Zentox.com

Journal of Food Protection, Vol. 68, No. 4, 2005, Pages 758–763Copyright ᮊ, International Association for Food Protection Monochloramine Versus Sodium Hypochlorite
as Antimicrobial Agents for Reducing Populations of Bacteria
on Broiler Chicken Carcasses
SCOTT M. RUSSELL1* AND STEPHEN P. AXTELL2
1Department of Poultry Science, The University of Georgia, Athens, Georgia 30602-2772; and 2Zentox Corporation, 538-A Wythe Creek Road, MS 04-347: Received 23 July 2004/Accepted 20 November 2004 ABSTRACT
Studies were conducted to compare the effect of sodium hypochlorite (SH) versus monochloramine (MON) on bacterial populations associated with broiler chicken carcasses. In study 1, nominal populations (6.5 to 7.5 log CFU) of Escherichiacoli, Listeria monocytogenes, Pseudomonas fluorescens, Salmonella serovars, Shewanella putrefaciens, and Staphylococcusaureus were exposed to sterilized chiller water (controls) or sterilized chiller water containing 50 ppm SH or MON. SH at 50ppm eliminated all (6.5 to 7.5 log CFU) viable E. coli, L. monocytogenes, and Salmonella serovars; 1.2 log CFU of P.
fluorescens; and 5.5 log CFU of S. putrefaciens. MON eliminated all (6.5 to 7.5 log CFU) viable E. coli, L. monocytogenes,S. putrefaciens, and Salmonella serovars and 4.2 log CFU of P. fluorescens. In study 2, chicken carcasses were inoculatedwith P. fluorescens or nalidixic acid–resistant Salmonella serovars or were temperature abused at 25ЊC for 2 h to increase thepopulations of naturally occurring E. coli. The groups of Salmonella serovar–inoculated or temperature-abused E. coli carcasseswere immersed separately in pilot-scale poultry chillers and exposed to tap water (controls) or tap water containing 20 ppmSH or 20 ppm MON for 1 h. The P. fluorescens–inoculated group was immersed in pilot-scale poultry chillers and exposedto tap water (controls) or tap water containing 50 ppm SH or 50 ppm MON for 1 h. Carcasses exposed to the SH treatmenthad nominal increases (0.22 log CFU) in E. coli counts compared with controls, whereas exposure to MON resulted in a 0.89-log reduction. Similarly, average nalidixic acid–resistant Salmonella serovar counts increased nominally by 34% (41 to 55CFU/ml) compared with controls on carcasses exposed to SH, whereas exposure to MON resulted in an average nominaldecrease of 80% (41 to 8 CFU/ml). P. fluorescens decreased by 0.64 log CFU on carcasses exposed to SH and decreased by0.87 log CFU on carcasses exposed to MON. In study 3, SH or MON was applied to the chiller in a commercial poultryprocessing facility. E. coli counts (for carcass halves emerging from both saddle and front-half chillers) and Salmonellaprevalence were evaluated. Data from carcasses exposed to SH during an 84-day historical (Hist) and a 9-day prepilot (Pre)period were evaluated. Other carcasses were exposed to MON and tested during a 27-day period (Test). E. coli counts forsamples collected from the saddle chiller were 25.7, 25.2, and 8.6 CFU/ml for Hist, Pre, and Test, respectively. E. coli countsfor samples collected from the front-half chiller were 6.7, 6.9, and 2.5 CFU/ml for Hist, Pre, and Test, respectively. Salmonellaprevalence was reduced from 8.7% (Hist ϩ Pre) to 4% (Test). These studies indicate that MON is superior to SH in reducingmicrobial populations in poultry chiller water.
