Rudra, Carole B.*†; Williams, Michelle A.*†; Lee, I-Min‡§; Miller, Raymond S.*; Sorensen, Tanya K.¶
Gestational diabetes mellitus is defined as the onset or first recognition of glucose intolerance during pregnancy and affects approximately 4% to 7% of pregnancies.1 This condition is associated with fetal sequelae such as macrosomia, jaundice, hypoglycemia, polycythemia, hypocalcemia, and birth trauma.1–4 Women with gestational diabetes are more likely than normoglycemic women to experience other pregnancy complications such as preeclampsia, to experience gestational diabetes in a subsequent pregnancy, and to develop overt diabetes after pregnancy.5–7
Although few modifiable risk factors for gestational diabetes have been identified,8–10 results from 4 studies suggest that efforts to increase maternal physical activity before and during pregnancy may reduce risk.10–13 Such an association is consistent with randomized clinical trials showing that leisure-time activity reduces the risk of type 2 diabetes in men and nonpregnant women.14,15 Additionally, aerobic dance and walking have been associated with a reduction in plasma insulin during pregnancy.16
Previous studies of this relation have used measurements based on absolute time and exertion in physical activity, neither of which captures the perception of required effort.10–13,17 For instance, brisk walking may require more physical effort by an overweight, unfit woman than by a lean, fit woman.18 The Borg scale is a measure of exertion that allows women to subjectively rate the intensity of physical effort.19 The measure is thus relative to a woman's physical fitness. The scale is commonly used during exercise stress testing and correlates with heart rate, blood lactate, and oxygen consumption in nonpregnant humans.20 We investigated whether perceived exertion during recreational physical activity in the year before pregnancy is associated with risk of gestational diabetes. We hypothesized that perceived exertion might capture aspects of maternal fitness that are not otherwise captured by measuring time spent in specific activities and that these aspects may be related to risk of gestational diabetes. To consider the biologic and epidemiologic relations among adiposity, gestational diabetes, and physical activity,10–13 we examined the relation of perceived exertion and gestational diabetes among subgroups of lean and overweight women. We also measured the association among women who did and did not meet physical activity guidelines in the year before pregnancy. By using information from both a case–control and a cohort study, we were able to examine the extent to which selection and recall biases may have affected the results from the case–control study while taking advantage of its increased statistical power.
This report is an expansion of published analyses of these 2 study populations.12,13 In the case–control population, we had previously detected an inverse association between energy expenditure and risk of gestational diabetes (comparing ≥30 metabolic-equivalent [MET]-hours/week versus none, adjusted odds ratio [OR] = 0.42; 95% confidence interval [CI] = 0.22–0.78).13 In the cohort population, the adjusted OR comparing ≥21.1 MET-hours/week versus none was 0.26 (CI = 0.10–0.65).12 We repeat these analyses in a slightly different form in this article to examine the relation between perceived exertion and gestational diabetes within subgroups of women categorized by energy expenditure. Also, this allows readers to directly compare the strengths of the perceived exertion and absolute exertion associations.
This analysis is based on 2 studies conducted at Swedish Medical Center and Tacoma General Hospital in Seattle and Tacoma, Washington.12,13 The procedures were in agreement with the protocols approved by the Institutional Review Boards of both institutions. All participants provided written informed consent. No subject participated in both studies, and no participant was observed over multiple pregnancies in the cohort study.
