United States
Environmental Protection    Office of Water (WH-550)    EPA 814-R-92-007
Agency              Washington DC 20460      August 1992
Simulation  of Microbial Occurrence,
Exposure, and Health Risks After
Drinking Water Treatment Processes

-------
THIS PAGE INTENTIONALL Y BLANK

-------
SIMULATION OF MICROBIAL OCCURRENCE, EXPOSURE AND HEALTH RISKS AFTER DRINKING
                          WATER TREATMENT  PROCESSES
             WILLIAM D.  GRUBBS,  BRUCE A.  MACLER*  and STIG  REGLI*


               Science Applications International Corporation

                                     and

                    *U.S.  Environmental Protection Agency

-------
                                   ABSTRACT
      For  the development of  the Disinfectant/Disinfection Byproduct Rule,  EPA
wishes to  compare human health risks from microbial  infection with  those  from
chemical disinfectants and  their byproducts.  A direct comparison using
available  data  is not possible at this  time.  Therefore, EPA is approaching
this problem with the use of  computer models that simulate occurrence  levels
of pathogenic organisms in  raw water, then simulate  disinfection and
production of disinfection  byproducts.  The microbial and chemical
concentrations  thus generated are then  used to estimate potential health
risks.  This paper presents the methodology used for these simulations and
estimations and discusses the assumptions and uncertainties inherent to this
modeling process.

      Two  distinct sources  of variation were examined in this analysis.
Summary measurements of existing data for Giardia occurrence from different
cities reflected a geographic variation.  Measurements from the same city but
on different days reflected a temporal  variation.  These variations were
characterized from data collected by Hibler (1988) and LeChevallier, et al
(1991).  The lognormal distribution was used to describe the geographic
variation, and  a combination  of two discrete distributions, the delta and the
negative binomial distributions, was used to describe the temporal variation.

      Annual averages of Giardia occurrence in raw surface water for 100
cities were simulated based on the geographic variation.  These averages in
raw surface water were the  basis for input to a simulation model, where
treatment  was applied as a  function of  the raw surface water quality.  This
simulation model used engineering and chemical equations to predict Giardia in
finished surface water.

      Giardia occurrence in finished surface water also exhibits a geographic
and temporal variation.  These types of variation were employed in conjunction
with a dose-response function that related the probability of infection to  the
number of  Giardia cysts in  finished surface water.  Quantities related to this
function were used to estimate endemic  levels and the frequency of an
outbreak.  Additional refinements to this analysis were performed to examine
the effects of  secondary infection and  system malfunctions on the results
based on Giardia occurrence in finished surface water.
                                 INTRODUCTION
      EPA is developing National Primary Drinking Water Standards for various
chemical disinfectants and their disinfection byproducts.   The goals of this
Disinfectant/Disinfection Byproducts (D/DBP) Rule are to ensure that drinking
water is microbiologically safe at any limits set for disinfectants and
byproducts,  and that the disinfectants and byproducts themselves do not pose
unacceptable risks at these limits.  EPA's approach in developing this rule is

-------
to consider different regulatory scenarios that achieve different definitions
of microbial safety and risk levels from disinfectants and byproducts.  These
risks are linked, in that any increase in disinfection to lower microbial
risks requires that use of more disinfectants and consequently yields higher
levels of byproducts, thus increasing chemical health risks.  Determination of
the magnitude of microbial and disinfectant/byproduct risks as a function of
different water treatment trains and source water qualities is essential to
crafting a rule that will minimize overall health risks from drinking water.

      The comparison of microbial health risks with those generated from
drinking water treatment for given treatments is not directly possible using
currently available data.  As a result, EPA is approaching this problem with
the use of computer models that simulate the occurrence levels of pathogenic
organisms in raw water, then simulate disinfection and production of certain
disinfection byproducts of health concern.  The microbial and chemical
concentrations generated for this "treated" water are then used to estimate
potential health risks.

      This paper presents the methodology used for these simulations and
estimations and discusses the assumptions and uncertainties inherent to this
modeling process.  Giardia lamblia was selected as the target organism for the
modeling effort since a) the existing data base for its occurrence is the most
extensive of any pathogenic microorganism found in drinking water; b) CT
values have been developed for predicting disinfection inactivation
efficiencies; c) it is much more resistent to disinfection than most other
waterborne pathogens and therefore changes in disinfection practice are more
likely to affect Giardia exposures than those for most other pathogens; and d)
dose-response data are available for Giardia for estimating risk from
exposure.
                             METHODOLOGY AND DATA
      A flowchart describing the process of estimating endemic levels and
outbreak frequency is presented in Figure 1.  Details of the methodology and
assumptions follow.


Geographic variation of Giardia occurrence in raw surface water data

      Data collected by LeChevallier, et al (1991) were used to characterize
the geographic variation of Giardia cyst concentrations in source waters for
different cities.  These data were used to represent the annual average cyst
concentrations for different cities.  These data are not appropriate for
assessing the temporal variation described below (i.e., the changes in the
number of Giardia cysts over some time period).   The listing of the 85
measurements in this data base is given in Appendix A.   The 15 measurements
with a '*' under the "OBSERVATION WAS DELETED" column were not included in the
analysis because these observations were estimated Giardia levels based on the
detection limit, rather than actual observations.   Each measurement was

-------
                                       Overview ofGiardia Modelling
     Hibler Data
      73 plants
  variation between
  plants: lognormal
                LeChevalier Data
                   46 plants


                variation between
                plants: lognormal
         t
  temporal variation
     within plants:
delta negative binomial
                      i
                distribution used
              for influent simulation
                        Assuming:

                        •  48% recovery
                        •  13% viability
                        •  deletion of
                          values estimated at
                          limit of detection
                                                  100
                                            simulated influents
                                                   1
                                            Treatment Model
                                                  T
                                                  100
                                       plant effluents (1st customer)
                                            summary statistics
                                  f
                                mean
                                     90th%
 distribution used
   for simulation
  of within plant
 temporal variation
                                 I
compute expected
value of Rose dose
response function
                                      I
compute expected value
 and variance of Rose
dose response function
                               estimate
                              of endemic
                              incidence
                                 rate
                                   estimate
                                  of outbreak
                                      risk
Assuming:

• 5% failure rate
  (loss of 1 log)
• 25% secondary
  infection rate
• 1% per 30 days
  outbreak threshold
• viable cysts
  infective in humans
• infection=illness

-------
multiplied by 2.08 to reflect a retrieval efficiency (i.e., the measurement
technique was not able to observe the true number of cysts present) and by
0.13 as a viability factor; that is, cysts that are counted, but because of
morphological characteristics are not considered actually alive.  Cities with
multiple measurements were averaged to obtain one measurement for each city.
These 46 city averages are also presented in Appendix A, along with the number
of annual averages from which these city averages were constructed.

      Data collected by Hibler (1988) also provide information on the
geographic variation, if, for example, the arithmetic average of the daily
measurements within a city is taken to construct a city average.  However, it
was determined that the LeChevallier data were more appropriate for
characterizing the geographic variation because the methods used in the
collection of the LeChevallier data were more advanced, and this data
encompassed a broader geographic region.
Distributional assumptions

      The 46 city averages based on the LeChevallier, et al (1991) data were
tested for normality and lognormality,  assuming independence among cities.
The Shapiro-Wilk (Shapiro and Wilk, 1965) test, as calculated in the SAS*
procedure PROC UNIVARIATE (SAS, 1990) was chosen to test the hypotheses of
normality and lognormality.   Lognormality was examined by testing the natural
logarithm of the 46 city averages for normality.  The results of this analysis
are shown in Appendix A.  The null hypothesis of normality of the city
averages was soundly rejected (p<0.001), but the null hypothesis of
lognormality could not be rejected (p=0.582).  Consequently the lognormal
distribution was used to characterize the geographic variation, and a
lognormal distribution was used to generate the input into the simulation
model, which predicted Giardia in finished surface water and is discussed in
further detail below.

      Although the Hibler data was not used in characterizing geographic
variation, an inspection of the 73 city averages constructed from this data
also support the hypothesis of lognormality of the city averages.


Temporal variation of Giardia occurrence in raw surface water data

      Data collected by Hibler (1988) were used to characterize the temporal
variation of Giardia cyst occurrence in raw surface water.  A summary of 1,515
measurements across 73 cities is given in Appendix B.  The number of cysts was
also multiplied by 2.0 to reflect a retrieval efficiency, as discussed by
Hibler (1988), and rounded to the nearest integer, since the nature of the
measurement (number of Giardia cysts) is inherently discrete.   No information
was available on the exact date of sampling to assess any correlation among
measurements across time.  The distributional methodology was consequently
developed based on an assumption of independence among daily measurements.
Cities were also assumed to be independent.  Selected summary statistics on
the Hibler data are provided in Appendix B.  An important statistic from this
data, which was crucial in the distributional assumption for characterizing

-------
the temporal variation of Giardia cyst occurrence, involves  the  large number
of zero Giardia cyst measurements.  Approximately 72 percent (1,087/1,515)  of
all Giardia cyst measurements in this data base are 0.
Distributional assumptions

      Typical discrete distributions for modeling the number of Giardia cysts
include the Poisson distribution and the negative binomial distribution.   If
the random variable X represents the number of Giardia cysts in raw  surface
water, then under the Poisson distribution,
Under the negative binomial distribution,
            Pr U=x) =       ._..__,      q>Q, m>0,  x=o,l,2.
                          !  \g+/n/  \ q+m)
where r(-) is the gamma function described in mathematical tables handbooks,
such as the CRC handbook  (Beyer, 1968) and x!-x-(x-1)•(x-2)•...1.  The A.
parameter for the Poisson distribution and the parameters m and q for the
negative binomial distribution are usually estimated from available data.

      A characteristic of the Poisson distribution is that the mean and
variance are equal.  The variance is larger than the mean for  the negative
binomial distribution.  In particular, if E(X) represents the mean of a
distribution of a random variable X, and V(X) represents the variance of X,
then for the Poisson distribution,

                            E(X)  = A. and  V(X)  =  X.

For the negative binomial distribution,

                         E(X) = m and V(X)  = m+(m2/q) .

      In analyzing the data, temporal distributions of Giardia cyst occurrence
in raw surface water were developed separately on a city-by-city basis.  It
was clear that neither the Poisson nor the negative binomial distribution in
their present form was able  to adequately represent the data, because of the
large number of zeroes.  A value of zero is permissible in both the Poisson
and negative binomial distributions, but to choose parameters that will
adequately model the high percentage of zeroes in this data would cause the
probability of larger values occurring to be extremely small.  Also an
inspection of the data reveals a larger variance than mean for many of the

-------
cities.  This aspect of the data favors the choice of the more flexible
negative binomial distribution over the Foisson distribution, which has an
equal mean and variance .

      Consequently, a modification of the negative binomial distribution was
used, whereby a point distribution placed at zero was combined with the
negative binomial distribution.  The point distribution is often referred to
as the delta distribution, and the mixture of these two distributions will
subsequently be referred to as the delta-negative binomial (DNB) distribution.
The form of the DNB distribution, where the random variable X represents the
number of Giardia cysts in raw surface water, is
                                                                   x=0,l,2,...
where I0(x)=l if x=0 and 0 otherwise.

The mean, E(X),  and the variance, V(X),  of the DNB distribution are

      E(X) = (1-5)-m, and
      V(X) = (l-5)-m-[(6-m)+l+(m/q)]

The derivations of these quantities are given in Appendix C.

