Noise Levels Associated With New York City's Mass Transit Systems

Abstract

Objectives. We measured noise levels associated with various forms of mass transit and compared them to exposure guidelines designed to protect against noise-induced hearing loss.

Methods. We used noise dosimetry to measure time-integrated noise levels in a representative sample of New York City mass transit systems (subways, buses, ferries, tramway, and commuter railways) aboard transit vehicles and at vehicle boarding platforms or terminals during June and July 2007.

Results. Of the transit types evaluated, subway cars and platforms had the highest associated equivalent continuous average (L_eq) and maximum noise levels. All transit types had L_eq levels appreciably above 70 A-weighted decibels, the threshold at which noise-induced hearing loss is considered possible.

Conclusions. Mass transit noise exposure has the potential to exceed limits recommended by the World Health Organization and the US Environmental Protection Agency and thus cause noise-induced hearing loss among riders of all forms of mass transit given sufficient exposure durations. Environmental noise–control efforts in mass transit and, in cases in which controls are infeasible, the use of personal hearing protection would benefit the ridership's hearing health.

For the first time in history, more than half of the world's population lives in cities, and it is projected that more than two thirds of the population will live in cities by 2030.¹ An important factor supporting the growth and viability of urban centers is mass transportation, which is rapidly expanding to keep pace with increasing demand. For example, in 2004 there were 95 subway systems worldwide; today there are 167, a 76% increase in only 5 years.² Although there are well-documented environmental and public health benefits associated with mass transit, interest in the health and safety effects of mass transit on urban communities is increasing.^3–5 A particular concern is the potential for mass transit to result in excessive exposure to noise.

Noise exposure is a function of 2 main factors: (1) the frequency-weighted exposure level, measured in A-weighted decibels (dBA), and (2) the exposure duration. The causal association between chronic exposure to excessive noise and permanent, irreversible, noise-induced hearing loss (NIHL) is well known, as are the adverse social, psychological, and occupational effects associated with the condition. Nonauditory adverse health effects have also been reported,^6–8 and recent research suggests that excessive noise exposure may be linked to hypertension and ischemic heart disease, disruptions in stress hormones, and sleep disorders.^9–12

There are no comprehensive national or international surveillance programs for hearing loss. Worldwide, more than 250 million people are estimated to suffer from hearing loss, of which at least 30 million cases represent NIHL.¹³ In the United States alone, between 3 to 10 million people are estimated to have NIHL.¹⁴ Hearing loss from all causes ranks among the top 10 most common serious health problems worldwide, and NIHL is the leading occupational disease in industrialized nations.^14,15 The limited data available suggest not only that NIHL prevalence and incidence rates are extraordinarily high but also that the associated costs are enormous.^16,17 Importantly, even though US occupational exposure regulations have been in place for decades, rates of NIHL-related workers' compensation cases remain high. Therefore, nonoccupational sources of exposure are coming under scrutiny, including mass transit.

The size of the population exposed to mass transit noise is of considerable magnitude. The US mass transit network, with an infrastructure encompassing subways, buses, commuter and light rail, ferry boats, trolleys, and tramways, is the largest in the world, with 9.7 billion passenger rides in 2006.¹⁸ There are 14 subway systems in the United States, with a combined daily ridership in excess of 10 million people.^19–21 Five of the US systems are more than 75 years old, and the largest, the New York City subway system, with over 4 million riders per weekday,²² is more than 100 years old. These older systems were designed before noise-control technologies were available. Worldwide, there are 2 subway systems with even greater ridership rates: Tokyo's is the largest at 2.6 billion passenger rides per year, and Moscow's is the second largest with 2.5 billion.^23,24

In a recent sound-level pilot survey on subways,³ we noted levels that potentially exceeded the community exposure limits initially recommended by the US Environmental Protection Agency (EPA) in 1974 and confirmed by the World Health Organization (WHO) in 1998. WHO and EPA recommended daily allowable exposure times are 24 hours at 70 dBA, 8 hours at 75 dBA, 2.7 hours at 80 dBA, 0.9 hours at 85 dBA, and 0.3 hours at 90 dBA. Chronic exposures that exceed these allowable combinations of duration and noise level are expected to produce NIHL in some members of the exposed population.^25,26

