Sampling size calculation

The number arrived at was 384. A further 10% was added to account for non-response or data recording errors. The number 422 thus obtained, was multiplied by 2 to counter the design effect. The final calculated minimum sample size was 844. This number was divided into 30 clusters, 15 clusters each from Grade 9 and Grade 10 (with a mean of 28 students in a cluster).

Sampling method

Multi-stage, cluster-sampling was used in identifying the study sample and to include a representative cross section of the school children. This was necessary because the geographical area included several categories of schools with wide ranging classroom numbers, resource availability and courses of study (Fig. 1).

A classroom with students between the minimum number [15] and the maximum number (40) was considered a cluster (mean 28). Allocation of clusters, to each category of school was done in proportion to the number of classrooms fulfilling the criteria for a cluster, in each category. This data were obtained from the relevant educational authorities. Thereafter cluster allocation was done randomly within the different categories of schools. While there were in total 96 schools in the education zone, after final cluster selection 46 schools were included in the study (type 1AB schools-14, type 1C-8 schools and type 2 schools 24).

Method of data collection

Students in the selected classrooms were given the questionnaire after obtaining prior informed consent from the school principal and class teachers. Consent was also obtained from the Director in charge of this education zone.

Prior, informed, written, consent was also obtained from the parents of the students in the selected classes. Information sheets about the study were also given along with the consent forms. Only those who returned the consent forms were included in the study. The assent of the participants themselves was asked for, before the questionnaire was given. Filling in the forms, on average, took around 20 min per questionnaire. The data collection was done within a 12 week period.

Assessment of mental health literacy

The adolescents included in the study responded to a questionnaire based on 4 clinical vignettes. The design of the questionnaire modelled that of the Australian National Survey on mental health literacy 2011 [5]. The current study used several vignettes, and questions based on the vignettes used in this study. The format of this survey had evolved and developed after several such surveys in the past years. The case vignettes used in our study, were on depression with suicidal ideation, social phobia, (Additional file 1) psychosis and diabetes mellitus. The vignette on diabetes mellitus (a physical health condition) was included for comparison. The questionnaire was initially written in English and translated to Sinhala and Tamil (the 2 main languages spoken in Sri Lanka). The questionnaire was tested for face and content validity using the Delphi method. Those participating in this process included, child and adolescent psychiatrists, psychiatrists and medical officers with experience in adolescent psychiatry (6 participants in all). The Delphi process was stopped only after all in the panel agreed on the content, structure, representativeness of the vignettes, cultural appropriateness of the questions and the suitability and appropriateness of the responses to the questions, after 5 rounds of discussions.

The questionnaire was first administered to a representative sample of students in another similar education zone. This pilot study was conducted to establish (i) acceptability of the (Additional file 2) questions, (ii) comprehension of the vignettes and questions, (iii) ease of administration, (iv) to refine the method of administration, (v) to assess the completeness of returned questionnaires, (vi) participation rates and (vii) to calculate the average time needed for completion. Test retest reliability was assessed by administering the questionnaire to a few of the pilot study participants a week later.

The final questionnaire consisted of close ended questions based on the vignettes. The questions required the respondents to give their opinion on, (i) whether the vignette depicted a mental health related problem, spiritual, physical related or other problem, (ii) what interventions could be helpful, (iii) what kinds of referrals would be helpful and (iv) possible outcomes for each vignette. Respondents were allowed multiple answers, as it was likely that this would be the situation in reality. The family income and the educational level of each parent, was included in the analysis as these socio-economic indicators were more likely to have an impact on mental health literacy.

The data was analysed for frequency distribution, cross tabulation and Chi square testing, using the Statistical Package for Social Sciences (SPSS) version 16 software package. The significance level for this study was set at 0.05.

Ability to recognise mental health problems

In this study, most respondents could recognise the depression, psychosis and social phobia vignettes as mental health problems. This was more than the number of respondents who recognised diabetes as a physical problem (Table 1).

However, this study assessed the ability of the participants to recognise a problem affecting the mental wellbeing, rather than an ability to give a diagnostic label of depression, social phobia or psychosis.

In Sri Lanka, there are no popular lay term for depression and other mental health disorders in local languages. Terms such as “vishadaya” for depression are unfamiliar to the general population [19–21]. This difficulty may explain why in a previous study, the recognition of depression among undergraduates in Sri Lanka was as low as 17.4% [17]. This rate fell further when medical undergraduates were excluded from the analysis.

In international studies, the recognition of disorders in vignettes varies widely, with depression, being the most easily recognised disorder [5]. Psychosis and social phobia had lower rates of recognition.

