Nausea and vomiting are common symptoms during early pregnancy,1–4 affecting as many as 80% of pregnant women.2,5 The condition hyperemesis gravidarum represents an extreme form of these symptoms6 and affects approximately 0.3 to 2.0% of pregnancies.1,5–9 Although there is inconsistency in the definition of this condition in the literature,6 hyperemesis is most often characterized by severe nausea and vomiting that interferes with nutritional intake and metabolism, causes fluid and electrolyte imbalances,9,10 and commonly requires hospital management.9
A number of risk factors associated with hyperemesis have been reported, including nulliparity,11,12 low maternal age,1,4,5,9,12 multiple gestation,1,4,6,7,10 fetal anomalies,1,10,12 a previous pregnancy complicated by hyperemesis,1 female sex,7,9,13 psychiatric conditions,11,14 and both high1,4,12 and low13,15 maternal prepregnancy weight. Smoking, on the other hand, has been associated with a reduced risk of hyperemesis.4,5,12,13
Many of the previous studies of hyperemesis have been limited in their statistical power and generalizability as a result of being facility-based and relatively small in size. Additionally, few have used multivariate analysis to identify independent risk factors for hyperemesis. In this study, we explored potential risk factors for this severe form of pregnancy-related nausea and vomiting using a large, population-based cohort of pregnant women during a 15-year period.
MATERIALS AND METHODS
This study included all deliveries to residents of Nova Scotia, Canada between 1988 and 2002. Deliveries of infants less than 20 weeks of gestation or less than 500 g were excluded. Data for this study were obtained from the population-based Nova Scotia Atlee Perinatal Database, which has collected data on all deliveries in the province since 1988. Nova Scotia Atlee Perinatal Database data were abstracted by trained coders from clinical forms and hospital records and have been found to be of high quality in previous studies.16,17
The Nova Scotia Atlee Perinatal Database contains maternal information on lifestyle, illnesses, reproductive history, and other characteristics as well as information on labor, delivery, and infant outcomes. In addition to the data abstracted from the prenatal and delivery admission records, information pertaining to all antepartum admissions during the pregnancy is also recorded by the database.
Each unique pregnancy contributed 1 observation to the study, including multiple gestation pregnancies (ie, twins, triplets) and those complicated by recurrent antepartum admissions for hyperemesis. From our initial study sample, we first identified all pregnancies with 1 or more antepartum admissions with hyperemesis (n = 1,515). To ensure that the primary reason for the admission was hyperemesis, we limited our case definition to women whose first admission occurred before 24 weeks of completed gestation, because hyperemesis is most common and most severe in early pregnancy.13 This resulted in the exclusion of 182 women, with an additional 32 women excluded for missing gestational age. Thus, 1,301 women with hospital admission for hyperemesis and a comparison group of 156,621 women with no antepartum admissions for hyperemesis during their pregnancy were included in the analysis.
Initially, gestational age–specific incidence rates of hospital admission for hyperemesis were calculated for each week in pregnancy from 8 to 24 weeks gestation. The rates were calculated by determining the number of women with an admission divided by the number of pregnant women at each gestational week. We used the odds ratio (OR) to estimate the relative risk (RR) of hyperemesis for each factor of interest. Logistic regression employing the method of generalized estimating equations generated standard errors that accounted for the nonindependence of separate pregnancies to the same woman. When observations are not independent (eg, more than 1 delivery per woman), generalized estimating equations have the effect of increasing the variance estimates, which will widen the confidence intervals. Unadjusted relative risks and 95% confidence intervals (CIs) were initially calculated for each potential risk factor, followed by a backward stepwise approach to determine the most parsimonious adjusted model. All factors with a χ2P value of .20 or less from unadjusted analyses were entered into a multivariate model. We began by removing the risk factor with the highest P value and used the likelihood ratio test to compare the model with and without the factor. If the removal of the factor did not result in a statistically significant likelihood ratio test (P < .05), the factor was eliminated and the process repeated. We tested for a period effect by introducing interaction terms between time and smoking, maternal age, and a variable for gender or multiple pregnancy. The contribution of each interaction term was assessed using the likelihood ratio test. All analyses were conducted with SAS 8.2 software (SAS Institute Inc., Cary, NC). This study received approval from the IWK Health Centre Research Ethics Board before commencing.
The rate of antepartum admissions in which there was a diagnosis of hyperemesis was significantly higher in the earlier years of the study (1.0% between 1988 and 1994) than in the later years (0.6% between 1995 and 2002) (P < .001), with an overall admission rate of 0.8%. Among the 1,301 women with at least 1 antepartum admission for hyperemesis, the majority were first admitted in the first trimester (79.5%) and had only 1 admission (75.6%). The median length of stay for women with only 1 hyperemesis admission that occurred during the first trimester was 4 days, with a range of 1–35 days. The gestational age–specific incidence rate of antepartum admission for hyperemesis was highest in early pregnancy, ranging from 1.0 per 1,000 pregnant women at 8 weeks of gestation to 0.1 per 1,000 pregnant women by the 20th week of gestation (Fig. 1).