The amount of NIHL anticipated to result from specific noise-exposure levels can be predicted with a model published by the International Organization for Standardization.²⁷ This model allows users to estimate the amount of NIHL expected to result from chronic 8-hour equivalent continuous average (L_eq) noise exposures between 75 and 100 dBA or 24-hour L_eq exposures between 70 and 95 dBA. The model permits the estimation of median values of expected NIHL as well as values for the 0.05 to 0.95 fractiles among an exposed population for given exposure levels and durations. Based on the WHO and EPA recommendations, chronic exposure to 80.3 dBA for more than 160 minutes per day may be expected to produce hearing loss in some exposed individuals, and a 90.2-dBA level likewise may cause hearing loss with just 18 minutes of exposure per day.

Few data involving dosimetry measurements of noise exposures associated with mass transit have been reported previously. In a study of the daily noise exposures experienced by 32 people in Madrid, Spain, Diaz et al.²⁸ measured noise levels associated with a variety of self-reported transportation exposures with noise dosimeters. Zheng et al.²⁹ conducted 24-hour noise dosimetry on 221 residents of Beijing, China, and assessed the noise levels associated with self-reported activities, including commuting. Nearly all other studies that have evaluated noise levels associated with subway equipment are decades old and based on sound level measurements rather than dosimetry. In 1931, Stanton conducted an unpublished noise-level survey of the New York City subways,³⁰ and in 1971, Harris and Aitken³¹ reported levels measured on specific New York City train line platforms and cars. A small sound level survey on a subway system in India was also recently reported.³²

Our current study expanded on our pilot study of subway noise and assessed average noise levels on a variety of types of mass transit to further evaluate noise exposure among transit riders.

METHODS

Noise levels were measured in the New York City area during June and July 2007 on various types of mass transit, including subways, buses, ferries, commuter railways, and the tramway. We measured equivalent continuous average (L_eq) and maximum (L_max) noise levels with type II noise dosimeters (Q–300; Quest Technologies, Oconomowoc, WI). L_eq levels represent the average exposure level over a measured period of time, and L_max levels represent the highest level reached during a measurement. Although point-in-time area measurements made with sound level meters—such as those collected in our pilot study³—can provide useful screening information for noise exposure potential, time-integrated L_eq measurements made via personal dosimetry are preferable for assessment of long-term average noise exposure levels, especially where noise levels vary widely over time and space, as is the case for transit noise exposures. L_max levels provide useful information about the maximum possible noise level in a given exposure scenario.

The dosimeters were configured according to the exposure standard recommended by the US National Institute for Occupational Safety and Health.³³ Research staff carried the dosimeters in backpacks during measurements, and microphones were located within 4 inches of the researcher's ear²⁷ to provide the most representative estimate of personal exposure.

Data Collection

We made measurements aboard transit vehicles and at vehicle boarding platforms or terminals from 7:00 am to 7:00 pm. Platform measurements had a target length of 2 minutes and captured noise levels of vehicles passing by a station (e.g., express trains at local stops) or arriving and departing from a station. Measurements aboard vehicles had a target length of 10 minutes while vehicles were in motion. To ensure measurement consistency and to mimic typical commuter noise exposures, the researchers sat approximately in the middle of the transit vehicle and stood at the center of the platform or terminal.

To account for variations in acoustics, passenger loads, and ambient noise levels, we made multiple in-vehicle and platform measurements for each mode of transit. We collected subway data on each of the 26 Metropolitan Transit Authority (MTA) subway lines and all 4 Port Authority Trans-Hudson subway lines. We made 6 measurements for each subway line: 2 in subway cars (1 on an aboveground and 1 on an underground track section) and 4 on station platforms (2 aboveground and 2 underground). We made platform measurements at a mixture of local stops and major high-traffic hubs.

We took measurements of commuter railway noise levels on the Metro-North Railroad (2 lines), Long Island Railroad (3 lines), and the Staten Island Railroad (1 line). We made 6 measurements for each commuter railway line: 2 in train cars (1 on an aboveground and 1 on an underground track section), and 4 on station platforms (2 aboveground and 2 underground). We made bus measurements aboard public New York (MTA) buses (13 lines) and while waiting at street-level bus stops at a variety of locations, including in residential neighborhoods, in commercial areas, and near airports. We also made measurements aboard the Staten Island ferry and at the ferry terminal, as well as aboard the Roosevelt Island Tramway to Manhattan and at the tramway terminal.

