The study employed a quantitative cross-sectional design. The design enables the assessment of the prevalence of an outcome and examines the relationship between exposures and outcome at a point in time [11] [24]. In this study, the outcome was unintended pregnancy and the exposures were individual sociodemographic and reproductive factors.

The study was conducted at Thyolo District Hospital in the Southern Region of Malawi. Thyolo District is one of the districts with the highest (5.7) fertility rates in the country [22] [23]. The hospital serves as the referral hospital for 37 health centres but also offers antenatal clinic services every day of the week to the surrounding communities.

The study was conducted from October, 2016 to December, 2016. The study population was pregnant women attending antenatal clinic at Thyolo District Hospital during the study period.

The subjects recruited were pregnant women attending antenatal clinic at Thyolo District Hospital between October, 2016 and December, 2016 and had given informed consent.

In this study, pregnant women who were unwell and required urgent medical care were excluded. Any woman who had more than one visit to the clinic during the data collection period was excluded to prevent the recruitment of the same participant more than once.

The Kish Leslie (1965) formula was used to calculate the sample size for both study objectives [25]. Based on previous studies, the proportion of 43% and the level of significance at 5%, a sample size of 378 was obtained [12]. In addition, 5% of the sample was added to mitigate the non-response rate.

The unit of sampling was the individual antenatal woman and a systematic random sampling strategy was used to recruit the women into the study. Systematic random sampling was used to recruit the required sample of 396 [25]. At the antenatal clinic, the clients are given numbers in order of the way they have reported unless the mother has a complication. Such numbers help to order when attending to clients as first come and first to be served unless the woman requires emergency care.

Data was collected through face-to-face interviews using a structured questionnaire administered by the researcher and trained research assistants. The questionnaire was developed following a review of the literature but also guided by World Health Organization literature on unintended pregnancies [9] [16]. In addition, some items were adapted with permission from the previous study conducted in Malawi [12].

The data was collected by the researcher and three trained research assistants in face-to-face interviews with the participants. The research assistants were trained on interpersonal communication skills, the method of eliciting responses from participants and upholding the rights of the subject in the study. During the training, the research assistants were also oriented to the research questionnaire and the method of filling the responses to reduce missing data, biases during data collection and the Hawthorne’s effect [26].

The questionnaire and consent forms were translated into the local language of Chichewa and pre-tested at another facility in the same district to assess the validity and the likelihood of the questions to elicit the intended responses [26]. The flow of questions and validity of the items were reviewed and proper adjustments were made to adapt the tool to the local population under study.

During the data collection period, each questionnaire was reviewed at the point of data collection before closing the interview to ensure completeness. The principal investigator who was also involved in data collection periodically reviewed the filled questionnaires by the assistants to monitor and control the procedure.

Data was entered into the SSPS version 21 data set after coding the variables into a numerical representation. Then data was exported to Stata version 13 for analysis.

Data set cleaning was performed using STATA through the identification of wild codes and outliers to correct mistakes made during data entry. Outliers and wild codes were followed in the questionnaire to make appropriate decisions. Analysis of data was conducted using STATA version 13 computer package for descriptive, bivariate and multivariate analysis.

At the bivariate level of analysis, the association between dependent and independent variables was established using Chi-square and binary regression. Strength of association was determined using odd ratios and a 95% confidence interval. A P-value of less than 0.05 was considered to be statistically significant.

At the multi-variate level, a logistic regression model was employed to assess the net effect of several independent significant variables on unintended pregnancy and variables with a P-value of less than 0.05 were again considered significant.

Ethical approval to do the study was sought from Makerere University School of Health Sciences Institutional Review Board (MakSH-IRB) with reference number SHSREC-2015-029 and the National Commission for Science and Technology of Malawi (NCST) with reference number NCST/RTT/2/6. Finally, approval was also obtained from Thyolo District Hospital’s Ethical Committees. In addition, a written informed consent was obtained from the study participants. Women who were below the legal adult age of 18 were recruited after getting an ethical assent from them.

While some of the adolescents had emancipated themselves through marriage below the legal age of 18. In general, the world health organization recommends that the parents have no veto ethical powers over adolescents’ decisions to access care and participate in reproductive health research [27]. Though the adolescents as minors were involved, a well-designed research on reproductive health involving adolescents was necessary in order to enhance scientific knowledge specific to these individuals [27].

Enrolment in the study was voluntary and patients had the right to withdraw at any time during the study. Participants were informed that there was no direct benefit from this study. Interviews were conducted in a secluded room away from other clients. Confidentiality was observed by using codes on the questionnaire instead of names.

The data set was secured with a code and information during the study was made accessible only to persons directly involved in the study. The questionnaires were kept safely in a closed cabinet to avoid unauthorized access. Anonymity was further observed by reporting the research findings as grouped data and not for the specific individual woman.

