The Fidgety Fetus Hypothesis: Fetal activity is an additional variable in determining birth weight of offspring of women with diabetes

Subjects were recruited through the Santa Barbara County Health Services Diabetes Clinic after being diagnosed with GDM using the criteria from the Fourth International Diabetes Workshop-Conference (8) between 24 and 28 weeks of gestation. Insulin therapy was initiated if needed based on the workshop guidelines. Prepregnancy maternal weights were not included in the analysis because the subjects’ medical records started at time of GDM diagnosis. All subjects signed informed consents approved by the Cottage Hospital Institutional Review Board. In phase I, 10 women with diabetes (mean age 32 ± 4 years, HbA_1c [A1C] 5.5 ± 0.6%, normotensive, nonsmokers, 7 with GDM, 2 with type 1 diabetes, and 1 with type 2 diabetes) were asked to perform fetal kick counts twice a day for 30 min and self-monitor their blood glucose before and 1 h after each meal. Subjects were instructed to schedule their monitoring sessions at the same time each day. The women were trained to detect gross fetal movements using fetal ultrasound for confirmation. The patients were asked to keep a diary documenting fetal movement and glucose values.

To improve our fetal monitoring techniques, in phase II, 13 women with GDM (mean age 32 ± 7 years, A1C 5.0 ± 0.4%, 8 treated on insulin and diet, and 5 controlled on diet alone) were recruited to test the feasibility of using a home fetal monitoring device. During session 1, the subjects were trained in the use of the Card Guard fetal monitor (CG-900P; Card Guard, Rehovot, Israel), which captures FHR, uterine contractions, and maternal perception of fetal movements. Previous studies report that the Card Guard fetal monitor has been easily and reliably operated by patients (9). Subjects performed a 20-min monitoring session daily at home for 4–6 weeks. Subjects were instructed to perform monitoring sessions at the same time everyday, between breakfast and lunch. The data were transmitted via modem and reviewed by a physician daily. FHR accelerations, defined as an increase in baseline fetal heart rate by at least 15 bpm sustained for at least 15 s, were employed as a surrogate marker to quantitate fetal activity since consistant linkage has been demonstrated between fetal body movements and FHR patterns (10,11). We chose to use a cutoff of FHR <4 to define the inactive group based on the collected data from phase II. There was an obvious separation of the inactive and active groups at this level. This definition was applied to both phases II and III (Table 1). An active fetus was defined as having ≥4 FHR accelerations in a 20-min period, and an inactive fetus was defined as having <4 FHR accelerations in a 20-min period. All subjects adhered to a program of diet and self blood glucose monitoring (four or more times per day: fasting and 1 h postprandial). Insulin was initiated, and frequency of blood glucose monitoring increased to seven per day (fasting and preprandial and 1 h postprandial) if the fasting blood glucose was >90 mg/dl and 1-h postprandial glucose concentrations were >120 mg/dl.

In phase III, we applied our definitions of an active and inactive fetus to a larger population of 46 women with GDM (mean age 32 ± 6 years, A1C 5.4 ± 0.5%) to examine the relationship between FHR accelerations and subsequent birth weight percentile, corrected for gestational age. We conducted a retrospective analysis of serial 20-min fetal monitoring strips (Hewlett-Packard HP 8041A) from women who visited the Santa Barbara County Health Services Diabetes Clinic between 9 and 11 am. Each woman’s total monitoring strips were then analyzed for FHR accelerations per 20 min. Characteristics of the patient population and infant outcome are listed in Table 2.

In all three phases of the study, there were no significant differences in the week that GDM was diagnosed, the week that insulin therapy was initiated, or the A1C levels between the inactive and active groups. In phase I, 71% of the recorded fetal movements and blood glucose records from the patients’ diaries contained adequate assessment. Evaluation of the women’s diaries showed little variability in fetal movements over a mean of 8.5 weeks (i.e., fetal kicks did not significantly deviate from one another on a day-to-day basis). There was no relationship between maternal weight and subsequent birth weight or the woman’s ability to perceive fetal movements. Maternal postprandial glucose concentration ranged from 83 to 111 mg/dl with a mean of 102. Mean maternal postprandial glucose levels positively correlated with increased birth weight (Fig. 1; r = 0.704). The ratio of 10 kicks per 1 mg/dl postprandial glucose value was calculated from the best correlation with birth weight. When the mean number of kicks per hour per day was divided by a factor of 10 and subtracted from the mean maternal 1-h postprandial glucose concentration, there was an improvement in the correlation between postprandial glucose and birth weight (Fig. 2; r = 0.773). The relationship between the neonatal birth weight and the “adjusted maternal postprandial glucose level” continued to improve as the number of fetal movements per glucose concentration increased.

Placental weights and umbilical cord lengths were collected from seven of the participants. The infant that weighed the most also had the heaviest placenta and the longest cord length. There was a significant positive correlation between placental weight and birth weight (r = 0.798, P = 0.032) and placental weight and cord length (r = 0.912, P = 0.004).

