The prevalence of gestational diabetes mellitus (GDM) ranges between 0.7% and 10.1% because of differences in race and diagnostic criteria, and it is increasing yearly.1–4 Although the early detection of GDM is important for reducing adverse pregnancy outcomes, its diagnosis is still controversial.5–7 The two-step approach, a 50-g glucose challenge test combined with a 100-g oral glucose tolerance test (OGTT), is supported by the American College of Obstetricians and Gynecologists and has been used at our hospital for 10 years. Our previous study showed significant relationships between adverse pregnancy outcomes and GDM diagnosed using the two-step screening approach (Wang P, Lu MC, Yan YH. Abnormal glucose tolerance is associated with preterm labor and increased neonatal complications in Taiwanese women. Taiwanese J Obstetrics Gynecol).
The American College of Obstetricians and Gynecologists has stated that the 50-g glucose challenge test can be performed “without regard to time of last meal or time of day.”8 Although this makes it convenient for pregnant women and physicians to schedule testing, it is unclear whether food intake influences the predictive value of the glucose screening. Few studies have evaluated the effects of fasting or feeding on the outcome of the glucose challenge test, and the results were discordant, likely because of differences in study design, diagnostic criteria, fasting time cutoffs, and race.9–12 The purpose of our study was to investigate the influence of prior food ingestion on the predictive value of the GDM screening test at 24–28 weeks of gestation in Taiwanese women. We hypothesized that fasting would lead to a higher predictive value than not fasting before the glucose screening test. We calculated the glucose challenge test positivity rate and the positive predictive value (PPV) to detect the overall predictive value of the GDM screening test.
MATERIALS AND METHODS
All of the pregnant women in this prospective study were recruited from the obstetrics clinic of one teaching hospital (Chia-Yi Christian Hospital, 1,000 beds) in Taiwan between March and October 2011. The study was approved by the institutional review broad of Chia-Yi Christian Hospital (No. 100030), and all of the participants gave informed consent to participate in the study. Blood glucose levels were measured at the central laboratory of Chia-Yi Christian Hospital using a standard clinical protocol. Plasma glucose levels were measured using a Hitachi 7170 automatic analyzer. We recorded the data from the medical charts and gathered data from a nutrition survey.
In Taiwan, based on the policy and cost coverage of the National Health Insurance, all women are asked to take the glucose challenge test. Besides, appointment scheduling in outpatient services is very busy and includes morning, afternoon, and night clinics (Fig. 1). Unlike in other countries, where most patients attend outpatient clinics in the morning, it is almost impossible to ensure that pregnant women in Taiwan who attend afternoon and night outpatient clinics and undergo a glucose screening test fast for more than 2 hours (without the traditional 8 hours or midnight fasting). However, in our setting, it is clinically feasible to ensure that patients avoid food intake for 1 or 2 hours before glucose screening. For this reason, we further compared the effects of no food intake for at least 2 hours (fasting greater than 2 hours), no food intake for at least 1 hour but less than 2 hours, and food intake within 1 hour of the screening test (fasting less than 1 hour) on the screening outcome.
All of the pregnant women at Chia-Yi Christian Hospital who underwent a 50-g glucose challenge test at 24–28 weeks of gestation, except those with prepregnancy diabetes, were invited to participate in the study. Women with multifetal pregnancies, a history of GDM, or positive 50-g glucose challenge test results but negative 100-g OGTT results were excluded (Fig. 2). The included women underwent the two-step approach for GDM screening. The glucose challenge test was considered negative if the screening value was less than 140 mg/dL. If the glucose challenge test was positive, the women subsequently underwent a 100-g OGTT. No instructions (ie, regarding eating or avoiding eating) were given to the participants before glucose challenge testing. All of the 100-g OGTT screenings were arranged and performed in the morning outpatient clinic, and the patients fasted starting at midnight (ie, no food intake after midnight) or for at least 8 hours. The test was scheduled for 8:00–9:00 AM. The participants’ prior carbohydrate intake was not monitored.
We administered a nutrition survey to investigate the recent dietary history, including the time and content of food intake, of the pregnant women who underwent the glucose challenge test. The nursing staff of the Department of Community Health at Chia-Yi Christian Hospital administered the dietary recall questionnaire (Appendix 1, available online at http://links.lww.com/AOG/A353) in face-to-face interviews with the women. A dietitian calculated the total calories and carbohydrate calories of the foods recorded in the dietary recall questionnaire based on the Dietary Reference Intakes and the Recommended Dietary Allowances in Taiwan. Such methods as the Handy model and such measuring tools as a standard bowl, teaspoon, and standard cup were used to improve the accuracy of the record.