Study Design and Population
The Alpha case–control study (1998–2002) was designed primarily to examine the epidemiology of preeclampsia. In addition to preeclamptic women, we recruited English-speaking patients in labor and delivery wards with a diagnosis of gestational diabetes. We used abstracts of medical records to identify women meeting the then-current diagnostic criteria for gestational diabetes mellitus described by the National Diabetes Data Group.21 Women were routinely screened for gestational diabetes between 24 and 28 weeks gestation using a 50-g 1-hour oral glucose tolerance test. Those with a posttest plasma glucose level >140 mg/mL participated in the diagnostic 100-g 3-hour oral glucose tolerance test. Women were classified as having gestational diabetes using 3-hour glucose tolerance test cut points according to National Diabetes Data Group criteria.21 Controls were English-speaking women who remained normotensive and who did not meet the criteria for gestational diabetes according to their abstracted medical records. Eighty-three percent of 288 eligible gestational diabetes cases and 58% of 866 eligible controls participated. Reasons for nonparticipation included lack of time, no interest in the study goals, and missed appointments. Enrollment continued after the publication of the previous case–control study analysis. Thus, the study population is slightly larger in this report.13
Data Collection and Physical Activity Assessment
Data collection occurred during the postpartum hospital stay. Trained interviewers administered a structured questionnaire on maternal sociodemographic, medical, reproductive, and lifestyle characteristics conducted in English. Self-reported height and weight 3 months before pregnancy were used to calculate body mass index (BMI; weight [kg]/height [m2]). A food frequency questionnaire was used to measure dietary intake during the 3 months before pregnancy as well as during pregnancy. The food frequency questionnaire has been validated in nonpregnant populations, but not in pregnant women.22 Maternal and infant medical records were reviewed to abstract detailed information on antepartum, labor, and delivery characteristics.
During interviews, we asked women questions regarding type, frequency, and duration of recreational activities in the year before pregnancy. These questions were derived from the Stanford Seven-Day Physical Activity Recall and the Minnesota Leisure-Time Physical Activity Questionnaire, which have been validated among men and nonpregnant women.23,24 Specifically, we asked the following: 1) “During the year before you became pregnant, what sports or recreation have you actively participated in?” For each activity, 2) “Which activity did you participate in on a regular basis during the year before you became pregnant?” We did not define “regular” for the participants. 3) “How many times per week?” 4) “How many months did you regularly participate in this activity?” 5) “How much time did you spend at the activity per episode?” To assess perceived exertion we asked, “During the year before you became pregnant, when you were exercising in your usual fashion, how would you rate your level of exertion?” Women based their score on the Borg scale of perceived exertion displayed for them during the interview.19 We classified perceived exertion as none to weak (0 to 2), moderate (3 or 4), strenuous (5 or 6), and very strenuous to maximal (7 to 10).
We assigned each activity an MET score using a standardized classification procedure.18 Using methods described previously, we calculated energy expenditure (MET-hours/week) during prepregnancy physical activity.13 This measure integrates time spent and the MET score of each activity. We designated inactive women (those reporting no regular activities) as the referent group. Using a priori cut points chosen for ease of interpretation, we categorized the remaining active women as follows: 0.1 to 14.9, 15.0 to 29.9, and ≥30.0 MET-hours/week. These cut points are slightly different than those presented in the previous report,13 which were based on approximate quartiles among active women.
We excluded 23 participating cases (10%) and 30 participating controls (6%) from this analysis due to missing information on physical activity, resulting in a final analytic population of 216 cases and 472 controls. We examined the frequency distributions of maternal characteristics according to case–control status. We used multivariable logistic regression to estimate the OR and 95% CI of gestational diabetes according to levels of perceived exertion and energy expenditure during prepregnancy recreational physical activity. We modeled these 2 activity measures separately using indicator variables. The covariates shown in Table 1 were evaluated as confounders on the basis of literature documenting their associations with physical activity or gestational diabetes.8–13 To assess confounding, we entered each covariate into a logistic regression model relating perceived exertion to gestational diabetes separately. We compared the models with and without adjustment for the covariate. We included in the final model covariates that altered unadjusted physical activity coefficients by 10% or more.25 For each covariate, we chose the coding specification that achieved the greatest control of confounding. When multiple specifications achieved similar control, we chose the most parsimonious one. We examined models with and without adjustment for prepregnancy BMI due to its plausible role as either a confounder or a causal intermediate.
