Household physical activity contributions to total physical activity and its associations with cardiometabolic risk factors: a Brazilian adult population analysis

Abstract

Background

Physical activity (PA) is well-recognized as a healthy behavior. However, there is insufficient evidence to differentiate the effect of accumulating PA in different PA domains (leisure-time, work, transportation, and household). We analyzed the contribution of household physical activity (HPA) to total physical activity (TPA) and its associations with cardiometabolic risk factors (CMRFs) in Brazilian adults.

Methods

Data from the 2019 Brazilian National Health Survey were used (n = 87,678). PA variables were constructed based on the weekly time spent on PA in all PA domains. Impact on prevalence, relative contribution of HPA, and Poisson regression models were conducted, reporting proportions and prevalence ratio values with 95%CI.

Results

When HPA was not considered in TPA, a significant reduction in the prevalence of active individuals was observed (-4.3p.p.), mainly among females (-6.8p.p.). However, the mean contribution of HPA to TPA was small (7.7%). Also, the associations between physical inactivity and CMRFs did not change when including or not HPA.

Conclusions

In summary, HPA has a modest impact on TPA and in the prevalence of active individuals, especially among females, and, in general, it did not modify the associations between PA and CMRFs.

Background

According to the World Health Organization (WHO), adults should practice at least 150–300 min per week of moderate aerobic physical activity (PA) or at least 75–150 min per week of vigorous aerobic PA or an equivalent combination of moderate and vigorous activities. This recommendation is associated with several positive health outcomes, such as decreased overall and cardiovascular mortality, reduced incidence of hypertension, cancer, and type 2 diabetes, reduced symptoms of anxiety and depression, and improved cognitive health, sleep, and adiposity [1].

The WHO advises that this recommendation can be achieved in all PA domains (i.e., leisure-time, occupation, transportation, and household) [1]. Although sufficient total PA is beneficial for health [2–4], there is still not enough evidence to differentiate the effect of accumulating PA in different domains across health outcomes [5], as well as the contribution of each domain to the prevalence of active individuals. While PA carried out in leisure-time and/or transportation domains seems more associated with positive outcomes, such as decreased mortality from all causes and cardiovascular events and reduced incidence of type 2 diabetes [6]– [7], PA can be associated with adverse health outcomes when accumulated in the occupation domain [6], which encompasses PA carried out during work, whether paid or voluntary [1]. This phenomenon constitutes what is called the PA paradox [8].

Concerning the household domain, the evidence seems even more scarce and inconclusive. Studies in high-income countries have not found associations between PA in the household domain and all-cause mortality [9], cardiovascular outcomes [10]– [11], or overweight [12]. In China, associations were found with a decreased risk of hypertension [13], decreased body weight [14], and increased cognitive function [15], but not with diabetes [16].

In other Regions, like South America, 10 out of 12 countries on the continent do not even investigate household PA in representative national surveys that assess the population’s level of PA [17]. In Brazil, which is the biggest country in the region and one of the two that includes the household domain of PA in its national representative health survey, it is plausible to admit that household PA can contribute to total PA, especially among females, considering they spend almost twice as much time on domestic activities as male [18]. However, the extent to which the household domain of PA contributes to total PA and to the prevalence of active individuals in Brazil still needs to be investigated, as well as the associations between household PA and health outcomes, like cardiometabolic diseases and risk factors. The knowledge of the contribution of household PA can help to support more equitable public health recommendations, considering that an essential part of the population does not have the opportunity to accumulate more PA in the leisure-time domain [19]– [20].

Therefore, we aimed to analyze the contribution of PA performed in the household domain, its contribution to the prevalence of active adults and to the associations between total PA with cardiometabolic risk factors in Brazilian adults. We also investigated this contribution according to sociodemographic variables. Our hypothesis is that household PA domain contributes little to total PA and to decreases the associations with cardiometabolic risk factors in the adult population in general. However, for certain groups of the population, such as women, it may be more significant in relation to other groups, which corroborates the WHO’s guidance to accumulate PA in this domain particularly for specific subgroups.

Methods

Design and sample

This observational cross-sectional study uses secondary data from the Brazilian National Health Survey (Pesquisa Nacional de Saúde – PNS in Portuguese) 2019. The PNS is a national survey representative of the Brazilian population designed to investigate the health profile of those aged 15 and over. This is a household-based survey carried out through interviews with household residents. Sampling is complex and includes three stages as follows: (1) the first stage aimed to select the primary sampling units (census tracts); (2) the second stage aimed to select the households by simple randomization in each primary unit; (3) the final stage consisted of a simple random sampling of residents at least 15 years old, obtained from the list of residents created at the time of the interview. Census tracts, households, and residents were randomly selected. PNS was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Comissão Nacional de Ética em Pesquisa (CONEP)/Conselho Nacional de Saúde (CNS), protocol code 3.529.376 and issued on August 23, 2019. Informed consent was obtained from all subjects involved in the study. More details about the methods have been previously published elsewhere [21]. The initial sample included data from 94,114 respondents. After excluding individuals under 18 years old (2,431) and with missing data on the variables of interest (4,005), the total sample consisted of 87,678 Brazilian adults. We excluded individuals under 18 years old since the target population of this study was adults and because PA recommendations differ according to age. In the analysis of the contribution of household PA among individuals who practiced PA in at least one domain, individuals with 0 min were excluded, with the sample consisting of 66,822 adults. Finally, in the analysis of the contribution of household PA to achieving PA recommendations, only active individuals (who practiced at least 150 min/week) were considered, totaling 52,620 adults.

