Effectiveness of a Multifaceted Intervention (TEMP) for Prevention of Occupational Heat-Related Illness among Outdoor Workers in the Power Grid Industry: A Cluster Randomized Controlled Trial

Abstract

Background: Occupational heat-related illness (OHI) is a health threat to workers that can be fatal in severe cases. Effective and feasible measures are urgently needed to prevent OHI. Objectives: We evaluated the effectiveness of a multifaceted intervention, TEMP, in reducing the OHI risk among outdoor workers in the power grid industry. Methods: A cluster randomized controlled trial was conducted with power grid outdoor workers in Southern China from 4 July 2022 to 28 August 2022. Work groups were randomly allocated (1:1) to the intervention or control groups. The multifaceted intervention TEMP comprised mobile application (app)-based education training (T), personal protective equipment [PPE (E)], OHI risk monitoring (M), and educational posters (P). Four follow-ups were conducted every 2 wk after the trial began. The primary outcome was the OHI incidence, and the secondary outcome was PPE usage. The app usage was considered as the compliance of intervention in the intervention group. The primary analysis used was intention-to-treat analysis. Multilevel analyses using random effects logistic regression models were performed to compare the odds of OHI between the two groups, adjusted for individual-level (education and work position) and work-related (including water intake when feeling thirsty, cooling measures, and poor sleep before work) covariates. Results: Of 528 participants, 422 (79.92%) were males, and the mean SD age was 36.36 8.18 y. The primary outcome, OHI incidence, was 1.80% in the intervention group and 4.82% in the control group at the end of the whole follow-up. OHI mainly occurred between 1100 and 1500 hours, with nausea, significantly increased heart rate, and oliguria being the top three reported OHI symptoms. Compared with the control group, the adjusted odds ratios between the intervention group and control group were 0.73 [95% confidence interval (CI): 0.30, 1.76] in the first follow-up wave, with 0.38 (95% CI: 0.15, 0.97), 0.29 (95% CI: 0.08, 1.05), and 0.39 (95% CI: 0.13, 1.19) in the following three follow-up waves, respectively. The intervention also significantly improved PPE usage in the intervention group. Discussions: This multifaceted intervention reduced the OHI risk among outdoor workers in the power grid industry. However, further research is needed to design a more flexible intervention strategy and evaluate its effectiveness in a larger population.

Headnote

Background: Occupational heat-related illness (OHI) is a health threat to workers that can be fatal in severe cases. Effective and feasible measures are urgently needed to prevent OHI.

Objectives: We evaluated the effectiveness of a multifaceted intervention, TEMP, in reducing the OHI risk among outdoor workers in the power grid industry.

Methods: A cluster randomized controlled trial was conducted with power grid outdoor workers in Southern China from 4 July 2022 to 28 August 2022. Work groups were randomly allocated (1:1) to the intervention or control groups. The multifaceted intervention TEMP comprised mobile application (app)-based education training (T), personal protective equipment [PPE (E)], OHI risk monitoring (M), and educational posters (P). Four follow-ups were conducted every 2 wk after the trial began. The primary outcome was the OHI incidence, and the secondary outcome was PPE usage. The app usage was considered as the compliance of intervention in the intervention group. The primary analysis used was intention-to-treat analysis. Multilevel analyses using random effects logistic regression models were performed to compare the odds of OHI between the two groups, adjusted for individual-level (education and work position) and work-related (including water intake when feeling thirsty, cooling measures, and poor sleep before work) covariates.

Results: Of 528 participants, 422 (79.92%) were males, and the mean SD age was 36.36 8.18 y. The primary outcome, OHI incidence, was 1.80% in the intervention group and 4.82% in the control group at the end of the whole follow-up. OHI mainly occurred between 1100 and 1500 hours, with nausea, significantly increased heart rate, and oliguria being the top three reported OHI symptoms. Compared with the control group, the adjusted odds ratios between the intervention group and control group were 0.73 [95% confidence interval (CI): 0.30, 1.76] in the first follow-up wave, with 0.38 (95% CI: 0.15, 0.97), 0.29 (95% CI: 0.08, 1.05), and 0.39 (95% CI: 0.13, 1.19) in the following three follow-up waves, respectively. The intervention also significantly improved PPE usage in the intervention group.

Discussions: This multifaceted intervention reduced the OHI risk among outdoor workers in the power grid industry. However, further research is needed to design a more flexible intervention strategy and evaluate its effectiveness in a larger population. https://doi.org/10.1289/EHP14172

Introduction

Occupational heat-related illness (OHI) is a spectrum of diseases caused by individuals working in adverse heat conditions. OHI can endanger workers' health and reduce their productivity,1'2 and severe OHI, such as heatstroke, can even cause death.3'4 The

International Labour Organization estimates that > 1 billion workers endure excessive heat exposure.2'5 A meta-analysis of 13,088 workers from 33 studies found that the prevalence of OHI in adverse heat conditions was 35% [95% confidence interval (CI): 31, 39].6 With the influence of global warming, increased temperature poses a significant threat of OHI and other heat-related illnesses in both occupational and nonoccupational populations.7'8 It is important to highlight that as global temperatures continue to rise rapidly9 the increasing number of vulnerable workers, especially those in outdoor labor, not only leads to economic losses but also makes OHI prevention more challenging.510

The United Nations has called for "urgent action to combat climate change and its impacts."11 Some countries have formulated preventive regulations and standards for heat-related exposure,12-14 which mainly include preventive education, risk monitoring and warning, ensuring hydration, setting occupational exposure limits for heat stress, and heat acclimatization training. Preventive education and risk monitoring are the foundation for ensuring the effectiveness of other preventive measures.15'16 However, studies have shown only a small percentage of workers have received heat-related preventive education (25.5%) or warnings (17.1%) in Italy,17 with similar results reported in the United States and Australia.18'19 These findings suggest that heat-related preventive regulations or standards might not be fully implemented in many workplaces,18 and more effective measures are needed to prevent OHI. A randomized controlled trial (RCT) found that after implementing the intervention based on a mobile application (app) among American agricultural workers and managers, the physiological heat strain in the intervention group was lower than that in the control group.20 Another study indicated that workers showed interest in using the mobile app for heat-related preventive education and warning,21 illustrating that preventive measures based on mobile apps might be feasible.

In hot seasons, outdoor workers in the power grid industry need to perform physically intense operations, such as emergency repair to ensure power supply, which may cause them to experience severe occupational heat stress (OHS). In a 2022 study in North America, ~ 66.7% of workers in the electric utility industry (including the power grid industry) were reported to have suffered moderate or above levels of OHS.22 Another research study found that heat-related preventive programs in the electric utility industry were inadequate.23 Severe OHS and inadequate preventive programs have led to high OHI risk among outdoor workers in the power grid industry. Based on the above situations, we hypothesized that implementing a multifaceted intervention for outdoor workers in the power grid industry could reduce their OHI risk, and we tested this hypothesis with a cluster RCT.