During measurements, researchers recorded on a paper time–location log the type of transit vehicle or boarding area being measured, their location on the transit route, the surrounding environment (aboveground or underground), the time and duration of the measurement, and any unusual circumstances during the measurement (e.g., musicians on platform or inside cars). At the conclusion of each measurement day, we entered time–location log data into an Excel file (Microsoft Corp, Redmond, WA). Dosimetry data was then downloaded using QuestSuite software (Quest Technologies, Oconomowoc, WI), exported into Intercooled Stata version 9.0 (Stata Corporation, College Station, TX) and combined with the time–location log data for analysis.

Analyses

We conducted analyses by individual transit system (e.g., MTA subway, Port Authority Trans-Hudson subway) as well as by transit type (e.g., subway, commuter rail).

We computed descriptive statistics for each system and type of transit by station and line and by measurement location (i.e., vehicle or station) and surrounding conditions (i.e., aboveground or underground). We also computed the percentage of measurements exceeding various exposure thresholds by transit system and type and by measurement location. We used 1-way repeated-measures analysis of variance (ANOVA) to compare statistical differences in L_eq levels by transit system and type between measurement locations, surrounding conditions, and borough or region within transit system and type as well as between individual stations and lines within transit systems. We considered statistical test results significant at the .05 level.

RESULTS

There were 243 valid dosimetry measurements. Table 1 provides the type and number of noise measurements collected and the L_eq noise levels associated with each transit type and system and by measurement location. Underground and aboveground measurements were combined in this table. Combined mean L_eq levels for the transit systems ranged from 75.1 dBA (Metro-North) to 80.4 dBA (MTA subway). The highest mean L_eq noise level inside a vehicle (79.3 dBA) was associated with MTA subway cars. MTA subway platforms also had the highest mean platform noise levels (81.1 dBA). The highest individual in-vehicle L_eq measurement (data not shown) was associated with an underground MTA subway car (87.9 dBA), and the highest individual platform L_eq measurement (90.2 dBA) was associated with an underground MTA subway station.

TABLE 1.

Average (L_eq) Noise Levels in dBA, by Transit Type and Measurement Location: New York City, June and July 2007

	Combineda Leq Levels		Leq Levels Inside Vehicle		Leq Levels at Platforms or Terminals
Transit Type or System	No.b	Mean dBA (SD)	No.b	Mean dBA (SD)	No.b	Mean dBA (SD)	Pc
Subway
MTA	156	80.4 (4.3)	60	79.3 (3.1)	96	81.1 (4.7)	.01
PATH	12	79.4 (3.3)	4	79.2 (4.2)	8	79.5 (3.1)	.89
Commuter rail
LIRR	18	74.9 (5.8)	6	71.4 (3.8)	12	76.6 (6.0)	.07
SIRR	3	76.7 (0.6)	2	76.5 (0.5)	1	77.2d	.43
Metro-North	11	75.1 (5.1)	4	71.9 (1.6)	7	77.0 (5.5)	.10
Bus	30	75.7 (3.7)	14	75.3 (2.6)	16	76.0 (4.4)	.62
Ferry	4	75.3 (3.1)	2	77.7 (2.1)	2	72.9 (1.1)	.09
Tram	4	77.0 (3.1)	2	77.5 (2.3)	2	76.6 (4.7)	.83

Note. L_eq = Equivalent continuous noise level; dBA = A-weighted decibel; MTA = Metropolitan Transportation Authority; PATH = Port Authority Trans-Hudson; LIRR = Long Island Rail Road; SIRR = Staten Island Rail Road; Metro-North = Metro-North Railroad.

^aAll L_eq levels across all measurement locations.

^bThe number of noise measurements taken.

^cCalculated with 1-way analysis of variance by measurement location.

^dSingle measurement.

Noise levels were 1 to 5 dBA higher at platforms and terminals than in vehicles for all types of transit except ferries and the tram. Vehicle and platform levels were significantly different only for the MTA subways. Subway L_eq levels differed significantly (data not shown; 1-way ANOVA, P < .001) between local stations (mean = 79.0 dBA) and major hubs (mean = 82.2 dBA) and across all subway stations (P < .001) and lines (P = .02). Overall, L_eq noise levels differed significantly across all transit types (1-way ANOVA, P < .001) but were not significantly different among the 3 commuter rail systems or the 2 subway systems.