Physical activity variables

The PA questionnaire used in the PNS was based on the one employed by VIGITEL (Surveillance of Risk and Protective Factors for Chronic Diseases by Telephone Survey), which had been previously validated in Brazil [22], and includes questions about the number of days per week (frequency) and the duration in hours/minutes of PA carried out in the leisure-time domain (“In the last three months, have you practiced some type of physical exercise or sport?”), occupation domain (“At your job, do you walk a lot? At your job, do you do heavy cleaning, carry weight, or do another heavy activity that requires intense physical effort?”), transportation domain (“To get to or from work, do you walk or cycle any distance? In your usual activities (such as going to a course, school or club or taking someone to a course, school or club, how many days a week do you do any activity that involves walking or cycling? Except work”)) and household domain (“In your domestic activities, do you do heavy cleaning, carry weight or do any other heavy activity that requires intense physical effort? Do not consider paid domestic activity”). According to PNS procedures, for the leisure-time domain, PA practiced in the last 3 months was considered, while for the other domains, habitual PA was adopted.

To construct the variables, first, the number of minutes of weekly PA performed in each of the domains separately (leisure-time, occupation, transportation and household) were considered, by multiplying the number of days per week, hours per day and minutes per day (Days*((Hours*60) + Minutes). Then, to construct the variable in all domains (PA-4Domains), the sum of the weekly time spent in the four domains was considered, while for the PA variable in the three domains (PA-3Domains), the time of PA in the household domain was disregarded. Finally, the PA variables (PA-4Domains and PA-3Domains) were categorized into active individuals (≥ 150 min/week) and inactive individuals (< 150 min/week), according to the recommendation of PA for adults, proposed by the WHO [1] and the Brazilian physical activity guideline [23].

Cardiometabolic risk variables

The variables of hypertension, diabetes, and high cholesterol were self-reported based on the following questions in the PNS survey: “Has a doctor ever diagnosed you with hypertension (high blood pressure)/diabetes/high cholesterol?”. These variables were analyzed in a dichotomous manner (yes or no). Despite the underestimation of self-reported of hypertension, diabetes and hypercholesterolemia compared to direct measures, these self-reported conditions, which were grouped in the biological score, presented good validity coefficients for sensitivity (92%) and lower for specificity (30%) compared to blood pressure measurement and laboratory tests [24]. Obesity was classified based on the Body Mass Index (BMI) (≥ 30 kg.m²) [25], according to self-reported body mass (“Do you know your weight?”) and height (“Do you know your height? Even if it is an approximate value”). Although subject to bias, a previous study carried out with the Brazilian population demonstrated that, despite some exceptions, self-reported measures of weight, height and BMI offer valid and reliable measures in adults [26].

Sociodemographic variables

In the stratified analysis, the following sociodemographic variables and their respective categorizations were used: sex (male or female), age groups (18 to 34 years, 35 to 49 years, 50 to 64 years, or 65 years or more), according to previous studies using the same Brazilian national survey [19, 27], ethnicity (white, black, mixed or other), schooling (no formal instruction, less than secondary educational level, secondary or more than secondary; i.e.,≥ high school degree), per capita household income (≤ 1 minimum wage, > 1 and ≤ 5 minimum wage or > 5 minimum wages), area (Urban or Rural), type of city (Capital or Non-capital) and country Region (North, Northeast, Southeast, South or Central-Western).

Confounders variables

In the adjusted analyses, in addition to sociodemographic variables, other confounding variables were used, chosen in accordance with previous studies carried out with Brazilians [27, 28]. The variables and their respective categorizations were: tobacco (current smoker, yes or no), alcohol (no/less than once a month or more than once a month), fruits (no/less than five a week or five/more a week), sugary drinks (no/less than once a week or more than once a week), and BMI (< 30 kg.m² or ≥ 30 kg.m²).