Methods

Study Design

This cluster RCT was conducted in the Ultra-high Voltage Branch of State Grid Chongqing Electric Power Company, located in Chongqing Municipality, China. Chongqing (east longitude 105 11' to 110 11', north latitude 28 10' to 32 13') has a subtropical humid monsoon climate, with a long and very hot summer and a short and warm winter, and is one of the "oven cities" in China.24'25 Temperature, humidity, and wind speed data were sourced from the Chongqing Meteorological Service. In 2022, the annual average temperature was 17.4 C.26

The trial was approved by the Public Health Ethics Committee of Shandong University (LL202303053) and the Biomedical Ethics Committee of Peking University (IRB00001052-21066). Electronic informed consent was obtained from the outdoor workers after details of the trial had been fully described. The trial was registered in the Chinese Clinical Trial Registry (ChiCTR2300071013) and followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines (Table SI).

Sample Size Calculation

The sample size calculation and assumption were based on the premise that the odds of OHI (OHI incidence/1 - OHI incidence) could be reduced in both control and intervention groups as a result of attention given to the study participants and heat acclimation.27-30 The null hypothesis was that the odds of OHI in the intervention group would be equivalent to that of the control group, and the alternative hypothesis was that the odds of OHI in the intervention group would be lower than that of the control group. Based on data from the previous study,31 the mean odds of OHI among power grid outdoor workers was -0.0906 [8.31%/(1 - 8.31%)], which was set as the baseline odds of OHI. We assumed that by the end of the follow-up period, the intervention group and control group in this study could reduce the odds of OHI by 90% and 20%, respectively. In other words, the odds of OHI for the intervention group was projected to be 0.0181, and for the control group, 0.0815. The study had 90% power to find a significant result (with a threshold one-sided p-value of 0.05). Given that the study was conducted in a relatively stable company, we assumed a dropout rate of 5%. The sample size was calculated using the Tests for Two Proportions module of PASS (version 15; NCSS, LLC), using a one-sided Z-test, with unspooled variance at a significance level of 0.05. The calculated sample size was 209 for both the intervention and control groups. Finally, the intervention group included 278 cases, and the control group, 250 cases.

Randomization and Blinding

We selected all seven outdoor work departments in the Ultrahigh Voltage Branch of State Grid Chongqing Electric Power Company, including three in urban areas and four in suburban areas (Figure SI). Workers in the urban departments were stratified by work positions and randomly allocated (1:1) to the intervention or control group by work groups. Because there was only one work group for the live-line work position, we combined it with the cable inspection and maintenance work position for randomization. For the suburban workers, randomization was conducted at the department level (1:1), given that the work positions within each department were similar. The randomization sequence was generated using R (version 4.2.0; R Development Core Team), with a fixed seed of 1. The results of randomization can be found in Table S2. Except for the researcher responsible for randomization, the participants and other researchers were blind to randomization and outcome measurements.

Participants

At the beginning of the study, 644 individuals from seven targeted outdoor-working departments of the recruited company were assessed for eligibility, representing the baseline population. The specific work positions included in urban and suburban areas are detailed in Table S3. Workers were included according to the following criteria: a) The work environment was located outdoors, b) the participant's age was >18 y, c) the participant had no language communication barriers, and d) the participant had a smartphone and could receive text messages. Of the 644 recruited workers, 528 were included and 116 were excluded owing to ineligibility (100 individuals worked indoors as office clerks or dispatchers, 11 did not have smartphones, and 5 declined to be involved).

Intervention

Based on literature reviews20'32-34 and interviews with key stakeholders, the multifaceted intervention, TEMP, was developed, including four main intervention strategies (Training, Equipment, Monitoring, and Poster). Specifically, T indicated OHI prevention capability training through app-based health education, E indicated OHI personal protective equipment (PPE), M indicated OHI risk monitoring and warning with recommendations, and P indicated OHI prevention posters. The intervention was implemented from 4 July 2022 to 28 August 2022, with all participants receiving instructions on using the measures during a training session before the trial began. Each measure of the intervention is described below.

Training OHI prevention capabilities through app-based health education. This measure relied on the popular science, prevention, and emergency functions in the mobile app named Emergency Event Assistant, which was specifically designed for this study. The main health education contents included a) heat-related knowledge, b) OHI preventive recommendations (personal and workplace preventive recommendations), and c) OHI emergency treatment recommendations (text and video). All contents were reviewed by experts before being published in the app. Detailed information on the app can be found in the Supplemental Material in "Supplementary File 1."

Equipping OHIPPE. Before selecting PPE as one of the interventions, we conducted convenience interviews with workers about their practical experience in the workplace. Based on safety guidelines within the company35 and previous literature,34'36 we finally selected widely used and effective PPEs for this study, including cooling gel stickers, portable fans, insulated cups, and sunshade brims that can be added to helmets (Figure 1). Each worker in the intervention group received 100 cooling stickers, one cup, one fan, and one brim.

Mobile app OHI prediction model building. This section discusses how the OHI risk monitoring and warning system was built for the mobile app. Data on OHI and related influencing factors were collected through a questionnaire for the workers in the State Grid Chongqing Electric Power Company. In total, 563 of the 1,284 participants interviewed in the survey reported OHI (Table S4). The influencing factors were determined by systematic review37-39 and Delphi methods. The Delphi methods were conducted in two rounds involving 25 experts from fields such as power, occupational diseases, and epidemiology, with an average of 26.32 9.95 y of professional experience. The predictors for this model included the following: sex, age, body mass index (BMI), preexisting disease, long-term or regular medication, frequency of drinking water when feeling thirsty in a high-temperature environment during the past 3 y, frequency of taking cooling or ventilation measures in a high-temperature environment, frequency of physically intense work in a high-temperature environment during the past 3 y, frequency of feeling fatigued or weak in a high-temperature environment work during the past 3 y, frequency of working in enclosed workspaces (poor ventilation) in a high-temperature environment during the past 3 y, frequency of poor sleep before starting work in a high-temperature environment during the past 3 y, frequency of feeling unwell before starting high-temperature work during the past 3 y, heat-related illness history during the past 3 y, heat resistance training experience during the past 3 y, education or training on heat-related illness prevention during the past 3 y, whether the workplace established any regulations on heat-related illness prevention during the past 3 y, longest duration of uninterrupted daily work in high-temperature environments during the past 3 y, temperature, humidity, and wind speed. The coding information is illustrated in Table S5.

For the model development, we randomly split the whole dataset into a training set (70%) and a testing set (30%). All categorical predictor variables were converted into dummy variables. Four algorithms were used to build the model: a) random forest, b) logistic regression, c) support vector machine, and d) XGBoost. Three rounds of 10-fold cross-validation were performed on the training set to optimize the hyperparameters of the classifiers (Table S6). In each iteration, the dataset was evenly divided into 10 subsets, with 1 subset used as the validation dataset and the remaining 9 used for training. This process was repeated until each of the 10 subsets served as the validation dataset, repeated three times. Hyperparameters were tuned using a grid search, with the average performance on the validation sets determining the final hyperparameter values for the model. Internal validation of the model's predictive performance was conducted on an independent test set not used in model development. Among the four algorithms, the random forest model had a higher accuracy than the other three models (Figure S2), and this was the model ultimately deployed in the mobile app. In addition, we reviewed and compared existing heat-related illness prevention and control apps, such as ClimApp,40 Heat-Health,41 HEAT-SHIELD,42 and HEAT awareness43 (Table S7).

