Physical activity in pregnancy and adverse birth outcomes

This study focused on the association between physical activity in the second trimester of pregnancy and adverse perinatal outcomes: low birth weight (LBW), preterm birth (PTB), and intrauterine growth restriction (IUGR). The study used a sample from the BRISA cohort, São Luís, Maranhão State, Brazil, which included women with singleton pregnancy, gestational age from 22 to 25 weeks confirmed by obstetric ultrasound performed at < 20 weeks, and re-interviewed in the first 24 hours postpartum (n = 1,380). Level of physical activity was measured by the International Physical Activity Questionnaire (IPAQ), short version, categorized as high, moderate, and low. A directed acyclic graph (DAG) was used to identify minimum adjustment to control confounding. High physical activity was not associated with LBW (RR = 0.94; 95%CI: 0.54-1.63), PTB (RR = 0.86; 95%CI: 0.48-1.54), or IUGR (RR = 0.80; 95%CI: 0.55-1.15). The results support the hypothesis that physical activity during pregnancy does not result in adverse perinatal outcomes. Pregnancy; Motor Activity; Low Birth Weight Infant; Premature Infant Correspondence

There are various risk factors for adverse perinatal outcomes ranging from classical factors such as hypertension, diabetes, and obesity 5 to physical activity during pregnancy 6 .
The American College of Obstetrics and Gynecology (ACOG) recommends that women who were already physically active should continue their physical activity during pregnancy and encourages pregnant women to engage in at least 30 minutes of physical activity per day, three or more days per week, in the absence of clinical contraindications 7 .
Although physical activity is recommended because of its benefits for the prevention and treatment of cardiovascular diseases, gestational diabetes, and other chronic diseases, there is no consensus on the potential risks to the mother and fetus 8 .Vigorous physical activity can result in reduced blood flow to the fetus, with a consequent reduction in essential substrates for its growth, and may result in adverse perinatal outcomes 9,10 .
Early systematic literature reviews yielded insufficient evidence on the risks and/or benefits of physical activity to the mother and fetus 8,11 .A more recent systematic review concluded that there are still significant gaps in knowledge on the safety of intense physical activity for the fetus 11 .This insufficient evidence may be due to the methodological difficulty in measuring gestational physical activity and to small study samples 8,11 .
The current study thus aimed to investigate the association between physical activity in pregnant women and adverse perinatal outcomes (low birth weight, preterm birth, and intrauterine growth restriction).

Study design
This was a nested study in a cohort study entitled The BRISA cohort data from the municipality of São Luís were used in the current study.This is a convenience cohort started during the prenatal period (PRENATAL BRISA), since it was not possible to draw a random sample, because no list of pregnant women was available.In 2009-2010, 1,447 pregnant women were recruited in the waiting room at their prenatal visit or for obstetric ultrasound at three major public maternity services and public and private ultrasound laboratories.Inclusion criteria were: singleton pregnancy and gestational age confirmed by obstetric ultrasound performed at less than 20 weeks gestational age.The women were then invited to attend the Clinical Research Center (CEPEC) to answer a questionnaire.The prenatal interviews took place from the 22 nd to the 25 th gestational week.At childbirth, 1,381 women (95.4%) were re-interviewed in 2010-2011, with 4.6% losses.One observation was excluded because of missing values on physical activity.The total sample for analysis amounted to 1,380 women.Methodological data have been published elsewhere 12 .

Sample calculation
It was estimated that a sample of 1,092 pregnant women would have 85% power to identify rates ratios ≥ 2.0 between physical inactivity and adverse birth outcomes, assuming 6-10% prevalence of LBW, PTB, and IUGR in women that are sedentary in the second trimester, a ratio of 1:1 between exposed and unexposed infants, and 5% probability of type I two-tailed error.After a 25% increment to adjust for potential confounding, the minimum sample size was set at 1,365 pregnant women.

Questionnaires
The first questionnaire was used to establish the women's socio-demographic profile and pattern of physical activity.The second questionnaire was applied in the first 24 hours postpartum.The following variables were recorded at birth: maternal hypertension, diabetes mellitus, vaginal bleeding, urinary infection, illicit drug use, alcohol consumption, hospitalization during pregnancy, and maternal smoking.All variables were categorized as yes or no.Parity (primipara, 2 or more children) and type of delivery (vaginal, non-elective cesarean section, elective cesarean section) were also recorded.

Socio-demographic variables
The target socio-demographic variables were maternal age categorized as younger than 20 years, 20 to 34 years, and 35 years or more.Maternal schooling was categorized as less than 5, 5 to 8, 9 to 11, and 12 or more years of school.Occupation was classified as non-manual, skilled/ semiskilled manual, or unskilled manual/unemployed/students).Economic class was defined according to the Criterion of Economic Classification -Brazil, elaborated by the Brazilian Association of Research Companies (ABEP) to estimate the purchasing power of urban families, categorized as A/B (highest income and education, and most household assets), C (intermediate), or D/E (lowest income and education) 13 .Married/living with a partner and living with children were both categorized as yes or no.

