Comparative Treatment Strategies for Obstructive Sleep Apnea and Their Relationships with Aberrant Driving Behaviors: Taxi Drivers in Taipei Metropolitan Area Authors: Zhihe Chen, Wen-Te Liu, Arnab Majumdar, Wen-Hua Hsu, Yi-Chun Kuan, Jiunn-Horng Kang, Yen-Ling Chen, Rachel Chien, Peizheng Wang, Kang-Yun Lee, Po-Hao Feng, Kuan-Yuan Chen, Hsin-Chien Lee, Yi-Chih Lin, Wun-Hao Cheng, Cheng-Yu Tsai Published: 2026 ABSTRACT AbstractStudy ObjectivesObstructive sleep apnea (OSA) is prevalent among professional drivers and is associated with an increased risk of aberrant driving behaviors (ADBs). This study evaluated and compared the effectiveness of continuous positive airway pressure (CPAP) and conservative treatments in mitigating ADBs and improving sleep health among urban taxi drivers.MethodsIn this prospective non-randomized comparative study, taxi drivers diagnosed with OSA through polysomnography were assigned to either the CPAP group or the conservative treatment group. Home-based sleep parameters (oxygen desaturation index [ODI], cyclic variation of the heart rate index [CVHRI], and chest effort index [CEI]) were measured using wearable devices. Driving data were recorded with a validated navigation application. Subjective measures such as Pittsburgh Sleep Quality Index (PSQI), Epworth Sleepiness Scale (ESS), and the 27-item Driving Behavior Questionnaire (DBQ) were also collected. Pre- and post treatment outcomes were compared within and between groups using non-parametric statistical tests following the 3-month treatment.ResultsTwenty-four taxi drivers were enrolled, with 13 in the CPAP group and 11 in the conservative treatment group. CPAP treatment significantly improved objective and subjective measures of sleep health and driving performance, evidenced by decreases in PSQI (p=0.04), ESS (p=0.01), and DBQ errors (p<0.01) and lapses (p=0.03). Additionally, the occurrence of ADBs per 20 km decreased (p<0.01), and sleep disorder indices improved (ODI, p<0.01; CEI, p=0.03).ConclusionThe conservative treatment group showed no significant changes in these measures. CPAP demonstrated its potential as a primary intervention for occupational drivers with severe OSA. Conservative treatment showed limited effectiveness and could serve as a supplementary approach.Keywords: continuous positive airway pressure, CPAP, sleep hygiene, obstructive sleep apnea, OSA, aberrant driving behavior, ADB, taxi driver Introduction Excessive daytime sleepiness is frequently associated with obstructive sleep apnea (OSA), which is defined as repeated nighttime respiratory disruptions due to upper airway obstruction.1 Fatigue and daytime sleepiness levels tend to be higher among individuals with OSA compared to those without.2 There is a significant prevalence of OSA among professional drivers, which substantially increases their risks of driving while fatigued and traffic crashes.3 For instance, taxi drivers diagnosed with severe OSA exhibited significantly higher occurrences of aberrant driving behaviors (ADBs), including speeding, abrupt steering, aggressive acceleration, and hard braking, compared to those without.4 Scores of the post-driving Karolinska Sleepiness Scale (KSS), which indicates levels of sleepiness, were significantly higher (mean score of 8.20 ± 0.40) for highway bus drivers with ADBs compared to those without ADBs (mean score of 7.20 ± 0.75), suggesting that increased sleepiness was associated with a higher risk of ADBs.5 These findings underscore the adverse impacts that OSA can have on road safety. Although relationships of fatigue and daytime sleepiness with OSA have been extensively studied, their specific impacts on taxi drivers have received less attention. The unique demands of this occupation render taxi drivers particularly susceptible to OSA. Prolonged periods of sedentary behavior and limited opportunities for physical exercise contribute to obesity, a well-established risk factor for OSA. In addition, long, irregular, and often nocturnal shifts disrupt circadian rhythms and reduce overall sleep duration.6 For instance, taxi drivers in New York City were found to be more likely than the average American male to sleep fewer than seven hours per night, and those working night shifts reported markedly worse sleep disturbance scores.7 While these occupational vulnerabilities are also present among other groups of commercial drivers, they tend to be more pronounced and less effectively managed among taxi drivers, thereby amplifying both their risk of OSA and its associated safety consequences. Given the heightened risk of crashes associated with OSA among this group, addressing OSA in taxi drivers could be a promising strategy for enhancing road safety. Continuous positive airway pressure (CPAP) has been widely utilized to mitigate the adverse impacts of hypoxia on cognitive function and improve sleep efficiency by eliminating respiratory events.8 Moreover, CPAP may reduce driving errors and lapses related to deficits in cognitive function and fatigue. A previous study showed that CPAP treatment can mitigate OSA symptoms, potentially enhancing transportation safety for drivers by significantly reducing incidences of traffic collisions post-treatment.9 For instance, a review of nine observational studies assessing the crash risk of drivers with OSA before and after CPAP treatment demonstrated a significant reduction in risk post-treatment (risk ratio [RR] = 0.28, 95% confidence interval [CI]: 0.22 to 0.35; p < 0.001).10 The aforementioned studies on treating OSA to enhance driving performance primarily focused on CPAP, the first-line treatment for moderate to severe OSA, despite low adherence rates due to discomfort and work-related factors. Taxi drivers with OSA often struggle with CPAP because of demanding schedules, short and fragmented sleep, discomfort, and fear of job loss, contributing to low long‑term adherence to CPAP.11 Alternative conservative treatment strategies, such