Biostatics assignment

profileBebisha
A_Meta_Analysis_of_Fall_Prevention_Programs_for.2.pdf

D ow

nloaded from

https://journals.lw w .com

/nursingresearchonline by

B hD

M f5eP

H K av1zE

oum 1tQ

fN 4a+kJLhE

ZgbsIH o4X

M i0hC

yw C X 1A

W nY

Q p/IlQ

rH D 3n4/ufkX

3x2fuU H L/2uH

b+41hdC IqH

W pV

0vizw okdtv4=

on 09/22/2020

Downloadedfromhttps://journals.lww.com/nursingresearchonlinebyBhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3n4/ufkX3x2fuUHL/2uHb+41hdCIqHWpV0vizwokdtv4=on09/22/2020

Nursing Research January/February 2002 Vol 51, No 1 http://www.nursingcenter.com 1

A Meta-Analysis of Fall Prevention Programs for the Elderly

How Effective Are They?

Elizabeth E. Hill-Westmoreland � Karen Soeken � Ann Marie Spellbring

� Background: Although fall prevention studies for the elderly have been reported, there is a paucity of work summarizing the effectiveness of these interventions.

� Objectives: The research question that guided this study was: “What are the effects of fall prevention programs on the pro- portion of falls in the elderly?”

� Method: Meta-analysis was employed to summarize findings of intervention studies of fall prevention in the elderly involving a comparison group and a quantifiable outcome. Studies were reviewed by two of the authors with the eligibility crite- ria in mind. Studies were then coded and an inter-rater reli- ability check was performed.

� Results: The overall mean weighted effect size for the 12 stud- ies included in the meta-analysis was .0779 (Z � 5.03, p � .001). For fall prevention intervention types, exercise alone had a mean weighted effect size of .0220 (Z � .5303, p � .5), exercise and risk modification had a mean weighted effect size of .0687 (Z � 3.41, p � .001), and comprehen- sive risk assessment intervention studies had an effect size of .1231 (Z � 3.97, p � .001). The mean weighted effect size size for community-based studies was .0972 (Z � 5.37, p � .001) and for institution-based studies was .0237 (Z � .7822, p � .22). Time to outcome measure analyses revealed that the mean weighted effect size for studies mea- suring proportion of falls at 12 months was .0905 (Z � 5.43, p � .001), and those measuring at four months or less was �.0972 (Z � �.005, p � .50).

� Conclusions: The results of this meta-analysis indicate that there was a 4% decrease in the rate of falls for individuals who were in the treatment groups receiving various fall pre- vention interventions. Additional intervention studies need to be conducted in the elderly population with a goal of pre- venting falls.

� Key Words: elderly • fall prevention • meta-analysis

alls are a common problem in the elderly population. In community-dwelling individuals age 65 and older,

approximately one-third fall every year (King & Tinetti, 1995).Once elderly individuals are institutionalized, their potential for falls multiplies. For elderly individuals resid- ing in long-term caring settings, the incidence of falls approaches 50% (Rubenstein, Josephson, & Robbins, 1994). Much attention has been devoted to the problem of falls in the elderly, as evidenced by numerous studies throughout the gerontology literature focusing on the risk factors for falls. Studies of risk have examined both intrin- sic factors (such as physiological changes and medica- tions), as well as extrinsic factors (environmental factors such as lighting, faulty equipment and floor surfaces) that might contribute to falls in this population. During the 1990s an increase in studies of fall prevention interven- tions began to appear in the gerontology journals.

Although a variety of intervention studies focusing on fall prevention have been conducted during the 1990s, with a trend toward targeted risk abatement and exercise, very little has been done to synthesize the results of these studies. Systematic reviews have been conducted to sum- marize the literature related to fall risk factors (Rawsky, 1998) and fall prevention programs (Anonymous, 1996) in the elderly. However, few attempts have been made to syn- thesize the results of studies of the effectiveness of fall pre- vention strategies using meta-analytic techniques. One meta-analysis combined the results of seven studies that were part of the Frailty and Injuries: Co-operative Studies of Intervention Techniques (FICSIT) project (Province et al, 1995). The FICSIT project combined various forms of

F

Elizabeth E. Hill-Westmoreland, MS, RN, CS, is a Doctoral Can- didate, University of Maryland at Baltimore School of Nursing. Karen Soeken, PhD, is Associate Professor, University of Mary- land at Baltimore School of Nursing. Ann Marie Spellbring, PhD, RN, is Associate Professor, Univer- sity of Maryland at Baltimore School of Nursing.

