Assessment of Airborne Exposure to Trihalomethanes from Tap Water in Residential Showers and Baths

Transcription

1 Risk Analysis, Vol. 20, No. 5, 2000 Assessment of Airborne Exposure to Trihalomethanes from Tap Water in Residential Showers and Baths B. D. Kerger, 1 * C. E. Schmidt, 2 and D. J. Paustenbach 3 This study evaluates airborne concentrations of common trihalomethane (THM) compounds in bathrooms during showering and bathing in homes supplied with chlorinated tap water. Three homes in an urban area were selected, each having three bedrooms, a full bath, and approximately 1,000 square feet of living area. THMs were concurrently measured in tap water and air in the shower/bath enclosure and the bathroom vanity area using Summa canisters. Chloroform (TCM), bromodichloromethane (BDCM), and chlorodibromomethane (CDBM) were quantified using U.S. Environmental Protection Agency (EPA) Method TO- 14. Air samples were collected prior to, during, and after the water-use event for 16 shower and 7 bath events. Flow rate and temperature were measured, but not controlled. The increase in average airborne concentration ( standard error) during showers (expressed as g/m 3 in shower enclosure or bathroom air per g/l in water) was for TCM, for BDCM, and for CDBM (n 12), and during baths was for TCM, for BDCM, and for CDBM (n 4). The relative contribution of each chemical to the airborne concentrations was consistent for all shower and bath events, with apparent release of TCM BDCM CDBM. The results are therefore consistent with their relative concentration in tap water and their vapor pressures. When the shower findings for TCM are normalized for water concentration, flow rate, shower volume, and duration, the average exposure concentrations in these urban residences are about 30% lower than those reported by other investigators using EPA analytical methods. This difference is likely attributable primarily to greater air exchange rates in residential shower/bath stalls compared to more airtight laboratory shower chambers. This appears to be the first field study to thoroughly evaluate THM exposures from residential showers and baths, and can be used to validate previously published models of tap water volatile chemical transfer to indoor air. KEY WORDS: Chloroform; trihalomethanes; airborne; shower; bath 1. INTRODUCTION Exposures to airborne vapors due to release of volatile organic compounds (VOCs) from domestic tap water use have been a topic of interest for more 1 Health Science Resource Integration, Inc., Tallahassee, FL. 2 Consultant, Red Bluff, CA. 3 Exponent, Menlo Park, CA. *Address correspondence to Brent D. Kerger, Health Science Resource Integration, Inc., 2976 Wellington Circle West, Tallahassee, FL 32308; brentkerger@worldnet.att.net. than a decade. (1 5) One reason for this is that a large fraction of the U.S. population is exposed to chloroform (trichloromethane, TCM) and other trihalomethanes (THMs) that are present in tap water that is chlorinated for disinfection purposes, particularly those water supplies derived from surface waters. (6 8) Published measurements and modeling exercises have suggested that inhalation exposures to THMs and other VOCs during showering may be appreciable compared to daily ingestion exposures. (5,9 19) Much of the published data pertaining to the air /00/ $16.00/ Society for Risk Analysis

2 638 Kerger, Schmidt, and Paustenbach borne vapor hazard associated with chlorinated VOCs in tap water are based on miniature, or otherwise modified, laboratory showers that may not reflect the likely exposures in the showers present in typical residences. (1,2,20 22) In addition, the available studies vary with respect to sampling and analytical procedures. These studies provide somewhat divergent results for compounds such as TCM and trichloroethylene (TCE) that have relatively similar physical/ chemical properties with respect to solubility, volatility, Henry s law constants, and estimated water-to-air transfer coefficients relative to radon (TCM 0.75; TCE 0.69). (3) TCM and TCE have been the subject of volatilization measurements in investigations using full-size (i.e., 1.5 m 3 ) shower stalls. (11,13,15,16,23) Key aspects of the experimental conditions for each of these studies are summarized in Table I. The present study, however, is the first report of THM measurements taken in homes using constrained conditions. The discrepancies in airborne concentrations of TCM and TCE volatilized from tap water are undoubtedly due, in part, to differences in shower configuration and the sampling and analysis methods utilized in each study. Depending on which study or data set is used for valid predictions of shower exposures, the calculated inhalation dose could vary by an order of magnitude or more (see Table I). If shower configuration is the key consideration, then one might rely more heavily upon the earlier study of Giardino et al., (13) which assessed showers in actual residences, where artificial laboratory shower conditions are not at issue. Another key difference though between the available studies is the accuracy of the sampling and analysis method. While the Miran infrared analyzer utilized by Giardino and colleagues is useful for providing near-real-time readings to examine relative concentration changes during a shower event, the accuracy of this instrument is limited, especially in humid environments and at relatively low analyte concentrations (e.g., below the detection limit range of 1,000 to 5,000 g/m 3 ). All of the measured concentrations reported by Giardino et al. (13) were near or below the typical detection limits for the instrument. Moreover, the results of Giardino and colleagues are consistently higher on average than those studies using gas chromatography (GC) quantitation (11,15,23), and the influence of humidity or the presence of other chemicals tends to cause an artificially higher instrument response on a Miran analyzer. In this case, interferences would cause an artificially higher reading for airborne concentration of TCM or TCE using the Miran analyzer, which could not occur using GC techniques. Accordingly, it seems likely that the discrepancies between the Giardino et al. (13) results and those of Jo et al. (11), Tancrede et al., (15) and Keating et al. (23) could be substantially due to analytical quantitation techniques. The purpose of the present study is to measure the average airborne concentrations of TCM and two other THMs generated from showering and taking baths in chlorinated tap water (public water supply) in three homes located in an urban area of the southwestern United States. To our knowledge, this is the first study that provides measurements of airborne exposure to VOC water contaminants during bath events. Several factors were examined that could appreciably influence the measured concentrations in each of the shower/bath enclosures, and at the bathroom vanity area during and after a bathing event, respectively. The testing protocol was designed to examine potential influences of (1) shower and bath characteristics (temperature, flow rate, duration, total water volume, bathroom ventilation), and (2) concurrent changes in hot and cold water concentrations of TCM. It was also designed to provide a relatively robust set of shower data for comparison to other studies using similar (GC) or different (Miran analyzer) quantitation techniques. 2. METHODS All procedures, including incorporation of appropriate blanks and spiked samples, were defined in a Quality Assurance Project Plan (QAPP) document that was written and peer reviewed prior to conducting the study. The exposure assessment methodology was designed to be consistent with that described by Lindstrom and Pleil (24) for residences using VOCcontaminated water Shower and Bath Event Protocols A standardized water-use and sampling protocol was implemented in each of three, three-bedroom homes of similar (950 1,100) square footage that were typical in age and construction to the majority of other homes located within an older (1950s 1960s) urban development. Each house had a similar size main bathroom with a standard, full-size bathtub. All three homes had the same size shower enclosure (2.8 m 3 ) equipped with a standard size plastic shower curtain hung at approximately 2 m, with the space above the curtain (about 18 cm 1.5 m) being open to the bathroom.