      Three parameters  (6, m, and q) need to be estimated for a DNB
distribution.  Parameter estimates for 42 cities in the Hibler data base are
presented in Appendix D.  All cities in this appendix had at least four
measurements; the other cities not included in this appendix had less than
four measurements.  The cities are listed in descending order by sample size.
These parameter estimates for each city were derived through the SAS*
procedure PROC NLIN.  The starting values (5S,  ms,  and  qs) used were

«s - n0/n
ms - meanpos
qs = meanpos2/(varpos-meanpos)  if varpos>meanpos,
   -        oo                  if varpos
-------
Relating Giardia occurrence in raw surface water to finished water

      A lognormal distribution based on the LeChevallier, et al (1991) data
was the basis for input into a simulation model that predicted Giardia in
finished surface water.  Based on the 46 city averages described previously, a
logmean of 5.39 and a logstandard deviation of 1.67 were used as input
parameters (see Appendix A for the derivation of these quantities) to the
model, and 100 city averages were generated with the SAS* function RANNOR.
The simulation model utilized engineering and chemical equations to apply
treatment as a function of raw water quality and generated 100 city averages
for Giardia in finished water as output. (See Gelderloos, et al (1992) and
Cromwell et al (1992) for a complete description of this simulation model.)
In general, this model simulated conventional water treatment of surface water
without lime softening (i.e., coagulation, sedimentation, filtration and
chemical disinfection), capable of meeting Surface Water Treatment Rule (SWTR)
requirements, with and without the use of an alternative disinfectant to
chlorine.  In addition, an "enhanced" SWTR level of treatment was simulated,
using EPA SWTR guidance for poorer quality waters, to consider the potential
value of increasing the level of disinfection as a function of poorer source
water quality.  Treatment performance under the enhanced SWTR was estimated by
assuming that filtration achieved 2.5 log removal of Giardia cysts and that
disinfection achieved additional log inactivation of cysts predicted by CT
equations.  The enhanced SWTR specified a level of treatment necessary to
achieve an approximately constant average Giardia cyst concentration at the
first customer (Gelderloos, et al, (1992)).

      Assuming that the removal efficiency for Giardia cysts between raw and
finished surface water is binomial and the distribution of Giardia cysts in
raw water is a DNB distribution, then the distribution of Giardia cysts in
finished surface water is also a DNB distribution.  The assumption of a
binomial removal process assumes a constant probability p of survival of a
Giardia cyst between raw and finished surface water.   The mathematical proof
behind this statement is included in Appendix F.   The log removal is directly
related to this probability p; in fact, log removal = -Iog10(p).  As seen by
this proof, if the number of Giardia cysts in raw water exhibits a DNB
distribution with parameters 6, m, and q, then the number of Giardia cysts in
finished water has a DNB distribution with parameters 6,  m«p,  and q.   In
particular, if the random variable Y represents the number of Giardia cysts in
finished surface water, then the mean, E(Y),  and the variance,  V(Y),  are

      E(Y) - (l-6)-m-p                          (1)
  and V(Y) = (1-6)-m-p-[(6-m-p)+l+((m-p)/q)]

      A DNB distribution was consequently assumed for each of the 100 cities,
using values for 6 and q as derived from the Hibler (1988) data,  along with
the annual averages as output from the simulation model.   A value for 6 of 0.7
was chosen, very similar to the approximately 72% of zero measurements found
in the Hibler data.   A value for q of 4.5 was selected to represent a typical
estimate for q from the cities in Hibler's data.   Since E(Y)  = (1-6)-m-p from

-------
formula (1), m*p was estimated as the annual average divided by (1-5).  A DNB
distribution with these values of 5, m-p, and q were consequently used, along
with the dose-response relationship described below, to estimate endemic
levels and outbreak frequency.

      In subsequent results, certain summary statistics from the 100 cities
were used to characterize the distribution.  In particular, for estimating
endemic levels,  results were based on the arithmetic average of the number of
cysts in finished water across the 100 cities.  For estimating outbreak
frequency under nominal SWTR conditions, results were based on the number of
cysts at the city closest to the 90th percentile of the 100 cities.  Under an
enhanced SWTR, no outbreaks should occur by definition, since the level of
treatment should ensure that the infection rate would be well below the
assumed outbreak threshold of greater than one percent of the population
becoming infected within a one-month period.  A small number of these 100
cities were not included in the analysis because the requirements of the SWTR
and taste and odor constraints could not be met (see Gelderloos, et al (1992)
for a further description of the deletion of cities from the Giardia
analysis).


Dose-response relationship

      Using a risk assessment model from Rose, et al (1991) based on human
infectivity studies (Rendtorff, 1954; Rendtorff and Holt,  1954), a dose-
response relationship was developed to estimate the risk of infection due to
waterborne exposure to Giardia.  In the Rendtorff studies,  a total of 40
volunteers were fed Giardia cysts in capsules, and a positive response was
measured by cyst excretion in the feces.  Infection was the measure of a
positive response and not illness.  While no infection resulted in illness in
the Rendtorff study, which used healthy male prisoners, we  used the
conservative assumption for the outbreak analysis that all  infections result
in illness.  This appears reasonable, since a substantial number of infected
individuals do become ill,  as indicated by over 100 reported waterborne
outbreaks of giardiasis in the U.S. since 1965.  Also, the  paper by Regli, et
al (1991),  which compared predicted infection rates with actual illness rates
in communities with waterborne outbreaks of giardiasis, appears to support
this assumption as being correct within one order of magnitude.

      An exponential dose-response function was used to relate the probability
of infection to the number of Giardia cysts ingested.   If the random variable
D represents the number of Giardia cysts ingested and the random variable PD
represents the probability of an infection, then the exponential dose-response
function can be related to the probability of infection as

                                PD=l-exp(-r-D),

where r is a parameter estimated from the data.  In this case, using data from
the Rose, et al (1991), a value for r of 0.02 was derived.   The data from
Rose, et al, are presented in Appendix G, along with the estimation of the
parameter r according to the exponential dose-response function using SAS*
PROC NLIN.

-------
Estimating endemic Levels

      The dose-response function and the DNB distribution were combined to
estimate the endemic level of infection in the population.  The endemic level
measures the frequency of occurrence of disease in a population that exists on
an ongoing basis.  In particular for this analysis, the estimate of endemic
levels was expressed as the expected number of infections per 10,000 per year.
A "case" was defined as an infection episode; that is, one person can have
more than one infection per year.

      A key assumption of this analysis regards the length of infection.  It
is assumed that once a person is infected, that person remains infected for a
30-day period.  That is, if a person were infected on day 1 then he would
remain infected until day 30, and not be infected on day 31.  This assumption
is aimed at eliminating implausible co-occurring infections resulting from the
probabilistic approach taken here.

      The algorithm for estimating the average number of cases per 10,000 per
year is described below:

      1)  The units of the Hibler (1988) and LeChevallier, et al (1991) data
are in cysts per 100 gallons, and the parameters developed from the Hibler
data for use in this analysis are based on the number of Giardia cysts per 100
gallons.  Consequently an outcome chosen from a finished surface water DNB
distribution will be expressed in cysts per 100 gallons.  Convert the dosage
from 100 gallons to 2 liters and calculate the probability of an infection;
that is,

                          PD*=l-exp(-r-D-0.005283)
                             -l-exp(-r*-D),                  (2)

where 2 liters-0.005283-100 gallons and r *~r-0.005283.

      2) Determine the average probability of an infection, which is the
expected value of the dose-response function in (2),  assuming D is an outcome
from the DNB distribution with parameters 6, m-p,  and q.  This expected value
is
                            ff+(ffl«p)«(l-e-* )
The derivation of this quantity is found in Appendix H.

-------
      3) Determine the probability of a person being infected for z days out
of a year, where z can range between 0 and 365.  The probability of being
infected on any given day, given that a person is not infected, is assumed to
be equal to the average probability of infection in equation (3).  The general
formula for a person being infected z days out of a year is
                                                             , ...365
where t = largest integer less than or equal to (i/30),
      exdr = average probability of infection given in (3), and
      v = smallest integer greater than or equal to (i/30).

The values t and v were accomplished using the SAS* functions FLOOR and CEIL,
respectively.  Specific examples of how this formula was derived are in
Appendix I.

      4) Translate the number of days being infected in a year to the number
of cases.  If a person is infected for 0 days in a year, that person has 0
cases for the year.  If a person is infected between 1 and 30 days in a year,
then that person has 1 case for the year.  Two cases in a year are for 31 to
60 days infected, and so on, up to 13 cases for between 360 and 365 days in a
year.

      5) Calculate the expected number of cases, which is the sum from 0 to
365 of the number of cases corresponding to day z times the probability of
being infected z days.  Multiply the result by 10,000.

      As additional modifications to this analysis, two extra issues were
investigated.  In particular, the influence of a secondary infection rate and
a reduction in system effectiveness were studied.

      A secondary infection rate assumes that a person being infected can pass
giardiasis to another person a certain percentage of the time.  Various
alternatives can be studied; in the results presented subsequently, a
secondary infection rate of 25% was assumed.  Using the principles of the
geometric series

                       1 +  a + a2 + a3 + ...  -  l/(l-a)

a 25% secondary infection rate translates to an increase in the endemic rate
by 33%.  The results for estimating endemic rates as described above were
adjusted accordingly to reflect a 25% secondary infection rate.

      The consequences of a reduction in system effectiveness were also
studied.  In particular,  subsequent results assumed that 5% of the time the
removal for a given city was 1 log (factor of 10) less than the nominal
removal.  The nominal removal was calculated from the annual average for the
raw and finished water at a given city.  For the purposes of this analysis,

-------
time is treated as a continuum; that is, the system had a 1 log reduction in
effectiveness 5Z of the time scattered over a period (the reduction is not
clustered on any particular day, for example).  The nominal removal is
calculated from the annual average for raw and finished water at a given city.

In particular, let

RAW   = average number of cysts in raw surface water at a particular city,
FIN 05 = average number of cysts in finished surface water at a particular
        city assuming a reduction in system effectiveness 5% of the time,
FINR  - average number of cysts in finished surface water at a particular city
        assuming no reduction in system effectiveness for a given (R) log
        removal from treatment, and
FINR.1 - average number of cysts in finished surface water at a particular
        city assuming a reduction in system effectiveness of 1 log.
Then
                         FIN os = 0.9
                                        0.05-10*1 RAW
                                  10*)          \ 10*
                                          o'
      Consequently, a 5% reduction in system effectiveness corresponds to a
45% increase in the average number of cysts in finished surface water at a
given city.  The results for estimating endemic rates as described above were
adjusted accordingly to reflect this 5% reduction in system effectiveness.


Estimating outbreak frequency

      The dose-response function and the DNB distribution were also combined
to estimate the frequency of an outbreak in the population.  For the purposes
of this analysis, an outbreak has been defined as observing giardiasis
infection in greater than one percent of the population within any given 30-
day period.  This assumption is based on observations that the awareness of
waterborne disease often does not occur unless at least one percent of the
population becomes ill within a time frame of about a month (Regli, et al,
1991).  The population considered in the analysis here was defined as first
customers closest to the treatment plant with consequently minimal

                                      10

-------
distribution system CT disinfection.  The assumption that outbreaks are
identified by a one percent infection rate at the first customer versus one
percent for the entire population is somewhat arbitrary and may be overly
conservative, depending upon the relative population density near the first
customer.

      For this analysis, the estimate of the frequency of an outbreak is
related to the probability that an outbreak occurs.  The results were based on
the simulated number of cysts closest to the upper 90th percentile of the
modeled distribution.

      The algorithm for estimating the outbreak frequency is described below:

      1)  Determine the average probability of an infection and the variance
in the probability of an infection.  The average probability of an infection
is the expected value of the dose-response function in equation (2), and as
given in equation (3), is
                        exdi= (l-») ll-f	2	  I*!
                                  (  [g+(m«p)»(l-e-r ) J J
The variance in the probability of an infection is the variance of the dose-
response function in equation (2),  and is
vajrdr=(l-&)'
» 1 - -(
[ [ U«
\* /
1 - fi-aii
g r
» i
                                                                (4)
The derivations of these quantities are found in Appendix H.

      2) Let I denote the event of being infected on a given day, and I30 the
event of being infected in any 30-day period.  Being infected in any 30-day
period can be considered as the sum of 30 consecutive events of being infected
on a given day.  Expressed mathematically, the probability that an outbreak
occurs is expressed as

                                Pr(I30 > 0.01),

Calculate the average probability of an infection in a 30-day period - exdr30
- 30-exdr.  Calculate the variance in the probability of an infection in a 30-
day period = vardr30 =  30-vardr.   The  formulas  for exdr and  vardr are given  in
equations (3) and (4),  respectively.
                                      11

-------
      3) The dose-response function in equation (2) is an exponential function
of X and X has a DNB distribution.  Consequently,  the dose-response function
will not have a normal distribution.  But by the Central Limit Theorem, it is
assumed that the mean (or sum, in this analysis) of 30 samples (I30)  of  the
dose-response function is normally distributed with mean exdr30 and variance
vardr30.  Consequently, using  the  Central  Limit  Theorem,


                                 Pr(I30>0.01) =

                            ( T..-
                          Pr
                             I3a - exdrto  0.01-exdr30
where »(•) represents the cumulative distribution function of the standard
normal distribution.  This was performed using the SAS* function PROBNORM.

