text new page (beta)
English (pdf)
Article in xml format
Curriculum ScienTI
Automatic translation
Cited by SciELO 
Cited by Google 
Similars in
SciELO 
Similars in Google 
Permalink
Introduction
Multimorbidity is the presence of two or more diagnosed diseases 1. Multimorbidity is common among older adults 2, but expressive rates of this condition are seen in adults. Chronic noncommunicable diseases (NCDs) individually reduce life expectancy 3 and are currently the leading death causes worldwide 4. The coexistence of NCDs aggravates the problems and can result in loss of quality of life and functional capacity, with a consequent increase in the use of health services, in the public spending and in the risk of death 2,5.
A strong association of NCD multimorbidity with behavior factors has been suggested 6. Among these factors, physical activity and sedentary behavior are important determinants for the development of NCDs and should not be confused since it has been established that they are independent behaviors characterized by distinct metabolic responses 7. In this respect, physical activity is considered an important health-protecting factor because it exerts a preventive effect even when performed at intensities below the current health-related recommendations (moderate to vigorous activities) 8. This effect appears to be greatly enhanced when the levels of activity increase 9. In contrast, sedentary behavior is a health risk factor 5, regardless of participation in physical activity, which is related to death from cardiovascular diseases (aggravation of NCD multimorbidity) and associated with the simultaneous diagnosis of different cardiovascular risk indicators 10,11.
Considering that health is a complex of dynamic and interconnected organ systems in which different diseases can occur simultaneously according to the influence of hereditability and behaviors adopted, this study aimed to investigate the association of NCD multimorbidity with physical activity and sedentary behavior in a representative Brazilian population.
Methods
This was a cross-sectional study, which performed a secondary analysis of data from the annual national survey Risk and Protective Factors Surveillance System for Chronic Non-Communicable Diseases Therough Telephone Interview (VIGITEL - Vigilância de Fatores de Risco e Proteção para Doenças Crônicas por Inquérito Telefônico). Between February and December 2013, data were collected in all 27 capitals of the Brazilian federative units from adults older than 18 years, who had a fixed residential telephone line.
The sampling was performed in three steps 12. Each capital had 2,000 interviewees as minimum sample; therefore, the variables could be estimated with a 95% confidence interval (95%CI) and a 3% maximum error. Weighting factors were used to compensate for bias of nonuniversal fixed-line coverage, adjusted to the adult Brazilian population based on the weight of each individual of the sample calculated by the raking method 12. The data were collected by telephone interview simultaneously using a computer, and all the calls were recovered. The supervisor checked, listed and made a re-interview in 10% of the interviews for quality control. The instrument applied, based on previous surveys 12, had been validated in pilot studies 13,14,15 with satisfactory validity and reliability considering five different cities in each macroregion of Brazil 14, and contained questions about sociodemographic, behavioral, nutritional, and health factors. The section used to assess physical activity was created based on instructions and models from the World Health Organization 16 and the centers for disease control and prevention 17, being validated for Brazilians 18.
The outcome variable was NCD multimorbidity, which was categorized in two ways according to the analysis proposed. For multinomial regression, the outcome had four categories: no occurrence (zero or one disease) and occurrence (two, three, or four diseases) of multimorbidity. For interaction analysis, the outcome was defined by the absence of disease and the presence of two or more diseases. To create the NCD variable, the four most prevalent noncommunicable diseases in Brazil were used, which were diabetes, dyslipidemia, arterial hypertension, and obesity. Diabetes, dyslipidemia, and arterial hypertension were considered when the participant answered yes to the following questions: “Has any doctor ever told you that you have diabetes?”; “Has any doctor ever told you that you have dyslipidemia?”; “Has any doctor ever told you that you have high blood pressure?”. Obesity was defined as a body mass index of 30kg/m2 or higher, calculated based on self-reported weight and height. For this variable, hot deck imputation of the data was used to include the blank answers (8.8%) in the analysis. The control variables were sex, age, marital status, ethnicity, demographic region, and education level. The following exposure variables were used: physical activity performed at least once a week in the commuting, domestic chore, leisure-time, and work domains; total physical activity level in the leisure-time and commuting domains categorized into inactive (none), insufficiently active (10-149 min/week), and active (≥ 150 min/week), and daily television time categorized into < 2 hours and ≥ 2 hours. We considered these activities because they are usually part of routine activities 19, and this cut-off for daily television time has been suggested in a previous study 20.
