Physical activity promotion in Latin American populations: a systematic review on issues of internal and external validity

The health consequences of physical inactivity are well documented [1, 2] and contribute to the global epidemic of non-communicable diseases [3]. Physical inactivity is the fourth leading risk factor for mortality and affects one third of the global adult population [4]. In Latin America, 43% of the population older than 15 years is inactive (defined as fewer than 30 min of moderate-intensity physical activity on at least five days every week), with the prevalence of physical inactivity ranging from 16% in Guatemala to 68% in Argentina [5].

A number of intervention strategies for improving physical activity among Latin American populations have been implemented to ameliorate this problem [6]. Initiatives like Ciclovia[7, 8], Agita[9], and GUIA[10] have focused on increasing active transportation, community-wide physical activity, and the implementation of evidence-based physical activity strategies across Latin American countries. A previous systematic review of physical activity promotion research across Latin America completed by Hoehner and colleagues summarized much of this research [6] from the perspective of the Guide to Community Preventive Services [11]. They found that there was insufficient evidence to make a recommendation for any intervention that focused on improving physical activity in adults. Poor methodological rigor resulting in low internal validity was noted as the primary limitation of the studies included in the review [6]. To improve the likelihood of having a public health impact it is necessary to also understand external validity and the influence that context may have on knowledge translation and engagement of participants who are representative of the population of interest [12–14]. Although Hoehner and colleagues updated their review to include the reporting of external validity factors [15], the majority of the studies they reviewed were from Brazil, a large, economically and culturally different country than the rest of Latin America.

The RE-AIM framework [16–18] was developed to provide researchers with an evaluation approach that balances internal and external validity factors. RE-AIM is an acronym that addresses reach and effectiveness at the individual level, adoption and implementation at the organizational level, and maintenance at the individual and organizational levels [19]. The RE-AIM framework has been used to review the reporting of internal and external validity of physical activity interventions using behavior change theories [20], school-based strategies [21], telephone-delivered strategies [22], workplace interventions [23], and interventions targeting cancer survivors [24]. Recommendations from these reviews are similar – the reporting of external validity factors should be improved to promote the translation of these strategies into practice. The purpose of this review was to examine the degree to which reports of physical activity interventions for Latin Americans focus on internal and external validity factors using the RE-AIM framework. For the purpose of this review, Latin Americans were defined as Mexicans, Mexican-Americans, Latinos and Hispanics receiving interventions in the United States or in a Spanish-speaking Latin American country.

The objective of this systematic review was to assess the degree to which the literature on physical activity interventions focusing on Latin American populations report on internal and external validity factors using the RE-AIM framework. Overall, the reviewed articles reported most frequently on the RE-AIM dimensions of reach, efficacy/effectiveness and implementation and least frequently on adoption and maintenance. Reporting was similar between English and Spanish language articles. However, Spanish articles were more likely to report on reach indicators, while English articles were more likely to report on efficacy/effectiveness indicators. In general, the reviewed articles focused on reporting RE-AIM indicators that captured internal validity.

Consistent with findings from similar reviews [20, 21, 23, 24], the method to identify the target population and inclusion criteria were the most commonly reported indicators of reach, and exclusion criteria, participation rate and representativeness were the least reported. In contrast, Hoehner and colleagues found that the studies that they reviewed mainly reported on the characteristics of the target population, participation rate and recruitment strategies [15]. Our findings show that reports from physical activity interventions for Latin American populations are lacking information relevant to generalizability.

Similar to findings from previous reviews [20, 21, 24], we found that the efficacy/effectiveness dimension was the most reported across studies. The most reported indicators within this dimension were primary outcome results and percent of attrition, and the least reported indicators were the use of imputation procedures and the use of quality of life and/or potential negative outcomes measures. Similarly, Hoehner and colleagues reported that outcomes comparable to clinical guidelines were often reported while quality of life was seldom reported [15]. These and our findings indicate that physical activity interventions for Latin Americans may be overlooking quality of life as an important public health indicator of intervention impact.

For implementation, our findings were consistent with previous reports [22–24]. Intervention dose delivered was almost always reported. Delivery of intervention as intended was sometimes reported, and cost of intervention was seldom reported. Similarly, Hoehner and colleagues found that description of intervention components/frequency and description of the delivery agent were the most reported implementation indicators [15]. These findings show that information about the fidelity and costs of physical activity interventions for Latin Americans is insufficient. Finally, adoption and maintenance indicators were the least reported, which aligns with previous findings [20, 21, 23, 24]. Within the adoption dimension, delivery agent expertise was the most reported indicator, followed by description of intervention location, which differs from a previous review [24]. Hoehner and colleagues report that the acknowledgement of intervention adoption in the setting was the most reported indicator in the articles reviewed [15]. The least reported adoption indicators were description of delivery agent, method to identify delivery agent and delivery agent/setting inclusion/exclusion criteria. Within the maintenance dimension, the most reported indicator was information on continued delivery, followed by outcomes assessed after six months, which differs from previous reviews [20, 21, 23, 24]. Hoehner and colleagues found that the most reported maintenance indicator was acceptability of intervention followed by sustainability of intervention, and similar to our findings, that intervention institutionalization was seldom reported [15].

