Maternal physical activity at term and spontaneous labor: a case-crossover study

Abstract

Background:

This study assessed associations between antenatal physical activity and the onset of spontaneous labor (SL).

Methods:

Data were from 541 participants in the third Pregnancy, Infection, and Nutrition Study who had no contraindications to antenatal physical activity. An interviewer-administered questionnaire assessed labor triggers, gestational age at birth, and physical activity within the week (24 hours to 7 days) and the 24-hour period (0 to 24 hours) prior to SL. A case-crossover design examined the association between physical activity (recreational, occupational, or any) and the risk of onset of SL within the subsequent 24 hours.

Results:

Overall, 21% (any), 26% (recreational), and 14% (occupational) of participants reported physical activity during the week, whereas 5% (any), 7% (recreational), and 3% (occupational) reported physical activity during 24-hour period, prior to SL onset. Participants who reported any or occupational physical activity during the 24-hour period had a decreased likelihood of SL within the subsequent 24 hours, while participants who reported at least 30 minutes of recreational physical activity had an increased likelihood. Results remained consistent among early, full, or post-term participants.

Conclusion:

Recreational, but not occupational, physical activity at term may increase the likelihood of SL; however, we cannot rule out reverse causality.

INTRODUCTION

Physical activity is recommended for adults and children, across life stages, to increase strength and cardiorespiratory fitness and decrease the risk of metabolic diseases associated with obesity.¹ The United States (US) Department of Health and Human Services recommends pregnant individuals without contraindications engage in at least 150 minutes of moderate-intensity aerobic activity per week.² Additionally, pregnant individuals who regularly engaged in vigorous activity prior to conception can continue those activities throughout pregnancy, while those naïve to vigorous activity are urged to commence light or moderate intensity activities. Regular physical activity during pregnancy has been associated with a decreased risk of excessive gestational weight gain, gestational diabetes, gestational hypertension, preeclampsia, and maternal postpartum depressive symptoms.^3–7

Maternity care in the US is characterized by an overuse of medical interventions such as induction of labor and cesarean birth.⁸ While these interventions are often necessary—the ideal rate of cesarean, for instance, is not zero, but rather between 10% and 15%, depending on patient risk profiles^9,10—there is consensus that they are used too frequently within standard US obstetric practice.^11,12 This is problematic because unnecessary interventions are associated with risks to both the pregnant person and the fetus/neonate, without corresponding benefit.^8,12,13 Not only is there an increased risk of morbidity and mortality that stems from unnecessary obstetric intervention, but the healthcare costs of complicated births are much higher.¹⁴ While identification of behaviors that can induce term spontaneous labor (delivery on or after 37 weeks completed gestation) is especially desirable, it is also important to note that preterm spontaneous labor (delivery prior to 37 weeks completed gestation) is not desirable, as it is associated with increased risk of infant mortality and morbidities throughout the lifespan (i.e., psychiatric disorders, academic problems, and social difficulties).¹⁵ Therefore, when considering the association between physical activity and spontaneous labor, it is important to assess whether the associations vary by gestational age at labor onset.

In addition to improved pregnancy outcomes, recreational physical activity, such as walking for exercise, swimming, or dancing, may also be related to labor and delivery outcomes, particularly onset of spontaneous labor at term (i.e., a labor that begins without induction). A quasi-experimental study found individuals who engaged in moderate-intensity recreational physical activity (e.g., strength and aerobic exercises), up to three times per week during the second and third trimesters of their pregnancy, had an increased likelihood of term spontaneous labor.¹⁶ Additionally, a randomized clinical trial conducted from 38 weeks’ gestation up to delivery found that walking for 30 minutes at 3 times/week increased the likelihood of term spontaneous labor.¹⁷ The results of these studies indicate there may be an association between physical activity near term as a trigger for spontaneous labor.