In the univariate analysis, nulliparity, maternal age less than 20 years, and multiple gestation were all risk factors for hyperemesis admission (Table 1). Women with a previous molar pregnancy, preexisting diabetes, depression or psychiatric illness, hyperthyroid disorder, ulcers, gastrointestinal disorders, respiratory disease, or asthma also had a higher risk of admission for hyperemesis than women without these conditions. Smoking was associated with a reduction in risk. Compared with pregnancies with a singleton male, pregnancies with a singleton female were at higher risk for a hyperemesis admission (unadjusted RR 1.3, 95% CI 1.1–1.4). However, all gender combinations among multiple gestation pregnancies had higher point estimates, particularly those involving all male fetuses (unadjusted RR 2.5, 95% CI 1.4–4.6) and those with a mix of male and female fetuses (unadjusted RR 3.0, 95% CI 1.7–5.4). Admission for hyperemesis in the pregnancy preceding the index pregnancy or in any previous pregnancy greatly increased the risk of admission for hyperemesis. Among a subset of women known to be nulliparous at their first recorded delivery in the database and who went on to have at least 1 more delivery, the risk of admission for hyperemesis was 29 times higher (95% CI 22.4–36.8) if the previous pregnancy also had an antepartum admission for hyperemesis. Because the analysis of this factor (previous pregnancy with an admission for hyperemesis) is based on a subset restricted to those with at least 1 previous delivery and whose previous pregnancies were all recorded in the database, this factor was not included in the multivariate analysis.
Following the backward stepwise modeling procedure, 12 variables remained in the final adjusted model (Table 2). The risk of antepartum hospital admission for hyperemesis was 70% higher during1988–1994 compared with 1995–2002 (adjusted RR 1.7, 95% CI 1.5–1.9). Among the maternal conditions studied, preexisting diabetes, psychiatric illness, hyperthyroid disorder, gastrointestinal disorder, and asthma all remained in the final model and were statistically significantly associated with an increased risk of hyperemesis admission. Women who smoked during pregnancy were less likely to be admitted for hyperemesis than nonsmoking women (adjusted RR 0.5, 95% CI 0.5–0.6), and women older than 30 years at delivery were less likely to be admitted for hyperemesis than younger women. Similar to the unadjusted results, multiple gestation pregnancies involving all male fetuses and those with a mix of male and female fetuses had a higher risk of antepartum admission for hyperemesis compared with pregnancies with a singleton male. A history of molar pregnancy was also a significant risk factor for hyperemesis admission (adjusted RR 3.3, 95% CI 1.6–6.8). The addition of interaction terms did not significantly contribute to the adjusted model; therefore none were retained.
To assess the sex ratio according to antepartum hospital admission for hyperemesis, an analysis was conducted on singletons only. The fetal sex ratio among women who did not have an antepartum admission for hyperemesis was 51.2% males and 48.8% females. Among women with 1–2 admissions for hyperemesis, the proportions changed to 44.4% males and 55.6% females and for women with 3 or more admissions, the fetal sex distribution was 37.5% males compared with 62.5% females.
We observed an overall admission rate for hyperemesis of 0.8%; however, there was a trend toward lower rates in the more recent years of the study. To evaluate the possible impact of a period effect, we tested several interaction terms and did not see any differences between earlier and later periods in the relationship between hyperemesis and smoking, maternal age or gender or multiple pregnancy. Decreasing antenatal admission rates have been noted both generally,18,19 and specifically20 for nausea and vomiting. Although it is possible that this is a reflection of reduced morbidity from hyperemesis possibly resulting from increasing medication use for nausea and vomiting of pregnancy20,21 and underlying changes in maternal characteristics, such as the decreasing number of deliveries to younger women,22 it is also very likely being influenced by changes in hospital admission policies and a greater tendency toward outpatient care in recent years.18,19 Time trend notwithstanding, the overall hyperemesis rate observed in this study approximates rates reported in the literature.1,5,6–9
Consistent with other studies, we found that hyperemesis was more common in young women1,5,9,12 and those with psychiatric illness14 and was less common among women who smoked during their pregnancy.5,12,13 Although both low13,15 and high1,12 maternal prepregnancy weight have been previously identified as risk factors for hyperemesis, this factor did not stay in our adjusted model, and we did not observe a consistent relationship in the univariate analysis. We found a slightly increased risk of hyperemesis among nulliparous women in unadjusted analyses; however, this was no longer evident after adjustment for maternal age (results not shown). A number of other studies have noted an increased risk of hyperemesis among nulliparous women,1,11,12 even after adjustment for maternal age.12 Nulliparous women and women pregnant for the first time have higher levels of estrogen than other women, which may predispose them to hyperemesis.12