Briefly, the app in this study monitored the OHI risk using climate, individual, and outdoor work-related information and provided OHI warning messages and recommendations to workers in two ways, a) The system regularly monitored the OHI risk for workers at 0700 hours and sent warning messages and recommendations to participants if the risk monitoring results were high or extremely high, both automatically, b) Workers could monitor the real-time OHI risk and get warning messages using the system that updated climate information hourly.

The OHI warning messages and recommendations were issued by the mobile app automatically, based on the artificial intelligence algorithms, when the app calculated the individual's daily risk assessment result as high or extremely high. In addition, according to the "Administrative Measures on Heatstroke Prevention,"44 the system would directly predict the risk as high or extremely high if the maximum temperature of the day was >37 C (high risk) or >40 C (extremely high risk). Warning messages and recommendations at the same risk level were sent randomly to participants. For instance, one of the warning messages and recommendations for extremely high OHI risk was, "[Emergency Event Assistant] Avoid prolonged outdoor work during the day and ensure hydration and rest!"

Posting OHI preventive posters. We posted OHI preventive posters in the offices of intervention work groups, reminding workers to use PPE, maintain hydration, and rest. An example of the poster can be found on GitHub (https://github.com/DRY0424/ Emergency-Event-Assistant-APP-TEMP.git).

Control group. Both the control and intervention groups followed the heat-related prevention program of the company, which included several components: a) conducting physical examinations annually and transferring out workers who were not suitable for outdoor high-temperature operations; b) implementing heat-related preventive and emergency training; c) working in shifts; d) allowing workers to take work-rest cycles by self-pacing; e) providing heat-relieving food and beverages; /) providing sunscreens, gloves, and goggles; g) carrying emergency kits; and h) allowing work group leaders to adjust or reduce outdoor operations based on weather conditions. The control group did not implement any measure from our intervention; however, participants could voluntarily use the same PPEs as those provided to the intervention group on their own.

Assessments

Before the trial began, baseline assessments were conducted from 27 June to 3 July 2022 simultaneously for both the intervention and control groups. Subsequently, four follow-up assessments were conducted every 2 wk after the trial began, which also guaranteed both the intervention and control groups being assessed at the same follow-up time points. All assessments were carried out online after detailed explanations to outdoor workers by uniformly trained researchers.

OHI information was assessed in two stages: First, participants reported information about OHJ occurrence (time and corresponding symptoms) through a follow-up questionnaire, and second, we further ascertained the time of occurrence (24-h clock) and corresponding symptoms of each case through telephone follow-up to comprehensively collect information on each occurrence of OHI. The telephone follow-up was conducted by well-trained investigators. The questionnaire and detailed information can be found in the Supplemental Material in "Supplementary File 2" and Table S8. Participants meeting any of the following criteria were included in the telephone follow-up study: participants who reported heat-related illness more than twice or during off-hours (e.g., weekends or beyond 0800-1700 hours) and participants who reported heat-related illness once, but the reported symptoms aligned with the diagnostic criteria of OHI in China (GBZ 41-2019). OHI was defined as workers reporting at least one symptom consistent with the OHI diagnosis in the diagnostic criteria of OHI in China (GBZ 41-2019),45 and the symptoms were caused by outdoor work and confirmed through telephone follow-up.

For the PPE use assessment, we defined it as workers reporting the use of at least one PPE (cooling stickers, fan, or brim) in the past 2 wk, collected via questionnaire: "In the past two weeks, have you used cooling stickers/fan/brim? (Yes/No)." Regarding app usage, we defined it as whether intervention group participants followed warning messages and recommendations sent by the app. This was assessed with the question: "In the past two weeks, have you mostly followed the warnings or recommendations suggested by the app? (Yes/No)."

Statistical Analysis

We divided the data into two datasets: the intention-to-treat (ITT) and the per-protocol (PP). Primary analyses were conducted with the ITT population according to their original allocation group. Baseline and follow-up characteristics are presented as means standard deviations (SDs) for continuous variables and counts (percentages) for categorical variables. Differences between intervention and control groups were assessed using the two-sample f-test for continuous variables and the chi-square/Fisher's exact test for categorical variables. Bar charts and violin plots were used to visualize the difference between the intervention and control groups. The primary outcome was the incidence of OHI. We used multilevel logistic regression models, accounting for clustering (at the work group level for urban workers and the work department level for suburban workers) as a random effect, to obtain unadjusted (crude model) and adjusted odds ratios (ORs) and 95% CIs of OHI. These models used the lem4 R package with the logit link function based on the binomial distribution. The covariates used in the adjusted models were education level (high school or below/above high school), work position (cable inspection and maintenance, electrical test, live-line working, substation inspection, substation maintenance, and transmission line inspection and maintenance), water intake when feeling thirsty (in a few cases/in about half the cases/in most cases), cooling measures (in a few cases/in about half the cases/in most cases), and poor sleep before work (in a few cases/in about half the cases/ in most cases). Because of the extremely low OHI incidence in each cluster, we also constructed a fixed effects logistic model46 to further evaluate the difference between the intervention and control groups as sensitivity analyses.

The secondary outcome was the use of PPE in both the intervention and control groups by using a chi-square test. In addition, app usage was considered as an indicator of compliance with the intervention in the intervention group. A post hoc subgroup analysis was performed for the effect of intervention in transmission line inspection and maintenance, mainly because those workers showed a higher OHI odds than participants in other work positions. The covariates used in the adjusted models were age (continuous variable), years of work (continuous variable), and smoking status (never/used to/current).

To minimize missing data, a) the questions in the baseline questionnaire and follow-up questionnaire were required and underwent logical validation during the study implementation, and b) the study was supported by the electric power company, with high cooperation from the participating workers. The variable characteristics and missing proportions of variables in this study are reported in Tables S9 and S10. We used the full dataset in both the ITT and the PP analysis. A two-sided p < 0.05 was defined as statistically significant. Statistical analysis was performed using Stata (version 17.0; StataCorp) and R (version 4.4.0; R Development Core Team).

Results

In 2022, the average temperature was 31.0 C and 34.0 C from July and August, respectively. Most days in these months reached the orange warning level (>37 C) or the red warning level (>40 C) of the heatwave (Figure S3). Overall, 22 clusters (urban, n = 18; suburban, n = 4) with 528 workers were included at baseline (n = 278, intervention group; n = 250, control group). During the study period, 2 participants were lost to follow-up, 1 was due to resignation, and the other changed from the intervention group to the control group due to job assignment during the third and fourth follow-up waves (Figure 2).

At baseline, 422 (79.92%) of 528 workers were males, with a mean SD age of 36.36 8.18 y and mean SD work years of 12.69 8.73 y (Table 1). There were 60 (21.58%) participants who reported a previous history of OHI in the intervention groups, and 54 (21.60%) in the control group. Compared with the control group, the intervention group had a higher proportion of live-line work positions (6.47% vs. 0.00%) and took cooling measures in a few cases (59.35% vs. 43.20%).