In contrast to the benefits incurred from recreational physical activity during pregnancy, studies have reported a less than favorable association between maternal occupational physical activity, such as walking, lifting, or carrying individuals or objects in a work setting, and preterm birth.^18,19 In a retrospective, nationally representative study of US births from 1997–2009, high levels of maternal occupational physical activity (e.g., climbing, standing, walking, stooping, and handling of materials) during pregnancy were associated with an increased odds of preterm birth.¹⁸ However, current studies on this topic do not always specify whether the preterm labor was spontaneous versus medically indicated, making it difficult to determine whether occupational physical activity is associated with clinical factors that necessitate a preterm delivery versus the onset of preterm spontaneous labor itself.¹⁹

To address the gaps in the current literature, we used a self-matched, case-crossover study design to assess whether physical activity triggered the onset of spontaneous labor. We hypothesized that recreational physical activity would be associated with an increased likelihood of term spontaneous labor, whereas occupational physical activity would be associated with an increased likelihood of preterm spontaneous labor.

SUBJECTS AND METHODS

Study Population

This analysis was performed using data from the third phase of the Pregnancy, Infection, and Nutrition (PIN3) study. PIN3 was a prospective cohort study conducted between January 2001 and June 2005 among 2,006 pregnant participants to collect data on prenatal lifestyle factors associated with preterm birth.²⁰ Singleton pregnancies, less than 20 weeks’ gestation, were recruited during their prenatal visit at 4 different clinics in North Carolina. Eligible participants were English speaking, older than 16 years, planning to receive prenatal care and deliver at a PIN3 study site, and able to access a telephone to complete telephone interviews. Further details regarding the PIN3 protocol have been published on the PIN3 study website.²¹ Data for this analysis came from a telephone interview at 17–22 weeks’ gestation, an interviewer-administered, in-hospital questionnaire completed by each participant after birth, before being discharged, and birth outcomes data abstracted from hospital medical records.²² Written, informed consent was obtained from all participants. Approval for the PIN3 study protocol and this analysis was granted by the Institutional Review Board of the University of North Carolina at Chapel Hill.

Overall, 2,006 pregnant individuals consented to participate in PIN3, although 47 became ineligible after recruitment (43 pregnancy losses, 4 pregnant with multiples), 121 were participating for the second or third time, 174 were lost to follow-up, and 156 completed the PIN3 in-hospital questionnaire before the labor trigger questions were added to the questionnaire, resulting in 1,508 participants. Of those 1,508, we further excluded participants who did not complete the questions on the in-hospital questionnaire required for this analysis (n=115) and for whom physical activity during pregnancy was contraindicated (n=68; determined through the telephone interview by asking each participant “at any time during this pregnancy has a doctor, nurse, or other health professional told you to change your physical activity rather than follow your regular activity routine?”). Thus, a final analytic sample of 1,325 participants was selected for analysis, of which 784 participants reported non-spontaneous labor and 541 reported spontaneous onset of labor (Supplementary Figure 1).

Study Design

This study was based on a self-matched, case-crossover design. This method is used to assess the risk of an acute event (i.e., risk of onset of spontaneous labor) after exposure to a transient risk factor (i.e., physical activity).^23,24 Using the self-matched approach, the case-crossover design compares the same participant at two different time periods, a control period (during which they did not experience the event) and a hazard period (during which they did experience the event). In this study, the hazard period, defined as physical activity performed during the 0 to 24 hours before the onset of spontaneous labor, was compared to the control period, defined as average daily physical activity performed during the 24 hours to 7 days before the onset of spontaneous labor (Figure 1). As a result of the self-matched approach, the need to adjust for within-person confounding is eliminated.

Figure 1.

Self-matched, case-crossover design to explore associations between physical activity and the onset of spontaneous labor^a

^aThe 7-day period before labor includes the 24-hour period before labor. To compare the 2 time periods (6 days versus 24 hours before labor onset), without overlap, we subtracted duration of physical activity during the hazard period out of the control period.

Measures

Spontaneous onset of labor

Spontaneous onset of labor was assessed after birth using an interviewer-administered questionnaire. This questionnaire assessed labor triggers by asking participants to self-report the day and time their labor began, and to identify what they perceived caused their labor to start. This analysis was limited to pregnant participants whose labors began spontaneously, as described previously.²⁵

Physical activity

After birth, during the in-hospital interview, participants were asked to report their moderate to vigorous physical activity. The recall was based on the PIN3 physical activity questionnaire, with evidence for validity and reliability of this method for measuring physical activity in this study population.²⁶ To assess physical activity during these two periods, participants were asked to report the type, frequency, duration, and intensity (“fairly light,” “somewhat hard,” “hard or very hard”) of any physical activity they had participated in during the control and hazard periods.