as enhancing sleep hygiene, lifestyle modifications, and pharmacological support (eg, nasal sprays), were demonstrated to effectively improve OSA severity.12 Precisely, adhering to good sleep hygiene practices, such as consistent schedules with sufficient sleep, environmental adjustments, limits on alcohol and caffeine, and incorporation of exercise and nutrition, can alleviate symptoms for patients with mild to moderate OSA.13 Another study indicated that a 1-year lifestyle intervention yielded sustained improvements in the apnea-hypopnea index (AHI) and reduced OSA symptoms over a 2-year follow-up period.14 As conservative interventions are generally less intrusive and more cost-effective than CPAP therapy, they may be particularly suitable for occupational drivers who face challenges to treatment adherence. However, the efficacy of both CPAP and conservative treatment specifically for taxi drivers in naturalistic driving scenarios requires further validation. We examined the association between OSA and ADBs, and assessed treatment strategies for drivers diagnosed with OSA. We hypothesized that taxi drivers with OSA might be at increased risk of presenting ADBs, and treating OSA could potentially help reduce the risk of presenting ADBs. This study employed CPAP and conservative treatments (ie, sleep hygiene enhancement, lifestyle modifications, and pharmacological support) for taxi drivers diagnosed with OSA for a span of 3 months. Following this period, both subjective and objective data were collected and analyzed to compare outcomes between pre- and post-treatment. These data may elucidate the benefits of various OSA treatment strategies in taxi drivers to improve subjective sleep quality, reduce fatigue levels, and potentially mitigate the risks of presenting ADBs to enhance road safety. Materials and Methods Research Ethics Participants were primarily recruited from the Taipei metropolitan area via taxi associations and public notices. Prior to commencing data collection, this study received approval from the Ethics Committee of the Taipei Medical University-Joint Institutional Review Board (TMU-JIRB: N202112077). This study was conducted in accordance with the principles of the Declaration of Helsinki. Prospective participants were informed of the study’s objectives, associated risks, and their right to withdraw, and were subsequently asked to provide written informed consent before initiating this study. All derived data were analyzed and preserved under the approved guidelines and privacy regulations. Study Protocol Figure 1 illustrates the flow of data collection and processing. The enrollment period of this prospective non-randomized comparative study was from February 2021 to July 2023. The inclusion criteria were as follows: (1) aged 20–65 years, (2) diagnosed with OSA by polysomnography (PSG), with an AHI of ≥ 5 events/h, (3) no prior treatment for OSA, and (4) held a commercial driving license for over 2 years. Participants were initially directed to the sleep center of Taipei Medical University-Shuang Ho Hospital (New Taipei City, Taiwan) to undergo PSG for an OSA diagnosis and were instructed to complete the baseline demographic survey. Due to the time required to process PSG scoring, participants typically had to wait 1 to 2 weeks for the results. This period was therefore designated as the pretreatment phase for data collection in this study. During this phase, participants were provided with a mobile phone equipped with the navigation application (app) UNIGO+ (version 2.0.18; TRADE-VAN Information Services, Taipei, Taiwan) to collect trajectory data and details of ADB events over a period of 3–4 consecutive days. The navigation app used in this study has been extensively described and validated in prior publications, ensuring accurate trajectory data and effective identification of ADB events such as speeding, abrupt steering, aggressive acceleration, and hard braking.15,16 For nighttime data collection, single-lead electrocardiogram (ECG) monitors (Rooti Rx System, Rooti Labs, Taipei, Taiwan) and oximetry wristwatches (ZT-Z2-B; ZoeTek, Taipei, Taiwan) were employed to record home-based sleep parameters over the same duration. After clinical consultation and patient-physician decision, treatment options were assigned for participants, a 3-month treatment regimen commenced. Subsequently, the same data collection procedure was repeated for another consecutive 3–4 days to capture data from the post-treatment phase. To consider factors such as weather conditions and congestion levels that may affect ADB occurrence, this study also incorporated public data on meteorological conditions (Central Weather Bureau, Taipei, Taiwan) and road congestion levels (vehicle detector data and public road cameras from the Freeway Bureau, Taipei, Taiwan). All of the derived data were used for group comparison analyses and subgroup analyses. Flowchart of data collection and processing. Questionnaires All participants completed a baseline demographic survey assessing body profiles, history of cigarette and alcohol use, sleep quality (eg, duration, latency, and sleep efficiency), driving-related details (eg, driving experience, working hours per day), and driving fatigue (eg, self-perceived reasons for fatigue, fatigue level during driving last week). Next, after completing 3–4 days of naturalistic driving tasks in both the pretreatment and post-treatment phases, participants were scheduled for an interview to complete the sleep-related, driving behavior, and subjective fatigue questionnaires. The extended 27-item Driving Behavior Questionnaire (DBQ), which comprises four categories (aggressive violations, ordinary violations, errors, and lapses), was employed to assess personality and driving behaviors.17 For the sleep-related questionnaires, the Epworth Sleepiness Scale (ESS),18 Pittsburgh Sleep Quality Index (PSQI),19 Patient Health Questionnaire (PHQ),20 and Insomnia Severity