Introduction

Frailty is a syndrome (sometimes with subclinical signs) that disturbs the function of several physiologic systems [1, 2]. This phenomenon has diverse etiological factors, possibly arising due to the interaction of biological (including genetics [3]), cognitive, social, clinical, psychological, and environmental aspects [4]. Researchers have been focusing their efforts in the study of the physical components of this syndrome, the so-called physical frailty. This review focuses on physical frailty. This health condition is directly associated with age [5] and enhances the risk of falls, hospitalization, morbidity, disability, and mortality [6, 7]. Its prevalence varies among researches, but in general, it is between 7% and 32% [8]. Despite its importance for a successful aging, there is a lack of consensus regarding operational definitions of frailty and “frail” older adults, which diminishes comparability among studies. Although there is a historical confusion in scientific publications concerning the usage of frailty, comorbidity, and disability [9], with the two latter often being utilized as a synonym for the former, current knowledge indicates that frailty is a singular entity different from disability and comorbidity [10, 11]. Frail older adults often present a comorbidity condition and functional limitations [6], which indicate a strong linkage among these conditions. Although the debate concerning operational definitions of frailty continues, the most utilized definition is the one proposed by Fried et al. [6]. These authors defined frailty as meeting at least three of the following five indicators: (1) low physical activity level; (2) weakness; (3) slowness; (4) shrinking; and (5) poor endurance and energy.

A physically inactive lifestyle decreases cardiorespiratory endurance, flexibility, strength, and mobility ability (balance and muscular coordination). In this sense, physical inactivity (or a low level of physical activity) is one of the most frequent indicators of frailty found in studies [12]. Growing scientific evidence has shown the influence of exercise on frailty [13]. However, the different exercise-related regimens (type, frequency, intensity, and duration) and some inconsistencies found in randomized controlled trials (RCT), such as differences in operational definitions of frailty or no definition at all of a supposed frail elderly population, make comparisons among researches difficult and add to further complications in the debate on this subject.

The aim of this review was to study the influence of exercise on health in elderly people (≥60 years old) defined as “frail.” In this way, RCTs that administered an exercise program to a frail older adult population and that presented an operational definition of frailty were compiled, and results were discussed. Due to relatively recent advances in research on the frailty phenomenon, the present article selected RCTs published from January 2000 onwards.

Methods

Search strategy

Electronic searches were made in October 2008, utilizing three electronic databases: PubMed, Cochrane Library, and SciELO. Words for searches were entered in blocks of four, each block with a specific terminology (related to: exercise, elderly, frailty, and RCT). Similar to the recent systematic review on frailty made by Chin A Paw et al. [13], exercise-related terms (e.g., exercise, physical activity, and training) were exhaustively interchanged in AND combinations with age-related (e.g., older adult, elderly, oldest, very old, and elders), frailty-related (e.g., frailty, frail), and study design-related words (e.g., randomized, randomization, controlled trial, intervention, evaluation study, and treatment outcome). Some articles from the author’s database and articles cited for another review [13] on exercise and frailtyFootnote 1 were incorporated when they met inclusion criteria.

Inclusion criteria

The inclusion criteria were: (1) RCT study design; (2) presence of at least one exercise group and one control group (receiving no intervention, non-exercise activities, or light intensity exercise such as a flexibility exercise program); (3) exercise intervention was not overlapped for another intervention; and (4) to present an operational definition of frailty. All full-text articles written in English, French, Spanish, and Portuguese were searched. However, all articles found for this review had been written in English, except for one article written in Spanish that was excluded because it was not a RCT.

Procedures

After electronic searches, articles with frailty-related words in the title were immediately accessed. Abstract and, if necessary, the methods topic of the text were read. This methodology allowed the author to immediately eliminate articles out of scope or that did not meet inclusion criteria. If inclusion criteria were met, with a frailty definition clearly indicated in the text, articles were selected (n = 13 [14–26]). If complementary information were needed, the authors were contacted by e-mail (n = 15; electronic address for one author was unavailable). Authors that did not answer the first e-mail were contacted once more 10 days after the first e-mail. Articles that met inclusion criteria were selected (n = 8 [27–34]). Studies in which full text was not available but that were possible candidates to meet inclusion criteria were preselected, and their authors were contacted by e-mail to provide full text. From the nine authors contacted, seven sent their articles, but just two articles [35, 36] met inclusion criteria and were selected. Two articles [37, 38] cited in another review about exercise effectiveness in frail older adults [13] were selected for the present study. Finally, three studies [39–41] from the author’s personal database took part in this review. Articles selected from other reviews and from the author’s personal database did not necessarily have the terms “frail” or “frailty” in the title, but authors of these articles defined the study sample as frail in the text.