The primary outcome, OHI incidence, was 7.91% in the intervention group and 10.40% in the control group during the first follow-up wave, whereas it was 1.80% in the intervention group and 4.82% in the control group in the fourth follow-up wave. The onset time of the worst-reported OHI was similar between the intervention and the control group during the four follow-up waves (Figure 3) and was mainly from 1100 to 1500 hours in the day (Figure S4). As for the different work positions, participants in the transmission line inspection and maintenance position reported higher OHI incidence than participants in other work positions, with 14.93% in the intervention group vs. 17.86% in the control group in the first follow-up wave and 2.99% in the intervention group vs. 7.14% in the control group in the fourth follow-up wave (Figure S5). The incidence of prodromal OHI symptoms and OHI symptoms are presented in Figures S6 and S7.

Specifically, the top three OHI symptoms reported by participants were nausea, significantly increased heart rate, and oliguria. For example, the incidence of nausea in the intervention group was 3.96%, 1.80%, 1.80%, and 0.36% from the first follow-up wave to the fourth follow-up wave, respectively, with the corresponding incidence in the control group being 4.40%, 4.00%, 2.40%, and 1.60%.

The adjusted OR between the intervention group and control group was 0.73 (95% CI: 0.30, 1.76) in the first follow-up wave, with 0.38 (95% CI: 0.15, 0.97), 0.29 (95% CI: 0.08, 1.05), and 0.39 (95% CI: 0.13, 1.19) in the following three follow-up waves, respectively (Table 2). In addition, the PP analysis (Table S11) and the results of the fixed model in both ITT analysis and PP analysis showed similar results (Tables S12 and S13). In the intervention group, the use of cooling stickers was relatively stable, with 43.53%, 41.73%, 43.17%, and 41.37% of participants reporting use during the four follow-up waves, respectively. In addition, the use of fans increased from 39.57% to 49.64%, but the use of brims slightly decreased from 43.17% to 40.65%. For secondary outcomes, compared with the control group, the use of PPE was significantly higher in the intervention group at the first follow-up wave (62.95% vs. 30.00%, p< 0.001), the second follow-up wave (63.31% vs. 31.60%, p< 0.001), the third follow-up wave (62.95% vs. 38.40%, p< 0.001), and the fourth follow-up wave (61.87% vs. 41.20%, p< 0.001) (Table 3).

In the post hoc analysis, the adjusted OR between the intervention group and control group was 0.73 (95% CI: 0.22, 2.40), 0.31 (95% CI: 0.09, 1.09), 0.36 (95% CI: 0.09, 1.51), and 0.54 (95% CI: 0.09, 3.24) during the four follow-up waves, respectively (Table S14). The compliance of app usage was 62.23% in the first follow-up wave, and it was 66.55% in the fourth follow-up wave (Table S15).

Discussion

We conducted this cluster RCT to implement a multifaceted intervention for power grid outdoor workers in Chongqing Municipality,

China. The results indicate that interventions, including training, equipping, monitoring, and posters, might have an effect on reducing OHI among power grid workers.

With the global temperature increasing,47 OHI prevention has become a serious public health challenge that needs more effective and feasible preventive measures to address it. The first and most important measure is monitoring the risk of high-temperature operations.48 Current heat-related prevention regulations mostly use the wet-bulb globe temperature (WBGT) index, heat index, or temperature as indicators to monitor the risk of high-temperature operations and guide OHI prevention, ignoring the impact of individual, work-related, and other climate factors on OHI,49'50 which may not apply to outdoor work environments.51 In Europe, the HEAT-SHIELD project conducted an OHS monitoring and warning system48 that incorporated climate, individuals, and heat acclimatization factors, using the heat risk level (defined as the percentage of the effective WBGT index over the recommended alert limit) to monitor OHS. However, the aforementioned system does not include someimportant individual factors (e.g., age, sex) and can only provide weekly ensemble forecasting32 in a style of a website platform rather than a mobile app. Except for the HEAT-SHIELD project, other well-trained mobile apps, such as ClimApp,40 Heat-Health,41 and HEAT awareness,43 also face challenges when applied to power grid workers. First, except for the HEAT awareness app, which conducted one RCT study to verify its effectiveness,20 most apps have not yet been used in practice. Second, from an intervention strategy perspective, it might be more efficient to directly monitor the OHI risk in addition to daily warnings and recommendations to workers. However, most current apps do not have such a function. For instance, the HEAT awareness app sent notifications to agricultural supervisors 1 or 6 d in advance rather than directly to workers,43 potentially affecting the timeliness and accuracy of warning messages. Third, from a predictor perspective, the factors included in our model are more targeted and comprehensive in reflecting the work characteristics of Chinese power grid workers. For example, other apps, such as ClimApp, do not consider the history of thermal exposure and assume the user starts from a neutral condition,40 and Heat-Health lacks important heat stress factors, such as age, sex, and medication use.41 Fourth, from a feasibility perspective, existing apps do not have a Chinese language setting, and some systems, such as HEAT-SHIELD, limit the collection of meteorological factors to Europe.42 In addition, some app intervention suggestions are unsuitable for the working environment of power grid workers. For example, HEAT-Health ran the model executing a condition-controlled loop to estimate the optimal clothing for an individual. The termination condition was set to either the core temperature being <37.5 C or the clothing thermal insulation being equal to 0.5 "clo" (very light clothing). However, power grid workers are required to wear long-sleeved uniforms and long pants regardless of the weather owing to their work environment (live electrical working environment). Last, from a functional orientation perspective,the app we present might be better and more comprehensive owing to its combining monitoring and warning with the health education and emergency response functions for OHI prevention. Most current apps lack these functions, which we tried to deliver to workers via the app in the multifaceted intervention strategy TEMP. To be noted, despite the challenges faced by current apps when applied to special occupational populations, these apps still hold considerable potential through relevant collaborations and model optimizations.

Our findings support the view that OHI can be reduced by implementing multifaceted preventive measures for outdoor workers. In our intervention, the OHI risk monitoring and warning system can monitor daily and real-time OHI risk using climate, individual, and work-related factors and provide OHI warning messages to each worker. Furthermore, several studies have shown that health education can help workers improve OHI-related knowledge and behavioral intentions52'53, and workers showed interested in mobile app-based health education in other studies.21'54 However, health education alone may not reduce the OHI risk efficiently,20 and more preventive measures should be integrated. A field study has shown that the proportion of agriculture workers reporting heat-related illness symptoms was reduced after using PPE.55 Thus, based on implementing health education and risk monitoring measures, we also selected PPEs that have been proven in previous studies to facilitate ventilation and reduce the radiative heat load56 to prevent OHI. However, compared with previous studies, only one study with agricultural workers and managers using health education and mobile app-based monitoring found that these preventive measures alone cannot significantly reduce the physiological heat strain of workers (e.g., the mean SD physiological strain index was 4.3 1.5 in the intervention group vs. 4.6 1.5 in the control group).20 This may be because their monitoring system used a fixed threshold to monitor high-temperature risk, but workers may not always be exposed to the same conditions. In addition, the warning messages were sent to managers rather than directly to the workers.20'43 This intervention strategy may lead to workers not being notified of their accurate risks and, as a result, implementing improper work plans and preventive measures, which is also the main reason we designed the intervention to be a daily direct intervention for the workers in this study. In addition, it is worth noting that the mixed model with clusters as random effects might be nonconverging when the outcome incidence is low and the sample size in each cluster is small,57'58 for which zero-inflation could be a potential solution.57 Moreover, in this study, the results of the fixed model and the mixed model may be quite similar, partly due to the relatively low intracluster correlation.

