Background of the field study
A total of 22 professional (21 males and 1 female) taxi drivers participated in a field exposure assessment project about air pollutant exposures inside taxi cabs [11]. The research team ran a recruitment/survey campaign at the Los Angeles Airport (LAX) taxi holding lot in February 2013 in order to recruit study participants and collect basic information about taxi drivers and their cabs. A questionnaire designed by the research team included the questions about sociodemographic, car model, work stress experience, health-related behaviors, health conditions, and the use of medication [11]. We initially handed out 2449 questionnaires and finally collected 316 complete questionnaires. Out of the 316 taxi drivers who completed the questionnaire, 121 non-smokers were eligible to participate in a field exposure assessment project about air pollutants and air exchange rates inside taxi cabs. To ensure the sampled taxi drivers/cabs are representative, stratified random sampling was conducted based on car models and drivers’ age. A total number of 22 non-smoking taxi drivers out of 121 eligible drivers were selected to participate in the field exposure assessment project [11]. The study design and protocol (#12-000845) were approved by the Institutional Review Board (IRB) of University of California Los Angeles. All of the taxi drivers who participated in the field exposure assessment project provided written informed consent.
13 male taxi drivers for this study
Among the above 22 taxi drivers, we excluded 9 taxi drivers from this study whose ambulatory heart rate on the first experimental day (for details, see below) was not be able to be collected (N = 5; cases T03, T07, T08, T19 and T20, including 1 female driver) or was collected, but only for a limited time period (<1 h) (N = 4; cases T05, T12, T18, and T21). Thus, we included 13 male taxi drivers in the current study. There were some differences between the excluded and included drivers: the excluded 9 drivers were older (53.0 years vs. 42.9 years), worked more years as a taxi driver (11.6 years vs. 7.4 years), and had higher prevalence of hypertension (33 % vs. 15 %) than the included 13 drivers.
6-hour experimental on-road driving
The 13 male drivers drove 6 h on each of four consecutive experimental days in the Greater Los Angeles area between April 2013 and November 2013 as he would typically do. One field technician rode along in the taxi cab operating and maintaining all of the air sampling instruments (see below). The starting time of each day was based on the driver’s availability. No actual fares were collected during the tests and the drivers’ time and effort were compensated by the research fund [11]. The driving routes were not specifically planned for each driver. Instead, each driver was asked to drive from the start location, University of California Los Angeles, to the area where he usually works. Each driver was allowed to take breaks as he would during a typical work day. The time and location of each break were recorded by hand by one field technician and confirmed by a GPS unit (Qstarz GPS BT-1000XT, Taipei, Taiwan). The experimental conditions with regard to taxi-cab air flow varied day to day during the four experimental days [11]. The experimental conditions on the first day were least intervening (most realistic) because the drivers had a full control over all the vehicle operations such as opening/closing windows, turning air conditioning on or off, setting ventilation to recirculation or outdoor air mode as they usually did on their typical work days. Thus we restricted our analyses for this study to the first experimental day.
Assessment of ambulatory heart rate: HRdriving, HRrest, and PMHRdriving
One trained field technician instructed the drivers to wear and use an ambulatory heart rate monitor (RS800CX, Polar Electro, Finland, with a sampling rate of 1,000 Hz) [29, 30] just before the experimental 6-hr on-road driving. Heart rate (HR) of the drivers was continuously measured based on the peak to peak interval of two consecutive QRS complex signals on electrocardiograms (called hereafter RR interval, milliseconds) with the monitor during the driving. The raw RR interval data of 13 drivers downloaded from the monitor were processed using the data analysis software (Kubios Version 2.2) [31] with the medium option for artifact correction. Then, we split the artifact-corrected RR interval data into 5-min segments (674 segments in total) from the starting time of the 6-hr driving and calculated the average HR of each 5-min segment (called hereafter 5-min HR). Afterwards, we manually examined each 5-min segment whether all RR intervals in the 5-min segment are within the normal range (300 to 2,000 milliseconds) and whether most (>95 %) of the ratios of two consecutive RR intervals in the 5-min segment are within the normal range (0.8 to 1.2) [30, 32]. Of the total 5-min 674 segments, 160 (23.70 %) segments were excluded by the last procedure from analyses in the current study.