Any type or mode of physical activity could be reported; the recall questionnaire²⁶ worked through the following broad categories: recreational, occupational, transportation, indoor and outdoor household activities, child and adult care, and any other activities. All responses were recorded verbatim and coded based on the Compendium of Physical Activities.²⁷ If the Compendium did not reflect a participant’s response, a new code was created to represent the specific activity described by the participant.²⁸ Using an inductive approach, the Compendium codes for specific physical activities were collapsed into broad domains. Based on the questionnaire responses, we had sufficient sample size to explore recreational and occupational modes, as well as “any physical activity” included from all modes combined.

Within the in-hospital questionnaire, as part of the case-crossover design, participants were asked to recall their physical activity during 2 different time periods, the 24 hours before labor (hazard period) versus the week before labor (control period). If physical activity was indeed a trigger of spontaneous labor, we would expect, using the case-crossover design, to observe more physical activity during the hazard period than the control period. During the hazard period, within each physical activity category (recreational, occupational, any), participants were coded as “yes” if they reported any physical activity, regardless of intensity, during the 24 hours before their labor began. Similarly, during the control period, participants were coded as “yes” if they reported any physical activity, regardless of intensity, during the week before labor began. Given that the 7-day period before labor includes the 24-hour period before labor, we subtracted duration of physical activity during the hazard period out of the control period so that we could compare 2 time periods (6 days versus 24 hours before labor onset) without overlap.

Covariates

Pre-gravid weight and height were obtained from the medical record and used to calculate pre-pregnancy body mass index (BMI). Race/ethnicity, marital status, education, parity (live plus still births after 20 weeks), number of individuals living in the household, and household income were collected during the first telephone interview (17–22 weeks’ gestation). These latter two variables were used to calculate the percent of the 2001 poverty level²⁹, as a proxy for socio-economic status. The mother’s age at time of conception was first determined from the medical record and verified during the telephone interview. Ultrasonography estimated the due date and was used to calculate gestational age at birth if the procedure was performed before 22 weeks’ gestation (>90% of the PIN3 sample); otherwise, the due date, and subsequently gestational age at birth, was estimated using the date of last menstrual period.

Statistical Methods

Descriptive statistics of analysis variables were calculated separately for the control and hazard periods. For the case-crossover analysis, each participant was in the analysis dataset twice: once using the hazard period as the exposure, after which she by design experienced the outcome, which was spontaneous onset of labor; and once using the control period as the exposure, after which by design she did not experience the outcome. Conditional (matched) logistic regression was then used to assess the relationship between physical activity and onset of spontaneous labor, within each participant by comparing the amount of physical activity in their control period to the amount in their hazard period. In these models, the onset of spontaneous labor was the outcome, and the exposure was physical activity during the hazard versus control periods. This analysis was repeated separately for recreational, occupational, and any physical activity.

To assess selection bias, differences in baseline characteristics between participants who reported onset of spontaneous labor and participants who reported non-spontaneous labor were assessed using paired t-tests for continuous variables and chi-squared tests for categorical variables. To address that daily probability of labor onset will increase in a person nearing full-term, we conducted a sensitivity analysis to explore whether the odds of onset of spontaneous labor differed when stratified by gestational age: preterm (<37 weeks), early term (37-<39 weeks), full term (39-<41 weeks), or post-term (>=41 weeks). The full term and post-term variables were collapsed together due to the small number of post-term births (n=45) in this sample. For all analyses, statistical significance was set at P<0.05 and analyses were conducted using Stata 15 (StataCorp, College Station, TX).

RESULTS

Sample characteristics

The majority of the 1,325 participants were non-Hispanic White, married, and had more than a high school education (Table 1). Almost half of the sample was nulliparous, and 56.1% of the sample had a normal BMI (18.5 kg/m²≤ BMI<25.0). Average gestational age at delivery was 38.7 weeks, and 10.4% of infants were born preterm (<37 weeks’ gestation).

Table 1.