We determined the percentage of L_eq measurements exceeding 2 threshold levels (the 24-hour 70-dBA WHO and EPA suggested exposure limit and the National Institute for Occupational Safety and Health 8-hour 85-dBA Recommended Exposure Limit) to evaluate the fraction of measurements with the potential to produce overexposure situations given sufficient exposure durations. Nearly all bus measurements, 3 out of 4 commuter rail measurements, and 100% of subway, ferry, and tram L_eq measurements exceeded the 70 dBA threshold. Almost 20% of the subway measures exceeded the 85 dBA threshold. Two subway lines (7%) had mean vehicle L_eq levels greater than 85 dBA, and 7 subway lines (23%) had mean L_eq platform levels greater than 85 dBA.

Table 2 shows the L_max levels associated with each transit type and system and by measurement location. As in Table 1, underground and aboveground measurements are combined. MTA subways had the highest maximum noise levels on average (90.4 dBA). The highest L_max level among all platform measurements was at an MTA subway station (102.1 dBA), followed closely by a bus stop measurement (101.6 dBA). The 2 highest L_max levels among all vehicle measurements were on an MTA subway car (97.8 dBA) and a bus (96.8 dBA). L_max noise levels were on average about 2 dBA higher at platforms and terminals than in vehicles for 2 transit types (commuter rail and buses) and 1 to 5 dBA lower for the other transit types. Roughly half the 30 measured subway lines had average vehicle and platform L_max levels that exceeded 90 dBA. Vehicle and platform levels were significantly different only for the ferry.

TABLE 2.

Maximum (L_max) Noise Levels in dBA, by Transit Type and Measurement Location: New York City, June and July 2007

	Combineda Lmax Levels			Lmax Levels Inside Vehicle			Lmax Levels on Platforms or Terminals
Transit Type or System	No.b	Mean dBA (SD)	Highest dBA Level	No.b	Mean dBA (SD)	Highest dBA Level	No.b	Mean dBA (SD)	Highest dBA Level	Pc
Subway
MTA	156	90.4 (4.6)	102.1	60	90.5 (3.6)	97.8	96	90.3 (5.2)	102.1	.75
PATH	12	88.1 (3.8)	94.9	4	88.3 (4.5)	94.9	8	88.0 (3.7)	92.6	.91
Commuter rail
LIRR	18	84.9 (6.0)	97.3	6	83.8 (5.2)	92.4	12	85.5 (6.5)	97.3	.59
SIRR	3	90.4 (3.2)	93.0	2	92.2 (1.2)	93.0	1	86.8d	86.8	.17
Metro-North	11	86.5 (6.1)	99.5	4	82.2 (1.1)	83.4	7	89.0 (6.5)	99.5	.07
Bus	30	86.8 (6.1)	101.6	14	85.6 (4.7)	96.8	16	87.8 (7.1)	101.6	.34
Ferry	4	89.9 (3.0)	92.5	2	92.5 (0.0)	92.5	2	87.4 (0.1)	87.4	<.001
Tram	4	88.7 (6.5)	93.9	2	90.9 (1.1)	91.7	2	86.6 (10.4)	93.9	.62

Note. L_max = maximum noise level; dBA = A-weighted decibel; MTA = Metropolitan Transportation Authority; PATH = Port Authority Trans-Hudson; LIRR = Long Island Rail Road; SIRR = Staten Island Rail Road; Metro-North = Metro-North Railroad.

^aAll L_eq levels measured inside vehicles and at platforms or terminals.

^bThe number of noise measurements taken.

^cCalculated with 1-way analysis of variance by measurement location.

^dSingle measurement.

Table 3 shows L_eq noise levels for subway and commuter rail measurements stratified by location (i.e., inside vehicle vs platform) and surroundings (i.e., aboveground vs underground). Mean vehicle and platform L_eq levels were equal for aboveground subway measurements. Mean aboveground vehicle L_eq levels were 1 to 5 dBA lower than were aboveground platform levels. Mean vehicle L_eq levels for underground measurements were always lower than those for underground platforms—in the case of commuter rail, vehicle levels were as much as 11 dBA lower. Aboveground vehicle and platform levels were not significantly different for any transit system; however, underground vehicle and platform levels differed significantly for the MTA subways, Long Island Railroad, and Metro-North systems.

