Risk of obstructive sleep apnea among healthcare workers: a systematic review and meta-analysis Authors: Gianluca Spiteri, Mamoona Arshad, Chiara Michela Rita, Riccardo Ruiu, Stefano Porru, Giuseppe Verlato Published: 2026 ABSTRACT Obstructive sleep apnea (OSA) is a very common but underdiagnosed health disorder. OSA-derived cognitive disorders could increase the risk of work-related injuries among healthcare workers (HW), representing a major threat both to them and their patients. This study aimed to conduct a systematic review and meta-analysis to assess the relevance of OSA in the healthcare sector. Studies reporting the prevalence of either high risk of OSA (HRO) or diagnosed OSA, assessed respectively by structured questionnaires or validated diagnostic tools among healthcare workers, published between 2000 and 2026 in PubMed and Scopus were included. A meta-analysis of the included papers was performed using random-effects models to estimate the pooled prevalence. Variability was evaluated by the heterogeneity test and the I² statistic. Statistical significance was set at p < 5%. Of the 211 articles identified, 9 with 8473 participants were included in the meta-analysis. OSA risk was assessed by the Berlin (BQ) and STOP-BANG questionnaires (SBQ) in 6 and 3 studies, respectively. Prevalence estimates for HRO were highly heterogeneous (I-squared = 99.1%), ranging from 7 to 37%. The pooled estimate was 23% (95% CI 13–33%). Prevalence estimates were significantly affected by geographic area, being higher in Europe and the Middle East than in South-East Asia and the US (p < 0.01), but neither by response rate nor by the questionnaire used. The prevalence of OSA diagnosis was instrumentally obtained only in two studies and ranged from 9.3 to 21.9%. Data on HRO prevalence among HW are sparse, including a few studies with limited sample sizes and variable response rates. Larger studies, based on home sleep tests or new validated digital tools, such as smartwatches, smart pillows, and microphones, are warranted.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-56765-2. Introduction Obstructive sleep apnea (OSA) is a health disorder with a very high prevalence, affecting over 50% of the general population1. Patients suffering from OSA can experience associated symptoms and signs, such as hypertension, insomnia, morning headaches, nocturia, and acid reflux2. Excessive daytime sleepiness (EDS) is detected in 14% of men and 5% of women with OSA3, and it is related to a worsening in quality of life, cognitive deficits, neurobehavioral impairments, and social functioning, as well as an increased prevalence of anxiety and depression4,5. As a result, workers affected by EDS are at higher risk for work-related injuries (WRI). In fact, Melamed reported a 2.23-fold increase in WRI due to EDS among a sample of 532 non-shift daytime workers6. This effect was confirmed in a meta-analysis by Chou et al., which showed that OSA and EDS were associated with an increased probability of a WRI of 1.64 and 1.68, respectively7. An even higher risk was found in another meta-analysis by Garbarino et al., (OR = 2.18)8. This phenomenon is relevant across sectors; however, a special focus should be placed on the healthcare sector, as workplace accidents in this context can affect not only workers themselves but also third parties, such as co-workers and patients. The major determinants of WRI among Healthcare Workers (HW) are well known, including profession, high-risk tasks, time of the day, workload, lack of work experience, and insufficient training9. However, some health conditions could also affect workers’ and patients’ safety. In particular, OSA, especially when associated with EDS, can cause memory deficits, difficulty concentrating, and reduced alertness10. As a consequence, it could have a primary role in the risk of WRI among HW. Along with this hypothesis, Hassani et al., reported that the HW high-risk group for sleep apnea had an OR of 2.74 (95% CI) for WRI compared with the HW low-risk group11. Moreover, the study by Suzuki et al., investigated daytime sleepiness and occupational accidents. The results showed a significant association between the incorrect operation of medical equipment and EDS, confirming that poor sleep quality can affect patients’ safety12. Despite this evidence, while data on the prevalence of OSA in the general population worldwide were extensively reviewed13, information on specific working population, such as HW, is scant, and neither reviews nor meta-analyses have been released on this topic. This knowledge gap limits understanding of the real impact of OSA in these settings and diminishes the awareness of the urgency of occupational physicians’ intervention to prevent health effects and reduce occupational risks. The aim of this study was to review the existing literature on this issue and conduct a meta-analysis to better understand the relevance of high risk of OSA (HRO) in the healthcare sector, detecting the