The Stata software (https://www.stata.com) was used for data statistical analysis. For descriptive analysis, absolute and relative frequencies were calculated, considering the prevalence estimates and 95%CI. For adjusted inferential analysis, multinomial logistic and Poisson regressions were performed, in which the data are expressed as odds ratio (OR) and prevalence ratio (PR), respectively. The first regression aimed to associate behavioral indicators with multimorbidity according to the number of existing diseases. Interaction analysis, defined by describing interdependent effects of two or more variables included in the model, considered total physical activity level and daily television time in multimorbidity occurrence. A multiplicative model was adopted, in which the product derived from the crossing of the tested variables was used in the Poisson regression with robust variance, considering the p-value of the Wald test for heterogeneity. Fitting the regressions, the first-level variables were sex, age, marital status, ethnicity, demographic region, and education level; the second-level were commuting, domestic shore, leisure-time and work, physical activity, total physical activity, and daily television time. Backward selection was adopted for statistical modeling, with a p ≤ 0.20 critical level for the variable to remain in the hierarchical regression model. A 5% significance level was adopted for all tests. All analyses considered the sample weight obtained by the inverse of the number of telephone lines existing in the interviewed household and the number of adults living in the interviewee’s household.
The National Research Ethics Committee of the Ministry of Health approved this study (protocol number 355.590/2013). All participants consented verbally to participate in the study, via telephone call.
Results
Among the eligible subjects (n = 74,005), 52,929 participated in the study (response rate of 71.5%). The adults’ mean age was 36.2 ± 11.2 years and most of them were women (52.9%) and lived without a partner (52%). Among the older adults, the mean age was 69.4 ± 13.3 years. Women (59.5%) and individuals living with a partner (56.9%) were predominant. The multimorbidity frequency of two, three, and four NCDs was 9.8%, 3.3%, and 0.6% in the group of adults, respectively, and 28%, 12.6%, and 2.7% among older adults.
Most adults did not perform physical activity in the domestic chore (59.9%), commuting (61.9%), work (56.4%), or leisure-time (53.3%) domain. These percentages were 66.8%, 86.6%, 82.8%, and 63.1%, respectively, among older adults. For total physical activity, most adults were physically active (45.5%) and most older adults were inactive (58.5%). In addition, adults and older adults watched television two or more hours per day (50.9% and 55.1%, respectively) (Table 1).
Table 1 Behavioral variables stratified by age. Brazil, 2013 (n = 52,929).
| Variable | Adults (n = 37,947) | Older adults (n = 14,982) | ||||||
|---|---|---|---|---|---|---|---|---|
| n | % * | 95%CI ** | Missing (%) | n | % * | 95%CI ** | Missing (%) | |
| Domestic physical activity | 0.3 | 0.4 | ||||||
| No | 21,981 | 59.9 | 59.0; 60.9 | 10,205 | 66.8 | 65.2; 68.4 | ||
| Yes | 15,903 | 40.1 | 39.1; 41.0 | 4,725 | 33.2 | 31.6; 34.8 | ||
| Commuting physical activity | 0.7 | 1.0 | ||||||
| No | 26,135 | 61.9 | 60.9; 62.9 | 13,248 | 86.6 | 85.3; 87.9 | ||