Chlorine in the form of sodium hypochlorite (SH), cal- number of substances such as fat, blood, fecal material, or cium hypochlorite tablets, or chlorine gas is the most com- protein. When chlorine is used in poultry processing facil- monly used disinfectant in the poultry industry in the Unit- ities to disinfect equipment surfaces, carcasses, or chiller ed States. The U.S. Department of Agriculture (USDA) systems, it encounters a very high organic load. Poultry Food Safety and Inspection Service (FSIS) allows for the process waters can have extremely high levels of total or- addition of chlorine to processing waters at levels up to 50 ganic carbon and a correspondingly high chemical oxygen ppm in carcass wash applications and chiller make-up water demand. Any free chlorine added to these high-demand wa- (9). Additionally, FSIS requires the application of chlori- ters is consumed rapidly, becoming unavailable for disin- nated water containing a minimum of 20 ppm available fection. If the chemical oxygen demand in these waters is chlorine on all surfaces of carcasses when the inner surfaces not satisfied, then a true free chlorine residual cannot be have been reprocessed (because of carcass contamination) established. A typical poultry chiller can have a chlorine other than solely by trimming (6). demand of 1,000 to 2,000 ppm that cannot be overcome by A major consideration when using chlorine as a dis- 50 ppm (maximum allowable by USDA) chlorine in the infectant is that free chlorine (hypochlorous acid, hypo- make-up water. Experiments conducted at the USDA–Ag- chlorite ion, or elemental chlorine) is highly reactive and ricultural Research Service (ARS) Western Regional Re- rapidly oxidizes, bleaches, or otherwise reacts with any search Center showed that a free chlorine residual could * Author for correspondence. Tel: 706-542-1368; Fax: 706-542-1827; not be established in a commercial poultry chiller even by adding up to 400 ppm of free chlorine (5). When chlorine MON VERSUS SH AS POULTRY CARCASS DISINFECTANTS reacts with organic material, it generally loses its microbio- Data were then analyzed with the Proc ANOVA procedure of cidal properties and can no longer act as a disinfectant (11). One disinfectant that has gained widespread acceptance The SH treatments throughout these studies were either 6 or and use in municipal potable water treatment facilities is 12.5% solutions. Monochloramine was manufactured at the timeof use by the controlled mixing of 6 or 12.5% SH and a solution monochloramine (MON). The controlled mixing of chlorine of Food Chemical Codex–grade ammonium chloride or 2% am- and ammonia in water generates this chlorine species.
monium hydroxide in tap water. Treatment concentrations were MON is tasteless, odorless, stable, highly soluble, persistent measured and verified by multiple methods, and devices including in water, and biocidal (10), and unlike free chlorine, it does ATI (Analytical Technology, Inc., Collegeville, Pa.) model A15/ not react readily with organic material (11). Because of 79 total Cl2 monitors, a Severn Trent (Charlotte, N.C.) 17T2000 these behavioral differences, many municipal potable water amperometric titrator, a Hach (Chicago, Ill.) DPD colorimetric an- plants switched from chlorine to MON in their distribution alyzer, a Hach Odyssey DR/2500 spectrophotometer, or Hach systems to lower the quantities of trihalomethanes (possible model CN-21P high-range chlorine test kits with sulfite I and sul- carcinogens) produced and, in so doing, have brought water famic acid powder pillows and sodium thiosulfate reagents.
plants into compliance with U.S. Environmental Protection Study 1: Effect of MON on pathogenic, indicator, and
Agency (EPA) requirements (11). Chloramine residuals of spoilage bacteria in a model system. E. coli, Listeria monocy-
4 ppm are approved by the EPA for potable water supplies togenes (LM), Salmonella serovars, and Staphylococcus aureus (8) and by the U.S. Food and Drug Administration for bot- were obtained from the Poultry Microbiological Safety Unit lab- oratory of the USDA-ARS. These pathogenic and indicator bac- Although the characteristics of MON have been well terial isolates were originally collected from commercial broiler known for many years, its use as a biocide has not been carcasses. Each isolate was assayed for Gram reaction, cyto- widely pursued beyond the potable water treatment arena chrome oxidase activity, and production of catalase and was iden- because it is simply not generally regarded as an efficacious tified with either the Vitek (bioMe´rieux Vitek, Inc., Hazelwood,Mo.), Biolog (Biolog, Inc., Hayward, Calif.), or Micro-ID (Or- water treatment. The reasons for this widely held view are ganon Teknika Corporation, Durham, N.C.) rapid identification twofold: MON is a slow-reacting biocide, and the specific lethality of MON is 200 times less than free chlorine (hy- Pseudomonas fluorescens and Shewanella putrefaciens spoil- pochlorous acid) in inactivating enteric bacteria (11). Al- age bacterial isolates were obtained by collecting broiler carcasses though there is no known research that addresses the use from processing plants in Georgia, Arkansas, California, and of MON to reduce microbial levels on food, the authors North Carolina. These carcasses were individually bagged in ster- hypothesize that, in systems in which long contact times ile polyethylene bags (3,000 ml O2 at 22.8 EC/m2/24 h at 1 atm) and high organic loads exist, such as in poultry processing and held on ice until arrival at the laboratory. Carcasses were plant immersion chillers, the increased efficacy and persis- allowed to spoil under controlled conditions at 3 Ϯ 0.5ЊC for 15 tence of MON make it a more effective disinfectant than days. After spoilage, the carcasses were rinsed with 100 ml ofsterile deionized water. The rinse fluid was diluted to 10Ϫ6, 10Ϫ7, and 10Ϫ8 with a sterile 1% solution of Bacto Peptone (Difco, The purpose of these studies was to compare the effi- Becton Dickinson, Sparks, Md.), and 1 ml of the diluent was cacy of SH with MON on populations of bacteria associated spread onto duplicate plate count agar (Difco, Becton Dickinson) with broiler chicken carcasses in a model system and to plates. Plates were incubated at 25ЊC for 48 h. Each isolate was compare the efficacy of SH to MON during immersion assayed for Gram reaction, cytochrome oxidase activity, and pro- chilling in a model system and in a commercial poultry duction of catalase and was identified by one of the same rapid identification methods used to identify the pathogenic and indi-cator bacterial isolates discussed previously. P. fluorescens and S. MATERIALS AND METHODS
putrefaciens isolates from these spoiled carcasses were obtainedand used in this study.