TABLE 3.

Average (L_eq) Noise Levels in dBA for Subway and Commuter Rail Systems, by Measurement Location and Surroundings: New York City, June and July 2007

	Leq Level Inside Vehicle		Leq Level at Platforms or Terminals
Transit Type or System	No.a	Mean dBA (SD)	No.a	Mean dBA (SD)	Pb
Aboveground
Subway
MTA	24	77.9 (3.1)	42	77.9 (3.3)	0.98
PATH	…	…	…	…	…
Commuter rail
LIRR	3	69.6 (3.9)	9	74.8 (5.8)	0.18
SIRR	2	76.5 (0.5)	1	77.2c	0.43
Metro-North	2	71.2 (1.3)	5	74.3 (3.5)	0.29
Underground
Subway
MTA	36	80.2 (2.8)	54	83.5 (4.3)	<.001
PATH	4	79.2 (4.2)	8	79.5 (3.1)	.88
Commuter rail
LIRR	3	73.1 (3.4)	3	82.0 (2.3)	.02
SIRR	…	…	…	…	…
Metro-North	2	72.5 (2.0)	2	83.9 (0.4)	.02

Note. L_eq = Equivalent continuous noise level; dBA = A-weighted decibel; MTA = Metropolitan Transportation Authority; PATH = Port Authority Trans-Hudson; LIRR = Long Island Rail Road; SIRR = Staten Island Rail Road; Metro-North = Metro-North Railroad.

^aThe number of noise measurements taken.

^bCalculated with 1-way analysis of variance by measurement location.

^cSingle measurement.

Table 4 shows mean L_eq noise levels for all transit systems and types by borough or region and environment. The effect of borough or region on L_eq noise levels was significant for MTA subways, the Long Island Railroad, buses, and Metro-North. For each of these transit types and systems, the highest measurements were associated with the borough of Manhattan, which had a mean L_eq level 3 to 9 dBA higher than other boroughs. Surrounding environment (aboveground or underground) had a statistically significant effect for only 1 transit system: MTA subways.

TABLE 4.

Mean (L_eq) Noise Levels in dBA, by Transit Type, Borough, and Surroundings: New York City, June and July 2007

	Borough						Surroundings
Transit Type or System	Bronx, Mean dBA	Brooklyn, Mean dBA	Manhattan, Mean dBA	Queens, Mean dBA	Staten Island, Mean dBA	Pa	Aboveground, Mean dBA	Underground, Mean dBA	Pa
Subway
MTA	79.0	77.8	82.5	79.3	…	<.001	77.9	82.2	<.001
PATH	…	…	79.4	…	…	…	…	79.4	…
Commuter rail
LIRR	…	…	82.0	73.4	…	.01	73.5	77.6	.17
SIRR	…	…	…	…	76.7	…	76.7	…	…
Metro-North	71.0	…	77.5	…	…	.03	73.4	78.2	.14
Bus	…	…	77.1	72.9	…	<.001	75.7	…	…
Ferry	…	…	…	…	75.3	…	75.3	…	…
Tram	…	…	77.0	…	…	…	77.0	…	…

Note. L_eq = Equivalent continuous noise level; dBA = A-weighted decibel; MTA = Metropolitan Transportation Authority; PATH = Port Authority Trans-Hudson; LIRR = Long Island Rail Road; SIRR = Staten Island Rail Road; Metro-North = Metro-North Railroad.

^aCalculated with 1-way analysis of variance by measurement location.

DISCUSSION

The results of this mass transit noise survey confirm our pilot study finding that transit noise levels can present a risk of NIHL given sufficient exposure duration. We found that subways had the highest mean L_eq noise levels (80.4 dBA) and the highest individual measured L_eq (90.2 dBA on a subway platform) among the types of transit assessed. On average, subways also had the highest mean L_max level (90.4 dBA) and the highest measured L_max (102.1 dBA on a subway platform) among the transit modes assessed. In comparison, 30 dBA is the noise level of a whisper, 60 to 70 dBA is normal conversation, 100 dBA is a chainsaw, and 140 dBA is gunfire. It is important to note that decibels are logarithms and that the risk of NIHL from noise rises quickly with small increases in exposure level. For example, a 95-dBA noise level is 10 times more intense than an 85-dBA noise level and 100 times more intense than a 75-dBA noise level.