Gestational diabetes mellitus was diagnosed according to the National Diabetes Data Group criteria. Glucose levels were considered abnormal at or greater than 105, 190, 165, and 145 mg/dL for the fasting, 1-, 2-, and 3-hour plasma glucose tests, respectively.13 A woman with a negative glucose challenge test was assumed to have normal glucose tolerance. A woman with a positive glucose challenge test and two or more abnormal 100-g OGTT values was diagnosed with GDM. Women with impaired glucose tolerance had a positive glucose challenge test but did not meet the GDM criteria.
The study participants were stratified into three groups according to the time of last food ingestion (fasting interval): 1 hour or less, 1–2 hours, and more than 2 hours before the glucose challenge test. The greater than 2 hours fasting interval group was defined as the “fasting” group. The 1 hour or less and 1–2 hours fasting interval groups were combined to form the “fed” group. Body mass index was calculated as weight (kg)/[height (m)]2 and classified according to the Bureau of Health Promotion, Department of Health, Taiwan, as follows: less than 18.5, underweight; 18.5–24, normal; 24–27, overweight; and 27 or greater, obese. The women who had consumed food or liquid less than 2 hours before a glucose challenge test were further classified into five groups based on 200-kcaL increments for total calories and 100-kcaL increments for carbohydrate calories. We calculated the positivity rate and PPV to detect the predictive value. Positive predictive value was calculated as the number of GDM participants divided by the number of glucose challenge test-positive participants.
We evaluated the normality of the distributions of all continuous variables. The differences between the fasting interval groups were analyzed using the Kruskal-Wallis test for continuous variables because the data were not normally distributed. An analysis of variance was used for continuous variables with normal distributions and homogenous variances, and Welch's analysis of variance was used to test for differences between group variances. If the difference was significant, Dunn's post hoc multiple comparison test was performed for the Kruskal-Wallis test, and the Tukey-Kramer post hoc test was performed for the analysis of variance. The χ2 test was used for categorical variables. The association between the fasting interval and the predictive value of the two-step approach was analyzed using the χ2 test, and the partition χ2 test was used to find the significant cutoff.15 For the multivariate analysis, a multinomial logistic regression was used to determine the relationship between the fasting and fed groups after adjusting for maternal age, nulliparous status, time the glucose challenge test was received, prepregnancy body mass index, and percentage of gestational weight gain. Associations were described in terms of an adjusted odds ratio (OR) with a 95% confidence interval. A two-sided P value <.05 was considered statistically significant. All of the data were merged, and analyses were performed using SAS 9.2, except for Dunn's post hoc multiple comparison test, which was performed using GraphPad Prism 5.01.
Of 1,681 women at 24–28 weeks of gestation without prepregnancy diabetes, 91.0% (n=1,530) underwent a glucose challenge test and were recruited into the study. Thirty-three women who did not have information about their fasting time were excluded; however, 15 women who were missing height, weight, caloric intake, or all these data were included in our univariate analyses. An additional 110 women were excluded from the analysis for the following reasons: 29 had multifetal pregnancies, 14 had histories of GDM, and 67 had positive glucose challenge tests but did not undergo the 100-g OGTT. Thus, a total of 1,387 women were included in the analysis. Of these, 1,076 (77.6%) had negative glucose challenge tests, 251 (18.1%) had impaired glucose tolerance, and 60 (4.3%) had GDM (Fig. 2).
Except for nulliparous status and percentage of gestational weight gain, the characteristics of the study participants did not differ significantly among the fasting interval groups (Table 1). A significantly greater number of nulliparous women were included in the 1-hour or less fasting time group (58.1%, P=.003). Women with greater weight gain were more prevalent in the 1 or less and 1–2 hours fasting groups (4.0% and 3.7%, P=.005).
Among the women with impaired glucose tolerance, the median plasma glucose level following the 50-g glucose challenge test was significantly higher in the 1-hour or less fasting interval group than in the greater than 2 hours fasting interval group (median [range] 159.5 [142–212] compared with 153 [141–238], P=.04; Table 2). There were no significant differences in the glucose challenge test among the three fasting interval groups for women with normal glucose tolerance, GDM, or for all participants combined (Table 2; Appendix 2, available online at http://links.lww.com/AOG/A354).