current issues to be addressed by future research for HW and patients’ safety. Methods A comprehensive search was performed in PubMed and Scopus using the following keywords: (“Obstructive sleep apnea” OR “Obstructive sleep apnoea”) AND (“healthcare workers” OR “healthcare professionals” OR “health workers”). The initial search yielded 86 articles on PubMed and 130 articles from Scopus and additional 2 articles from other sources. After removing 7 duplicate records found in both databases, the remaining 211 articles were screened (Fig. 1). The full search strategy for PubMed and Scopus was: Studies were eligible if they: involved healthcare workers as the study population,reported prevalence for HRO, using a validated screening tool, such as STOP-Bang Questionnaire (SBQ) or Berlin Questionnaire (BQ) for OSA risk;reported prevalence for OSA, diagnosed through home sleep test, with AHI (Apnea-Hypopnea Index) criteria ≥ 5 events/hour for OSA diagnosis.were written in English or French.were published between 2000 and 2026. involved healthcare workers as the study population, involved healthcare workers as the study population, reported prevalence for HRO, using a validated screening tool, such as STOP-Bang Questionnaire (SBQ) or Berlin Questionnaire (BQ) for OSA risk; reported prevalence for HRO, using a validated screening tool, such as STOP-Bang Questionnaire (SBQ) or Berlin Questionnaire (BQ) for OSA risk; reported prevalence for OSA, diagnosed through home sleep test, with AHI (Apnea-Hypopnea Index) criteria ≥ 5 events/hour for OSA diagnosis. reported prevalence for OSA, diagnosed through home sleep test, with AHI (Apnea-Hypopnea Index) criteria ≥ 5 events/hour for OSA diagnosis. were written in English or French. were written in English or French. were published between 2000 and 2026. were published between 2000 and 2026. Studies that did not report OSA-related outcomes in healthcare workers were eliminated from the analysis. Two authors (M.A., G.S.) screened titles and abstracts, evaluated full texts, and extracted data. Extracted variables included sample size, job type, screening/diagnostic tool, prevalence of HRO, and adjustments for confounding variables. The prevalence of respiratory disorders is strongly influenced by the response rate in epidemiological studies14. To account for possible selection bias, the response percentage was also considered. The quality of retrieved articles was graded according to Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Studies Reporting Prevalence Data15. Five studies presented a moderate quality (score 6)16–20 and four a high quality (scores 7–9)21–24 (Table S1). The primary analysis included all the 9 articles, while a sensitivity analysis was performed on the four articles with the highest scores (Fig. S1). Finally, PRISMA checklist was filled-in and enclosed to the present manuscript as an online supplement (Table S2 for the abstract and Table S3 for the main body)25. Statistical methods A meta-analysis was performed using random-effects models to estimate pooled prevalence. To evaluate variability among studies, a test of heterogeneity was applied, and the I² statistic was computed, which indicates the proportion of total variation among the effect estimates of different studies attributed to heterogeneity rather than sampling error. As the test of heterogeneity was significant (p < 0.050) and I² was less than 30% in all analyses26,27, a random-effects model was used, and results were pooled using the DerSimonian and Laird method28. The level of statistical significance was set at 5%, and confidence intervals (CI) were calculated at 95%. The results were displayed graphically using forest plots. Sensitivity analysis was accomplished by considering only articles with the highest JBI score (7 or 9). As prevalence estimates can be affected by response percentages29, meta-regression was applied to address the relationship between these variables. Stata® software 18 (StataCorp LP, College Station, Texas, USA) was used for statistical analysis. Results The survey identified seven articles dealing with the prevalence of HRO among a total of 7642 HW workers17,18,20–24. Two articles investigated the same topic among 831 nurses16,19 (Table 1). All studies were cross-sectional, as expected for prevalence assessments. Fig. 1PRISMA 2020 flow diagram for updated systematic reviews which included searches of databases and registers only. PRISMA 2020 flow diagram for updated systematic reviews which included searches of databases and registers only. Risk of OSA in HealthCare workers The articles included in the systematic review and meta-analysis were recent, published in the last decade: 3 in 2016, 2 in 2019, and 1 in each of the following years: 2020, 2022, 2024, 2025. Two studies had been carried out in Europe21,24, two in the Middle East18,23, two in South-Eastern Asia17,22 and one in the US20. The studies were heterogeneous as regards gender and age. Indeed, the proportion of men