      4) The quantity in equation  (5) is an estimate of the probability of an
outbreak in any 30-day period.  To estimate the expected number of outbreaks
in a year, multiply this probability by 365.  The number 365 is used to
estimate a years' worth of 30-periods (for example, days 1-30, days 2-31, days
3-32, and so on).  To estimate the expected number of years to an outbreak,
take the reciprocal of the expected number of outbreaks in a year.

      The impacts of a 25% secondary infection rate and a 5% reduction in
system effectiveness were also incorporated into the estimates of outbreak
frequency.  The threshold for an outbreak was changed from 1% to 0.75% to
account for a 25% secondary infection rate and the average number of cysts in
finished water was increased by 45% to reflect a 5% reduction in system
effectiveness.  The subsequent results for estimating outbreak frequency
reflect the 25% secondary infection rate and 5% reduction in system
effectiveness.
                            RESULTS AND DISCUSSION


      Table 1 presents estimates of endemic levels and outbreak frequency of
microbial infection using the methodology described above, with respect to the
simulated systems' ability to attain given levels of haloacetic acids and
total trihalomethanes.  A more detailed set of results is presented in
Appendix J.  Results are presented for systems complying with SWTR and
enhanced SWTR disinfection levels and consider the use of alternative
disinfectants to chlorine.  Endemic levels are expressed in number of cases
per 10,000 per year, and outbreak frequency is expressed as the average number
of years between outbreaks.

      The results indicate that systems only minimally meeting SWTR standards

                                      12

-------
(3-log removal and inactivation of Giardia. 4-log removal and inactivation of
viruses, and maintenance of a disinfectant residual in the distribution
system) could produce water yielding significant endemic levels of microbial
illness under different potential TTHM or HAA drinking water standards.
Additionally, efforts to lower HAA or TTHM levels by reducing chlorine
disinfection could not only increase endemic illnesses, but might lead to
frequent outbreaks of illness in the community.  The implications for systems
in the upper percentiles of the distribution, which likely represents systems
with poorer quality source waters, are particularly worrisome.  The data
indicate a precipitous decrease in the time between outbreaks as the
distributions moves from the mean into the upper percentiles.

      It is important to note that these results represent what might occur if
all systems were only to minimally meet the SWTR requirements and treatment
constraints as described by Gelderloos, et al (1992).  Since many systems now
provide higher levels of inactivation from the minimums used for this modeling
analysis, the predicted infection rates and outbreak frequency rates may
significantly overestimate what actually might occur, especially under the
current TTHM MCL of 100 ug/1 and corresponding high MCL target of 50-60 ug/1
for HAAs.

      However, as the DBF MCL targets decrease and it becomes more difficult
for more systems to meet such a target, there should be greater likelihood for
systems to only minimally meet the SWTR requirements.  Therefore, greater
significance should be given to the predicted infection and outbreak incidence
at the lower DBF MCLs and to the difference between the predicted values under
high versus low DBF MCL targets.

      On the other hand, increasing the level of disinfection for poorer
source waters (i.e., in accordance with EPA guidance to the SWTR) to an
"enhanced" SWTR could reduce endemic illness and outbreaks to de minimus
levels.  The data from modeling show typically 1,000-fold lower numbers of
infections per year.

      An important question for any computer modeling effort is whether the
simulation comes close to matching reality as well as we can estimate it.  The
statistical models used to describe the observed occurrence data of Hibler
(1988) and LeChevallier, et al (1991) were seen to be poorest in their fit
towards the high occurrence end of the distributions.  This is critical,  since
the data from the model suggest that the majority of outbreaks occur from
systems described by this part of the distribution.   However,  we believe that
any overestimation of outbreaks from the model can be compensated for in the
interpretation of the data.

      The data from these simulations indicate a 2-5% annual risk of Giardia
infection to the first customer drinking nominal SWTR-treated water.   We
expect that the increased disinfection CT farther in the distribution system
should yield a much lower infection rate over the entire population served.
For the population at large benefiting from this additional CT,  we estimate
that this infection rate would probably be reduced by an order of magnitude,
i.e.,  the average infection rate would range from 0.2-0.5% per year for the
population as a whole, if systems have only to meet  the SWTR requirements.

                                      13

-------
For the 103 million people in the U.S. represented by our model simulation,
this translates to an endemic level of about 200,000-500,000 infections from
Giardia each year.  The treatment system modeled here was chosen by
Gelderloos, et al (1992) to represent conventional treatment systems using
surface water most likely to produce drinking water capable of meeting SWTR
standards.  Treatment plants for the remaining 60 million people in the U.S.
served by surface water are not believed to be as effective.  We believe that
these systems are likely to yield an additional 200,000-500,000 infections
from Giardia each year, for a total U.S. infection rate of about 400,000-1
million per year.  (Systems using groundwater exclusively as their source
water contribute few, if any, Giardia infections.)  If our assumptions on cyst
viability and illness/infection rates are valid (see also Regli, et al, 1991),
then perhaps 10% of these predicted infections will result in illness, or
about 40,000-100,000 cases per year.  Data from the Centers for Disease
Control (Bennett, et al, 1987) indicate that Giardia contributes about 70,000
cases of illness in the U.S. each year, in very good agreement with our
estimates.

      The effectiveness of an enhanced SWTR disinfection is amplified by the
consideration of all waterborne microbial illness.  It is estimated that
940,000 cases of waterborne microbial illness occurred in 1985, resulting in
some 900 deaths  (Bennett, et al, 1987). (While Giardia is not considered to
contribute to microbially-related deaths, the overall death rate from
waterborne microbial illness is about 0.1%.)  Giardia was chosen for our
calculations in  part due to its resistance to disinfection, which is generally
greater than that for bacteria and viruses.  Yet modeling of enhanced SWTR
versus nominal SWTR treatment indicated an additional 3-log decrease in
giardiasis by employing the enhanced  SWTR.  This could reasonably be expected
to apply to disinfection of bacteria  and viruses as well, which could result
in substantial decreases in overall endemic microbial illness from drinking
water and reduce related deaths to a  de minimus level.  Even if outbreak
occurrence rates predicted from the model for minimal SWTR treatment are
overestimated, it is clear that increased disinfection to the enhanced SWTR
will eliminate treatment-system derived outbreaks.
                                  REFERENCES
Bennett JV, SD  Holmberg, MF  Rogers  and  SL Solomon  (1987).  Infectious and
papsitic diseases.  Am J Prev  Med 3:  102-114.  In  RW Amler and HB Dull  (eds),
Closing the gapt  the  burden of  unnecessary  illness.  Oxford University Press,
pp  102-114.

Beyer, WH, ed.  (1968).  CRC  Handbook  of Tables for Probability and Statistics.
The Chemical  Rubber Company, Cleveland.

Cromwell, JE, X Zhang,  FJ  Letkiewicz, S Regli and  BA Macler (1992), Analysis
of  Potential  Trade-offs in Regulation of Disinfection By-Products, U.S.
Environmental Protection Agency.

                                      14

-------
Gelderloos, AB, et al (1992). Simulation of Compliance Choices for the
Disinfection By-Products Regulatory Impact Analysis, Seminar on Disinfection
and Disinfection By-Product Control, 1992 Annual Conference of the American
water works Association, Vancouver, B.C.

Hibler CP (1988).  Analysis of municipal water samples for cysts of Giardia.
In; Wallis P and B Hammon (eds).  Advances in Giardia research.  Univ of
Calgary Press, ppp 237-245.

Jolley LBW (1961).  Summation of Series.  Dover Publications, Inc., New York.

LeChevallier MW, WD Norton and RG Lee (1991).  Occurrence of Giardia and
Cryptosporidium spp. in surface water supplies.  Appl Env Microb 57: 2610-
2616.

Regli S, JB Rose, CN Haas and CP Gerba (1991).  Modeling the risk from Giardia
and viruses in drinking water.  J Am Water Works Assoc 83:11: 76-84.

Rendtorff RC (1954).  The experimental transmission of human intestinal
protozoan parasites.  II.  Giardia lamblia cystes given in capsules.  Am J Hyg
59: 209-220.

Rendtorff RC and CJ Holt (1954).  The experimental transmission of human
intestinal protozoan parasites.  IV.  Attempts to transmit Endamoeba coli and
Giardia lamblia by water.  Am J Hyg 60:  327-328.

Rose JB, CN Haas and S Regli (1991).  Risk assessment and control of
waterborne giardiasis.  Am J Pub Health 81: 709-713.

Shapiro SS and MB Wilk (1965).  An analysis of variance test for normality
(complete samples).  Biometrika 52: 591-611.

SAS Procedures Guide, version 6. third edition (1990).  SAS Institute,  Inc.,
Gary, North Carolina.
                                      15

-------
TABLE 1.  SUMMARY OF ENDEMIC LEVELS AND OUTBREAK FREQUENCY FOR HALOACETIC ACID
AND TOTAL TRIHALOMETHANES REGULATORY ALTERNATIVES
        Disinfection  HAA
Average Number
 of Cases per
10.000 per Year
Mean Average
Number of Years
for an Outbreak
90th Xile Average
Number of Years
for an Outbreak
ii^i^i^^
SWTR
SWTR
SWTR
SWTR
SWTR
SWTR
SWTR
SWTR
SWTR
SWTR
SWTR
SWTR
ESWTR
ESWTR
ESWTR
ESWTR
ESWTR
ESWTR
ESWTR
ESWTR
ESWTR
ESWTR
ESWTR
ESWTR
With
With
With
With
With
With
Without
Without
Without
Without
Without
Without
With
With
With
With
With
With
Without
Without
Without
Without
Without
Without
60
50
40
30
20
10
60
50
40
30
20
10
60
50
40
30
20
10
60
50
40
30
20
10
210
230
280
320
400
560
240
250
270
320
390
490
0.26
0.26
0.28
0.31
0.31
0.31
0.25
0.26
0.27
0.31
0.32
0.32
CO
00
2.5 E13
9.3 E9
5.3 E4
4.0
00
00
CO
1.1 E10
1.3 E5
73
CO
00
CO
00
00
00
00
00
CO
CO
00
00
3,100
2.5
0.66
0.17
0.018
0.006
210
1.1
0.66
0.17
0.034
0.008
00
00
00
00
00
CO
00
CO
00
00
00
03
Assumptions and definitions:

        •predicted incidence at first customer
        •arithmetic average number of cysts across cities for average number
         of cases per 10,000 per year
        •90th percentile of distribution of number of cysts across cities for
         average number of years for an outbreak
        •25% secondary infection rate
        •1-log reduction in treatment performance 5% of the time
        •SWTR = surface water treatment rule
        •ESWTR = enhanced surface water treatment rule
        •HAA - haloacetic acid target MCL 
-------
TABLE 1. (continued)  SUMMARY OF ENDEMIC LEVELS AND OUTBREAK FREQUENCY FOR
HALOACETIC ACID AND TOTAL TRIHALOMETHANES REGULATORY ALTERNATIVES

Alternative
Rule Disinfection
SWTR With
SWTR With
SWTR With
SWTR With
SWTR Without
SWTR Without
SWTR Without
SWTR Without
ESWTR With
ESWTR With
ESWTR With
ESWTR With
ESWTR Without
ESWTR Without
ESWTR Without
ESWTR Without


TTHM
100
75
50
25
100
75
50
25
100
75
50
25
100
75
50
25
Average Number
of Cases per
10.000 oer Year
330
380
460
500
340
370
460
500
0.26
0.29
0.30
0.32
0.25
0.28
0.30
0.32
Mean Average
Number of Years
for an Outbreak
5.1 E8
5.7 E5
440
53
1.9 E8
1.0 E6
380
58
00
03
CD
00
CO
CO
CD
CD
90th Xile Average
Number of Years
for an Outbreak
3.0
0.18
0.035
0.008
1.1
0.18
0.035
0.008
GO
CO
CO
CO
CD
CO
CO
CO
Assumptions and definitions:

      •predicted incidence at first customer
      •arithmetic average number of cysts across cities for average number of
       cases per 10,000 per year
      •90th percentile of distribution of number of cysts across cities for
       average number of years for an outbreak
      •25% secondary infection rate
      •1-log reduction in treatment performance 52 of the time
      •SWTR = surface water treatment rule
      •ESWTR - enhanced surface water treatment rule
      •TTHM - total trihalomethane level (pg/1)
                                      17

-------
APPENDIX A

-------
    LISTING OF RAW WATER DATA FROM LE CHEVALLIER USED TO CHARACTERIZE
             GEOGRAPHIC VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.08 TO REFLECT RETRIEVAL EFFICIENCY
   MEASUREMENTS HAVE ALSO BEEN MULTIPLIED BY 0.13 AS A VIABILITY FACTOR


CITY
1
2
101
102
109
109
109
302
305
306
307
307
307
307
307
307
307
307
310
310
311
312
312
312
314
315
401
402
404
405
406
407
409
410
410
411
414
501
502
502
502
503
504
504
504
504
504
506
ANNUAL AVERAGE
NO. OF CYSTS OBSERVATION WAS
IN RAW WATER DELETED (*)
506.73
1758.79
184.44 *
22.48
86.40
49.13 •
75.16 *
16.25
720.42
4.33
3087.50
174.15
1084.42
2143.38
1760.42
1760.42
1706.25
1706.25
151.67
61.48
189.31
102.65
894.02
894.02
80.71 *
22.56 *
54.17 *
22.48
121.87
12.59
4.93
110.77
3096.98
1407.79
135.42
182.81
270.83
6770.83
39.54 *
366.55 *
157.76 *
331.77
1692.71
108.33
13.54
595.83
54.17
1780.73

-------
    LISTING OF RAW WATER DATA FROM LE CHEVALLIER USED TO CHARACTERIZE
             GEOGRAPHIC VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.08 TO REFLECT RETRIEVAL EFFICIENCY
   MEASUREMENTS HAVE ALSO BEEN MULTIPLIED BY 0.13 AS A VIABILITY FACTOR


CITY
508
509
511
512
512
513
514
516
517
518
519
602
603
604
605
605
605
605
605
605
605
605
606
606
608
609
610
611
612
613
614
615
616
618
619
703
703
ANNUAL AVERAGE
NO. OF CYSTS OBSERVATION WAS
IN RAW WATER DELETED (*)
32.50
108.33
89.92
195.00
514.58
1468.73
37.92
410.85
124.96
162.50
541.67
338.54
90.46
2499.25
601.79
622.37
855.02
270.83
542.21
1303.52
277.60
1516.67
542.21
277.60
376.19 *
180.65 *
492.37
29.01 *
801.94
1269.53
297.92
24808.33 *
98.58
108.87
60.94
52.00 *
19.34 *

-------
    LISTING OF RAW WATER DATA FROM LE CHEVALLIER USED TO CHARACTERIZE
             GEOGRAPHIC VARIATION (IN CYSTS PER 100 GALLONS)
                 MULTIPLE MEASUREMENTS AT A CITY AVERAGED
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.08 TO REFLECT RETRIEVAL EFFICIENCY
   MEASUREMENTS HAVE ALSO BEEN MULTIPLIED BY 0.13 AS A VIABILITY FACTOR


CITY
1
2
102
109
302
305
306
307
310
311
312
402
404
405
406
407
409
410
411
414
501
503
504
506
508
509
511
512
513
514
516
517
518
519
602
603
604
605
606
610
612
613
614
616
618
619
NUMBER
OF
MEASUREMENTS







8
2
1
3
1
1
1
1
1
1
2
1
1
1
1
5
1
1
1
1
2
1
1
1
1
1
1
1
1
1
8
2








CITY
AVERAGE
506.73
1758.79
22.48
86.40
16.25
720.42
4.33
1677.85
106.57
189.31
630.23
22.48
121.87
12.59
4.93
110.77
3096.98
771.60
182.81
270.83
6770.83
331.77
492.92
1780.73
32.50
108.33
89.92
354.79
1468.73
37.92
410.85
124.96
162.50
541.67
338.54
90.46
2499.25
748.75
409.91
492.37
801.94
1269.53
297.92
98.58
108.87
60.94

-------
                                 SUMMARIES OF GIAROIA AND NATURAL LOGARITHM OF GIARDIA IN RAW WATER





                                                        UNIVARIATE PROCEDURE
Variable=GIARRAW       GIARDIA IN RAW WATER




                 Moments
Ouantiles(Def=5)
Extremes
N
Mean
Std Dev
Skeuness
USS
CV
T:Hean=0
Sgn Rank
Nun "= 0
U: Normal
46
657.3852
1149.549
3.812403
79344952
174.8668
3.878568
540.5
46
0.572595
Sun Wgts
Sun
Variance
Kurt os is
CSS
Std Mean
Prob>|T|
Prot»|s|
ProbtU
46
30239.72
1321462
17.89069
59465807
169.4917
0.0003
0.0001
0.0001
100% Max
75X 03
SOX Med
25% 01
OX Min

Range
03-01
Mode

6770.833
720.4167
284.375
90.45833
4.333333

6766.5
629.9583
22.47917

99X
95X
90X
10X
5X
IX


6770.833
2499.25
1758.792
22.47917
12.59375
4.333333


Lowest
4. 333333 (
4.929167C
12.59375(
16.25(
22. 4791 7(


Obs
7)
15)
14)
5)
12)


Highest
1758. 792(
1780.729(
2499. 25 (
3096.979(
6770.833(


Obs
2)
24)
37)
17)
21)



-------
                                 SUMMARIES OF GIARDIA AND NATURAL LOGARITHM OF GIARDIA IN RAW WATER





                                                        UNIVARIATE PROCEDURE
Variable=LGGIAR
                       NATURAL LOG OF GIARDIA IN RAU WATER
                 Moments
                                                          Quantiles(Def=5)
                                                                                                           Extremes
N
Mean
Std Dev
Skeuness
USS
CV
T:Mean=0
Sgn Rank
Nun *= 0
W:Normal
46
5.393429
1.668332
-0.37578
U62.852
30.93842
21.92203
540. 5
46
0.975227
Sun Wgts
Sun
Variance
Kurtosis
CSS
Std Mean
Prob>|T|
Prob>|S|

Prob
-------
APPENDIX B

-------
             SUMMARY OF RAW UATER DATA FROM H1BLER USED TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED

                                    NUMBER
                                      OF
                            CITY     CYSTS    FREQUENCY

                               1
                              11
                              20
                              32
0
4
6
12
24
30
0
2
3
5
8
12
36
63
65
0
3
24
0
1
2
4
5
6
8
28
0
1
6
7
9
1
2
1
1
2
9
1
1
2
2
1
1
1
1
21
1
1
49
1
2
4
1
2
2
1
19
1
1
1

-------
             SUMMARY OF RAU WATER DATA FROM HIBLER USEO TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED

                                    NUMBER
                                      OF
                            CITY     CYSTS    FREQUENCY

                              40
                              61
0
1
I
3
t.
6
7
8
9
10
11
12
15
14
15
17
30
0
2
5
6
12
13
20
48
50
51
0
1
2
6
10
12
13
22
40
50
162
0
13
49
35
2
6
1
5
1
2
2
3
1
2
3
2
1
2
1
1
29
4
1
1
1
1
1
' 1
1
1
28
4
3
2
2
2
1
1
1
1
1
2
1
1

-------
             SUMMARY OF RAW WATER DATA FROM H18LER USED TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED


CITY
64




69

71



73

NUMBER
OF
CYSTS
0
3
4
5
6
6
13
0
1
5
20
0
1


FREQUENCY
24
1
1
2
2
1
1
5
1
1
1
3
1

-------
             SUMMARY OF RAW WATER DATA FROM HI8LER USED TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED
NUMBER
OF
CITY CYSTS
74 0
2
3
4
5
6
9
11
12
13
15
20
22
29
34
36
37
38
58
66
67
95
105
130
134
147
160
204
220
252
740
86 0
1
6
9


FREQUENCY
15
7
2
2
1
2
2
4
1
1
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
6
2
1
1

-------
             SUMMARY OF RAW WATER DATA FROM HI BIER USED TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUKOEO
NUMBER
OF
CITY CYSTS
89 0
2
3
4
5
6
8
10
11
12
13
14
17
21
24
38
42
90 0
1
2
3
5
7
9
10
11
13
16
18
23
35
52


FREQUENCY
90
1
4
1
S
2
1
2
2
2






1
184
3
2
1
3
2
1
3
1
1
1
1
1
1
1

-------
             SUMMARY OF RAU WATER DATA FROM H1BLER USED TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED


CITY
91















93


















100


NUMBER
OF
CYSTS
0
1
2
3
4
6
7
8
9
10
11
14
17
IB
30
88
0
2
10
13
48
50
58
59
63
88
149
191
406
433
441
448
800
336
1301
0
32
45


FREQUENCY
170
2
2
3
1
3
1
1
2
2
2
1
2
1
1
1
59
1
1
1
1
1
1
1
2
1
1
1
1
1
1
1
1
1
1
7
1
1

-------
             SUMMARY OF RAW WATER DATA FROM HIBLER USED  TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100  GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED 8Y 2.0 TO REFLECT RETRIEVAL  EFFICIENCY  AND ROUNDED

                                    NUMBER
                                      OF
                            CITY     CYSTS    FREQUENCY

                             101
                             102
                             105
                             107
                             109
                             111
                             118
                             120
                             121
0
40
200
352
0
10
0
102
2
16
0
1
2
3
4
5
6
7
8
9
11
13
16
17
24
0
2
9
93
0
1
2
0
2
5
2
1
1
1
1
1
4
1
1
1
52
2
3
1
2
3
1
1
1
1
1
1
1
1
1
4
1
1
1
52
1
1
16
1
1

-------
             SUMMARY OF RAW WATER DATA FROM HIBLER USED TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE SEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED

                                    NUMBER
                                      OF
                            CITY     CYSTS    FREQUENCY
122


1JO






136
139
U2

U7






U9


0
I
243
0
2
3
15
17
19
22
6
0
0
10
0
2
11
16
20
28
30
0
27
37
2
1
1
9
1
2
1
1
1
1
1
4
16
1
2
1
1
1
1
1
1
1
1
1

-------
             SUMMARY OF RAW WATER DATA FROM HIBLER USED TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED
NUMBER
OF
CITY CYSTS
150 0
1
2
3
5
6
7
8
9
11
13
14
16
17
22
24
28
29
37
40
133
146
152 0
1
2
4
5
6
7
8
9
15
16
20
23
24
26
28
31
36
106


FREQUENCY
49
5
4
1
1
2
2
1
1
1
1
1
1
2
1
2
1
1
1
1
1
1
27
. 1
1
2
1
2 '
1
1





3
1
1
3
1
1

-------
             SUMMARY OF RAW WATER OATA FROM HI3LER USED TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED
                                                                      10
                            CITY

                             159
NUMBER
  OF
 CYSTS

    0
   16
   27
FREQUENCY

     2
     1
     1
160
162




2
0
2
6
16
19
1
7
1
1
1
2
164
165

168




2
0
47
0
2
5
3
30
1
3
1
3
1
1
1
1
169
173
174





175



195

12
542
0
12
43
46
47
55
0
3
4
31
50
54
1
1
8





4
1
1
1
1
1

-------
             SUMMARY OF RAW WATER DATA FROM HI BIER USED TO CHARACTERIZE
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED


CITY
199






201
207


208


212
215

217



229
230

234
236
241
246
248
. 256
NUMBER
OF
CYSTS
0
2
4
6
10
13
17
18
0
10
14
0
6
7
6
0
14
0
2
6
11
20
17
19
18
17
3
6
648
26


FREOUENI
34
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
1
1
1
1 '
1
1
1
1
1
1
1
1

-------
             SUMMARY OF RAW WATER DATA FROM H1BLER USED TO CHARACTERIZE                                   12
                   TEMPORAL VARIATION (IN CYSTS PER 100 GALLONS)
MEASUREMENTS HAVE BEEN MULTIPLIED 3T 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED


CITY
257
258
259

268
269
295


299
302

308

NUMBER
OF
CYSTS
36
(
0
5
27
66
0
2
12
10
0
8
0
16


FREQUENCY
1
1
1
1
1
1
7
1
1
1
1
1
7
1

-------
                 SUMMARY STATISTICS FOR RAW WATER DATA FROM HI8LER
MEASUREMENTS HAVE BEEN MULTIPLIED BY 2.0 TO REFLECT RETRIEVAL EFFICIENCY AND ROUNDED


CITY
T
8
11
20
32
40
41
61
62
64
69
71
73
74
86
89
90
91
93
100
101
102
105
107
109
111
118
120
121
122
130
136
139
142
147
149
150
152
159
160
162
164
165
168
169
173
174
175
195
199
NUMBER
OF CYST
MEASUREMENTS
16
19
23
62
22
70
41
46
4
30
2
8
4
59
10
117
206
195
78
9
5
2
5
2
72
5
2
54
18
4
16
1
4
17
8
3
81
51
4
1
12
1
4
7
1
1
13
7
2
41
PERCENTAGE OF
MEAS. SHOWING
ZERO CYSTS
56.250
47.368
91.304
79.032
86.364
50.000
70.732
60.870
50.000
80.000
0.000
62.500
75.000
25.424
60.000
76.923
89.320
87.179
75.641
77.778
40.000
50.000
80.000
0.000
72.222
80.000
0.000
96.296
88.889
50.000
56.250
0.000
100.000
94.118
25.000
33.333
60.494
52.941
50.000
0.000
58.333
0.000
75.000
42.857
0.000
0.000
61.538
57.143
0.000
82.927
MINIMUM
CYST
MEASUREMENT
0
0
0
0
0
0
0
0
0
0
6
0
0
0
0
0
0
0
•0
0
0
0
0
2
0
0
9
0
0
0
0
6
0
0
0
0
0
0
0
2
0
2
0
0
12
542
0
0
50
0
MAXIMUM
CYST
MEASUREMENT
30
65
24
28
7
30
51
162
48
6
13
20
1
740
9
42
52
88
1301
45
352
10
102
16
24
2
93
2
5
243
22
6
0
10
30
37
146
106
27
2
19
2
47
30
12
542
55
31
54
17
MEDIAN
CYST
MEASUREMENT
0.0
2.0
0.0
0.0
0.0
0.5
0.0
0.0
6.5
0.0
9.5
0.0
0.0
9.0
0.0
0.0
0.0
0.0
0.0
0.0
40.0
5.0
0.0
9.0
0.0
0.0
51.0
0.0
0.0
1.5
0.0
6.0
0.0
0.0
13.5
27.0
0.0
0.0
8.0
2.0
0.0
2.0
0.0
2.0
12.0
542.0
0.0
0.0
52.0
0.0
MEAN OF
CYST
MEASUREMENTS
7.000
10.789
1.174
1.323
0.636
4.071
5.195
7.674
15.250
0.967
9.500
3.250
0.250
46.712
1.700
2.547
1.194
1.518
69.987
8.556
118.400
5.000
20.400
9.000
2.014
0.400
51.000
0.056
0.389
61.500
5.062
6.000
0.000
0.588
13.375
21.333
7.840
9.549
10.750
2.000
5.167
2.000
11.750
6.429
12.000
542.000
15.615
5.429
52.000
1.317
STANDARD DEVIATION
OF CYST
MEASUREMENTS
11.027
20.376
5.015
3.995
1.916
5.906
13.312
25.326
22.677
2.025
4.950
6.985
0.500
109.488
3.164
6.674
5.230
7.241
212.689
17.285
154.444
7.071
45.616
9.899
4.564
0.894
59.397
0.302
1.243
121.008
8.045

0.000
2.425
12.153
19.140
22.898
17.715
13.200

7.964

23.500
10.830


22.681
11.400
2.828
3.698

-------
                                   SUMHARV STATISTICS FOR  RAU WATER  DATA  FROM  H1BLER
                  MEASUREMENTS HAVE BEEN MULTIPLIED 3V 2.0 TO REFLECT  RETRIEVAL  EFFICIENCY  AND  ROUNDED
14
CITY

 201
 207
 208
 212
 215
 217
 229
 230
 234
 236
 241
 246
 248
 256
 257
 258
 259
 268
 269
 295
 299
 302
 308
NUMBER
OF CtST
MEASUREMENTS
f
1
3
1
2
5
1
2
1







2
1
1
9
1
2
8
PERCENTAGE OF
ME AS. SHOWING
ZERO CYSTS
0.0000
33.3333
33.3333
0.0000
50.0000
40.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
50.0000
0.0000
0.0000
77.7778
0.0000
50.0000
87.5000
MINIMUM
CTST
MEASUREMENT
18
0
0
6
0
0
20
17
18
17
3
6
648
26
86
4
0
27
68
0
10
0
0
MAX I MUM
CYST
MEASUREMENT
18
14
7
6
14
11
20
19
18
17
3
6
648
26
86
4
5
27
68
12
10
8
16
MEDIAN
CTST
MEASUREMENT
18.0
10.0
6.0
6.0
7.0
2.0
20.0
18.0
18.0
17.0
3.0
6.0
648.0
26.0
86.0
4.0
2.5
27.0
68.0
0.0
10.0
4.0
0.0
MEAN OF
CTST
MEASUREMENTS
18.000
8.000
4.333
6.000
7.000
3.800
20.000
18.000
18.000
17.000
3.000
6.000
648.000
26.000
86.000
4.000
2.500
27.000
68.000
1.556
10.000
4.000
2.000
STANDARD DEVIATION
Of CYST
MEASUREMENTS
.
7.21HO
3.78594
.
9.89949
4.71169
.
1.41421
.
.
.
.
.
.
.
.
3.53553
.
.
3.97213
.
5.65685
5.6568S

-------
APPENDIX C

-------
      Expected Value  [E(X>] of  the Delta-Negative  Binomial  Distribution
                                                     _.
                                                    \ g+m
                                         r(g)-x!






                                                 ./_2L\
                                      T(g) •(*-!)!






                                  Let j -  x-1






                                                 kl./-™-
         o(l-6).(-3-\V-gL]    (rj}'*lrl} 'f-5-)".  since r(a+l)=a.r(a)
                \g+m/ \g*m/^s    T(g)»j!    I q+n)






               = (i-6, .g.(_2_)*.(_2_U_2_r"*"    (see note
                        I g*m/ \ g+m/ \ g+m/
From Jolley (1961), formula 1015:
                                   .
                   2!       g*m        1   g*m/         g*m

-------
          Variance [V(X)] of  the  Delta-Negative Binomial Distribution
                             V(X) -  E(X2) -  [E(X)]2

                                  -  E(X2) -  [2
                                                                (see note below)
                                         g+m



                                    Let j-x-2
                           [since r(a+2)=(a+l)*a*r(a)]
= (i-8).(_2_)ff.(_2_)2.(
-------
APPENDIX D

-------
City
90
91
89
150
93
109
40
20
74
120*
152
61
41
199
64
11*
32
8
121
142*
1
130
174
162
86
100
295
71
147
308
168
175
105*
101
111
217
62
73
122
139*
159
165
Sample
Size
206
195
117
81
78
72
70
62
59
54
51
46
41
41
30
23
22
19
IB
17
16
16
13
12
10
9
9
8
8
8
7
7
5
5
5
5
4
4
4
4
4
4
q
0.717917
0.677301
1.625189
0.209461
0.553427
1.341256
1.874561
1.301950
0.256774
00
1.102640
0.112547
0.441110
1.953953
00
0.880435
2.707007
0.570479
00
00
2.038864
1.355463
5.520679
2.272272
0.254426
376.89806
1.446761
0.294975
2.346651
OB
0.914368
0.621825
00
1.653133
OB
3.696419
2.932655
00
0.281353
00
50.514247
OB
m
9.527345
10.028699
10.619969
11.264430
278.938986
6.579215
7.781179
5.592729
46.231592
1.500000
19.489594
6.507630
13.936982
7.351698
4.833013
13.500000
4.302894
17.162917
3.519917
10.000000
15.809209
10.995573
40.699701
12.213228
1.811588
38.499536
6.684823
5.006120
17.580409
15.999823
10.568594
10.508781
102.000000
198.003206
1.633307
6.163064
30.968027
0.326039
117.737349
0.000000
21.498798
46.979735
£
0.874538
0.8487S1
0.760072
0.304341
0.749330
0.693713
0.477132
0.758171
0
0.962963
0.510031
0
0.628038
0.821713
0.800007
0.913043
0.849712
0.381082
0.885147
0.941176
0.557102
0.543141
0.626873
0.577371
0
0.775875
0.765451
0.358253
0.251390
0.874989
0.420023
0.487783
0.800000
0.402287
0.752298
0.383741
0.510882
0.222424
0.488077
1.000000
0.515855
0.750057
Note: * denotes starting values for these parameters.  Nonlinear algorithm
      failed to detect any optimum parameter estimates away from the starting
      values.

-------
APPENDIX E

-------
ACTUAL CITY 109 GIARDIA OBSERVATIONS


                  (N=72)
   1.0 -i
 W
 W
 o
 o
   0.8 -
 0 0.6 -
 
-------
APPENDIX F

-------
SIMULATION BASED ON THE DELTA-NEGATIVE BINOMIAL. MODEL


             PARAMETERS FOR CITY 109 (N=10,000)
         w
         w
         OS
         o

         fa
            1.0 n
            0.8 -
         PQ

         m  0.4
         o
         PS
         a,
         a
         w
            0.2 -
0.0
•M'f f-f-f-T-T-TT-T-T-TT-TT
                                      T
                 o i z :i 4 s «
                             10 II 12 IJ 14 Ifi IB IT IB I!) <» 
-------
APPENDIX 6

-------
     Application of a Binomial Removal  Process  to  a

          Delta-Negative Binomial Distribution



          Let X - number of cysts in raw water

          and K - number of cysts in finished water.



              For k-0 and x-0, Pr(K-0)-6.



                 For k>0 and xX>  (x£k) :




.TUN* -.I-,, £ ^.^-.^.a-p,".^^^)'.^
                      Lee J - x-k




          P*  J  a\'J  a \4f* l.(1-

         flXfl) A g*«j  \^«)     js  '

           V-flL\*» r^^ [l-f {1-P^|rtri>   (see note
             lo>*/  *t»r
-------
APPENDIX H

-------
                                       ESTIMATION OF  r  IN EXPONENTIAL DOSE-RESPONSE FUNCTION    14:32 Thursday, February 20, 1992
                                     BASED ON  HUNAN  INFECT1VITT STUDY AS CITED IN ROSE ARTICLE
                                                                                             Method: Gauss-Newton
Non-Linear Least Squares Iterative
Iter
0
1
2
3
4
5
6
7
8
9
10
11
12
13
Phase Dependent variable RESP
R Sum
0
0.0000014043
0.00000434*0
0.0000110043
0.0000259919
0.0000837003
0.000190
0.000343
0.001451
0.012521
0.017625
0.017917
0.017907
0.017907
of Squares
21.000000
16.593611
14.226365
12.735403
11.781231
10.500940








.928268
.755801
.036939
.940740
.729831
.729345
.729345
.729345
NOTE: Convergence criterion net.
                               Non-Linear Least Squares Suwasry Statistics     Dependent Variable RESP

                                  Source                DF SIM of Squares     Mean Square
                                  Regression
                                  Residual
                                  Unconnected Total
 1   15.270655199
39    5.729344801
40   21.000000000
                                 15.270655199
                                  0.146906277
                                  (Corrected Total)     39    9.975000000
                                  Parameter     Estimate
                Asysptotfc             Asysptotfc 95 X
                Std. Error         confidence interval
                                   Lower         Upper
0.0179074539 0.00502972826 0.00773391748 0.02808099039
                                                   Asymptotic Correlation Matrix

                                                     Corr                     R

                                                            R                 1

-------
APPENDIX I

-------
Variance of the Exponential Dose-Response  Function where Dose (K)
      Ha* a Delta-Negative Binomial Distribution (continued)

                    Now,  E(X2)  -  E(XZ)  -  [E(X)]Z
 So'EiX*)  - (1-*)
    1 -
                 - (I-*)1
 	2	r*
 m-p) •(!-«•'•»;
 -(	2	rf
   lo>«m.p)« J   J
      ,  „ /         «.        \t   „  ..,  /	<(fli«p>»(l-e-**}); )