Statistical analysis

For the comparison of active individuals (PA-4Domains versus PA-3Domains), prevalences and their respective 95% confidence intervals (95%CI) in each variable were calculated using frequency distribution, considering the proportion of individuals within each category of the sociodemographic variables and cardiometabolic risk factors, using the command “svy linearized: proportion” of Stata. The absolute differences between prevalences were presented in percentage points (p.p.), inferring their significance in comparing the prevalence of PA-4Domains and PA-3Domains using the McNemar test. To assess the contribution of time spent on household PA to total PA time, the percentual contribution for each individual was first calculated by dividing the number of minutes per week of household PA by the number of minutes per week of total PA ((household PA/total PA)*100)). Then, the mean of these percentual, with their respective standard deviations were calculated, according to the stratifications of sociodemographic variables and cardiometabolic risk factors, analyzing all individuals and individuals who perform some PA (PA-4Domains > 0 min). To compare the dependence on PA household to achieve minimum PA recommendations, absolute and relative values were presented, with the distribution compared within each category using the chi-square test. To calculate the associations between categories and between PA-4Domains, PA-3Domains, and cardiometabolic risk factors, crude and adjusted Poisson regression models with robust variance were performed, presenting prevalence ratios (PR) and their respective 95% Confidence Interval (CI), using the command “svy” of Stata. In adjusted analyses, all confounding variables were used. BMI was used in all analyses except for obesity. The analyses between PA-4Domains, PA-3Domains, and cardiometabolic risk factors were stratified by sex because this variable shows the highest difference between prevalences and percentage of the contribution of PA-Household observed in this study. Furthermore, in Brazil women tend to do more activities in the domestic environment than men [18]. The sampling weight was considered in all analyses. The significance level adopted was less than 5%, and the statistical package used was Stata 15.0 software.

Results

Table 1 presents the distribution of the sample, stratified by sociodemographic characteristics and cardiometabolic risk factors, in addition to comparing the prevalence of active individuals considering PA performed in all domains (PA-4Domains) and PA performed in the three domains, disregarding the household one (PA-3Domains). Considering the total sample, a significant reduction of 4.3 p.p. in the prevalence of active individuals was observed when the household domain was not considered. In stratified analyses, a decrease in prevalence was observed in all categories of the variables analyzed, with more significant reductions among females (−6.8 p.p.), individuals with obesity (−5.4 p.p.), with high cholesterol (−5.1 p.p.), from rural area and Central-western Region (−5.0 p.p. each), aged 50 to 64 years (−4.8 p.p.), aged 35 to 49 years (−4.7 p.p.), with secondary schooling (−4.7 p.p.) and with income ≤ 1 minimum wage (−4.7 p.p.).

Table 1.

Comparison between the prevalence of active individuals considering total physical activity (PA-4Domains) and only the three domains, except household (PA-3Domains), according to sociodemographic characteristics and cardiometabolic risk factors

Sociodemographic variables and cardiometabolic risk factors		Active PA-4Domains		Active PA-3Domains		Absolute difference
	n	%	95%CI	%	95%CI	p.p.
All	87,678	62.6	62.1–63.2	58.3	57.7–58.9	−4.3*
Sex
Male	41,662	67.1	66.3–67.9	65.5	64.7–66.3	−1.6*
Female	46,016	58.7	57.9–59.4	51.9	51.1–52.7	−6.8*
Age groups
18 to 34 years	23,459	69.1	68.0-70.1	64.9	63.8–66.0	−4.2*
35 to 49 years	25,844	69.2	68.2–70.1	64.5	63.5–65.5	−4.7*
50 to 64 years	22,451	61.5	60.4–62.6	56.7	55.6–57.9	−4.8*
65 years or more	15,924	38.2	36.9–39.5	34.9	33.6–36.2	−3.3*
Ethnicity
White	32,161	61.7	60.8–62.6	57.2	56.3–58.2	−4.5*
Black	10,017	66.3	64.7–67.8	62.4	60.8–63.9	−3.9*
Mixed	44,169	62.7	61.8–63.4	58.3	57.5–59.1	−4.4*
Other	1,322	60.4	55.1–65.4	57.4	52.1–62.5	−3.0*
Schooling
No formal instruction	7,614	38.6	36.5–40.7	35.8	33.8–37.9	−2.8*
Less than secondary	39,598	58.9	58.1–59.8	54.8	53.9–55.6	−4.1*
Secondary	26,980	68.2	67.2–69.2	63.5	62.5–64.5	−4.7*
More than secondary	13,486	69.8	68.4–71.1	65.3	63.8–66.6	−4.5*
Per capita household income
≤ 1 minimum wage	47,715	60.3	59.5–61.0	55.6	54.8–56.4	−4.7*
> 1 and ≤ 5 minimum wage	34,931	64.8	63.9–65.6	60.6	59.7–61.5	−4.2*
> 5 minimum wage	5,032	68.2	65.9–70.3	65.6	63.3–67.9	−2.6*
Area
Urban	67,567	63.3	62.7–63.9	59.1	58.4–59.7	−4.2*
Rural	20,111	58.5	57.4–59.5	53.5	52.4–54.6	−5.0*
Type of city
Capital	31,806	64.6	63.7–65.4	60.5	59.7–61.4	−4.1*
Non-capital	55,872	62.0	61.3–62.7	57.6	56.9–58.3	−4.4*
Region
North	16,722	58.9	57.7–60.1	55.0	53.7–56.2	−3.9*
Northeast	30,415	59.6	58.7–60.4	55.6	54.7–56.5	−4.0*
Southeast	19,285	65.4	64.3–66.4	60.8	59.7–61.8	−4.6*
South	11,186	62.0	60.7–63.2	57.9	56.6–59.1	−4.1*
Central-Western	10,070	62.9	61.5–64.2	57.9	56.5–59.3	−5.0*
Cardiometabolic risk factors
Hypertension Yes	22,496	52.5	51.4–53.6	48.2	47.1–49.4	−4.3*
Hypertension No	65,182	65.9	65.2–66.5	61.5	60.8–62.2	−4.4*
Diabetes Yes	7,076	46.7	44.7–48.8	43.2	41.2–45.2	−3.5*
Diabetes No	80,602	64.0	63.4–64.6	59.6	59.0-60.2	−4.4*
High cholesterol Yes	13,349	57.9	56.4–59.3	52.8	51.3–54.2	−5.1*
High cholesterol No	74,329	63.5	62.9–64.1	59.3	58.6–59.9	−4.2*
Obesity Yes	18,026	59.8	58.5–61.0	54.4	53.1–55.7	−5.4*
Obesity No	69,652	63.4	62.8–64.0	59.4	58.7–60.0	−4.0*

