Study design and location

This is a cross-sectional study, conducted among public primary HCWs in the State of Negeri Sembilan. Negeri Sembilan is situated in the western region of Peninsular Malaysia and covers an area of 6,658 km² with a population of approximately 1.13 million [30]. The location was chosen in accordance to the National Health Morbidity Survey 2019, Negeri Sembilan was found to be the top three states having the highest prevalence of depression among adult population [31]. Negeri Sembilan State Health Office oversees seven district health offices and 51 health clinics. One district health office was designated for this pilot study, while the other remaining district health offices and clinics participated in the study.

Inclusion and exclusion criteria

Inclusion Criteria:

1. (1) HCWs working in district health offices and health clinics in Negeri Sembilan
2. (2) Four job categories including (i) doctors, (ii) nurses, (iii) assistant medical officers, (iv) assistant environment health officers and environment health officers.

Exclusion Criteria:

1. (1) HCWs with working experience of fewer than three months,
2. (2) those on maternity leave during the study period and
3. (3) HCWs who refuse to participate in the study.

Sample size

The sample size was calculated using two independent proportion formula by Lwanga and Lemeshow 1991 [32]. The reason of using the largest sample size possible and with the most feasible sample size in conducting the research as it is within the achievable time frame. The sample size was chosen based on a study by Tian et al. (2021). P1 is the proportion of respondents with depressive symptoms who have used mental health services (20.2%), whereas P2 is those who did not have depressive symptoms but have used mental health services (8.7%) [4]. Based on the calculation, sample size estimation (n) is 146. The total sample size required is 352 after the adjustment of comparison of two groups and 20% non-response rate.

Sampling method

All six district health offices were included in the study, with the Port Dickson District Health Office reserved for the pilot study. Sampling frames were obtained from Negeri Sembilan State Health Department and all the district health offices. The required samples were selected using proportionate stratified random sampling with job categories served as strata. The strata consist of four job categories which are (i) doctor, (ii) nurse, (iii) assistant medical officer, and (iv) assistant environment health officer & environment health officer. This method was chosen due to the heterogeneity of the study sample. The number of samples required in each stratum is derived by calculating the entire sample size divided by the total population in sampling frame times with the total numbers of HCW in each stratum. The selection of eligible respondents from each stratum was done using simple random sampling. Identification of the data number was assigned to each respondent in the sampling population and random numbers were generated using Microsoft Excel.

Study instruments and quality control

A Malay Language (Bahasa Malaysia) self-administered paper-based questionnaire was used to reduce the non-response rates, as demonstrated in a previous study [33]. The questionnaires consist of six sections. The first and second sections include self-constructed questions on sociodemographic factors like age, gender, history of mental illness, and a work-related section consisting of place of work, job category, working hours, overnight work, and mental health training. The third section consists of questions on MHSU adapted from Wang et al. (2007) [34]. Participants were asked if they had used mental health services for screening, counselling, treatment, and rehabilitation in the past 12 months. If yes, they were asked about the purpose, frequency, type of service, and professional met. If no, they were asked about the main reasons for non-use. Forth section measures “Stigma by Others”, using “Perception of Stigmatisation by Others for Seeking Help (PSOSH)” tool, developed by Vogel et. al (2009), translated and validated into Malay version with a good Cronbach’s alpha value of 0.84 [35]. PSOSH scale consists of 5 questions, measured on 5-point Likert scale and total score ranges from 5 to 25. The fifth section measures the enabling factors, such as household income per month, awareness of mental health services, co-workers, and supervisor support. The co-worker and supervisor support were adapted from Malay Language version Job Content Questionnaire (JCQ) by Amin et al. (2015). It has good internal consistency with Cronbach alpha value of 0.83 [36]. Supervisor support and co-worker support consist of four questions in each domain, measured on 4-point Likert scale with total scores ranging from 4 to 16. The sixth section comprised need factors including self-rated health and mental health status. Self-rated health is a single question, measured using 5-point Likert scale ranging from 1–5. The question asked respondents to rate their general health in the past two weeks. The scales 1–2 were categorised as good and 3–5 were categorised as poor self-rated health status [37]. Mental health status was measured using Malay Depression, Anxiety and Stress 21-item (DASS-21) questionnaire. It consisted of 21 questions, each question using 4- point Likert scale (from 0 to 3). A higher score indicates higher level of the respective subscale’s symptoms. This tool was adopted and validated by Musa et al. (2007) with acceptable to good Cronbach’s alphas values 0.84, 0.74, and 0.79 for depression, anxiety, and stress, respectively [38]. Scores of more than 5 were interpreted as “yes for depression”, more than 4 as “yes for anxiety”, and more than 7 as “yes for stress”, respectively, based on the items of each subdomain [39].