Information about study population, frailty criteria, exercise program, principles of exercise training, randomization, main and secondary outcome measures, study follow-up, and control group characteristics were extracted from each article. Some exercise regimen-related principles of training (progression, individualization, frequency, duration, intensity, and specificity—the latter was achieved when the exercise program was theoretically able to induce improvements on main outcome measures. For example, if strength is the main outcome measure of a study, and the exercise program consisted in resistance training in this case, exercise specificity was achieved) were evaluated herein on their “presence” or “absence” in the exercise program design (just when main outcome measures were physiological measures or physical performance-based measures). This evaluation, at least for frequency, duration, and intensity principles, was based on recommendations from the American College of Sports Medicine (ACSM) and the American Heart Association (AHA) for older adults [42] (for strength training, the number of repetitions per set and the number of exercises per session were not evaluated). Continuity was not evaluated because exercise adherence is a very complex health behavior, and it is not the aim of this study. Balance/coordination and task-specific exercises do not have any precise prescription regarding frequency, duration, and intensity in elderly population [42]; that is why, they were considered as “present” herein if those exercises were met, at least twice a week during 20 min in a moderate to high intensity (as described by authors).

Results

The final sample was composed of 28 articles. Table 1 shows all studies arranged in alphabetical order, with information concerning the study sample, randomization, dropouts, and the operational criteria of frailty. Six studies [14, 19, 24, 37, 39, 41] based their criteria of frailty, at least in part, on functional limitation degree as measured by activities of daily living (ADL) and instrumental activities of daily living; disability in executing ADLs was the principal criteria for two of them [39, 41]. Performance in some functional fitness tests was the most utilized criteria to define frailty, being present in at least 19 articles [14, 15, 19–28, 31–34, 36, 37, 40], while physiological measures were found to characterize frailty in nine studies [14, 19, 20, 22, 25, 26, 34–36] and fall-related aspects in six studies [24, 28, 31–33, 38]. Nutritional status and metabolic aspects contributed to define a frail elderly population in six articles [16–18, 20, 29, 38], and a low physical activity level participated in ten definitions [16–18, 20, 22, 25, 26, 29, 34, 36].

Table 1 Study sample, randomization, dropouts, and frailty criteria of all 28 articles
Full size table

Five articles [16–18, 20, 29] presented an operational definition of frailty that is different from disability and in accordance with current knowledge about this syndrome. They met two or more of the following aspects: physiological (for example, handgrip strength as measured by a dynamometer) and metabolic (for example, unintentional weight lost) aspects, poor endurance and energy, physical performance or mobility factors (for example, walking speed), and physical inactivity.

Sample size varied from 13 [23] to 311 [25, 26, 36] individuals among studies. In eight articles [16, 20, 22, 24–26, 34, 36], it was ≥200 persons, and in five others [19, 23, 27, 30, 35], it was ≤46. Mean age varied from 78.5 [17] to 86.8 [27] years of age.

Exercise program, study follow-up and intervention length, control group characteristics, and training principles can be seen in Table 2, arranged by type of exercise program. Most of the studies (n = 25) utilized supervised exercise training, and the other three articles chose home-based exercise [21, 24, 40]. Multicomponent training was the choice for nine studies [14, 15, 19, 28, 29, 31–33, 37]. Tai chi exercise [22, 25, 26, 34, 36], resistance training [23, 24, 27, 35, 38], as well as skills and functional training [16–18, 39, 41] were found in five articles, and balance/gait/coordination in four other studies [20, 21, 30, 40]. Intervention length varied from a 48-week period with tai chi exercise [22, 25, 26, 36] to a 10-week intervention with resistance/balance program [24, 27, 35] or multicomponent training [31–33]. Control group characteristics also varied among studies, but most of them utilized group activities (either low-intensity exercise—commonly focused on flexibility or social groups) to prevent influence from “socializing and attention effects” on the results achieved. Regarding training principles, exercise specificity and frequency were the most prevalent principles being found in all studies followed by exercise individualization (except for one study [27]); exercise intensity was the least prevalent principle being “present” in 12 articles [14, 16, 19, 24, 28, 29, 31, 32, 35, 38, 39, 41].