Comparison of sociodemographic characteristics between pregnant participants who went into spontaneous labor and pregnant participants who did not; Pregnancy, Infection, and Nutrition 3 Study

	Overall (N=1,325)	Spontaneous Labor (N=541)		Non-spontaneous Labor (N=784)
Sociodemographic characteristics	n, %	Number of Cases	Cumulative Incidence	Number of Cases	Cumulative Incidence
Age (years)
<25	250 (18.9)	106	42.4	144	57.6
25–34	857 (64.8)	355	41.4	502	58.6
>35	218 (16.5)	80	36.7	138	63.3
Race/ethnicity a
Non-Hispanic, White	982 (74.2)	414	42.2	586	57.8
Non-Hispanic, Black	222 (16.8)	70	31.5	152	68.5
Asian or Pacific Islander	44 (3.3)	18	40.9	26	59.0
Other (not specified)	76 (5.7)	39	51.3	37	48.5
Marital status
Single	246 (18.6)	96	39.0	150	61.0
Married	1,039 (78.4)	433	41.7	606	58.3
Separated or Divorced	38 (2.9)	12	31.6	26	68.4
Widowed	2 (0.2)	0	0.0	2	100.0
Income (% of poverty level for 2001)
≤185	238 (18.7)	84	35.3	154	64.7
186–350	254 (19.9)	101	40.0	153	60.2
>350	784 (61.4)	340	43.4	444	56.6
Education ≤ 12 years a	236 (17.8)	72	30.5	164	69.5
Pre-pregnancy BMI (kg/m2) a
Underweight (<18.5)	57 (4.4)	30	52.6	27	47.4
Normal (18.5–24.9)	734 (56.1)	342	46.6	392	53.4
Overweight (25.0–29.9)	252 (19.3)	89	35.3	163	64.7
Obese (≥30.0)	266 (20.3)	72	27.1	194	72.9
Nulliparous a	644 (48.6)	284	44.1	360	55.9
Gestational age at birth a
Preterm (<37 weeks)	138 (10.4)	55	39.9	83	60.1
Early term (37–<39 weeks)	381 (28.8)	128	33.6	253	66.4
Full or post term (≥39 weeks)	806 (60.8)	358	44.4	448	55.6

^aStatistically significant difference (P <0.05) between pregnant participants who experienced a spontaneous onset of labor compared to those pregnant participants who did not. Statistical significance was determined using a chi-square test for categorical variables.

Physical activity as a labor trigger

Participants (N=1,325) reported performing several different types of physical activity during the week before their labor began, including recreational, household, child-care, occupational, and transportation. Examples of each of these types of activities can be found in Supplementary Table 1.

Of the 541 participants who reported spontaneous onset of labor, twenty-one percent, 26%, and 14% reported participating in any, recreational, and occupational physical activity during only the control period (e.g., 24 hours to 7 days prior to spontaneous onset of labor) (Table 2). Participants who reported physical activity during the control period were more likely to be older (any activity), have over 12 years of education (any, recreational, and occupational), be above 350% of the poverty level (recreational activity), give birth at a later gestational age (any and recreational activity), be nulliparous (recreational and occupational activity), have a lower pre-pregnancy BMI (recreational activity) compared to participants who did not report physical activity during the control period (Supplementary Table 2, 3, and 4).

Table 2.

Distribution of pregnant participants who experienced spontaneous onset of labor in the Pregnancy, Infection, and Nutrition 3 Study according to physical activity type by hazard and control period (N=541)

≥5 minutes/day of physical activity type, % yes	Hazard Period Onlyb, n (%)	Control Period Onlyc, n (%)	Both (Hazard and Control Period), n (%)	Neither, n (%)
Anya	27 (5.0)	116 (21.4)	349 (64.5)	49 (9.0)
Recreationala	40 (7.4)	139 (25.7)	138 (25.6)	224 (41.4)
Occupationala	14 (2.6)	77 (14.2)	31 (5.7)	419 (77.4)

^aStatistically significant difference (P < 0.05) between physical activity during the hazard and control period. Statistical significance was determined using and a chi-square test for categorical variables.

^bHazard period defined as 0 to 24 hours prior to spontaneous onset of labor

^cControl period defined as 24 hours to 7 days prior to spontaneous onset of labor

Five percent, 7%, and 3% of participants reported participating in any, recreational, and occupational physical activity during only the hazard period (e.g., 0 to 24 hours prior to spontaneous onset of labor). During the hazard period, participants who reported any activity were more likely to have over 12 years of education (any and recreational activity), give birth at a later gestational age (any and recreational activity), have a lower pre-pregnancy BMI (recreational activity), and be nulliparous (recreational activity) (Supplementary Table 2 and 3). There were no statistically significant differences between participants who reported any occupational physical activity during the hazard period compared to participants who did not (p>0.05) (Supplementary Table 4).

