Combining four screening tools for cost effective screening of OSA in train drivers: A UK experience

Abstract

Background: 

Obstructive sleep apnoea (OSA) in drivers/workers has been implicated in railway and road traffic safety incidents; however, there are insufficient data on its prevalence and cost-effective screening methods.

Aim: 

This pragmatic study examines four OSA screening tools: the Epworth sleepiness scale (ESS), the STOP-Bang (SB), the adjusted neck circumference (ANC) and the body mass index (BMI), exploring their suitability and effectiveness separately and in combination.

Method: 

Using all four tools, 292 train drivers were opportunistically screened between 2016 and 2017. A polygraph (PG) test was carried out when OSA was suspected. Patients with an apnoea–hypopnea index (AHI) ≥5 were referred to a clinical specialist and reviewed annually. Those who had continuous positive airway pressure (CPAP) treatment were evaluated for compliance and control.

Results: 

Of the 40 patients who had PG testing, 3 and 23 participants met the ESS >10 and SB >4, criteria, respectively, whereas 25 participants each had an ANC >48 and a BMI >35 with a risk factor or ≥40 with none. OSA was detected in 3, 18 and 16 of them who met the ESS, SB and ANC criteria, respectively, and was positive for OSA in addition to 16 others who met the BMI criteria. A total of 28 (72%) were diagnosed with OSA.

Conclusion: 

Although when used individually, these screening methods are less effective/inadequate, combining them is easy, feasible and offers the maximum chance of OSA detection in train drivers.

INTRODUCTION

Obstructive sleep apnoea (OSA) is a common condition with a prevalence of 14%–49%.^[1] Obesity is a major risk factor for OSA and the rising rate of obesity indicates that the prevalence of OSA is likely to rise.

OSA is associated with sleepiness, inattention, poor concentration, cognitive dysfunction, slow reaction time, and other secondary morbidities.^[2] Cognitive impairments and slow reaction times have been reported to contribute to a higher risk of road traffic accidents, occupational incidents and commercially sensitive task errors and failures.^[3] Up to 20% of large vehicle accidents have been attributed to sleepiness.^[4] OSA has been directly implicated in road traffic and safety incidents as well as railway derailments.^[5,6] Previous study estimated a high prevalence of OSA in commercial truck drivers.^[7]

Individual studies and systematic reviews have observed that OSA is associated with two to five times more risk of having a work-related accident.^[8] In vocational driving populations, undiagnosed OSA is a potential threat to public safety. Burks et al.’s^[9] retrospective study reported that poor adherence to continuous positive airway pressure (CPAP) treatments (<4 h) was associated with nearly five times more risk of crashes and good adherence reduced the risk to levels seen in individuals without OSA. A Swedish case–control study also reported similar findings.^[10] Although those studies are not randomized control studies, they are strongly persuasive about the benefits of screening and treatment of OSA.

In practice, most organizations cannot afford to test safety-critical workers for OSA using polysomnography, often referred to as the gold standard test. Therefore, a variety of less expensive screening tools have been used to identify at-risk populations. The UK rail industry has no specific tool mandate for such screening; however, at the time of our study, the widely used methods for detecting OSA were the Epworth Sleepiness Scale (ESS) and clinical history.

The ESS, which assesses daytime sleepiness, is a subjective tool that is used as a proxy to screen for the likelihood of OSA, and it is often susceptible to intentional and unintentional underreporting.^[11,12] The British Thoracic Society’s statement on driving and OSA perhaps provides a more focused interpretation of ESS by recommending its use in combination with a history of incidents to decide on risk.^[13] The Schneider study showed that sleepiness seems underreported in driving populations.^[9] Therefore, using only the ESS combined with a history of mishaps is not enough for organizations of sensitive public safety, such as the railway industry. The UK Railways Safety and Standard Board, the United States, the Canadian and Australian railway bodies have all recommended more robust strategies involving anthropometric measures and medical history for screening for OSA in train drivers.^[14,15] The relevant anthropometric measures are the adjusted neck circumference (ANC) and body mass index (BMI).^[16,17] The STOP-Bang (SB) is a screening tool used by anesthetists to assess the risk of OSA and it has also been used in driving populations.^[18,19] A number of reviews have noted the need for a better screening tool despite those currently in use.^[20]

No one tool seems to be perfect for OSA screening. We hypothesized that combining these tools might provide a cost-effective, easy method to improve the detection rate when implemented. In considering the best screening strategy for a train operating company, we reviewed the existing methods, focusing on those that did not require computer algorithms, copyright, or patents and that could be easily administered in an office setting. Although we examined other popular screening methods, we preferred those that have already been used in the railway environment and available options that use easily collected anthropometric measurements. We discarded the MAPI, Somni-Sage, Berlin Questionnaire and Sleep Disorder Questionnaire for the above reasons. We selected four tools: the ESS, SB, ANC (as used by the Canadian Rail Association) and BMI (as used by the Australian Rail Commission).^[19,17,21] ESS was included because of its widespread use. We sought to explore whether one method was more suitable than the other and whether combining them would be more effective than using any tool individually.

