Determinants of Hyperemesis Gravidarum Among Pregnant Women in Public Hospitals of Mekelle City, North Ethiopia, 2019: Unmatched Case-Control Study

Background: Globally, hyperemesis gravidarum affects 0.3 to 3.6% of pregnant women. The etiology of hyperemesis gravidarum is unclear. In Ethiopia, limited studies have been conducted on the determinants of hyperemesis gravidarum. Therefore this study is aim to identify the determinants of hyperemesis gravidarum among pregnant women in public hospitals of Mekelle city, North Ethiopia. Methods: An unmatched case-control study was conducted from May to October 2019, with the ratio of 1:2(109 cases and 218 controls). Cases were women with hyperemesis gravidarum and controls were women who had no hyperemesis gravidarum. Cases were enrolled using consecutive sampling techniques and controls were selected by a systematic random sampling technique. Data were collected by interview using a structured questionnaire and analyzed using SPSS version 22. Statistical signicance was considered at p- 0.05, and the strength of association was assessed by odds ratio and 95% condence intervals.

by interview using a structured questionnaire and analyzed using SPSS version 22. Statistical signi cance was considered at p-0.05, and the strength of association was assessed by odds ratio and 95% con dence intervals. Conclusion and Recommendations: In this study; being a housewife, unplanned pregnancy, had a family history of hyperemesis gravidarum, H. pylori infection, high perceived stress, and being in the 1 st and 2 nd trimester of pregnancy were found to be the determinants of hyperemesis gravidarum. We recommend stress should be minimized through psychological support during follow up of pregnancy. Screening for H. Pylori should be taken as routine investigations for pregnant women who complain of nausea and vomiting.

Background
Hyperemesis gravidarum (HG) is a severe type of vomiting of pregnancy which has got deleterious effect on the health of the mother and/or incapacitates her in day-to-day activities (1). Its adverse effect includes dehydration, weight loss and starvation ketosis proved by positive ketone in urine at least once during pregnancy (1). This may result in uid and electrolyte imbalance and affecting the nutritional status and is potentially lethal if left untreated (2,3). Hyperemesis gravidarum was considered for a long time to be a disorder caused by psychological distress, fear of childbirth, or resentment of the pregnant state (4,5). Globally, the prevalence of HG was found to be between 0.3 and 3.6% with an average of 1.1% (6). It's prevalence in Ethiopia also ranges from 4.4% to 8.2% (7)(8)(9).
Hyperemesis gravidarum is one of the health problems during pregnancy and it affects women's daily lives and was found to have adverse effects on daily life functioning. Clinical complications from nutritional de ciencies, psychological disorder, gastrointestinal trauma, and neurologic damage are the most common morbidities seen in women with HG (10,11). It also contributes to the termination of wanted pregnancies that affects the couple family planning. Women with HG often require hospitalization, as a result, they lose time to both paid employment and housework with a signi cant economic burden (9,10,12). Hyperemesis gravidarum also increases the risk for adverse pregnancy outcomes such as low birth weight, preterm birth, small-for-gestational-age infants, and low APGAR score (13). It also hurts women's ability to care for their children, affects their relationship with partner and work capacity for most of the women (10).
The cause of HG is unclear, but it is widely agreed that it is a multifactorial condition with sociodemographic, obstetrics, medical, and physiological contributing determinants (12,14,15). However, a con ict between studies was found. Limited studies have been conducted to identify the determinants of hyperemesis gravidarum in Ethiopia moreover; no study is done in the Tigray region. Therefore, this study aimed to identify determinants of HG, which could reduce the adverse perinatal outcome, hospitalization; time lost from paid employment, physical, psychological, economic, and social burdens of HG on women's lives, through prevention and early initiation of intervention

Study area and period
The study was conducted in public hospitals of Mekelle city namely Ayder, Mekelle, and Quiha hospital from May 1 to October 30, 2019. Mekelle is the capital city of Tigray regional state located 780 km from Addis Ababa the capital city of Ethiopia. These public hospitals are serving a population of 9 million from Tigray, Afar, and southeast Amhara. According to 2018, the hospital's health management information system reports; 5322 pregnant women had their rst antenatal care follow up.

Study design
A facility-based unmatched case-control study design was employed.

Population
Source population: The source population was all pregnant women who visited the maternity center of the study hospitals during the study period.

Study population
All pregnant women who were admitted to the hospitals with the diagnosis of hyperemesis gravidarum during the study period were study population for cases, whereas all pregnant women who visited the hospitals for antenatal care service during the study period were study population for controls.

Inclusion and exclusion Criteria
Inclusion Criteria Pregnant women who were hospitalized due to HG diagnosed by the clinicians were included as cases and pregnant women who were visiting antenatal care services at the same time but not have symptoms of HG were included as controls.