3 Table I. Summary of Key Aspects and Results of Published Studies of Chloroform and Trichloroethylene Exposures in Full-Size Showers Study Analytical method Shower type and volume Chemical tested n Water concentration ( g/l) Flow rate (L/min) Shower duration (min) Temp. Range ( C) Unadjusted unit exposure a ( g/m 3 per g/l) (mean std. dev.) THM Exposure in Showers and Baths 639 Adjusted unit exposure b ( g/m 3 per g/l) (mean std. dev.) TCM min Jo et al. (1990) (11) GC/ECD TWA Lab shower, volunteer entry, 1.5 m 3 Tancrede et al. (1992) (15) GC/ECD TWA Lab shower, no entry, 1.5 m 3 TCM , TCE , Keating et al. (1997) (23) GC/ECD TWA Lab shower, no entry, 1.53 m 3 TCM ( jet) 3 55,600 21, c TCM (spray) 3 56,600 7, c Current study Giardino et al. (1992) (13) Giardino et al. (1996) (16) GC/ECD & GC/MS TWA Miran 1A infrared analyzer (instant) Miran 1A infrared analyzer (instant) Home shower, volunteer TCM (shower 1) entry, 2.8 m 3, 3 homes TCM (shower 2) TCM (shower 3) TCM (all) composite Home shower, no entry, TCE min m 3, 4 homes Lab shower, no entry, 1.5 m 3 TCM , TCE Note: GC/ECD is gas chromatography with electron capture detection; GC/MS is gas chromatography with mass spectrometric detector; TWA is time-weighted average; TCM is chloroform; TCE is trichloroethylene. a The unadjusted unit exposure concentration represents the measured shower air level representing the TWA g/m 3 exposure during the shower event divided by the water concentration, in g/l. b The adjusted unit exposure concentration is derived by making linear adjustments in the unadjusted unit exposure concentration to standardize the shower event to 1.5 m 3 shower volume and 100 L total water volume (flow duration). If only peak (end-of-shower) or mass emission rate data were reported, TWA exposure levels for t 0 to 10 min were estimated based on estimated mass emitted, divided by shower volume adjusted for reported air exchange rate. If shower-specific data were not reported for a given parameter, a midrange value was assumed for the calculation. For instantaneous measurements, area under the concentration versus shower time curve was integrated to calculate exposure concentration for t 0 10 min. c Variance after adjustment could not be calculated based on the data provided by the authors.