By design, case-crossover analyses of physical activity as a labor trigger are limited to those participants whose physical activity behaviors during the control and the hazard period are discordant. Compared with the 24 hours to 7 days before labor (control period), immediate odds of spontaneous labor were lower with at least 5 to at least 30 minutes per day of any physical activity (OR_{≥30 minutes}: 0.74, 95% CI: 0.54–1.00), at least 5 to at least 10 minutes of recreational physical activity (OR_{≥10 minutes}: 0.45, 95% CI: 0.32–0.62), or at least 5 to at least 45 minutes of occupational physical activity (OR_{≥45 minutes}: 0.39, 95% CI: 0.18–0.85) in the hazard period. However, the immediate odds of spontaneous labor were increased with at least 30 to at least 60 minutes of recreational physical activity (OR_{≥30 minutes}: 1.54, 95% CI: 1.09–2.15). There were no associations with onset of spontaneous labor after 45 minutes of any physical activity, between 15 to 20 minutes of recreational physical activity, or after 60 minutes of occupational physical activity in the hazard period (Table 3; Left-hand column in Figure 2).

Table 3.

Odds ratios of spontaneous labor by duration and physical activity type for pregnant participants in the Pregnancy, Infection, and Nutrition 3 Study

	Any physical activity			Recreational physical activity			Occupational physical activity
Duration (minutes/day)	Hazard Onlya	Control Onlyb	Odds Ratio (95% CI)	Hazard Onlya	Control Onlyb	Odds Ratio (95% CI)	Hazard Onlya	Control Onlyb	Odds Ratio (95% CI)
≥5	27	116	0.23 (0.15–0.35)	40	139	0.29 (0.20–0.41)	14	77	0.18 (0.10–0.32)
≥10	33	125	0.26 (0.18–0.39)	53	119	0.45 (0.32–0.62)	11	66	0.17 (0.09–0.32)
≥15	50	114	0.44 (0.31–0.62)	68	93	0.73 (0.53–1.00)	17	57	0.30 (0.17–0.51)
≥20	60	118	0.51 (0.37–0.69)	84	77	1.09 (0.80–1.49)	17	47	0.36 (0.21–0.63)
≥30	73	99	0.74 (0.54–1.00)	86	56	1.54 (1.09–2.15)	14	29	0.48 (0.26–0.91)
≥45	91	85	1.07 (0.80–1.44)	70	32	2.19 (1.44–3.32)	9	23	0.39 (0.18–0.85)
≥60	86	77	1.12 (0.82–1.52)	58	20	2.90 (1.74–4.82)	7	16	0.44 (0.18–1.06)

^aHazard only indicates participant reported physical activity during the 0 to 24 hours prior to spontaneous onset of labor, but not during the 24 hours to 7 days prior to labor onset.

^bControl only indicates participant reported physical activity during the 24 hours to 7 days prior to spontaneous onset of labor, but not during the 0 to 24 hours prior to labor onset.

Figure 2.

The odds of onset of spontaneous labor by type and duration of physical activity and gestational age at delivery among pregnant participants in the Pregnancy, Infection, and Nutrition 3 Study. Error bars indicate 95% confidence intervals.

Sensitivity analyses

After stratification by gestational age, the associations between any, recreational, or occupational physical activity and the onset of spontaneous labor among participants who delivered preterm were imprecise and inconsistent across gestational age categories (Figure 2; Supplementary Tables 5–7). There were no meaningful associations identified between physical activity (any, recreational, or occupational) and the onset of spontaneous preterm labor, though small sample sizes and resulting wide confidence intervals were an issue.

Among participants who delivered early term the decreased likelihood of spontaneous labor onset after moderate amounts of any, recreational, or occupational physical activity in the hazard period remained significant (Column 3 in Figure 2; Supplementary Tables 5 and 7). Among those whose pregnancies ended at 39 weeks or later (full term or post term), all durations of occupational activity were associated with decreased odds of spontaneous labor, but only moderate durations of any or recreational activity were associated with this decrease (Column 4 in Figure 2; Supplementary Tables 6). Large amounts of recreational activity remained associated with increased odds of spontaneous onset of labor during the full term and post term periods.

