instructors trained under previously established criteria delivered all classes and were paid $25 to $35 hourly.¹¹

Evaluation

Adoption and reach.

We calculated the adoption rate by dividing the number of centers that agreed to participate by the total number of centers approached within the dissemination geographic area. Using information on (1) the total number of responses received and (2) the total number of individuals who agreed to participate and enrolled in classes, we calculated a proportion of reach by dividing the number of participants who were enrolled by the total number of responses received during the entire study period. We evaluated the representativeness of the enrolled participants by comparing the demographic characteristics of enrolled participants with the general characteristics of the targeted population (e.g., age, gender, living situation, ethnicities) reported by the National Council on Aging.¹⁹

Implementation.

We defined implementation as the extent to which instructors successfully implemented the following prespecified program protocol components: (1) 60-minute sessions delivered twice weekly over a 48-week period, (2) adherence to the teaching and training protocols specified in the teaching plan, and (3) a class participation rate of 75% or better. We ensured fidelity of implementation by using previously established criteria.¹¹ The first author or an experienced peer instructor conducted in-class fidelity checks every 2 months.

Effectiveness.

The primary outcome was falling, which was defined as “landing on the floor or the ground, or falling and hitting objects such as stairs or pieces of furniture, by accident.”^6(p188) Information on falls was collected during monthly visits to classes by research staff or phone calls made to participants at the end of each month. We included information on falls for participants who completed the monthly falls calendar until they withdrew from the study, reached the end of the 48-week intervention, or died. In addition, we collected 4 physical performance measures: (1) timed “Up & Go” test,²⁰ (2) functional reach,²¹ (3) chair stands,²² and (4) 50-foot speed walk.

Maintenance.

We documented maintenance at 2 levels: (1) participating centers and (2) participants. For participating centers, we recorded information on whether they continued to offer the program (verified by the participating center’s staff) during the 6 months following the 48-week class. For participants, we collected information (via self-report) on whether they continued to practice during the 6 months (24 weeks) following the end of the 48-week class.

Data Analysis

We calculated descriptive statistics as means and standard deviations or percentages for demographic and study outcome measures. To evaluate effectiveness, we calculated the fall incidence rate as the number of falls that occurred during the 48-week intervention period divided by the total person-months of participation. In addition, we estimated change and annual rate of change using a latent growth curve modeling approach²³ on the repeated measures of physical performance outcomes from baseline to 24 weeks (midpoint assessment) to the program’s termination at 48 weeks. Intraclass correlations²⁴ for the repeated measures of each physical performance outcome variable were small (range = 0.02–0.08; Table A, available as a supplement to the online version of this article at http://www.ajph.org). To account for nonindependence of observations due to cluster sampling (i.e., participants were sampled within centers that were located in the same county), we analyzed a 2-level model with 3-level analysis (i.e., repeated measures of the same participants across time [level 1] that were nested within participants [level 2] who were nested within centers [level 3]), with the highest clustering level (county) used as a weight.²⁵ Estimates of the model in each of the outcome measures focused primarily on change in the latent mean of the linear slope (Figure A, available as a supplement to the online version of this article at http://www.ajph.org). The analysis accounted for incomplete data due to attrition or inability to complete the follow-up assessment (11% at 6 months; 27% at 12 months).

Given the single-group design, we used an average cost-effectiveness approach to derive a cost-effectiveness ratio on the outcome of interest: cost per fall prevented, with the ratio calculated by dividing the total direct costs of the intervention by the effect (i.e., the total number of falls prevented by the intervention). We evaluated the latter effect as the difference between the total number of falls occurring 6 months prior to intervention (i.e., “doing nothing”—the base-case alternative) and the total number of falls recorded at the end of the 48-week intervention (i.e., the effectiveness of the TJQMBB program). We also calculated a similar measure for multiple falls (i.e., participants who reported recurring falls).