Index (ISI)21 were administered to collect subjective details. The license to use the ESS was obtained for this study, and all questionnaires were administered in validated Taiwanese Mandarin by trained research staff via interview-assisted self-administration. The outcomes of these questionnaires served to evaluate self-reported sleep quality, insomnia severity, and symptoms of OSA, including snoring or excessive daytime sleepiness. The KSS was applied to assess the subjective fatigue level at three time points (ie, before driving, in the middle of driving, and after driving). Respondents were instructed to provide a score that represented their mental fatigue state (ranging from 0 = “extremely alert” to 5 = “neither alert nor sleepy” and 9 = “extremely sleepy”). Hospital-Based and Home-Based Sleep Parameters The severity of OSA was assessed using the AHI obtained through PSG, which is regarded as the gold standard for diagnosing OSA. The AHI quantifies the average number of apnea and hypopnea events occurring per hour of sleep and serves as the primary metric for evaluating disease severity.22 According to the American Academy of Sleep Medicine, OSA severity is categorized as mild (5≤AHI < 15 events/h), moderate (15 ≤ AHI < 30 events/h), and severe (AHI ≥ 30 events/h).23 Due to ongoing debates regarding the accuracy of AHI for measuring OSA severity,24 this study also employed two types of wearable devices to monitor home-based sleep parameters, which served as surrogate indices for the AHI. These devices included an oximetry wristwatch and a single-lead ECG monitor (chest patch) (Figure S1). The wristwatch measures oxygen saturation, enabling calculation of the oxygen desaturation index (ODI, ≥3%), which quantifies the average number of episodes per hour where oxygen saturation decreases by at least 3% from the baseline for a minimum of 10s.10 The chest patch device can record both continuous ECG data from the single-lead ECG and chest wall breathing movements via an integrated three-axis accelerometer. Two sleep disorder indices—cyclic variation of the heart rate index (CVHRI) and the chest effort index (CEI)—were derived from the heart rate variability (HRV) and chest wall movements, respectively. More technically, the CVHRI quantifies the average number of cyclic variations in heart rate events per hour, characterized by heart rate fluctuations associated with respiratory events,25 while the CEI measures the number of chest effort events per hour, identified by changes in chest wall motion during respiratory efforts against airway obstruction.26 Both the CVHRI and CEI have been validated with PSG to serve as surrogate indices for the AHI to assess the severity of OSA,27 Tsai,28 and both indices were used to evaluate CPAP and conservative treatment outcomes. OSA Treatment Participants diagnosed with OSA (ie, AHI ≥5 events/h) were scheduled for subsequent outpatient consultations to confirm selected treatment approaches. For those undergoing CPAP treatment, a titration study following American Academy of Sleep Medicine guidelines was conducted to determine the optimal CPAP pressure. Once established, participants underwent CPAP therapy using an iCH-Auto-2 device (Wellell, New Taipei City, Taiwan) for 3 months. For conservative treatment, a written brochure was provided that detailed strategies for sleep hygiene education, sleep environment adjustment, smoking cessation, alcohol and caffeine reduction, exercise and nutrition incorporation, pharmacological support (ie, nasal spray), and stress management. To support treatment efficiency and provide further support to increase treatment adherence, follow-up phone calls (twice a week) and outpatient services (once a month) were scheduled for all participants. Specifically, patients in the CPAP treatment group were instructed to use the CPAP device for at least 4 h on a minimum of 5 nights per week. For participants in the conservative treatment group, sleep hygiene education and medication usage consultations were provided during the weekly follow-up phone calls. Statistical Analysis All statistical analyses were conducted using Python (vers. 3.6.8) and the open-source library Scikit-learn (version 0.21.2). Collected data were initially divided into two groups based on the treatment strategies, namely the CPAP treatment group (group 1) and conservative treatment group (group 2). Subsequently, mean scores from the questionnaires, sleep parameters (both from home- and hospital-based settings), and details from recorded navigation (ie, driving data and ADB events) were compared between pre- and post-treatment for each group. Since the Shapiro–Wilk test indicated that the difference scores were not normally distributed (p < 0.05), Wilcoxon signed-rank test were applied for continuous or ordinal variables, and Chi-squared tests were used to analyze categorical variables. To evaluate differences in pre- and post-treatment variables and explore the efficacy of each treatment strategy, detailed paired comparisons were conducted. Next, a subgroup analysis was conducted by restricting the analysis to patients with severe OSA to assess whether the treatment effects were consistent within this subgroup. Results Demographics and Self-Reported Sleep Quality of Participants A total of 38 taxi drivers were initially approached, but 14 were not enrolled because they did not have OSA or were unwilling to undergo treatment (Figure 2). Twenty-four participants were enrolled, and no participants opted out during the subsequent data collection process. Thirteen (all male) were assigned for CPAP treatment, while 11 (10 males and 1 female) were assigned for conservative treatment. No statistically significant differences were observed in age, sex distribution, body mass index, neck and waist circumference, cigarette and alcohol usage, or outcomes of self-reported sleep quality between groups 1 and 2 (Table 1). Data flowchart of participant recruitment. Table 