2 Meta-Analysis of Fall Prevention Intervention Studies Nursing Research January/February 2002 Vol 51, No 1 http://www.nursingcenter.com

exercise with other interventions. The results of the meta- analysis revealed that those in the exercise group had an estimated 10% lower risk of falling than the controls (Province et al, 1995).

The meta-analysis of the FICSIT project, however, did not include the results of other studies of fall prevention and was limited primarily to studies of exercise interven- tions. To date, the impact of the various types of fall pre- vention programs on the rate difference of falls in the elderly population has not been reported and provides the rationale for conducting a meta-analysis in this area.

Purpose, Research Question, and Variable Definitions

The purpose of the study was to examine the effectiveness of strategies for preventing future falls by synthesizing the indi- vidual studies of fall prevention. The research question that guided the study was: “What are the effects of fall preven- tion programs on the proportion of falls in the elderly?”

Falls were conceptually defined as coming to rest on the ground, floor, or other lower level, unintentionally (Buchner et al., 1993). In this meta-analysis the specific outcome of interest is the rate of difference or the difference in the pro- portion of falls in the experimental group versus the control group (Shadish & Haddock, 1994). The independent vari- able in this study included any type of intervention strategy that was performed with the goal of fall prevention.

Methods

Literature Search: A variety of search techniques were employed to identify published studies of fall prevention pro- grams in the elderly. Searches were performed using Medline (1966-1999) and Cumulative Index for Nursing and Allied Health Literature (CINAHL) (1982-1999) computerized databases. The database searches were limited to articles published in the English language using combinations of the keywords “aged or elderly,” “falls,” “fall prevention,” “acci- dental falls,” “interventions,” “intervention studies,” “experimental studies,” and “results.” The database searches of Medline and CINAHL revealed a total of 139 articles. Each of these 139 abstracts was reviewed with the research question in mind. Eleven of the 139 abstracts that appeared to include the variables of interest were used for citation searching (Reed & Baxter, 1994) of the Social Sciences Cita- tion Index (SSCI) and the Science Citation Index (SCI) com- puterized databases. The citation search of the SSCI and SCI revealed three additional abstracts that appeared to include the variables of interest for this meta-analysis.

Additional search strategies were used to supplement the computerized databases, which can miss as many as 50% of studies when used in isolation (Cooper & Hedges, 1994). The ancestry approach or footnote chasing (White, 1994), utilized to uncover additional studies from refer- ence lists, yielded three articles that were reviewed for inclusion in the meta-analysis. The third author located another two studies.

Study Selection: Nineteen studies identified through the search were obtained and were independently reviewed (by

the first and third authors) for inclusion in the meta-analy- sis. Studies were included if they met the following eligibil- ity criteria: (a) sample included only older adults 60 years of age or older; (b) inclusion of some type of fall preven- tion intervention; (c) use of a comparison group on out- come measure; (d) sufficient information to determine the rate of difference in falls as the outcome measure; and, (e) the number of participants in the study groups was speci- fied in the article. The reviewers agreed that 11 studies met these eligibility criteria. Reviewers disagreed on a 12th study that was then reviewed by the second author who deemed the study met the eligibility criteria. These 12 stud- ies were included in this meta-analysis.

Seven studies failing to meet the inclusion criteria were eliminated from this meta-analysis. One study by Mitchell & Jones (1996) was eliminated because some of the study participants were less than 60 years of age, some were as young as 38. Six of the studies failed to provide enough information for the proportion of fall rate difference to be calculated (Fiatarone et al, 1994; Kilpack, Boehm, Smith, & Mudge, 1991; Province et al, 1995; Shumway-Cook, Gruber, Baldwin, & Liao, 1997; Wolf et al, 1996; Mosley, Galindo-Ciocon, Peak, & West, 1998).

