Effect of antenatal depression on ANC service utilization in northwest Ethiopia

Maternal morbidity and mortality remain high among women who did not attend antenatal care (ANC). Antenatal care is one of the interventions given to pregnant women to detect existed problems or problems that can develop during pregnancy, which harm the health of pregnant women and fetuses. In Ethiopia, however, there is limited evidence that revealed the effect of antenatal depression on ANC service utilization. Hence, this study aimed to see the effect of antenatal depression on ANC visits among women in urban northwest Ethiopia. A population-based, prospective cohort study was done from June 2019 to March 2020. The Edinburgh postnatal depression scale was administered to 970 women in the second and third trimesters of pregnancy to screen for antenatal depression. Additional data were collected on ANC visits, the mother’s socio-demographic, obstetric, clinical, psychosocial, and behavioral factors. A logistic regression model was used to adjust confounders and determine associations between antenatal depression and inadequate ANC visits. The cumulative incidence of inadequate ANC visits was 62.58% (95% CI: 59.43, 65.63). The cumulative incidence of inadequate ANC visits among depressed pregnant women was 75% as compared to 56% in non-depressed. The incidence of inadequate ANC visits in the exposed group due to antenatal depression was 25.33%. After multivariable analysis, antenatal depression at the second and third trimesters of pregnancy remained a potential predictor of inadequate ANC visits (AOR = 1.96: (95% CI 1.22, 3.16)). In addition, antenatal depression, long travel time for ANC visits (AOR = 1.83 (95% CI 1.166, 2.870)), and late initiation of ANC visits (AOR = 2.20 (95% CI 1.393, 3.471)) were the predictors of inadequate ANC visits as compared to their counterpart. This study suggested that antenatal depression affects ANC visits in Ethiopian urban settings. Therefore, early detecting and treating depression symptoms during the antenatal period reduced significantly the impacts of depression on the health of the mother and fetus.

eligible for this study.The data collectors were interviewed through home-to-home visits and declared untraceable after three recruiting visits considered as unsuccessful.Totally, 970 women were identified within five months (June 2019-October 2019) and followed up to March 2020.Potentially all cohort pregnant women who gave birth were eligible (Fig. 1).

Data quality management.
A data collection tool was pilot tested in a similar context.The data collectors were Diploma and BSc holders in nursing who had experience in data collection in the field.Before the commencement of data collection, the data collectors and supervisors took comprehensive training with practical exercises for two days that help them to understand the contents of the questionnaire, objectives, and ethical issues relevant to the study.Similarly, Master's degree holder supervisors were assigned during data collection.Besides, the principal investigator, field supervisors made regular on-site supervision during the data collection periods.The questionnaires were rechecked by the supervisors and the principal investigator for completeness and consistency in a weekly based face to face meetings and telephone communications.Accordingly, incomplete or incorrectly filled questionnaires were identified and managed.For the data entry, templates with proper check codes were designed and employed to minimize data entry errors.The experts who did the data entry had also ample experience on using the Epi-data software.Besides, double entry of the data, 10% of the questionnaires were done to detect and manage the mistakes likely to occur during data entry.Also, the principal investigator randomly cross-checked the correctness of the data entered to the Epi-data software and the data captured in the questionnaire.
Measurement.Primary exposure.Antenatal depression was the primary exposure variable; which was assessed using the EPDS.The EPDS has a sensitivity, specificity, and internal reliability Cronbach's alpha of 78.9%, 75.3%, and 0.71 respectively; from a validation study in Ethiopia 31 .It includes ten items with a Likert scale of responses scored from 0 to 3 to a maximum score of 30; which was then coded as Yes 1 if there is depression (scores ≥ 12) and No (0) if otherwise (< 12 scores) to indicate scores that are likely to suggest depressive disorder.
EPDS is a preferable scale over other depression scales to screen for depression during pregnancy; because it removes the physical symptoms of depression associated with pregnancy 32 .In these local kebeles, a woman with suicidal ideation was referred to a psychiatric unit in Debre Tabor Hospital for further assessment and appropriate treatment.
Outcome variables.The primary outcome variable was inadequate ANC visits.ANC visits were categorized as inadequate (< four visits) and adequate (≥ four visits) visits.ANC visits were obtained by interviewing women after delivery 33 .