The experimental design for study 1 was 3 ϫ 3 ϫ 6 ϫ 5 Chiller water was collected from a commercial processing (replication, treatment, bacterium, and tube, respectively). The ex- facility and was autoclaved to eliminate background microflora.
perimental design for study 2 was 3 ϫ 4 ϫ 10 (replication, treat- The water was then compared with chiller water that had not been ment, and carcass, respectively). In study 3, data were collected autoclaved by adding chlorine to the water and measuring the during three phases: historical (Hist), prepilot (Pre), and pilot depletion from the reaction with organic material in the water.
(Test). The experimental design was 2 ϫ 127 or 75 (treatment and Both autoclaved and unautoclaved chiller water had the same carcass [Hist ϩ Pre and Test]) for Salmonella and 2 ϫ 3 ϫ 421, characteristics with regard to depletion of chlorine. Thus, auto- 39, or 110 (sample location, treatment, and carcass [Hist, Pre, and claved chiller water was deemed acceptable as a chiller water sub- Test]) for Escherichia coli (EC) in the front-half chiller and 2 ϫ stitute to provide background organic material.
3 ϫ 651, 60, or 216 (sample location, treatment, and carcass [Hist, To determine the effect of MON or SH on each isolate or on Pre, and Test]) for E. coli in the saddle chiller. Results were an- indicator populations of bacteria, E. coli, L. monocytogenes, Sal- alyzed by subjecting the data to t tests with SAS software (4) in monella serovars, and S. aureus were individually placed into studies 1 and 2. Treatment means were separated by Fisher’s least brain heart infusion broth (Difco, Becton Dickinson) at 35ЊC, and significant difference option (study 2) of SAS (4). For study 3, P. fluorescens and S. putrefaciens were individually placed into bacterial count data were transformed by log transformation, brain heart infusion broth at 25ЊC for 24 h. One 10-␮l loopful of NCFU ϭ ln(CFU ϩ 0.1). The purpose of the addition of 0.1 was each of these actively growing cultures was placed into 10 ml of to keep NCFU defined when CFU ϭ 0. This transformation leads sterile chiller water as controls or into sterile chiller water con- to more symmetric distributions and homogeneity of variance.
taining MON or SH at concentrations of 50 ppm, and the suspen- TABLE 1. The effect of water, sodium hypochlorite (50 ppm), and monochloramine (50 ppm) on pathogenic, indicator, and spoilagebacteria associated with chicken carcassesa a Means within a row with different letters are significantly different (P Յ 0.05). DT, time in hours required for bacterial growth to exceed the detection threshold of approximately 106; —, calibration curves have not been established for this bacterium.
b n ϭ 5 for each of three repetitions for each bacterium. Estimates were calculated on the basis of preestablished calibration curves for each bacterium, in which detection times were regressed against log CFU per milliliter.
sions were allowed to remain for 1 h to mimic commercial chill resistant Salmonella serovars or P. fluorescens were diluted, and 0.1 ml was placed onto the breast of each carcass and spread over After the exposure period, 1 ml of the suspension was placed the breast of each chicken with a sterile bent glass rod. The bac- into 9 ml of sterile brain heart infusion broth containing 0.16 g teria were allowed to attach for 15 min before treatment. Ten of sodium thiosulfate per liter and vortexed. One milliliter of this carcasses each were inoculated with Salmonella serovars or P. mixture was placed into a bactometer module well in duplicate.