PA-4Domains physical activity in all domains, PA-3Domains physical activity in all domains, except the household domain, p.p. percentual points, 95%CI 95% confidence interval; *p < 0.05 in comparing PA-4Domains and PA-3Domains, for all variables investigated, using the McNemar test

Table 2 presents the mean percentage values of household PA contribution to total PA. Considering the total sample (n = 87,678), the mean contribution of household PA time in total PA time was 7.7%, with the highest contribution percentages being found among females, 35 and 49 years old, white and mixed, secondary schooling, lower income, rural area, non-capital cities, Central-Western Region, without hypertension and diabetes, and with high cholesterol and obesity. When considering only individuals who perform PA in at least one of the domains (n = 66,822), the mean contribution of household PA time in total PA time was 9.9%. Observing the categories of sociodemographic variables, the highest average contribution percentages were found in the same categories as the total sample, except in hypertension and diabetes, whose contribution percentages were highest in individuals with the diagnosis.

Table 2.

Mean household physical activity time percentage contribution to total physical activity time, according to sociodemographic characteristics and cardiometabolic risk factors

Sociodemographic variables and cardiometabolic risk factors	All individuals		Individuals who do any PA
	n	Mean (SD)%	n	Mean (SD)%
All	87,678	7.7 (22.1)	66,822	9.9 (24.5)
Sex
Male	41,662	3.4 (14.7)	32,819	4.2 (16.3)
Female	46,016	11.6 (26.5)	34,003	15.2 (29.4)
Age groups
18 to 34 years	23,459	7.3 (21.3)	19,487	8.7 (23.0)
35 to 49 years	25,844	9.0 (23.0)	21,220	10.8 (24.8)
50 to 64 years	22,451	8.2 (23.1)	17,138	10.6 (25.7)
65 years or more	15,924	5.4 (19.9)	8,977	9.2 (25.5)
Ethnicity
White	32,161	7.8 (22.3)	24,364	10.0 (24.8)
Black	10,017	7.6 (21.1)	7,802	9.5 (23.1)
Mixed	44,169	7.8 (22.3)	33,632	9.9 (24.7)
Other	1,322	5.6 (18.6)	1,017	7.5 (21.2)
Schooling
No formal instruction	7,614	4.9 (19.0)	4,247	8.6 (24.6)
Less than secondary	39,598	7.4 (22.1)	29,261	9.9 (25.0)
Secondary	26,980	8.4 (22.6)	21,939	10.1 (24.4)
More than secondary	13,486	8.2 (22.2)	11,375	9.6 (23.8)
Per capita household income
≤ 1 minimum wage	47,715	8.1 (22.8)	35,309	10.6 (25.5)
> 1 and ≤ 5 minimum wage	34,931	7.6 (21.8)	27,298	9.5 (24.0)
> 5 minimum wage	5,032	4.9 (17.3)	4,215	5.8 (18.6)
Area
Urban	67,567	7.7 (21.8)	52,570	9.7 (24.1)
Rural	20,111	8.2 (23.9)	14,252	11.2 (27.3)
Type of city
Capital	31,806	7.5 (21.5)	24,953	9.4 (23.7)
Non-capital	55,872	7.8 (22.3)	41,869	10.0 (24.8)
Region
North	16,722	6.9 (21.4)	12,273	9.3 (24.4)
Northeast	30,415	7.0 (21.3)	23,007	9.1 (23.8)
Southeast	19,285	8.3 (22.6)	15,155	10.3 (24.8)
South	11,186	7.4 (21.7)	8,654	9.5 (24.2)
Central-Western	10,070	8.4 (23.4)	7,733	10.7 (25.9)
Cardiometabolic risk factors
Hypertension Yes	22,496	7.4 (22.1)	15,185	10.6 (25.8)
Hypertension No	65,182	7.8 (22.1)	51,637	9.7 (24.2)
Diabetes Yes	7,076	6.6 (21.1)	4,349	10.4 (25.7)
Diabetes No	80,602	7.8 (22.2)	62,473	9.8 (24.5)
High cholesterol Yes	13,349	9.0 (24.0)	9,787	11.9 (26.9)
High cholesterol No	74,329	7.5 (21.8)	57,035	9.5 (24.1)
Obesity Yes	18,026	9.2 (24.2)	13,200	12.3 (27.1)
Obesity No	69,652	7.3 (21.5)	53,622	9.2 (23.8)
PA Physical activity, SD Standard Deviation