We hypothesized that maternal adiposity or absolute physical activity intensity may modify the relation between perceived exertion and risk of gestational diabetes. Thus, we measured this relation among subgroups defined by prepregnancy BMI and prepregnancy energy expenditure. We categorized women as overweight (≥25 kg/m2) or lean (<25 kg/m2). We dichotomized energy expenditure at 10 MET-hours/week, a value approximately equivalent to a U.S. national physical activity recommendation.26
We repeated all analyses after excluding 34 cases of gestational diabetes that also had preeclampsia or pregnancy-induced hypertension. Because this exclusion did not appreciably change the results, we report here the analyses based on the entire study population.
Study Design and Population
The Omega Study, an ongoing cohort study designed primarily to examine dietary risk factors for preeclampsia, began in 1996. Women who initiated prenatal care before 16 weeks gestation at hospital-affiliated clinics were eligible. Women were ineligible if they were less than 18 years old, did not speak and read English, did not plan to carry the pregnancy to term, or did not plan to deliver at either hospital. Emphasis was on recruitment of nulliparous women, although multiparous women were enrolled when study personnel were available. Of the 1,219 women invited through 2002, 1,000 (82%) consented to participate.
Data Collection and Physical Activity Assessment
Data collection occurred at or soon after enrollment (13.2 weeks gestation, on average). Trained interviewers administered a questionnaire nearly identical to that used in the Alpha Study and provided the same food frequency questionnaire.22 Diagnosis of gestational diabetes was made after medical record abstraction of glucose tolerance test results.21 We used methods described here to assess prepregnancy activity. As a result of small numbers, we combined energy expenditure categories, resulting in categories of 0, 0.1 to 14.9, and ≥15.0 MET-hours/week. These cut points are different than those of the previous related study.12 Again due to small numbers, we categorized perceived exertion as none to moderate (score 0 to 4), strenuous (5 or 6), and very strenuous to maximal (7 to 10).
We excluded 103 participants (10%), resulting in an analytic population of 897 women. Of these 103 women, 36 were lost to follow up, 25 experienced spontaneous abortions, 6 voluntarily terminated their pregnancies, 6 had preexisting diabetes mellitus, 3 were missing glucose tolerance test data, and 27 were missing data on perceived exertion. (In the previous published analysis of this population,12 12 of these 27 women were included because they reported prepregnancy activities used in calculating absolute exertion measures. Thus, the sample size for that analysis was slightly larger.) We examined the frequency distributions of maternal characteristics according to perceived exertion. We used logistic regression as described here to estimate ORs of gestational diabetes according to levels of prepregnancy perceived exertion and energy expenditure separately. Due to the small number of cases, fitting regression models with more than a few covariates resulted in near-infinite or immeasurable coefficients. However, exploratory modeling revealed little evidence of confounding. Like in the Alpha Study, we chose the most parsimonious model and covariate specifications that achieved the greatest control of confounders.
The Alpha Study
Table 1 shows distributions of selected maternal characteristics according to case status. We found differences according to case status for several of these distributions, including age, race, nulliparity, and body mass index, among others.
Perceived exertion was inversely associated with risk of gestational diabetes mellitus after adjustment for maternal age, race/ethnicity, prepregnancy hypertension, and nulliparity (data not shown). Further adjustment for BMI did not meaningfully change the estimated ORs, which are shown in Table 2 (range of changes = 3–15%). Women reporting moderate usual exertion had a 59% risk reduction compared with women reporting negligible or weak exertion (OR = 0.41; CI = 0.21–0.82). Those reporting strenuous exertion also were at decreased risk compared with the referent group (0.26; 0.14–0.47). Women engaging in very strenuous to maximal exertion were 81% less likely to develop gestational diabetes than the referent group (0.19; 0.15–0.50). This pattern in risk reduction was somewhat stronger than that observed after classifying women according to energy expenditure.
In a post hoc analysis, we measured the association between perceived exertion and gestational diabetes while adjusting for energy expenditure. The purpose of this analysis was to see whether perceived exertion remained associated with risk of gestational diabetes after adjustment for absolute exertion. Perceived exertion and energy expenditure were moderately correlated within cases (Spearman correlation [ρ] = 0.54) and controls (ρ = 0.43). When both measures were modeled together and adjusted for confounders, the perceived exertion association remained essentially the same, whereas the energy expenditure association weakened (data not shown). For instance, the OR for strenuous to maximal, compared with negligible, perceived exertion was 0.26 (0.12–0.54), whereas the OR for ≥30.0 MET-hours/week versus none was 0.96 (0.49–1.90).