Potential confounding variables.
The most common potential confounding factors were long travel time for ANC visits (categorized as 'Yes' if the time takes greater than or equal to 30 min, or 'No' if otherwise), threatening life events, lack of time for ANC visit, social support, intimate partner violence, waiting a long time for ANC service (categorized as 'Yes' if the time takes greater than or equal to 120 min or 'No' if otherwise, initiation of ANC visit (If a mother starts ANC visit within 3 months of pregnancy period = early initiation).In addition, any chronic medical condition, integration of ANC influence ANC service, history of complications current www.nature.com/scientificreports/pregnancy, history of alcohol misuse, socio-demographic and economic variables, including marital status, wealth index, and level of education also assessed.Experiences of stressful life events during the six months before the assessment was assessed using the List of Threatening Experiences (LTE).The scale contains twelve items and includes questions about death, illness, conflicts, and loss of property 34 .The presence of LTE ascertained by the presence of at least one or more LTE.The test-retest reliability of the LTE was good, with a Kappa of 0.61-0.87 35.
The Oslo Social Support Scale (OSSS-3) 36 was used to measure maternal social support during pregnancy.The level of social support classified as "poor support" [3][4][5][6][7][8] , "moderate support" [9][10][11] , and "strong support" 12,14 scores.The OSSS-3 consists of three items assessing the number of close intimates, perceived level of concern from others, and perceived ease of getting help from neighbors.The OSSS-3 has good convergent and predictive validity 37 .Both the LTE-12 and the OSSS-3 have been used in a population-level study in Ethiopia 38 .
Intimate Partner Violence (IPV) during pregnancy was assessed by asking three questions related to emotional, physical, and sexual violence.The presence of IPV was ascertained by the presence of at least one type of IPV 39 .The history of physician-diagnosed chronic medical conditions, including cardiac disease, hypertension counted for each woman, recorded as "No" for those without any chronic medical conditions, and "Yes" otherwise History of current pregnancy complications: anemia, hypertension, edema, Antepartum haemorrhage (APH) were counted for each woman and recorded as "yes/no." Alcohol use was assessed using the four-item scale; the Fast Alcohol Screening Test (FAST) 40 , which ranges from 0 to 16. Hazardous drinking refers to a score of three or more on a FAST scale 40 .
Analysis strategy.We used Stata software (version 14) for analysis.We run descriptive statistics for all study variables.Percentage values, with their corresponding 95% confidence intervals were used to summarize categorical variables.We used the odds ratio to measure the effect of antenatal depression on ANC visits.We carried out univariate logistic regression analyses for each risk factor to identify possible predictors associated with inadequate ANC visits.During univariate analysis, variables were selected for multivariable analysis if the p-value was < 0.2.The difference was deemed to be significant if p < 0.05.The total number of losses to follow up was 32 (3.2%).We also used a complete case analysis as it was suggested less than 5% lost to follow-up was with little concern 41,42 .We used a principal component analysis for the wealth index and checked its assumption of whether it fulfills or not.There were 50 or more valid cases, a case-to-variable ratio of at least 5-1, two or more correlations of 0.30 or higher, a sampling adequacy measure of at least 0.50 for each variable, an overall sampling adequacy measure of at least 0.50, and a probability of the Bartlett test of sphericity less than the level of significance for each variable.
Ethics approval and consent to participate.We obtained ethical approval from the Amhara region public health institute and Institutional review board (IRB) of the University of Gondar.We received permission from Debre Tabor and Woreta towns' health department administrat office.All methods were carried out in accordance with the Helsinki declaration guidelines and regulations.Before we want to collect the data, participation information sheet read for all participants and we obtained written informed consent from all participants.The participants were told that participation is fully voluntary and they had the right to refuse, withdraw at any time and refrain from answering any question that they don't want to without any reprisal.Information gathered from the participants was stored in a secured cabinet by the principal investigator.Besides these, the data obtained from participants kept confidential.Those women who were expressing sever suicidal ideation referred to Debre Tabor referral Hospital psychiatry unit for further diagnosis and appropriate management.The decision for referral was made by the principal investigator based on the review of all women who were expressing suicidal ideation.

Result
The detailed recruitment profile of women showed in the flow chart in Fig. 1. Between June 2019 and October 2019, 970 pregnant women were eligible to participate.Of this thirty-two (3.3%), declined to participate because of refusal to participate, and moving from the area.Finally, 938 women were included in the analysis, with a response rate of 96.7%.Our study revealed that the cumulative incidence of inadequate ANC visits was 62.58%.Of 938 participants, 858 (91.47%) women attended at least one antenatal care (ANC).However, from those who attended antenatal care services, one-fourth of 173 (20.16%) participants made their first visit during their first trimester.Of these, only 351(37%) of women had four or more ANC visits during their last antenatal period.The mean (± SD) of ANC visits was 3.1 ± 1.3, ranging from 0 to 6.

Socio-demographic characteristics of the study participants.
Out of the 938 mothers, 403 (43%) of them were found between the ages of 25-29 years.The mean (± SD) age was 27.1 ± 4.8 years.In this study, women who were married was 860 (92%).Regarding the level of education, 797 (85%) of the participants had formal education (Table 1).