It has been well established that pregnancies complicated by hyperemesis have altered sex ratios in favor of female offspring7,13 and maternal hormonal levels associated with female offspring likely play a role in the occurrence of hyperemesis.6,9,12,23,24 Although we observed a higher risk of admission for hyperemesis among singleton gestations with a female fetus compared with a male fetus, we found that the effect was larger among multiple gestations, particularly among those with all–male fetuses or a mix of male and female fetuses. That the relative risks were larger for all gender combinations among multiple gestations was not surprising, because this is in itself a risk factor for hyperemesis.1,4,6,7,10 However, the finding that multiple gestations with all–female fetuses had the lowest point estimate among multiple gestation pregnancies was unexpected. We are aware of only 1 other study that has examined the effect of gender in this manner. Basso et al7 looked at the combined effect of gender and twinning on hyperemesis among pregnancies of more than 28 weeks of completed gestation. They found increased risks for all gender combinations among twin pregnancies compared with those with a singleton male; however, odds ratios for twin pregnancies with 1 or 2 females were 41% and 35% higher, respectively, than pregnancies with 2 male offspring.7 Our results were unchanged when we restricted the cohort to more than 28 weeks of completed gestation. Given that multiple gestation pregnancies are independently associated with hyperemesis1,4,6,7,10 and are characterized by different maternal hormone levels than singleton pregnancies,7,25 the expected gender association with hyperemesis may be obliterated in the context of a multiple pregnancy.
In addition to fetal sex, several of the risk factors that we identified are also thought to be associated with hormonal changes in pregnancy. Asthma in pregnancy may be influenced by changes in hormonal levels,26 as seen by an apparent association with fetal sex.26,27 An association between molar pregnancy and hyperemesis in a current pregnancy has been reported.1,6,10,23 If the direction of the association between hCG and hyperemesis is one of higher hCG levels preceding the development of hyperemesis, as has been proposed by Goodwin,23 then the higher risk for hyperemesis that we observed among women with a past history of a molar pregnancy may be identifying women who are predisposed to abnormal placental hormone levels during pregnancy. It is also possible that this association is the result of a type 1 error, because many factors were tested. Hyperthyroidism has been associated with hyperemesis,28 although the exact mechanism is not known. Some thyroid conditions, eg, transient hyperthyroidism, are associated with higher than normal hCG levels, a characteristic of hyperemesis.29
Strengths of this study included the availability of a large, population–based cohort with data that were unaffected by recall bias. By restricting our case definition to hyperemesis requiring hospital admission, it is likely that we identified the subjects that would meet the diagnostic criteria for hyperemesis as opposed to milder forms of nausea and vomiting of pregnancy.
Limitations of this study included temporal trends (women in the earlier years with hyperemesis were more likely to be admitted), absence of consistent diagnostic criteria for hyperemesis in the database, lack of information on drug use and treatment in hospital, and lack of information on terminations or losses before 20 weeks of completed gestation. It is possible that there were subjects in the comparison group who had hyperemesis but were not admitted to hospital, especially in later years; thus, we may have underestimated the true burden of hyperemesis as well as biasing the relative risks toward the null value.
Using a large, population–based cohort, this study has found some of the expected associations with severe hyperemesis gravidarum such as maternal age, smoking, and psychiatric illness. It also identified several potential risk factors that merit future study. In particular, maternal preexisting diabetes, asthma, gastrointestinal disorders, and previous molar pregnancy, although not modifiable, could potentially assist in identifying women at high risk for developing hyperemesis during their pregnancies if they are established as risk factors. Finally, although the relationship between female fetal gender and hyperemesis can be considered established for singleton pregnancies, this study does not support prevailing theories for the relationship in the context of a multiple pregnancy.
1. Eliakim R, Abulafia O, Sherer DM. Hyperemesis gravidarum: a current review. Am J Perinatol 2000;17:207–18.
2. Gadsby R, Barnie–Adshead AM, Jagger C. A prospective study of nausea and vomiting during pregnancy. Br J Gen Pract 1993;43:245–8.
3. Lacroix R, Eason E, Melzack R. Nausea and vomiting during pregnancy: a prospective study of its frequency, intensity and patterns of change. Am J Obstet Gynecol 2000;182:931–7.
4. Klebanoff MA, Koslowe PA, Kaslow R, Rhoads GG. Epidemiology of vomiting in early pregnancy. Obstet Gynecol 1985;66:612–6.
5. Kallen B, Lundberg G, Aberg A. Relationship between vitamin use, smoking, and nausea and vomiting of pregnancy. Acta Obstet Gynecol Scand 2003;82:916–20.