4 640 Kerger, Schmidt, and Paustenbach During 4 days of sampling, exposures were assessed for a total of 12 shower and six bath events following the protocols illustrated in Fig. 1. For each shower or bath event, an adult volunteer entered the shower/bath and went through normal washing activities during the timed event including using soap for body washing and shampoo/conditioner for hair washing. Water temperature and flow rate were measured, but not controlled. All of the tested homes had standard (not water-saver) showerheads. Each home had a central air-conditioning unit that was operating during the study, and the bathroom doors and windows remained shut during all measurements except for the ventilated shower event. Four additional showers were monitored on days that were considered to be off protocol due to home equipment failures; these data are presented separately. As illustrated in Fig. 1, the first shower of the day was an average-duration (6.8 min) shower with no room ventilation. The 6.8-min duration represents the 50th-percentile value for time spent in the shower according to the U.S. Environmental Protection Agency (EPA). (25) For this first shower, a 6.8-min preshower sample was collected via a sampling tube hung over the shower rod at a height of 1.5 m from the bathtub bottom. The shower water was then turned on and the temperature adjusted to the volunteer s preference. Within 2 min, the water flow rate and temperature were measured, the volunteer entered the shower, and the during-shower sample (again at a height of 1.5 m within the bath enclosure) commenced and was shut off at 6.8 min. As the volunteer exited the shower, the postshower sample was collected for 6.8-min, located at a height of 1.5 m to simulate a person standing directly in front of the bathroom vanity. Reference in subsequent figures to not ventilated means that there were no operating bathroom exhaust fans and the windows and doors to the bathroom were all shut during sample measurement. In all cases, the bathroom door was shut during the measurement for all on-protocol days. However, the central air-conditioning system, including a vent to the bathroom, was operating in each test residence. The protocol for the second shower event started approximately 1 hour after the first shower event, following identical operational and sampling methods with the exception that durations of the three sample types were varied. In this second shower event, a long -duration shower lasting 12 min was assessed, which corresponds to the 90th-percentile duration for time spent in the shower according to the EPA. (25) The preshower sample was also 12 min, and the postshower sample was 20 min in duration. The protocol for the third shower event was identical to that of the first shower (an average-duration shower) with the exception that this event was scheduled in the midafternoon and involved some degree of active ventilation. Two of the homes, Residences 1 and 3, had small bathroom exhaust fans (30 f 3 /min, no windows) that operated during this shower event, Fig. 1. Water use and sampling events timeline.

5 THM Exposure in Showers and Baths 641 and the third home, Residence 2, had no fan but did have a window that was opened 30 cm to facilitate ventilation. Two protocols were developed to examine THM exposures during bath events. A simulated babybath protocol involved taking a prebath sample of 6.8-min duration, filling the bathtub with approximately 8 cm of warm (35 37 C) water, and sampling the air in a simulated supine position breathing zone at 30 cm above the water surface at the center of the tub for 6.8 min for the during-bath sample. One of the volunteers splashed and mixed the water gently to simulate infant activity during the sampling event, but did not enter the bath. Bath water temperature was measured before and after the during-bath air sample was taken, and the water surface area was determined. The after-bath sample was omitted because it wasn t considered relevant to the infant exposure being assessed. To simulate an adult bath, a 20-min duration prebath sample was collected, followed by filling the tub to about 18 cm with hot (38 42 C) water, and sampling at 30 cm above the water surface at one end of the tub for 20 min. A volunteer entered the bath during the event, and used soap and shampoo to wash, as in the shower protocol. Water temperature was measured before and after the during-bath air sample was taken, and the water surface area was determined. The after-bath sample was subsequently taken at a height of 1.5 m, in front of the bathroom vanity, for 20 min. 2.2 Analytical Methods Separate hot and cold water samples were drawn from the kitchen sink at designated intervals and were sealed in VOA (volatile organic analysis) vials for analysis using EPA Method 601. Airborne vapor samples were collected at the designated intervals utilizing Summa canisters. Airborne concentrations of TCM were quantified from the Summa canister by GC using electron capture detection (EPA Method TO-14). EPA Method TO-14 using GC with mass spectroscopy (MS, Gelms) was used to confirm the presence of TCM and affirm the approximate concentration in approximately 20% of the air samples. EPA Method TO-14 is the state-of-the-art method for analyzing organic compounds in air. (26) The method used polished, stainless-steel Summa canisters as the sample collection and storage media. The canisters are cleaned and prepared in the laboratory and are shipped to the field with a 29 mercury vacuum. Ambient air samples are collected without the use of a pump by interfacing a low-flow air flow controller to the canister under vacuum and drawing an ambient air sample for a designated sample collection interval. As such, this method of sample collection is preferred over other integrated sample collection approaches because the air sample does not pass through a sample collection pump; the canister media is provided at ultraclean levels ( 20 parts per billion by volume [ppbv] total hydrocarbon compounds as methane and 0.2 ppbv individual compound levels); the sample collection media is whole air and does not rely on adsorption on media and desorption, which can result in artifact formation and sample bias; and the canister media has a recommended 14-day hold time (minimum) with excellent stability for extended hold time. Laboratory sample preparation includes pressurizing and transferring it to the gas chromatograph (GC) for separation. The canister is pressurized with ultra-high-purity (UHP) nitrogen to pounds per square inch gauge (psig). Nitrogen is added to the canisters to provide positive pressure for removing the sample, to dilute oxygen and moisture in the sample, and to minimize sample component reactions. VOCs from the air samples are concentrated at cryogenic temperatures by passing a known volume of sample into a cryotrap loop submersed in liquid nitrogen. The sample is then transferred by thermal desorption from the cryotrap into a capillary GC system for separation and analysis. The cryotrap sample preparation provides for a concentration of sample affording ultralow levels of analysis (i.e., sub-ppbv). After separation, the organic compounds are detected by an electron capture detector (ECD) with a method detection limit of 0.01 ppbv (about 0.05 g/ m 3 for TCM) for each THM compound studied Data Analysis and Statistical Methods The data, expressed as unit exposure concentrations (UECs), are calculated as follows: UEC ( g/m 3 per g/l) THM level during event ( g/m 3 ) THM level prior to event ( g/m 3 ) concurrent THM water concentration ( g/l), where the concurrent THM water concentration is the average of hot and cold measurements nearest in time to the monitored event. A data normalization procedure was applied in Table I to develop comparable estimates of duringshower exposure rates across the available published