Table 2 Exercise program, study follow-up, control group characteristics, and training principles
Full size table

Table 3 presents the main outcome measures and the secondary outcome measures, as well as respective results, arranged by type of main outcome measures. The results indicate significant differences that favored participants in the exercise group with relation to subjects in the control group. For some studies, results of secondary outcome measures were not reported because either there were no secondary outcome measures or authors did not clearly specify those measures in the articles. Most of the articles had physical function (self-reported and performance-based) and mobility ability measures as the main outcome of the study [14–16, 21, 27–32, 34, 36–41] followed by physiological measures [14, 16, 18, 19, 23, 29, 32, 34, 35, 38, 41]; psychological and mental aspects were found in seven articles [17, 22, 24, 25, 33, 34, 38] and fall risk in three studies [20, 24, 26]. Four articles [17, 24, 31, 38] did not present any difference between exercisers and control group participants regarding the results of the main outcome measures. Just in two studies, exercise training had a negative effect for frail elderly regarding main outcome measure (fall-risk [20] and cardiovascular endurance [34]). Results on secondary outcome measures were not achieved for six [17, 18, 21, 24, 28, 34] of the 14 studies concerned (i.e., those that clearly presented secondary outcome measures).

Table 3 Main outcome measures and secondary outcome measures with respective results
Full size table

Discussion

This review provides some evidence of exercise effectiveness on health in a population of older adults defined as frail. The results presented for the analyzed 28 articles show that physiological factors, as well as functional fitness (FF, especially performance-based measures) and mobility ability (balance and coordination) and psychological aspects may be improved by exercise intervention in several vulnerable elderly populations.

Frailty criteria varied very much among studies. It is not surprising that the operational definition of frailty was not the same among articles because frailty is a recent subject in this research area. Furthermore, a standardized criterion of frailty has not yet been achieved by experts in this field [43]. Although it is very early to propose a single definition of frailty and then to standardize its operational criteria, efforts must be done to clearly define a frail elderly sample based on current knowledge in this field, thus, dissociating this syndrome from “disability.” Just five articles [16–18, 20, 29] presented an operational definition of frailty in accordance with current knowledge about this syndrome. In these studies, exercise was able to improve FF, mobility, lean body mass, and strength. One of them [20] presented contradictory results (FF and risk of falls were improved in prefrail older adults, but the opposite results were found among frail individuals). Although these results suggest a positive influence of exercise on health in frail older adults, it is too early to support this. More RCT studies are needed to confirm exercise benefits in a well-defined frail elderly population.

Future RCT researches on physical frailty must pay attention to the selection process of the study sample. At least three studies [27, 28, 41] had a sample partially composed of cognitively impaired or demented individuals. As indicated by Ferrucci et al. [44], frailty as a result of reduced cognition is considered a distinct clinical entity, although, decline in cognition may be found in frail persons. Intervention length is also a very important aspect for achieving improvements by exercise intervention. Frail older adults constitute a more vulnerable population when compared to healthy elderly; then, to provide improvements in physiological factors (e.g., strength, cardiovascular endurance, and flexibility), as well as in FF, they may need much time to adapt to and to progress in exercise volume.

Multicomponent training was the most utilized type of exercise program followed by tai chi, and skills and functional training, and resistance exercises. However, some incoherencies in relation to exercise programs were found. First, some studies did not clearly expose the exercise regimen-related principles of training. Exercise frequency, intensity (and how it is monitored), specificity, progression, individualization, and session duration are basic data indispensable to analyze exercise effectiveness. Thus, they have to be clearly exposed in the text even when the training program has been detailed elsewhere. This does not mean that some of the researches evaluated herein did not base their exercise program on principles of training, but these data were not precise enough to be extracted from the text.