Data Collection Methods: After initially reviewing the 12 studies to be included in the meta-analysis, variables were selected for inclusion in the codebook. Coded were: author, year, retrieval source, setting, country, time to out- come measure, study design, type of fall prevention inter- vention, number of subjects in the study groups, mean age of the study sample, proportion of falls for the treatment and control groups, and study quality.

A modified version of the “Research Quality Scoring Method” by Sackett and Haynes (1976) was used to code study quality (Table 1). Using this quality rating scale, the range of total quality points was 0 to 6. For coding pur-

1. Study Design (0 - 3) 0 = one group pre-test/post-test 1 = nonrandomized retrospective data 2 = nonrandomized prospective experiment 3 = randomized experiment

2. Clarity of Outcome Construct Definition (0 - 1) 0 = no definition of “falls” 1 = “falls” clearly defined

3. Outcome Measure (0 - 1) 0 = subjective measure of “falls” (i.e., self-report, report via

survey, or retrospective review of records) 1 = objective measure of “falls” (i.e., documented by staff

during course of study as fall events occurred) 4. Indication of Time Until Falls Outcome Measure (0 - 1)

0 = not specified

1 = clearly indicated

TABLE 1. Study Quality Ratings for the Meta- Analysis

Nursing Research January/February 2002 Vol 51, No 1 http://www.nursingcenter.com Meta-Analysis of Fall Prevention Intervention Studies 3

poses the studies with scores 0-3 were considered low qual- ity and those with scores 4-6 were considered high quality.

All studies were rated for quality by the first author. To determine the reliability of the ratings, two other raters applied the quality rating scale to four randomly selected studies and the Intraclass Correlation Coefficient (ICC) was calculated as a measure of the consistency among the three raters. The single measure ICC was .905 (95% CI: .80 � .96) with the average measure ICC of .97 (95% CI: .92 � .99). When the primary rater was paired with each of the other two raters, the ICC values were 1.0 and .86.

After variables were selected, data coding forms were developed to extract data from the articles. Because only one coder extracted the data from all 12 articles, another coder extracted data for three randomly selected articles (25%) for inter-rater reliability. Coder agreement was ini- tially 91%. Coders then reviewed those items for which there was lack of agreement and consensus was reached. The data were then entered into an SPSS version 10.0 (SPSS, Inc., Chicago) data file.

Statistical Methods: An effect size (d’) was calculated for each of the individual studies using the difference between population proportions formula (Becker, 1994). Specifi- cally, in this meta-analysis the effect size for each study was calculated by subtracting the proportion of falls for the experimental group from the proportion of falls for the control group so that a positive effect size favors the exper- imental group. An overall mean weighted effect size for the 12 studies was calculated weighting for study variance. As a means of sensitivity analysis, another mean weighted

effect size was computed only for those studies considered high quality, again using differences in proportions and weighting by variance. In an effort to determine whether results of the synthesis were sensitive to the inclusion or exclusion of low quality studies, the three studies not receiving a high quality rating were not included in this cal- culation: Schoenfelder & Van Why, 1997; Tinetti et al., 1994; Uden, Ehnfors, & Sjostrom, 1999. In addition, 95% confidence intervals were calculated for each effect size and for the overall mean effect size.

Results

General Description: The 12 study samples that meet eligi- bility criteria (Table 2) included a large proportion of women (overall 79.5%). The samples in this meta-analysis had a mean age of 76.5 years. Overall, the studies were of high quality, with nine of the 12 studies scoring at least a 4 of 6 on the quality rating measure previously described.

