Assessing Physical Activity in Public Parks in Brazil Using Systematic Observation

Abstract

Objectives. We assessed park use in Recife, Brazil, and differences in physical activity and occupation rates in public parks with and without the Academia da Cidade Program (ACP), which provides cost-free, supervised physical activity classes.

Methods. We used the System for Observing Play and Recreation in Communities (SOPARC) in 128 targeted areas in 10 park sites (5 ACP sites, 5 non-ACP sites) to obtain data on the number of users and their physical activity levels and estimated age. Each area was assessed 4 times a day for 11 days over a 4-week period.

Results. A total of 32 974 people were observed during 5589 observation visits to target areas. People using ACP parks were more likely to be seen engaging in moderate-to-vigorous (64% vs 49%) and vigorous (25% vs 10%) physical activity. Relatively more participants in ACP sites than in non-ACP sites were females (45% vs 42% of park users) and older adults (14.7% vs 5.7% of park users).

Conclusions. On the basis of systematic observation, ACP appears to be a useful strategy in promoting park use and physical activity among the population in Recife.

Physical activity during leisure time has particular relevance for public health practitioners because of its important role in preventing chronic disease and improving mental health, perceived health status, and quality of life.^1,2 Leisure-time physical activity can also contribute to increased social interactions and social support and promote a greater sense of community cohesion.¹ However, despite its well-known benefits, the prevalence of leisure-time physical activity continues to be low in many populations, particularly in low-income countries.³

Numerous studies have found that the built environment plays an important role in promoting active living and its corresponding health benefits,^4–6 and public parks have been identified as important environmental resources for promoting leisure-time physical activity.⁷ Public health authorities have emphasized that access to and the promotion of parks, trails, and other public recreational facilities help community members reach recommended levels of physical activity.^8,9 Systematic literature reviews have also found that access to public recreational facilities promotes physical activity at the community level.^10,11 Specifically, the possibility of being physically active is greater among those living in areas with a larger proportion of land dedicated to public parks.^12,13 Studies have also recommended promoting programs within the community through the use of outreach activities, some of which have been implemented by programs in Latin America.^14–16

A systematic review of community-based physical activity interventions in the Latin American public health literature (in Spanish and Portuguese) was conducted in 2006 during the first phase of Project GUIA (Guide for Useful Interventions for Activity in Brazil and Latin America).¹⁷ The review identified 3 physical activity intervention categories that had not previously been included in the US Guide to Community Preventive Service review.¹⁰ One of these strategies was the offering of physical activity classes in community settings, including public parks. This strategy has been employed by the Academia da Cidade Program (ACP; in English, “City Gyms”), a municipal program sponsored by the Secretariat of Health in Recife, Brazil. This public policy strategy makes use of environmental resources, such as parks, that are currently available in communities to deliver physical activity classes and health and nutritional counseling free of charge.¹⁴ The second phase of Project GUIA included the evaluation of 2 ongoing community-based physical activity promotion interventions in Brazil, with the ACP in Recife being one of the programs chosen for a comprehensive evaluation using both qualitative and quantitative methods.¹⁴

The use of systematic observation methods has been promoted as a strategy suitable for monitoring levels of physical activity in the community.^18–20 One instrument, SOPARC (System for Observing Play and Recreation in Communities), has been used to obtain reliable information on physical activity and related variables in parks in the United States. There is a need for alternative evaluation approaches to physical activity at the population level, and methods like SOPARC could be good alternatives.^21,22 We used the SOPARC method to compare physical activity levels, park use, and park contextual characteristics in sites with and without the ACP.

METHODS

Recife, the capital in Pernambuco State in northeastern Brazil, is the country's fifth largest city, with over 1.5 million people. It has a tropical climate and warm temperatures that contribute to outdoor physical activity year round. Although mortality from cardiovascular disease has been declining in Brazil, states in the northeast region, including Pernambuco, have shown reverse trends, with mortality rates increasing among both men and women.²³ Additionally, a study by the Brazilian Ministry of Health in 2007 found 43% of Recife adults (≥ 18 years) to be overweight (body mass index ≥ 25 kg/m²) and 11% to be obese, rates that are among the highest of the country.²⁴ Recife also has the fifth lowest prevalence of leisure-time physical activity of all state capitals,³ and a recent survey found that only 19% of its population (26% of males, 14% of females) met recommended levels.¹⁴