| Yes | 11,539 | 38.1 | 37.1; 39.1 | 1,588 | 13.4 | 12.1; 14.7 | ||
| Work physical activity | 0.0 | 0.0 | ||||||
| No | 22,895 | 56.4 | 55.4; 57.4 | 12,781 | 82.8 | 81.4; 84.2 | ||
| Yes | 15,038 | 43.6 | 42.6; 44.6 | 2,198 | 17.2 | 15.8; 18.6 | ||
| Leisure-time physical activity | 0.0 | 0.0 | ||||||
| No | 18,868 | 53.3 | 52.3; 54.3 | 8,141 | 63.1 | 61.5; 64.7 | ||
| Yes | 19,079 | 46.7 | 45.7; 47.7 | 6,568 | 36.9 | 35.3; 38.5 | ||
| Total physical activity | 0.7 | 1.0 | ||||||
| Inactive | 15,347 | 41.4 | 40.5; 42.4 | 7,827 | 58.5 | 56.8; 60.1 | ||
| Insufficiently active | 4,922 | 13.1 | 12.4; 13.7 | 2,201 | 14.0 | 12.9; 15.2 | ||
| Active | 17,405 | 45.5 | 44.6; 46.5 | 4,808 | 27.5 | 26.1; 29.0 | ||
| Daily television time (hours) | 0.0 | 0.0 | ||||||
| Up to 2 | 19,090 | 49.1 | 48.1; 50.1 | 6,947 | 44.9 | 43.1; 46.6 | ||
| ≥ 2 | 18,857 | 50.9 | 49.9; 51.9 | 8,035 | 55.1 | 53.4; 56.9 | ||
95%CI: 95% confidence interval.
* Percentage in the weighted sample;
** 95% confidence interval in the weighted sample.
Tables 2 and 3 show the behavioral indicators associated with the number of NCDs in multimorbidity in adults and older adults, respectively. Among 37,947 adults, when comparing subjects with none and one disease, the occurrence of two chronic diseases was more frequent among those active in the leisure-time domain (OR = 1.46, 95%CI: 1.16; 1.84) and with longer daily television time (OR = 1.20, 95%CI: 1.05; 1.38). An inverse trend between the presence of two diseases (p < 0.001) and total physical activity was observed. The occurrence of three chronic diseases was less frequent in subjects active in the work domain (OR = 0.60, 95%CI: 0.46; 0.78) and more frequent in those with longer daily television time (OR = 1.50, 95%CI: 1.20; 1.88). The occurrence of four diseases was less frequent in subjects active in the work domain (OR = 0.48, 95%CI: 0.27; 0.82).
Table 2 Association * of behavioral indicators with the number of chronic noncommunicable diseases in the diagnosis of multimorbidity in adults. Brazil, 2013 (n = 37,947).
| Variable | 2 vs. 0 and 1 | 3 vs. 0 and 1 | 4 vs. 0 and 1 | |||
|---|---|---|---|---|---|---|
| Adjusted OR (95%CI) | p-value | Adjusted OR (95%CI) | p-value | Adjusted OR (95%CI) | p-value | |
| Domestic physical activity | 0.187 | 0.067 | 0.321 | |||
| No | 1.00 | 1.00 | 1.00 | |||
| Yes | 1.10 (0.96; 1.25) | 0.80 (0.64; 1.02) | 1.32 (0.77; 2.26) | |||
| Commuting physical activity | 0.801 | 0.448 | 0.379 | |||
| No | 1.00 | 1.00 | 1.00 | |||
| Yes | 0.98 (0.83; 1.16) | 0.89 (0.66; 1.20) | 0.76 (0.41; 1.40) | |||
| Work physical activity | 0.461 | < 0.001 | 0.008 | |||
| No | 1.00 | 1.00 | 1.00 | |||
| Yes | 0.95 (0.82; 1.10) | 0.60 (0.46; 0.78) | 0.48 (0.27; 0.82) | |||
| Leisure-time physical activity | 0.001 | 0.162 | 0.745 | |||
| No | 1.00 | 1.00 | 1.00 | |||
| Yes | 1.46 (1.16; 1.84) | 1.33 (0.89; 1.98) | 0.81 (0.22; 2.94) | |||
| Total physical activity | < 0.001 | 0.048 | 0.545 | |||
| Inactive | 1.00 | 1.00 | 1.00 | |||
| Insufficiently active | 0.66 (0.48; 0.91) | 0.79 (0.47; 1.31) | 1.15 (0.33; 4.06) | |||
| Active | 0.54 (0.41; 0.71) | 0.66 (0.43; 1.02) | 0.79 (0.25; 2.47) | |||
| Daily television time (hours) | 0.007 | < 0.001 | 0.022 | |||
| Up to 2 | 1.00 | 1.00 | 1.00 | |||
| ≥ 2 | 1.20 (1.05; 1.38) | 1.50 (1.20; 1.88) | 1.97 (1.10; 3.52) | |||
95%CI: 95% confidence interval; OR: odds ratio.