-------
           Expected Value  of the Exponential Dose-Response  Function Where
                 Dose  (K)  Has a Delta-Negative Binomial Distribution
                          -e<-*-»] [8-1 „(*) +(i-6)
                                  I
             -8.(l-exp(0)) * VHi-e <-'••*'] (1-8)
               (1 expiujj     tl  •     j  i o>
                        g   \»
                (1-8)  - (1-8) •
                        -8) -
                         o;
                                     «7*(in»p}
                            (!-*)  -  (1-8) •(
                                          \
                               (1-«4-[ - 2 -
                                    I  [gMfli-PMl-e -'*))
                    • (1-8) - (1-8) •(	3—p\TC JT/yJj!  •fffl'p.'e'i.
From Jolley,  1961, formula 1015:





                         ^ ^ MM-L^^^»*I •••^^•^WB J » 1 *^*|
                2!
                                        _f gi>p»e -*')'. f ar*(in»p) -

-------
         Variance of Che Exponential Dose-Response Function where Dose (K)
                     Has a Delta-Negative  Binomial Distribution
   (from derivation of the expected value of the exponential dose-xesponse function)
              (l-8)-2«(l-«)«f - 2 - )%{!-«) 4 - 2 - ._]'
                           \ 
-------
APPENDIX J

-------
                       exdr = 11
-a,U
   I  L
                                      <7+
-------
       SVTR DATA --  FIRST CUSTOMER --  SURFACE:   NO  SOFTENING  -- HAAs-60
                        WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2. 1992 VERSION

                                                 Percentage of Time  Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 252
SOX
OX
90th Xile 25X
50Z
OX
99th Xile 25X
SOX
Removal is Reduced bv 1*

OX
124
16S
248
245
326
490
1,098
1,464
2,196

5X 10X
180 235
239 313
359 470
355 464
473 619
709 928
1,586 2,070
2,115 2,760
3,172 4,139
Percentage of Time Log



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 2SX
SOX
Removal is Reduced bv 1*

OX
00
00
GO
00
1.17xl09
32.S68
0.007
0.004
0.003

5X 10X
00 00
oo 13.4xl09
502,984 84.384
70.6x10* 697.352
212.089 0.700
0.098 0.014
0.004 0.003
0.003 0.003
0.003 0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SWTR DATA --  FIRST CUSTOMER --  SURFACE:   NO SOFTENING --  HAAs-50
                        WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2. 1992 VERSION

                                                 Percentage of Time Log
Average Number
of Cases per
10,000 per Year
Average Number
of Years for
an Outbreak
Secondary
Infection
Rate
OZ
Mean 25Z
50Z
OZ
90th Zile 25Z
50Z
OZ
99th Zile 25Z
SOZ
Secondary
Infection
Rate
OZ
Mean 25Z
SOZ
OZ
90th Zile 2SZ
SOZ
OZ
99th Zile 25Z
SOZ
Removal is Reduced
OZ 5Z
129 187
172 250
259 375
312 451
416 602
623 903
1,098 1,586
1,464 2,115
2,196 3,172
Percentage of Time
Removal is Reduced
OZ SZ
0 OD
CO 00
24.7xl012 94,107
249x10' 1,754
13,775 1.091
0.442 0.016
0.007 0.004
0.004 0.003
0.003 0.003
bv 1*
10Z
245
327
491
591
788
1,181
2,070
2,760
4,139
Log
bv 1*
10Z
00
1.02x10'
30.900
1.910
0.043
0.006
0.003
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SWTR DATA -- FIRST CUSTOMER --  SURFACE:   NO SOFTENING --  HAAs-40
                        WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage of Time Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 25Z
SOZ
OZ
90th Zile 2SZ
SOZ
OZ
99th Zile 25Z
SOZ
Removal

OZ
139
186
278
322
429
644
1,098
1,464
2,196
is Reduced

SZ
202
269
404
466
621
932
1,586
2,115
3,172
Percentage of Time



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 2SZ
SOZ
OZ
90th Zile 2SZ
SOZ
OZ
99th Zile 2SZ
SOZ
Removal

OZ
00
00
224x10'
is Reduced

SZ
GO
00
6,746
27.1x10' 616.588
4,392
0.290
0.007
0.004
0.003
0.658
0.014
0.004
0.003
0.003
bv 1*

10Z
264
352
529
610
813
1,220
2,070
2,760
4,139
Log
bv 1*

10Z
oo
17.7xl06
6.498
1.089
0.033
0.005
0.003
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SWTR DATA -- FIRST CUSTOMER --  SURFACE:   NO SOFTENING --  HAAs-30
                        WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage of Time Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 25Z
SOZ
OZ
90th Zile 2SZ
50Z
OZ
99th Zile 25Z
SOZ
Removal

OZ
163
217
326
355
474
710
1,098
1,464
2,196
is Reduced bv 1*

5Z 10Z
236 309
315 412
472 618
514 673
686 897
1,029 1,346
1,586 2,070
2,115 2,760
3,172 4,139
Percentage of Time Log



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25Z
SOZ
OZ
90th Zile 25Z
SOZ
OZ
99th Zile 25Z
SOZ
Removal

OZ
is Reduced bv 1*

52 10Z
co to 457x10'
« 10.8x10' 18,872
38.9x10*
64.5xl06
202. 563
0.096
0.007
0.004
0.003
77.375 0.496
37.963 0.249
0.174 0.017
0.009 0.004
0.004 0.003
0.003 0.003
0.003 0.003
*Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SVTR DATA --  FIRST  CUSTOMER -- SURFACE:  NO SOFTENING  -- HAAs-20
                       WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                Percentage of Time  Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 25X
50Z
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal

OX
203
271
406
420
561
841 1
1,642 2
2,189 3
3,284 4
is Reduced bv 1*

5X 10X
294 38S
392 514
588 770
609 796
812 1,062
,217 1,593
,367 3,082
,155 4,110
,733 6,165
Percentage of Time Log



Average Number
of Years for
an Outbreak






Secondary
J
Infection
Rate
OX
Mean 2SX
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 25X
SOX
Removal

OX
« 12.
is Reduced bv 1*

5X 10X
3xl012 1.02xl06
« 130,966 25.222
5,548
23,236
3.861
0.024
0.003
0.003
0.003
1.018 0.046
1.127 0.041
0.034 0.008
0.005 0.004
0.003 0.003
0.003 0.003
0.003 0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SVTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING  -- HAAs-
                        WITHOUT ALTERNATE DISINFECTION
                            JANUARY 2,  1992  VERSION
-10
                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 2SX
SOX
Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal is Reduced bv 1*

OX
255
339
509
548
731
1,096
3,736
4,982
7,473

5X
369
492
738
793
1,057
1,586
5,344
7,126
10,689
Percentage of Time
Removal is Reduced

OX
00
127x10*
13.848
8.591
0.086
0.007
0.003
0.003
0.003

5X
8.69x10*
73.707
0.067
0.-043
0.008
0.004
0.003
0.003
0.003

10X
483
644
966
1,037
1,383
2,074
6,909
9,212
13,818
Log
bv 1*

10X
212.187
0.394
0.011
0.009
0.004
0.003
0.003
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SWTR DATA  -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- HAAs-60
                        WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal

OX
0.129
0.172
0.258
0.342
0.455
0.683
0.702
0.936
1.404
is Reduced

5X
0.187
0.249
0.374
0.495
0.660
0.991
1.018
1.358
2.036
Percentage of Time
Removal is Reduced

OX
00
00
00
GO
09
00
OB
00
CD

5X
09
00
00
00
00
00
00
00
00
by 1*

10X
0.245
0.327
0.490
0.649
0.865
1.298
1.334
1.779
2.668
Log
by 1*

10X
00
00
00
00
00
OB
00
00
00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
   ENHANCED  SVTR DATA -- FIRST CUSTOMER  -- SURFACE:  NO SOFTENING  -- HAAs-50
                        WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2. 1992 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25Z
50Z
OX
90th %ile 25Z
SOX
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 2SX
SOX
OZ
90th Zile 25Z
SOX
OX
99th Xile 25X
SOX
Removal

OZ
0.134
0.179
0.268
0.346
0.461
0.692
0.702
0.936
1.404
is Reduced

SZ
0.195
0.259
0.389
0.502
0.669
1.003
1.018
1.358
2.036
Percentage of Time
Removal is Reduced
•
OX
00
00
00
00
CD
OB
00
OP
00

5X
00
00
00
00
00
00
00
00
00
by 1*

10Z
0.255
0.340
0.510
0.657
0.877
1.315
1.334
1.779
2.668
Log
by 1*

10Z
00
00
00
00
00
00
00
00
00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SVTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- HAAs-AO
                        WITHOUT  ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage  of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
02
Mean 25X
SOX
02
90th Xile 25Z
50Z
01
99th Xile 25Z
502
Secondary
Infection
Rate
OX
Mean 2SX
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 2SX
SOX
Removal is

OX
0.141 0.
0.188 0.
0.282 0.
0.346 0.
0.461 0.
0.692 1.
0.702 1.
0.936 1.
1.404 2.
Percentage
Removal is

OX
00
00
00
00
00
09
CO
00
oo
Reduced

5X
205
273
410
502
669
003
018
358
036
of Time
Reduced

5X
00
00
00
00
00
00
m
00
00
bv 1*

10X
0.268
0.358
0.537
0.657
0.877
1.315
1.334
1.779
2.668
Log
by 1*

10X
00
00
CD
CO
to
00
00

00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SVTR DATA  -- FIRST CUSTOMER  -- SURFACE:  NO SOFTENING  -- HAAs-30
                        WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2. 1992 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 252
SOX
OX
90th %ile 25X
SOX
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 2SX
SOX
Removal

OX
0.162
0.216
0.324
0.372
0.496
0.745
0.702
0.936
1.404
is Reduced

5X
0.235
0.313
0.470
0.540
0.720
1.080
1.018
1.358
2.036
Percentage of Time
Removal is Reduced

OX
09
09
00
oo
as
00
00
00
00

5X
a
00
00
09
09
09
09
09
09
by 1*

10X
0.308
0.410
0.615
0.707
0.943
1.415
1.334
1.779
2.668
Log
by 1*

10X
03
00
00
to
CO
00
CD
03
'03
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SVTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- HAAs-20
                        WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage  of  Time Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 25X
SOX
Removal

OX
0.164
0.218
0.328
0.355
0.473
0.710
0.702
0.936
1.404
is Reduced

5X
0.238
0.317
0.475
0.514
0.686
1.029
1.018
1.358
2.036
Percentage of Time



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal

OX
CO
00
00
00
00
OB
00
00
00
is Reduced

5X
00
00
OB
CD
00
00
OB
00
00
by 1*

10X
0.311
0.415
0.622
0.674
0.899
1.348
1.334
1.779
2.668
Log
by 1*

10X
CO
00
00
00
a
00
00
00
CD
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SVTR DATA  -- FIRST CUSTOMER  -- SURFACE:  NO SOFTENING  -- HAAs-10
                        WITHOUT ALTERNATE DISINFECTION
                           JANUARY 2. 1992 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25Z
50Z
OZ
90th Zile 25Z
50Z
OZ
99th Zile 25Z
50Z
Secondary
Infection
Rate
OZ
Mean 25Z
50Z
OZ
90th Zile 25Z
50Z
OZ
99th Xile 25Z
50Z
Removal is Reduced

OZ 5Z
0.164 0.237
0.218 0.317
0.327 0.475
0.356 0.517
0.475 0.689
0.712 1.033
0.710 1.029
0.946 1.372
1.419 2.058
Percentage of Time
Removal is Reduced

OZ 5Z
00 CO
00 00
00 00
00 -00
00 OD
OD OD
GO 00
OB 00
00 00
by 1*

10Z
0.311
0.415
0.622
0.677
0.902
1.354
1.348
1.797
2.696
Log
by 1,*

10Z
CO
OB
CO
OB
00
00
OD
CD
OB
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SVTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- HAAs-60
                         WITH ALTERNATE DISINFECTION
                           JANUARY 2,  1992 VERSION

                                                 Percentage of Time Log
Secondary
Removal is Reduced bv 1*
Infection
Rate
Average Number
of Cases per Mean
10,000 per Year