Table 3 presents the absolute and percentage number of active individuals who depend on household PA to achieve 150 min/week of PA. Significant differences were observed in the distributions of individuals in all sociodemographic variables investigated. Considering the prevalence ratio values in adjusted analyses, females were almost five times more likely to depend on household PA than males. Older individuals are more likely to depend on household PA than younger people, between 30% and 53%. Regarding income, those who earned ≤ 1 minimum wage were more than twice as likely as those who earned more than five wages. Significant prevalence ratio values were also found in individuals who earned > 1 and < 5 minimum wages compared to those who earned more than 5. Those who live in rural areas are 48% more likely to depend on household PA than the urban area, and those who live in the Central-Western Region have more than 36% than those who live in the Northeast Region. Those who had obesity were 28% more likely to depend on household PA than those who did not have this outcome.

Table 3.

Comparison of active individuals who depend and do not depend on household physical activity to be active, according to sociodemographic characteristics and cardiometabolic risk factors

Sociodemographic variables and cardiometabolic risk factors	Active individuals n	Depend on household PA n (%)	Crude model PR (95%CI)	Adjusted model 1 PR (95%CI)	Adjusted model 2 PR (95%CI)
All	52,620	3489 (6.9%)*
Sex*
Male	26,949	674 (2.4%)	Ref	Ref	-
Female	25,671	2815 (11.6%)	4.87 (4.24–5.61)	4.92 (4.27–5.68)	-
Age groups*
18 to 34 years	15,895	1036 (6.0%)	Ref	Ref	-
35 to 49 years	17,493	1104 (6.7%)	1.13 (0.98–1.29)	1.05 (0.92–1.21)	-
50 to 64 years	13,436	902 (7.8%)	1.30 (1.12–1.52)	1.30 (1.12–1.52)	-
65 years or more	5796	447 (8.7%)	1.45 (1.20–1.75)	1.53 (1.26–1.85)	-
Ethnicity*
White	19,140	1355 (7.3%)	1.49 (0.97–2.28)	1.52 (0.99–2.35)	-
Black	6323	368 (5.8%)	1.20 (0.77–1.87)	1.20 (0.76–1.88)	-
Mixed	26,344	1710 (6.9%)	1.41 (0.92–2.16)	1.38 (0.90–2.13)	-
Other	806	56 (4.9%)	Ref	Ref	-
Schooling*
No formal instruction	2872	191 (7.1%)	1.10 (0.83–1.46)	0.88 (0.64–1.20)	-
Less than secondary	22,658	1595 (7.1%)	1.09 (0.92–1.29)	0.94 (0.78–1.13)	-
Secondary	17,767	1179 (6.9%)	1.06 (0.89–1.27)	0.99 (0.82–1.20)	-
More than secondary	9323	524 (6.5%)	Ref	Ref	-
Per capita household income*
≤ 1 minimum wage	27,311	2103 (7.7%)	2.08 (1.51–2.86)	2.28 (1.59–3.28)	-
> 1 and ≤ 5 minimum	21,856	1277 (6.4%)	1.74 (1.26–2.40)	1.80 (1.28–2.55)	-
> 5 minimum wage	3453	109 (3.7%)	Ref	Ref	-
Area*
Urban	41,421	2595 (6.7%)	Ref	Ref	-
Rural	11,199	894 (8.5%)	1.27 (1.13–1.43)	1.48 (1.31–1.67)	-
Type of city*
Capital	19,968	1195 (6.3%)	Ref	Ref	-
Non-capital	32,652	2294 (7.1%)	1.14 (1.03–1.27)	1.07 (0.95–1.20)	-
Region*
North	9771	585 (6.7%)	1.01 (0.87–1.18)	1.10 (0.94–1.27)	-
Northeast	17,615	1177 (6.6%)	Ref	Ref	-
Southeast	12,172	793 (7.1%)	1.06 (0.93–1.21)	1.16 (1.02–1.33)	-
South	6875	505 (6.6%)	0.99 (0.86–1.14)	1.05 (0.90–1.22)	-
Central-Western	6187	429 (7.8%)	1.18 (1.00-1.38)	1.36 (1.16–1.59)	-
Cardiometabolic risk factors*
Hypertension Yes	11,086	840 (8.1%)	1.22 (1.08–1.38)	0.99 (0.87–1.13)	0.96 (0.84–1.10)
Hypertension No	41,534	2649 (6.6%)	Ref	Ref	Ref
Diabetes Yes	3096	235 (7.5%)	1.09 (0.89–1.35)	0.90 (0.73–1.12)	0.90 (0.73–1.11)
Diabetes No	49,524	3254 (6.9%)	Ref	Ref	Ref
High cholesterol Yes	7405	614 (8.8%)	1.33 (1.16–1.53)	1.05 (0.92–1.21)	1.05 (0.92–1.21)
High cholesterol No	45,215	2875 (6.6%)	Ref	Ref	Ref
Obesity Yes	10,193	858 (8.9%)	1.40 (1.23–1.59)	1.27 (1.11–1.44)	1.28 (1.13–1.46)
Obesity No	42,427	2631 (6.4%)	Ref	Ref	Ref