6 642 Kerger, Schmidt, and Paustenbach studies due to varied shower conditions. The possible parameters of interest might include water temperature, water flow rate, duration of shower, volume of shower enclosure, air exchange rates within the shower, and showerhead type. No adjustment factor was applied for differences in water temperature, based on the observation of McKone and Knezovich (26) that there was no significant difference between the transfer efficiency of TCE in water to shower air measured with hot (37 C) or cold (22 C) shower water. Water flow rate, duration of shower, and volume of the shower enclosure are all parameters that may be linearly related to the mass of VOC transferred from water to air within the shower. (3,15) McKone and Knezovich (27) determined that duration of the shower does not significantly affect the transfer efficiency for TCE, with all other parameters held constant. The cumulative loading of VOCs must, however, be dependent on both water flow rate and duration of flow, and concentration must be dependent upon the air volume into which the VOCs are being loaded. Accordingly, a simple first-order (linear) relationship was assumed to exist between each of these factors and the measured shower air concentrations of TCM and/or TCE in the current study and the comparison studies. (11,13,15,16,23) Similar to the conditions tested in the most robust set of shower data to date, (11) all data were normalized to a shower enclosure volume of 1.5 m 3 and a shower flow duration of 100 L (10 L/min 10 min). An example normalization calculation is as follows: 100 L 2.8 m 3 (3.4 g/m 3 per g/l) 13.2 L/min 6.8 min 1.5 m g/m 3 per g/l, or measured value times measured values equals normalized value. Statistical comparisons were made using GB- STAT for Macintosh Version PPC (Dynamic Microsystems, Inc., Silver Spring, Maryland) student s t- test routines with a minimal significance level set at p RESULTS All of the water and airborne chemical measurements reported in the main (on-protocol) data set were determined to be valid with respect to the original study plans and the corresponding quality control samples (trip blanks, laboratory blanks, dupli- cates, and spiked samples) and instrument calibration data per procedures outlined in the QAPP. Every fifth air sample was also subjected to confirmation using MS, confirming the presence of the THM compounds reported (data not shown). Only a few of the air samples were rejected, due primarily to laboratory error or problems with the timing of sampling events (see annotations on the raw data in Tables III through V, discussed below) Water THM Concentrations THM concentrations in the tap water of each residence are summarized in Table II. The average of concurrent hot and cold water concentrations of each THM was used to calculate UECs from the airborne concentration data (i.e., expressing them in units of g/m 3 per g/l). In general, the hot and cold water THM measurements were similar generally within 15% to 20% although the hot water concentration was significantly higher (p.05) on average than the cold water concentration for all THMs on two test days and for two of three THMs on a third day (See Table II). Day-to-day differences in THM levels were also small generally 15% although average THM levels on the first test day were significantly lower than on the other days at the same or different residences ( p.05). The occurrence and relative concentrations of the three THMs assessed were consistent throughout the study. All three THMs were detected in all water samples. A very high degree of data consistency was found, as suggested by statistics for the ratio of TCM to bromodichloromethane (BDCM) concentrations (mean ratio of concurrent hot/cold averages, standard error of the mean [SEM]) as well as for the ratio of TCM to chlorodibromomethane (CDBM, ) Airborne THM Concentrations As seen in the water THM data, there was a high degree of data consistency with respect to the occurrence and relative concentrations of the three THMs measured during the shower/bath (in the bath enclosure) and after the shower/bath (at the bathroom vanity). Each THM was detected in both the duringand after-shower/bath events for the four on-protocol days. The ratios of TCM to BDCM ( SEM) and TCM to CDBM ( ) were consistent in the during- and after-shower sampling events (see Table II, outlier data based on 1.1 g/m 3,

7 THM Exposure in Showers and Baths 643 Table II. Concentrations of Trihalomethanes in Tap Water ( g/liter) Chloroform (TCM) Bromodichloromethane (BDCM) Chlorodibromomethane (CDBM) Location and date Time Cold Hot Average Cold Hot Average Cold Hot Average Residence 1 6:36 a.m :10 a.m :32 a.m :10 p.m :30 p.m Avg a,b,c c b,c SEM Residence 2 d 6:51 a.m :14 a.m :15 a.m :07 p.m :33 p.m Avg e SEM Residence 1 f (2nd sampling) 6:42 a.m :52 a.m :00 a.m :45 p.m :05 p.m Avg e e e SEM Residence 3 f 6:44 a.m :15 a.m :31 p.m :44 p.m :38 p.m Avg c e SEM Aggregate Overall avg Overall avg Overall avg average SEM 1.5 SEM 1.1 SEM 0.8 Note: U.S. Environmental Protection Agency Method 601 was used. SEM Standard error of the mean. a Statistically significant difference for average data set in comparison to average data set for Residence 2 based on results of two-tailed t-test and p.05. b Statistically significant difference for average data set in comparison to average data set for Residence 3 based on results of two-tailed t-test and p.05. c Statistically significant difference for average data set in comparison to average data set for Residence 1 (second sampling) based on results of two-tailed t-test and p.05. d Statistically significant difference for hot versus cold data set comparison for BDCM and CDBM based on results of two-tailed t-test and p.05. e Statistically significant difference for average data set in comparison to average data set for Residence 1 based on results of two-tailed t-test and p.05. f Statistically significant difference for hot versus cold data set comparison of TCM, BDCM, and CDBM based on results of two-tailed t-test and p.05. CDBM excluded from the calculation). Similar but slightly higher ratios were observed for TCM to BDCM ( ) and TCM to CDBM ( ). These ratios were not significantly different for the before- compared to the after-event measurements, or for the shower compared to the bath events (p.05) Shower Event Airborne THM Measurements Table III provides a summary of airborne THM concentrations measured before, during, and after each event for the three shower protocols, as explained in the Methods section. The release of THM vapors from water during all shower events showed a consis-