The ACP, defined as a health promotion policy with a focus on physical activity, leisure, and healthy eating, was first implemented in 2002 by the health secretary of Recife. The program currently has approximately 30 000 participants in its 19 “polos” (i.e., settings where the intervention occurs, commonly public parks). The program takes place daily from 5:30 to 8:30 am and from 5:00 to 8:00 pm; it includes aerobic and dance classes and organized jogging groups, as well as exercise and diet orientation sessions for healthy people and also those with hypertension, obesity, diabetes, and heart disease.²⁵

Study Design

Our study, which was conducted in 2007, used the SOPARC methodology²¹ to objectively quantify levels of physical activity as well as user and park characteristics in 10 small to large parks (2400–85 926 m²) located in urban neighborhoods in Recife. Table 1 gives the size of ACP and non-ACP parks selected for the study, as well as the socioeconomic status (SES) of their neighborhoods. We used purposeful sampling to select an equal number of ACP and non-ACP parks. We attempted to match ACP and non-ACP parks by neighborhood SES and type of setting (there were 6 inland parks and 4 beach sites that included both parks and trails, each group divided equally between ACP and non-ACP sites), because previous literature showed evidence that these factors may affect leisure-time physical activity.^26–28 In addition, we selected parks to represent each of the 6 political administrative regions of the Municipality Office. These political administrative regions represent a balanced distribution of Recife's diversity in local (neighborhood) political representation and socioeconomic stratification as identified by type of residential developments (e.g., neighborhoods with slums or poor dwellings) and the mean monthly income of neighborhood residents.^29,30 Taking practical considerations (such as safety) into account, we distributed the final sample of 10 sites throughout the city (Table 1).

TABLE 1.

Characteristics of Parks With and Without the Academia da Cidade Program (ACP): Recife, Brazil, 2007

Site	SESa	Size, m2
ACP
Jaqueirab	High	84 100.16
Hipodromob	Middle	11 835.75
Sitio da Trinidadeb	Low	79 497.15
Boa Viagemc	High	3267.89
Brasilia Teimosac	Low	2400
Non-ACP
Casa Forteb	High	7891.78
Treze de Maiob	Middle	85 926.31
Ypirangab	Low	6783.73
Piedadec	Middle	4325.78
Boa Viagemc	High	2785.30

Note. SES = socioeconomic status.

^aSES was determined on the basis of the city administration's classification, which uses the type of residential developments (e.g., neighborhoods with slums or poor dwellings) and monthly mean income of the residents. High SES was considered a monthly mean income of 4000 Brazilian reais, equivalent to US $2300; middle SES was considered a monthly mean income of 1800 Brazilian reais, equivalent to US $1035; low SES was considered a monthly mean income of 500 Brazilian reais, equivalent to US $287.

^bInland location.

^cBeach location.

Using specially designed coding forms, SOPARC observers classified park users according to their physical activity level, sex, activity modes or types, and estimated age group. SOPARC also provided information on individual park activity areas, such as their levels of accessibility, usability, supervision, and organization. Prior to data collection and the SOPARC protocol, each site was mapped and divided into smaller target areas for measurement. A total of 128 target areas were selected for observation (mean = 12.8 areas per park); these included green spaces and picnic areas, playgrounds, multipurpose fields, sport-specific fields or courts, and gymnastic or fitness areas (all of these were standard locations for the ACP physical activity classes). Certified assessors then visited the target areas at specific times on randomly scheduled days, both weekdays and during weekends.

Observation Protocol

During each visit to a target area, assessors used the SOPARC observation protocol to scan slowly from left to right (taking approximately 1 second per person) while using a mechanical counter to score each person in the area by sex, age group (child, adolescent, adult, or older adult—that is, apparently older than 60 years), and physical activity–level code, which included sedentary (lying down, sitting, or standing), walking, and vigorous (activities producing a heart rate faster than during ordinary walking). Data were then transferred to prepared observation forms. During each visit, assessors also rated the contextual characteristics of target areas according to the following categories, classifying them as yes or no: accessible (i.e., open to the general public), usable (i.e., suitable for physical activity), supervised (i.e., directly monitored by local personnel such as park rangers, security guards, teachers), and equipped (i.e., had “loose” equipment such as balls provided).