PA Physical activity, PR Prevalence Ratio, 95%CI 95% confidence interval, Ref reference category. Adjusted model 1– Adjusted by all sociodemographic variables (sex, age groups, ethnicity, schooling, per capita household income, area, type of city, and country region). Adjusted model 2– Adjusted by all sociodemographic and other confounders variables (sex, age groups, ethnicity, schooling, per capita household income, area, type of city, country region, tobacco, alcohol, fruits, sugary drinks, and BMI, except for obesity); *p < 0.05 in comparing who depend and do not depend on household physical activity to be active within each variable, using the chi-squared test

Values highlighted in bold were considered significant

Table 4 presents crude and adjusted models of the associations between PA performed in all domains (PA-4Domains) and the three domains (PA-3Domains), with cardiometabolic risk factors (hypertension, diabetes, high cholesterol, and obesity) in the total sample and stratified by sex. Being inactive in PA-4Domains was directly associated with hypertension and obesity (except for females) and diabetes. Also, PA-4Domains was inversely associated with high cholesterol (only in females). Similar results were found with PA-3Domains, except for females with high cholesterol and obesity, whose associations were respectively insignificant and significant when compared to PA-4Domains. Despite these changes in the significance of values, the magnitude of associations found in PA-4Domains with the analyzed cardiometabolic risk factors remained when only the PA-3Domains were considered.

Table 4.

Association between total physical activity (PA-4Domains) and only the three domains, except household (PA-3Domains) with cardiometabolic risk factors, according by sex

Cardiometabolic risk factors		PA-4Domains PR (95%CI)		PA-3Domains PR (95%CI)
		Active	Inactive	Active	Inactive
Crude model
Hypertension	All	Ref	1.52 (1.46–1.58)	Ref	1.50 (1.44–1.56)
	Male	Ref	1.51 (1.42–1.61)	Ref	1.50 (1.40–1.59)
	Female	Ref	1.47 (1.40–1.55)	Ref	1.44 (1.36–1.51)
Diabetes	All	Ref	1.91 (1.77–2.06)	Ref	1.84 (1.70-2.00)
	Male	Ref	1.96 (1.74–2.20)	Ref	1.91 (1.70–2.15)
	Female	Ref	1.83 (1.65–2.03)	Ref	1.72 (1.54–1.92)
High cholesterol	All	Ref	1.22 (1.15–1.29)	Ref	1.25 (1.18–1.32)
	Male	Ref	1.22 (1.11–1.33)	Ref	1.22 (1.11–1.35)
	Female	Ref	1.15 (1.07–1.23)	Ref	1.15 (1.08–1.23)
Obesity	All	Ref	1.13 (1.08–1.18)	Ref	1.17 (1.12–1.23)
	Male	Ref	1.20 (1.11–1.29)	Ref	1.20 (1.11–1.33)
	Female	Ref	1.05 (0.99–1.12)	Ref	1.11 (1.04–1.18)
Adjusted model
Hypertension	All	Ref	1.06 (1.02–1.10)	Ref	1.06 (1.01–1.10)
	Male	Ref	1.09 (1.03–1.16)	Ref	1.09 (1.02–1.15)
	Female	Ref	1.03 (0.98–1.08)	Ref	1.02 (0.98–1.07)
Diabetes	All	Ref	1.25 (1.16–1.35)	Ref	1.22 (1.12–1.32)
	Male	Ref	1.34 (1.19–1.51)	Ref	1.32 (1.17–1.48)
	Female	Ref	1.17 (1.05–1.30)	Ref	1.13 (1.01–1.26)
High cholesterol	All	Ref	0.97 (0.92–1.03)	Ref	0.99 (0.94–1.05)
	Male	Ref	1.03 (0.93–1.13)	Ref	1.04 (0.94–1.14)
	Female	Ref	0.92 (0.86–0.99)	Ref	0.95 (0.88–1.01)
Obesity	All	Ref	1.14 (1.09–1.20)	Ref	1.18 (1.13–1.24)
	Male	Ref	1.26 (1.16–1.35)	Ref	1.25 (1.16–1.35)
	Female	Ref	1.04 (0.98–1.11)	Ref	1.10 (1.03–1.17)