6. Goodwin TM. Hyperemesis gravidarum. Clin Obstet Gynecol 1998;41:597–605.
7. Basso O, Olsen J. Sex ratio and twinning in women with hyperemesis or pre–eclampsia. Epidemiology 2001;12:747–9.
8. Tsang IS, Katz VL, Wells SD. Maternal and fetal outcomes in hyperemesis gravidarum. Int J Gynaecol Obstet 1996;55:231–5.
9. Bashiri A, Neumann L, Maymon E, Katz M. Hyperemesis gravidarum: epidemiologic features, complications and outcome. Eur J Obstet Gynecol Reprod Biol 1995;63:135–8.
10. Hod M, Orvieta R, KaplanB, Friedman S, Oviada J. Hyperemesis gravidarum: a review. J Reprod Med 1994;39:605–12.
11. Atanackovic G, Wolpin J, Koren G. Determinants of the need for hospital care among women with nausea and vomiting of pregnancy. Clin Invest Med 2001;24:90–3.
12. Depue RH, Bernstein L, Ross RK, Judd HL, Henderson BE. Hyperemesis gravidarum in relation to estradiol levels, pregnancy outcome, and other maternal factors: a seroepidemiologic study. Am J Obstet Gynecol 1987;156:1137–41.
13. Schiff MA, Reed SD, Daling JR. The sex ratio of pregnancies complicated by hospitalisation for hyperemesis gravidarum. BJOG 2004;111:27–30.
14. Swallow BL, Lindow SW, Masson EA, Hay DM. Psychological health in early pregnancy: relationship with nausea and vomiting. J Obstet Gynaecol 2004;24:28–32.
15. Rochelson B, Vohra N, Darvishzadeh J, Pagano M. Low prepregnancy ideal weight:height ratio in women with hyperemesis gravidarum. J Reprod Med 2003;48:422–4.
16. Fair M, Cyr M, Allen AC, Wen SW, Guyon G, MacDonald RC. An assessment of the validity of a computer system for probabilistic record linkage of birth and infant death records in Canada. The Fetal and Infant Health Study Group. Chronic Dis Can 2000;21:8–13.
17. Fair M, Cyr M, Allen AC, Wen SW, Guyon G, MacDonald RC. Validation study for a record linkage of births and infant deaths in Canada. Catalogue No. 84F0013XIE. Ottawa (Canada): Statistics Canada; 1999.
18. Bacak SJ, Callaghan WM, Dietz PM, Crouse C. Pregnancy–associated hospitalizations in the United States, 1999–2000. Am J Obstet Gynecol 2005;192:592–7.
19. Bennett TA, Kotelchuck M, Cox CE, Tucker MJ, Nadeau DA. Pregnancy–associated hospitalizations in the United States in 1991 and 1992: a comprehensive view of maternal morbidity. Am J Obstet Gynecol 1998;178:346–54.
20. Neutel IC. Variation in rates of hospitalization for excessive vomiting in pregnancy by Benedictin/Diclectin use in Canada. In: Koren G, Bishai R, editors. Nausea and vomiting of pregnancy: state of the art 2000. Toronto (Canada): Motherisk, Hospital for Sick Children; 2000. p. 54–9.
21. Lee J, Einarson A, Gallo M, Okotore B, Koren G. Longitudinal change in the treatment of nausea and vomiting of pregnancy in Ontario. Can J Clin Pharmacol 2000;7:205–8.
22. Fell DB, Joseph KS, Dodds L, Allen AC, Jangaard K, Van den Hof M. Changes in maternal characteristics in Nova Scotia, Canada from 1988 to 2001. Can J Public Health 2005;96:234–8.
23. Goodwin TM, Hershman JM, Cole L. Increased concentration of the free beta–subunit of human chorionic gonadotropin in hyperemesis gravidarum. Acta Obstet Gynecol Scand 1994;73:770–2.
24. Goodwin TM, Montoro M, Mestman JH. Transient hyperthyroidism and hyperemesis gravidarum: clinical aspects. Am J Obstet Gynecol 1992;167:648–52.
25. Mashiach R, Orr–Urtreger A, Yaron Y. A comparison between maternal serum free beta–human chorionic gonadotropin and pregnancy–associated plasma protein A levels in first–trimester twin and singleton pregnancies. Fetal Diagn Ther 2004;19:174–7.
26. Tan KS, Thomson NC. Asthma in pregnancy. Am J Med 2000;109:727–33.
27. Gardner MO, Doyle NM. Asthma in pregnancy. Obstet Gynecol Clin North Am 2004;31:385–413.
28. Vohra S, Koren G. Hypothetical framework for a relationship between maternal thyroid function, nausea and vomiting of pregnancy, and congenital heart disease. Med Hypotheses 2001;56:392–4.
29. Tan JY, Loh KC, Yeo GS, Chee YC. Transient hyperthyroidism of hyperemesis gravidarum. BJOG 2002;109:683–8.