METHODS

Train drivers need to attain an acceptable medical fitness standard to be considered fit for duty. In the UK, this is defined by the Train Driver Licensing and Certification Regulations (TDLCR) and which specify that drivers must not have a condition that is likely to cause a loss of concentration. This statutory instrument also requires drivers to be screened triennially until the age of 55 years and then annually thereafter. All train operators or drivers who presented for a medical checkup between August 2016 and August 2017 were screened for OSA using the ESS, the SB, the ANC and the BMI criteria.

The ESS is an eight-item questionnaire that asks the patient to rate subjective sleepiness in a variety of daytime situations. The items have a scale of 0 to 3 and the maximum score is 24. A patient with an ESS greater than 10 was referred for further OSA investigation.^[22]

The SB is an eight-item form that records symptoms, medical history and anthropometric measures with a score of zero or one for each item.^[12,23] All drivers with a score of five or higher were deemed to be at high risk for OSA and they were referred for further investigation.

ANC is a clinical prediction rule that combines four known clinical features that predict OSA with the patient’s measured neck circumference as follows: ANC score = NC (neck circumference, in cm) +4 (if a history of hypertension [HT]); +3 (if a history of frequently reported snoring); +3 (if a history of frequently reported choking, gasping and/or witnessed apnoeic events). “Frequent” means that the behavior or event occurs five or more nights in a week. A score greater than 48 indicates a high risk of OSA and requires further investigation.^[17] ANC is of diagnostic value in screening for OSA.^[21] It is the preferred method of screening by the Canadian Railway Association.^[17]

A joint task force of the American College of Occupational and Environmental Medicine (ACOEM), the American College of Physicians (ACP), and the National Sleep Foundation recommended the BMI criteria in an effort to address the high prevalence of untreated OSA in the US commercial driving population.^[24] These criteria have also been adopted by the Australian Railway Commission as follows:

1). A driver with a BMI of 40 or more has a high risk for OSA and should be referred for investigation.

2). A driver with a BMI greater than 35 and one or more of the following conditions has a high risk for OSA and should be referred for investigation:

a) Type 2 diabetes; b) High blood pressure requiring two or more medications for control; c) A history of habitual loud snoring during sleep or of witnessed apnoeic events.^[14]

All drivers attending medical checkups during the study period were screened using all four tools. Those who fulfilled any of the above criteria were tested for OSA either by private assessment or by a general practice physician on the National Health Service (NHS). The NHS criteria often require the patient to be symptomatic based on reported EDS.^[25] Testing for OSA in all cases was via home sleep apnoea testing using the Limited Channel Polysomnography (PG) (level 3 test/type 3 devices). The measured apnoea–hypopnea index (AHI) (events per hour) was used to diagnose OSA, which is categorised by the American Association of Sleep Medicine (AASM) guidelines^[26] into:

None/Minimal: AHI <5 per hour;

Mild: AHI >5, but <15 per hour;

Moderate: AHI ≥15, but <30 per hour;

Severe: AHI ≥30 per hour

All drivers diagnosed with sleep apnoea (AHI ≥5) were further assessed at 6 and 12 weeks following the medical checkups and then annually irrespective of age. Control and compliance evaluations were done for those that were treated with CPAP. The occupational health management protocol is summarised in Figure 1.

Figure 1: Obstructive Sleep Apnoea Screening Flow Chart

RESULTS

A total of 292 train drivers were screened during the study. There were 285 males and 7 females. The age range was 24–70 years, with a mean age of 48.6 years. The BMI range was 19–49 kg/m², with a mean BMI of 29.6 kg/m². A total of 59 drivers had hypertension and 34 of them were on two or more antihypertensive medications. Although 57 drivers had a history of snoring, 9 drivers had witnessed apnoeic events, and 19 drivers had type 2 diabetes.

Seven of the 292 drivers who were screened had a pre-existing diagnosis of OSA. Of the remaining 285, 40 met at least one criterion for home PG. Although 3 drivers met the ESS criteria, 23 had an SB ≥5, 25 drivers had an ANC >48 and 25 had a BMI >35 with a risk factor or ≥40 without a risk factor. Some drivers met multiple criteria.

The 40 drivers who had screened positive were more obese (mean BMI of 36 vs. 29) and had a higher ESS score (mean ESS of 4.5 vs. 3.4). They also had a higher SB (a mean of 4.7 vs. 2.4) and ANC (a mean of 48 vs. 42) [Table 1].

Table 1: Screen positive numbers per tool

Of the 40 who had PG, 28 were diagnosed with OSA in which 14, 9 and 5 drivers had mild, moderate, and severe OSA, respectively. The PG was negative in 11 drivers, and one was lost to follow-up. Thus, 28 (72%) drivers had a positive sleep study.

Of 40 train drivers who underwent PG, 3 out of 3 with ESS >10 were diagnosed with OSA and 18 out of the 23 with a high SB had OSA. Of the 25 with ANC >48, 16 had OSA and of the 25 who met the BMI criteria, 16 had OSA.