Compared with other workers, workers in the transmission line inspection and maintenance position had to perform heavy, physically intense operations, such as manual pole work,59 under high OHS conditions, resulting in a higher OHI risk than others,22 which required more targeted measures, such as implementing structured mandatory rest breaks,34 to reduce OHI risk. The trial's success carries implications for the prevention of heat-related occupational injuries. Reducing OHI risk will not only decrease the OHI incidence but may also reduce the risk of other heat-relatedoccupational injuries, such as kidney injury, and will have a positive effect on curbing the huge heat-related health and productivity losses.61 In addition, the intervention was feasible, low-cost, and sustainable, suggesting its potential for widespread adoption across other outdoor work industries. Future studies should determine the effectiveness of our intervention in the entire power grid industry and other outdoor work industries. Moreover, we also recommend that more preventive measures should be considered and implemented to improve the OHI prevention programs of outdoor work industries based on the intervention, such as conducting heat acclimatization training and work-rest cycles.56'62 Perfect OHI preventive programs will effectively reduce the OHI risk among outdoor workers in the increasingly hot work environment in the future.

Strengths

To our knowledge and based on recent reviews,34'56 our trial was the first RCT to effectively reduce the OHI risk among outdoor workers in the power grid industry using a multifaceted intervention. The main strength of this study lies in its multifaceted nature, including both mobile app-based health education and risk monitoring and warning, as well as real-world-based PPE and preventive posters. All these measures were effective, economical, and sustainable.

Limitations

The study also has several limitations. First, self-reported OHI symptoms from workers may have been affected by recall bias and misclassification, both differential and nondifferential.63'64 Previous research has shown that longer recall periods result in larger differences,65 thus we selected a 2-wk time frame to minimize recall bias after considering the feasibility of data collection.66 The possibility that participant responsiveness may have been influenced by the intervention, either positively or negatively, cannot be ruled out.67 However, potential misclassification may be mitigated by the systematic and comprehensive OHI prevention and control knowledge provided to participants in the intervention group,68'69 including regular company OHI training, local emphasis on heat-related illness prevention, a two-stage OHI assessment, and rigorous quality control in this study. Second, although we recruited all eligible workers, only ~ 20% of the participants were female, which relates to the small number of female workers assigned to outdoor operations in the power grid industry. Results of a review has suggested that female outdoor workers were more likely to suffer OHI.70 Future studies should recruit additional female participants. Third, the blinding of workers to the intervention group might not have been completely effective. We used cluster randomization to minimize contamination, but workers might have talked to each other or share PPE when working together, which could shift the behavior of the control group toward the intervention group and reduce the intervention effect.71 However, we still observed a reduction of the OHI risk in the intervention group, suggesting that the potential intervention effect might have been larger. Fourth, some indicators, such as the time spent while using the app and the evaluation of the usage of each app function, were not recorded in the present study to measure engagement with digital technologies, and this index could be an important variable to reflect the effectiveness of the app in future study.72 Finally, we did not include physiological data to evaluate the effectiveness of the app on occupational heat stress/strain and its influence on OHI incidence, considering the cost, feasibility, and cooperation of our study with a relatively large number of participants.73-75 Future research can further incorporate physiological indicators into the study design to explore effective interventions that can influence these indicators and ultimately reduce OHI.76 Conclusions This multifaceted, app-based intervention might reduce the OHI risk among outdoor workers in the power grid industry. Further studies are needed to assess its effectiveness in the entire power grid industry and other outdoor work industries. Acknowledgments R.D., Q.G., and S.W. designed the study. Q.G. performed randomization. R.D., Y.T., J.W., X.M., H.L., and H.Z. enrolled participants and collected data. R.D. and S.W. performed the analysis and made the figures. R.D. drafted the manuscript. S.W., P.L., Y.W., Y.Y., and X.M. revised the manuscript. All authors read and approved the final manuscript. This work was supported by the National Nature Science Foundation of China (grant numbers: 82173616, 72342015) and the State Grid Corporation of China (State Grid) (grant number: 520626210025). We sincerely thank all power grid outdoor workers who participated in this study. We also thank the managers of the Ultrahigh Voltage Branch of State Grid Chongqing Electric Power Company for their full cooperation and communication. De-identified data for use in future studies will be available upon reasonable request in the premise of approvals from the Expert Panel of TEMP, Science and Technology Project of the Head Office of State Grid Grid Co., LTD. Collaborations and external investigations on the TEMP dataset are welcomed to make more contributions to occupation-related health promotions. The analysis code of the study may also be obtained by reasonable application. The application form can be found in the Supplemental Material in "Supplementary File 3." The Expert Panel of TEMP will contact you via email once your application is considered meaningful and the data is approved by the above committees.

References

References 1. Morrissey MC, Brewer GJ, Williams WJ, Quinn T, Casa DJ. 2021. Impact of occupational heat stress on worker productivity and economic cost. Am J Ind Med 64(12):981-988, PMID: 34590324, https://doi.org/10.1002/ajim.23297. 2. Kjellstrom T, Maitre N, Saget C, Otto M, Karimova T. 2019. Working on a Warmer Planet: The Impact of Heat Stress on Labour Productivity and Decent Work. Geneva, Switzerland: International Labour Organization, https:// www.humanitarianlibrary.org/sites/default/files/2023/09/Working%20on%20a% 20Warmer%20Planet%20-%20The%20lmpact%20of%20Heat%20Stress%20on% 20Labour%20Productivity%20and%20Decent%20Work.pdf [accessed 2 July 2023]

3. Bouchama A, Abuyassin B, Lehe C, Laitano 0, Jay 0, O'Connor FG, et al. 2022. Classic and exertional heatstroke. Nat Rev Dis Primers 8(1):8, PMID: 35115565, https://doi.org/10.1038/s41572-021-00334-6. 4. Yang M, Li Z, Zhao Y, Zhou F, Zhang Y, Gao J, etal. 2017. Outcome and riskfac-tors associated with extent of central nervous system injury due to exertional heatstroke. Medicine (Baltimore) 96(44):e8417, PMID: 29095276, https://doi.org/ 10.1097/M D.0000000000008417.

5. Ebi KL, Capon A, Berry P, Broderick C, de Dear R, Havenith G, et al. 2021. Hot weather and heat extremes: health risks. Lancet 398(10301 ):698-708, PMID: 34419205, https://doi.org/10.1016/S0140-6736(21)01208-3.