Assessment of selection bias

Non-spontaneous labor was more common among participants who delivered preterm, early term, full or post term. There were few differences between participants who went into spontaneous labor compared to those who did not (Table 1). Compared to participants who did not experience spontaneous labor, fewer participants in the spontaneous labor group reported 12 years or less of education (13.3% versus 20.9%), and participants in the spontaneous labor group were more likely to be Non-Hispanic, White (76.5% versus 72.5%), nulliparous (52.5% versus 46.0%), have a lower mean pre-pregnancy BMI (24.2 ± 5.5 kg/m² versus 26.7 ± 7.7 kg/m²). Age, marital status, percent of poverty level for 2001, and gestational age at birth were not significantly different between participants who experienced onset of spontaneous labor and those who did not.

DISCUSSION

Based on previous studies that have reported an association between recreational physical activity at term and spontaneous labor onset^16,17, we hypothesized that engaging in recreational physical activity at term would result in an immediate increased likelihood of spontaneous labor. Among participants who reported engaging in recreational physical activity for at least 30 minutes, the odds of spontaneous labor were increased during the subsequent 24 hours. Additionally, based on the association between occupational physical activity and preterm birth^18,19,30,31, we hypothesized that engaging in occupational physical activity would result in an immediate increased likelihood of spontaneous preterm labor. We found that participants who reported engaging in occupational physical activity for at least 5 to at least 45 minutes had decreased odds of spontaneous labor within the subsequent 24 hours.

Reverse causality is a possible explanation for the observed association between any physical activity and decreased odds of spontaneous labor onset. Late in pregnancy, individuals have commonly reported a lack of energy, difficulty sleeping, and shortness of breath.^32–34 Imminent labor may exacerbate these symptoms; whereas in a person whose body is not actively preparing for labor, she may feel more able and willing to be physically active. Likewise, a person who suspects labor is imminent, due to symptoms like the loss of their cervical mucus plug³⁵ or mild contractions, might try to reserve their energy in preparation for the labor itself. However, we were unable to discern the direction of these relationships, as our available data did not ask participants to discuss why they chose to or chose not to engage in physical activity.

Similarly, the immediate odds of spontaneous labor were decreased within 24 hours of engaging in occupational physical activity. After stratification by gestational age, these associations remained significant among those whose pregnancies were delivered early, full or post term. After stratification by gestational age, we rejected our hypothesis that occupational physical activity was associated with labor onset among preterm pregnant individuals. However, the statistical power to detect such an association was reduced leading to imprecise results. The association between occupational physical activity and a decrease in the likelihood of spontaneous labor may be explained by reverse causality and healthy worker bias. In the general population, individuals who can work are considered healthier than the overall population. Simply put, the overall population includes both individuals who can work as well as individuals who are too sick to work.³⁶ In this study, pregnant individuals who were able to engage in occupational physical activity may have been feeling more energetic and had fewer, or less severe, symptoms of imminent labor compared to those pregnant individuals did not engage in occupational physical activity. Therefore, it is likely that the choice to engage in occupational physical activity does not reduce ones’ chances of spontaneous labor, rather indicates that labor is not imminent.

Consistent with our hypothesis and results from a recent randomized clinical trial,¹⁷ the immediate odds of spontaneous labor were increased within 24 hours of engaging in at least 30 minutes of recreational physical activity among full or post term pregnant individuals. Although the link between recreational physical activity at term and the onset of spontaneous labor is largely unexplained, the physiological changes required for labor to begin may help provide a potential mechanism for this finding.^37,38 For labor to begin, progesterone levels decrease, while estrogen levels increase to allow for cervical ripening and promotion of uterine contractions. This rise in estrogen leads to increased production of oxytocin to stimulate uterine contractions. Human and animal studies have observed oxytocin levels also increase in response to recreational physical activity.^39,40 Furthermore, animal studies have reported females have a greater increase in oxytocin after exercise, compared to males.⁴⁰ Therefore, among term pregnancies that have already started to undergo the hormonal shifts required for birth to occur, it may be possible for a bout of recreational physical activity to further increase oxytocin levels, allowing for contractions, and subsequently spontaneous labor, to begin. We also observed odds of spontaneous labor were decreased with 24 hours of engaging in up to 10 minutes of recreational physical activity. We suggest future studies seeking to assess physical activity as a trigger of labor employ the use of prospective data collection in which participants prospectively record which physical activities they engage in, noting reasons why they stopped each session, as it would provide more insight into this relationship.