Note: Analysis adjusted for sex, age, marital status, ethnicity, demographic macroregion, and education level (first level), and for physical activity in the commuting, domestic, leisure-time and work domains, total physical activity and daily television time (second level).
* Values weighted for the inverse of existing telephone lines and the number of adults living in the interviewee’s household.
Table 3 Association * of behavioral indicators with the number of chronic noncommunicable diseases in the diagnosis of multimorbidity in older adults. Brazil, 2013 (n = 14,982).
| Variable | 2 vs. 0 and 1 | 3 vs. 0 and 1 | 4 vs. 0 and 1 | |||
|---|---|---|---|---|---|---|
| Adjusted OR (95%CI) | p-value | Adjusted OR (95%CI) | p-value | Adjusted OR (95%CI) | p-value | |
| Domestic physical activity | 0.004 | 0.334 | 0.025 | |||
| No | 1.00 | 1.00 | 1.00 | |||
| Yes | 0.77 (0.65; 0.92) | 0.89 (0.70; 1.13) | 0.56 (0.34; 0.93) | |||
| Commuting physical activity | 0.797 | 0.101 | 0.427 | |||
| No | 1.00 | 1.00 | 1.00 | |||
| Yes | 0.96 (0.69; 1.33) | 0.65 (0.39; 1.09) | 1.39 (0.62; 3.14) | |||
| Work physical activity | 0.548 | 0.139 | 0.002 | |||
| No | 1.00 | 1.00 | 1.00 | |||
| Yes | 0.92 (0.69; 1.22) | 0.72 (0.46; 1.11) | 0.26 (0.11; 0.62) | |||
| Leisure-time physical activity | 0.889 | 0.541 | 0.359 | |||
| No | 1.00 | 1.00 | 1.00 | |||
| Yes | 1.02 (0.74; 1.42) | 0.86 (0.54; 1.39) | 0.53 (0.14; 2.04) | |||
| Total physical activity | 0.129 | 0.424 | 0.878 | |||
| Inactive | 1.00 | 1.00 | 1.00 | |||
| Insufficiently active | 0.72 (0.39; 1.31) | 0.91 (0.47; 1.78) | 1.03 (0.23; 4.60) | |||
| Active | 0.67 (0.39; 1.15) | 0.77 (0.42; 1.41) | 1.11 (0.18; 6.79) | |||
| Daily television time (hours) | 0.004 | < 0.001 | 0.068 | |||
| Up to 2 | 1.00 | 1.00 | 1.00 | |||
| ≥ 2 | 1.29 (1.09; 1.53) | 1.54 (1.25; 2.01) | 1.54 (0.97; 2.45) | |||
95%CI: 95% confidence interval; OR: odds ratio.
Note: analysis adjusted for sex, age, marital status, ethnicity, demographic macroregion, and education level (first level), and for physical activity in the commuting, domestic, leisure-time and work domains, total physical activity and daily television time (second level).
* Values weighted for the inverse of existing telephone lines and the number of adults living in the interviewee’s household.
Among 14,982 older adults, the occurrence of two diseases was less frequent in those active in the domestic domain (OR = 0.77, 95%CI: 0.65; 0.92) and in those with four diseases in the domestic (OR: 0.56, 95%CI: 0.34; 0.93) and work domains (OR = 0.26, 95%CI: 0.11; 0.62). The presence of two and three chronic diseases was consistently more frequent in subjects with longer daily television time (OR = 1.29, 95%CI: 1.09; 1.53, OR = 1.54, 95%CI: 1.25; 2.01, respectively).