Second, some inadequacies were found in exercise program design, which led some authors to suggest that exercise was not effective. In a home-based quadriceps resistance exercise program, Latham et al. [24] concluded that this “form of resistance exercise was harmful to patients, as evidenced by the higher incidence of musculoskeletal injuries.” However, the exercise program started with a high-intensity resistance training (60–80% of 1RM), which probably led to musculoskeletal injuries. This kind of problem can be prevented if exercise training is taken in a gradual or stepwise approach [42] (principle of load progression) in the beginning of the intervention, which Latham et al. [24] did after participants’ complaints. Moreover, this inadequacy may have negatively influenced results achieved in relation to self-rated physical health, a main outcome measure of the study (there was no difference between exercisers and control individuals). In the same way, Faber et al. [20] indicated that the “absence of significant positive training effects (regarding frail individuals) might also be attributed to inadequacy of training intensity, frequency, duration, and/or specificity of the exercise mode.” Brown et al. [15] did not find any difference on flexibility between exercisers and control subjects, though the control group met an exercise program focused on range of motion, 1 h per session, three times a week (more than the minimum prescribed for older adults by ACSM and AHA) during the intervention period. Furthermore, both groups presented improvements when compared to baseline assessments, which suggests that the exercise program was able to improve flexibility. In the study of Wolf et al. [34], tai chi exercisers reduced their cardiovascular endurance (distance ambulated in a 12-min walking test) after a 15-week period intervention. In a review study, Kuramoto et al. [45] indicated that “Tai Chi may be an additional form of aerobic exercise.” However, two meta-analyses concerning the effects of tai chi on aerobic capacity [46, 47] concluded that the average effect size was small and nonsignificant for subjects enrolled in the experimental studies. Further studies are needed to know the real influence of tai chi on cardiovascular endurance.

Third, several of these studies were, apparently, part of the same trial (same sample and exercise training) with different main outcome measures. Thus, the exercise program may have been more adequate for some of the outcome measures in one study than it was for other measures in another study. This may have influenced some of the results. In those cases, the absence of positive results of exercise training on outcome measures does not necessarily mean an absence of the effect (see comments about Brown et al. [15] just above).

Furthermore, general guidelines for balance exercise, such as ACSM and AHA, are not precise enough to guide the design of RCT studies. As indicated by Nelson et al. [42], “the preferred types, frequency, and duration of balance training are unclear,” which makes it difficult to design a study based on current recommendations to improve balance in a frail older adult population. It may partially explain the ineffectiveness of intervention in some trials (in main outcome measures [20], as well as in secondary ones [24]).

The present study has some limitations. First, in order to be selected through electronic searches, articles had to contain the terms “frail” or “frailty” in the title, which may have limited the final sample. Second, articles had to present an operational definition of frailty to meet the inclusion criteria, which may also have reduced the final sample. Finally, the several operational definitions of frailty found among studies characterized a large range of vulnerable populations. It is possible that other studies, with similar samples, have not been found by electronic searches and, therefore, were not included because authors did not label their population as “frail.”

However, to the author’s knowledge, it is the first review that compiles RCTs that studied exercise effects on health in frail elderly populations with a defined criterion of frailty. The study methodology, with direct contacts with authors of the original RCTs to clarify some points of the trials, was a strong aspect of the present review. This permitted the compilation of articles that had not presented a clear definition of frailty in the text.

Conclusion

Frailty is a physiological syndrome that increases the risk of poor health. Researches that studied exercise benefits in frail elderly populations utilized different operational definitions of frailty, which renders comparisons among them difficult to make. However, an increasing amount of evidence supports the affirmation that exercise training is an important tool to improve health in various at-risk populations. This review suggests that exercise is able to improve physiological factors (e.g., strength, cardiovascular endurance, and flexibility), as well as FF (especially performance-based measures) and mobility ability (balance and gait), and psychological (e.g., perceived health status, fear of falling, and mood) aspects in diverse vulnerable elderly populations. However, some efforts must be done to strengthen result consistencies from RCTs that study frail older adults. A frail elderly population must be defined based on current knowledge in this field before further conclusions can be drawn. Furthermore, the exercise program must be designed in accordance with current guidelines regarding exercise frequency, duration, and intensity, and it must take into account population heterogeneity (individualization) and other training principles, such as progression and specificity. Experimental researches that focus on long-term exercise adherence in frail older adults are also needed. In sum, exercise training seems to be a safe and effective tool to promote and maintain optimal health levels in a large variety of vulnerable older adults. However, lack of RCTs that utilize a well-defined frail older population based on current knowledge in this field does not permit us to establish that exercise influences health in the elderly. To confirm exercise benefits on health in frail older adults further researches are needed.