ranged from 20 to 40%, but the Thai study involved only male HCW. The mean age was reported in only 4 studies, ranging from 34 years18 to 50 years24. The mean BMI reported by three studies ranged from 24.5 to 25.2 kg/m2. In addition, Spiteri et al., reported median BMI as a function of OS: 22.5, 25.2, 27.7 kg/m2 in HW with low, intermediate, and high risk of OSA, respectively24. OSA risk was assessed by the Berlin questionnaire in 6 studies and by the SBQ in 3 studies. Response percentage varied widely, from 5.7%20 to 95.5%19. Full information on the included articles is reported in (Table 1). Table 1Screening methodology across studies on OSA in healthcare Workers.CountryAuthor (year)SettingMean age, yearsGenderBMI, kg/m2Screening toolResponse percentage Healthcare professionals IranSeyedmehdi et al.23Hospital StaffF: 62.8%M:34.4%25.2Berlin Questionnaire715/800 = 89.4%ThailandArunsurat et al.17Male Health Workers47100% Male24.5Berlin Questionnaire135/273 = 49.5%FranceCadelis et al.21University Hospital42F: 67%M:32.7%Berlin Questionnaire773/3803 = 20.3%TurkeyAydın Guclu et al.18Healthcare workers34.94F:60%M:40%24.8Berlin Questionnaire604/850 = 71%USAPascoe et al.20Healthcare workersSTOP-BANG questionnaire, ESS2851/50,000 = 5.7%ItalySpiteri et al.24University Hospital50F:80.2%M:19.8%STOP-BANG & Home sleep test1564/5031 = 31.1%VietnamDuong-Quy et al.22Healthcare workersF:69.1%M:30.9%ESS*, STOP-BANG & home sleep test1721/2250 = 76.5% Nurses GreeceAlexandropoulou et al.16Nurses from secondary & tertiary hospitalF:41.4M:42.9F:87.4%M:12.6%F: 25.1M:27.2Berlin Questionnaire444/530 = 83.8%ItalyDe Nunzio et al.19Multicenter,Professional NursesF:75.4% M:24.5%25.4Berlin Questionnaire387/405 = 95.5%*ESS = Epworth sleepness scale. Table 1Screening methodology across studies on OSA in healthcare Workers.CountryAuthor (year)SettingMean age, yearsGenderBMI, kg/m2Screening toolResponse percentage Healthcare professionals IranSeyedmehdi et al.23Hospital StaffF: 62.8%M:34.4%25.2Berlin Questionnaire715/800 = 89.4%ThailandArunsurat et al.17Male Health Workers47100% Male24.5Berlin Questionnaire135/273 = 49.5%FranceCadelis et al.21University Hospital42F: 67%M:32.7%Berlin Questionnaire773/3803 = 20.3%TurkeyAydın Guclu et al.18Healthcare workers34.94F:60%M:40%24.8Berlin Questionnaire604/850 = 71%USAPascoe et al.20Healthcare workersSTOP-BANG questionnaire, ESS2851/50,000 = 5.7%ItalySpiteri et al.24University Hospital50F:80.2%M:19.8%STOP-BANG & Home sleep test1564/5031 = 31.1%VietnamDuong-Quy et al.22Healthcare workersF:69.1%M:30.9%ESS*, STOP-BANG & home sleep test1721/2250 = 76.5% Nurses GreeceAlexandropoulou et al.16Nurses from secondary & tertiary hospitalF:41.4M:42.9F:87.4%M:12.6%F: 25.1M:27.2Berlin Questionnaire444/530 = 83.8%ItalyDe Nunzio et al.19Multicenter,Professional NursesF:75.4% M:24.5%25.4Berlin Questionnaire387/405 = 95.5%*ESS = Epworth sleepness scale. Screening methodology across studies on OSA in healthcare Workers. F: 62.8% M:34.4% F: 67% M:32.7% F:60% M:40% F:80.2% M:19.8% F:69.1% M:30.9% F:41.4 M:42.9 F:87.4% M:12.6% F: 25.1 M:27.2 Multicenter, Professional Nurses *ESS = Epworth sleepness scale. Prevalence estimates for HRO ranged from 7%23 to 37%20. The pooled estimate of the prevalence of HRO among HCW was 23%. The corresponding confidence interval was quite large (95% CI 13–33%) with an extremely high heterogeneity among studies (I-squared = 99.1%) (Fig. 2). Fig. 2Forest plot on the prevalence of high OSA risk among Healthcare professionals. The meta-analysis was performed using a random-effects model. Forest plot on the prevalence of high OSA risk among Healthcare professionals. The meta-analysis was performed using a random-effects model. The prevalence estimate for HRO did not change when considering the four articles with the highest JBI scores (7 or 9). The pooled estimate was 20% (95% CI 9–31%) and the I-squared was 99.0% (Fig. S1). Meta-regression showed a weak negative correlation between prevalence estimates of HRO, normalized through logit transformation, and response percentage (Fig. 3). However, the correlation was minor (adjusted R-squared = 6.73%) and not significant (p = 0.281). To address heterogeneity, studies were grouped according to the instrument used (SBQ versus BQ) and by geographic area (Europe, Middle East, South-Eastern Asia, or US). Pooled estimate of HRO was higher in studies using the SBQ (29%, 95% CI 12–45%) than the BQ (19%, 95% CI 9–28%), but the difference was not significant (p = 0.29) (Fig. 4). On the other hand, prevalence estimates were significantly affected by geographic area (p < 0.01), being lower in the Middle East and Europe, and higher in South-East Asia and the United States (Fig. 5). Fig. 3Meta-regression of the prevalence estimates for OSA high risk on the corresponding response percentage in the seven studies dealing with healthcare professionals. The logit transformation was used to normalize the prevalence distribution. Meta-regression of the prevalence estimates for OSA high risk