1Demographic Details of the Recruited Taxi DriversVariablesRecruited Taxi Drivers (N=24)PGroup 1 CPAPTreatment (N=13)Group 2 ConservativeTreatment (N=11)Age (years) a45.08 ± 7.9347.55 ± 10.440.52Sex (male/female) b13/012/10.99BMI (kg/m2) a30.32 ± 4.9228.06 ± 5.630.31Neck Circumference (cm) a40.86 ± 2.8439.09 ± 4.360.24Waist Circumference (cm) a103.62 ± 11.5694.91 ± 13.740.11Cigarette usage b0.18 None12 (50%)8 (33.33%) Current smoker1 (4.17%)3 (12.5%)Alcohol usage b0.99 None9 (37.6%)7 (29.17%) 1–3 times (month)4 (16.67%)4 (16.67%)Self-reported sleep quality b Duration (hours/day)0.97 4–66 (25%)5 (20.83%) 7–97 (29.17%)6 (25%) Latency (mins)0.13 <207 (29.17%)3 (12.5%) 20–294 (16.67%)2 (8.33%) 30–392 (8.33%)6 (25%) Sleep efficiency (%)0.07 <60%11 (45.84%)5 (20.83%) 60–80%2 (8.33%)3 (12.5%) >80%-3 (12.5%) Early Awakening (times/week)0.68 0–17 (29.17%)5 (20.83%) 2–36 (25%)6 (25%)Notes: Data are expressed as mean ± standard deviation or as number (percentage).a Mann–Whitney U-test was applied for statistical analysis (P-value). b Chi-square test is applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group.Abbreviation: BMI, body mass index. Demographic Details of the Recruited Taxi Drivers Notes: Data are expressed as mean ± standard deviation or as number (percentage).a Mann–Whitney U-test was applied for statistical analysis (P-value). b Chi-square test is applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group. Abbreviation: BMI, body mass index. Driving-Related Questionnaire Outcomes of Participants The analysis of the driving-related questionnaire outcomes revealed no statistically significant differences between groups 1 and 2 in terms of working hours or fatigue levels (Table 2). Specifically, 37.5% of group 1 and 20.83% of group 2 reported working 9–12 h/day. Similarly, 54.17% of group 1 and 41.67% of group 2 reported that fatigue affected their driving 1–2 times per week. Notably, sleep deficiency was identified as the primary reason for fatigue by 33.33% of group 1 and 25% of group 2. Table 2Driving-Related Questionnaire Outcomes of the Recruited Taxi DriversVariablesRecruited Taxi Drivers (N=24)PGroup 1 CPAPTreatment (N=13)Group 2 ConservativeTreatment (N=11)Driving-related detailsDriving experience (years)0.22 2–54 (16.67%)2 (8.33%) 5–101 (4.17%)4 (16.67%) >108 (33.33%)5 (20.83%)Working hours (/day)0.41 5–84 (16.67%)6 (25%) 9–129 (37.5%)5 (20.83%)Rest days (/month)0.35 <33 (12.5%)4 (16.67%) 3–56 (25%)2 (8.33%) 5–74 (16.67%)5 (20.83%)Traffic violation ticket (last year)0.43 0–1 times9 (37.6%)5 (20.83%) 2–4 times4 (16.67%)6 (25%)Road traffic crash (last year)0.99 0–1 times8 (33.33%)7 (29.17%) 2–4 times5 (20.83%)4 (16.67%)Self-reported driving fatigueFeel fatigued during driving (/week)0.39 0 times-2 (8.33%) 1–2 times13 (54.17%)9 (37.6%)Fall asleep during driving (/week)0.99 0 times3 (12.5%)3 (12.5%) 1–2 times10 (41.67%)8 (33.33%)Fatigue affecting driving (/week)0.93 0 times-1 (4.17%) 1–2 times13 (54.17%)10 (41.67%)Self-perceived reason for fatigue0.74 Sleep deficiency8 (33.33%)6 (25%) Excessive driving time4 (16.67%)3 (12.5%) Monotonous environment1 (4.17%)2 (8.33%)Notes: Data are expressed as number (percentage). Chi-square test was applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group. Driving-Related Questionnaire Outcomes of the Recruited Taxi Drivers Notes: Data are expressed as number (percentage). Chi-square test was applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group. Sleep Details in Hospital- and Home-Based Settings Group 1 exhibited significantly more severe OSA compared to group 2, highlighted by both hospital- and home-based sleep parameters (Table 3). Group 1 also had a significantly higher mean AHI compared to group 2 (group 1: 50.84 ± 16.25; group 2: 27.42 ± 16.79; p<0.01). Specifically, 50% of group 1 experienced severe OSA, with no cases of mild OSA, whereas in group 2, 16.67% experienced severe OSA and 8.33% had mild OSA (p=0.02). Home-based sleep parameters also showed a significantly higher ODI (group 1: 52.34 ± 35.09; group 2: 25.07 ± 12.6; p=0.04), CVHRI (group 1: 27.58 ± 17.08; group 2: 10.82 ± 6.6; p=0.02), and CEI (group 1: 21.22 ± 14.02; group 2: 7.44 ± 4.68; p=0.01) in group 1. Table 3Sleep Parameters of the Recruited Taxi Drivers in Hospital- and Home-Based SettingsVariablesRecruited Taxi Drivers (N=24)PGroup 1 CPAPTreatment (N=13)Group 2 ConservativeTreatment (N=11)PolysomnographySleep architecture Sleep efficiency (%)77.73 ± 13.4472.6 ± 13.150.36 Wake (% of SPT)10.31 ± 9.8113.76 ± 9.630.41 NREM (% of SPT)57.05 ± 33.2957.32 ± 28.420.28 REM (% of SPT)9.78 ± 7.5810.92 ± 8.070.72 WASO (mins)49.65 ± 30.6850.92 ± 25.670.86Sleep disorder indices (events/h) AHI50.84 ± 16.2527.42 ± 16.79<0.01 ArI33.08 ± 14.9321.56 ± 13.60.06 Snoring index398.51 ± 207.92249.66 ± 239.20.12Wearable devicesSleep disorder indices (events/h) ODI52.34 ± 35.0925.07 ± 12.60.04 CVHRI27.58 ± 17.0810.82 ± 6.60.02 CEI21.22 ± 14.027.44 ± 4.680.01OSA severity (n, %)0.02 Mild–2 (8.33%) Moderate1 (4.17%)5 (20.83%) Severe12 (50%)4 (16.67and)Notes: Data are expressed as mean ± standard deviation. Mann–Whitney U-test was applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group.Abbreviations: NREM, non-rapid eye movement; REM, rapid eye movement; SPT, Sleep period of time; WASO, wake time after sleep onset; AHI, apnea-hypopnea index; ArI, arousals index; Oxygen desaturation index (≥3%); CVHRI, cyclic variation of heart rate index; CEI, chest effort index. Sleep Parameters of the Recruited Taxi Drivers in Hospital- and Home-Based Settings Notes: Data are expressed as mean ± standard deviation. Mann–Whitney U-test was applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group. Abbreviations: NREM, non-rapid eye movement; REM, rapid eye movement; SPT, Sleep period of time; WASO, wake time after sleep onset; AHI, apnea-hypopnea index; ArI, arousals index; Oxygen desaturation index (≥3%); CVHRI, cyclic variation of heart rate index; CEI, chest effort