In our population of 292 drivers who attended a medical checkup within the study period, 35 [28 new +7 previously diagnosed] drivers were confirmed to have OSA. This gives a prevalence of approximately 12%. We could not subject all drivers to sleep study, positive predictive value (PPV)/negative predictive value (NPV) could not be calculated.

DISCUSSION

In this pragmatic study, using the individual screening tools would have identified a maximum of 25 eligible drivers for OSA testing, but combining all four tools increases this number to 40 (60%). Similarly, using a single tool would have identified a maximum of 18 drivers (6.16%) with OSA (identified with SB). The outcome of combining the four screening tools is that nearly twice as many train drivers, 35 (12%), were diagnosed with OSA. This demonstrates the value of our approach. The mean ESS was 3.4 in those who did not fulfil the screening criteria compared to 4.5 in those who were positive by at least one other screening tool. This highlights that relying solely on the ESS score limit of >10 for sleep study referral would have underestimated the frequency of OSA in our population.

Our aim was to explore a range of screening tools used worldwide. The Australian and Canadian rail boards use ANC and BMI, respectively. Both approaches focus on identifying obese individuals at risk of OSA. The Australian railway screening criteria additionally highlight patients with diabetes as having high risks for OSA. Other factors, such as craniofacial morphology and upper airway physiology, also contribute to the likelihood of OSA.^[27] Although the SB questionnaire is used mostly for preoperative evaluations, it has become useful for the screening of commercial drivers,^[19] and with additional information on diabetes, we used all three screening tools. A score limit of 3 on the SB has the risk of over-sensitivity because of its low specificity in identifying moderate to severe sleep apnoea. We used a limit of five because of its higher specificity for moderate to severe OSA.^[23] However, combining this cut-off with other screening tools mitigates the increased risk of false negatives. In our study, 11 patients (7 positive and 4 negative for OSA) with an SB score of 4 were missed by using a limit of 5, but were picked up as high risk by other methods and all three patients who had diabetes alongside raised BMI were diagnosed with OSA. Though combining questionnaires introduces a risk of false positives, in a safety-critical population, a false positive is a more acceptable outcome. The procedure we followed, however, minimized the occurrence of false positives.

In this study, a group of drivers was identified as high risk by all the tools; others were positive with one screening tool and negative with another. Using the composite screening method, however, increased the yield with minimal extra cost. A combination of tools also broadens sensitivity, which is more reassuring in our safety-critical population. No single tool was superior to the other in the detection of OSA.

Routine inquiry about sleepiness and fatigue on duty, near-misses and safety line incidents form a part of the operational assessment of drivers. Similar policies should be part of the standard operating procedures for other train operating companies and safety-critical organizations.

The driver and vehicle licensing authority (DVLA) in the UK publishes a guide for medical professionals for assessing fitness to drive motor vehicles and it is often used as a reference by occupational physicians assessing individuals’ fitness to drive trains. The current DVLA recommendation is based on declared excessive daytime sleepiness rather than the risk of OSA.^[28] Using this guide alone would be equivalent to using only the ESS for screening for OSA in our population, leading to a very high proportion of false negatives.

The mean AHI for those involved in crashes in the Schneider study was 31.3.^[9] This study would recommend that drivers with severe OSA (an AHI ≥30) should be treated with CPAP before resuming driving duties. This is consistent with the practice guidelines for Canadian and Australian railways. We would also suggest that those with moderate OSA and sleepiness, fatigue, and/or a history of safety incidents should also be treated.

Limitations

We are likely to have underestimated the true prevalence of OSA, as not all drivers were tested. Therefore, we are unable to calculate the positive and negative predictive values of the composite tool. We do not have an accurate idea of the true prevalence in our study.

The screening methods used did not focus on those with low BMI or neck circumference but on those with other clinical features of a narrowed upper airway. We used home PG rather than in-patient polysomnography. This could underscore the severity and miss some diagnoses of OSA, as there is no guarantee that someone with normal polysomnography was actually asleep.

The use of overnight oximetry was considered as it is a relatively inexpensive tool. However, after the use of the composite tool, we needed a test that would be robust at ruling out moderate to severe sleep apnoea. A negative finding on oximetry would not obviate the need for home polysomnography.

This was a pragmatic study rather than a research exercise where the entire population could be tested for sleep apnoea. Though it potentially underestimates the true prevalence of OSA, it adds to the body of knowledge about how screening tools fare against each other.

CONCLUSIONS

ESS is inadequate as the only screening tool for OSA. We recommend using a combination of screening tools that allowed us to maximise the detection of OSA in train drivers. The composite tool identified 28 drivers out of 40 (72%) as high risk and they tested positive for OSA, whereas 14 had moderate to severe OSA. ESS alone would have detected 3 patients and any other single tool would have detected a maximum of 18, thereby validating our approach.

Ethics approval

Ethical approval was not required as this is an observational study detailing clinical practice as part of their routine medical. All participants were given information on OSA and informed as to the reason for asking them to complete questionnaires and sleep tests. They consented to their medical, which was looking at all aspects of health including OSA.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.