The primary studies included a broad range of inter- ventions (Table 3). Eight of the studies included some form of an exercise intervention in addition to some other fall prevention strategies. These eight studies were divided into two subgroups. One group included three studies that had exercise-focused interventions only (Campbell et al., 1997; Lord, Ward, Williams, & Strudwick, 1995; Mulrow et al., 1994). The second group included five studies with exer- cise interventions, as well as other forms of risk modifica- tion (Reinsch, MacRae, Lachenbruch, & Tobis, 1992; Schoenfelder & Van Why, 1997; Tinetti et al., 1994; Uden et al., 1999, Wagner et al., 1994). Three of the studies per-

Proportion of Subjects Who Fell

Number of First Author Year Subjects Mean Age Design Control Experimental

Campbell 1997 233 84.1 R 0.530 0.457 Close 1999 397 78.2 R 0.521 0.321 Lord 1995 169 71.7 R 0.479 0.454 Mulrow 1994 180 80.6 R 0.432 0.478 Ray 1997 417 82.7 Q 0.566 0.482 Reinsch 1992 230 74.4 Q 0.380 0.400 Rubenstein 1990 160 87.1 R 0.753 0.709 Ryan 1996 45 78 R 0.200 0.100 Schoenfelder 1997 14 74.6 Q 0.462 0.143 Tinetti 1994 291 77.9 Q 0.472 0.354 Uden 1999 379 74.4 Q 0.170 0.205 Wagner 1994 1559 72.5 R 0.368 0.283

Note. Design: R = Randomized Experiment (subjects randomly assigned); Q = Quasi-Experiment.

TABLE 2. Studies Included in Meta-Analysis of Fall Prevention in the Elderly

4 Meta-Analysis of Fall Prevention Intervention Studies Nursing Research January/February 2002 Vol 51, No 1 http://www.nursingcenter.com

formed interdisciplinary comprehensive risk assessments with recommendations made for specific interventions based upon these assessments (Close et al., 1999; Ray et al., 1997; Rubenstein, Robbins, Josephson, Schulman, & Osterweil, 1990). A final study used community education related to fall prevention with environmental modifica- tions made by subjects (Ryan & Spellbring, 1996).

Pooled Results: Effect sizes and 95% confidence intervals were calculated for each individual study and for the over- all mean weighted effect size (MWES) for the 12 studies (Figure 1). The overall MWES for the 12 studies is .0779 (Z � 5.03, p � .001) based on N � 4074 with a 95% CI .0475 to .1083. Thus across the 12 studies the results indi- cate the intervention is effective in reducing the proportion of falls incurred.

Subgroup Analyses: The variability of effect sizes for the complete set of studies (N � 12) and within subsets of studies was examined using the Q-statistic. When the test for heterogeneity of variance in effect sizes of the 12 stud- ies (QT � 17.70, df � 11) was found to be significant at p � .10, partitioning of the variance was performed. Het- erogeneity was then partitioned according to the four types of intervention groupings and homogeneity of variance of effect sizes was found within all of the intervention groups. Using alpha � .05, the Q-statistic suggest the 12 studies have similar findings.

To determine the effectiveness of the three major types of interventions combined in this meta-analysis, a MWES was calculated for each of the following groups of studies: (a) three studies using strictly exercise interventions (EX); (b) five studies using exercise and risk factor modification

Interventions 1 2 3 4 5 6 7 8 9 10 11 12

Exercise of some type Walking or low-intensity X X X X X X X Resistance or muscle strengthening X X X X X X Range of motion X X X X Reaction time X X

Mobility Gait training X X X X X Balance or coordination X X X X X X X X Transfers X X X X Proper use of walking aids X X

Educational information Fall risk and prevention X X X X Relaxation training X X Staff training on prevention X X

Screening and referrals Vision X X X X Hearing X X Alcohol abuse X Mental status X X X X X Depression X X X

Biological and physical assessments Blood pressure measurement X X X X X Various lab tests X Carotid sinus studies X EKG X 24-hour holter monitor X

Modifications Medication recommendations X X X X X X Personal or environmental safety X X X X X X X X X

Note. Numbers in the heading indicate the first author of each study: 1 = Campbell, 2 = Close, 3 = Lord, 4 = Mulrow, 5 = Ray, 6 = Reinsch, 7 = Rubenstein, 8 = Ryan, 9 = Schoenfelder, 10 = Tinetti, 11 = Uden, 12 = Wagner.

