The term mental health literacy (MHL) was coined in the late 1990s by Anthony Jorm, and since then, it has been extensively researched in Australia and other developed countries.¹ Literacy in this context entails seven features: the capacity to recognise mental disorders; how to seek mental health information; awareness of mental health risk factors; appreciation of the aetiology of mental disorders; knowledge of self-treatment; knowledge of professional help available; and attitudes that promote recognition of appropriate health-seeking behaviour.² While the emerging MHL literature shows much promise for the mental health field, there are several challenges, especially regarding the measurement of MHL.

Most MHL measurements fail to simultaneously account for all the features of MHL and describe adequate psychometric properties.^3,4 Another challenge is the unknown status of the significant contextual contributions to MHL, such as culture, ethnicity, geographic location, as well as the educational background of its practitioners and their health status.⁵ As a remedy to address these unknowns, recent studies have attempted to bridge the gaps by developing strong psychometric measures that quantify MHL, assess its many components and target specific populations such as healthcare workers.

O’Connor and Casey⁴ developed the Mental Health Literacy Scale (MHLS), the first scale-based measure to assess all attributes of MHL that were found to have acceptable psychometric properties and can be easily administered and scored. Although MHL comprises the seven features mentioned above, O’Connor and Casey⁴ developed their scale to measure only six attributes, as feedback from a clinical panel suggested that, generally, there is a lack of knowledge to distinguish between risk factors for mental illness and its causes. These traits have therefore been integrated into one: knowledge of risk factors and causes. Although the MHLS was recently revised and content-validated in South Africa and Zambia by a diverse expert panel comprising primary healthcare (PHC) staff and research professionals,⁶ it has not been administered to a group of nursing students to assess its psychometric properties, as reported here.

As the PHC nurse is often the first contact with a patient, it is imperative that these healthcare workers have sufficient command of MHL. Unfortunately, Lahti and colleagues⁷ found that these workers’ recognition and screening of mental health problems at the PHC level is poor. Indeed, Dirwayi⁸ reported that the majority of nurses showed subtle negative attitudes toward people with mental illness. To improve the MHL of nurses, therefore, we need to be able to measure these attributes accurately. If we can identify deficits in MHL in nursing students early, we can adopt curricula in their training to empower them to fulfil this need; hence, the study reported here investigated the viability of using the revised MHLS scale in a group of nursing students in South Africa.

To date, there has been no investigation of the level of MHL among South African nursing students. Indeed, there have been only two studies that have validated the MHLS in a South African context,^6,9 thereby necessitating the need for its critical review in current circumstances. The aim of our study was to investigate the psychometric properties of the revised MHLS in a group of undergraduate nursing students and to assess the level of MHL in the group under investigation.

A total of 121 student nurses, whose characteristics are reported in Table 1, completed the online questionnaire, with no questionnaires excluded. The gender distribution of the participants was as follows: 92% (n = 111) were female, and 8% (n = 10) were male. Most of the participants (69%) were between the ages of 20 and 24. In terms of home language, 44% (n = 53) listed Afrikaans and 36% (n = 44) gave their home language as Setswana. Most of the participants (33%) were in their fourth year, followed by those in their first year (27%), their third year (21%) or their second year (19%). Over 82% (n = 100) of the students reported having less than 1 year of professional nursing experience; the remaining 17% (n = 21) reported a corresponding 1–5 years.

This study set out to investigate the usefulness of the MHLS in a group of nursing students in South Africa.

While Korhonen and colleagues⁹ reported that the revised version of the scale has a valid structure with strong internal consistency, our results do not support this finding. The identification of a 5-factor model, consisting of 17 items, provides a potentially more appropriate framework for understanding MHL in this context. These factors represent key domains, such as attitudes at the individual and societal level, information behaviour and recognition and treatment of mental illness.

The internal consistencies for mental health attitudes on a micro level, mental health information-seeking and sharing, and mental health attitudes on a macro level were all high; however, the internal consistencies for recognition of mental illness and treatment of mental illness were found to be very low and negative, respectively. It should be noted that Korhonen and colleagues⁹ had a total of 343 participants in their study, whereas our study had a total of 121 participants. Furthermore, Korhonen and colleagues⁹ used PCA, whereas we used not only EFA, including PCA, but also PAF, as well as CFA. Recurrent findings have indicated that PCA inflates factor loadings, changing the factor structure as a result, especially when there are few variables.^12,13 A PCA helps identify the possible number of underlying factors in a data set, whereas a PAF evaluates the item loadings on the number of factors specified (in this case, an oblique Geomin rotation was used). Neither of the two provides an actual model or specific fit statistics. In Mplus, the two steps are combined to provide all the necessary information in one step. From the EFA results, a CFA indicates whether a specifically specified model fits the data, what the relationships between factors look like (and if they exist), and where possible issues with the model or data could be. This step is not possible in SPSS, and AMOS can take long, intensive work; thus, Mplus is ideal for the CFA process.

Preliminary findings using the revised 5-factor model, as presented in this study, indicate that the sample of undergraduate nursing students had a satisfactory level of MHL. Crisp and colleagues¹² proposed that this may have been influenced by their greater exposure to university education in general, a conclusion that has been confirmed by Furnham and Hamid,¹³ who also found that education had a positive effect on MHL. The lower scores obtained specifically pertaining to information-seeking and treatment of mental illness are consistent with the literature; these nurses did not consider themselves to be sufficiently prepared to work with people with mental illnesses.^14,15,16 According to Sand-Jecklin,¹⁷ incorporating MHL education and skills training into nursing and health curricula would allow students to evaluate patients’ MHL and offer interventions or support to make sure the patients understand the value of health information and make the most effective use of their resources. These findings should be interpreted with some caution, though, as, because of the small sample size, the CFA had to be conducted on the same data as the EFA.

Although this was a small cross-sectional study, with a low response rate, which limited the generalisability of the results, its contribution in terms of the development of the MHLS to be useful in the education sector is clear. For future use, adjusting the Likert scales to use the same range throughout the questionnaire (e.g. by removing the neutral option from items 16–35) could strengthen the reliability of all the factors, especially the mental health treatment factor. Future research on the scale should focus on adapting or removing or adding applicable items through consultation with experts to improve content validity, and further clarifying the factor structure by conducting a larger-scale study, using EFA and CFA to investigate the validity of the structure.

Increasing the MHL of healthcare students could assist in integrating mental health treatment into PHC services, particularly in low- and middle-income countries.¹⁸ This study has shown that the revised MHLS is not valid for a group of nursing students and requires more research to confirm the 5-factor structure and improve its reliability. Furthermore, we propose the use of CFA in future studies to explain relationships between variables and to analyse factor structures.