PA-4Domains = physical activity practicing in all domains (leisure-time, work, transportation, and household), considering activity ≥ 150 min/week; PA-3Domains = physical activity practicing in three domains, except household domain, considering activity ≥ 150 min/week. PR = Prevalence Ratio; 95%CI = 95% confidence interval; Ref = reference category. Adjusted model considering sex (only in all individuals), age groups, ethnicity, schooling, per capita household income, area, type of city, country region, tobacco, alcohol, fruits, sugary drinks, and BMI (except for obesity). Values highlighted in bold were considered significant

Discussion

In this novel study based on nationally representative data, our study revealed that household PA has a modest contribution of around 4 p.p. to the prevalence of active individuals, mainly among females. Also, the mean contribution of household PA to total PA is close to 10%, but the percentage of females’ contributions is more than three times higher than males. Stratified analyses showed that the subgroups that most needed household PA to meet the minimum recommended PA were females, older individuals, those with an income of up to 1 minimum wage, those living in rural areas, Central-Western Region, and those with obesity. Also, in general, the household domain did not change the association between PA and cardiometabolic risk factors. To the best of our knowledge, this is the first study to comprehensively evaluate the contribution of the household PA based on a nationally representative health survey of a country, Brazil, which is one of the top 10 in the world in terms of the economy and population, with high inequality in the access and opportunities for PA [29].

Despite the differences in the prevalence of active individuals, when household PA is considered or not, the interpretation of its magnitude is not simple to define. When comparing the prevalence of active individuals among countries, the overall difference of 4.3 p.p. might have a small impact since it might be smaller than the usual 95% confidence intervals reported in PA prevalence studies, such as in South America, which variation for total PA, considering all countries, was 7.5 p.p [30]. Also, the mean contribution of HPA in this nationally representative study could be considered modest (7.7% for the entire sample and 9.9% among those with some PA during the week). On the other hand, in the comparison between groups/countries and/or in terms of trends of physical activity over time, this magnitude of contribution might be considered.

Even though WHO recommends the accumulation of PA in all domains, it is quite reasonable to consider household PA as a type of activity that must be done by any house resident rather than a domain in which PA should be increased as a good way of achieving the minimum PA per week. So, in that scenario, our findings indicate that the percentage of household PA usually performed by adults tends to be below 10% of the total amount of PA accumulated during the week, which was very similar among the different sociodemographic groups, except for females (15.2%) and those with low income (10.6%) (Table 2).

Considering the daily time distribution (e.g., hygiene, family carries, transport, work, leisure, etc.), it is not surprising the relatively low contribution of household PA to the total PA. However, we could identify structural differences in that distribution, such as sex and income differences. Even though still around 10%, the percentage of contribution of household PA was around three times higher among females than males. Also, one can consider that the magnitude of the contribution of household PA should be considered crucial since the prevalence of being physically active depending on household PA was around 5-fold higher for females, more than 2-fold higher for those with lower incomes, and almost 50% higher for older individuals and those leaving on rural areas. Simultaneously, the finding that the association between PA and the evaluated cardiometabolic risks was almost unaffected by household PA reinforces the understanding that household PA should be considered as having only a modest impact. Again, depending on the context, such as in longitudinal or comparative studies, the impact of household PA might be considered.

Our findings align with results found with a sample of adults in Northern Ireland, where household PA, mainly among female and older adults, was responsible for a significant proportion of moderate to vigorous daily activity [31]. In Brazil, a previous study [20] had already identified sociodemographic correlates of engagement for household PA, identifying greater engagement among females compared to males, the age group of 25 to 59 years compared to younger people, and among individuals with intermediate education compared to those with less formal education. The authors also identified lower engagement of household PA among individuals with higher incomes compared to those with lower ones. Nevertheless, our study advanced the understanding of how much the household domain contributed to total PA recommendations in these categories, reinforcing differences and possible social inequalities between groups.

A previous study carried out with adults from Brazilian capitals confirms that, of all the PA domains, the household domain is the only one whose prevalence among females is higher than among males [32]. Although the difference between males and females in overall household activity has decreased over the years, Brazilian females still spend almost twice as much time on domestic tasks as males [18]. Among these activities, vigorous PA, such as heavy cleaning, helps females achieve the minimum PA recommended by the WHO [12, 31, 33]. But, the extent to which this type of PA may be associated with health outcomes still needs to be further investigated, as there are psychological, emotional, social, and economic factors that are linked to the performance of this type of activity and which can compromise physical and mental health, in parallel, or even in contradiction, to the benefits of increased energy expenditure [12, 31, 33].