on the corresponding response percentage in the seven studies dealing with healthcare professionals. The logit transformation was used to normalize the prevalence distribution. Fig. 4Forest plot presenting prevalence of high risk of OSA among HCW as a function of type of questionnaire (Berlin versus STOP-BANG). Forest plot presenting prevalence of high risk of OSA among HCW as a function of type of questionnaire (Berlin versus STOP-BANG). Fig. 5Forest plot presenting prevalence of high risk of OSA among HCW as a function of geographic area (Europe, Middle East, South-East Asia, United States). Forest plot presenting prevalence of high risk of OSA among HCW as a function of geographic area (Europe, Middle East, South-East Asia, United States). Only one study investigated HRO by job title. It was more prevalent among administrative staff (20.3%) than among health technicians (19%), physicians (13.4%) and nurses (10.5%)24. Prevalence of OSA diagnosis among healthcare workers One study directly reported not only the prevalence of HRO but also the prevalence of OSA itself, diagnosed by home sleep test, while another study provided the information necessary to estimate it (prevalence of HRO * PPV = 0.136*0.688 = 0.093)22,24. The prevalence of OSA diagnosis ranged from 9.3 to 21.9% (Table 2). Table 2Prevalence estimate for OSA diagnosis, posed on the basis of home sleep test.CountryAuthor (year)SettingResponse percentagePrevalence of OSAItalySpiteri et al.24University Hospital1564/5031 = 31.1%9.3%VietnamDuong-Quy et al.22Healthcare workers from several hospitals1721/2250 = 76.5%21.9% Table 2Prevalence estimate for OSA diagnosis, posed on the basis of home sleep test.CountryAuthor (year)SettingResponse percentagePrevalence of OSAItalySpiteri et al.24University Hospital1564/5031 = 31.1%9.3%VietnamDuong-Quy et al.22Healthcare workers from several hospitals1721/2250 = 76.5%21.9% Prevalence estimate for OSA diagnosis, posed on the basis of home sleep test. Discussion OSA, reducing attention, memory, concentration, and alertness, is associated with increased frequency and severity of WRI7,8. While several studies examined OSA prevalence among professional drivers7,8,30,31, data on prevalence among other job categories are scant, and we found only 9 articles on the prevalence of OSA among HW published since 2000. Prevalence Meta-analysis showed an overall prevalence of OSA of 23%. This data is significantly lower than the worldwide OSA prevalence estimated for the adult population (excluding the elderly) in the meta-analysis by de Araujo Dantas (54%)1. On the other hand, the estimates reported by Benjafield et al., which include only adults aged 30–69, are only slightly higher than ours. Indeed, they calculated that worldwide 936 million people belonging to this category are affected by mild to severe OSA, accounting for 25% of the whole population13. In any case, in some work categories such as HW, a lower prevalence, as compared to the general population, could be expected due to a higher awareness of health issues and to the healthy worker effect, which assumes that individuals with severe diseases are less likely to be employed32. However, the same trend is not observed across all job categories. Indeed, estimates for some professional groups are even higher than in the general population (e.g., bus drivers: 78%, truck drivers: 87%, shift workers: 84%). It is therefore plausible that the key factor is that HW are more aware of health issues and lead healthier lifestyles, thereby being less prone to comorbidities that increase the OSA risk. Another important issue is that the results showed significant heterogeneity among studies, ranging from 7 to 37%. This effect could be mainly due to the factors described previously (sample sizes, selection bias, and detection method). On the other hand, although they were conducted in different countries (Italy and Turkey) and used different questionnaires (SBQ and BQ), the studies by Spiteri et al., and Guclu et al., reported a similar prevalence (15 and 14%, respectively)18,24. A possible explanation is that, despite differences due to ethnic characteristics, both studies included large populations (1564 and 604 subjects) and all HW job categories. Among the studies, HRO prevalence was reported only for the entire healthcare worker population, except for the study by Spiteri et al., which also examined HRO by occupational category and found significant differences. In fact, they found the highest values among administrative staff and technicians (20.3% and 18.9%), while nurses had the lowest prevalence (10.5%)24. These data highlight the relevance of considering job titles in analyses to prioritize screening policies and targeted health surveillance for the highest-risk categories. Data quality Data quality was non-optimal across all studies for the following reasons: (1) Selection bias and sample size; (2) Methods used for detecting OSA. Selection bias and sample size The study by Geiger-Brown