SOPARC has been used successfully and with high reliability and validity^21,22,31 in the United States; for our study, some physical activity categories were adapted for cultural relevance to Brazil (e.g., American football was omitted and dance was added). The observation protocol and forms were translated by a native Portuguese speaker, fluent in English, and checked for accuracy.

Observation Schedule

Observations were conducted over a period of 28 days in September 2007. Each target area (n = 128) was scheduled to be observed on 11 different days at each park during 4 one-hour observation periods (starting at 6:30 am, 9:30 am, 2:30 pm, and 5:30 pm) on both weekdays and weekends. Data were collected during 5589 visits to the target areas (i.e., 5632 scheduled visits minus 43 visits not made [0.7%]). The observation times permitted comparisons to be made throughout the day and at times when ACP activities were in place and when they were not. A set rotation ensured that observations at each site included each day of the week twice (i.e., 2 Sundays, 2 Mondays, and so on). Observations were conducted only during clement weather, and missing observations were rescheduled to take place on the same day and hour during a later week.

Observer Training

Nine trained SOPARC observers conducted all observations under the supervision of 2 field coordinators. Observer training was conducted by 4 core team members of the GUIA Project who had been trained by the creator of the SOPARC tool.²¹ Observer training consisted of a 2-day workshop. On day 1, trainees participated in interactive classroom sessions designed to familiarize them with operational definitions, instrument notation, and coding conventions, and to help them to discriminate among various physical activity and contextual characteristics. Trainees practiced coding and received feedback on their scoring using examples contained in the SOPARC training DVD, which was translated from English into Portuguese. On day 2, trainees reviewed the materials and practiced their skills in diverse field exercises in local parks.

Data analysis.

Descriptive statistics (SPSS version 14.0; SPSS Inc, Chicago, IL) were used to examine differences in the frequencies and percentages of park users by sex, age, physical activity level, park characteristics, and presence or absence of ACP. Separate analyses of park use were conducted by time of observation (e.g., early morning) and ACP in operation as well as park use patterns by neighborhood SES, ACP presence, and type of setting (i.e., inland, beach). Statistical tests (χ²) were calculated, and significance was set at a α ≤ .05.

Reliability.

Reliability data were collected during 24 different 1-hour time periods by pairs of observers who made simultaneous and independent observations. During reliability observations, 28 825 people were seen, an average of 3353 per park (range = 792–5286). Reliability analyses were conducted to assess interobserver agreement on (1) characteristics of target areas (i.e., accessibility, usability, supervision, organization, and whether they were equipped), (2) number of females and males in target areas, and (3) age group and physical activity levels of females and males in target areas. Interobserver agreement scores on all area characteristics were high (Pearson product moment correlation = 0.87). Agreement between assessors on the total number of individuals, as well as their sex, age, and physical activity levels, was also high, exceeding 90%.

RESULTS

A total of 5589 visits to target areas were made and 32 974 people were observed (mean users per park = 3297; range = 935–9885). Table 1 shows use patterns by presence of ACP. Overall, more males than females were observed in the target areas (56% vs 44%; P < .001); adults (64%) were seen most frequently, followed by adolescents (13%), children (13%), and older adults (11%). When observed, 43% of the people were sedentary, 39% were walking (moderate physical activity), and 18% were engaged in vigorous activity (P < .001).

A significantly greater proportion of park users in ACP sites than in non-ACP sites were females (45% vs 42%; P < .001) and older adults (15% vs 6%; P < .001). In addition, people in ACP sites were observed to be sedentary less often (36% vs 51%; P < .001) and more likely to be engaged in vigorous physical activity (25% vs 10%; P < .001) (Table 2).

TABLE 2.