index. Outcome Comparisons Between Pre- and Post-Treatment After 3 months of treatment, group 1 had significantly lower mean PSQI (pretreatment: 9.77 ± 3.85; post-treatment: 7.08 ± 3.88; p=0.04) and ESS scores (pre-treatment: 12.46 ± 4.81; post-treatment: 7.69 ± 6.03; p=0.01) (Table 4). KSS scores decreased significantly both during (pretreatment: 6.15 ± 0.9; post-treatment: 4.38 ± 1.56; p<0.01) and after driving (pretreatment: 7.46 ± 1.2; post-treatment: 5.0 ± 1.87; p<0.01) in group 1. Conversely, group 2 showed a significant increase in KSS scores in the middle of driving (p=0.03) and after driving (p=0.03). Regarding the subjective evaluation of driving behaviors, group 1 demonstrated significant reductions in DBQ scores for errors (p<0.01) and lapses (p=0.03), whereas group 2 exhibited no significant change in any driving-related questionnaire score. Table 4Comparison of Questionnaire Outcomes Between Pre- and Post-TreatmentCategorical VariablesGroup 1CPAP Treatment (N=13)PGroup 2 Conservative Treatment (N=11)PPre-TreatmentPost-TreatmentPre-TreatmentPost-TreatmentSleep-related Questionnaires PSQI (scale from 0 to 3)9.77 ± 3.857.08 ± 3.880.049.09 ± 2.958.73 ± 2.940.94 ESS (scale from 0 to 3)12.46 ± 4.817.69 ± 6.030.019.64 ± 4.789.45 ± 7.240.63 ISI (scale from 0 to 4)9.77 ± 4.258.85 ± 5.840.4411.45 ± 5.479.55 ± 5.240.14 PHQ (scale from 0 to 1)7.38 ± 6.475.92 ± 5.60.125.64 ± 4.674.36 ± 4.570.26KSS (scale from 0 to 9) Before driving4.69 ± 0.484.08 ± 1.320.144.18 ± 0.874.82 ± 0.980.11 In the middle of driving6.15 ± 0.94.38 ± 1.56<0.014.73 ± 1.566.27 ± 1.560.03 After driving7.46 ± 1.25.0 ± 1.87<0.015.82 ± 1.897.36 ± 1.360.03DBQ (scale from 0 to 6) Aggressive violations (3 item)2.23 ± 1.542.38 ± 2.930.733.55 ± 2.584.01 ± 2.720.39 Ordinary violations (8 item)13.15 ± 4.614.46 ± 5.640.1913.27 ± 5.8513.09 ± 4.780.88 Errors (8 item)9.08 ± 3.255.62 ± 2.4<0.017.18 ± 3.47.27 ± 2.570.77 Lapses (8 item)7.92 ± 4.094.46 ± 3.180.036.36 ± 3.615.27 ± 2.10.63Notes: Data are expressed as mean ± standard deviation. Wilcoxon signed-rank test was applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group.Abbreviations: PSQI, Pittsburgh sleep quality index; ESS, Epworth sleepiness scale; ISI, insomnia severity index; PHQ, patient health questionnaire; KSS, Karolinska sleepiness scale; DBQ, driving behaviour questionnaire. Comparison of Questionnaire Outcomes Between Pre- and Post-Treatment Notes: Data are expressed as mean ± standard deviation. Wilcoxon signed-rank test was applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group. Abbreviations: PSQI, Pittsburgh sleep quality index; ESS, Epworth sleepiness scale; ISI, insomnia severity index; PHQ, patient health questionnaire; KSS, Karolinska sleepiness scale; DBQ, driving behaviour questionnaire. The number of ADBs per 20 km was significantly reduced after the 3-month treatment in group 1 (pretreatment: 1.19 ± 1.41; post-treatment: 0.29 ± 0.15; p<0.01), while nonsignificant changes were observed in driving durations and distances (Table 5). Home-based sleep parameters revealed significant decreases in the ODI (pretreatment: 48.43 ± 32.67; post-treatment: 27.04 ± 21.7; p<0.01) and CEI (pretreatment: 16.23 ± 11.03; post-treatment: 8.29 ± 6.96; p=0.03). In the subgroup analysis of patients with severe OSA, the results mirrored the overall group analysis, with significant improvements observed in group 1 for both driving performance (eg, reduced ADB occurrence) and sleep health parameters (eg, PSQI, ESS, ODI, and CEI), while no significant changes were noted in group 2 (Tables S1 and S2). Regarding the types of ADBs, hard braking remained the dominant ADB for both groups pre- and post-treatment (Table S3). Table 5Comparison of Aberrant Driving Behaviour and Home-Based Sleep Parameters Between Pre- and Post-Treatment (Various Strategies)VariablesGroup 1 CPAP Treatment (N=13)PGroup 2 Conservative Treatment (N=11)PPre-TreatmentPost-TreatmentPre-TreatmentPost-TreatmentRecorded day trip43404036Driving time (hr/day)11.24 ± 4.0512.75 ± 3.830.219.15 ± 5.4211.82 ± 5.070.33Driving distance (km/day)112.63 ± 71.57143.83 ± 48.040.0673.3 ± 52.3493.97 ± 36.390.25ADB related detailADB event (n)1666610088ADB occurrence (n/20 km)1.19 ± 1.410.29 ± 0.15<0.011.24 ± 1.130.57 ± 0.30.08Latency of ADB (mins)85.26 ± 92.5118.79 ± 162.740.5966.29 ± 33.78139.79 ± 125.610.17Accumulated distance of ADB (km)16.19 ± 15.5730.18 ± 31.630.1316.88 ± 14.6521.36 ± 11.670.24Home-based parameter (event/hr)ODI48.43 ± 32.6727.04 ± 21.7<0.0124.13 ± 8.8430.86 ± 19.660.28CVHRI19.06 ± 13.3818.92 ± 25.280.2410.55 ± 6.2210.46 ± 7.590.99CEI16.23 ± 11.038.29 ± 6.960.036.38 ± 3.435.91 ± 2.60.58Notes: Data are expressed as mean ± standard deviation. Wilcoxon signed-rank test was applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group. Latency of ADB: Time since the last 30-minute rest until ADB occurs; Accumulated distance of ADB: Distance travelled since the last 30-minute rest to ADB occurrence.Abbreviations: ADB, aberrant driving behaviour, ODI, oxygen desaturation index (≥3%); CVHRI, cyclic variation of heart rate index; CEI, chest effort index. Comparison of Aberrant Driving Behaviour and Home-Based Sleep Parameters Between Pre- and Post-Treatment (Various Strategies) Notes: Data are expressed as mean ± standard deviation. Wilcoxon signed-rank test was applied for statistical analysis (P-value). Bold P values indicate statistical significance at P < 0.05. N denotes the number of participants in each treatment group. Latency of ADB: Time since the last 30-minute rest until ADB occurs; Accumulated distance of ADB: Distance travelled since the last 30-minute rest to ADB occurrence. Abbreviations: ADB, aberrant driving behaviour, ODI, oxygen desaturation index (≥3%); CVHRI, cyclic variation of heart rate index; CEI, chest effort index. Discussion This study investigated the effectiveness of different OSA treatment strategies in improving both subjective and objective outcomes in taxi