6. Flouris AD, Dinas PC, loannou LG, Nybo L, Havenith G, Kenny GP, et al. 2018. Workers' health and productivity under occupational heat strain: a systematic review and meta-analysis. Lancet Planet Health 2(12):e521-e531, PMID: 30526938, https://doi.org/10.1016/S2542-5196(18)30237-7.

7. Fatima SH, Rothmore P, Giles LC, Varghese BM, Bi P. 2021. Extreme heat and occupational injuries in different climate zones: a systematic review and metaanalysis of epidemiological evidence. Environ Int 148:106384, PMID: 33472088, https://doi.Org/10.1016/j.envint.2021.106384.

8. Faurie C, Varghese BM, Liu J, Bi P. 2022. Association between high temperature and heatwaves with heat-related illnesses: a systematic review and meta-analysis. Sci Total Environ 852:158332, PMID: 36041616, https://doi.org/10. 1016/j.scitotenv.2022.158332. 9. Matthews HD, Wynes S. 2022. Current global efforts are insufficient to limit warming to 1.5 C. Science 376(6600): 1404-1409, PMID: 35737785, https://doi.org/ 10.1126/science.abo3378.

10. Cheng L, Gu K, Zhao L, Wang H, Ji JS, Liu Z, et al. 2023. Projecting future labor losses due to heat stress in China under climate change scenarios. Sci Bull (Beijing) 68(22):2827-2837, PMID: 37858411, https://doi.Org/10.1016/j.scib.2023. 09.044. 11. ESCAP (Economic and Social Commission for Asia and the Pacific). 2021. SDG 13: Climate Action: Take Urgent Action to Combat Climate Change and Its Impacts. Bangkok, Thailand: ESCAP. https://repository.unescap.org/res1/bitstreams/ed8b9696-45b1-463e-9857-a3d1958ac714/retrieve [accessed 8 September 2023].

12. Zhou Z. 2018. Understanding the administrative regulation on occupational health and trend in China. J Occup Health 60(2):126 131, PMID: 29311440, https://doi.org/10.1539/joh.17-0229-RA. 13. Schmidt CW. 2022. Hot pursuit: new efforts to prevent heat-related illness on the job. Environ Health Perspect 130(9):092001, PMID: 36094404, https://doi.org/ 10.1289/EHP11954.

14. Acharya P, Boggess B, Zhang K. 2018. Assessing heat stress and health among construction workers in a changing climate: a review. Int J Environ Res Public Health 15(2):247, PMID: 29389908, https://doi.org/10.3390/ijerph15020247. 15. Epstein Y, Druyan A, Heled Y. 2012. Heat injury prevention-a military perspective. J Strength Cond Res 26(suppl 2):S82-S86, PMID: 22614224, https://doi.org/ 10.1519/JSC.0b013e31825cec4a.

16. Chirico F, Magnavita N. 2019. The significant role of health surveillance in the occupational heat stress assessment. Int J Biometeorol 63(2):193-194, PMID: 30519957, https://doi.org/10.1007/s00484-018-1651-y. 17. Bonafede M, Levi M, Pietrafesa E, Binazzi A, Marinaccio A, Morabito M, et al. 2022. Workers' perception heat stress: results from a pilot study conducted in Italy during the C0VID-19 pandemic in 2020. Int J Environ Res Public Health 19(13):8196, PMID: 35805854, https://doi.org/10.3390/ijerph19138196.

18. Arbury S, Lindsley M, Hodgson M. 2016. A critical review of 0SHA heat enforcement cases: lessons learned. J Occup Environ Med 58(4):359-363, PMID: 27058475, https://doi.org/10.1097/JOM.0000000000000640. 19. Xiang J, Hansen A, Pisaniello D, Bi P. 2016. Workers' perceptions of climate change related extreme heat exposure in South Australia: a cross-sectional survey. BMC Public Health 16:549, PMID: 27402077, https://doi.org/10.1186/ S12889-016-3241-4.

20. Chavez Santos E, Spector JT, Egbert J, Krenz J, Sampson PD, Palmandez P, etal. 2022. The effect of the participatory Heat Education and Awareness Tools (HEAT) intervention on agricultural worker physiological heat strain: results from a parallel, comparison, group randomized study. BMC Public Health 22(1):1746, PMID: 36104813, https://doi.org/10.1186/s12889-022-14144-2. 21. LuqueJS, BossakBH, DavilaCB,Tovar-AguilarJA. 2019. "I think the temperature was 110 degrees!": work safety discussions among Hispanic farmworkers. J Agromedicine 24(1):15-25, PMID: 30317928, https://doi.org/10.1080/1059924X.2018. 1536572.

22. Flouris AD, loannou LG, Notley SR, Kenny GP. 2022. Determinants of heat stress and strain in electrical utilities workers across North America as assessed by means of an exploratory questionnaire. J Occup Environ Hyg 19(1 ):12 22, PMID: 34731074, https://doi.org/10.1080/15459624.2021.2001475. 23. Kaltsatou A, Notley SR, Flouris AD, Kenny GP. 2021. An exploratory survey of heat stress management programs in the electric power industry. J Occup Environ Hyg 18(9):436^45, PMID: 34406910, https://doi.org/10.1080/15459624. 2021.1954187.

24. Li Y, Li C, Luo S, He J, Cheng Y, Jin Y. 2017. Impacts of extremely high temperature and heatwave on heatstroke in Chongqing, China. Environ Sci Pollut Res Int 24(9):8534-8540, PMID: 28191617, https://doi.org/10.1007/s11356-017-8457-z. 25. Qian Z, He Q, Lin H-M, Kong L, Bentley CM, Liu W, et al. 2008. High temperatures enhanced acute mortality effects of ambient particle pollution in the "oven" city of Wuhan, China. Environ Health Perspect 116(9):1172-1178, PMID: 18795159, https://doi.org/10.1289/ehp.10847.

26. Chongqing Statistics Bureau. 2023. Chongqing Statistical Yearbook 2023 [in Chinese], https://tjj.cq.gov.en/zwgk_233/tjnj/tjnj.html7urhhttps//tjj.cq.gov.cn/zwgk_ 233/tjnj/2023/2023cqtjnj.pdf [accessed 29 June 2024].

27. Sedgwick P, Greenwood N. 2015. Understanding the Hawthorne effect. BMJ 351:h4672, PMID: 26341898, https://doi.org/10.1136/bmj.h4672. 28. McCambridge J, Witton J, Elbourne DR. 2014. Systematic review of the Hawthorne effect: new concepts are needed to study research participation effects. J Clin Epidemiol 67(3):267-277, PMID: 24275499, https://doi.org/10.1016/ j.jclinepi.2013.08.015.

29. Periard JD, Racinais S, Sawka MN. 2015. Adaptations and mechanisms of human heat acclimation: applications for competitive athletes and sports. Scand J Med Sci Sports 25(suppl 1)20-38, PMID: 25943654, https://doi.org/10. 1111/sms.12408. 30. Periard JD, Travers GJS, Racinais S, Sawka MN. 2016. Cardiovascular adaptations supporting human exercise-heat acclimation. Auton Neurosci 196:52-62, PMID: 26905458, https://doi.Org/10.1016/j.autneu.2016.02.002.

