A cross sectional study was carried out in field antenatal clinics in Anuradhapura District from 7th August 2018 to 7th September 2018. Anuradhapura is one out of 25 administrative districts in the country with a population close to 0.9 million . The district is having very high suicide rates  and suicide is reported as the leading causes of maternal deaths. There are 20 medical officer of health (MOH) areas covering entire district, which conduct field antenatal clinics at village level where pregnant women are provided with an antenatal care package under national maternal health programme. These field clinics are conducted at village level. Attending the antenatal clinics at least once was reported by 96% of antenatal women .
Study Population and Study Sample
All pregnant women with a period of gestation (POG) more than 12 weeks and visiting field antenatal clinics in Anuradhapura district were included in the study population. Pregnant women attending antenatal clinics in Anuradhapura, but not residing in Anuradhapura district (temporary visits) and those who are mentally handicapped and are unable to provide reliable answers for the questions were excluded.
Minimum sample size was estimated assuming that the prevalence was 50%, level of confidence was 95% and precision of 5%. Design effect (de) was selected as 1.5 as some homogeneity between clusters is expected. Response rate of 95% was applied. Estimated minimal sample size was 606.
Multi-stage cluster sampling with probability proportionate to the size was used for sampling. In the first stage of sampling, five (05) out of the 20 MOH areas in Anuradhapura district were selected randomly using a computer generated random number sequence. Only five MOH areas were selected due to feasibility issue. Clusters were selected from field clinics. Based on clinic attendance data in selected MOH areas, cluster size was determined to be 25. Total of 25 clusters were required to achieve the minimum sample size.
All the field clinics providing antenatal care in the selected MOH areas were listed in alphabetical order. Size of the clinic was determined by number of second and third trimester pregnant women attending each clinic during month of June 2018. Probability proportionate to size method was used to select 25 clusters.
From each clinic, 25 consecutive, consenting pregnant women fulfilling eligibility criteria were recruited for the study.
Measuring stigma and perceived threat
A vignette describing experience of ‘Ama’, a postpartum mother of a one month old baby who is experiencing symptoms of peripartum depression (without suicidal thoughts) was given in part A. In part B, the vignette was modified as the same mother getting suicidal thoughts frequently. Selected aspects of stigma and perceptions about symptoms of peripartum depression were included in the questionnaire as statements.
For conceptualizing stigma in the current study, mental illness stigma framework was used as it is more focused on individual experiences regarding stigma . According to this framework, there are three main types of stigma; perspective of stigmatizer, perspective of the stigmatized and perceived stigma. From the perspective of the stigmatizer, there can be three aspects of stigma; stereotypes, prejudice and discrimination. They are respectively the cognitive, affective and behavioral responses towards the affected persons. Stereotypes are beliefs about characteristics of the affected people. Prejudice is how people feel about affected person and discrimination is unjust or unfair treatment towards affected person. Perspective of the Stigmatized has been described under experienced stigma (experienced in real life), anticipated stigma (anticipated to experience) and internalized stigma (endorse the negative beliefs and feelings for self). Perceived stigma is the perception about stereotypes, prejudice and discrimination from the society, which is stigma shared by both people with and without mental illness. All three constructs of stigma from perspective of stigmatizer as per mental illness stigma framework were included as statements.
Similarly to assess the perceived severity, statements regarding what participants may think about symptoms of PPD (part A of vignette) and suicidal thoughts (part B of the vignette) were included.
A five point likert scale was used to assess level of agreement (from strongly disagree to strongly agree) to each statement considering how participants personally feel after going through the vignette. Responses given as strongly disagree and disagree were classified as disagreement to the statements which meant not having that particular stigma/perception.
Participant perceived possibility of them developing a mental health problem during antenatal or postpartum period of current pregnancy was marked in an adjectival scale. Responses given as impossible were classified as not perceiving susceptibility and all other responses (may be possible, possible and definitely possible) were classified as perceiving susceptibility.
Considering the sensitive nature of questions and high (94.6%) female literacy rates in the district , a self-completed questionnaire was used for data collection. Participants who found it difficult complete the questionnaire by themselves were assisted by data collectors.
To increase the internal validity a vignette should be constructed following assessment of existing literature, validated by a panel of experts and pretested .
This vignette was designed based on the DSM V diagnosis criteria for PPD . Survivor stories presented in scientific literature, qualitative studies, and internet based postpartum depression support groups were also referred when designing the vignette and the statements on stigma and perception. Opinion of reproductive age females in the same community and midwives, and doctors in the public health sector were also considered during the process.
The tool was validated by a panel of multidisciplinary experts which included a Consultant Psychiatrist, a Consultant Community Physician, a Medical Anthropologist, a Social Scientist, and two Medical Officers of Health (MOH) and a panel of public health midwives (PHMs).
Cognitive validation of the questionnaire was conducted with ten pregnant women. Pregnant women were asked to read the vignette and answer questions in the tool that were asked verbally from them. Then the thought process of respondents in giving a particular answer was assessed using probing questions on three aspects; comprehension of key phrases in the question/ vignette, applicability of the question to the community and sensitivity of the question/vignette. Answers were documented, and the questionnaire was changed accordingly.
The original vignette and the questionnaire were developed in Sinhala language and were translated to English and Tamil by professional translators. Consensus was obtained from three native Tamil health professionals regarding the Tamil translation.
The questionnaire was then pretested among pregnant women.
A team of medical undergraduates were recruited and trained as data collectors.
Use of vignette can introduce socially desirability bias to the study. To minimize this, participants were given a structured introduction before data collection. They were reassured that identification information are not collected in the questionnaire and responses regarding perceptions will not be checked when they handover the filled questionnaire. They were also informed about the value of providing responses genuinely representing their perception and thoughts without worrying about being right or wrong.
Data was entered by single entry technique. Manual verification was undertaken for 10% of the data. Data was analyzed using IBM SPSS version 22. Graphs were prepared using Microsoft Xcel.
Responses to stigma statements were subjected to an exploratory factor analysis using maximum likelihood extraction method. Oblimin rotation was performed. Prior to performing factor analysis correlation matrix was examined and many coefficients of 0.3 or above were observed. Kaiser-Meyer-Olkin Measure of Sampling Adequacy and Bartlett's Test of Sphericity were performed to confirm the suitability of data for factor analysis.
Direct logistic regression was performed using perceived susceptibility as the dependent variable. Factor scores obtained by factor analysis, age, average monthly family income and period of gestation (POG) were entered as continuous predictor variables. Having herd about PPD, knowing someone with PPD, having past diagnosis of a mental illness, having higher education and agreement to statements regarding perception were entered to the model as binary variables. Number of children they have was added as a categorical variable.
Informed written consent was obtained from all the participants. Ethical clearance for the study was obtained from the ethics review committee, Faculty of Medicine and Allied Sciences, Rajarata University of Sri Lanka.