An interaction was observed between total physical activity level and television time in adults and older adults with NCD multimorbidity (Figures 1 and 2). Among active adults, multimorbidity frequency was higher in those spending more time watching television compared with those spending less time doing it (Figure 1). Among older adults, multimorbidity frequency was higher in inactive subjects spending less time watching television compared with those spending more time doing it, as well as in active subjects spending more time watching television when compared with those spending less time in this activity (Figure 2). Both age groups were similar regarding multimorbidity frequency between insufficiently active subjects who spend more and less time watching television.
Figure 1 Interaction * of total physical activity with sedentary behavior in the occurrence of multimorbidity in adults. Brazil, 2013 (n = 37,947).
Discussion
This study aimed to investigate the association of NCD multimorbidity with indicators of physical activity and sedentary behavior, as well as the interaction between these behaviors. Multimorbidity increased the odds of longer television time in adults and older adults. Likewise, multimorbidity reduced the odds of being active and insufficiently active, as well as the participation in commuting- and work-related physical activity among adults. In older adults, multimorbidity reduced the odds of participating in domestic- and work-related physical activity.
Sedentary behavior, represented here by television time, was associated with a higher risk of NCD multimorbidity in adults and older adults from the capitals of the Brazilian states. Similar findings have been reported in studies conducted in middle-income 12 and high-income countries 21. Watching television is one of the main activities performed by adults during leisure time 22 and deserves attention because it characterizes a risk profile that is strengthened by the combination with some behaviors affecting energy balance, such as inadequate eating and low levels of physical activity 22,23,24,25.
An inverse association was observed between NCD multimorbidity and physical activity in the work domain. Considering work-related physical activity, Mabry et al. 26 also found a reduction in the occurrence of three diseases. The work place is a social and environmental determinant of health 27 and comprises active behaviors involving occupational and commuting routine. Thus, individuals with multimorbidity who do not perform physical activities resulting from occupational demands (carrying weight or getting stand up for a long time) at work would be an important target for preventive measures.
Leisure-time physical activity, on the other hand, was associated with a risk for multimorbidity. This domain exists in the literature as a prevention 28 or without association with multimorbidity 26. The protection relationship can be attributed to leisure activity practices, which should be done with higher intensity, mainly due to the short relative duration of the equivalent training sessions 28. A possible justification for the findings of this study is that attributing a causal relationship to the variables and raising the assumption that those with two diseases started physical activity in this domain after the conditions occured is not possible. Another justification is that the benefit of practicing physical activity outside working hours for health seems not to be enough for the deleterious effects of sedentary behavior throughout the day 11. In other words, even if activities were performed outside the occupational context, the association of long periods during work journey seems to be stronger.
Total physical activity considering the leisure-time and commuting domains was also inversely associated with multimorbidity in adults. This finding agrees with the study by Wu et al. 29, which indicated that combining several domains permits identifying protection against multimorbidity, considering mainly the total cumulative physical activity time. Habitual physical activity is an excellent approach to prevent multimorbidity 30, in which the diversification of physical capacities 31 and the constant regulation of the increase in intensity and volume have benefits 32,33. However, light physical activities should also be encouraged because of their proven preventive effect 9. Thus, physical activity along the day, regardless of domain, type or intensity, is inversely associated with NCD multimorbidity occurrence. Regarding the interaction between leisure-time and commuting total physical activity and sedentary behavior in adults, increasing physical activity slightly reduced the proportions of multimorbidity, which was significant in active adults who spend less time watching television compared with those spending more time doing it. Lopronzi et al. 9 also demonstrated an association of multimorbidity with sedentary behavior, regardless of physical activity. The lower prevalence of multimorbidity detected with the sufficient physical activity practice, as also reported by Bertrais et al. 34, may increase the time allocated to the activity and its intensity. Thus, physical activity seems to attenuate the occurrence of multimorbidity in relation to television time 9,34. Within this context, physical activity does not play a preventive role, but it may control a reduction in the occurrence of multimorbidity in more active subjects, regardless of television time.