Before the actual study, the face and content validity were evaluated by four public health physicians and a consultant psychiatrist. The details of the responses to the questionnaire were thoroughly evaluated and revised accordingly. All 38 appropriate items were retained based on their content validity index. For the reliability test, a pilot study was performed on 30 HCWs at the Port Dickson Health Office in Negeri Sembilan, who were not sampled in the actual study. The overall result demonstrated good to excellent internal consistency with Cronbach alpha’s values 0.81 for stress, 0.78 for anxiety, 0.86 for depression, 0.92 for supervisor support, 0.89 for co-worker’s support and 0.93 for PSOSH scale. The MHSU questions also demonstrated an excellent test-retest reliability with Cohen’s kappa = 1. It is imperative to verify that the DASS-21, PSOSH scale, and JCQ scale were reliable instruments for evaluating depression, anxiety, and stress, stigma, supervisor and coworker support respectively. This validation helps to account for cultural, demographic, and contextual disparities within the study population that may influence the interpretation. Furthermore, MHSU questions signifies perfect test-retest reliability assessment and confirm its stability and consistency in measuring the utilisation of mental health services over time.

Data collection method

Data was collected from eligible HCWs from December 2022 to April 2023. Eligible HCWs were approached individually at their workplace, with their approved consent, and personal information sheet before eventually answering the questionnaire. The questionnaires were completed in 20 minutes, and the researcher thoroughly checked for missing information upon submission.

Data analysis

Data was analysed using IBM Statistical Package for Social Sciences version 26. Descriptive statistics were presented in frequency and percentage for the categorical data, mean and standard deviation or median with inter-quartile range for continuous data. This study utilised Multiple Logistic Regression (MLR) to determine predictors of MHSU. Variables that demonstrated a p-value of less than 0.25 in simple logistic regression were selected for inclusion in the MLR model [40]. For determining statistical significance in the MLR, a p-value threshold of 0.05 and a 95% confidence interval (CI) were applied to all tests.

Ethical consideration

The study received ethical approval from the National Medical Research Register and the Medical Research and Ethics Committee of the Ministry of Health, Malaysia with approval number NMRR-ID-22-02258-CTA (IIR). Following this approval, formal notifications were sent to Negeri Sembilan State Health Office, and all participating District Health Offices. Participants were provided with an information sheet and were required to sign a written informed consent form prior to participation. Throughout the study, the researcher adhered to the Declaration of Helsinki and Malaysian Good Clinical Practice Guidelines.

Prevalence of mental health service utilisation (MHSU)

A total of 294 out of 352 respondents completed the self-administered questionnaire, representing an 83.5% response rate. The 12-month prevalence of MHSU was 45.6%. Among those utilising mental health service, majority used it once (91.0%), for screening (96.3%) and treatment purposes (28.4%), primarily accessed through health clinics (85.1%), and involved interaction with paramedics (44.0%) and medical officers (38.8%).

Conversely, 54.4% of respondents did not use mental health services. The main reasons cited were no mental health problems (76.3%), problems not severe enough (26.9%), not enough time (7.5%), privacy concerns (5.6%), and others (see Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics and prevalence of MHSU (N = 294).

https://doi.org/10.1371/journal.pone.0317654.t001

Characteristics of respondents

Table 2 shows the characteristics of respondents based on predisposing, enabling and need factors. For predisposing factors, the median age of respondents was 35 (IQR = 9), and the median years of service were 11 years (IQR = 9). Most respondents were female (73.5%), married (82.7%) and had no history of mental illness (94.9%). For the job characteristics, 50.3% of respondents were nurses, 23.5% doctors, 16.3% assistant medical officers, and 9.9% assistant environmental health officers and environmental health officers. Most of respondent worked at health clinics (89.5%) and did not receive mental health training in the past 12 months (66.0%). The median day of overnight work in previous week was zero (IQR = 0), median of average working hours was 45 hours per week (IQR = 5) and median score for stigmatisation by others for seeking help was 9 (IQR = 8). For the enabling factors, most of them are from M40 household income (54.4%), majority aware of available mental health services (99.7%) and most of respondents had received high supervisor and high co-worker support with 81.3% and 57.5%, respectively. In this study, household income is defined as the respondents’ monthly gross household income. Household incomes were categorised into three groups based on gross household income in Negeri Sembilan: less than RM 4,210 for B40, RM 4,210 to RM 9,299 for M40, and RM 9,300 or more for T20 [41]. As for the need factors, most respondents were in a good self-rated health status (82.7%). In relation to the mental health status using DASS-21 tool, the highest prevalence was anxiety (15%), followed by depression (11.6%) and stress (10.9%).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of respondents according to predisposing, enabling and need factors (N = 294).