fluorescens, and 10 were temperature abused to increase E. coli Samples were monitored with the bactometer microbial monitor- populations. The Salmonella serovars and E. coli groups of the ing system M128 (bioMe´rieux, Inc., Hazelwood, Mo.). The path- inoculated or temperature-abused carcasses were immersed sepa- ogens and E. coli were monitored at 35ЊC. Spoilage bacteria were rately in pilot-scale poultry chillers containing tap water, tap water monitored at 25ЊC. Impedance was measured on all samples for with 20 ppm SH, or tap water with 20 ppm MON. The P. fluo- rescens group of inoculated carcasses was immersed in pilot-scalepoultry chillers containing tap water, tap water with 50 ppm SH, Study 2: Effect of SH versus MON on E. coli, Salmonella,
or tap water with 50 ppm MON. The carcasses for all groups were and Pseudomonas on broiler carcasses during immersion chill-
exposed for 1 h at 5ЊC and sampled. Six additional carcasses per ing. Ninety-six broiler chicken carcasses were collected just after
replicate were inoculated—two with each type of bacteria—and evisceration and just before the rinse cabinets at a commercialprocessing facility. The carcasses were transported to The Uni- allowed to attach or temperature abused and tested immediately versity of Georgia Poultry Research Center pilot processing fa- to determine how many of the bacteria could be recovered from cility for inoculation, treatment, and testing.
inoculated or temperature-abused carcasses. Three replicate trials A marker strain of nalidixic acid–resistant Salmonella was were conducted for treated and control carcasses.
obtained from the USDA-ARS Poultry Microbiological Safety In studies 1 and 2, carcasses were sampled by rinsing ac- Unit laboratory. These isolates were originally collected from cording to the procedure described by Cox et al. (1), except that commercial broiler carcasses and were selected for resistance to 100 ml of Butterfield’s phosphate buffer was used instead of de- nalidixic acid. The Pseudomonas used in study 2 was collected, ionized water. For the evaluation of nalidixic acid–resistant Sal- identified, and cultured as described in study 1. For the E. coli monella serovars, 0.1 ml of carcass rinsate was placed onto the evaluation, broiler carcasses were subjected to temperature abuse surface of brilliant green sulfa agar (Difco, Becton Dickinson) to increase the populations of naturally occurring E. coli on their containing 200 ppm nalidixic acid. The plates were incubated at surfaces. These populations were enumerated for control and treat- 35ЊC for 24 h. Nalidixic acid–resistant Salmonella serovars were Actively multiplying (24-h-old) cultures of nalidixic acid– For the E. coli evaluation, 5 ml of carcass rinse was placed TABLE 2. The effect of tap water, sodium hypochlorite (50 ppm), and monochloramine (50 ppm) in a mock-scale immersion chilleron log Escherichia coli on broiler chicken carcassesa a Means within a row with different letters are significantly different (P Յ 0.05).
b n ϭ 10 for each of the three repetitions.
MON VERSUS SH AS POULTRY CARCASS DISINFECTANTS TABLE 3. The effect of tap water, sodium hypochlorite (50 ppm), make-up water during processing. The concentration of chlorine and monochloramine (50 ppm) in a mock-scale immersion chiller used was 50 ppm. During the prepilot (Pre) phase, SH was in- on nalidixic acid–resistant Salmonella serovar counts on broiler jected directly into the red water chiller return lines and controlled to total chlorine levels of between 10 and 20 ppm in the chillers.
During the pilot (Test) phase, MON was injected directly into the redwater chiller return lines and controlled to total chlorine levels between 10 and 20 ppm in the chillers.
Carcasses were sampled by rinsing with 400 ml of Butter- field’s phosphate buffer as required by the USDA-FSIS. E. coli was assayed with E. coli plate counts according to the Official Methods of Analysis of AOAC International method 990.12 (CFU). Salmonella was assayed with The Official Methods of Analysis of AOAC International method 2000.07 and reported aseither positive or negative.
a n ϭ 10 for each of the three repetitions.