Obstetric and clinical characteristics of the study participants.
The majority of the participants, 858 (91.47%) had one or more ANC visits during their last pregnancy.The result of this study revealed that 109 (12%) of the study population experienced current pregnancy complications.Among the participants, 160 (19%) did not have enough time to visit the ANC service.Around one-fifth of women 182 (19.40%) spent less than thirty minutes to arrive at the health facility (Table 2).
Factors associated with inadequate ANC visits.Antenatal depression, previous pregnancy complication, marital status, life-threatening events, history of depression, intimate partner violence, family history of depression, unplanned pregnancy, Social support, alcohol use, hunger, debt, long travel time for ANC visit, and late initiation of ANC visit all had a bivariate relationship with inadequate visits with the P-value < 0.20.All these variables were a candidate for multivariable analysis.
Multivariable logistic regression analyses showed that, after adjustment for covariates, women with antenatal depression symptoms (AOR = 1.96 (95% CI 1.220, 3.162)) were twice more likely to be inadequate ANC visits than women without antenatal depression symptoms.In addition, women with a history of late initiation of ANC visits (AOR = 2.20 (95% CI 1.393, 3.471)) and long travel time for ANC visits (AOR = 1.83 (95% CI 1.166, 2.870)) were increased odds of inadequate ANC visits as compared with their counterpart (Table 4).

Discussion
The main finding was antenatal depression has effect on ANC visits in the Ethiopian urban settings.This is the first to our knowledge population-based, prospective cohort study examining specifically the effect of antenatal depression on ANC visits in our country.After controlling for known risk factors for inadequate ANC visits, antenatal depression (AOR = 1.96 (95% CI 1.220, 3.162)) remained significantly associated with inadequate ANC visits.The incidence of inadequate ANC visits in the exposed group due to antenatal depression was 25.33%.Similarly, the positive effect of antenatal depression on inadequate ANC visits was reported in the United States of America, England, and East Asia region 22,23,25,26 .The possible explanation might be depression symptoms have loss of interest, and fatigue that disturb normal mother-infant bonding 43,44 .However, the association that we have reported about the increased odds of inadequate ANC visits among women with antenatal depressive symptoms did not replicate the findings of the recent cohort study from Ghana and the US of America 21,27 .These conflicting results might arise due to differences in socio-cultural, demographic factors, and sample size.
Furthermore, results from multivariable logistic regression analyses indicated that long travel time to health facility found to be highly influential over the use of ANC visits.In this study, women who travel more than 30 min to health facilities were 1.83 times more likely to have inadequate ANC visits as compared to their counterparts.This finding is supported by several other studies including in Ethiopia [45][46][47][48][49][50][51][52][53][54] .It is a fact that the accessibility of a health facility is the main barrier to maternal health service.However, our finding is in contrast to the study finding in Zambia 55 .The possible explanation might be due to Geographic, socio-cultural, and level of awareness differences.
Study findings from Cameroon and South Africa revealed that timely initiation of ANC visits is a key method to decrease pregnant women's morbidity and mortality 3,56 .Also, pregnant women who received an early ANC check-up were much more likely to receive four or more WHO-recommended ANC visits 57 .Despite this importance, the results of this study revealed that only 173 (20.16%) of pregnant women were found to early initiate ANC visits.This estimate reflects under-utilization of ANC, and this could contribute to high maternal mortality.Those pregnant women who were late initiated ANC visits were 2.2 times more likely to have inadequate ANC visits as compared to their counterparts.The result was consistent with the research finding from Belgium 58 .
Limitations.This study has limitations that prevent the full understanding of adequate antenatal care for ANC attendance in the study area; specifically, we did not address the content adequacy of antenatal care.Besides, we had no information on antidepressant medication for those referred to a psychiatric unit, thus it could not be assessed its roles as a confounder in multivariable analysis.
Despite these limitations, the results of this study have important implications.As this study confirmed that antenatal, depression has an adverse effect on ANC visits.
The health care providers and policymakers should think about the efforts to reduce the effect of antenatal depression such as routine screening of all pregnant women for depressive symptoms and treating them at the primary health care level by integrating the service; since screening is an effective approach for reducing morbidity in depressed people 59 , as a result, adequacy of ANC visits could be improved or maintained.

Conclusions
This study suggests that antenatal depression affects ANC visits in Ethiopian urban settings.The incidence of inadequate ANC visits in the exposed group that is due to antenatal depression was 25.33%.Therefore, early detecting and treating depression during the antenatal period reduced significantly the impacts of depression on the health of the mother and fetus.Discussing antenatal depression and its effects on ANC visits for the community as a whole is warranted.
Flow chart of recruited pregnant women and the outcome of inadequate ANC visits.

Table 1 .
Frequency distribution of Socio-demographic factors among pregnant women at Debre Tabor and Woreta towns, Northwest Ethiopia, 2020.

Table 2 .
Frequency distribution of obstetric and clinical factors among pregnant women at Debre Tabor and Woreta towns, Northwest Ethiopia, 2020.

Table 3 .
Frequency distribution of psychosocial and behavioral factors among pregnant women at Debre Tabor and Woreta towns, Northwest Ethiopia, 2020.

Table 4 .
Bivariate and multivariable analysis for factors associated with inadequate ANC visits among women at Debre Tabor and Woreta towns, Northwest Ethiopia, 2020.NB Abbreviations, CI confidence interval,