8 Table III. Summary of Airborne Trihalomethane Measurements and Related Data for Shower Events Water Water Chloroform ( g/m 3 ) Bromodichloromethane ( g/m 3 ) Chlorodibromomethane ( g/m 3 ) Residence flow temp. Shower type a location b (L/min) ( C) Preshower Shower Postshower Preshower Shower Postshower Preshower Shower Postshower Shower 1, 6.8 min, not vented Shower 2, 12 min, not vented c d d d Shower 3, 6.8 min, vented e a Shower types 1 and 3 involve 6.8-min samples before and during the shower event, and 6.8-min samples at the vanity after the shower. Shower 2 involves 12-min samples before and during the shower event, and 20-min samples afterwards. All samples taken at breathing zone height (1.5 m) with an adult actually bathing in the shower enclosure. b For Residences 1, 2, and 3, the total bathroom volumes were measured as 10.8, 11.3, and 9.1 m 3, respectively. Bath stalls each had shower curtains and had a measured volume of 2.8 m 3 at all three locations. c Laboratory error prevented quantitation of this compound; pressurization and cryotrap procedure error. d Sampler error led to lost sample (sample collection continued erroneously when shower began). Substituted average of other three preshower samples in this data subset. e Vented shower event implies that either the bathroom exhaust fan was on or the bathroom window was opened (about 30 cm) during the sampling events. The bathroom door was shut for all shower and bath sampling events. 644 Kerger, Schmidt, and Paustenbach

9 THM Exposure in Showers and Baths 645 Fig. 2. Comparison of airborne trihalomethane (THM) unit exposure concentrations in the shower enclosure during three shower protocols. Unit exposure concentrations are calculated as the air concentration g/m 3 divided by the concurrent water concentration g/l of the THM. Note: Error bars represent standard error. tent concentration gradient of TCM BDCM CDBM. This gradient is in accordance with expected trends based on physical and chemical properties of these THMs, such as their vapor pressures, Henry s law constants, and molecular weight and size. Figure 2 illustrates the during-shower THM measurements expressed as UECs (average air concentration in g/m 3 per water concentration in g/ L). There were no statistically significant differences or apparent trends between the three shower protocols or between the three test residences with respect to the average unit exposure rates. The second shower protocol, however, involving the longer (12- min) shower duration, showed considerably less variance in UECs compared to the other two protocols involving an average (6.8-min) shower duration (see Fig. 2). Immediately after the shower, average airborne THM concentrations were measured over a similar duration (6.8 or 12 min) at adult breathing height in front of the bathroom vanity. As shown in Fig. 3, the UECs for each THM were about 30 to 50% lower on average than those observed in the shower enclosure during the event. No statistically significant differences were observed (at p.05) when comparing average unit exposure rates for the after-shower measurements between the three shower protocols, or between the three tested residences. A considerably lower variance was again observed, however, for the second shower protocol involving the 12-minduration shower. Two additional days of shower sampling were conducted, but were not included with the primary data set due to problems that occurred with the water heater, air-conditioning, or other house equipment that was functional during all on-protocol test days. These are called off-protocol days and are included for the sake of thoroughness. Table IV summarizes data collected for four additional showers, three of which were conducted at relatively cool water temperatures. Although there are other factors at issue for each measurement, the UECs for TCM and BDCM in the cooler showers both during (TCM average SEM; BDCM average ) and after the event (TCM average ; BDCM average ) are relatively consistent with the corresponding values in the on-protocol shower data set. In addition, the TCM and BDCM UECs calculated for the 12-min during-shower data (TCM, 3.4 and 3.7; BDCM, 2.3 and 1.7 g/m 3 per g/l) exhibited lower variance compared to the 6.8-min showers (TCM, 4.0 and 1.6; BDCM, 2.4 and 0.6 g/m 3 per g/l), consistent with the main shower data set Bath Event Airborne THM Measurements This study examined average airborne THM concentrations for two different bath-taking proto-