Level of physical activity
Physical activity was assessed using the International Physical Activity Questionnaire (IPAQ)short version 14 , validated for the Brazilian population by Matsudo et al. 15 .This instrument is used to obtain data on any type of physical activity.Women were asked, referring to the previous seven days, on how many days they walked and did moderate or vigorous physical activity for at least 10 minutes at a time.
Physical activity was classified in three levels -high, moderate, and low -on the basis of frequency, duration of each activity, and energy expenditure as metabolic equivalents (MET)minutes/week.For walking, 1 MET = 3.3 x time (minutes) x day; for moderate activity, 1 MET = 4.0 x time (minutes) x day, and for high activity, 1 MET = 8.0 x time (minutes) x day.Total METminutes/week is calculated as the sum of partial MET-minutes/week 16 .
Physical activity was considered high when practiced vigorously for at least three days, reaching at least 1,500 MET-minutes/week, or for at least seven days with any combination of walking, moderate or vigorous activity, reaching at least 3,000 MET-minutes/week.Moderate was defined as vigorous activity for at least 20 minutes for three or more days or moderate for five or more days and/or involving a walk of at least 30 minutes per day, or consisting of five or more days of a combination of walking, moderate activity, or vigorous activity reaching at least 600 MET-minutes/week.Low activity was defined as opposed to high or moderate levels of physical activity, < 600 MET-minutes/week 16 .

Outcome variables
The dependent variables were: gestational age categorized in weeks as preterm (PTB < 37 weeks) or term (≥ 37 weeks); LBW (< 2,500g), and IUGR classified according to the birth weight ratio (BWR) proposed by Kramer ete al. 17 , calculated by dividing the newborn's weight by the sexspecific median weight for gestational age from the Canadian curve 18 .Neonates were defined as not presenting IUGR when the BWR was ≥ 0.85, and as presenting IUGR when BWR was < 0.85 17 .Gestational age was estimated by two criteria, i.e., date of last menstruation (DLM) and obstetric ultrasound (OU) performed before 20 weeks gestational age (GA).When GA measured by DLM differed by more or less than 10 days from the value estimated by OU, GA was calculated by DLM; when GA differed by more than 10 days between these two indicators, it was estimated by OU 19 .Information on date and time of birth was obtained from the infant's medical records.

Statistical analysis
Data were analyzed using Stata, version 12.0 (StataCorp LP, College Station, USA).Descriptive analyses were performed with estimates of absolute and percent frequencies.Frequencies were compared between groups by the chi-square test.Associations between physical activity and three different outcomes (LBW, PTB, and IUGR) were estimated by Poisson multiple regression models with robust estimation of variance 20 .Rates ratios (RR) and 95% confidence intervals (95%CI) were used as measures of association.Physical activity was treated as an ordinal variable categorized in three levels: low, moderate, and high.Moderate physical activity was taken as the reference.Sensitivity analysis was also performed to check for consistency, where physical activity was treated as a continuous variable (minutes or MET-minutes/week) or divided into tertiles, quartiles, or quintiles.To check for possible non-linearity, a quadratic term was also added to the models, where physical activity was treated as a continuous variable.
A directed acyclic graph (DAG) was constructed to identify a minimum set of confounding adjustment in the DAGitty program (Figure 1) 21 .Interrelations between variables were constructed based on Takito & Benicio's theoretical hierarchical model 22 .Models were then adjusted for maternal schooling, economic class, living with partner, and living with children.
The study was approved by the Institutional Review Board of the University Hospital of UFMA
Taking low physical activity as the reference, there were no associations between moderate or vigorous physical activity and LBW, PTB, or IUGR.No associations were observed between physical activity treated as a continuous variable, either in minutes or in MET-minutes/week, and adverse birth outcomes.A quadratic term for MET-minutes/week was non-significant.Even when considering physical activity in MET-minutes/week divided in tertiles, quartiles or quintiles, no association between physical activity and adverse birth outcomes was detected (Table 2).Note: models adjusted for maternal schooling, economic class, living with a partner, and living with children.