In this study too, the symptom complex of depression was the most easily recognised, and that of social phobia the least recognise as a mental health problem. It is possible that adolescents in the study population found it difficult to differentiate between normal social anxiety and social phobia.

Cultural influences may explain the high rate of responses for the psychosis vignette as being a spiritual problem. Belief in the supernatural and dissociative disorders, presenting with psychosis like hallucinations and delusions are well recognised in the Indian sub-continent [22].

However, in comparison to other such questionnaires our questionnaire did not include open ended questions and the response choices were limited. Having open ended questions and a broader response choice may have yielded more information. While our study was on the recognition of a mental health problem, other studies required the identification of specific mental health disorders (such as depression) as well. This may have an influence in comparing our study with other such studies.

Helpful interventions

In the current study, the highest response rate for the helpfulness of referral to a health service was for the diabetes vignette (69.6%), followed by psychosis (52.7%), depression (46.4%) and social phobia (41.61%) vignettes.

A relatively high response rate, across all the mental health related vignettes was seen for “talking to the person”, as a helpful intervention. This was highest for the social phobia vignette (50%) followed by depression (49.8%), psychosis (39.5%) and diabetes (19.5%) (Table 1). In comparison, in the 2011 National Mental Health Literacy Survey in Australia (youth component), the rates for “helping by talking to the person” were 52.2% for depression, 48.9% for psychosis and 46.4% for social phobia [5]. The differences in rates in the current study may again be due to cultural influences. It may be perceived that talking to a person with anxiety is more appropriate and acceptable, than talking to a person exhibiting delusions and hallucinations.

Interestingly although depression was better identified as a mental health, problem, those with psychosis were more likely to be referred to a health service. This may be an indication that depression is considered a less severe condition, with less urgency for referral to health services.

Surprisingly the number, who had responded that exercise would be beneficial in depression, was less than that for all other vignettes. In contrast, in a study among university undergraduates in Sri Lanka, 85.6% responded that physical exercise is a helpful intervention in relation to the given depression vignette [17]. It is possible that the link between mental well-being and physical activity was not understood by adolescents as in the undergraduate age group.

Helpful referral options

The response rate for referral to a medical doctor in the physical health vignette (diabetes) was significantly higher than for the mental health vignettes (Table 1). This indicates that respondents were likely to be more comfortable in seeking the support of a doctor for physical symptoms, as opposed to behavioural, emotional and cognitive symptoms, which are the common manifestations of mental ill health.

While the response rate for referral to a medical doctor for the psychosis vignette was 48.5%, the response rate for the benefit of a Bodhi pooja was 31.7%. A Bodhi pooja is a religious ritual centred on a Bo tree (Ficus religiosa). This ritual is derived from Buddhist cultural practices. This finding is consistent with the relatively higher culturally influenced response rates for a spiritual causation for psychosis. Similarly, a study in an adult population in India highlighted how cultural factors effect referral and help seeking options [23]. In this study 74% responded that mental illness is related to evil spirits or black magic and possibly due to sins in one’s past life. The same percentage also responded that going to a traditional healer would improve the condition [23].

Outcomes

A majority responded that the persons in all the vignettes would become better with treatment (Table 1). In the Australian National Mental Health Literacy Survey 2011; respondents in the general community survey were also more likely to believe in full recovery in relation to similar vignettes with problems re-occurring for those who received appropriate treatment for mental disorders [5].

Interestingly 33.63% had responded that the person in the social phobia vignette would recover on their own, but the corresponding response was 5.8% for depression. Even so, 48.2% considered a doctor in the government sector as a referral option in the social phobia vignette, which was similar to that for the depression vignette.

Around 20% also responded that the person depicted in the psychosis vignette would recover on their own. The reason for this is unclear, but it is possible that these adolescents are more familiar with transient psychotic like states, rather than chronic psychotic disorders such as schizophrenia.

Socio economic variables and mental health literacy

A higher socio-economic level was significantly associated with better recognition of mental health problems and appropriate referral (Table 2). Meanwhile when parental education levels were considered, the fathers’ education level was significantly associated with better recognition of mental health problems in only the depression and psychosis vignettes. The mothers education level was not significantly associated with recognition of any of the mental health related issues.

These finding contrasts with the finding of studies in other parts of the world, where increasing parental education levels have been significantly associated with higher mental health literacy levels [24]. The differences in the finding of the current study may be related to the differences in the parental education levels between the sample populations. In the current sample only 52.4% of fathers and 52% of mothers were reported to have continued education beyond the G.C.E O/L exam.

These findings however, point to the need for mental health initiatives in the future to specifically target populations in the lower income range.