In older adults, NCD multimorbidity was directly associated with increased sedentary time in front of the television. Similar results have been reported for Australian older adults 35. Harvey et al. 36 found sitting time in front of the television is a potential predictor of poor health in older adults, attributed to lower quality of life and poor diet indicators.
The domestic and work physical activity domains were inversely associated with multimorbidity in older adults. A study has shown benefits for cardiovascular health in older adults who practice physical activity in the domestic and work domains 37. Older adults who perform physical activities in their daily life have better functional capacity, which is considered a protective factor against multimorbidity 38. On the other hand, the commuting or leisure-time physical activity domain was not significant for the older adults studied here. A longitudinal study investigating older adults who performed physical activity at different intensities found a lower prevalence of multimorbidity in the vigorous physical activity group 39. Thus, the hypothesis can be raised that the intensity of physical activity rather than its duration is an important indicator of multimorbidity prevention.
Analysis of the interaction between behaviors in older adults indicates a trend towards an inversely proportional decrease in multimorbidity with increasing physical activity duration, regardless of television viewing time. Similar results have been reported by Gardiner et al. 35. The behaviors investigated in the interaction are characterized by independent organ responses 7. However, physical activity seems to play a key role in the lower frequency of NCD multimorbidity in older adults, suggesting an effect of biological aging on this epidemiological profile 39.
Another characteristic of the interaction between these behaviors in older adults was the higher frequency of NCD multimorbidity among inactive subjects with little television viewing time. This finding does not suggest the frequency of multimorbidity is inversely associated with less television viewing time in inactive subjects, but rather with other sedentary behaviors performed by this age group. Gardiner et al. 35 pointed the association between the risk of multimorbidity and overall sitting activities in this population. These sitting activities could result from two findings: first, subjects’ functional capacity that renders them inactive 40 and, second, types of sitting activities not performed in front of a screen 41.
This was the first study to investigate the association of NCD multimorbidity with physical activity and sedentary behavior in the Brazilian scenario, which separately analyzed adults and older adults and the respective associations with behavioral indicators. Even considered a continental country, Brazil has similar Human Development Index (HDI) values in all capitals, possibly reflecting in the multimorbidity occurrence in each city, once HDI considers health information. In addition, Brazilian population is formed by different ethnicities and the results may be important to future recommendation based on evidences.
A main limitation of this study was the small number of NCDs comprising multimorbidity, which were restricted to the most prevalent diagnoses in the country, hampering the comparison with studies conducted in other countries. Self-reported diseases and body mass index (BMI), although usually applied in cross-sectional studies, need to be considered in the interpretation of the results. In it, the values of occurrence of diseases may be overestimated or underestimated and the BMI may have a potential information bias, according to the interviewee's knowledge. However, Brazil has public health programs that provide this kind of information in primary health care in all country 42. The physical activity and sedentary behavior were also self-reported, and the physical activity may have been overestimated as well as the sedentary behavior may have been underestimated. On the other hand, physical activity questions provide different domains, offering important contributions to the analyses. Considering sociodemographic information, the retirement was not an option in the work physical activity domain, the analysis disregarded differences in results between gender, and the seasonal particularities during data collection were not included. Finally, the cross-sectional design did not allow us to exclude the possibility of reverse causality between the findings. However, considering physical activity and sedentary behavior as precedents of the conditions studied here is plausible 43.
Conclusion
Multimorbidity was associated with increased sedentary time and physical activity in the commuting and work domains in adults from the capitals of Brazil. In addition, it was also associated with insufficient (10 to 149 minutes) and sufficient physical activity in the leisure-time and commuting domains. Among older adults, NCD multimorbidity was associated with increased television viewing time and physical activity in the domestic and work domains. Analysis of the interaction between sedentary behavior and physical activity permitted identifying a lower frequency of multimorbidity in subjects who are active in the leisure-time and commuting domains and who spend less time watching television. A reduction in television time should be encouraged particularly in adults, with the concomitant incentive of more active behaviors without considering the volume and intensity of physical activity. We suggest measuring physical activity using objective measures in future studies to close the gaps found between the results of this study and the international literature.