90th Xile


99th Xile

OX
25X
SOX
OX
25X
SOX
OX
25X
SOX
OX
107
143
214
224
299
449
793
1,058
1,587
5X
155
207
310
325
434
650
1,147
1,529
2,294
10X
203
271
406
426
568
851
1,499
1,998
2,997
Average Number
of Years for
an Outbreak
Mean
                          Secondary
                          Infection
                            Rate
 OX
25X
SOX
                               Percentage of Time Log
                               Removal is Reduced bv 1*
                             OX
OD

GO
                             5X
   00

448xl06
                       10X
  00

5,380
                  90th Xile
             OX
            25X
            SOX
             287xl09
             280.611
        13.6xlO»
          3,089
          0.2S6
            14,402
             3.052
             0.022
                  99th Xile
             OX
            25X
            SOX
               0.043
               0.008
               0.004
          0.006
          0.004
          0.003
             0.004
             0.003
             0.003
*Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SWTR  DATA --  FIRST  CUSTOMER  -- SURFACE:  NO SOFTENING  --  HAAs-50
                         WITH ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage of Time  Log
Average Number
of Cases' per
10,000 per Year
Average Number
of Years for
an Outbreak
Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 2SX
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal is Reduced
OX 5X
120 173
159 231
239 346
297 430
396 574
594 861
793 1,147
1,058 1,529
1,587 2,294
Percentage of Time
Removal is Reduced
OX 5X
GO 00
00 00
» 2.27xl06
8.23x10" 9,557
86,925 2.496
0.873 0.021
0.043 0.006
0.008 0.004
0.004 0.003
bv 1*
10X
227
303
454
563
751
1,127
1,499
1,998
2,997
Log
by 1*
10X
00
137x10'
209.821
4.796
0.066
0.007
0.004
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SVTR DATA -- FIRST CUSTOMER --  SURFACE:   NO SOFTENING -- HAAs-40
                         WITH ALTERNATE DISINFECTION
                           JANUARY 2. 1992 VERSION

                                                 Percentage of Time Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Elate
OX
Mean 25Z
SOZ
OZ
90th Zile 25Z
50Z
OZ
99th Zile 25Z
SOZ
Removal

OZ
142
190
284
322
429
644
1.101
1.468
2,202
is Reduced

5Z
206
275
412
466
621
932
1.590
2.120
3,180
Percentage of Time



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25Z
50Z
OZ
90th Zile 25Z
SOZ
OZ
99th Zile 25Z
SOZ
Removal

OZ
00
is Reduced

5Z
CD
c» 24.7x10"
61.5xl09
3,422
27.1x10' 616.588
4,392
0.291
0.007
0.004
0.003
0.658
0.014
0.004
0.003
0.003
bv 1*

10Z
270
360
540
610
813
1,220
2.075
2.767
4,150
Log
bv 1*

10Z
00
6.26x10*
4.369
1.089
0.033
0.005
0.003
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       S¥TR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- HAAs-30
                          WITH ALTERNATE  DISINFECTION
                            JANUARY 2,  1992  VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year





Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 2SX
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal is Reduced bv 1*

OX
163
218
326
355
474
710
1,101
1,468
2.202

5X
237
315
473
514
686
1,029
1,590
2,120
3,180
Percentage of Time
Removal is Reduced

OX
OD
CD 9
34.7x10*
64.5x10'
202.563
0.096
0.007
0.004
0.003

SX
CO
.26x10'
72.955
37.927
0.174
0.009
0.004
0.003
0.003

10X
310
413
619
673
897
1,346
2,075
2,767
4.150
Log
bv 1*

10X
386xlO»
17,249
0.480
0.249
0.017
0.004
0.003
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SVTR DATA --  FIRST CUSTOMER --  SURFACE:   NO SOFTENING --  HAAs-20
                         WITH ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage of Time Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal

OX
208
277
415
463
617
925
1,642
2,189
3,284
is Reduced

SX
301
401
601
669
893
1,339
2,367
3.155
4,733
Percentage of Time



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 2SX
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 25X
SOX
Removal

OX
oo 3
24.7x10"
2,730
785.963
0.740
0.014
0.003
0.003
0.003
is Reduced

5X
.53x10"
53,577
0.730
0.268
0.018
0.005
0.003
0.003
0.003
by 1*

10X
394
525
787
876
1,168
1,752
3,082
4,110
6,165
Log
by 1*

10X
368,842
15.173
0.039
0.021
0.006
0.003
0.003
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SVTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- HAAs-10
                          WITH ALTERNATE DISINFECTION
                            JANUARY 2.  1992  VERSION

                                                 Percentage of Time Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 2 52
SOX
OX
90th Xile 25X
50Z
OX
99th Xile 25X
SOX
Removal is Reduced bv 1*

OX
290
386
580
593
790
1,186
4,405
5,873
8,810

5X
420
560
840
858
1,144
1,715
6,285
8,380
12,571
Percentage of Time



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 2SZ
SOX
OX
99th Xile 25X
SOX
Removal is Reduced

OX
24.7xl012
244,127
1.284
1.795
0.042
0.006
0.003
0.003
0.003

5X
24.844
3.991
0.025
0:024
0.006
0.003
0.003
0.003
0.003

10X
549
733
1,099
1,121
1,495
2,243
8,107
10,809
16,213
Log
bv 1*

10X
8.092
0.084
0.007
0.007
0.004
0.003
0.003
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SWTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- HAAs-60
                         WITH ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 2SX
SOX
Removal is

OX
0.132 0.
0.176 0.
0.26S 0.
0.342 0.
0.455 0.
0.683 0.
0.702 1.
0.936 1.
1.404 2.
Percentage
Removal is

OX
CO
CO
00
CD
00
CO
CD
CO
00
Reduced

5X
192
2S6
384
495
660
991
018
358
036
of Time
Reduced

5X
00
00
00
00
00
CD
00
CD
00
by 1*

10X
0.251
0.335
0.503
0.649
0.865
1.298
1.334
1.779
2.668
Log
by 1*

10X
CO
00
00
00
00
00
CO
00
00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
   ENHANCED SWTR DATA --  FIRST CUSTOMER --  SURFACE:  NO  SOFTENING --  HAAs-50
                          WITH ALTERNATE DISINFECTION
                            JANUARY 2,  1992 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
50Z
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 2SX
50X
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal

OX
0.136
0.182
0.272
0.346
0.461
0.692
0.702
0.936
1.404
is Reduced

5X
0.197
0.263
0.395
O.S02
0.669
1.003
1.018
1.3S8
2.036
Percentage of Time
Removal is Reduced

OX
OB
00
00
00
00
00
00
00
00

SX
CO
CO
00
00
00
00
00
00
00
by 1*

10X
0.259
0.345
0.517
0.657
0.877
1.315
1.334
1.779
2.668
Log
bv 1*

10X
00
00
09
00
00
00
00
00
00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
   ENHANCED  SVTR DATA --  FIRST  CUSTOMER -- SURFACE:  NO SOFTENING  -- HAAs-40
                         WITH  ALTERNATE DISINFECTION
                            JANUARY  2,  1992 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25Z
SOZ
OZ
90th Zile 25Z
50Z
OZ
99th Zile 2SZ
SOZ
Secondary
Infection
Rate
OZ
Mean 2SZ
SOZ
OZ
90th Zile 25Z
SOZ
OZ
99th Zile 2SZ
SOZ
Removal

OZ
0.143
0.190
0.285
0.346
0.461
0.692
0.702
0.936
1.404
is Reduced

5Z
0.207
0.276
0.414
0.502
0.669
1.003
1.018
1.358
2.036
Percentage of Time
Removal is Reduced

OZ
00
00
00
00
00
00
00
OB
00

5Z
CO
00
00
09
CO
CO
CD
00
09
bv 1*

10Z
0.271
0.361
0.542
0.657
0.877
1.315
1.334
1.779
2.668
Log
by 1*

10Z
00
00
00
00
00
00
00
00
00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
   ENHANCED  SWTR DATA  -- FIRST CUSTOMER  -- SURFACE:  NO SOFTENING  --  HAAs-30
                         WITH ALTERNATE DISINFECTION
                           JANUARY 2; 1992 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25Z
50Z
OX
90th Zile 2SX
SOX
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Zile 25X
SOX
OX
99th Zile 25Z
50Z
Removal is Reduced

OZ SZ
0.162 0.235
0.216 0.313
0.324 0.470
0.368 0.533
0.491 0.711
0.736 1.067
0.702 1.018
0.936 1.358
1.404 2.036
Percentage of Time
Removal is Reduced

OZ 5Z
09 00
00 00
00 00
00 00
00 00
00 00
00 09
OD GO
00 09
by 1*

10Z
0.308
0.410
0.615
0.699
0.932
1.398
1.334
1.779
2.668
Log
by 1*

10Z
00
00
00
00
09
00
00
00
00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SVTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- HAAs-20
                         WITH ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak





Secondary
Infection
Rate
OX
Mean 25Z
502
OX
90th Zile 25X
SOX
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile *" 25X
SOX
OX
99th Xile 2SX
SOX
Removal

OX
0.162
0.216
0.325
0.349
0.46S
0.698
0.702
0.936
1.404
is Reduced

SX
0.235
0.314
0.471
0.506
0.675
1.012
1.018
1.358
2.036
Percentage of Time
Removal is Reduced

OX
00
00
00
OB
00
GO
OB
00
00

5X
CD
00
00
CO
00
DO
00
00
00
by 1*

10X
0.308
0.411
0.617
0.663
0.884
1.326
1.334
1.779
2.668
Log
by 1*

10X
00
00
03
00
00
00
to
at
a
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SWTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- HAAs-10
                         WITH ALTERNATE DISINFECTION
                           JANUARY 2, 1992 VERSION

                                                 Percentage  of Time Log



Average Number
of Cases per
10,000 per Year










Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25Z
50Z
OZ
90th Zile 25Z
SOZ
OZ
99th Zile 25Z
SOZ

Secondary
Infection
Rate
OZ
Mean 2SZ
50Z
OZ
90th Zile 2SZ
SOZ
OZ
99th Zile 25Z
SOZ



0
0
0
0
0
0
0
0
1













Removal is

OZ
.159 0.
.212 0.
.318 0.
.350 0.
.467 0.
.701 1.
.693 1.
.925 1.
.387 2.
Percentage
Removal is

OZ
09
CD
CO
OB
CO
GO
00
00
IB
Reduced

5Z
230
307
461
508
677
016
006
341
Oil
of Time
Reduced

5Z
a
00
00
00
00
00
00
00
00
by 1*

10Z
0.302
0.403
0.604
0.666
0.888
1.331
1.318
1.757
2.635
Log
by 1*

10Z
00
00
00
00
00
00
00
00
00
*Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
      SVTR DATA --  FIRST CUSTOMER --  SURFACE:   NO SOFTENING --  TTHMs-100
                        WITHOUT ALTERNATE DISINFECTION
                          DECEMBER 12, 1991 VERSION

                                                 Percentage of  Time Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 252
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal

OX
174
233
349
312
416
623
3,723 5
4,964 7
7,445 10
is Reduced bv 1*

5X 10X
253 331
337 441
506 662
451 591
602 788
903 1,181
,325 6,885
,100 9,179
,650 13,769
Percentage of Time Log



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 2SX
SOX
Removal

OX
00
00
1.67x10*
252xl09
13,775
0.442
0.003
0.003
0.003
is Reduced bv 1*

5X 10X
oo 4.27x10'
187x10* 1,706
16.058 0.206
1,754 1.912
1.091 0.043
0.016 0.006
0.003 0.003
0.003 0.003
0.003 0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
      SWTR DATA  -- FIRST CUSTOMER  -- SURFACE:  NO SOFTENING  -- TTHMs-75
                        WITHOUT ALTERNATE DISINFECTION
                          DECEMBER 12, 1991 VERSION

                                                 Percentage of Time Log



Average Number
of Cases per
10,000 per Year










Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
02
Mean 252
SOX
OX
90th Zile 25*
SOX
OX
99th Xile 25X
SOX

Secondary
J
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 2SX
SOX


OX
194
2S8
387
354
472
707
3,723
4,964
7,445



OX
GO
00
28,134
Removal is Reduced

5X
280
374
561
512
683
1,025
5,325
7,100
10,650
Percentage of Time
Removal is Reduced