https://doi.org/10.1371/journal.pone.0317654.t002

Predictors of MHSU

Seven variables with p-value <0.25 from simple logistic regression (SLR) were tested in multiple logistic regression (MLR) namely, mental health training (p-value = 0.020), job category (p-value = 0.073), household income (p-value = 0.001), average working hours (p-value = 0.171), supervisor support (p-value = 0.075), co-worker support (p-value = 0.034) and self-rated health status (p-value = 0.246). It has been suggested that the variables with p-value<0.25 in the univariate analysis and clinically important variables should be included in the multivariate analysis [40]. This is due to the weak association with the outcome in univariate analysis that can contribute significantly when they are combined.

Table 3 shows four significant predictors for MHSU in the final model, using forward LR variable selection method. The model assumption was checked. The overall classification percentage was 63.3%. No multicollinearity between the variables. The model had significant model fit (Hosmer and Lemeshow test, p-value = 0.373) with Cox and Snell R squared and Nagelkerke R squared were 10.3% and 13.8% respectively. Based on the Receiver Operating Characteristic (ROC) curve, the model was significantly discriminated in 68.7% of cases (ROC = 0.687, p-value<0.001).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Predictors of MHSU using multiple logistic regression analysis (N = 294).

https://doi.org/10.1371/journal.pone.0317654.t003

The result showed that respondents with household income in B40 and M40 groups had 3.426 times and 3.781 times more likely to utilise mental health services as compared to T20 groups, respectively (aOR = 3.426, 95% CI: 1.588, 7.393, p-value = 0.002; aOR = 3.781, 95% CI: 1.916, 7.460, p-value<0.001). Respondents with low supervisor support were 2.302 times more likely to utilise mental health services as compared to those who received high supervisor support (aOR = 2.302, 95% CI: 1.206, 4.392, p-value = 0.011). Respondents who received mental health training were 2.058 times more likely to utilise mental health services as compared to those who did not receive mental health training (aOR = 2.058, 95% CI: 1.221, 3.469, p-value = 0.007). Respondents with high co-worker support were 1.701 times more likely to utilise mental health services as compared to those who received low co-worker support (aOR = 1.701, 95% CI: 1.034, 2.798, p-value = 0.03).

Mental health service utilisation (MHSU)

To our knowledge, this study represents the inaugural attempt in Malaysia to determine the prevalence and predictors of MHSU among public primary HCWs. The current study found that 45.6% of public primary HCWs in Negeri Sembilan had MHSU in the past 12 months. The prevalence is notably high as compared to lifetime MHSU by nurses in China (10.9%) and MHSU for treatment among frontline HCWs during COVID-19 pandemic in United States (19.1%) [4, 8]. The high prevalence of service utilisation in our study was attributed to screening purposes, which accounted for 96.3% of cases among those who utilised mental health services. Studies in China and United States did not include MHSU for screening. Notwithstanding this, our study observed a relatively high prevalence of MHSU for treatment at 28.5%, in comparison to prior studies [4, 8]. Mental health screening is an essential component of preventive healthcare as it enables the prompt detection of mental health issues. It facilitates treatment access, safeguards confidentiality, and encourages the normalisation of mental health discussions, thereby reducing stigma. Furthermore, the implementation of the KOSPEN WOW (Wellness of Workers) by Malaysian Ministry of Health since 2016 had increased accessibility and encouraged MHSU among HCWs [42]. This program provides mental health training, voluntary mental health screenings, and a connection to active intervention [43]. This is consistent with the findings of a systematic review and meta-analysis on mental health screening in the workplace, which concluded that screening followed by appropriate treatment interventions had a more beneficial impact on the mental health of employees [28].

Moreover, the COVID-19 pandemic acted as a catalyst for a significant surge in awareness pertaining to mental health, resulting in heightened levels of self-awareness and advocacy. These developments in turn encouraged HCWs to participate in screening activities and seek treatment. Significant efforts have been made to enhance mental health services in Malaysia during the COVID-19 pandemic, such as digitalising mental health screening, creating of national helpline service “Talian HEAL” and establishment of National Centre of Mental Health Excellence to coordinate mental health services across public, private, and non-governmental organisations [44].