TABLE 3. Description of Study Interventions

Nursing Research January/February 2002 Vol 51, No 1 http://www.nursingcenter.com Meta-Analysis of Fall Prevention Intervention Studies 5

(EXRM); and (c) three studies using an interdisciplinary comprehensive risk assessment (CRA) approach (Figure 2). Because the interventions used in Ryan & Spellbring (1996) were considerably different from the intervention approaches of the other study groupings they were omitted from this subgroup analysis. The mean weighted effect size for the EX intervention studies was .0220 (Z � .5303, p � .5) with 95% CI of �.0593 to .1033. The mean weighted effect size for the EXRM studies was .0687 (Z � 3.41, p � .001) with a 95% CI of .0292 to .1082. The mean weighted effect size for the CRA studies was .1231 (Z � 3.97, p � .001) with a 95% CI of .0623 to .1839.

The studies were grouped according to two major types of settings, community-based and institutional, to examine whether differences in effect sizes occurred. For the com- munity-based studies the MWES was .0972 (Z � 5.37, p � .001) with a 95% CI of .0617 to .1327 (Figure 2). The MWES for the institution-based studies was .0237 (Z � .7822, p � .22) with a 95% CI of �.0357 to .0831. The analysis suggests that community-based interventions demonstrate a significant effect on proportion of falls where institution-based interventions did not.

Because studies differed in the time point at which the outcome (proportion of falls) was measured, the studies were grouped based on time. One group was comprised of eight studies that measured proportion of falls 12-months

into the study period; the other consisted of the four stud- ies measuring the outcome at four months or less. The MWES for the 12-month group was .0905 (Z � 5.43, p � .001) with a 95% CI of .0578 to .1232 and the MWES for the 4-month or less group was �.0972 (Z � �.005, p � .50) with a 95% CI of �.0810 to .0806 (Figure 2). Thus, studies that measured proportion of falls at 12-months demonstrated significant effects, but those that measured them at four months or less did not.

Sensitivity Analyses: A sensitivity analysis was performed based upon the study quality rating (range of scores 0 to 6). Nine of the 12 studies scoring between 4 and 6 on the quality rating scale were assigned a high quality rating while the remaining three studies were assigned a low quality rating (Table 4). The MWES for the high quality studies was .0812 (Z � 4.86, p � .001) with 95% CI of .0485 to .1139 while the MWES for the low quality stud- ies was .0593 (Z � 1.55, p � .07) with a 95% CI of �.0193 to .1379 (Figure 2). Results indicate that inter- ventions in high quality studies demonstrated a significant reduction in proportion of falls whereas low quality stud- ies did not.

Publication Bias: A fail-safe N � 56 was computed for the standardized Z of the 12 studies, using the “meta” com-

FIGURE 1. Fall prevention programs and rate difference of falls: Meta-analysis results. Effect sizes and 95% confi- dence intervals.

6 Meta-Analysis of Fall Prevention Intervention Studies Nursing Research January/February 2002 Vol 51, No 1 http://www.nursingcenter.com

puter program (http://nw3.nai.net/~dakenny/meta.htm) by David Kenny (1999), indicating the number of studies hav- ing an average effect of zero needed to reduce the overall effect to nonsignificance (Becker, 1994). Another means of identifying publication bias is by examining a funnel plot of sample size versus effect size (Figure 3). In the funnel graph the spout is pointing up and the funnel is not skewed. The characteristics of this plot indicate that the

problem of publication bias is unlikely in this meta-analy- sis (Begg, 1994).

Discussion

In this study, the effects of fall prevention interventions on the proportion of falls were found to be small (Cohen, 1992). Experimental mortality within each of the studies

FIGURE 2. Fall prevention programs and rate difference of falls for various subgroups. Mean effect sizes and 95% confidence intervals. EXRM � Exer- cise and Risk Modification, CRA � Comprehensive Risk Assessment.

FIGURE 3. Funnel graph of 12 fall prevention studies.