10 646 Kerger, Schmidt, and Paustenbach Fig. 3. Comparison of airborne trihalomethane (THM) unit exposure concentrations in the bathroom after three shower protocols. Unit exposure concentrations were calculated as the air concentration g/m 3 divided by the concurrent water concentration g/l of the THM. Note: Error bars represent standard error. cols involving a total of seven bath events: three baby-bath simulations (6.8-min duration), and four adult bath simulations (20-min duration). The raw data for airborne THM concentrations measured during each bath event are provided in Table V. The data suggest that the adult bath protocols resulted in somewhat lower average UECs (e.g., for TCM, average of SEM) compared to the baby-bath protocol (for TCM, ), likely due to expected decreasing THM levels over the longer adult bath event (about three times longer than the baby bath). No statistically significant difference (p.05) was found between UECs for the two bath protocols. The adult bath UECs for TCM and BDCM, however, were significantly lower ( p.05) on average than those for the 12-min shower event both during and after the event. The during- and after-bath THM UECs calculated for the adult bath are summarized in Fig. 4. The average THM UECs during the adult bath were about threefold lower than those observed for the various shower events. As in the shower data, a trend of decreasing UEC of TCM BDCM CDBM was again observed. 4. DISCUSSION One objective of this study was to identify the relative difference in airborne THM exposure for persons taking a typical bath compared to a shower. The data suggest that THM release and corresponding average airborne concentrations are about three times higher on average for a typical shower compared to a bath. For both baths and showers in a residential setting, it appears that shorter event duration is associated with higher average THM exposure concentrations and greater variability from one event to the next. These observed trends are likely due to the multiple variables that influence water-to-air transfer efficiency, atmospheric conditions, and air exchange characteristics within the shower/bath enclosure, also discussed by other investigators. (15,23,27) A second objective of this study was to examine the trends for selected THM concentrations in an urban water supply at different residences, over a series of days. In general, higher THM concentrations were observed in hot water samples, a small but in some cases statistically significant trend compared to concurrently sampled cold water from the same test residence. It is possible that separate sampling of hot and cold water supplies may be a source of error in calculating UECs for airborne TCMs in those instances where hot water concentrations were significantly greater than those for cold water. Not measured was the THM content of the mixed shower water; it was assumed that the average of hot and cold water THM concentrations would be an adequate surrogate. The significance of this possible source of

11 Table IV. Summary of Airborne Concentrations of Three Trihalomethanes During Showering for Off-Protocol Days Water Water Chloroform ( g/m 3 ) Bromodichloromethane ( g/m 3 ) Chlorodibromomethane ( g/m 3 ) Residence flow temp. Shower type location (L/min) ( C) Preshower Shower Postshower Preshower Shower Postshower Preshower Shower Postshower Shower 1, 6.8 min, not vented 1 a,c ND ND 12.9 ND Shower 2, 12 min, not vented 1 a,d ND Shower 1, 6.8 min, not vented 3 b,e Shower 2, 12 min, not vented 3 b,e ND ND Not detected (limit of detection was 0.05 g/m 3 ). a Off-protocol day represents data collected during days when there was intermittent failure of the gas hot water heater and central air conditioner. b Off-protocol day due to failure of electric hot water heater and door missing from bathroom. c Concurrent trihalomethane (THM) concentrations in water were 35.1, 29, and 20 g/l, respectively. d Concurrent THM concentrations in water were 38, 30, and 22 g/l, respectively. e Concurrent THM concentrations in water were 36, 40, and 32 g/l, respectively. Table V. Summary of Airborne Trihalomethane Measurements and Related Data for Bath Events Water Water Chloroform ( g/m 3 ) Bromodichloromethane ( g/m 3 ) Chlorodibromomethane ( g/m 3 ) Residence volume temp. Bath type location (L) ( C) Prebath Bath Postbath Prebath Bath Postbath Prebath Bath Postbath Baby bath b, 6.8 min, not vented NS NS ND 5.1 NS NS NS NS 1 c NS NS NS Adult bath d, 12 min, not vented e e e NS No sample taken, per protocol for baby bath event. ND Not detected (limit of detection was 0.05 g/m 3 ). a Surface area of water in bathtub measured as 0.71, 0.84, and 0.80 m 2 at Residences 1, 2, and 3, respectively. b Baby bath includes 6.8-min prebath sample and 6.8-min during-bath sample, with limited stirring and splashing with hands in an approximately 8-cm depth of bathwater to simulate an infant bath event. Air samples taken approximately 30 cm above water surface at center of tub. c Sampling was performed on an off-protocol day, but after problems with water heater and air conditioner were resolved. d Adult bath includes 20-min prebath sample, 20-min sample during the adult bath event in an approximate 18-cm depth of bath water, and a 6.8-min postbath sample at the vanity. Bathtub air samples were taken at approximately 30 cm above the water surface at the end of the tub opposing the side where fixtures were located. e Sampler error, no valid data for prebath background sample. Average of three other values in this data subset were substituted. THM Exposure in Showers and Baths 647

