Study design

A quantitative cross-sectional study design [14] was conducted to assess the socio-cognitive determinants of PPE use among DWCs (n = 344) in the Ho Municipality of Ghana. A cross-sectional survey was deemed appropriate given that data would be collected at a single time point without any follow-up. Data were collected via structured face-to-face administered questionnaires to assess PPE utilization patterns and associated risk factors.

Study setting and population

The research was carried out in the Ho Municipality of the Volta Region of Ghana. The Volta Region currently has 18 administrative districts, with the Ho Municipality serving as the administrative capital. The projected 2018 population of the Ho Municipality was 213,960 at the time of data collection, with 51 percent of the population being female [15]. The municipality is known as the commercial hub of the Volta Region due to its strategic location and proximity to the Republic of Togo, as well as easy access to district and regional capitals such as Accra. The municipality was chosen for the study because of its solid waste management issues and the state of the municipality’s environment, both of which need to be addressed significantly. Data were collected from the two waste management companies in the Ho Municipality (see reference [37] for more information on the waste management companies).

The study was conducted among DWCs in Ho Municipality of Ghana who worked for two solid waste companies. The waste management companies and respondents were kept anonymous (the two participating companies are known as Company A and B). Waste collection, waste processing, waste transportation, and waste disposal in designated dumping sites are all part of DWCs’ daily work schedules. The two solid waste companies analyzed in this study employ DWCs with notably low educational attainment and wages compared to other sectors in the region [25], a characteristic that may influence occupational health behaviors and risk exposure. The estimated study population consisted of 1,025 DWCs.

Inclusion and exclusion criteria

The specific inclusion criteria were that DWCs must work for at least one year and above to be eligible to take part in the study. DWCs who did not provide written consent and were ill or absent at the time of data collection were excluded from the study. In addition, respondents were restricted to adult DWCs aged ≥18 years, as Ghana’s labor laws prohibit formal employment in hazardous waste handling for minors. No respondents under 18 were enrolled, aligning with international ethical guidelines for occupational health research. Thus, workers less than 18 years, on medical/annual leave, or declining written consent were excluded.

Study instruments

The construction of the questionnaire was informed by a review of the literature on the determinants of taking protective measures by DWCs, previous qualitative research by the team of authors in the Ho municipality on health risks and preventive measures among DWCs and their managers and supervisors [25,37], and the theoretical framework of the HBM, and the RAA. The study utilized a structured questionnaire which was divided into seven (7) sections. The first section of the questionnaire included socio-demographic information such as gender, age, marital status, ethnicity, educational level, job roles, and the number of years at the workplace. These items were placed at the beginning of interviews on purpose to foster understanding between the research team and the study participants. The subsequent sections measured constructs of the HBM and RAA that used five-point Likert scales (1 = strongly disagree; 5 = strongly agree). These included measures of perceived susceptibility to occupational-related health problems; perceived severity of occupational-related health problems; perceived benefits of using PPE; perceived barriers to use PPE; identified cues to action; perceived self-efficacy of using PPE; attitude toward PPE use; subjective norms towards PPE use; behavioral beliefs (i.e., normative beliefs are individuals’ beliefs about the extent to which other people who are important to them think they should, or should not perform particular behaviors) towards PPE use, and intention to use PPE (for the full questionnaire, see S1 Appendix). Intention was the principal outcome variable; socio‑cognitive constructs served as independent variables.

Quality control measures

To ensure construct and face validity, the questionnaires were developed in English, reviewed by a panel of 5 experts (occupational health specialists, and waste management professionals) to assess relevance and clarity using a 4-point Likert scale (CVI = 0.82), translated into Ewe, then translated back into English. Discrepancies between the two first and second English versions of the questionnaire were resolved through discussion among the research team. Two (2) research assistants and two (2) interpreters who are natives of the study setting with prior knowledge of waste management practices were trained to assist in all field data collection, standardized question delivery, ethical protocols (e.g., Informed consent) data recording, and also to minimize interviewer bias, in addition to mock interviews with feedback. A pilot study was done among 50 participants with similar socioeconomic backgrounds to the respondents of the study population to test for validity and reliability from 01/05/2018 to 15/05/2018. This was done using the research instrument developed for the study to assess how easy it would be for respondents to understand the questions. Except for minor textual changes, no content changes were made to the instrument.