Park Use Patterns, by Presence or Absence of the Academia da Cidade Program (ACP): Recife, Brazil, 2007

	All Sites, No. (%)	ACP Sites, No. (%)	Non-ACP Sites, No. (%)
Total users	32 974 (100)	18 007 (55)	14 967 (45)
Sex
Male	18 494 (56)	9889 (54.9)	8605 (57.5)
Female	14 480 (44)	8118 (45.1)	6362 (42.5)
Age group
Children	4147 (13)	2373 (13)	1774 (11.9)
Adolescents	4294 (13)	2388 (13.3)	1906 (12.7)
Adults	21 033 (64)	10 892 (60.5)	10 141 (67.8)
Older adults	3500 (11)	2644 (14.7)	856 (5.7)
Physical activity level
Sedentary	14 129 (43)	6532 (36.3)	7597 (50.8)
Moderate	12 862 (39)	7042 (39.1)	5820 (38.9)
Vigorous	5983 (18)	4433 (24.6)	1550 (10.4)

Table 3, which describes the contextual characteristics of target areas, shows that both ACP and non-ACP sites were accessible, usable, and had organized activities nearly all the times they were visited. Target areas in ACP sites, however, were significantly more likely to be equipped (9.0% vs 0.2% of the time; P < .001) and supervised (88% vs 46%; P < .001).

TABLE 3.

Contextual Characteristics of the Target Areas in Parks With and Without the Academia da Cidade Program (ACP): Recife, Brazil, 2007

Characteristic of Site	All Sites (n = 5589), No. (%)	ACP Sites (n = 2855), No. (%)	Non-ACP Sites (n = 2734), No. (%)
Accessible	5395 (97)	2664 (93)	2731 (99)
Usable	5386 (96)	2655 (93)	2731 (99)
Equipped	262 (5)	255 (9)	7 (0.2)
Supervised	3775 (66)	2512 (88)	1263 (46)

Note. Numbers and percentages are based on total number of observation visits.

Park Use Patterns by Time of Observation

Overall, significantly more park users were seen during the 2 afternoon periods than the 2 morning periods (54% vs 46%; P < .001). Figure 1 shows that there was also variation in the proportion of people engaged in the 3 levels of physical activity (i.e., sedentary, walking, and vigorous) by time of day in both ACP and non-ACP sites. A greater proportion of people in ACP sites were engaged in vigorous physical activity during all observation periods, but this was particularly evident at the specific times during which ACP activities were being conducted; that is, at 6:30 am (25.3% at ACP sites vs 16% at non-ACP sites; P < .001) and 5:30 pm (28.2% at ACP sites vs 9.5% at non-ACP sites; P < .001). In addition, the proportion of people engaging in sedentary behaviors was substantially smaller in ACP sites during all time periods, especially during the 5:30 pm period (33.9% at ACP sites vs 59.7% at non-ACP sites; P < .001).

FIGURE 1.

Patterns of park use in parks with and without the Academia da Cidade Program (ACP): Recife, Brazil, 2007.

Note. ACP = Academia da Cidade Program

Park Use Patterns by Presence versus Absence of Intervention

Taken together, Figures 2 and 3, which illustrate physical activity levels at beach (parks and trails) and inland park locations by ACP presence and neighborhood SES, show moderate-to-vigorous physical activity (walking plus vigorous activity) to be higher at beach locations. Figure 2 shows that the proportion of people engaged in moderate-to-vigorous physical activity at beach locations was higher at non-ACP sites than at ACP sites in both low-to-middle-SES neighborhoods (92.7% vs 71.3%; P < .001) and high-SES neighborhoods (98.9% vs 71.4%; P < .001).

FIGURE 2.

Physical activity levels in parks at beach locations (n = 4), stratified by neighborhood socioeconomic status: Recife, Brazil, 2007.

Note. ACP = Academia da Cidade Program; SES = socioeconomic status.

FIGURE 3.

Physical activity levels in parks at inland locations (n = 6), stratified by neighborhood socioeconomic status: Recife, Brazil, 2007.

Note. SES = socioeconomic status.

Figure 3 shows that a greater proportion of people in ACP parks were engaged in vigorous physical activity in both low-to-middle-SES neighborhoods (20.8% vs 6.1%; P < .001) and high-SES neighborhoods (26.6% vs 10.3%; P < .001). Figure 3 also shows that a substantially large proportion (64.7%) of people in inland parks were sedentary, particularly in the non-ACP sites in the low-to-middle-SES neighborhoods.

DISCUSSION

To our knowledge, this investigation is the first to use direct observation on a large scale to study physical activity levels in Latin America. The results suggest that SOPARC is a reliable and feasible instrument for assessing physical activity and associated contextual variables (e.g., time of day, use of equipment, park characteristics) in community settings in Latin America. In our study, the presence of the ACP was associated with a higher prevalence of vigorous physical activity, particularly at inland sites. Previous research showed that former and current ACP participants were more likely to meet physical activity recommendations than non-ACP participants.¹⁶ The provision of locally funded programs may be an effective approach for increasing levels of physical activity at the population level, especially for lower-income populations. Public parks are potential resources for the promotion of physical activity during leisure time, and they are especially relevant to low-SES populations, which rely on cost-free facilities.

