Data were collected by means of a self-administered questionnaire. The questionnaire consisted of three parts. The first part of the questionnaire included information on the following: age, gender, number of years practising medicine, area of medical practice (state or private sector), contact with a close family member with mental illness. The second section of the questionnaire consisted of 10 statements describing attitudes towards mental illness in general. This component was developed after looking at other similar studies, which used Likert-scale questionnaires to assess mental illness attitudes.^10,30 Ten statements covering stigma concepts such as social distance, negative attitudes, stereotyped beliefs and diagnostic overshadowing were compiled and used in the questionnaire. Respondents were asked to grade their level of agreement with the statements using a 5-point Likert scale (strongly agree, agree, uncertain, disagree and strongly disagree). The third section of the questionnaire consisted of 10 questions designed to assess the general knowledge of mental illness. Each question had three possible answer options (correct answer, incorrect answer and an option if the participant was unsure of the answer). This component was developed after perusing published mental illness literacy scales that assessed the general knowledge of mental illness.^31,32,33 – these scales were not used in their entirety, with the aim of maintaining the brevity of the study questionnaire, so that participants would not find it tedious to complete. The questions were designed to cover factual aspects of mental illness that generalist doctors would likely encounter frequently, such as depression and anxiety, substance use, dementia, personality disorders and treatment non-adherence.

A link to the study questionnaire (an online questionnaire using a freely available online survey platform), information about the study, details of the researcher and matters pertaining to informed consent were emailed to potential participants. Participants accessed the questionnaire through the electronic link and submitted their responses electronically. The questionnaire was disseminated to potential participants as follows: (1) by the administrators of professional practitioner bodies including Independent Practitioners Association, the Islamic Medical Association, the General Practitioners Management Group and others; (2) by clinical managers at state facilities (primary healthcare clinics and district hospitals); and (3) by the researcher directly inviting eligible individuals to participate in the study or to extend the invitation to other colleagues. Data collection took place from December 2017 to September 2018. Data responses were originally recorded automatically in an online database and were then manually exported and recorded in a Microsoft Excel spreadsheet.

Ethical approval was obtained from the University of the Witwatersrand Human Research Ethics Committee (protocol number M170733). Owing to the nature of the data collection, it was understood that written consent could not be obtained from each participant. An information leaflet with relevant information about consent for study participation was distributed along with the questionnaire, which made participants aware that tacit consent to participate was implied once an individual accessed the questionnaire and began the survey. Participants were informed that they had the option of not continuing with the questionnaire at any point or choosing not to submit their answers. In addition, they were not obliged to answer any question that they did not wish to.

The mean age of the SS group was 34.4 years (range: 23–56 years), which was significantly lower than that of the PS group (44.6 years; range: 29–71 years) and the SPS group (44.3 years; range: 31–64 years) (Wald χ² = 90.73, p < 0.010). The majority of participants in the SS group were in the < 35 year age group (52.3%, n = 38), between 35 and 45 years in the PS group (36.2%, n = 21) and between 35 and 45 years in the SPS group (50%, n = 5) (Table 1). Of note was that 17.2% (n = 10) of participants in the PS group were over 55 years of age compared with only 1.4% (n = 1) in the SS group.

Majority of the participants were male in the PS group (60.3%, n = 35) (χ² = 90.73, p = 0.115) and the SPS group (80%, n = 8) (χ² = 3.60, p = 0.060), whilst the majority were female in the SS group (66.6%, n = 48) (χ²₂ = 8.00, p = 0.005) (Table 1). The majority of participants in the SS group were practising for less than 5 years (40.3%, n = 28), for more than 20 years in the PS group (34.5%, n = 20) and for between 11 and 15 years in the SPS group (50%, n = 5) (χ² = 12.96, p = 0.044). Significantly more respondents in the SS group were employed for less than 5 years and significantly fewer for 20 years or more, compared with the other two groups (χ² = 27.71, p < 0.001) (Table 1). There were no statistically significant differences with respect to personal or family contact with mental illness (p > 0.05).