Nursing Research January/February 2002 Vol 51, No 1 http://www.nursingcenter.com Meta-Analysis of Fall Prevention Intervention Studies 7

may have contributed to this overall small effect. Five of the 12 studies (Schoenfelder & Van Why, 1997; Tinetti et al., 1994; Mulrow et al., 1994; Ray et al., 1997; and Lord et al., 1995) reported that the proportion of falls was calculated after individuals who dropped out of the study were removed from the number of individuals in each of the study groups. Thus, the majority of the studies either included individuals who had dropped out in computing their outcome measures or failed to report study dropouts. These factors may contribute to an overall small effect size. Original study authors may have included dropouts in the numerators upon which proportion of falls calculations were based; it was not possible to remove dropouts from the denominator when calculating proportion of falls because the number of dropouts was not known.

When analyses were performed to examine the EX, EXRM and CRA programs, the results of the EX grouping were not statistically significant. However, the MWES for the EXRM programs and the CRA programs were statisti- cally significant. The importance of using an approach to fall prevention that does not focus solely on exercise is supported by the larger effect sizes for the EXRM and CRA programs and their statistical significance.

When analyses were performed examining effect sizes according to study setting, the MWES was larger for the community-based studies than for the institution-based studies. These findings indicate that when compared, the community-based fall prevention programs were more effective. Generally, individuals residing in the community tend to be less frail than those residing in institutional set- tings (Rubenstein, Josephson, & Osterweil, 1996) which helps to explain this finding.

The analyses performed based upon time-to-outcome- measure demonstrated a small but statistically significant MWES for studies measuring proportion of falls at 12- months. However, the MWES was negative and not statis- tically significant for studies measuring proportion of falls at four months or less. The results suggest that fall preven- tion interventions show the greatest effects when the out- come is measured for the long-term effect, in this case, 12- months after the interventions had been initiated. Finally analyses were performed according to type of study design. The MWES for randomized experiments was larger than those for the quasi-experiments. Only the randomized grouping was statistically significant. These findings sug- gest that the more tightly controlled studies were more effective at preventing falls. The practical importance of the overall effect size of .0779 is best presented in terms of a Binomial Effect Size Display (Rosenthal, 1994). The interpretation of this finding is that there was a 4% (from 52% to 48%) decrease in the rate of falls for individuals who were in the treatment groups receiving various fall prevention interventions.

Most of the research in the area of falls in the elderly in the 1980s had the goal of identifying the risk factors related to falls. During the 1990s and in the current decade studies of the effectiveness of fall prevention programs have become more common. Based upon the overall MWES found in this meta-analysis, additional studies are needed to enhance the effectiveness of fall prevention pro- grams. Consistently including the rate difference of falls as an outcome measure will permit additional meta-analyses to be conducted and promote the development of knowl- edge in this important area. �

First Author Study Quality Study Setting* Intervention Type† Time to Outcome Effect Size Variance

Campbell High C EX 12 months 0.073 0.004 Close High C CRA 12 months 0.201 0.002 Lord High C EX 12 months 0.025 0.006 Mulrow High LTC EX 4 months �0.047 0.006 Ray High LTC CRA 12 months 0.084 0.002 Reinsch High C EXRM 12 months �0.020 0.006 Rubenstein High R CRA 12 months 0.044 0.005 Ryan High C ED 3 months 0.100 0.014 Schoenfelder Low C EXRM 3 months 0.319 0.029 Tinetti Low C EXRM 12 months 0.119 0.003 Uden Low H EXRM Not specified �0.035 0.004 Wagner High C EXRM 12 months 0.086 0.001

*Study setting: C = community; LTC = long term care; R = residential care with varying levels; H = hospital †Intervention Type: EX = exercise; EXRM = exercise and risk modification; ED = education & environmental modification; CRA = comprehensive risk assessment

TABLE 4. Study Effect Size

8 Meta-Analysis of Fall Prevention Intervention Studies Nursing Research January/February 2002 Vol 51, No 1 http://www.nursingcenter.com

Accepted for publication July 11, 2001. The authors thank John Westmoreland, Information Systems Business Analyst, First Union National Bank, for his assistance with the graph- ical presentation of the data presented in this article. Corresponding author: Elizabeth E. Hill-Westmoreland, MS, RN, CS, University of Maryland at Baltimore, School of Nursing, Room 462, 655 West Lombard Street, Baltimore, MD 21201 (e-mail: [email protected]).