included only a very small sample of full-time, 12-hour-shift female nurses (n = 21)33. The same critical issues are shared by the study by Yasin, which involved only 52 subjects34 who, moreover, were aged less than 40 years. De Nunzio et al., and Alexandropoulou et al., collected data only from nurses, excluding other job categories16,19. As a result, the data are not representative of the entire healthcare workforce. One of the major limitations of other 3 studies was the low response rate (range 6–31%), which significantly affected the representativeness of the results (Cadelis and Fayad Y Monteagudo 2016;20,24. Arunsurat et al., enrolled only male HW, so no data on prevalence among females, who are the majority of the labour force in health settings, are available17. Only 3 studies seem not to be influenced by selection bias18,22,23. Indeed, they were characterized by large sample sizes (604–1720 subjects), included all hospital job categories, and achieved solid response rates. Methods used for detecting OSA Another important limitation to the comparability of the studies was the variability in the methods used to detect OSA cases. They included screening questionnaires (SBQ, BQ), and a home sleep test. The major difference between questionnaires and instrumental tests is that the former only classify subjects as high or low risk for OSA, whereas the latter directly measure apnoea episodes, making them a more accurate and reliable method22. Moreover, no validated questionnaires are available for healthcare occupational contexts. Indeed, while their effectiveness as a screening tool has been demonstrated in clinical settings, especially among patients with multimorbidity, data on specific working populations are very scarce. Geiger-Brown investigated the suitability of the BQ in a small sample of nurses and found a sensitivity of 0.33 and a specificity of 0.83, confirming that this tool is less useful in a low-prevalence population than in sleep, cardiovascular, bariatric, and pulmonary clinics33. In these contexts, the sensitivity ranged from 0.62 to 0.9333. On the other hand, 2 studies included in this review, using the SBQ as a screening tool, reported moderate-to-high positive predictive values (61.2% and 68.8%)22,24. In particular, Spiteri et al., demonstrated that this value significantly improved when the questionnaire was administered by trained personnel (96.9%) compared with self-administration, suggesting that the instrument’s effectiveness can increase through specialized personnel intervention. Anyway, in both studies, the home sleep test was performed only in a subsample at HRO, potentially affecting the reliability of the results. Another key issue related to the use of questionnaires is the impact of demographic and ethnic characteristics on their predictability. Yasin et al., found that the true prevalence of OSA in a small sample of young HW without hypertension or obesity was twice the percentage of HRO HW identified by the SBQ34. The Authors assumed that the lack of comorbidity and the young age of the population could significantly reduce its negative predictive value, as well as other determinants not investigated by the questionnaire, such as the particular craniofacial structure of Asians, so that the usefulness of this tool in this specific setting would be reduced. Another crucial issue is the comparability of the results among different questionnaires. Chiu et al., conducted a meta-analysis to understand the sensitivity and specificity of each tool35. They found a significant difference between BQ and SBQ across all OSA severity levels. Indeed, Specificity (Sp) was 88%, 90%, and 93% for SBQ, and 76%, 77%, and 84% for BQ, in mild, moderate, and severe OSA, respectively. On the other hand, Sensitivity (Se) was 42%, 36%, and 35% for SBQ, and 59%, 44%, and 38% for BQ. Similar results were reported in another review and meta-analysis by Bernhardt36. These findings confirm that results from different studies cannot be directly compared. Moreover, they are derived from studies involving the general population and clinical patients, so there may be an even greater discrepancy in the occupational settings. To address this issue, Chen et al., and Seraratna et al., included studies on the validity of the SBQ and BQ in occupational populations in their reviews. Among commercial drivers, the SBQ showed a Se of 91% and a Sp of 43% for moderate-to-severe OSA37. The only 2 studies on the validity of BQ included in Seraratna’s meta-analysis were Geiger-Brown (nurses) and Firat (highway bus drivers). Both reported low Se (33% and 46%) and high Sp (83% and 85%)38. These data suggest that SBQ may be more reliable than BQ in occupational settings. However, performance does not appear optimal, suggesting the need to implement a specific screening questionnaire for workers. Only 4 studies used instrumental assessment to diagnose OSA. The research by Geiger-Brown et al., involved only a very small sample of nurses, so