We used SPSS statistical software version 23.0 (IBM Corp, Armonk, NY) for tabulating center and participant characteristics and calculating descriptive statistics of the study outcomes. We used Excel MSO Professional Plus 2016 (Microsoft Corp, Redmond, WA) to calculate all cost-related information or data. We analyzed change in physical performance data with the maximum likelihood estimator with robust standard errors via Mplus version 7.4 (Muthén & Muthén, Los Angeles, CA).

RESULTS

The following provides a summary of the results on the components of RE-AIM and cost-effectiveness analysis.

Adoption

Of the 36 centers initially contacted, 32 agreed to offer the TJQMBB program, resulting in an 89% adoption rate. Of the 4 centers that were unable to participate, 1 center declined because of a conflict with an ongoing tai ji quan class, 1 was unwilling to work with a research-based project, 1 was unable to schedule classes because it lacked a room, and 1 did not respond. A survey of the 32 participating centers’ administrative staff indicated that these centers collectively served approximately 63 000 local older adults annually (excluding duplicated service visits) in their communities (range = 50–19 000). The characteristics of participating centers are shown in Table B (available as a supplement to the online version of this article at http://www.ajph.org).

Reach

Overall enrollment flow is presented in Figure 1. Between March 2012 and September 2014, a total of 569 individuals who responded to the study program promotion were screened. Of these, 58 individuals (10%) were not enrolled because of poor health conditions (n = 13), inability to commit to the length of class time (n = 17), or time conflicts (n = 28). Thus, the recruitment effort reached 90% (511 of 569) of all respondents.

FIGURE 1—

Study Flow of Participant Recruitment, Participation, and Data Collection Status: Tai Ji Quan: Moving for Better Balance, Oregon, 2012–2016

Participants’ characteristics at baseline are shown in online Table C. The mean age for the study population was 75 years; 80% were women, 56% were living alone, and 91% were White. These demographic characteristics were similar to those of senior center participants nationally.¹⁹ A total of 417 participants (82%) reported having 2 or more chronic conditions, 279 (55%) were taking 2 or more medications (prescribed or over-the-counter), 179 (35%) reported having had 2 or more falls over the previous 6 months, and 322 (63%) reported fear of falling.

Implementation

Of the 511 participants enrolled at baseline, 119 (23%) withdrew from the study and 392 (77%) completed the 48-week intervention. Primary reasons for dropping were inability to commit to time required (53%) and problems with health (38%). Median class attendance (percentage) for the whole sample (n = 511) was 75%, compared with 80% for those who completed the entire 48-week intervention (n = 392). The median number of class sessions attended per participant, based on the whole sample, was 72 (range = 5–96). Average class attendance after the removal of dropouts was 73 sessions (SD = 15; median = 77; range = 21–96), with 258 participants (66%) attending more than 75% (72 sessions) of the 96 total intervention sessions.

We observed no severe adverse events related to the intervention. Two participants died of medical conditions unrelated to the study. A total of 3 falls without injuries were reported in class during the intervention period; all 3 participants continued class participation. At the end of the 48 weeks, 95% of those who completed an exit survey (393 of 413) described the movements in the program as “safe to do.”

Program fidelity was well maintained during implementation, as demonstrated by the fact that (1) all instructors successfully delivered the 48-week program per the protocol and (2) an annual class participation rate of 75% or better was achieved. Of the 413 participants who completed an exit survey at week 48, a large majority indicated that they (1) enjoyed the program (94%) and (2) felt that the exercises helped their balance (93%), leg strength (81%), and mobility (84%). A total of 337 participants (95%) expressed their intent to continue with the exercises.