References Anonymous. (1996). Preventing falls and further injury in older

people. Nursing Standard, 10(47), 32-33. Becker, J. B. (1994). Combining significance levels. In H. Cooper

& L. V. Hedges (Eds.), The Handbook of Research Synthesis (pp. 215-230). New York: Russell Sage.

Begg, C. B. (1994). Publication bias. In H. Cooper & L. V. Hedges (Eds.), The Handbook of Research Synthesis (pp.399- 409). New York: Russell Sage.

Buchner, D. M., Hornbrook, M. C., Kutner, N.G., Tinetti, M. E., Ory, M. G., Mulrow, C. D., Schechtman, K. B., Gerety, M. B., Fiatarone, M. A., Wolf, S. L. (1993). Development of the com- mon data base for the FICSIT trials. Journal of the American Geriatrics Society, 41(3), 297-308.

Campbell, A. J., Robertson, M. C., Gardner, M. M., Norton, R. N., Tilyard, M. W., & Buchner, D. M. (1997). Randomised controlled trial of a general practice programme of home based exercise to prevent falls in elderly women. BMJ, 315(7115), 1065-1069.

Close, J., Ellis, M., Hooper, R., Glucksman, E., Jackson, S., & Swift, C. (1999). Prevention of falls in the elderly trial (PRO- FET): A randomised controlled trial. Lancet, 353(9147), 93- 97.

Cohen, J. (1992). Quantitative methods in psychology: A power primer. Psychological Bulletin, 112(1), 155-159.

Cooper, H., & Hedges, L. V. (Eds.). (1994). The handbook of research synthesis. New York: Russell Sage Foundation.

Englander, F., Hodson, T. J., & Terregrossa, R. A. (1996). Eco- nomic dimensions of slip and fall injuries. Journal of Forensic Science, 41(5), 733-746.

Fiatarone, M. A., O’Neill, E. F., Ryan N. D., Clements, K. M., Solares, G. R., Nelson, M. E., Roberts, S. B., Kehayias, J. J., Lipsitz, L. A., & Evans, W. J. (1994). Exercise training and nutritional supplementation for physical frailty in very elderly people. The New England Journal of Medicine, 330(25), 1769- 1775.

Kilpack, V., Boehm, J., Smith, N., & Mudge, B. (1991). Using research-based interventions to decrease patient falls. Applied Nursing Research, 4(2), 50-56.

King, M. B., & Tinetti, M. E. (1995). Falls in community- dwelling older persons. Journal of the American Geriatrics Society, 43,1146-1154.

Lord, S. R., Ward, J. A., Williams, P., & Strudwick, M (1995). The effect of a 12-month exercise trial on balance, strength, and falls in older women: A randomized controlled trial. Jour- nal of the American Geriatrics Society, 43,1198-1206.

Mitchell, A., & Jones, N. (1996). Striving to prevent falls in an acute care setting-action to enhance quality. Journal of Clinical Nursing, 5(4), 213-220.

Mosley, A., Galindo-Ciocon, D., Peak, N., & West, M. J. (1998). Initiation and evaluation of a research-based fall prevention program. Journal of Nursing Care Quality, 13(2), 38-44.

Mulrow, C. D., Gerety, M. B., Kanten, D., Cornell, J. E., DeNino, L.A., Chiodo, L., Aguilar, C., O’Neil, M. B., Rosenberg, J., & Solis, R. M. (1994). A randomized trial of physical rehabilitation for very frail nursing home residents. JAMA, 271(7),519-524.

Province, M. A., Hadley, E. C., Hornbrook, M. C., Lipsitz, L. A., Miller, J. P., Mulrow, C. D., Ory, M. G., Sattin, R. W., Tinetti, M. E., & Wolf, S. L. (1995). The effects of exercise on falls in elderly patients: A preplanned meta-analysis of the FICSIT tri- als. JAMA, 273(17), 1341-1347.

