Determinants of work capacity (predicted VO_2max) in non-pregnant women of reproductive age living in rural India

Human survival is dependent upon the capacity to perform the work necessary to ensure adequate food, shelter, and clothing [1]. The current reliance on modern technology in most developed countries has resulted in a steady decline in daily work-related energy expenditure [2]. This is not the case in still-developing societies, however, in which survival remains dependent on human power and prolonged physical work. In India, approximately 55% of women engage in agricultural labor, with the majority of them being of reproductive age [3], suggesting that most women carry a dual burden of outside labor and family care. Work capacity, therefore, has important financial and social implications for these women.

The dual work burden is made worse by the fact that more than half of Indian women have anemia – and this is especially so in rural areas [4]. Anemia is defined by a low (< 12 g/dL) concentration of hemoglobin (Hb) in the blood [5], which then contributes to a diminished work capacity due to the reduction in the oxygen-transport ability of blood in response to prolonged and heavy exercise (i.e., work). The effects of a diet lacking in sufficient caloric intake, micronutrient, and macronutrient content on anemia and consequent work capacity are well-known [6‐9], and recent qualitative research also conducted in Odisha, India, indicates that women tend not to prioritize their own health, relative to the health of their husbands or children [10]. For example, they tend to eat only after everyone else in the family has eaten and thus, are often left with inadequate and poor quality food. Anemia is then exacerbated in these women, as they tend not to seek medical care for their own health and physical weakness is understood to be part of their identity as women [11].

A number of other behavioral factors influence work capacity [8] and physical activity is one such behavior. Insufficient daily physical activity results in poor physical and cardiorespiratory fitness. Workers with lower levels of fitness may be able to perform a similar work output, compared with their fitter counterparts; however, the energy expenditure and physiologic cost of performing that work will be higher [12]. On the other hand, sitting and reclining behaviors may aid in recovery from prolonged work in the heat, thus augmenting subsequent work capacity.

Work capacity can be expressed as maximal aerobic capacity (VO_2max) [12]. A low VO_2max will limit the oxygen-delivery capacity of the cardiovascular system, as well as the oxygen-uptake capacity of the muscle, and these limitations are compounded by anemia. Whereas, VO_2max is most precisely measured in the laboratory setting using indirect calorimetry, several field-based tests have been validated that rely on nomograms based on heart rate responses to a standardized sub-maximal exercise test. The Queen’s College Step-Test (QCST) [13] is the simplest of these measures and uses prediction equations to calculate VO_2max based on recovery heart rate.

The impact of anemia on work capacity (VO_2max) has been studied extensively in male and female workers [7]. We know of only one other study that has examined multiple determinants of work capacity simultaneously [8]; however, anemia was not considered in this analysis, there were no objective measures of physical activity, and work capacity was determined by the number of rice bundles produced per hour. Therefore, we used cross-sectional data to examine the multivariable contributions of several sociodemographic, physiologic, and behavioral variables to work capacity (VO_2max) in women of reproductive age living in rural India. We hypothesize that hemoglobin concentrations and physical activity will be strong independent correlates of predicted VO_2max, but that these associations would vary by age and/or by level of education.

The Reduction in Anemia through Normative Innovations (RANI) project is a norms-based, clustered randomized controlled trial to reduce anemia among women of reproductive age in Odisha, India. The project is being implemented in two blocks (Athamalik and Kishorenagar) in the Angul district of Odisha. Angul was selected for study because it is primarily rural and the prevalence of anemia in that region (44%) is similar to that of Odisha [4]. In the two blocks of Athamalik and Kishorenagar, nearly 25% of people are tribal, 33% are literate, and about half of women work outside of the home [14]. Detailed information on the cluster sampling and randomization methods are described elsewhere [14]. Briefly, 89 clusters of villages were randomized into treatment and control groups on a 1:1 basis by the program implementers, using a random number generator. Women (2055 per group) living in 15 selected clusters (40–41 villages) were then randomly selected for data collection. The overall sample size was based on a predicted 7% improvement in anemia prevalence, assuming an alpha-level of 0.05, with statistical power (1-beta) of 0.80 and accounting for a design effect of 2.0 (clustering effects within the villages) and a 20% loss to follow-up. All women between 15 and 49 years old, who resided in one of the villages selected for study, and who speak Odia were eligible to participate in the larger trial. The sampling design also randomly-generated a subset (187 per group) of non-pregnant participants who performed the QCST and who wore an activity monitor for 3 days. Prior formative research on the cultural norms of this study population indicated that 3 days was the maximal number of days that women would tolerate wearing the ActivPAL. We performed the current cross-sectional analysis using data from this subset of participants, all of whom had complete QCST and physical activity data (n = 330). Data collectors and program implementers were blinded with regard to the treatment and control status of the villages. Written informed consent was obtained in Odia by the local data collectors. In the case of participants under the age of 18 years, written permission of one parent or legal guardian and assent of the participants were obtained. All procedures were approved by Institutional Review Boards (IRB) at the George Washington University, as well as Sigma Science and Research, an independent IRB located in New Delhi, India, and the Indian Council for Medical Research’s (ICMR’s) Health Ministry’s Screening Committee (HMSC).