The data for this study were collected through researcher-administered, face-to-face interviews using structured questionnaires. This approach was chosen to ensure clarity and completeness for trained research assistants to verbally explain questions to DWCs, address ambiguities in real-time, and minimize missing responses. Also, to account for literacy variability given potential variations in literacy levels among DWCs, verbal administration improved accessibility and data quality. In addition, to enhance contextual validity to allow face-to-face interactions with the researchers to observe non-verbal cues and cross-check responses (e.g., verifying self-reported PPE use with visual inspections where possible). However, the rationale for the research team not using Self-Administered/Online Methods were due to low feasibility with regards to limited smartphone/internet access among the target population (DWCs) made tools like Google Forms impractical. Moreover, the issue of research team engagement in face-to-face interactions with DWCs fostered trust and higher participation rates compared to self-administered methods.

Sampling and data collection procedure

The study’s field survey was from May to July 2018 covering typical working hours for DWCs in the Ho Municipality (e.g., 5:00 am to 10:00 am), coinciding with their active waste collection schedules. This timing ensured that DWCs were available and engaged in their routine tasks, enhancing the ecological validity of responses regarding PPE use. The survey instruments (e.g., questionnaires, and observational checklists) were administered on-site during short breaks or upon completion of shifts to minimize disruption in the work. Each interview/questionnaire session lasted approximately 10–15 minutes, while direct observations spanned 2–3 hours per work session to assess real-time PPE compliance. Adjustments were made for DWCs who opted to participate before or after shifts to accommodate their preferences.

A multi-stage sampling procedure was used to select the sample [16]. In the first stage, we divided the Ho Municipality into five (5) zones purposely after visiting the area and consulting with the local Municipal Assembly and regulatory bodies such as Environmental Health inspectors. In the second stage, 20 sub-zones (four from each Zone) were chosen at random, and finally, DWCs were chosen at random from a list submitted by solid waste company managers and supervisors [37]. The use of a multi-stage sampling procedure allows practically for primary data collection for large study respondents that are geographically dispersed, reduces the costs and time associated with data collection, and provides flexibility, for the research team to break down the participants as often as necessary to create the sample population needed.

Using Cochran’s single-population proportion formula, the preliminary requirement was 384 respondents [39]. Applying the finite-population correction for N = 1,025 produced n = 278. To offset an anticipated 30% non-response, the target was inflated to 400 (278/ 0.70). Ultimately, 344 fully completed questionnaires were returned, yielding an 86% response rate; only these complete cases were included in the analyses.

Data analyses

SPSS version 25 and SmartPLS 3.3.9 [40] were used for the data analysis. Only fully completed questionnaires (n = 344) were retained for analysis. In the analysis, descriptive statistics of the study respondents such as socio-demographic characteristics were generated into percentages and frequencies as demonstrated in Table 1. Partial least squares structural equation modeling (PLS-SEM) procedure, which has the predictive ability to evaluate several regression models or equations concurrently [26], was then used to examine the study model and its paths [27]. Thus, the measurement model including the reliability and validity of measures was examined using the PLS algorithm with default settings. Internal consistency was satisfactory: Cronbach’s α ranged from 0.673–0.906, composite reliability from 0.815 – to 0.934, and average variance extracted (AVE) from 0.511 to 0.779. Next, the structural model was evaluated using the bootstrapping technique with 5000 resamples, and finally the blindfolding procedure [17,27]. The figure of 5,000 resamples was adopted in line with Hair et al.’s [27] guidance that at least 5,000 subsamples are necessary for stable estimates and resamples were generated by random sampling with replacement within SmartPLS [40]. Before bootstrapping, univariate normality was evaluated using the Shapiro–Wilk test; most indicators departed from normality (p < .05). This justified the use of PLS-SEM, which does not assume multivariate normality.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographics characteristics of respondents(n = 344).

https://doi.org/10.1371/journal.pone.0334542.t001

Ethical approval

The study was approved by The Ethical Review Committee Psychology and Neuroscience (ERCPN) of the Faculty of Psychology and Neuroscience, Maastricht University (approved research line ERCPN-188_10_02_2018), and the Ghana Health Service Ethics Review Committee (GHSERC 08/05/17). Permission was also obtained from the waste management companies from which the respondents were recruited. The first page of the questionnaire provided information about the project, the survey, and the voluntary nature of their participation. The description outlined the project’s goal and the risks associated with not wearing PPE while emphasizing the anonymity and confidentiality of participants. In addition, trained research assistants explained the project and the survey to participants, who were informed that they could withdraw from the study at any moment without penalty. The respondents provided written informed consent before participating in the study. Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research is included in the S2 Appendix.