Effectiveness and Maintenance

Information on the number of falls across the 48-week intervention period is displayed in Figure 2, which shows a reduction trend over time. For all the participants who provided data on monthly falls (511 participants at baseline; 479 at 6 months; 434 at 12 months), we documented 345 falls for 173 participants (34%) over a total of 5632 months of follow-up during the 1-year (48-week) intervention period. The average time from baseline to the first fall was 1.4 months (interquartile range = 2 months), with an incidence rate of 6 falls per 100 person-months. Of the 263 participants who reported at least 1 fall at baseline, 141 reported no falls during the 12-month intervention period, representing a 54% reduction. Similarly, at the end of the intervention, there were 327 fewer falls than at baseline (n = 672), resulting in a 49% reduction in the number of falls.

FIGURE 2—

Incidence of Falls Across the 48-Week Study Period: Tai Ji Quan: Moving for Better Balance, Oregon, 2012–2016

Note. TJQMBB = Tai Ji Quan: Moving for Better Balance.

Descriptive statistics of physical performance measures and the rate of change for each outcome across time are shown in Table 1. Latent slope means indicated a significant change in all outcomes at each 6-month interval; the estimated reduction rate was 0.13 seconds for timed Up & Go, 0.10 seconds for chair stands, and 0.14 seconds for 50-foot speed walk. We also found an increase of 0.09 inches in functional reach at each 6-month interval. The estimated average rate of change across the 48 weeks is also shown in Table 1.

TABLE 1—

Change in Physical Performance From Baseline to 48 Weeks: Tai Ji Quan: Moving for Better Balance, Oregon, 2012–2016

Measures	Baseline, Mean (SD)	24 Weeks, Mean (SD)	48 Weeks, Mean (SD)	Latent Intercept Meana	Latent Slope Meana	Estimated Mean at 6 Monthsb	Estimated Mean at 12 Monthsb
Timed Up & Go, sec	10.46 (4.25)	9.36 (3.48)	9.04 (3.29)	10.36**	−0.13*	9.61	8.86
Chair stands, sec	13.07 (4.87)	21.05 (3.86)	12.06 (4.26)	12.95**	−0.10*	12.36	11.77
50-ft walk, sec	14.46 (5.16)	13.75 (4.66)	12.73 (4.61)	14.51**	−0.14*	13.47	12.79
Functional reach, in	8.11 (2.4)	8.79 (2.45)	9.12 (2.35)	8.15**	0.09*	8.67	9.20

Note. Total number of participants was 511.

^aIn the context of a latent growth curve analysis with a linear trend, the latent mean intercept value represents the estimated average initial level of the variable at baseline (also known as a constant from which change is measured) whereas the latent mean slope represents the estimated average rate of change (i.e., an increase in the outcome variable per unit of measurement—24 weeks [6 months], 48 weeks [12 months]).

^bCalculated as (Mean_Intercept + Mean_Slope × Month of assessment). For example, the predicted mean value for timed Up & Go at 12 months is calculated as 10.36 s + (−0.13 s × 12 mo) = 8.86 s. (Month is used as the unit of change.)

*P < .005; **P < .001.

Twenty-four weeks (6 months) after the termination of the 48-week implementation, 17 centers (55%) continued offering the TJQMBB program delivered by the same trained instructors as in the original project. In a 6-month postintervention follow-up survey completed by 330 participants (65% of the total sample), 264 participants had continued their tai ji quan practice for an average of 91 minutes (median = 90 minutes) each week, 179 of them (68%) at a TJQMBB class and 85 (32%) on their own at home.

Costs and Cost-Effectiveness

The total cost for implementing the twice weekly, 48-week TJQMBB program with 511 participants in 32 classes was $307 170 ($601 per participant or $6.25 per person per class session). The average cost-effectiveness ratio for implementing the 48-week program was $917 per fall prevented; for participants who reported multiple falls at baseline and during the 48-week intervention period, the ratio was an estimated $676 per fall prevented.

DISCUSSION

In this study, we evaluated implementation of the evidence-based TJQMBB fall prevention program in a real-world setting—community senior centers. The impact of the study is demonstrated by (1) the level of the program’s adoption, (2) the extent to which it reached its intended population, (3) the delivery of the program with good fidelity, which in turn resulted in reductions in the number of falls and positive changes in physical performance, and (4) the sustainability of the program in terms of its continued use by participating centers and individual participants, as observed at a 6-month postintervention follow-up.