The majority of the participants (> 50%) in all three groups agreed with the following statements in the questionnaire: ‘people with mental illness rarely pose a risk to their families and communities’, ‘employers should hire a person with a managed mental illness if he or she is the best person for the job’, ‘I feel that I am adequately trained and have sufficient knowledge to be able to diagnose and initiate treatment in a person with mental illness’ and ‘I would not mind if a person with mental illness lived next door to me’ (Table 2).

The majority of participants (> 50%) in all three groups disagreed with the following statements posed in the questionnaire: ‘most people with mental illness don’t try hard enough to get better’, ‘I sometimes perceive people with mental illness as being weak or lacking in willpower and self-discipline’ and ‘if a person with mental illness complained of physical symptoms, I would likely attribute this to their mental illness’ (Table 2).

The only significant difference found in responses for individual questions was that participants in the SS group were more likely to agree that they were more comfortable treating someone with a physical illness than a mental illness, when compared with the PS group (Wald χ²₁ = 15.691, p = 0.047).

The most overall positive attitude towards mental illness was observed in the PS group (total modal score = 11) followed by the SS group (total modal score = 8) and the SPS group (total modal score = 6). The differences between the three groups were not statistically significant (χ²₂ = 1.52, p = 0.468).

The majority of participants in all three groups (> 50%) gave the following correct responses to questions in the survey: ‘psychotherapy may be the most beneficial intervention in personality disorders’, ‘someone who has a close family member with depression is at risk of developing depression’, ‘people with schizophrenia who do not take their treatment regularly most likely do so because they do not recognise that they have an illness’, ‘people with post-traumatic stress disorder are at high risk for developing alcohol abuse’, ‘a woman with borderline personality disorder who frequently becomes suicidal because of an ongoing fear is most likely fearful of abandonment’ and ‘an elderly patient who is brought to the doctor by her family as she has become difficult and unmanageable at home over the past few months, hiding things away and accusing family of stealing things from her, is most likely suffering from dementia’ (Table 3).

The majority of the doctors in the SS group (> 50%) and all of the doctors in the PS group and SPS group responded correctly to the question, which stated that people who have had one episode of depression are most at risk for another episode of depression.

In addition, the majority of doctors in the SS and PS groups (> 50%) gave correct responses to the following questions: ‘a man at the airport who is talking and laughing very loudly, passing money to people and talking about his plans to fly around the world is most likely suffering from mania’, ‘a woman who has been isolating herself from friends and family and no longer pursues hobbies and interests as before is most likely suffering from depression’ and ‘mental illness can most often be attributed to chemical imbalances or structural damage to the brain’.

The sum of the modal scores for all three groups was the same, indicating that there was no statistical difference between the knowledge scores of the three groups.

The SPS group was excluded from this analysis because of the small sample size. There was no statistically significant difference in the overall attitude towards mental illness between the SS and PS groups with respect to age (Wald χ²₁ = 9.564, p = 0.089), gender (Wald χ²₁ = 0.636, p = 0.425), duration of practice (Wald χ²₁ = 0.292, p = 0.999) and having a family member with mental illness (Wald χ²₁ = 3.248, p = 0.197).

There were also no significant associations between overall knowledge score between the SS and PS groups and any of the sociodemographic variables (p > 0.05), as well as no significant influences of sociodemographic variables on differences in individual responses to any of the knowledge questions (p > 0.05).

The study found that the overall attitude towards mental illness of both SS and PS doctors was positive. This is similar to the findings of published international studies.^34,35,36,37

In one such study, Wang et al.³⁸ found that over 80% of study respondents (who were Chinese non-psychiatric hospital doctors) felt that they had a responsibility towards their patients in managing their psychological distress. To the best knowledge of the researchers, there are no published studies amongst non-specialist medical doctors working in South Africa. However, Eksteen et al.³⁹ found that psychiatrists had the least stigmatising attitudes when compared with pre-clinical and post-clinical medical students. Mausling et al.⁴⁰ showed that second-year medical students had positive attitudes towards psychiatric illness and psychiatry, whilst Dube et al.²³ demonstrated that nurses working in primary healthcare clinics were positive about the management of psychiatric patients. Studies have also shown positive attitudes in other healthcare workers, such as medical and nursing students⁴¹ and nurses, social workers and healthcare assistants.⁴² Similar to this study, it appears from the literature that doctors and other healthcare professionals exhibit positive attitudes towards people with mental illness.