Discussion
The present study did not show any association between women's level of physical activity during the second trimester of pregnancy and adverse perinatal outcomes (LBW, PTB, and IUGR), regardless of level of activity, i.e., low or high, taking moderate as the reference.The findings are similar to those of a study in North Carolina from 2004 to 2007 23 , in which pregnant women responded to a questionnaire by telephone, providing information on their physical activity in various domains (leisure-time, job-related, housework, and child and adult care), with no association detected between high physical activity in the first trimester and LBW or PTB.
The study also agrees with results of population-based surveys that failed to identify physical activity as a risk factor for adverse perinatal outcomes 24,25 .Two population-based cohort studies in 79,692 pregnant women in Denmark 24 between 1996 and 2002 and in 61,098 pregnant women in Norway 25 between 2000 and 2006 also showed no positive association between physical activity during pregnancy and prematurity.Instead, the studies actually showed a reduction of preterm delivery among women who had practiced some type of exercise during the second and third trimesters.The Norwegian study also showed a protective effect (OR = 0.82; 95%CI: 0.73-0.91)against prematurity in women that exercised 3 to 5 times a week.The results also agree with a randomized study of 105 sedentary primiparous pregnant women (mean age: 30.7 ± 4.0 years and pregestational body mass index (BMI): 23.8 ± 4.3), which concluded that aerobic and resistance exercises are not associated with preterm birth 26 .It should be noted that the women in the study were sedentary and therefore had no previous physical conditioning.Another study, which included pregnant women with hypertension and/ or previous preeclampsia, showed similar results, namely no increased risk of adverse maternal and neonatal outcomes, notably hypertensive complications, preterm birth, and low birth weight among women who had exercised during pregnancy 27 .In addition, a, systematic review indicated that physical activity during pregnancy had a protective effect against preeclampsia 28 .
However, a cohort study in India in 546 rural pregnant women, assessing physical activity in various domains (job-related, leisure-time, caregiving, or housework) and during the three gestational trimesters detected an association between the highest tertile of physical activity and LBW 29 .
A systematic review indicated that most studies did not detect an association between physical activity during pregnancy and LBW or PTB, but that some studies found an association between specific activities (e.g., climbing stairs or standing for long periods) and low birth weight and preterm birth 11 .Our study did not ask pregnant women about specific physical activities.However, few women in our sample reached very high activity levels, and most physical activity in our sample probably involved household chores 30 .Thus, based on our data it seems that physical activity during pregnancy is not harmful to the fetus.
In the current study, a high percentage of pregnant women (60.3%) were physically active in the second trimester.A large share of these women (42.8%) reported moderate physical activity according to ACOG recommendations that might result in benefits for maternal-fetal health 7, i.e., walking at least 30 minutes on 5 or more days a week.
A cross-sectional study in the city of Rio Grande, Rio Grande do Sul State, Brazil, in 2007 in a sample of 2,557 pregnant women showed that 32.8% were physically active 31 .The study excluded activities at school, on the job, and in the household, which may explain the lower percentage of physical activity compared to our study.In Pelotas, Rio Grande do Sul State, only 12.9% of a sample of 4,471 pregnant women practiced some type of leisure-time physical activity in any trimester of pregnancy 32 , indicating the impor-tance of considering on-the-job physical activity in pregnant women.
Until the 1990s, physical activity was discouraged for pregnant women 7 .Today there has been a change, with moderate physical activity recommended 7 , so that more women are physically active during pregnancy 31 .However, more pregnant women may be considered sedentary if physical activity is defined as planned exercising; in addition, a more sedentary routine may tend to increase in the third trimester 32 due to lack of time, fatigue, and discomfort with progression of the pregnancy 33 or a feeling of insecurity during physical activity 34 .Some studies indicate that pre-gestational physical exercise is an important predictor of higher levels of physical activity during pregnancy 32,35 .
The current study's strengths were its large sample size, low percentage of losses and refusals, and more precise calculation of gestational age based on a combination of gestational ultrasound before the 20th week and date of last menstruation.
The study's limitations include use of the short version of the IPAQ to assess physical activity.The short version does not distinguish between domains of physical activity (job-related, leisure-time, caregiving, or housework).In addition, physical activity was assessed only once in the second trimester, so it was not possible to verify the level throughout the pregnancy.Finally, no objective measure of physical activity was employed, which may produce measurement bias.Although objective measures of physical activity are less prone to error when compared to questionnaires 36 , they are still rarely used in population-based studies with large samples because of their high cost and complex logistics.Since data are collected on several days, close monitoring is required to avoid dropouts and supervise correct machine positioning.However, the IPAQ has been properly validated in the Portuguese language 15 and has been used previously to measure physical activity in pregnant women 37,38 .It is used internationally because it affords good stability, low cost, and acceptable accuracy in population studies 14 .Furthermore, the results of the reproducibility study of type test / retest support the use of IPAQ to assess physical activity levels in different populations 14 .
Based on the current study's findings, physical activity of any level during the second trimester of pregnancy is not associated with adverse perinatal outcomes such as low birth weight, premature birth, or intrauterine growth restriction.

Figure 1 Directed
Figure 1Directed acyclic graph of the association between physical activity during pregnancy and adverse birth outcomes.

Table 1
Socioeconomic, clinical, and lifestyle characteristics of pregnant women studied in the city of São Luís, Maranhão State, Brazil, 2009-2011.

Table 2
Adjusted analysis of the association between physical activity and preterm birth, low birth weight, and intrauterine growth restriction.São Luís, MaranhãoState, Brazil, 2009-2011.