31. WangX, Gong Q, Ding R, Shelenko, Dou D,Wang K, etal. 2023. Investigation of occupational hazard emergencies in power grid enterprises in the three places from 2018 to 2020 [in Chinese]. Chin Occup Med 50(01):53-56, https://doi.org/10. 20001/j.issn.2095-2619.20230208. 32. Grifoni D, Messeri A, Crisci A, Bonafede M, Pasi F, Gozzini B, et al. 2021. Performances of limited area models for the WORKLIMATE heat-health warning system to protect worker's health and productivity in Italy. IntJ Environ Res Public Health 18(18):9940, PMID: 34574860, https://doi.Org/10.3390/ijerph 18189940.

33. Joubert D, Thomsen J, Harrison 0. 2011. Safety in the Heat: a comprehensive program for prevention of heat illness among workers in Abu Dhabi, United Arab Emirates. Am J Public Health 101 (3):395-398, PMID: 21307373, https://doi.org/10. 2105/AJPH.2009.189563. 34. Chicas R, Xiuhtecutli N, Dickman NE, Scammell ML, Steenland K, Hertzberg VS, et al. 2020. Cooling intervention studies among outdoor occupational groups: a review of the literature. Am J Ind Med 63(11):988-1007, PMID: 32886396, https://doi.org/10.1002/ajim.23175.

35. General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China; Standardization Administration of the People's Republic of China. 2011. Safety code of electric power industry-Electric part of power plants and transformer substations. Beijing, China: General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China; Standardization Administration of the People's Republic of China, https:// www.chinesestandard.ne1/PDF/English.aspx/GB26860-2011 [accessed 18 October 2024]. 36. loannou LG, Foster J, Morris NB, Piil JF, Havenith G, Mekjavic IB, et al. 2022. Occupational heatstrain in outdoor workers: a comprehensive review and metaanalysis. Temperature (Austin) 9(1):67-102, PMID: 35655665, https://doi.org/10. 1080/23328940.2022.2030634.

37. Zhang Y, Nitschke M, Bi P. 2013. Risk factors for direct heat-related hospitalization during the 2009 Adelaide heatwave: a case crossover study. Sci Total Environ 442:1-5, PMID: 23168533, https://doi.Org/10.1016/j.scitotenv.2012.10.042.

38. Epstein Y, Yanovich R. 2019. Heatstroke. N Engl J Med 380(25):2449-2459, PMID: 31216400, https://doi.org/10.1056/NEJMra1810762.

39. Peiris AN, Jaroudi S, Noor R. 2017. Heat stroke. JAMA 318(24):2503, PMID: 29279936, https://doi.org/10.1001/jama.2017.18780.

40. Kingma BRM, Steenhoff H,Toftum J, Daanen HAM, Folkerts MA, Gerrett N, et al. 2021. ClimApp-integrating personal factors with weather forecasts for individualised warning and guidance on thermal stress. Int J Environ Res Public Health 18(21 ):11317, PMID: 34769832, https://doi.Org/10.3390/ijerph182111317. 41. loannou LG, Ciuha U, Fisher JT, Tsoutsoubi L, Tobita K, Bonell A, et al. 2023. Novel technological advances to protect people who exercise or work in thermally stressful conditions: a transition to more personalized guidelines. Appl Sci [Basel] 13(15):8561, https://doi.org/10.3390/app13158561.

42. Morabito M, Messeri A, Noti P, Casanueva A, Crisci A, Kotlarski S, et al. 2019. An occupational heat-health warning system for Europe: the HEAT-SHIELD platform. IntJ Environ Res Public Health 16(16):2890, PMID: 31412559, https://doi.org/ 10.3390/ijerph16162890. 43. Krenz J, Chavez Santos E, Torres E, Palmandez P, Carmona J, Blancas M, et al. 2021. The multi-level heat education and awareness tools [HEAT] intervention study for farmworkers: rationale and methods. Contemp Clin Trials Commun 22:100795, PMID: 34169175, https://doi.Org/10.1016/j.conctc.2021.100795.

44. Department of Occupational Health. 2023. National Health Commission of the People's Republic of China. Notice on issuing Administrative Measures on Heatstroke Prevention [in Chinese], http://www.nhc.gov.cn/zyjks/zcwj2/201207/ 3cda5ca273484f0b994733bccc819aac.shtml [accessed 25 August 2023]. 45. National Health Commission of the People's Republic of China. 2023. Diagnosis of occupational heat illness, http://www.nhc.gov.cn/wjw/pyl/201902/ 4c0de7e8b63142fba837b5d18872e6c0.shtml [accessed 25 August 2023].

46. Perkins G D, Lall R, Quinn T, Deakin CD, Cooke MW, Horton J, et al. 2015. Mechanical versus manual chest compression for out-of-hospital cardiac arrest (PARAMEDIC): a pragmatic, cluster randomised controlled trial. Lancet 385(9972):947-955, PMID: 25467566,https://doi.org/10.1016/S0140-6736(14)61886-9.

47. IPCC (Intergovernmental Panel on Climate Change). 2021: Summary for policymakers. In: Climate Change 2021: The Physical Science Basis. Contribution of Working Group 1 to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Masson-Delmotte V, Zhai P, Pirani A, Connors SL, Pean C, Berger S, et al., eds. New York, NY: Cambridge University Press, 3-32.

48. Morris NB, Piil JF, Morabito M, Messeri A, Levi M, loannou LG, et al. 2021. The HEAT-SHIELD project-perspectives from an inter-sectoral approach to occupational heat stress. J Sci Med Sport 24(8):747-755, PMID: 33757698, https://doi.org/ 10.1016/j.jsams.2021.03.001. 49. Notley SR, Flouris AD, Kenny GP. 2019. Occupational heat stress management: does one size fit all? Am J Ind Med 62(12):1017 1023, PMID: 30791115, https://doi.org/10.1002/ajim.22961. 50. Roelofs C. 2018. Without warning: worker deaths from heat 2014-2016. New Solut 28(2):344-357, PMID: 29790832, https://doi.org/10.1177/1048291118777874.

51. Heidari H, Golbabaei F, Shamsipour A, Rahimi Forushani A, Gaeini A. 2018. Consistency between sweat rate and wet bulb globe temperature for the assessment of heat stress of people working outdoor in arid and semi-arid regions. IntJ Occup Environ Med 9(1):1-9, PMID: 29319050, https://doi.org/10. 15171/ijoem.2018.1204. 52. Grzywacz JG, Gonzales-Backen M, Liebman A, Marin AJ, Trejo M, Gudino CO, et al. 2019. Attending to pesticide exposure and heat illness among farmworkers: results from an attention placebo-controlled evaluation design. J Occup Environ Med 61(9):735-742, PMID: 31205205, https://doi.org/10.1097/ JOM.0000000000001650.