it cannot be considered representative of the entire healthcare population33. Similarly, the study by Yasin et al., enrolled only a few subjects (n = 52) and selected them based on age (< 40 years old)34. Spiteri et al., performed home sleep test in a subsample of HW who tested at HRO through SBQ and accepted to undergo the exam24. On the other hand, in the study by Sy Duong-Quy et al., all HW who were classified at high-risk via questionnaire completed home sleep tests (n = 358)22. However, neither study performed the diagnostic tests in low-risk subjects. This made it unfeasible to determine specificity, potentially leading to an underestimation of cases. Finally, as pointed out by Lechat et al., single night instrumental assessment could have affected the reliability of the tests. Indeed, they reported that a multinight in-home monitoring reduced misclassification by about 20%, highlighting that even the instrumental evaluation, in the lack of proven test reproducibility, does not guarantee sufficient accuracy for diagnostic purposes39. Perspectives Overall, the number of the studies on this topic is insufficient and their quality is not optimal. Moreover, the screening questionnaires used are not harmonized or validated for this specific work population. In the last few years, new digital tools that could improve the effectiveness of screening policies have been released. They are affordable, easy to wear and use, and reporting them is less time-consuming than reporting polygraphies. A systematic review by Duarte et al., investigated digital tools for this purpose and reported that electronic devices, such as smartwatches, smart pillows, and microphones, showed good to excellent sensitivity and specificity when AHI > 30 was used as a cut-off40. On the other hand, a recent review by Pinilla et al., focused on wearables (e.g. wrist and finger sensors), nearables (e.g. bedside sensors), and airables devices (e.g. audio and video recordings), showing that they could be a reliable alternative to assess sleep breathing disorders also in mild and moderate OSA. Indeed, very high sensitivity and specificity were reported for the majority of these devices, even if the AHI detected throughout polysomnography was lower than 10 events per hour41. However, the quality of studies affected the reliability of these results, in particular the small sample sizes, and the lack of multiple studies using the same device. To overcome these issues, high-quality research on this topic is needed, and outcome studies in large, diverse populations, including HW, should be urgently conducted42. Considering the potential impact on the safety of healthcare workers and patients, this research is particularly crucial in the healthcare sector. The role of occupational physicians is fundamental in this effort, as they can evaluate the working population through preventive and periodic health surveillance, for example, by collecting clinical data on the prevalence of HRO using validated questionnaires and prescribing instrumental tests to confirm or not the diagnosis. These evaluations are essential for early diagnosis and for identifying a therapeutic approach to address clinical symptoms and related cognitive deficits, with the aim of ensuring occupational health and safety, as well as a comprehensive appraisal of fitness for work, in line with the evolving role of occupational physicians in healthcare settings. Strengths and limitations To the best of our knowledge, this review is the first to focus on the prevalence of HRO among HW. We analyzed the results of the included studies, taking into account differences in HRO assessment, target population and sample size. This approach enabled us to highlight some weaknesses of the same studies, and will be useful in designing new investigations that could achieve a higher degree of scientific reliability. On the other hand, the very low number of articles available limits the power of the review’s findings. Therefore, it should be reminded that the estimates of HRO prevalence and its determinants among HW should be interpreted with caution. Conclusion Data on the prevalence and the influence of OSA on WRI among HW are insufficient and methodologically flawed. Therefore, rigorous research is needed to address open issues. To achieve this goal, researchers should focus on validating screening questionnaires in HW populations. In particular, it is crucial to evaluate not only the sensitivity but also the specificity of the tool through home sleep test, even in subjects who tested negative, and to investigate the role of OSA in WRI risk. New digital tools, such as smartwatches, smart pillows, and microphones, could improve the effectiveness of screening policies, also in occupational settings. However, before their use as screening tools, their suitability and usefulness need to be evaluated in large working populations, with measurement in real work contexts. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4