Exclusion Criteria
Pregnant women with severe nausea and vomiting, but diagnosed to have a concurrent medical and surgical illness (pyelonephritis, dyspepsia, and acute abdomen) were excluded.

Sample size determination and sampling technique
Sample size determination The sample size was calculated with the assumption of a double population proportion formula for an unmatched case-control study using Epi Info software version 7.0. The following parameters were considered: 95% con dence interval, power was 80%; control to case ratio was 2:1 and different sample sizes were produced from previously identi ed determinants of HG and the maximum and manageable sample size was obtained by taking perceived stress as a determinant factor from a previous study done in south Ethiopia; where the proportion of exposure among cases to be 17.7% and among controls 6.1% with an odds ratio of 3.3 (16). This yields a maximum sample size of 296 (99 cases and 198 controls). By adding 10% non-response rate, the nal sample size required for the study was 327 (109 cases and 218 controls).

Sampling technique
Cases were selected by a consecutive sampling technique with daily monitoring of all new admissions until the sample size was ful lled. For each case, two controls were selected and this procedure was continued until the required sample size was attained. A systematic sampling technique was employed to select controls. Data collection procedure Data were collected through face-to-face interviews and chart reviews using structured and pretested questionnaires adapted from existing literature (14)(15)(16). The questionnaire includes four sections; sociodemographic, obstetric and gynecologic characteristics, medical and psychological characteristics, and questions that measured perceived stress status. The data were collected by ve BSc midwives using the Tigrigna version ques tionnaire.

Data quality control
The questionnaire was prepared originally in English and was translated to Tigrigna (local language) and back to English by two independent persons to keep the consistency of the questionnaires. Data collectors and supervisors were trained. Close supervision was done by the principal investigator and supervisor during data collection. Completeness of the data was checked at the eld level.

Data analysis
The collected data were checked for completeness and entered into Epi-data version 3.1. Then the data were exported to SPSS version 22 and cleaned, coded, and collapsed before subsequent analysis. Summary statistics such as median, mean, interquartile range, and standard deviation were computed for cases and control groups. The independent variables were cross-tabulated among cases and controls. Then bivariate analysis was carried out to assess the crude association between the independent and outcome variable. The variables with p-value <0.05 in bivariate analysis were entered into a multivariable logistic regression to assess the net effect by controlling confounders. The variables with p < 0.05 in multivariable logistic regressions were considered as statistically signi cant determinants for HG. The adjusted odds ratio (AOR) with 95% con dence interval (CI) was used to assess the strength of association. As a result, the model used was good as indicated by the Omnibus model goodness of t test (p 0.001 with the cut point <0.05) and Hosmer & Lemeshow model goodness of t test (p=0.618 with cut point >0.05). Multicollinearity test was done and all determinants had a variance in ation factor less than 10 indicating that there was no high correlation between the independent variable.

Operational de nition
Cases: Pregnant women who were hospitalized due to hyperemesis gravidarum diagnosed by the clinicians based on the clinical and laboratory nding.
Controls: Pregnant women who visited antenatal care services in the study facilities, but did not have symptoms of hyperemesis gravidarum.
Perceived stress: It was measured with the perceived stress scale (PSS). PSS is a 10-item multiplechoice self-report psychological instrument for measuring the perception of stress. Each answer was scored 0 (never) to 4 (very often). Individual scores on the PSS can range from 0 to 40 with higher scores indicating higher perceived stress. Scores ranging from 0-13 would be considered low stress.
Scores ranging from 14-26 would be considered moderate stress. Scores ranging from 27-40 would be considered high perceived stress (17).

Ethical consideration
All the methods carried under this study were approved by the Ethics and Research Committee of Mekelle University (ERC1444/2018). The anonymity and con dentiality of the women were maintained in accordance with current legislation. Written informed consent was obtained from all participants. A letter of cooperation was obtained from Tigray regional health bureau. Information on the study was explained to the participants, including the objective, procedure, and bene t of the study. The respondents were informed that they have the right to refuse or decline participation in the study at any time.

Maternal sociodemographic characteristics
In this study, a total of 105 cases and 210 controls were successfully interviewed with a response rate of 96.3%. The same proportion of cases (88.6%) and controls (88.6%) were in the age group of 20-34 years. The median age of case and control was 27 (IQR =5) and 28 (IQR = 6) respectively. The majority of cases (86.7%) and controls (93.8%) were living in urban. The majority (81.9%) of cases and (79.0%) of controls were Orthodox. Eighty-six percent of cases and 95.2% of controls were married. About forty-six percent of cases and 25.2% of controls were housewives. (Table 1)   Medical and psychological characteristics of respondents Concerning medical characteristics, the proportion of history of diabetic mellitus among cases was 1.9% and that of controls was 1.0%. Twenty percent of the cases and 9% of controls were reporting a history of HG in their mothers and sisters. About forty-six percent of cases and 21.4% of controls were seropositive for Helicobacter pylori (H. pylori) infection. Pre-pregnancy motion sickness was reported by 13.3% of cases and 10% of controls. According to the perceived stress scale, it was observed that 19% of cases and 9% of controls had high perceived stress (Table 3).