Table 1 displays the participant characteristics. On average, women were 30 ± 8 years of age with a BMI of 21.1 ± 3.6 kg/m². Although 18% of the women had no education, approximately 46% had 8 or more years of schooling. The majority of women (59%) were from the “other backward castes” (socially-disadvantaged people that belong to non-Hindu religions) and about 70% had between one and three children. The average hemoglobin concentration was 11.5 ± 1.3 g/dL and the prevalence of anemia in this study population was 61%, which is above the national average of 53% [4]. Of the 201 women having anemia, only four met criteria for severe anemia. Therefore, anemia status was dichotomized into “mild/moderate/severe” and “none” for the regression modeling.

Average predicted VO_2max in the women was 42.9 ± 3.2 ml∙(kg∙min)^− 1, indicating an excellent level of cardiorespiratory fitness in women of this age [25]. Participants were able to complete the QCST at about 78% of their age-predicted maximal heart rate; however, the range in physiological cost was 57% (more fit) to 100% (less fit). Overall, women performed an average of about 36 MET-h/day of physical activity and 13,425 steps/day. For reference, 36 MET-h can be achieved by performing work activity of moderate intensity (e.g., 4 METs) for 9 h over the course of the day. Time spent sitting and reclining were low, relative to more affluent and sedentary populations in the United States [26]. Table 2 displays mean levels of predicted VO_2max and age-predicted maximal heart rate, as well as the physical activity variables by anemia status. As indicated, there were no significant differences in any of these variables between those women with and without anemia, suggesting that the presence of anemia did not impair work capacity or increase the physiological cost of the QCST. Moreover, we did not observe any behavioral adjustments to anemia, such as lower amounts of overall physical activity or greater amounts of daily sitting and reclining. Data on dietary diversity were available on only 65 women in this subsample of participants. There was no difference in diet diversity score between those with (4.0 ± 0.91) and without (4.3 ± 0.25) anemia.

The results of the multivariable regression modeling indicated that age², BMI, educational attainment, and physical activity were significant correlates of work capacity in our study participants (Table 3). Predicted VO_2max was 0.19 ml∙(kg∙min)^− 1 lower with each unit increase in BMI (about 2.4 kg in this study population) and was 1.35 ml∙(kg∙min)^− 1 lower in women with 8–11 years of education compared with those women having no education. Each MET∙h/day increase in physical activity was associated with an increase in predicted VO_2max of 0.19 ml∙(kg∙min)^− 1. Hemoglobin concentration was marginally associated with work capacity in the presence of the other covariables (p = 0.067). Predicted VO_2max was 0.22 ml∙(kg∙min)^− 1 higher with each g/dL increase in hemoglobin concentration. When hemoglobin concentration was replaced with anemia status the results were essentially unchanged. Women with anemia had a predicted VO_2max 0.23 ml∙(kg∙min)^− 1 lower than women without anemia (beta = − 0.23; 95% CI: − 0.90, 0.44, although this was not statistically significant (p = 0.50). Also, when MET∙h/day was replaced with steps/day as the indicator of physical activity, the estimates also were unchanged. Hours per day spent reclining, sitting, or standing and dietary diversity score were not significantly associated with work capacity in this model. Contrary to our hypothesis, we observed no statistically significant interactions between the study variables of interest and either age or education level in their joint associations with work capacity.

Fig. 1 shows the inverse curvilinear nature of the adjusted relationship between age and work capacity. Maximal aerobic capacity appeared to increase slowly until reaching its peak value at about age 30 years (approximately the sample mean). Thereafter, VO_2max began to decline in an accelerated manner.

Haas and Brownlie [7] present a comprehensive framework to conceptualize the effects of anemia on the physiological, psychosocial, and socioeconomic aspects of work. As stated previously, the underlying mechanism of a diminished work capacity with anemia is a reduction in the oxygen-transport ability of the blood in response to prolonged and heavy exercise. This altered work performance (i.e., capacity, efficiency, and endurance), in turn may affect both the quantity and quality of time allocated to other family responsibilities (cooking and child care) and to leisure activities that allow for recovery from outside work activities. Kalasuramath and colleagues [6] studied 600 working Indian women (18–55 years) and observed that women having mild-to-moderate anemia (Hb > 8 to < 12 g/dL) had a significantly lower VO_2max [38.7 ± 6.2 mL∙(kg∙min)^− 1)], compared with those without anemia [45.5 ± 5.3 mL∙(kg∙min)^− 1; p < 0.05]. In contrast, we observed no difference in predicted VO_2max or age-predicted maximal heart rate from the QCST between women with and without anemia is our study population, suggesting that anemia did not impair work capacity or increase the physiological cost of the step-test itself. Moreover, we did not observe any behavioral adjustments to anemia, such as lower amounts of overall daily physical activity or greater amounts of daily sitting and reclining. This is contrary to what we hypothesized and may be explained by the fact that we did not include pregnant women in our study sample and that the mean hemoglobin concentrations between those with and without anemia were very similar. On the other hand, inequitable gender norms may make it very difficult for women to rest during the day if they are feeling fatigued due to anemia. Indeed, these women must bear the burden of both outside labor and family care, and therefore, there is little opportunity (or encouragement) to sit or recline during the day [11].