The fasting intervals and the predictive value of the two-step approach were significantly associated (Table 3). As shown in Table 3A, the GDM rate was 6.9% in the greater than 2 hours fasting interval group and higher than the GDM rate in the 1 hour or less (2.5%) and 1–2 hours (3.1%) fasting interval groups (P=.01). The impaired glucose tolerance rate was similar in the three groups (17.7–18.5%). We further divided Table 3A into four tables, Tables B1 through B4, according to the rule of the partition χ2 (Appendix 3, available online at http://links.lww.com/AOG/A355).14 There was no significant difference among the women with 1 hour or less and 1–2 hours fasting intervals; in neither group could all participants be classified according to glucose tolerance as “glucose challenge test (−)” or “glucose challenge test (+)” (Table 3[B2]) or as “impaired glucose tolerance” and “GDM” in the glucose challenge test (+) group (Table 3[B4]). The significant cutoff was the fasting intervals 2 hours or less and greater than 2 hours (Table 3[B1] and [B3]); thus, the women in those groups were classified as “fed” and “fasting,” respectively. Table 3(B1) shows that the fasting group had a higher glucose challenge test positivity rate compared with the fed group (25.3% compared with 20.7%, P=.047), and Table 3(B3) shows that the fasting group had a higher PPV than the fed group (27.1% compared with 13.7%, P=.003). In summary, the fasting group (no food intake for at least 2 hours before screening) had an increased glucose challenge test positivity rate and PPV. The groups with 1 hour or less (food intake near the screening test) and 1–2 hours (no food intake within 1 hour before the screening test) fasting intervals were not significantly associated.
As Table 4 shows, the univariate analyses demonstrated that food intake state (fasting or fed) and baseline characteristics were significant predictors of impaired glucose tolerance and GDM, although nulliparous status and time the glucose challenge test was received were not. Furthermore, in the multivariate analysis adjusted for the baseline factors, the adjusted OR of diagnosis for GDM was significantly higher in the fasting group than in the fed group (adjusted OR 2.86, 95% CI 1.65–4.95) when compared with the normal glucose tolerance group. Food intake state was not a significant predictor of impaired glucose tolerance (adjusted OR 1.17, 95% CI 0.87–1.56).
Table 5 and Appendix 4 (available online at http://links.lww.com/AOG/A356) show that there were no significant differences among women with normal glucose tolerance, impaired glucose tolerance, and GDM in 200-kcaL increments in total calories (P=.95) or 100-kcaL increments in carbohydrate calories (P=.65) consumed less than 2 hours before the glucose challenge test.
A screening test should be convenient, safe, and inexpensive. To improve the predictive value of the screening test for GDM, some studies have recommended changing the threshold, whereas others have recommended changing the timing of blood collection after glucose is consumed during the glucose challenge test. We were concerned about the potential effect of eating before the test. Few studies have investigated this issue, and their results were discordant because of differences in design, diagnostic criteria, fasting time cutoff, and the participants’ race.9–12 The previous studies compared the mean glucose levels of normal and pregnant women with GDM who had fasted or eaten before the test. Our study combined the normal and GDM groups and used positivity rate and PPV to detect the predictive value of the screening test for GDM. In addition, we used multinomial logistic analysis to determine whether the food intake state before the glucose challenge test influenced the diagnosis of GDM, impaired glucose tolerance, or both. We also investigated the relationship between total and carbohydrate caloric intake and the results of the screening test for GDM in the fed group.
This study investigated a clinically important topic. The strengths of the study include its prospective (although not randomized) design, its large sample size, the use of a face-to-face dietary questionnaire, and the clinically relevant analysis. The main finding was that women who fasted for at least 2 hours before the glucose challenge test had a higher positivity rate for GDM and a higher PPV compared with women in the fed group, whereas the two groups were almost similar at baseline. After adjusting for baseline bias, food intake state was still the predictive factor of GDM diagnosis but not impaired glucose tolerance. The results did not differ significantly with respect to carbohydrate or total caloric intake within 2 hours before the glucose challenge test. The results of our study may have implications for clinical practice and guidelines. It may be clinically feasible and beneficial to ensure that pregnant women fast for 2 hours (ie, no food intake for 2 hours) before the glucose challenge test.