After the above data cleaning process, each 5-min segment was classified into the following two groups: 5-min HR while driving (5-min HRdriving) and 5-min HR while taking breaks (5-min HRbreaking) based on the record by one field technician on the on-road driving. In the current study, we used only 5-min HRdriving values (of 344 5-min segments in total) for analyses as consistent with the previous environmental science studies in taxi drivers [10, 11]. The driving times noted by one field technician were confirmed by matching the driving times with the vehicle speed information (i.e., ≥ 1 km∙hr−1) during the same time periods. The breaking times recorded by the field technician included the times when the drivers took a rest inside taxi cabs, or went outside taxi cabs for a stretch, walk, or meal.
Resting HR (HRrest) of each taxi driver was estimated conservatively based on the following two-step process: (1) identifying the lowest 5-min HRdriving value of each driver [28]; and then (2) choosing the age- and gender-specific 25th, 50th, or 75th HRrest percentile value in the United States (US) adult population reference data from the 1999–2008 National Health and Nutrition Examination Surveys [33] that was lower than, but closest to the identified lowest 5-min HRdriving value.
As a way to control for the individual differences in age and resting and maximum HR, we estimated the 5-min percent maximum heart rate range (PMHRdriving) of each taxi driver during the driving times using the following equation [9]:
$$ 5\hbox{-} \min\ {\mathrm{PMHR}}_{\mathrm{driving}}\left(\%\right) = \frac{5 \min {\mathrm{HR}}_{\mathrm{driving}}-{\mathrm{HR}}_{\mathrm{rest}}}{{\mathrm{HR}}_{\max }-{\mathrm{HR}}_{\mathrm{rest}}} \times 100 $$
The maximum heart rate (HRmax) of each driver was estimated by using the formula, 205.8 − (0.685 × age) [34, 35].
Assessment of five physical (environmental) hazards while driving inside taxi cabs
Five physical hazards (PM2.5, CO2, relative humidity, temperature, and noise) were continuously assessed inside taxi cabs during the experimental 6-hr on-road driving. One DustTrak (Model 8520, TSI Inc., St. Paul, MN) was used to measure the in-cabin PM2.5 concentrations. One Q-trak monitor (Model 8554, TSI Inc., St. Paul, MN) was also used to measure the in-cabin CO2 concentrations, relative humidity, and temperature simultaneously. The noise level in cabin was measured by a Quest 2900 Sound Level Meter (3 M, St. Paul, MN). All of these instruments were synchronized with the heart rate monitor and were set to record one reading every second, to provide data with high time resolution. All instruments were calibrated to ensure the quality of measurement. After the experimental driving, data were downloaded. Data were then observed in Excel for visualization and obvious outliers - mostly caused by instruments malfunctioning - were removed before further data analysis. Then as with the HR data, the physical hazards data were also split and averaged into 5-min segments for analyses and only the 5-min averaged data while driving (excluding the 5-min averaged data while breaking) was used in the current study.
Assessment of work hours, work stress, and other covariates
Work hours per week were calculated using the two questions in the recruitment survey questionnaire (“Typically, when do you start and end your work day?” and “How many days do you typically work as a taxi driver each week?”). Work stress experience of taxi drivers on a typical workday was also measured with one question (“How often do you find your work stressful?). Age, race/ethnicity, exercise during leisure-time (the frequency of the moderate or vigorous level of aerobic exercise: frequent (2 or more times per week) and infrequent (0–1 times per week))), health condition (“In general, would you say your health is: Excellent; very good; good; fair; and poor), body weight and height, hypertension (“Have you ever been diagnosed with hypertension?”), and the use of anti-hypertensive medication were also assessed with the recruitment survey questionnaire.
Data analysis
At first, the distributions of five physical hazards inside taxi cabins during the 5-min HRdriving periods for each driver were examined. Then we examine the distribution (maximum, mean, and minimum) of the 5-min HRdriving values for each driver with a focus on the extent of the elevation of HR and the threshold HR value for a sustainable 8-hour work (i.e., 35 beats per min above HRrest) [9]. The above analyses were replicated with the 5-min PMHRdriving values among the 13 taxi drivers with a focus on the threshold PMHR value for a sustainable 8-hour work (i.e., 30 % of percent maximum heart rate range) [27, 28]. As a sensitivity test, we tested whether the above analyses would be affected by the status of hypertension, obesity, and exercise frequency of the taxi drivers. All data analyses and graphs were performed using PASW version 23.0 (SPSS Inc., Chicago, IL, USA) and Sigma Plot software, version 12.5 (Systat Software Inc. USA), respectively.