We examined the relation between perceived exertion and risk of gestational diabetes according to BMI and energy expenditure (Table 3). The association did not differ substantially between overweight women (≥25 kg/m2) and lean women (<25 kg/m2). Similarly, the association did not meaningfully differ according to energy expenditure. Among women who did not meet physical activity recommendations in the year before pregnancy (<10 MET-hours/week),26 risk of gestational diabetes among those reporting very strenuous to maximal perceived exertion was 71% lower than that for women reporting negligible or weak exertion (0.29; 0.12–0.68). Among women meeting or exceeding recommendations (≥10 MET-hours/week), risk reduction associated with very strenuous to maximal perceived exertion was similar (0.14; 0.02–1.13).
The Omega Study
Distributions of selected maternal characteristics according to usual perceived exertion are presented in Table 4. Nine percent of Omega Study participants (78 of 897) reported no regular physical activity, and 32% (286 of 897) reported negligible to moderate usual exertion (Table 5).
Both perceived exertion and energy expenditure were inversely associated with risk of gestational diabetes in the Omega Study (Table 5). In contrast to the Alpha Study, the association with energy expenditure was stronger than the perceived exertion association. Adjusted ORs among women reporting strenuous and very strenuous to maximal exertion were 0.63 (0.31–1.29) and 0.57 (0.24–1.37), respectively, compared with negligible to moderate exertion. Women reporting ≥15.0 MET-hours/week experienced an 86% risk reduction (0.14; 0.05–0.38) versus inactive women.
Perceived exertion during usual recreational physical activity in the year before pregnancy appears to be strongly and inversely related to the risk of subsequent gestational diabetes mellitus. Perceived exertion appears to be associated with reduced risk of gestational diabetes even among women who are overweight before pregnancy. These results also suggest that women who do not meet physical activity recommendations in the year before pregnancy may benefit from engaging in activity that they experience as strenuous.
By analyzing data from 2 study designs, we were able to capitalize on the strengths of each while minimizing the impact of inherent limitations. The Alpha Study afforded the statistical power to examine the relation of interest among subgroups. The low participation rates, particularly among controls, suggest the possibility that our case and control groups may not have been representative of the underlying populations from which they were sampled. However, characteristics of participating control subjects were similar to those of all women delivering at the study hospitals.13 Case and control participants may have also differed in their ability and willingness to report physical activity characteristics. Although the magnitude of the point estimates differed between the 2 studies, the general consistency of the associations suggests that recall and selection biases are unlikely explanations of the Alpha Study findings.
Among Alpha participants, the association with perceived exertion was stronger than that with energy expenditure. The opposite pattern occurred among Omega participants. The extent to which we can interpret this difference is limited by the overlap of the perceived exertion and energy expenditure confidence intervals within each study, as well as the difference in activity measure categorizations between the 2 studies.
The findings from our post hoc analysis of Alpha Study data using a 2-measure model are consistent with results from stratified analyses. However, inferences based on the 2-measure model may be limited by collinearity between perceived exertion and energy expenditure. Nevertheless, both results suggest that perceived intensity of usual prepregnancy activity is related to risk of gestational diabetes independent of energy expenditure. We speculate that perceived exertion captures a component of physical fitness that is not captured by integrating the amount and absolute intensity of usual exercise. Overall fitness may be equally or more strongly related to risk of gestational diabetes, compared with prepregnancy activity patterns.
We note that these measures of perceived exertion and absolute exertion are not entirely concordant, ie, women could report no regular participation in activities yet also report a Borg rating greater than zero. Perceived exertion among women who report no regular activity likely reflects the effort they experienced during sporadic activity that they did not deem as usual.