A greater proportion of the park users in Recife were observed to be engaged in moderate-to-vigorous physical activity than were users of 8 Los Angeles parks in a study using the same instrument (57% vs 34%).³¹ Our results, however, were similar to those of McKenzie et al.²¹ and Cohen et al.²² in that we observed more males than females, more adults than any other age groups, and more users in the afternoon and evening periods than at other times. In addition, our results were similar to those of Ferreira et al.,³² who used the SOPARC methodology in the city of Curitiba, Brazil, a year after we conducted our study. They also observed more men than women, as well as more adults and adolescents than older adults and children, which is similar to what we found in our study.

According to a recent physical activity population prevalence survey in the capitals of the main Brazilian states, walking is the most common—and the preferred—physical activity for both males and females.³ We observed a large proportion of the park users engaging in walking, particularly in beach locations. It is important to mention that 3 of the 4 beach sites included walking and running trails alongside the beach; this could have contributed to our findings, since most of the people observed at these locations were engaging in walking, jogging, or running. This helps to explain the large percentage of people observed to be engaged in walking and vigorous activity at beach sites as compared with inland sites, particularly at non-ACP sites. A study conducted by Bauman et al.²⁶ documented a coastal effect on physical activity; they found that proximity to the coast was independently associated with higher levels of physical activity after adjusting for potential confounders.

We found that more people used ACP parks than non-ACP parks, and that people in ACP parks were more active, particularly during the hours that program activities were in session and in sites away from the beach. Our results support the findings of a randomized population phone survey,¹⁴ which reported that people who heard about or participated in the ACP program were more likely to meet leisure-time physical activity recommendations for health. Our results and an earlier evaluation of the program¹⁴ suggest the effectiveness of the ACP. Findings suggest that availability of cost-free and open spaces, such as walking and running trails, could be a good public investment for the promotion of physical activity. This has implications for governmental planning and policy in Brazil and Latin America, suggesting that investments in programs like the ACP should be considered.

With large coverage at a relatively low cost, SOPARC provides an objective tool for assessing physical activity and numerous associated contextual characteristics. The results of this study suggest that direct observation of selected public spaces may be useful in providing population estimates of physical activity as well as examining program effects in an urban developing country where there are issues of safety, low income, and limited options for recreational physical activity. Because SOPARC can assess the number and demographic characteristics of people using public spaces, and the types and intensity of physical activity they engage in, it is a useful alternative to traditional surveys.

Limitations of this study include the absence of data on the condition of facilities and their amenities, the proximity of people to the selected sites, and their perceptions of safety. The confounding effect of characteristics of the built environment and social characteristics cannot be determined by the data gathered in this study. As with most studies using direct observation methods, there was the potential for participants to react to the presence of observers; however, the locations of the observers were purposefully selected to guarantee the lowest visibility to the participants. Very few people responded with curiosity or questioned observers about why they were there. Observers reported that they were typically ignored, and none of them reported that any of the people they observed displayed territoriality.

Our findings show the promise of parks in promoting physical activity. Nonetheless, although parks served many people, they were rarely filled to capacity, and individual target areas were occupied during only 41% of the observation visits. Assessing community feedback relative to park features and programming would go a long way to determining how funds for public parks should be allocated. By using feedback, it might be possible to modify facilities and create programs to meet the specific needs of community members, thus attracting more people to engage in physical activity.

ACP programs may have been a contributor to vigorous physical activity, particularly in low-income areas in sites away from the beach, where parks with an ACP program were likely to have a greater proportion of users participating in vigorous activity. Additionally, recommendations for physical activity are frequently focused on leisure-time physical activity. However, this recommendation rarely resonates with lower socioeconomic populations, where cultural and economic conditions restrict leisure-time physical activity (i.e., they have less leisure time and fewer safe areas in which to exercise).³¹ Therefore, providing free or low-cost physical activity programs at the community level, such as the ACP, may be a good way to reduce disparities in physical activity rates.