Socio-demographic profile of the respondents

The final sample consisted of 344 DWCs. Close to half of the participants (n = 157, 45.6%) were aged < 20 y (M = 22.18, SD = 11.09; range = 18–64 years). Most participants were male (n = 175, 50.9%), identified as Ewe (n = 292, 84.9%), professed Christianity (n = 293, 85.2%), and were married (n = 204, 59.8%).

Furthermore, most respondents were DWCs (83.2%; n = 175) and achieved a junior high school education (n = 125, 36.5%). Lastly, most had 1–5 years of experience (n = 167, 48.8%), and belonged to Company A (n = 329, 95.6%).

Measurement Model Assessment (Reliability and Validity Results)

This section reports the reliability and validity of the construct measures as advocated by [27]. The item loadings, VIF, construct reliability coefficients, and convergent validity (AVE) reported in Table 2 prove that all construct indicators have significant factor loadings (p < 0.001) with no multicollinearity issues (VIF < 5), and all the latent variables exhibit adequate reliability and convergent validity as the values of composite reliability (CR), and the average variance extracted (AVE) for each latent variable are greater than the minimum threshold limit of 0.70 and 0.50, respectively. Thus, there is acceptable reliability and convergent validity of the constructs [27,35,36]. Further, to check the discriminant validity of the constructs, the Heterotrait-Monotrait ratio (HTMT) criterion was used as reported in Table 3. The HTMT values are less than 0.90 as recommended [41,42], thus, confirming the presence of constructs’ discriminant validity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Construct reliability and convergent validity results.

https://doi.org/10.1371/journal.pone.0334542.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Discriminant validity using Heterotrait-Monotrait Ratio of Correlations (HTMT) Criterion.

https://doi.org/10.1371/journal.pone.0334542.t003

The results in Table 4 demonstrate that all the predictors have significant positive correlations with intention with r = .10−.23 signifying a small effect, r = .24−.36 demonstrating a moderate effect, and r ≥ .37 signifying a large effect [43]. Specifically, we found strong positive univariate associations with the intention for attitude, susceptibility, behavioral beliefs, subjective norm, perceived benefits, cues to action, self-efficacy, and management. Severity and perceived barriers had moderate positive associations with intentions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Descriptive statistics and correlational matrix.

https://doi.org/10.1371/journal.pone.0334542.t004

Predictors of intention to use PPE (Structural model assessment)

Succeeding the fulfillment of the reliability and validity of the measurement model, the structural model was examined including the structural relationships, model’s predictive power, and relevance [27,44]. As revealed in the model’s summary section of Table 5 and Fig 1, the beliefs, susceptibility, severity and perceived benefit explained 37.4% of the variance in attitude (Q²= 0.261), and attitude, cues to action, management, self-efficacy, subjective norm accounted for 66.6% of the variation in intention (Q²= 0.424). Lastly, perceived barriers contribute to 14.2% of the variance in self-efficacy (Q²= 0.074). Thus, the structural model exhibits good predictive power, accuracy, and relevance [27,45–47].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Structural model results.

https://doi.org/10.1371/journal.pone.0334542.t005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Structural model of factors that affect the use of PPE.

https://doi.org/10.1371/journal.pone.0334542.g001

Furthermore, the complete bootstrapping procedure using 5,000 resamples with no significant changes revealed that beliefs (β = 0.342; p = 0.000; BCa 95% CI [0.224; 0.450]), severity (β = 0.244; p = 0.000; BCa 95% CI [0.141; 0.340]), and perceived benefits (β = 0.209; p = 0.002; BCa 95%CI [0.074; 0.340]) had significant positive direct effects on attitude. However, susceptibility (p = 0.839) failed to have significant direct influence on attitude.

Regarding the predictors of intention, intention was significantly positively and directly influenced by attitude (β = 0.174; p = 0.001; BCa 95%CI [0.070; 0.278]), cues to action (β = 0.500; p = 0.000; BCa 95%CI [0.392; 0.602]), management (β = 0.114; p = 0.013; BCa 95%CI [0.028; 0.212]), and self-efficacy (β = 0.199; p = 0.002; BCa 95%CI [0.076; 0.325]). But, the effect of subjective norm on intention was not significant (p = 0.396). Lastly, as expected, the barriers perceived had a significant positive direct effect on self-efficacy (β = 0.377; p = 0.000; BCa 95%CI [0.264; 0.449]).