Nearly 1 in 5 subway L_eq measures exceeded 85 dBA, and roughly half of subway L_max measurements exceeded 90 dBA. L_eq levels associated with the other transit types evaluated—commuter rail, buses, ferries, and a tramway—were 3 to 5 dBA lower than subway noise levels but still potentially presented a risk of NIHL. The transit noise levels reported by Diaz et al.²⁸ (average L_eq noise levels from the study were 75.6 dBA for buses, 78.8 dBA for subways, and 76.0 dBA for commuter trains) are remarkably consistent with those we measured. Diaz et al. found that transportation noise accounted for about 13% of participants' total noise exposure, and for participants older than 60 years, transportation noise was the primary source of noise exposure.

Zheng et al.²⁹ found that the average amount of time spent commuting was 40 minutes and the mean L_eq noise level was 76.1 ±7.8 dBA—consistent with the measures for several of the transit types examined here. Commuting noise contributed approximately 8% of the total noise exposure among this group of participants. Zheng et al. did not report the types of transit assessed in the study, but presumably, in a large urban setting such as Beijing, a variety of transit types, including some of those assessed here, were utilized.

A relatively recent (1996) study of stations and cars in the underground subway system in Calcutta, India, by Bhattacharya et al.³² found point-in-time sound pressure levels of 84 to 87 dBA in the 3 stations assessed, with the highest level measured in the only aboveground station. These levels were 7 to 10 dBA higher than the time-integrated L_eq levels we measured. The maximum sound pressure level in a station was 95 dBA, similar to the L_max levels in our current study. L_eq noise levels of 92 to 99 dBA (much higher than the levels we measured) were found aboard the subway cars during operation, with vehicle levels dropping to 72 to 75 dBA during stops. The generally higher levels measured in the Calcutta subway system may be at least partly explained by differences in system construction, public-address system configuration, and subway car design.

When compared with our current results, the 1931³⁰ and 1971³¹ New York City subway studies suggest that subway noise levels have declined over time, although such a comparison must consider possible differences in measurement equipment and protocols. Sound pressure levels in the 1931 study by Stanton³⁰ ranged from 87 to 97 dBA, with measurements taken on subway platforms and on the tracks themselves. The 1971 study by Harris and Aitken³¹ found sound pressure levels that ranged from 87 to 110 dBA, with the highest levels on Queens and Manhattan lines, both at the platform level and inside cars. Harris and Aitken found that certain subway cars, especially those manufactured before 1970, had higher levels than newer cars.³¹

A number of studies have assessed noise exposures associated with commuting in automobiles, the primary alternative to mass transit in an urban setting. In a study of nonoccupational noise exposures among 112 construction workers, Neitzel et al.³⁴ found that commuting by car or bus had mean L_eq levels of 76 to 78 dBA. In another study of nonoccupational noise, Schori and McGatha³⁵ found that many of the highest L_eq sound levels measured on 50 participants were associated with riding in cars. Automobile noise levels in that study ranged from 76.9 to 78.3 dBA.

In an early report,³⁶ the EPA estimated that passenger cars have mean interior noise levels of 67 dBA at 30 miles per hour and 77 dBA at 60 miles per hour. More recently, Diaz et al. found a mean L_eq noise level for passenger cars of 79.7 dBA.²⁸ Although limited, these data suggest that noise levels associated with automobile commuting are comparable to the L_eq levels we measured for buses, ferries, trams, and commuter rails and are lower than those measured for subways.

Comparison of Current Results To Pilot Study

To assess the validity of the noise levels estimated in our pilot study,³ we compared them to the levels we measured for the current study. The mean L_eq measured on underground subway platforms in the current study (83.5±4.3 dBA) was significantly different (Student t test, P = .006) from the mean sound pressure level in the pilot study (85.7±3.9 dBA), though the absolute difference was only about 2 dBA. The mean L_max measured on underground subway cars in the current study (91.0±3.2 dBA) was also significantly different (P = .01) from the mean in the pilot study (94.9±7.1 dBA). Measured L_max levels at aboveground bus stops did not differ significantly between the current study (87.8±7.1) and the pilot study (84.1±4.5 dBA).