Nevertheless, there are two possible alternative explanations to explain the association between recreational physical activity and spontaneous labor. First, each day a pregnant person remains pregnant, their likelihood of labor increases. Based on the case-crossover design of this study, the hazard period (0 to 24 hours before labor onset) occurs after the control period (24 hours to 7 days before labor onset); meaning a person’s likelihood of labor during the hazard period is, by definition, greater than the control period. While we were able to stratify our results by gestational age, the case-crossover study design makes it impossible to control for exact gestational age in the model.

Second, like any and occupational physical activity, the association between recreational physical activity and onset of spontaneous labor could be due to reverse causality. While systematic reviews have been unable to find sufficient evidence to support a relationship between recreational physical activity and the onset of labor,^41,42 empirical evidence among popular media sources suggest recreational physical activity aids in labor onset.⁴³ A person who has reached full term may wish to stimulate labor and be more inclined to engage in recreational physical activity in hopes of inducing labor sooner. In this scenario, the recreational activity may be leading to labor onset, but we cannot ascertain the direction of the association with the data we have. Moreover, the imminent labor would be the cause of the person deciding to engage in recreational physical activity, rather than the recreational physical activity leading to labor onset.

Strengths and Limitations

To our knowledge, this study is among the first to assess physical activity as a trigger of spontaneous labor onset. Methodologically, it can be especially difficult to collect maternal physical activity data during the week prior to labor. While some studies have evaluated whether consistent physical activity during pregnancy is associated labor outcomes, our study is the first to evaluate whether physical activity may trigger the onset of spontaneous labor, providing potential insight into behaviors that can induce spontaneous labor. Additional strengths of this study include a large sample size, detailed physical activity and labor onset data collected in-hospital immediately after birth, and the use of a self-matched, case-crossover design which eliminated the need to adjust for confounders.

Limitations to this analysis should be considered. First, PIN3 took place in a teaching hospital and required participants to be receiving prenatal care at an affiliated clinic, which may limit generalizability of these findings. Additionally, data collection occurred before the establishment of the Affordable Care Act, meaning individuals, especially low-income individuals, were less likely to have access to prenatal care. Second, the physical activity exposure and spontaneous labor onset outcome measures were self-reported; thus, we were unable to estimate measurement error for our exposure and outcome measure. Additionally, validity of the spontaneous labor onset outcome was not assessed. Yet, given that participants were not aware of this study hypothesis, any misclassification is likely to be non-differential-except among participants who experienced spontaneous preterm labor as they may be subject to recall bias since they may remember their physical activity in the 7 days prior to spontaneous labor differently than participants who experienced full or post term labor, potentially leading to differential misclassification. However, if this were indeed the case, we would have expected more participants who experienced spontaneous preterm labor to report increased durations of physical activity during the hazard period versus the control period which would ultimately have biased the results away from the null. This was not observed in our data. Third, we were unable to assess the relationship between physical activity intensity and labor onset because of small sample sizes within the moderate-to-vigorous category, as most participants preparing for labor were not engaging in moderate-to-vigorous physical activity. Lastly, data were collected retrospectively for only two time periods (24 hours prior to labor onset and 7 days prior to labor onset). Prospective physical activity data collection, perhaps using an objective method of data collection such as accelerometry, would allow researchers to evaluate multiple hazard periods (e.g., 12 hours instead of 24) and lessen the potential for reverse causation.

CONCLUSION

Participants who reported any or occupational physical activity during pregnancy had an immediate decreased likelihood of spontaneous labor within the subsequent 24 hours, while participants who reported at least 30 minutes of recreational physical activity had an immediate increased likelihood of spontaneous labor. Future investigations should aim to evaluate causality within these findings. It is plausible that a lack of energy to engage in any or occupational physical activity may be a potential sign of approaching labor. Similarly, pregnant individuals who suspect their labor is near, or who have reached term, may feel inclined to engage in recreational activity to encourage labor onset. These results support current physical activity recommendations for pregnant individuals and potentially indicate a use for recreational physical activity to aid in the natural induction of labor at term.