Further examination of the paths in Table 5 revealed three significant indirect effects. Specifically, beliefs (β = 0.060; p = 0.002; BCa 95% CI [0.026; 0.104]), severity (β = 0.042; p = 0.004; BCa 95% CI [0.018; 0.077]) and perceived benefit (β = 0.036; p = 0.065; BCa 95% CI [0.009; 0.087]) had significant indirect effects on intention through attitude. Thus, attitude mediated the influence of beliefs, severity, and perceived benefit on intention. Similarly, self-efficacy mediated the effect of perceived barriers on intention (β = 0.075; p = 0.010; BCa 95%CI [0.023; 0.131]).

Theoretical implications

Our findings underscore the value of integrating the HBM and RAA frameworks – along with organizational factors to advance the theoretical understanding of PPE use behavior. The combined model demonstrated that both HBM constructs (e.g., perceived severity, benefits, barriers, cues to action) and RAA constructs (e.g., attitude, self-efficacy as a proxy for perceived behavioral control, subjective norm) jointly explain a substantial proportion of variance (67%) in DWCs’ intention to use PPE. This high explanatory power suggests that a multi-theory approach can capture a wider range of socio-cognitive influences on protective behavior than any single theory alone. For instance, risk perception variables from HBM (such as the perceived likelihood and seriousness of injury) were found to shape workers’ attitudes toward PPE, while behavioral belief-based attitude and self-efficacy from RAA directly drove intention. This indicates that both the perception of risk and the evaluation of PPE’s benefits/feasibility are critical motivators. Notably, our study showed that perceived management rule enforcement, an external contextual factor, had a significant positive effect on PPE use intention (comparable in magnitude to personal attitudinal factors). The inclusion of this managerial enforcement element – conceptually akin to a strong cue to action or an environmental facilitator – extends traditional behavior models by acknowledging that institutional safety rules and their enforcement can directly motivate worker compliance. This theoretical extension is particularly important in occupational settings where formal regulations and supervision play a role in shaping individual behavior, adding a new dimension to HBM and RAA in the context of workplace safety.

Comparing our results with prior studies further highlights the theoretical contributions. In line with our findings, a study of factory workers in Ethiopia reported that higher perceived susceptibility and severity of hazards, greater self-efficacy, and fewer perceived barriers were associated with better PPE usage. Similarly, research using HBM in an agricultural context (Iran) and safety behavior studies in China have observed that internal drives (like risk awareness and confidence) and external triggers can jointly influence protective behavior. These parallels across diverse contexts suggest that our integrative socio-cognitive model has broad applicability in occupational health theory, reinforcing the generalizability of HBM/RAA constructs for PPE compliance. However, our results also revealed a nuanced divergence: subjective norms did not significantly predict PPE intention in our sample, whereas some studies in Asia did find social normative influences to be important. This discrepancy implies that the role of peer influence may depend on context – for Ghanaian waste collectors, coworkers’ opinions or example might be less salient than personal conviction and management directives. From a theoretical standpoint, this finding invites refinement of the RAA/TPB normative component for similar settings, perhaps by distinguishing between injunctive norms (what others think one should do) and descriptive norms (what others actually do) or considering the effect of direct peer pressure. In sum, the integration of HBM and RAA, enriched with an organizational enforcement factor, advances our theoretical understanding by providing a more comprehensive model of PPE-use behavior. It demonstrates the importance of multi-level determinants – individual cognitive factors and institutional cues – and sets the stage for future occupational safety research to adopt and further test such combined theoretical frameworks in explaining and predicting protective behaviors in various work environments.

Practical implications

The study’s findings translate into actionable insights for public health officials, waste-management companies, and safety training providers aiming to improve PPE use among DWCs. First, education and training interventions should be prioritized to address the identified socio-cognitive gaps. Because positive attitude toward PPE was a strong predictor of intent, and attitude in turn was shaped by risk perceptions and beliefs about benefits, training programs must emphasize hazard awareness and the tangible benefits of consistent PPE use. Public health and safety trainers can develop targeted modules that vividly communicate the severity and susceptibility of work-related injuries (for example, sharing local case studies of injuries prevented by proper PPE) to increase risk perception and personal relevance. Such trainings should also highlight the protective advantages of gloves, masks, boots, and other gear in preventing disease or injury, thereby fostering more positive outcome expectations. Indeed, previous studies in Ghana and other low-resource settings have shown that tailored PPE education and safety training significantly improve workers’ PPE usage rates. These educational efforts will not only improve knowledge but can serve as continuous cues to action reinforcing the habit of PPE use through reminders, demonstrations, and visual signage at the workplace.