However, there is also a significant body of literature, which shows that doctors have negative attitudes towards mental illness.^11,13,43,44

One such review,³⁶ looking at stigma and discrimination towards mental illness in healthcare professionals found negative attitudes in psychiatrists, nurses, psychologists and medical students; it appeared that the most negative attitudes were described in psychologists. It was also found that chronically ill patients with more hospital admissions were viewed more negatively than patients with milder symptoms. In a South African study, Jury⁴⁵ described negative attitudes towards mental illness in specialist doctors practising in state sector teaching hospitals. In addition, a study of South African nurses by Mavundla et al.⁴⁶ found that participants held mainly negative attitudes towards the mentally ill.

The positive attitudes found in this study could have a number of explanations. It may have been partly as a consequence of improved training and exposure to psychiatry at the undergraduate level, and exposure to psychiatric patients during the internship period.⁴⁷ In addition, there is increasing recognition that the contribution of mental illness to the global disease burden is on the rise⁴⁸ with a consequence of this being that healthcare workers have increased exposure to mental illness and thus may have more positive perceptions. With mental health issues such as the Life Esidimeni tragedy and the recent suicides of prominent figures being widely covered in the mainstream and social media, awareness and advocacy for mental health has improved. This has likely led to a more sympathetic stance towards mental illness amongst doctors and in other sectors of the general population. Mental illness is also on the rise within the medical profession, with recent statistics showing a considerable prevalence in doctors (at the time of writing this article, statistics for mental illness in South African doctors were not available).^49,50

This may be another contributing factor to positive attitudes, as doctors may be more likely to have a favourable opinion of mental illness if they have personally experienced such symptoms.

Although the difference was not statistically significant, this study found that SS doctors were slightly less positive towards mental illness than their PS counterparts. Similar findings have been reported in other studies.^7,9,51 Jury’s⁴⁵ study of specialist doctors practising in South African state sector teaching hospitals described negative attitudes towards mental illness, in particular towards schizophrenia and borderline personality disorder. There are, however, published studies with contrasting findings.^15,52,53

The less positive attitude amongst SS doctors may be because of inadequate exposure to mental illness because of a lack of integration of mental healthcare at primary healthcare level, a demanding workload and a tendency to focus consultations on physical complaints with less time devoted to mental health complaints. In addition, as the SS doctors in this study were generally younger and less experienced than the PS doctors, it is likely that age and lesser experience contributed towards less positive attitudes, as in other studies.^11,54 Studies have shown that increased exposure to mentally ill people leads to more positive attitudes.^55,56,57 It is also likely that PS doctors are more open to advocating for their mentally ill patients because of the lesser demands and time constraints of their work as compared with state-employed doctors. Within private practice, it is easier to develop long-lasting and enduring doctor–patient relationships, as patients most often consult with the same doctor multiple times. This is unlike the state sector, where many doctors work in a single department and doctors rotate through different units, hospitals or clinics, meaning that they often do not consult regularly with the same patient. This fragmented relationship may exacerbate feelings of social distance and contribute to less positive attitudes towards patients with mental illness.

This study found that the factors of age, gender, duration of practice and having a family member with mental illness were not significantly associated with overall attitudes towards mental illness. Similarly, Sujaritha et al.⁵² and Sri et al.⁵³ reported that gender and level of qualification did not influence mental illness attitudes, whilst Vistorte et al.⁵⁸ found that gender, age, training and years of experience did not hold any association with stigmatising attitudes.