53. Marquez D, Krenz JE, Chavez Santos E, Torres E, Palmandez P, Sampson PD, et al. 2023. The effect of participatory heat education on agricultural worker knowledge. J Agromedicine 28(2): 187-198, PMID: 35345983, https://doi.org/10. 1080/1059924X.2022.2058667. 54. Luque JS, Becker A, Bossak BH, Grzywacz JG, Tovar-Aguilar JA, Guo Y. 2020. Knowledge and practices to avoid heat-related illness among Hispanic farmworkers along the Florida-Georgia line. J Agromedicine 25(2)190-200, PMID: 31544652, https://doi.Org/10.1080/1059924X.2019.1670312.

55. Chicas R, Xiuhtecutli N, Elon L, Scammell MK, Steenland K, Hertzberg V, et al. 2021. Cooling interventions among agricultural workers: a pilot study. Workplace Health Saf 69(7):315-322, PMID: 33357122, https://doi.org/10.1177/ 2165079920976524.

56. Morris NB, Jay 0, Flouris AD, Casanueva A, Gao C, Foster J, et al. 2020. Sustainable solutions to mitigate occupational heat strain - an umbrella review of physiological effects and global health perspectives. Environ Health 19(1):95, PMID: 32887627, https://doi.org/10.1186/s12940-020-00641-7. 57. Long DL, Preisser JS, Herring AH, Golin CE. 2015. A marginalized zero-inflated Poisson regression model with random effects. J R Stat Soc Ser C Appl Stat 64(5):815-830, PMID: 26635421, https://doi.org/10.1111/rssc.12104.

58. Huang M, Anderson C. 2021. A Bayesian solution to non-convergence of crossed random effects models. In: Proceedings in Mathematics & Statistics, Quantitative Psychology. IMPS2020, vol. 353. Wiberg M, Molenaar D, Gonzalez J, Bockenholt U, Kim JS, eds. Cham, Switzerland: Springer International Publishing, 297-307. 59. Meade RD, Lauzon M, Poirier MP, Flouris AD, Kenny GP. 2015. An evaluation of the physiological strain experienced by electrical utility workers in North America. J Occup Environ Hyg 12(10):708-720, PMID: 26011148, https://doi.org/ 10.1080/15459624.2015.1043054.

60. Glaser J, Hansson E, Weiss I, Wesseling C, Jakobsson K, Ekstrom U, et al. 2020. Preventing kidney injury among sugarcane workers: promising evidence from enhanced workplace interventions. Occup Environ Med 77(8):527-534, PMID: 32404530, https://doi.org/10.1136/oemed-2020-106406. 61. CaiW, Zhang C, Suen HP, Ai S, Bai Y, Bao J, et al. 2021. The 2020 China report of the Lancet Countdown on health and climate change. Lancet Public Health 6(1):e64-e81, PMID: 33278345, https://doi.org/10.1016/S2468-2667(20)30256-5.

62. loannou LG, Mantzios K,Tsoutsoubi L, Nintou E,Vliora M, Gkiata P, etal. 2021. Occupational heat stress: multi-country observations and interventions. Int J Environ Res Public Health 18(12):6303, PMID: 34200783, https://doi.org/10.3390/ ijerph 18126303. 63. Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al. 2019. Cochrane Handbook for Systematic Reviews of Interventions. Hoboken, NJ: John Wiley & Sons.

64. Yland JJ, Wesselink AK, Lash TL, Fox MP. 2022. Misconceptions about the direction of bias from nondifferential misclassification. Am J Epidemiol 191(8):1485-1495, PMID: 35231925, https://doi.org/10.1093/aje/kwac035. 65. Mendoza TR, Dueck AC, Bennett AV, Mitchell SA, Reeve BB, Atkinson TM, et al. 2017. Evaluation of different recall periods for the US National Cancer Institute's PR0-CTCAE. Clin Trials 14(3):255-263, PMID: 28545337, https://doi.org/10.1177/ 1740774517698645.

66. Stewart WF, Ricci JA, Leotta C. 2004. Health-related lost productive time (LPT): recall interval and bias in LPT estimates. J Occup Environ Med 46(suppl 6): S12-S22, PMID: 15194892, https://doi.org/10.1097/01.jom.0000126685.59954.55.

67. Bauhoff S. 2011. Systematic self-report bias in health data: impact on estimating cross-sectional and treatment effects. Health Serv Outcomes Res Methodol 11(1-2):44-53, https://doi.org/10.1007/s10742-011-0069-3. 68. Cheveldayoff P, Chowdhury F, Shah N, Burow C, Figueiredo M, Nguyen N, et al. 2023. Considerations for occupational heat exposure: a scoping review. PLOS Clim 2(9):e0000202, https://doi.org/10.1371/journal.pclm.0000202.

69. Robson LS, Stephenson CM, Schulte PA, Amick BC III, Irvin EL, Eggerth DE, et al. 2012. A systematic review of the effectiveness of occupational health and safety training. Scand J Work Environ Health 38(3):193-208, PMID: 22045515, https://doi.org/ 10.5271/sjweh.3259. 70. El Khayat M, Halwani DA, Hneiny L, Alameddine I, Haidar MA, Habib RR. 2022. Impacts of climate change and heat stress on farmworkers' health: a scoping review. Front Public Health 10:782811, PMID: 35211437, https://doi.org/10.3389/ fpubh.2022.782811.

71. Stamp E, Schofield H, Roberts VL, Burton W, Collinson M, Stevens J, et al. 2021. Contamination within trials of community-based public health interventions: lessons from the HENRY feasibility study. Pilot Feasibility Stud 7(1):88, PMID: 33771233, https://doi.org/10.1186/s40814-021-00805-3.

72. Perski 0, Blandford A, West R, Michie S. 2017. Conceptualising engagement with digital behaviour change interventions: a systematic review using principles from critical interpretive synthesis. Transl Behav Med 7(2):254-267, PMID: 27966189, https://doi.org/10.1007/s13142-016-0453-1.

73. Khakurel J, Melkas H, Porras J. 2018. Tapping into the wearable device revolution in the work environment: a systematic review. Inf Technol People 31(3):791-818, https://doi.org/10.1108/ITP-03-2017-0076. 74. Fuller D, Colwell E, Low J, Orychock K, Tobin MA, Simango B, et al. 2020. Reliability and validity of commercially available wearable devices for measuring steps, energy expenditure, and heart rate: systematic review. JMIR Mhealth Uhealth8(9):e18694, PMID: 32897239, https://doi.org/10.2196/18694.

75. Kritzler M, Backman M, Tenfalt A, Michahelles F. 2015. Wearable technology as a solution for workplace safety. In: Proceedings of the 14th International Conference on Mobile and Ubiquitous Multimedia. 30 November-2 December 2015. New York, NY: Association for Computing Machinery, https://doi.Org/10.1145/2836041.28360.

76. Fang W, Wu D, Love PED, Ding L, Luo H. 2022. Physiological computing for occupational health and safety in construction: review, challenges and implications for future research. AdvEng Inform 54:101729, https://doi.Org/10.1016/j.aei.2022.101729.