Two previous studies that used fasting cutoff times that differed from those used in the present study found no difference in glucose challenge test glucose levels between fasting and fed states in normal participants,9,10 but two other reports showed conflicting results. The study by Lewis et al11 reported significantly higher glucose levels in the fasting than in the fed state,11 whereas the study by Wu et al reported lower fasting glucose levels.12 In participants with GDM, two studies have shown significantly higher glucose responses in the fasting state,10,11 whereas the study by Wu et al did not.12 Our impaired glucose tolerance and GDM results are in agreement with those of Wu et al.12 Additionally, our normal glucose tolerance is concordant with the results of two U.S. studies (Table 2).9,10
Regarding the baseline characteristics of pregnant women, although nulliparity was more prevalent in the fasting 1-hour or less group, this difference did not influence the diagnosis of GDM in the study (Table 4; P=.33). The fact that women with greater weight gain (30% or greater) were more prevalent in the fasting 1 or less and 1−2 hours groups may have biased these groups toward higher pretest probabilities of GDM; however, the probability was actually significantly lower in the fed group than in the fasting group (2.6−3.1% compared with 6.9%). Although the influence of the participants’ baseline characteristics was minor, we nevertheless used multinomial logistic regression, which adjusted the baseline factors, to detect the relationships between food intake and the diagnosis of GDM and impaired glucose tolerance. The result was concordant with the univariate analysis shown in Table 3. Because some confounding factors are unknown or are not quantitative, randomized control trials are needed to determine whether randomly assigning women to one group or another improves diagnostic accuracy before a widespread change in approach is recommended.
Although we used a face-to-face dietary recall questionnaire and a single dietitian standardized the caloric intake calculation, the accuracy of all forms of dietary assessment has been criticized. Future studies should include a more accurate dietary log to determine the exact number of calories consumed needed. To address the possible problem of interobserver bias, more than two dietitians should also calculate the caloric intake.15,16
Our findings suggest that food intake influences the predictive value of the gestational diabetes screening test. The results of our study may have implications for clinical practice and guidelines.
1. Fadl HE, Ostlund IK, Magnuson AF, Hanson US. Maternal and neonatal outcomes and time trends of gestational diabetes mellitus in Sweden from 1991 to 2003. Diabet Med 2010;27:436–41.
2. Pedula KL, Hillier TA, Schmidt MM, Mullen JA, Charles MA, Pettitt DJ. Ethnic differences in gestational oral glucose screening in a large US population. Ethn Dis 2009;19:414–9.
3. Zhang F, Dong L, Zhang CP, et al.. Increasing prevalence of gestational diabetes mellitus in Chinese women from 1999 to 2008. Diabet Med 2011;28:652–7.
4. Ferrara A. Increasing prevalence of gestational diabetes mellitus: a public health perspective. Diabetes Care 2007;30:S141–6.
5. Basevi V, Di Mario S, Morciano C, Nonino F, Magrini N. Comment on: American Diabetes Association. Standards of medical care in diabetes—2011. Diabetes Care2011;34:S11–61.–
6. Screening and diagnosis of gestational diabetes mellitus. Committee Opinion No. 504. American College of Obstetricians and Gynecologists. Obstet Gynecol 2011;118:751–3.
7. HAPO Study Cooperative Research Group, Metzger BE, Lowe LP, Dyer AR, Trimble ER, Chaovarindr U, Coustan DR, et al.. Hyperglycemia and adverse pregnancy outcomes. N Engl J Med 2008;358:1991–2002.
8. Management of diabetes mellitus in pregnancy. ACOG Technical Bulletin No 92. American College of Obstetrics and Gynecology. Washington, DC: ACOG; 1986.
9. Berkus MD, Stern MP, Mitchell BD, Newton ER, Langer O. Does fasting interval affect the glucose challenge test? Am J Obstet Gynecol 1990;163:1812–7.
10. Coustan DR, Widness JA, Carpenter MW, Rotondo L, Pratt DC, Oh W. Should the fifty-gram, one-hour plasma glucose screening test for gestational diabetes be administered in the fasting or fed state? Am J Obstet Gynecol 1986;154:1031–5.
11. Lewis GF, McNally C, Blackman JD, Polonsky KS, Barron WM. Prior feeding alters the response to the 50-g glucose challenge test in pregnancy. The Staub-Traugott effect revisited. Diabetes Care 1993;16:1551–6.
12. Wu LF, Liu DY, Huang XH, Zu XS, Yang M, Liu WJ, et al.. Multicenter study on screen method for gestational diabetes [in Chinese]. Zhonghua Fu Chan Ke Za Zhi 2003;38:132–5.
13. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. National Diabetes Data Group. Diabetes 1979;28:1039–57.
14. Agresti A. Partitioning chi-squared. An introduction to categorical data analysis. 2nd ed Hoboken (NJ): John Wiley & Sons; 2007. p. 39–40.
15. Beaton GH, Burema J, Ritenbaugh C. Errors in the interpretation of dietary assessments. Am J Clin Nutr 1997;65:1100S–7S.
16. Kubena KS. Accuracy in dietary assessment: on the road to good science. J Am Diet Assoc 2000;100:775–6.
Supplemental Digital Content
© 2013 The American College of Obstetricians and Gynecologists