Determinants of hyperemesis gravidarum
The bivariate analyses revealed that marital status, residence, maternal occupation, gestational age, gravidity, parity, unplanned pregnancy, had a family history of HG, Helicobacter pylori infection, and perceived stress were found to have an association with the development of hyperemesis gravidarum.
After controlling for potential confounds on multiple logistic regression analysis; being a housewife, unplanned pregnancy, being in the rst and second trimester of pregnancy, family history of HG, H. pylori infection and perceived stress were identi ed as determinants of hyperemesis gravidarum among pregnant women.

Discussion
The current study investigates the determinates of hyperemesis gravidarum among pregnant women in public hospitals of Mekelle City and the nding shows that being a housewife was signi cantly associated with the development of hyperemesis gravidarum as compared to employed women. This nding is supported by a study conducted in the Bale zone, southern Ethiopia, and Netherland (15,18).
This study also shows that unplanned pregnancy was signi cantly associated with hyperemesis gravidarum as compared to a planned pregnancy. The possible explanation for the observed association could be that pregnant women may perceive the negative effect of unplanned pregnancy on their job, education, income, relationship with their partner and motivation to achieve their goal, which may expose them to anxiety and tension, which could lead them to be at the more severe end of nausea and vomiting.
However, this nding contradicts a study done in southern Ethiopia in which no association was seen between HG and unplanned pregnancy (15). The reason may be due to the difference in the study area and sample size. However, the nding is consistent with a study conducted in Turkey (19).
A family history of hyperemesis gravidarum was also found to have a signi cant association with HG. This nding contradicts a study conducted in Bale zone, south Ethiopia in which no association was seen between HG and family history of hyperemesis gravidarum (15). The possible reason for the difference could be that the family history of hyperemesis gravidarum was based on self-report, which may lead to misclassi cation. However, this nding is consistent with a study conducted in Norway which shows that women with hyperemesis were at signi cantly higher risk of developing hyperemesis in their own pregnancy compared with daughters of women without hyperemesis (20). This nding also consistent with a study conducted in the USA which shows that a signi cantly higher risk of hyperemesis in women whose sisters or mothers had the disorder (21). The possible explanation for the observed association could be due to either shared environmental determinants among families or inheritance of some factors that can contribute to the development of hyperemesis gravidarum.
Women in the rst and second trimester of pregnancy were signi cantly more likely to suffer from HG than those women who were in the third trimester of pregnancy. The nding is consistent with a study conducted in Jimma University Medical Center (southwest Ethiopia) and Bale zone (south Ethiopia) and the University of Michigan(United States) (7,15,22). However, the nding contradicts a study conducted in Turkey, which shows that gestational age had no signi cant association with HG (23). This may be due to the difference in sample size in which the sample size used in Turkey was half of the sample size of the current study.
Similarly, H. pylori infection was found to have a signi cant association with hyperemesis gravidarum. This nding is consistent with a study conducted in Addis Ababa (Ethiopia), Egypt, and China (24)(25)(26).
This may be due to that H. pylori may aggravate the hormone-induced changes in the chemoreceptor trigger zone in the brain stem including the vomiting center and electric functioning of the stomach, which could lead an infected pregnant woman to develop severe nausea and vomiting (27). However, the nding is in contradicts a study conducted in Europe (28). This might be explained by the difference in socioeconomic status between the two study areas in which the prevalence rate of H. Pylori-infection is much higher in women with low socioeconomic status (29). Pregnant women with moderate and high perceived stress were found to have a statistically signi cant association with hyperemesis gravidarum as compared to women who have no low stress. This nding is consistent with a study done in south Ethiopia, Egypt, Turkey, and the USA (15,22,29,30).

Conclusion
This study concludes that being a housewife, being in the rst and second trimester of pregnancy, unplanned pregnancy, had a family history of hyperemesis gravidarum, Helicobacter pylori infection, and high perceived stress were found to be the determinants of hyperemesis gravidarum among pregnant women. We recommend that health promotion messages are needed to focus to improve the utilization of contraceptives methods to prevent unplanned pregnancy and screening for H. Pylori should be taken as routine investigations for pregnant women who complain of nausea and vomiting. Healthcare providers should look for the emotional stress during pregnancy and should minimize the stresses through psychological support during follow up.