12 648 Kerger, Schmidt, and Paustenbach Fig. 4. Unit exposure concentrations for three trihalomethanes (THMs) during a bath and in the bathroom after an adult bath. Unit exposure concentrations were calculated as the reported air concentration g/m 3 divided by the concurrent water concentration g/l of the THM (n 4). Note: Error bars represent standard error. error is probably not appreciable, however, given the relatively small differential between the surrogate value (hot and cold average value) and the feasible actual value, which could approach but not exceed the hot water average (which was always less than 12% higher than the hot and cold average). The water concentration data for THMs assessed in this study are consistent with the study of Weisel and Chen, (28) who reported that heating water with a relatively high chlorine content can lead to increased THM concentrations over time compared to unheated tap water. Although free or total chloride concentrations were not assessed in the current study, the relatively high THM content of the cold water supply suggests a fairly high degree of chlorination. Relatively moderate (6 28%) increases in THM content of hot water were observed, however, while Wiesel and Chen (28) reported nearly doubled levels (e.g., for TCM) following 8 hr of water heating. A third objective of this study was to examine average exposure concentrations for THMs volatilized from tap water during a series of shower and bath events in unaltered residences. Selected variables that could affect water-to-air transfer of THMs were analyzed by monitoring key parameters (temperature, flow rate, enclosure volume, air and water THM content, and ventilation) and controlling the event duration and ventilation conditions, but otherwise allowing human volunteers to operate the shower according to their personal preferences. This study design led to fairly wide variations in measured flow rates and temperatures that may have affected all three shower protocols to different degrees. Although water temperature appears to have relatively little influence on shower exposure concentrations, (22,27) water flow rate and shower duration are key to the mass loading of THMs from water into the shower enclosure. Use of typical bathroom ventilation mechanisms (open window, exhaust fan) in the current study apparently had little effect on average THM exposure concentrations during a shower event. The bathroom door remained shut for all four shower events within the ventilated 6.8-min shower protocol, however, and effective air flow through the bathroom and shower enclosure was probably limited by this condition. In addition, the relatively high water flow rates measured in this data subset ( L/min) may have inadvertently masked an influence of ventilation by providing greater aerosolization and more rapid THM accumulation within the shower enclosure. It should also be noted that outside temperatures during the study were generally 90 F, which led to relatively continuous operation of the central air conditioners in these homes. Since each home had a small air-conditioning vent leading into the bathroom, a relatively consistent cooled air flow likely had some influence on the results.

13 THM Exposure in Showers and Baths 649 The shower configuration and parameters in the current study in many instances differ substantially from the potentially comparable full-size-shower studies published to date. (11,15,16,23) For example, most of the comparison studies utilized small (approximately 1.5 m 3 ) laboratory shower enclosures, whereas the current study examined three residential bath/ shower enclosures (2.8 m 3 ). As discussed in the introduction of this article, analytical quantitation methodology may also be an important consideration when comparing the current study results to the other relevant published literature. The current study found TCM UECs within the shower enclosure that averaged between 3 and 4 g/ m 3 per g/l in water for residential shower enclosures with a volume of 2.8 m 3. These data are consistent with the observations of Jo et al., (11) who reported relative TCM concentrations of 6 7 g/m 3 per g/l for a smaller (1.5 m 3 ) and more tightly enclosed laboratory shower. Linear adjustment of data in the current study to normalize the results using the smaller shower enclosure volume employed by Jo et al. (11) leads to an equivalent TCM UEC in the range of g/m 3 per g/l in water. Further adjustment for cumulative water flow into the shower enclosure (duration flow rate) results in an equivalent TCM UEC averaging between 3.3 and 5.6 g/m 3 per g/l in the shower enclosure, again based on the assumption that a linear relationship may exist between cumulative water flow and average UEC. For a literature comparison exercise, these simple normalization calculations for the current study findings were applied to five other studies of TCM and/or TCE volatilization in full-size showers (11,13,15,16,23) ; see Table I). After applying these normalization procedures, the present data compare favorably with the three other studies that utilized GC for quantifying airborne TCM, TCE, or both. (11,15,23) It should be noted that the Jo et al. (11) and Tancrede et al. (15) studies are most similar to the current study in terms of using GC for airborne chemical measurements, having water concentrations 60 g/l, and reporting a data set with at least four shower measurements. The Keating et al. (23) study, which also applied GC for airborne chemical measurements, differs substantially in having tested only one domestic showerhead in jet and spray modes, and using TCM concentrations averaging 50,900 and 60,000 g/l. The latter study also used water-saver devices that limited the water flow rate to only L/min, compared to the typical L/min flow rates tested in the other published studies. Table I illustrates that the most robust studies, by Jo et al. (11) and the current study, provide relatively close agreement in terms of UECs of TCM when normalized to represent similar shower conditions. The data reported by Giardino et al. (13) from measurements taken in actual residences and those by Giardino and Andelman (16) from a laboratory shower study are above those from the other available studies when normalized to standard shower conditions (Table I, adjusted unit exposure). One possible explanation for these discrepancies is inaccuracy of the airborne chemical quantitation method utilized. These two earlier studies utilized a Miran infrared analyzer to quantitate airborne chemical concentrations that apparently were not confirmed by GC techniques in either study. A second possibility is that the higher water concentrations utilized (primarily in the laboratory shower study) may exhibit different transfer efficiencies compared to the lower concentrations tested in most other studies. Table I shows that the Keating et al. (23) data collected for water TCM concentrations averaging 56,600 g/l correspond to an adjusted UECs averaging 8.4 (spray mode), similar to the range reported for all other GC/ECD studies examined (i.e., ). In contrast, the average of the adjusted UECs calculated for the Giardino et al. (16) data is 17, more than double that for the Keating et al. (23) study. Thus, it appears that use of the less rigorous infrared analyzer quantitation method may be the primary reason for the higher shower exposure data reported by Giardino and colleagues. (13,16) Shower enclosure air exchange characteristics are also possible explanations for difference between (and within) the available studies, as these are factors not included in the normalization procedures. The residential shower study of Giardino et al. (13) found lower average UECs compared to Giardino and Andelman s subsequent laboratory shower studies. (16) Similarly, the current study data show somewhat lower average UECs compared to the laboratory shower study of Jo et al. (11) Such trends might be expected based on the more open design of typical residential shower enclosures compared to the smaller laboratory enclosures that are designed to control the air flow characteristics within the chamber. Unfortunately, no air exchange rate data were collected in the current study to assess this more directly. Showerhead type is also probably an important factor in comparing the current study data to other available data. The interaction of water pressure, water flow rate, and the flow path resulting in shower spray may be important in determining the timing and extent of aerosolization within the shower enclo-