The study recruited a total of 396 pregnant women. The population was generally youthful with the majority (63%) below the age of 25 years and the mean age of the participants was 23.2 years (SD 5.9). The participants were mainly from the rural setting and a little over half (60%) were educated up to the primary level. The majority (85%) of subjects were married and most (69%) of the subjects were reported to have married before their Twentieth birthday (Table 1).

CCAP means Church of Central Africa Presbyterian.

Most (53.7%) of the spouses of the married participants had secondary education. Economically, most (67.7%) of families were living below 1$ per day. Very few women (6.6%) had salaried employment. The main source of income for a larger proportion (60%) of participants was farming. Slightly more than half (51.6%) of the spouses of the participants were employed (Table 2).

The majority of the women 189 (48%) were prim-gravida, 178 (45%) were gravida 2 - 4, and 29 (7.3%) were gravidae five and above. Half of the women had not carried a pregnancy to term. A significant proportion of 171 (43.2%) was para one to three, and 27 (6.8%) had delivered more than 3 times. Most women 205 (52%) had no living child, while 174 (43%) had one to three children, and, 17 (4%) had four children and above. Amongst women who had children, 46 (24%) had a boy or only boys only, 58 (30.2%) had a girl or girls only while 88 (45.8%) had both a girl and a boy child or children.

In the study, 54% of pregnant women were reported to have unintended pregnancies (Figure 1). Furthermore, a proportion of 163 (76%) of unintended pregnancies were observed among the women residing in rural parts of the district compared to those 52 (24%) from the peri-urban area (Figure 1).

There was a significant association between age of the participant and unintended

Employment included jobs in tea and coffee estates, professional job, and craft industry * one participant did not respond.

pregnancy. Teenage mothers had a higher proportion 90 (42%) of all unintended pregnancies. Women below the age of 20 were 70 percent more likely to report unintended pregnancy compared to the women who were of age between 20 and 24 years (COR 0.30; 95% CI, 0.16 - 0.50) (Table 3).

COR for crude odds ratios and AOR for adjusted odds ratios, *statistically significant results, P < 0.05.

Single women were 96% more likely to report having unintended pregnancy compared to married women (COR = 0.04, 95% CI, 0.02 - 0.16). The age at which the woman got married was also found to be significant, women who got married before 20 years were 2.7 times more likely to have unintended pregnancy compared to those who got married above the age of twenty (COR = 0.37, 95% CI, 0.21 - 0.68). After adjusting for marital status, type of marriage, level of education, gravidity and parity and the number of living children, age at marriage was significantly associated with unintended pregnancy (AOR 0.41; 95% CI, 0.20 - 0.82) (Table 3).

Comparatively, the women in polygamous relationships were four times more likely to report unintended pregnancy than those in monogamous relationships (COR = 4.2; 95% CI, 1.34 - 12.9). After adjusting for age, education level, gravidity, parity and number of living children, women in polygamous relationships were 3.5 times more likely to report unintended pregnancy than women in monogamous relationships (AOR 3.5; 95%, 1.08 - 11.36).

In relation to reproductive characteristics of the women, the Primigravidas were 3 times more likely to report unintended pregnancy than women of gravida 2 - 4 (COR = 0.36; 95% CI, 0.23 - 0.56). The education level of the woman was another socio-demographic factor that remained a significant predictor of unintended pregnancy (AOR 0.53; 95% CI, 0.31 - 0.89) (Table 4).

MK 50,000 equivalents to almost US 30 dollars at the time of study, COR for crude odds ratios and AOR for adjusted odds ratios, *statistically significant results, P < 0.05, Note that the - (spaces), is not omitted but dropped.

There was a significant association between the age of the participant and unintended pregnancy. Teenage mothers had a higher proportion of 90 (42%) of all unintended pregnancies. Women below the age of 20 were 70 percent more likely to report unintended pregnancy compared to the women who were of age between 20 and 24 years (COR 0.30; 95% CI, 0.16 - 0.50).

Single women were 96% more likely to report having unintended pregnancy compared to married women (COR = 0.04, 95% CI, 0.02 - 0.16). The age at which the woman got married was also found to be significant, women who got married before 20 years were 2.7 times more likely to have unintended pregnancy compared to those who got married above the age of twenty (COR = 0.37, 95% CI, 0.21 - 0.68). After adjusting for marital status, type of marriage, level of education, gravidity and parity and the number of living children, age at marriage was significantly associated with unintended pregnancy (AOR 0.41; 95% CI, 0.20 - 0.82).

The study found that pregnant women who were employed or whose partner was employed were less likely to have unintended pregnancies (Table 4). Furthermore, the families who earned at least half a dollar per day were less likely to report unintended pregnancies (Table 4).