Second, the role of management is crucial: waste-management companies should strengthen enforcement of PPE policies and ensure supportive working conditions. Our findings that perceived rule enforcement by management is a significant motivator for PPE use indicate that stricter supervision and clear organizational safety rules can drive behavior change. Company managers and supervisors need to consistently enforce PPE requirements – for instance, by mandating PPE at the start of each shift, conducting spot checks, and instituting accountability measures for non-compliance. Importantly, enforcement should be paired with facilitation: firms must provide adequate PPE supplies (of the right sizes and comfort) to all workers at no cost. Prior research in Ghana revealed that many waste workers lack or avoid PPE due to cost, discomfort, and insufficient supply from employers. Addressing these barriers through budget allocation for PPE procurement and replacement will improve workers’ self-efficacy and willingness to use the equipment. Additionally, establishing a safety culture where supervisors lead by example (wearing PPE themselves and encouraging peer support for compliance) could compensate for the weaker influence of subjective norms in this context. Public health authorities and labor regulators also have a part to play by setting and enforcing industry-wide standards – for example, integrating waste sector PPE compliance into municipal contracts or labor inspections. Clear policy guidelines (such as municipal bylaws or company-level Standard Operating Procedures) that mandate PPE use, coupled with training on their importance, will create an environment where wearing protective gear is the default expectation.

Finally, these implications extend to other low-resource occupational settings beyond urban waste collection. In many developing regions, workers face similar challenges of limited resources, low risk awareness, and lax enforcement of safety rules. The approaches indicated by our study – enhanced hazard communication, empowerment of workers through training, and diligent managerial enforcement of safety protocols – are broadly applicable to improve occupational health outcomes where PPE compliance is suboptimal. Interventions built on the socio-cognitive determinants (like shaping attitudes and boosting self-efficacy) can often be implemented with relatively low cost but high impact, making them ideal for resource-constrained settings. By combining behavior-change education with organizational policy action, stakeholders can significantly raise PPE usage and protect workers’ health even in the face of economic limitations. In short, public health officials, company managers, and safety trainers should collaborate to translate these findings into practice: develop comprehensive PPE programs that integrate training, consistent rule enforcement, and ongoing risk communication. Such multi-faceted efforts, informed by our study’s evidence, will improve safety behavior among DWCs and can be adapted to enhance occupational safety across similar high-risk industries.

Strengths and limitations

This research contributes to the literature by examining PPE use determinants in an understudied population (DWCs in Ghana) using a comprehensive theoretical approach. A key strength of our study is the integration of constructs from both the HBM and RAA, which allowed us to capture a wide range of influences on PPE use intention. This integrated model explained a high proportion of variance in intention, suggesting it can offer richer explanatory power than single-theory models. However, the use of an extensive combined framework also introduced certain challenges. The survey instrument was quite lengthy (107 items in total), raising the potential for respondent fatigue or reduced data quality. Future studies might streamline the measurement of similar constructs to minimize burden. In addition, our cross-sectional study design limits causal inference – we can identify associations but cannot establish the direction of cause and effect between socio-cognitive factors and PPE use intentions. Longitudinal studies or interventions would be useful to determine how changes in these determinants lead to changes in PPE use behavior over time. All data were self-reported, so responses may be subject to biases such as social desirability or recall bias. This means that some DWCs might have overestimated their intentions or provided socially acceptable answers regarding safety practices. Another limitation is the generalizability of our findings. The study sample was drawn from a single municipality, and the majority of participants shared similar backgrounds within that region. As a result, the results may not fully represent DWCs in other parts of Ghana or in different countries. Finally, while our integrated theoretical model proved useful, many of the significant predictors aligned with well-established HBM components (e.g., risk perceptions, perceived benefits, self-efficacy). This overlap suggests that the added value of incorporating all RAA elements (such as subjective norms) should be interpreted with caution. In contexts like ours where peer norms showed little effect, a more parsimonious model focusing on threat appraisal and efficacy beliefs might perform nearly as well. Nevertheless, using the combined approach provided a holistic view and helped confirm that multiple socio-cognitive pathways – from individual threat appraisal to organizational cues – jointly influence DWCs’ motivation to use PPE.