Participants of both studies were more likely to engage in physical activity than the overall American population. Twenty percent of Alpha control subjects and 9% of Omega participants reported no regular prepregnancy physical activity, whereas in a 1998 national survey, 59% of American women aged 18 to 44 reported never engaging in activity lasting 10 or more minutes a week.27 Our study participants were more similar in physical activity patterns to Washington State women; in a 2000 state survey, 18% reported no participation in recreational physical activity during the previous month.28 Additionally, sociodemographic and behavioral characteristics such as race/ethnicity, education, BMI, and smoking among our participants were more similar to those among Washington State women than those among American women in general.27
Previous studies have documented relations between physical activity intensity and risk of gestational diabetes.10–13 In one large registry-based study, women who were inactive during pregnancy experienced a 1.9-fold increased risk (95% CI = 1.2–3.1) compared with women who exercised regularly. However, this association was limited to women with a prepregnancy BMI of 33 kg/m2 or more.11 Among 14,613 participants in the Nurses’ Health Study II, women who reported vigorous activity 4 or more times a week during the year before pregnancy were less likely to develop gestational diabetes than those reporting less than once a week (OR = 0.8, 95% CI = 0.6–1.3).10 In analyzing data from the same prospective cohort examined in this analysis, Dempsey et al12 noted inverse relations between risk of gestational diabetes and physical activity in the year before pregnancy, as well as physical activity in the first 20 weeks of pregnancy.12 A previous analysis of absolute exertion using the case–control data analyzed here produced similar results.13
Measures of relative intensity have recently been examined in relation to chronic diseases. One study demonstrated an inverse association between perceived exertion and risk of coronary heart disease among a cohort of middle-aged men, even among those who did not meet current activity recommendations.29 The associations with gestational diabetes observed here among reproductive-aged women who did not meet activity recommendations before pregnancy are similarly compelling.
These results should not be extrapolated to early and midpregnancy activities, because we did not assess usual perceived exertion during pregnancy. Nonetheless, these findings motivate further examination of perceived exertion in pregnancy in relation to pregnancy outcomes. Physiological changes such as nausea and fatigue, weight gain, and increased joint laxity may alter both participation in activity and the perception of exertion.30 Pregnancy-associated perceptual changes would not be captured using absolute exertion measurements.
This study adds to others supporting pregnancy-related benefits of physical activity both before and during pregnancy.10–13 Collectively, this literature motivates additional epidemiologic research to document the pregnancy-related effects of physical activity. Given the activity habits of most American women, examination of the potential impact of low- and moderate-intensity activities on pregnancy-related outcomes is particularly warranted. These results also suggest that perceived exertion may be a useful addition to the epidemiologic assessment of physical activity and fitness during pregnancy—a period of great physiological and, often, behavioral change.
We thank the staff at the Center for Perinatal Studies for their expert assistance.
1. American Diabetes Association. Gestational diabetes mellitus. Diabetes Care. 1999;22(suppl):S74–S76.
2. O'Sullivan JB. Long term follow-up of gestational diabetes. In: Camerini Davalos RA, Cole HS, eds. Early Diabetes. New York: Academic Press; 1984:1009–1027.
3. Jovanovic-Peterson L, Peterson CM. Is exercise safe or useful for gestational diabetic women? Diabetes. 1991;40(suppl):179–181.
4. National Diabetes Data Group of the National Institute of Diabetes and Digestive and Kidney Disease. Diabetes in America, 2nd ed. Bethesda: National Institutes of Health; 1995.
5. Suhonen L, Teramo K. Hypertensive and preeclamptic women with gestational glucose intolerance. Acta Obstet Gynecol Scand. 1993;72:269–272.
6. Foster-Powell KA, Cheung N. Recurrence of gestational diabetes. Aust N Z J Obstet Gynaecol. 1998;38:384–387.
7. Damm P. Gestational diabetes mellitus and subsequent development of overt diabetes mellitus. Dan Med Bull. 1998;45:495–509.
8. Dooley SL, Mertzger BE, Cho N, Liu K. The influence of demographic and phenotypic heterogeneity in the prevalence of gestational diabetes mellitus. Int J Gynaecol Obstet. 1991;35:13–18.