Differences in the subway car and platform levels measured in the 2 studies are likely caused by measurement protocol and equipment differences. In our pilot study, measurements were made with a sound level meter held approximately 3 feet off the ground and in front of the researcher's body, whereas researchers in our current study used noise dosimeters with microphones mounted next to their ears. Having the microphone of the sound level meter in front of the researcher's body and relatively close to the ground may have produced acoustic reflections that increased the noise levels measured by the sound level meter, whereas the dosimetry protocol in the current study collected levels that were more representative of the true exposure at the researcher's ear. We made measurements in the pilot study during nonpeak commuting hours (10:00 am to 4:00 pm), whereas measurements in the current study included rush hour (7:00 am to 7:00 pm). The more-densely packed rush hour cars may have lower ambient noise levels because of sound absorption by clothing and human bodies.

Our pilot study focused on high-noise events (i.e., measurements were made only when subway cars were entering or leaving a station, and only maximum levels were recorded in subway cars), whereas measurements in our current study focused instead on obtaining average exposure levels and included times of relative quiet. The sound level meters used in the pilot study could not make time-integrated L_eq measurements; instead, the researcher read and recorded the noise level on the sound level meter display at 5-second intervals. We then arithmetically averaged these interval readings to determine an average level for each measurement, which may have introduced error into the results. The dosimeters in the current study made time-integrated L_eq measurements, and no additional averaging or manipulation was required. Overall, the time-integrated personal noise levels measured in the current study must be considered a more robust representation of commuter exposures than the area noise levels measured in our pilot study, because the current study involved a much larger number of measurements made at a more representative measurement location (i.e., the ear).

Conclusions

Our results confirm that, given sufficiently long exposure durations, noise levels associated with mass transit are high enough to produce NIHL in riders. We noted significant differences between the mean levels of various transit types evaluated and between subway lines, stations, and station types. One borough (Manhattan) consistently had the highest associated L_eq levels. Subways (including cars and platforms) had the highest associated mean L_eq and L_max noise levels (80.4 and 90.4 dBA, respectively) of all transit types evaluated. At the noise levels measured in the subway, exposures of a few hours to as little as 2 minutes a day (in the case of the highest L_max level measured, 102.1 dBA) would be expected to cause hearing loss for some people given chronic exposure. Other types of transit had mean L_eq noise levels 3 to 5 dBA lower than the subway system but still above the NIHL risk threshold of 70 dBA averaged over a 24-hour period.³⁷

L_eq noise levels were higher at platforms than on vehicles for subways, commuter rail, and buses, whereas ferries and the tram had higher L_eq noise levels on vehicles than at stations. L_eq levels in underground subway and commuter rail cars and stations were higher—in some cases by 10 dBA or more—than those in aboveground cars and stations. L_max noise levels were higher aboard vehicles than at stations for the subway, ferry, and tram systems. The effect of surroundings (i.e., aboveground vs underground), borough or region, and measurement location (i.e., platform vs in vehicle) are important considerations for subway and commuter rail transit noise-exposure assessments.

Additional data are needed on average commute duration by transit type for transit riders. By combining exposure durations with the noise levels measured in our current study, estimates of annual rider noise exposures can be developed. These estimates can then be incorporated into the available International Organization for Standardization model for the prediction of hearing impairment from noise.²⁷ This model will allow us to infer the percentage of transit riders with probable NIHL resulting from transit noise. These inferences can then be used to assess the need for transit noise-control efforts or the use of hearing protection, which have the potential to reduce both the risk of NIHL and other adverse effects of excessive noise.

In accordance with the “hierarchy of controls” for public health hazards,³⁸ engineering noise-control efforts, including increased transit infrastructure maintenance and the use of quieter equipment, should be given priority over use of hearing protection, which requires rider motivation and knowledge of how and when to wear it. Given the various nonauditory health effects associated with noise exposure and the large percentage of US residents exposed to transit noise, noise-abatement efforts have the potential to benefit the public's health.