14 650 Kerger, Schmidt, and Paustenbach sure, which may in turn affect THM transfer efficiency and air exchange characteristics in a typical residential shower. (23) In contrast to the water-saving ( L/min) showerheads examined by Keating et al., (23) most of the other studies generated data using primarily standard domestic showerheads that are typically operated in the range of L/min. In any case, the lack of specific information to account for differences in showerhead performance between studies is another limitation of the comparisons illustrated in Table I. Differences in flow rate and showerhead type between residences in the current study might account for some of the variability observed within data for the three shower types, and overall. As a final note, the current study data can be used to examine the accuracy or conservatism involved in making certain simplifying assumptions regarding the average daily dose from inhalation of TCM or TCE vapors for shower or bath events. For example, one common risk assessment assumption is that bathing-related dose (including inhalation and dermal contact) may be approximated by simply doubling the ingestion dose at an assumed rate of 2 L of water ingested per day. For TCM exposures during and after a shower of 20 min (exposed to elevated vapor levels in shower and bathroom) per day at a UEC of 3.3 g/m 3 per g/l, an inhalation rate of 1 m 3 /hr, and an inhalation bioavailability of 60%, the estimated unit exposure rate would be about 0.65 g/day (per g/l), compared to an assumed complete absorption of 2.0 g/day (per g/l) for ingestion. A similarly calculated inhalation component for bath TCM exposure based on a UEC of 1.2 g/m 3 per g/ L would be only 0.24 g/day per g/l. Although the current study doesn t address dermal contact exposures, it appears that the inhalation component for shower and bath exposures is less than the ingested dose by a factor of approximately three to four. The current study data sets are useful for applying more sophisticated shower exposure modeling techniques and uncertainty analysis for THMs and similar VOCs. 5. CONCLUSIONS This study provides measurement data on THM concentrations in water and air that can be used to further examine exposure models and potential health risks from use of chlorinated water in residential showers and baths. It has been demonstrated that THM exposure concentration during a long (20-min) bath event in most cases is at least three times less on average than that for a long (12-min) shower. The present study residential shower THM measurements are consistent with, but slightly less than, the most robust study of TCM exposure concentrations in a laboratory shower setting, (11) as well as others that used similar laboratory shower chambers and GC quantitation of airborne levels of THMs and other chlorinated VOCs having similar Henry s law constants to TCE. (15,23) The slightly lower average THM unit exposure concentrations ( g/m 3 in air per g/l, in water) calculated for the 16 showers included in the current study are likely due to the greater air exchange rates in typical residential shower enclosures compared to more air-tight laboratory shower chambers. The current data also suggest that the considerable variability in residential shower measurements could be attributable to water flow rate, shower duration, and shower enclosure volume, all of which may strongly influence water-to-air transfer efficiency, atmospheric conditions, and air exchange characteristics within the shower/bath enclosure. ACKNOWLEDGMENTS This article is based on information presented at the 1998 Annual Meeting of the Society of Toxicology, March 1 5, 1998, in Seattle, Washington. The data collection and preliminary data analysis for this study were funded by defendants in a toxic tort lawsuit who solicited these measurements to evaluate the accuracy and relevance of modeling assumptions and results proffered by the plaintiffs experts. Publication of this article was funded by the individual authors and/or their employers. We gratefully acknowledge the assistance of Tom Flahive, Gwen Corbett, David Suder, and Mike Manning for their valuable contributions. REFERENCES 1. Decker, D., DiMardi, S. R., & Calabrese, E. J. (1984). Does chloroform exposure while showering pose a serious public health concern? Medical Hypotheses, 15, Andelman, J. B. (1985). Inhalation exposure in the home to volatile organic contaminants of drinking water. Science of the Total Environment, 47, McKone, T. E. (1987). Human exposure to volatile organic compounds in household tap water: the indoor inhalation pathway. Environmental Science and Technology, 21(12), Pleil, J. D., McClenny, W. A., & Oliver, K. D. (1989). Temporal variability measurements of specific volatile organic compounds. International Journal of Environmental and Analytical Chemistry, 37, Lindstrom, A. B., Highsmith, V. R., Buckley, T. J., Pate,