This is contrary to the findings of other published studies, which have reported both positive and negative associations with demographic characteristics. Female doctors have been found to be more positive and much less stigmatising than male doctors.^45,59,60,61 Stuber et al.⁶² noticed that younger female mental health professionals were less stigmatising than older male professionals. Ewalds-Kvist et al.⁶¹ reported that even though older individuals were more open-minded towards mentally ill people, they were less likely to be supportive of community psychiatric services and integration. Those with more clinical experience⁶³ or a family member or friend with mental illness^7,11 were more sympathetic towards the mentally ill. However, Chandramouleeswaran et al.⁵⁵ found that personal contact with an individual with a psychiatric condition did not increase the positivity of mental illness attitudes.

Within available literature, the overall impression is that there is no consistent sociodemographic predictor of mental illness attitudes as the relationship between variables and attitudes differs across studies, suggesting that these interactions may be complex and multifaceted. The interplay between demographic factors and cultural influences is also likely to play a role in how mentally ill people are viewed by others.

The study found that doctors had adequate knowledge of mental illness, as evidenced by the high modal scores. Similar findings have been reported in other studies.^64,65,66

In contrast, some studies have reported low levels of mental illness knowledge amongst doctors,^15,19,67,68 particularly in those who had little or no formal psychiatric training. It would appear that non-psychiatric doctors are, at the very least, comfortable in diagnosing common mental illnesses such as depression but feel less confident in being able to treat mentally ill patients effectively.^69,70 Inadequate mental illness knowledge appears to be prevalent in medical students and nurses as well.^20,22

The findings of adequate levels of knowledge in this study may have been, in part, because of training and exposure to psychiatry in medical teaching facilities in South Africa. South African universities have established training programmes for medical and psychiatric teaching, with four of South Africa’s medical schools being ranked in the top 300 best universities globally for clinical medicine.⁷¹ Another explanation is that the research questions may have been too simple and straightforward and focused only on diagnosis and symptomatology – this is a retrospective finding. As literature shows, non-psychiatric doctors reportedly have most difficulty with instituting treatment for mental illness and appropriate pharmacotherapy. Hence, there appears to be significant gaps in primary care and non-specialist doctors’ knowledge of mental illness with regard to interventions. Including questions on treatment and pharmacotherapy may have yielded different, albeit more representative, results.

Overall, no significant associations between sociodemographic variables and mental illness knowledge could be made in any of the study groups. Significant associations were made in other studies. Younger age,⁶⁶ engagement in mental health training not long before participation in a research study,^69,70 as well as being a family physician (compared to GPs or other specialists)⁶⁴ are associated with better diagnostic and practical knowledge. Aruna et al.²⁰ found that the level of knowledge in undergraduate medical students increased with progressive years of study, such that students in their final years had better mental health literacy than those in their first year of study. The general finding appears to be that doctors who engaged in some form of compulsory psychiatric training, over and above standard medical school teaching, have better knowledge of mental illness and are thus better able to manage psychiatric conditions.

It is noteworthy that positive attitudes towards mental illness appear to be associated with a better knowledge of mental illness and self-perceived competence in dealing with mental illness.^55,58

Negative attitudes appear to be associated with a lack of mental health training and a self-perceived lack of competence in managing mental illness.¹⁵ Also, better knowledge appears to be associated with a more positive attitude. Whilst it may not be possible to discern whether better knowledge is a consequence of positive attitudes or vice versa, it seems clear that the two share a significant association and both are likely to have an important contribution towards patients receiving adequate mental healthcare. Therefore, it stands to reason that in order to reach an end goal of improved healthcare for mentally ill patients, both doctors’ attitudes and mental health knowledge need to be adequately addressed.

Focusing on improving both factors (attitudes and mental health knowledge) is likely to lead to significantly better health outcomes for the mentally ill. Training of non-specialist doctors, focused on recognition and management of mental illnesses, may contribute to reducing the stigma of mentally ill patients within healthcare services. Consequently, this may lead to an increased willingness for families and patients to seek help, leading to earlier diagnosis, better recognition of illnesses and improved outcomes. A focus on fostering positive attitudes towards mental illness in non-specialist and primary healthcare doctors may make doctors more receptive to learning more about mental illnesses. This, in turn, may mean that doctors are willing to spend more time with mentally ill patients to determine their psychiatric symptoms and stressors and manage their illnesses better.