Regarding the specific associations between total PA and PA accumulated only in three domains (except the household one), we observed that when the household domain was disregarded, being physical inactivity became significantly associated with obesity only among females. On the other hand, high cholesterol was only associated with being physical inactivity when all domains were considered, but the direction of this association was inverse, that is, being inactive considering all domains was associated with not having high cholesterol. The finding of the first result could mean that the energy expenditure promoted by household PA may contribute to reducing the likelihood of obesity outcome among females, who are the stratum, in our data, that most needed household PA to meet the total PA recommendations. In spite of that, because of the second result, the interpretation should be made with caution, because in addition to the small differences between the magnitudes of the associations, there may be other factors, which were not taken into account in this study, that have modified the strength of these associations.

Generally speaking, we can interpret that Individuals who do not comply with PA recommendations in all domains are more likely to present cardiometabolic risk factors, except high cholesterol. The magnitude of these associations remains even when time spent on household PA is disregarded. In this way, even contributing to an increase in the total prevalence of active individuals, the household domain does not appear to have a significant effect on the relationship between PA and the evaluated cardiometabolic risk factors. The general lack of impact of household PA on the association between PA and the evaluated cardiometabolic risk factors reinforces, at least theoretically, the interpretation that this is a type of PA that someone must do but that it does not necessarily improve cardiometabolic health despite its contribution to higher energy expenditure. This is a critical hypothesis that emerged from our data but deserves further studies specifically designed to address this complex issue.

Our study has some limitations. Firstly, all variables investigated were obtained through self-report, which may significantly overestimate the time spent in PA in each domain. Also, compared to leisure and transport domains, the household PA tends to be less clear as a manifestation of PA, which could impact the estimative, especially when using this as a continuous indicator [34]. However, the fact that the PNS has a single specific question about household PA mitigates the difficulty of reporting activities in this domain. Furthermore, a study with data from the 2013 PNS [24] shows that, despite discrepancies found between self-reported measures of hypertension, diabetes, and hypercholesterolemia and actual health conditions, the biological score, which includes these measures, in addition to smoking, showed a sensitivity of 92% and specificity of 30% in relation to direct measures. Also, self-reported measures of health conditions have been widely used in other studies and population surveys, mainly due to the difficulty of performing direct measures on a population scale. Secondly, the PNS does not differentiate the intensity of each PA between moderate and vigorous performed in each domain; it only classifies the intensity of leisure-time PA based on the modality. Therefore, using the cutoff point of 75 min of vigorous PA to classify active individuals was impossible. Despite that, a study based on three different random and representative Brazilian population samples did not show a significant difference between the prevalence of PA when time spent on vigorous PA was doubled or not, which mitigates the possible impact of the inherent limitation in our findings [35]. Thirdly, the fact that the associations were made with categories. Although continuous analysis may be preferable in terms of statistical analysis, we believe it is relevant for our study to consider the recommendations for PA as a cutoff point for the prevalence of active individuals. Furthermore, as this is a secondary study, many of the covariates used were already categorized. Finally, as this is a cross-sectional study, inferring causality in the associations is impossible. Also, causality is beyond our scope, which minimizes its effects on the study aims. Despite this, the present study analyzed data from a nationally representative sample, which, due to the sampling methodology, may allow the findings to be generalizable to the Brazilian adult population. The study also provides evidence from a middle-income country whose lack of research is known [36].

Conclusion

Physical activity in the household domain impacted the prevalence of active Brazilian adults, especially among females, but contributed modestly to the total amount of physical activity and achieving physical activity recommendations. Besides, household physical activity did not modify the associations between total physical activity and the cardiometabolic risk factors investigated. Our findings reinforce the need for caution in interpretation and comparison between studies that consider different domains of physical activity, and further studies are needed to understand better the association of household physical activity with health outcomes and its interaction with other domains of physical activity. To this end, using the South American example, where only Brazil and Peru include the assessment of physical activity in the household domain in their national health surveys, our findings must be considered when the prevalence of physical activity levels from different countries in this region are compared. Also, concerns must be considered when subgroups are evaluated, mainly when categorized by sex and income.

Authors' contributions

LLSB: Concept and study design, data analysis, interpretation of the data and drafted the initial manuscript. LGGP: Concept and study design, interpretation of the data, critical revision and approval of the manuscript for important intellectual content. MRA, and DRPS: critical revision and approval of the manuscript for important intellectual content. All authors have read and approved the final version of the manuscript and agree with the order of presentation of the authors.

Funding

L.B. was supported by the Coordination for the Improvement of Higher Education Personnel (CAPES) with a Ph.D. scholarship (CAPES process:88887.694146/2022-00 - Finance Code 001). This research received financial support from the public notice UnB-FEF-PPGEF nº. 07/2024 This paper presents independent research. The views expressed in this publication are those of the authors and not necessarily those of the acknowledged institution.

Data availability

Data from the National Health Survey is available on the Brazilian Institute of Geography and Statistics website (https://www.ibge.gov.br/en/home-eng.htmlhttps://www.ibge.gov.br/en/home-eng.html).

Declarations

Ethics approval and consent to participate

All procedures performed in the original study involving human participants were approved by the Comissão Nacional de Ética em Pesquisa (CONEP: 3.529.376).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.