Rawsky, E. (1998). Review of the literature on falls among the elderly. Image: Journal of Nursing Scholarship, 30(1), 47-52.

Ray, W. A., Taylor, J. A., Meador, K.G., Thapa, P.B., Brown, A. K., Kajihara, H. K., Davis, C., Gideon, P., & Griffin, M. R. (1997). A randomized trial of a consultation service to reduce falls in nursing homes. JAMA, 278(7), 557-562.

Reed, J. G., & Baxter, P. M. (1994). Using reference databases. In H. Cooper & L. V. Hedges (Eds.), The Handbook of Research Synthesis (pp. 57-70). New York: Russell Sage.

Reinsch, S., MacRae, P., Lachenbruch, P. A., & Tobis, J. S. (1992). Attempts to prevent falls and injury: A prospective community study. Gerontologist, 32(4), 450-456.

Rosenthal, R. (1994). Parametric measures of effect size. In H. Cooper & L. V. Hedges (Eds.), The Handbook of Research Synthesis (pp. 231-244). New York: Russell Sage.

Rubenstein, L. Z., Josephson, K. R., & Osterweil, D. (1996). Falls and fall prevention in the nursing home. Clinics in Geriatric Medicine, 12(4), 881-902.

Rubenstein, L. Z., Josephson, K. R., & Robbins, A. S. (1994). Falls in the nursing home. Annals of Internal Medicine, 121,442-451.

sRubenstein, L. Z., Robbins, A. S., Josephson, K. R., Schulman, B. L., & Osterwell, D. (1990). The value of assessing falls in an elderly population. Annals of Internal Medicine, 113(4),308- 316.

Ryan, J. W., & Spellbring, A. M. (1996). Implementing strategies to decrease risk of falls in older women. Journal of Geronto- logical Nursing, 22(12), 25-31.

Sackett, D. L., & Haynes, R. B. (Eds.). (1976). Compliance with therapeutic regimens. Baltimore: Johns Hopkins University Press.

Schoenfelder, D. P., & Van Why, K. (1997). A fall prevention edu- cational program for community dwelling seniors. Public Health Nursing, 14(6),383-390.

Shadish, W. R., & Haddock, C. K. (1994). Combining estimates of effect size. In H. Cooper & L. V. Hedges (Eds.), The Handbook of Research Synthesis (pp. 261-281). New York: Russell Sage.

Shumway-Cook, A., Gruber, W., Baldwin, M., & Liao, S. (1997). The effect of multidimensional exercises on balance, mobility, and falls risk in community-dwelling older adults. Physical Therapy, 77(1),46-57.

Tinetti, M. E., Baker, D. I., McAvay, G., Claus, E. B., Garrett, P., Gottschalk, M., Koch, M. L., Trainor, K., & Horwitz, R. I. (1994). A multifactorial intervention to reduce the risk of falling among elderly people living in the community. New England Journal of Medicine, 331(13) 821-827.

Uden, G. Ehnfors, M., & Sjostrom, K. (1999). Use of initial risk assessment and recording as the main nursing intervention in identifying risk of falls. Journal of Advanced Nursing, 29(1),142-152.

Wagner, E. H., LaCroix, A. Z., Grothaus, L., Leveille, S. G., Hecht, J. A., Artz, K., Odle, K., & Buchner, D. M. (1994). Pre- venting disability and falls in older adults: A population-based randomized trial. American Journal of Public Health, 84(11),1800-1806.

White, H. D. (1994). Scientific communication and literature retrieval. In H. Cooper & L. V. Hedges (Eds.), The Handbook of Research Synthesis (pp. 41-55). New York: Russell Sage.

Wolf, S. L., Barnhart, H. X., Kutner, N. G., McNeely, E., Coogler, C., Xu, T. (1996). Reducing frailty and falls in older persons: An investigation of Tai Chi and computerized balance training. Journal of the American Geriatric Society, 44,489-497.