In phase II, the fetal monitoring strips separated into two distinct groups determined to be “active” or “inactive,” consistently illustrating that the active fetuses were always active and the inactive fetuses were always inactive. Mean birth weight percentile in the active group was 37 vs. 63% in the inactive group (P = 0.05). This significant difference in birth weight percentile remained despite the fact that the mean maternal A1C levels were not significantly different in the two groups (5.1 vs. 4.9%, respectively, P = 0.47).

In phase III, we observed an inverse correlation between the mean number of FHR accelerations and the birth weight percentile (Fig. 3). We found that the active fetus was more likely to be associated with a birth weight <50th percentile. In contrast, the inactive fetus was more likely to be associated with a birth weight >50th percentile. In addition, macrosomic fetuses were only observed in the inactive group. The mean birth weight percentile in the active group was 37 vs. 62% in the inactive group (P = 0.0017). This difference remained despite the fact that the mean maternal A1C levels were not significantly different in the two groups (5.3 vs. 5.4%, respectively, P = 0.92) (Table 2).

The Fidgety Fetus Hypothesis suggests that fetal activity is a determinant of birth weight. Previous reports have examined the “fidget gene” in lean and obese adults and suggest that individual differences in activity level are biologically determined (12). Based on this notion, the Fidgety Fetus Hypothesis states that if physical activity is an intrinsic characteristic, then it should be expected to manifest itself in utero. If the fetus is overnourished by either intermittent or sustained maternal hyperglycemia, then those fetuses that are intrinsically active can minimize their risk of becoming macrosomic. Alternatively, those fetuses that are less active would be more affected by the hyperglycemia and thus develop an increased likelihood of becoming macrosomic.

The pediatric literature (6,13–15) supports the Fidgety Fetus Hypothesis, lending insight into the fact that only 42% of the neonates born to women with diabetes are macrosomic, while the remaining 58% are not. The Framingham Children’s Study found that low levels of activity during the preschool years had a moderately strong effect on increasing the child’s level of adiposity during later childhood. The Bogalusa Heart Study and a literature review suggest that childhood obesity is related to both overnutrition and lack of physical activity, and the degree of physical activity seems to be intrinsic to each. The established factors that contribute to birth weight of the fetus are maternal age, race, genetics, and maternal/paternal physiognomy. Gestational diabetes, maternal obesity, parity, postdatism, and previous birth weights contribute to the development of macrosomia (16–19). We propose that fetal activity is an additional variable in determining birth weight.

The obstetric literature also supports the fidgety fetus hypothesis, claiming that fetal motor activity is an important factor in predicting neonatal motor activity (20,21). DiPietro et al. (20) discovered that fetal motor activity shows evidence of stability before birth and predicts aspects of temperament associated with regulatory control in early childhood. In addition, fetal motor activity was consistent with neonatal motor activity at 2 weeks, 4 weeks, and at 1 and 2 years. Thus, individual differences in motor activity level in the 1st month following birth were probably present during fetal life.

Froen (22) reinforced the importance of fetal movement counting as a method of screening for fetal well-being. His study determined that reduced fetal movements are associated with an adverse pregnancy outcome in both high- and low-risk pregnancies. He concluded that screening for fetal well-being by maternal fetal movement counting can reduce fetal mortality rates. Additionally, Pearson et al. (23) determined that 12-h fetal movement counts <10 (the lower limit of normal) were associated with fetal demise.

Previous short-term clamp studies (24–27) analyzing glucose extremes have shown a negative correlation between fetal movements and maternal glycemia. Fetal movements significantly increased with low maternal glycemia in pregnancies complicated by diabetes. Although maternal hyperglycemia was associated with low fetal movement counts, this may have been associated with a sick fetus.

Our three-phased study has shown that intrinsic fetal activity minimizes the impact of maternal hyperglycemia and puts the fetus at a decreased risk of becoming macrosomic. In phase I, although we confirmed that both placental weight and cord length relate positively to birth weight, our new observation was that the more active the fetus was, in spite of postprandial hyperglycemia, the less likely he or she was large for gestational age. In phase II, this observation was confirmed by home fetal telemetric monitoring. In addition, phase III of our study confirmed that glucose-mediated macrosomia was only present in the inactive fetus group.

In summary, the fetus appears to play a role in determining its own destiny. Increased fetal activity, as a response to increased maternal glucose or as an intrinsic trait, can negate the impact of maternal hyperglycemia on birth weight. The inactive fetus appears to be at a higher risk for glucose-mediated macrosomia. Although normoglycemia is the goal of therapy in all women with diabetes, in those women whose fetuses manifest less movement, increased surveillance and treatment of hyperglycemia is of utmost importance. Perhaps an indicator of fetal movement could be combined with indicators of maternal glycemia to improve the ability of predicting birth weight in pregnancies complicated by diabetes.