9. Berkowitz GS, Lapinski RH, Wein R, Lee D. Race/ethnicity and other risk factors for gestational diabetes. Am J Epidemiol. 1992;135:965–973.
10. Solomon CG, Willett WC, Carey VJ, et al. A prospective study of pregravid determinants of gestational diabetes mellitus. JAMA. 1997;278:1078–1083.
11. Dye TD, Knox KL, Artal R, Aubrey RH, Wojtowycz MA. Physical activity, obesity, and diabetes in pregnancy. Am J Epidemiol. 1997;146:961–965.
12. Dempsey JC, Sorensen TK, Williams MA, et al. Prospective study of gestational diabetes mellitus risk in relation to maternal recreational physical activity before and during pregnancy. Am J Epidemiol. 2004;159:663–670.
13. Dempsey JC, Butler CL, Sorensen TK, et al. A case–control study of maternal recreational physical activity and risk of gestational diabetes mellitus. Diabetes Res Clin Pract. 2004;66:203–215.
14. Laaksonen DE, Lindstrom J, Lakka TA, et al. Physical activity in the prevention of type 2 diabetes: the Finnish Diabetes Prevention Study. Diabetes. 2005;54:158–165.
15. Knowler WC, Barrett-Connor E, Fowler SE, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002;346:393–403.
16. Clapp JF III, Capeless EL. The changing glycemic response to exercise during pregnancy. Am J Obstet Gynecol. 1991;165:1678–1683.
17. Howley ET. Type of activity: resistance, aerobic and leisure versus occupational physical activity. Med Sci Sports Exerc. 2001;33(suppl):S364–369.
18. Ainsworth BE, Haskell WL, Leon AS, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc. 1993;25:71–80.
19. Borg GAV. Psychophysical bases of perceived exertion. Med Sci Sports Exerc. 1982;14:373–381.
20. American College of Sports Medicine. The recommended quantity and quality of exercise for developing and maintaining cardiorespiratory and muscular fitness and flexibility in healthy adults. Med Sci Sports Exerc. 1998;30:975–991.
21. Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the Expert Committee on the Diagnosis and Classification of diabetes Mellitus. Diabetes Care. 1997;20:1183–1197.
22. Patterson RE, Kristal AR, Tinker LF, Carter RA, Bolton MP, Agurs-Collins T. Measurement characteristics of the Women's Health Initiative food frequency questionnaire. Ann Epidemiol. 1999;9:178–187.
23. Jacobs DR Jr, Ainsworth BE, Hartman TJ, et al. A simultaneous evaluation of 10 commonly used physical activity questionnaires. Med Sci Sports Exerc. 1993;25:81–91.
24. Richardson MT, Leon AS, Jacobs DR Jr, et al. Comprehensive evaluation of the Minnesota Leisure Time Physical Activity Questionnaire. J Clin Epidemiol. 1994;47:271–281.
25. Rothman KJ, Greenland S. Modern Epidemiology, 2nd ed. Philadelphia: Lippincott-Raven Publishers; 1998.
26. Department of Health and Human Services. Healthy People 2010: Understanding and Improving Health. Washington, DC: US Department of Health and Human Services; 2001.
27. National Center for Health Statistics. Summary Health Statistics for US Adults: National Health Interview Survey, 1998. Hyattsville, MD: National Center for Health Statistics; 2002.
28. Behavioral Risk Factor Surveillance System [database online]. Atlanta: US Department of Health and Human Services, Centers for Disease Control and Prevention; 2000.
29. Lee IM, Sesso HD, Oguma Y, Paffenbarger RS Jr. Relative intensity of physical activity and risk of coronary heart disease. Circulation. 2003;107:1110–1116.
30. Wolfe LA, Weissgerber TL. Clinical physiology of exercise in pregnancy: a literature review. J Obstet Gynaecol Can. 2003;25:473–483.
© 2006 Lippincott Williams & Wilkins, Inc.