drivers. Twenty-four participants were assigned to two groups based on clinical consultation and patient-physician decision: a CPAP group and a conservative treatment group. Pretreatment and post-treatment data for each group were compared to evaluate the effectiveness of each treatment strategy in reducing ADBs, improving home-based sleep disorder indices, enhancing self-reported driving behavior, and alleviating sleep-related fatigue symptoms. The baseline demographic and sleep characteristics suggested that group 1 had moderate to severe OSA while group 2 had mild to severe OSA, and both groups identified sleep deficiency as the primary reason for driving fatigue. The current results seem to be indicative of the fact that inherent occupational challenges faced by taxi drivers make them especially vulnerable to OSA. The extended and unpredictable working hours, frequent night shifts that disrupt circadian rhythms, and, most notably, the sedentary nature of the job, which limits physical exercise and causes weight gain, contribute to their vulnerability to OSA.29 These occupational factors expose taxi drivers to complex driving environments and extended periods on the road, both of which contribute to sleep deprivation. With regard to sleep deficiencies and vulnerability to OSA, our findings confirmed those of previous studies indicating that driving fatigue in taxi drivers is affected by their occupational challenges. CPAP treatment led to significant improvement in objective and subjective evaluations of OSA-related symptoms and driving performance. Regarding subjective evaluations, group 1 demonstrated significant reductions in PSQI, ESS, and KSS scores, indicating improvements in sleep quality, daytime sleepiness, and driving fatigue after 3 months of treatment. Notably, baseline sleep parameters obtained from PSG and home settings indicated that group 1 had more-severe OSA than group 2, aligning with clinical guidelines that advise patients with moderate to severe OSA to consider active interventions such as CPAP.30 The observed improvements in OSA-related symptoms in group 1 may be partially attributed to the effects of CPAP, which eliminates nocturnal oxygen desaturation and enhances sleep efficiency by reducing respiratory events during sleep.31 By reducing respiratory events and nocturnal oxygen desaturation, CPAP may preserve sleep continuity and alleviate daytime sleepiness and fatigue. Moreover, the significant reductions in DBQ scores for errors and lapses in group 1 highlight the cognitive benefits of CPAP, potentially due to increased nocturnal oxygenation and prevention of hypoxemia-related brain damage.8 These findings are consistent with a previous study that demonstrated that OSA may lead to cognitive deficits, while using CPAP may help correct those deficits. Next, the observed improvements in clinical symptoms and self-reported driving behaviors in group 1 suggest that CPAP can potentially enhance the overall health and driving safety of individuals with OSA, emphasizing its importance as a primary treatment option. Regarding objective measurements, our study provides evidence of the effectiveness of CPAP treatment in alleviating OSA-related symptoms, as group 1 showed significant decreases in the ODI and CEI from pre- to post-treatment. CPAP enables reductions in proinflammatory cytokines and nighttime acid contact that likely contribute to improved airway conditions and reduced sleep disorder indices.32 Furthermore, group 1 demonstrated a significant reduction in ADBs per 20 km, while nonsignificant changes in daily driving durations and distances were observed. We acknowledge that the observed improvements in driving behavior may have been partially influenced by increased participant self-awareness arising from study participation and repeated monitoring. However, the concurrent improvements in objective sleep disorder indices suggest that the reduction in ADBs was not solely attributable to behavioral modification. These findings suggest that CPAP treatment was effective in mitigating aberrant driving risks associated with OSA over a 3-month treatment duration. Previous studies supported these findings, showing reduced traffic collision risks in real-world scenarios and improved driving performance in simulated driving after CPAP treatment.33 Considering the imbalanced distribution of OSA severity between treatment groups, subgroup analysis focusing only on patients with severe OSA yielded consistent results regarding significant improvement in the aforementioned measurements, further validating the treatment effects of CPAP. Taken together, these findings highlight the potential of CPAP as a primary treatment option for occupational drivers with severe OSA. Importantly, the effectiveness of CPAP is highly dependent on treatment adherence. Although CPAP demonstrated significant improvements in both sleep-related symptoms and driving performance in the present study, previous research has consistently shown that long-term compliance remains a major challenge among occupational drivers due to irregular work schedules, fragmented sleep opportunities, device discomfort, and concerns regarding employment consequences following OSA diagnosis.34 Therefore, the favorable outcomes observed in this study should be interpreted within the context of treatment adherence. Conservative treatment was significantly less effective than CPAP in terms of both objective and subjective measurements of OSA-related symptoms and driving performances. Regarding subjective evaluations, group 2 demonstrated non-significant decreases in PSQI and ESS scores but increased KSS scores. The abnormal increase in driving fatigue highlights the inherent challenges of conservative treatment in achieving genuine lifestyle changes within 3 months among fatigued patients. This likely reflected a discrepancy between patient expectations and the benefits received.35 Ballester et al,36 demonstrated that