An average predicted VO_2max of about 43 ml∙(kg∙min)^− 1 indicated an excellent level of cardiorespiratory fitness for women of this age [25]. For comparison, the average VO_2max in untrained, but healthy U.S. women of comparable age is typically between 27 and 31 ml∙(kg·min)^− 1 [25]. Women in the current study achieved 36 MET-h/day and 13,425 steps/day of physical activity, which is well above current WHO guidelines of 150 min/week of moderate-intensity activity (or about 7–8 MET-h/week) [27]. This volume of physical activity in our study sample presumably is due to work activity of moderate intensity (3.5 to 5 METs) that may occur for up to 9 or more hours/day.

We also observed an inverse association between predicted work capacity and level of educational attainment, which is commonly observed in developing countries and presumably due to the fact that more educated women work at more sedentary professions. Other studies of more affluent female university or medical students living in India report predicted VO_2max levels between 35 and 38 mL∙(kg∙min)^− 1 [18, 22], which is lower that what we observed in our rural sample. As is observed in other Indian populations [18, 23] and most populations globally, predicted VO_2max was lower with increasing BMI, and this association was independent of the confounding effects of age, parity, and lower levels of physical activity.

Maximal aerobic capacity declines between 5 and 15% per decade after age 25 years [12, 15]. The slightly accelerated decline in VO_2max that we observed in our participants after age 30 years has important implications for their health and function. Indeed, if work productivity is expected to remain stable through the reproductive years, then it would be performed at an exponentially higher physiology cost.

Factors other than anemia may affect work capacity in women, and despite the important contributions of age, BMI, education, and physical activity, our final regression model explained only 16% of the variance in predicted VO_2max. This suggests that other unmeasured behavioral or health-related factors may contribute to predicted VO_2max in these women. When the dietary diversity score was added to the regression modeling, the predicted variance of the model decreased further and with no significant contribution from diet. It is possible that total energy intake is a more important correlate of work capacity than is the diversity of one’s diet. Moreover, we did not study iron deficiency (Fe < 15 μg/L), as these measurements are quite expensive to perform in the field. Evidence indicates that the effects of iron deficiency on work performance are partially distinct from those of anemia [7]. Anemia targets oxygen delivery and thus impairs aerobic capacity (which is best captured by the determination of VO_2max), whereas, iron deficiency affects tissue oxidative capacity (i.e., diffusion of oxygen to the muscle mitochondria for energy production), which is related more to work efficiency and endurance. We attempted to capture endurance by objectively measuring volume (MET∙h/day) of physical activity and by steps/day, while efficiency was assessed by age-predicted maximal heart rate achieved at the conclusion of the QCST; however, we have no way of knowing whether these physical performance outcomes differed between those with and without iron deficiency.

It is important to consider that from an economic standpoint, work capacity may differ from actual work productivity for many reasons [28]. Work productivity (i.e., output) depends on intelligence, body stature, energy intake, and strength [8], but also on incentives and motivation [29]. Unfortunately, we did not gather information on these latter two factors. Furthermore, not all work requires maximal aerobic capacity; in fact, the rural agricultural, cooking, and family-care labor of our study population may rely more on endurance (i.e., moderate-intensity activity over many hours of the day).

There are several strengths to this study; namely, a sampling design that produced a representative sample of women of reproductive age living in the Angul district and the objective measurement of physical activity and maximal aerobic capacity by validated methods. Nonetheless, the analysis was cross-sectional from baseline RANI data and therefore the temporal sequencing between study factors such as physical activity and hemoglobin concentrations and the primary outcome of VO_2max could not be established. Pregnant women were not included in our study sample and perhaps we would have observed stronger associations between physical activity, hemoglobin concentrations, and VO_2max if they were included, due to the added physiologic challenge of pregnancy. Also, the difference in mean hemoglobin concentrations between those with and without anemia was only 2 g/dL, which would explain the similarity in regression estimates when using the continuous hemoglobin variable versus the dichotomous anemia variable. Unfortunately, the small number of severe anemia cases (n = 4) precluded our ability to study the impact of anemia on work capacity over a range of levels. Finally, due to the small stature of our participants and the fact that they wore a sari while performing the QCST, the height of the step was lowered from 16.25 in. to 12 in.. This adjustment allowed all participants to complete the test; however, it limits our ability to compare our VO_2max data with those from other studies using different step heights. In addition, this adjustment may result in an over-estimation of maximal aerobic power in our participants.

Factors other than anemia may contribute to a lower work capacity in women of reproductive age. Our data indicate that older age, higher BMI and educational attainment, and lower levels of physical activity are important correlates of predicted VO_2max. Government and non-governmental agencies should account for these factors when creating policies directed toward promoting the economic capacity of women living in rural India.