RESULTS AND DISCUSSION
into 5 ml of sterile double-strength CM medium (bioMe´rieux) Study 1. With the exception of its performance against
supplemented with 2% dextrose, which acts as a selective growth S. aureus, MON equaled or outperformed SH in reducing medium, for E. coli conductance assays according to the proce- populations of pathogenic, indicator, and spoilage bacteria dure described by Russell (3), and the mixture was vortexed. One in chiller water. MON at 50 ppm reduced Salmonella to a milliliter of this mixture was placed into a bactometer module well level such that no growth in the bactometer was detected in duplicate. Samples were monitored with the bactometer micro- in any of the repetitions over a period of 24 h, which bial monitoring system M128. All of the bacterial isolates tested equates to the elimination of all viable organisms in the were monitored at 44ЊC. All samples were monitored for 48 h byconductance. E. coli conductance detection times were converted initial population (6.8 log CFU). The SH treatment elimi- (log CFU per milliliter) with a previously developed calibration nated all viable organisms in five of six samples evaluated.
curve. For Pseudomonas, 1 ml of carcass rinse was placed into 9 Both the SH and MON treatments at 50 ppm resulted in ml of sterile brain heart infusion broth and vortexed. One milliliter complete reductions of the initial populations of Listeria of this mixture was placed into a bactometer module well in du- (7.5 log CFU) and E. coli (6.9 log CFU) in all of the rep- plicate. Samples were monitored with the bactometer microbial etitions such that no growth occurred in the bactometer.
Neither SH nor MON significantly reduced initial S. aureus Three batch chillers were filled with 30 gal (114 liters) of tap water. Ice and 1 liter of fresh chicken blood were added to MON-treated Pseudomonas resulted in no growth in the chillers. No chemicals were added to the first chiller as a tap one of the repetitions and long detection times in the other water control. SH was added to the second chiller to produce a two repetitions, which equated to an average reduction of final concentration of 50 ppm chlorine. MON was added to thethird chiller to produce a final concentration of 50 ppm. Ten car- the initial 7.5-log CFU populations of 4.2 log CFU. The casses inoculated with Salmonella were added to each of the chill- SH treatment had no significant effect on Pseudomonas in ers and treated for 1 h. The carcasses were removed, and whole two repetitions and reduced the initial 7.5-log CFU popu- carcass rinses were conducted by the procedure described in Cox lations by an average of 1.2 log CFU. No calibration curves et al. (1), except that Butterfield’s phosphate buffer was used in- had been established for S. putrefaciens at the time of this experiment, so the recorded detection times could not beregressed to determine initial bacterial populations or re- Study 3: Effect of SH compared with MON on Salmonella
prevalence and E. coli counts on broiler carcasses in a com-
ductions (log CFU). However, detection times revealed a mercial processing facility. Data were collected during three
pattern similar to that observed with the other bacteria, with phases. The first phase was termed historical (Hist). During this the MON-treated S. putrefaciens producing longer average phase, SH was preseeded into the chillers and added to the fresh detection times and hence outperforming SH. The MON TABLE 4. The effect of tap water, sodium hypochlorite (50 ppm), and monochloramine (50 ppm) in a mock-scale immersion chilleron Pseudomonas fluorescens counts on broiler chicken carcassesa a Means within a row with different letters are significantly different (P Յ 0.05).
b n ϭ 10 for each of the three repetitions.
TABLE 5. Escherichia coli counts from postchill carcasses treated with sodium hypochlorite or monochloramine in a commercialpoultry processing planta a Means within a column with different letters are significantly different (P Յ 0.05) when CFU data were transformed by the log- transformation NCFU ϭ ln(CFU ϩ 0.1).
treatment resulted in no S. putrefaciens growth in one rep- tistical significance (P ϭ 0.1433). On the saddles, however, etition and significant reductions in initial populations in after converting the data to a log scale, the decrease in the other two repetitions. The SH treatment resulted in no standard deviation was highly statistically significant (P Ͻ S. putrefaciens growth in one repetition, a significant re- duction in initial populations in one repetition, and no sig- Salmonella prevalence was also lower with MON (3 nificant reduction in initial populations in the third repeti- of 75, 4.0%) compared with the historical data for SH (11 tion. Overall, MON was statistically equal or superior to of 127, 8.7%). Thus, MON appears to have a beneficial SH in reducing populations of pathogenic, indicator, and effect on reducing Salmonella prevalence over and above spoilage bacteria in the model chiller water system (Table that observed for SH. Again, this is hypothesized to be because most SH is bound by organic material in the chillerand rendered inactive.