although ESS scores significantly decreased in both groups of patients with moderate to severe sleep apnea/hypopnea syndrome after 3 months of treatment, the reduction was greater in the group receiving CPAP combined with conservative treatment compared to the group receiving conservative treatment alone (ie, sleep hygiene education and weight management). A more recent study found that ESS scores significantly decreased in both groups of patients with mild to moderate OSA undergoing CPAP treatment combined with conservative treatment and those receiving conservative treatment alone (ie, weight management and sleep hygiene education) after 10 weeks of treatment, with CPAP showing superior effects.37 Given the similar treatment durations, the decreased effectiveness of conservative treatment compared to studies by Ballester et al and Lam et al may be partially attributed to the lower OSA severity of patients in group 2, and the exclusion of weight loss as conservative treatment in our study. Group 2 exhibited nonsignificant improvements in terms of objective measurements of sleep disorder indices and driving performance. Although Redline indicated that conservative treatment (eg, sleep hygiene education and pharmacological support) significantly improved oxygen saturation in patients with mild OSA,38 other studies suggested that longer treatment durations (eg, 1 year) are required to observe significant improvements in OSA-related symptoms.39 The reduced effectiveness observed in our study compared to Redline may be attributed to the shorter, 3-month treatment duration. Overall, conservative treatment demonstrated less effectiveness in improving both subjective and objective OSA-related symptoms and could be considered a supplementary intervention alongside CPAP therapy. This study has several strengths. First, we employed a systematic approach to improve treatment acceptance among participants. More specifically, a rigorous follow-up mechanism was implemented, involving twice-weekly phone calls and monthly outpatient medical appointments. Furthermore, alterations in sleep parameters, occurrences of ADBs, and subjective evaluations of sleepiness and OSA symptoms were further compared between pre- and post-treatment for different treatment strategies. To our knowledge, no study has holistically compared the efficacy of CPAP treatment with conservative treatment in addressing OSA-related clinical symptoms together with driving behaviors. Such statistical outcomes may potentially serve as evidence to shed light on the efficacy of different interventions for OSA in addressing sleep problems, thereby enhancing transportation safety. The study presents noteworthy findings while acknowledging several limitations. First, while taxi drivers were assigned to the two treatment groups, baseline demographic and driving-related characteristics showed no statistically significant differences between the groups. However, the distribution of OSA severity was imbalanced between treatment groups, with the CPAP group comprising participants with more severe symptoms at baseline. This imbalance may have influenced treatment responsiveness and introduced residual confounding when comparing outcomes between groups. To mitigate this concern, a subgroup analysis restricted to participants with severe OSA was conducted and yielded findings consistent with those of the overall analysis, suggesting that the observed treatment effects were not solely attributable to differences in baseline disease severity. Nevertheless, future research should adopt stratified randomization to ensure an even distribution of OSA severity levels across treatment groups. Second, the relatively modest sample size may have limited the statistical power to detect treatment-related changes, particularly within the conservative treatment group. We acknowledge that the sample size was modest; however, this study was designed as an exploratory investigation in a difficult-to-recruit occupational cohort, requiring extensive clinical assessments, home-based sleep monitoring, and naturalistic driving data collection. Consequently, the present findings should be considered preliminary and interpreted with caution until validated in larger cohorts. Third, although this study accounted for weather conditions (ie, daily rainfall, relative humidity, and daily temperature) and traffic congestion, it omitted consideration of other influential environmental factors. In addition, although the navigation app currently in use was validated to guarantee acceptable trajectory data for ADB determination, recording driving with a dashcam might offer more-reliable identification of ADBs, aiding a more-comprehensive analysis. Last, while this study conducted regular phone call follow-ups to maintain adherence to sleep hygiene recommendations during the treatment period, it lacked objective evidence to confirm the level of adherence. Altogether, these limitations underscore the need for future research that incorporates stratified randomization to address baseline imbalances, integrates additional environmental variables, and includes objective measures to verify adherence to treatment. Such efforts would enhance the robustness of findings and deepen the understanding of ADBs in the context of taxi driving. The present findings have important implications for occupational health management among commercial drivers. First, the observed improvements in sleep-related symptoms and driving performance following CPAP treatment suggest that OSA should be recognized as a modifiable occupational health risk factor rather than solely a clinical sleep disorder. Given the elevated prevalence of OSA among professional drivers and its association with fatigue-related driving risks, routine OSA screening could be incorporated into occupational health surveillance programs to facilitate earlier identification and intervention. Second, advances in telemedicine, remote