Mental health training

In the current study, mental health training emerged as the only significant predictor for predisposing domain from ABMHSU. This finding is aligned with a cross-sectional study among public health workers in China demonstrated that psychological training enhances the likelihood of seeking professional help during the COVID-19 pandemic. The beneficial impact of mental health training enhances knowledge, and awareness, fostering positive attitudes and reducing the stigma towards MHSU. This is explained by a mixed-method study among doctors in Australia, which emphasized that the recognition and awareness of symptoms of mental health problems are required to prompt doctors to seek help [45]. Besides, the positive view about mental health has encouraged NHS doctors to utilise mental health services [23]. Similarly, a qualitative study by Spiers et al. (2017) mentioned how openness and low stigma could encourage MHSU. The willingness to discuss mental health issues and how they benefited from seeking help encourages others with similar issues to seek professional help [22]. Therefore, it is recommended that regular mental health training integrated with screening programs presents a golden opportunity to enhance MHSU and optimise mental health outcomes among HCWs.

Household income

Three significant predictors identified in the enabler domain of ABMHSU encompassing the M40 and B40 household income group, high co-worker support and low supervisor support. A strong predictor of MHSU is the B40 and M40 household income. There is a growing chance that people from lower and middle household incomes use mental health services more frequently because of financial stress, lead to mental health problems and raises the need for such services. A systematic review reveals that lower household-income employees are more likely to have poor mental health outcomes such as depression, anxiety, stress and psychological distress [46]. A nationwide survey of Malaysia’s adult population supports this statement, finding that the proportion of depressed people in the B40 (2.7%) and M40 (1.7%) groups was higher compared to T20 groups (0.5%) [47]. In addition, the majority of respondents (85.1%) reported that they had utilised mental health services at public health clinics. Public mental health services in Malaysia are provided free of charge to all HCWs by government-affiliated entities. Additionally, they are familiar with the system and have easy access to internal services. This is supported by a cross-sectional study by 7846 Australian HCWs showing that nurses and allied health workers are more likely to utilise internal mental health services [48]. However, the findings contrast with the mixed method study findings among pharmacists in the United States which found that low income is a barrier to utilising mental health services as the informant explained that adequate finances are important for sustainable follow-up and treatment [19].

Supervisor support

Another factor that predicts MHSU in public primary HCWs is low supervisor support. This can be explained by low supervisor support in terms of lack of emotional and practical support as well as poor relationship with supervisor that can lead to the increased risk of poor mental health outcomes. The findings of a systematic review and meta-analysis of 22 studies demonstrated that receiving low supervisor support was associated with 1.45 and 1.16 times higher risk of suicidal ideation and suicide, respectively [49]. Additionally, interpersonal conflicts at work, such as violence and workplace bullying, have been shown to elevate the risk of stress, anxiety, and depression [50]. Furthermore, an effort-reward imbalance, where high effort at work is coupled with low appreciation, has been linked to an increased risk of depression [51]. These factors could indirectly contribute to an increase in MHSU. Besides, the Job Demand-Resource Model elucidates the role of supervisor support in the workplace. Specifically, it indicates that low supervisor support serves as a risk factor, exacerbating the negative effects of job demands and increased workplace stressors [52].

However, this finding differs from a qualitative study among NHS doctors, which discovered that supportive supervisors can encourage help-seeking behaviour due to their elevated status within the hierarchy, which allows them to successfully encourage junior doctors to use mental health services [23]. In this study, the low supervisor support acts as a stressor, thereby influencing MHSU differently. The distant relationship and poor supervisor support would impede open conversation and result in reluctant to interact with supervisor [53]. As a result, HCWs opt to communicate with their colleagues, who link them to mental health services or they directly used other preferable mental health [45, 54, 55]

Co-worker support

Another significant predictor for MHSU is high co-worker support. The result aligns with the qualitative studies among doctors which explained that informal conversations between colleagues are able to facilitate access to mental health services [54, 55]. Most doctors prefer informal consultation and are likely to contact friends, colleagues and family who are also doctors if they have mental health problems or difficulties [55]. Another qualitative study among doctors in Norway mentioned that co-worker support provides psychological safety, trust, time flexibility, informal talk and raising awareness to seek help and MHSU [54]. Furthermore, most HCWs are likely to turn to their co-workers when encountering difficulties [56, 57]. Considering these findings, it is critical to develop robust peer support networks to promote informal support mechanisms as essential elements of MHSU and to create a supportive work environment.