The results on the outcomes implementing the components of the RE-AIM model are consistent with our prior implementation studies. This is especially evident in terms of the program’s adoption by both community service providers and clinical providers and by its successful reach to the target audience of community-dwelling older adults and those at high risk for falling.^10,11 We also show that there was a significantly greater reduction in the incidence of falls in the second 6 months of the intervention compared with the first 6 months. This finding is congruent with our previous randomized controlled trials^6,8 and implementation studies^10,11 and corroborates our earlier finding of a sustained training effect observed at a 6-month postintervention follow-up.²⁶

In terms of implementation costs, the largest anticipated costs were related to expenses involving room rental and instructor pay. Overall, program implementation costs remained quite low, which is consistent with our prior work.^10,27,28 Although the economic analysis did not involve cost comparisons with a competing program, the average cost-effectiveness ratio for TJQMBB—$916 per fall prevented for single falls and $676 per fall prevented for multiple falls—indicates that the program has the potential to be used by community senior centers as a cost-effective way of training for balance and reducing falls among community-dwelling older adults.

Of the 14 exercise-based fall prevention programs that have been reviewed in the literature,^3,9 TJQMBB is the only one that has undergone a systematic implementation and evaluation process in both community and clinical settings,^10,11 as well as administrative evaluations by public health organizations or senior service agencies (e.g., AAA, state departments of health, independent evaluators) in several other states.^29–31 The accumulated knowledge derived from this study and others can help improve the application of research-to-practice strategies and further strengthen efforts aimed at widening the program’s dissemination by the broad community of public health and aging services providers, thus increasing the adoption of effective fall prevention programs nationwide.

This study has caveats and limitations. First, the implementation was limited to a single state where tai ji quan is relatively well-known and where the program has been well received in local communities. Therefore, it is unknown whether the findings from this study are generalizable to other settings (e.g., housing sites, community centers, YMCAs). Independent evaluation of this program by others has provided some initial indication that the program is feasible for wide adoption,^29–31 especially in multicultural settings.²⁹ Second, because of the lack of a comparison group, we cannot conclusively confirm the effectiveness of our program on falls and other secondary outcomes. Third, although cost data derived from this study appear promising, the robustness of the cost-effectiveness of the TJQMBB program remains to be evaluated through more rigorously designed trials.

Results from this study, complemented by results from other studies,^{10,11,29–31} suggest that TJQMBB, as a fall prevention program, can be successfully implemented and broadly disseminated in community senior centers where older adults are most often connected to vital programs and services that help them “age in place” in their homes and communities.¹⁹ Because TJQMBB does not require equipment and offers flexibility in terms of space needs, there are preliminary indications that the program may be delivered at a low cost. A major implementation challenge may involve fidelity to the key elements of the program training protocol, which includes compliance with teaching elements, class frequency (2–3 sessions per week), duration (a full 60 minutes per session), and a minimum of 48 doses required to achieve health benefits.

In this study, TJQMBB was implemented in community senior center settings. The degree to which this program can be successfully implemented by other community stakeholders or partners that provide routine preventive services remains to be determined. These stakeholders include Accountable Care Organizations, Patient-Centered Medical Homes, health insurance plans, and integrated health care entities. To increase the program’s public impact, further dissemination efforts are crucial for developing an integrated referral and enrollment system that fulfills the clinical need to refer older adults at risk for falls to TJQMBB classes organized by community service providers.⁹

In conclusion, the use of TJQMBB at existing community senior centers for primary prevention of falls among older adults is feasible and is associated with reductions in the rate of falls. Community-based aging services can be an effective platform for delivering exercise-based fall prevention interventions and should be more frequently leveraged as an underused public health resource for reducing falls among community-dwelling older adults.