The difference between findings from the present and the Hoehner and colleagues review may be explained by the fact that we included different studies, employed a different data extraction tool and focused on different Latin American countries. For instance, only five of studies we included in this review were also included in the Hoehner and colleagues review, which exposes work that might not have been captured previously. Further, our extraction tool was focused capturing external validity items according to the RE-AIM framework, while the tool employed by Hoehner and colleagues focused did not employ the full RE-AIM model. Last, the majority of the studies included in the Hoehner and colleagues review were from Brazil, whereas in this review most of the studies came from the United States, followed by Chile, Mexico and Colombia. Nevertheless, our conclusions are similar in that more attention to, and reporting of, external validity factors is needed the literature reporting physical activity interventions for Latin Americans.

From this review, we have drawn the following recommendations that can help improve the reporting of intervention findings and ultimately their translation into public health practice in Latin America. For improving the generalizability of findings, the reporting of intervention findings should include information about the representativeness of the sample; that is, participation rate and the characteristics of participants and non-participants. A good example of such reporting is provided in Atehortúa et al., where authors reported how patients were identified and recruited, the inclusion/exclusion criteria, that 65% of the patients who were invited to participate entered the intervention and that these patients were similar to the population at the clinic [47]. Regarding effectiveness, physical activity interventions should include a quality of life measure, which is an important intervention outcome, especially from a public health perspective, since mental and physical wellbeing provide a critical check on the impact of a program [16]. Further, such interventions should report both positive and negative outcomes in order to ensure that program-related harms do not outweigh program benefits [16]. Except for the reporting of program negative outcomes, the paper by Atehortúa et al., is a good example where the authors reported having observed positive changes in the primary outcome (patients’ cardiorespiratory fitness) and no changes in quality of life at program completion [47].

For implementation, the reporting of intervention fidelity and costs should be improved in order to promote the translation of physical activity interventions for Latin Americans across the continent. For example, Salinas et al., reported that 96 physical activity workshops were delivered over eight months (three 60-minute sessions per week), where 97% of the professors delivered the intervention as intended, and the cost per workshop was $1,200 US dollars [72]. Information about the adoption and maintenance of physical activity interventions among Latin Americans is scarce, thereby limiting their transferability and potential for sustainability. The reporting of intervention adoption should include a description of intervention location, the method used to identify the staff who delivered the intervention and the rate of adoption at the staff and setting levels. For instance, Kain et al., reported that a school-based intervention to prevent obesity in children was delivered in a small municipality of Chile by dieticians and physical educators, where 100% of the schools selected actually participated in the intervention and 57% of teachers within the schools also adopted the intervention [61]. Finally, reports on the maintenance of intervention effects and implementation should improve; outcomes after six months, institutionalization of intervention and continuation of delivery should be reported. For instance, Ingram et al., reported physical activity outcomes two years after the program was complete and reported that internal financial resources to continue the program were obtained [60].

The implications of the present findings are threefold. First, our findings show that the reviewed literature focused on reporting internal validity factors and underscore the need to improve reporting on external validity factors associated with generalizability. These findings can inform the design, implementation and reporting of future physical activity interventions by outlining the external validity factors that are likely to promote their transferability across populations and settings. For instance, the Ciclovia program [8], in which automobile streets are closed for one day a week to promote active transportation, could be evaluated with special attention to the external validity factors highlighted in this review. Further, the Ciclovia program has been shown to be a cost-beneficial program [7] that could be adopted and promoted in other Latin American countries if information critical for generalizability were available. Second, our review showed a discrepancy in reporting proportions between English and Spanish studies, which suggests that a different priority may be given to internal and external validity factors between languages. This inconsistency could be addressed by using a validated framework, such as RE-AIM, and standardizing the reporting of physical activity interventions focusing on Latin Americans. Last, this review makes and important contribution to the existing research by exposing the research gap that exists between larger, more economically advanced countries and other Latin American countries. As shown in this and the Hoehner and colleagues review [15], large counties such as Brazil and United States have conducted many more studies than other, under-resourced Latin American countries.

This systematic review provides relevant information for the physical activity promotion field that can be used to ensure that interventions are representative of Latin Americans and intervention effects are consistent and replicable across settings. This review builds onto a body of knowledge that is still in its early stages and contributes to the development of the Latin American physical activity literature. Regardless of language of publication, physical activity intervention research for Latin Americans should increase the attention to, and measurement of, external validity and cost factors that are critical in the decision making process in practice settings and can increase the likelihood of translation into community or clinical practice.