monitoring, and wearable technologies offer promising opportunities to improve the accessibility and effectiveness of occupational health services for professional drivers. Commercial drivers frequently encounter barriers to healthcare access because of irregular schedules, long working hours, and geographic mobility. Future occupational health programs may benefit from integrating wearable devices capable of continuously monitoring physiological indicators. Integrating wearable technologies, home-based sleep monitoring systems, and naturalistic driving behavior data could facilitate the development of proactive occupational health management systems capable of detecting emerging safety risks before adverse driving events occur. Improved driving safety may additionally yield economic benefits through reductions in crash-related costs, healthcare expenditures, productivity losses, and vehicle damage. Collectively, these findings support a shift toward continuous, data-driven occupational health management to enhance both driver well-being and transportation safety. Conclusions Although the efficacy of CPAP has been questioned by some researchers, a growing body of evidence supports its effectiveness. This study confirmed and extended previous findings on the effectiveness of CPAP in improving objective and subjective evaluations of OSA-related symptoms and driving performances, suggesting that CPAP can potentially be utilized as a primary treatment option for occupational drivers with severe OSA. Conversely, conservative treatment demonstrated limited effectiveness in alleviating OSA-related symptoms or reducing ADBs, suggesting that it may be more suited as a supplementary strategy rather than a standalone solution. These findings underscore the potential of CPAP treatment in enhancing transportation safety, suggesting a need for integrating such interventions in taxi driver health programs. Although the present findings provide preliminary evidence supporting the benefits of CPAP treatment, the moderate sample size may limit their generalizability. Therefore, future studies should incorporate larger and more heterogeneous cohorts, randomized controlled designs, and extended longitudinal observations to validate the long-term effects of CPAP adherence on driving performance and health outcomes. Acknowledgments We are grateful to the technologists at the Sleep Center of Shuang Ho Hospital for gathering and managing the raw data. We appreciate the technological support from TRADE-VAN.COM and Rooti Lab. We also thank all of the participants in this study for their contributions. The Epworth Sleepiness Scale was used with the appropriate licence obtained for this study. Funding Statement This study was funded by the National Science and Technology Council of Taiwan (NSTC 112-2634-F002-003, NSTC 113-2222-E-038-003, NSTC 114-2634-F-002-005, NSTC 114-2622-E-038-002), Taipei Medical University (TMU113-AE1-B03), and the Natural Environment Research Council [NE/S007415/1]. Abbreviations ADB, Aberrant driving behavior; AHI, apnea-hypopnea index; ArI, arousal index; CEI, chest effort index; CPAP, continuous positive airway pressure; CVHRI, cyclic variation of the heart rate index; DBQ, driving behavior questionnaire; ECG, electrocardiogram; ESS, Epworth Sleepiness Scale; insomnia severity index (ISI), Karolinska Sleepiness Scale (KSS), OSA, obstructive sleep apnea; ODI, oxygen desaturation index; PHQ, patient health questionnaire; PSQI, Pittsburgh Sleep Quality Index; PSG, polysomnography. Data Sharing Statement The data for this study were collected at the Sleep Center of Taipei Medical University-Shuang Ho Hospital between February 2021 and July 2023. Due to the inclusion of personal clinical information, the data are not provided as a supplementary file. Interested parties may contact the corresponding authors to request access to the dataset and relevant documents. Ethics Approval and Consent to Participate This study was approved by the Ethics Committee of the Taipei Medical University-Joint Institutional Review Board (TMU-JIRB: N202112077). All participants received written information about the study procedures and provided their informed consent through a signed agreement to enrolment. Author Contributions Z.C.: Methodology; Formal analysis; Visualization; Writing - original draft. W.-T.L.: Formal analysis; Writing - original draft; Supervision; Funding acquisition. A.M.: Formal analysis; Supervision; Writing - review & editing. W.-H.H.: Data curation; Investigation; Writing - review & editing. Y.-C.K.: Formal analysis; Writing - review & editing. J.-H.K.: Conceptualization; Resources; Project administration; Supervision; Writing - review & editing. Y.-L.C.: Data curation; Investigation; Writing - review & editing. R.C.: Data curation; Investigation; Validation; Writing - review & editing. P.W.: Writing - original draft; Validation. K.-Y.L.: Software; Validation; Writing - review & editing. P.-H.F.: Conceptualization; Methodology; Writing - original draft. K.-Y.C.: Data curation; Investigation; Writing - review & editing. H.-C.L.: Data curation; Investigation; Writing - review & editing. Y.-C.L.: Investigation; Writing - review & editing. W.-H.C.: Data curation; Investigation; Writing - review & editing. C.-Y.T.: Conceptualization; Methodology; Visualization; Writing - review & editing; Supervision. All authors gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. Disclosure The authors declare that there are no conflicts of interest regarding the publication of this article. References Associated Data Data Availability Statement The data for this study were collected at the Sleep Center of Taipei Medical University-Shuang Ho Hospital between February 2021 and July 2023. Due to the inclusion of personal clinical information, the data are not provided as a supplementary file. Interested parties may contact the corresponding authors to request access to the dataset and relevant documents.