Study 2. When used as an antimicrobial agent in im-
The data from these studies indicate that MON is an mersion chillers, MON outperformed SH in reducing the excellent alternative to SH for disinfecting chiller systems counts of E. coli, Salmonella serovars, and P. fluorescens in poultry processing plants. Variance in process control as on broiler carcasses. MON significantly reduced E. coli determined by oscillating bacterial prevalence and counts populations on broiler chicken carcasses, whereas SH had and resulting from increasing or decreasing organic loads no effect in this study (Table 2). The authors hypothesize can be mitigated with the use of MON because it is not that this result occurred because, within several minutes of affected nearly as much by organic material. Although an adding SH to the chiller water, the organic material in the extensive literature search did not produce references that water bound the disinfectant, rendering it inactive; with specifically address the efficacy of MON in a poultry chiller MON, only approximately 10 ppm was lost to organic bind- application, the conclusions reached in this study are ing, leaving 40 ppm available for disinfection.
strongly supported by the experience of a number of po- MON nominally reduced Salmonella populations by table water treatment plants that found that MON both re- 80%, whereas populations nominally increased by 34% on mained more persistent in distribution systems than free carcasses treated with SH (Table 3). MON likewise out- chlorine and maintained or improved the antimicrobial ef- performed SH in reducing Pseudomonas populations (Table ficacy achieved under a chlorination regime (2, 11). The naturally occurring organic material found in all potable It is believed that because of the high concentration of water distribution systems reacts with and therefore de- organic material often encountered in poultry chiller water, pletes residual free chlorine far more quickly than it can MON would be advantageous to use as opposed to sodium react with and deplete MON. This same principle explains or calcium hypochlorite. These differences would be em- the difference in efficacy between SH and MON treatments phasized in industrial situations because of the higher level experienced in these studies and supports the conclusion of organic material in industrial chillers compared with the that the controlled use of MON is a safe and more effica- level used in this study (1,500 versus 90 ppm biological cious approach to disinfection during chilling.
oxygen demand as tested, respectively).
REFERENCES
Study 3. E. coli counts and Salmonella prevalence data
were collected over a period of 84 days (historical data for Cox, N. A., J. E. Thomson, and J. S. Bailey. 1981. Sampling of SH), 9 days (prepilot data for SH), and 27 days (pilot for broiler carcasses for Salmonella with low volume water rinse. MON). E. coli counts were significantly reduced (P Ͻ Geldenhuys, J. 1995. Chloramination to preserve microbiological 0.0001) on both the back halves (saddles) and the front halves (fronts) of carcasses when using MON compared Russell, S. M. 2000. Comparison of the traditional three-tube MPN with historical and prepilot data obtained when SH was method with the PetrifilmJ, SimPlateJ, BioSys optical, and Bacto- used as the chiller antimicrobial. In addition, the variability meter conductance methods for enumerating Escherichia coli fromchicken and ground beef. of E. coli counts showed a marked reduction with MON SAS Institute. 1994. SAS/STAT7 guide for personal computers. Ver- compared with SH. Although the decrease in standard de- sion 7 edition. Statistical Analysis Systems Institute, Inc., Cary, N.C.
viation on the fronts was substantial, it did not achieve sta- Tsai, L.-S., B. T. Molyneux, and J. E. Schade. 1992. Chlorination of MON VERSUS SH AS POULTRY CARCASS DISINFECTANTS poultry chiller water: chlorine demand and disinfection efficiency.
2000. Sanitation performance standards Directive 11,000.1. U.S. De- Poultry Sci. 71:188,194–195.
partment of Agriculture, Food Safety and Inspection Sevice, Wash- U.S. Code of Federal Regulations. 2003. Poultry products inspection U.S. Environmental Protection Agency. 1999. Microbial and disin- U.S. Code of Federal Regulations. 2003. Bottled water. 21 CFR fection byproduct rules simultaneous compliance guidance manual, p. 2–11. EPA 815-R-99-015. U.S. Environmental Protection Agency, U.S. Code of Federal Regulations. 2001. Maximum residual disin- White, G. C. 1992. The handbook of chlorination and alternate dis- U.S. Department of Agriculture, Food Safety and Inspection Service.
infectants, 3rd ed. Van Nostrand Reinhold, New York.

Source: http://www.zentox.com/news-articles/PathX-JFP.pdf