5X
bv 1*

10X
367
490
735
670
894
1,341
6,885
9,179
13,769
Log
bv 1*

10Z
o> 10.7x10*
1.02xl06 81.852
2.199
Sl.OxlO6 42.112
227.327
0.100
0.003
0.003
0.003
0.182
0.009
0.003
0.003
0.003
0.070
0.263
0.018
0.004
0.003
0.003
0.003
*Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
      SWTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- TTHMs-50
                       WITHOUT ALTERNATE DISINFECTION
                          DECEMBER 12, 1991 VERSION

                                                 Percentage of Time  Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX
Removal is Reduced

OX
240
320
480
419
559
838
3,723
4,964
7,445

5X
348
463
695
607
809
1,213
5,325
7.100
10,650
Percentage of Time



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 2SX
SOX
bv 1*

10X
455
607
910
794
1,058
1,587
6,885
9,179
13,769
Log
Removal is Reduced bv 1*

OX
00
3 . 84xl09
51.775
26,626
4.132
0.025
0.003
0.003
0.003

5X
220xl06
376.753
0.120
1.195
0.035
0.005
0.003
0.003
0.003

10X
1,333
0.955
0.015
0.043
0.008
0.004
0.003
0.003
0.003
*Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SWTR DATA --  FIRST CUSTOMER  -- SURFACE:  NO SOFTENING  -- TTHMs-25
                        WITHOUT ALTERNATE DISINFECTION
                          DECEMBER 12,  1991 VERSION

                                                 Percentage of Time Log



Average Number
of Cases per
10,000 per Year










Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 2SX
SOX

Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 2SX
SOX


OX
257
343
Removal is Reduced

5X
372
496
514 744
548
731
1,096
3,736
4,982
7,473



OX

-------
 ENHANCED SWTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- TTHMs-100
                       WITHOUT ALTERNATE DISINFECTION
                          DECEMBER 16, 1991 VERSION

                                                 Percentage  of  Time  Log



Average Number
of Cases per
10,000 per Year










Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25Z
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX

Secondary
J
Infection
Rate
OX
Mean 2SX
SOX
OX
90th Xile * 25X
SOX
OX
99th Xile 2SX
SOX
Removal is

OX
0.132 0.
0.175 0.
0.263 0.
0.342 0.
0.455 0.
0.683 0.
0.702 1.
0.936 1.
1.404 2.
Percentage
Removal is

OX
00
CO
00
CO
OB
CD
CO
00
00
Reduced

5X
191
254
381
495
660
991
018
358
036
of Time
Reduced

5X
00
00
CO
00
00
00
00
00
00
by 1*

10X
0.250
0.333
0.500
0.649
0.865
1.298
1.334
1.779
2.668
Log
by 1*

10X
00
00
00
00
00
00
00
00
OB
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SWTR DATA -- FIRST CUSTOMER -- SURFACE:   NO SOFTENING -- TTHMs-75
                        WITHOUT ALTERNATE DISINFECTION
                          DECEMBER 16, 1991 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25Z
SOX
OZ
90th Zile 25Z
50Z
OZ
99th Zile 25Z
50Z
Secondary
Infection
Rate
OZ
Mean 25Z
50Z
OZ
90th Zile 25Z
SOZ
OZ
99th Zile 2SZ
SOZ
Removal

OZ
0.144
0.193
0.289
0.346
0.461
0.692
0.702
0.936
1.404
is Reduced

SZ
0.209
0.279
0.419
0.502
0.669
1.003
1.018
1.358
2.036
Percentage of Time
Removal is Reduced

OZ
00
00
00
CD
00
CD
00
OR
00

5Z
00
CO
CD
-00
OR
00
CD
CO
OB
bv 1*

10Z
0.274
0.366
0.549
0.657
0.877
1.315
1.334
1.779
2.668
Log
by 1*

10Z
00
00
CO
00
00
00
00
00
00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SWTR DATA -- FIRST CUSTOMER -- SURFACE:   NO SOFTENING -- TTHMs-50
                        WITHOUT ALTERNATE DISINFECTION
                          DECEMBER 16, 1991 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 252
SOX
OX
90th Xile 25X
50X
OX
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 2SX
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX


0.
0.
0.
0.
0.
0.
0.
0.
1.

OX
CD
00
00
00
CD
00
OD
00
OD
Removal is Reduced

OX SX
157 0.227
209 0.303
313 0.454
346 0.502
461 0.669
692 1.003
702 1.018
936 1.358
404 2.036
Percentage of Time
Removal is Reduced

5X
00
00
CD
00
00
00
00
00
00
by 1*

10X
0.298
0.397
0.595
0.657
0.877
1.315
1.334
1.779
2.668
Log
by 1*

10X
00
00
00
00
00
00
CD
00
CO
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SWTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- TTHMs-25
                        WITHOUT ALTERNATE DISINFECTION
                           DECEMBER 16, 1991 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25X
SOX
OZ
90th Zile 25Z
50Z
OZ
99th Zile 25Z
50Z
Secondary
Infection
Rate
OZ
Mean 25Z
50%
OZ
90th Zile 25Z
SOZ
OZ
99th Zile 25Z
SOZ


0.
0.
0.
0.
0.
0.
0.
0.
1.

OZ
GD
00
00
00
00
00
00
00
00
Removal is Reduced

OZ 5Z
165 0.240
221 0.320
331 0.480
358 0.519
477 0.692
715 1.037
702 1.018
936 1.358
404 2.036
Percentage of Time
Removal is Reduced

5Z
00
00
00
00
00
00
00
n
00
bv 1*

10Z
0.314
0.419
0.629
0.680
0.906
1.359
1.334
1.779
2.668
Log
by 1*

10Z
OB
09
09
09
00
to
00
09
09
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
      SWTR DATA --  FIRST CUSTOMER --  SURFACE:   NO SOFTENING --  TTHMs-100
                         WITH ALTERNATE DISINFECTION
                          DECEMBER 16, 1991 VERSION

                                                 Percentage of  Time Log



Average Number
of Cases per
10,000 per Year






Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 25X
SOX
Removal is Reduced

OX
171
228
343
294
392
588
3,723
4,964
7,445

5X
248
331
497
426
568
852
5,325
7,100
10,650
Percentage of Time



Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 2SX
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 25X
SOX
Removal is Reduced

OX
00
00
3.65xl06
12.3x10"
129,871
1.014
0.003
0.003
0.003

5X
by 1*

10X
325
434
650
SS8
744
1,116
6,885
9,179
13,769
Log
bv 1*

10X
* 13.6xl09
510x10*
23.657
13,851
2.994
0.022
0.003
0.003
0.003
3,084
0.255
5.870
0.072
0.007
0.003
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
      SWTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- TTHMs-75
                         WITH ALTERNATE DISINFECTION
                          DECEMBER 16. 1991 VERSION

                                                 Percentage of Time Log



Average Number
of Cases per
10,000 per Year










Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25X
50%
OX
90th Xile 2SX
SOX
OX
99th Xile 2SX
SOX

Secondary
Infection
Rate
OX
Mean 25X
50X
OX
90th Xile 25X
SOX
OX
99th Xile 25X
SOX


OX
196
261
392
354
472
707
3,723
4,964
7,445



OX
CD
00
17,841
Removal is Reduced bv 1*

5X 10X
284 372
379 496
568 744
512 670
683 894
1,025 1,341
5,325 6,885
7,100 9,179
10,650 13,769
Percentage of Time Log
Removal is Reduced bv 1*

5X 10X
o 5.52xl06
573,404 58.721
1.769 0.062
81.0x10* 42.112 0.263
227.327
0.100
0.003
0.003
0.003
0.182 0.018
0.009 0.004
0.003 0.003
0.003 0.003
0.003 0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
      SWTR DATA -- FIRST CUSTOMER -- SURFACE:  NO SOFTENING -- TTHMs-50
                         WITH ALTERNATE DISINFECTION
                          DECEMBER 16, 1991 VERSION

                                                 Percentage  of Time Log

Average Number
of Cases per
10,000 per Year




Average Number
of Years for
an Outbreak




Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 2SX
SOX
Secondary
Infection
Rate
OX
Mean 2SX
SOX
OX
90th Xile 2SX
SOX
OX
99th Xile 25X
SOX
Removal is Reduced bv 1*

OX
239
318
477
419
559
838
3,723
4,964
7,445

OX

5X 10X
346 453
461 604
692 906
606 794
809 1,058
1,213 1,587
5,325 6,885
7,100 9,179
10,650 13,769
Percentage of Time Log
Removal is Reduced bv 1*

5X 10X
» 293xl06 1,574
5.21x10' 436.376 1.035
58.281 0.126 0.016
26,626
4.132
0.025
0.003
0.003
0.003
1.195 0.043
0.035 0.008
0.005 0.004
0.003 0.003
0.003 0.003
0.003 0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
       SVTR DATA --  FIRST CUSTOMER --  SURFACE:   NO  SOFTENING --  TTHMs-25
                          WITH ALTERNATE DISINFECTION
                           DECEMBER 16,  1991  VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year


Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 25Z
SOZ
OX
90th Xile 25Z
SOZ
-ox
99th Xile 25X
SOX
Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Zile 25X
SOZ
OZ
99th Zile 25Z
SOZ
Removal is Reduced bv 1*

OZ
2S8
344
S16
549
733
1,099
3.7M
4,982
7,473

5Z-
-^^••"
374
499
747
795
1,060
1,590
3,344
7,126
10,689

10Z
489
652
978
1,040
1,386
2,079
6,909
9,212
13,818
Percentage of Time Log
Removal is Reduced bv 1*
OZ
5Z
CD 4.45xl06
62.8x10' 52.690
10.547
8.112
0.084
0.007
0.003
0.003
0.003
0.060
0.042
0.008
0.004
0.003
0.003
0.003
10Z
145 . 502
0.329
0.011
0.009
0.004
0.003
0.003
0.003
0.003
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
     ENHANCED SWTR DATA -- FIRST CUSTOMER --  SURFACE:   NO SOFTENING --  TTHMs-100
                             WITH ALTERNATE DISINFECTION
                              DECEMBER 16.  1991 VERSION

                                                 Percentage  of  Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OX
Mean 252
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 25X
50X
Secondary
Infection
Rate
OX
Mean 25X
SOX
OX
90th Xile 25X
SOX
OX
99th Xile 2SX
SOX
Removal

OX
0.134
0.179
0.268
0.342
0.455
0.683
0.702
0.936
1.404
is Reduced

SX
0.195
0.259
0.389
0.495
0.660
0.991
1.018
1.358
2.036
Percentage of Time
Removal is Reduced

OX
CD
00
00
00
OB
00
00
00
00

5X
00
00
CD
00
00
00
00
00
00
by 1*

10X
0.255
0.340
0.510
0.649
0.865
1.298
1.334
1.779
2.668
Log
by 1*

10X
00
00
00
00
00
00
00
00
00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SVTR DATA -- FIRST CUSTOMER -- SURFACE:   NO SOFTENING -- TTHMs-75
                         WITH ALTERNATE  DISINFECTION
                           DECEMBER  16, 1991 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25Z
502
OZ
90th Zile 25Z
50Z
OZ
99th Zile 2SZ
50Z
Secondary
Infection
Rate
OZ
Mean 2SZ
SOZ
OZ
90th Zile 25Z
SOZ
OZ
99th Zile 25Z
SOZ


0.
0.
0.
0.
0.
0.
0.
0.
1.

OZ
00
CD
CD
CO
CO
00
CO
CD
CD
Removal is Reduced

OZ 5Z
148 0.214
197 0.286
295 0.428
346 0.502
461 0.669
692 1.003
702 1.018
936 1.358
404 2.036
Percentage of Time
Removal is Reduced

52
CD
CD
CO
CO
CD
CO
00
CD
CD
bv 1*

10Z
0.281
0.374
0.561
0.657
0.877
1.315
1.334
1.779
2.668
Log
by 1*

10Z
CD
CD
CD
CD
CD
00
00
CD
CO
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SWTR DATA -- FIRST CUSTOMER --  SURFACE:   NO SOFTENING --  TTHMs-50
                         WITH ALTERNATE DISINFECTION
                          DECEMBER 16, 1991 VERSION

                                                 Percentage of Time  Log

Average Number
of Cases per
10,000 per Year







Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25Z
50Z
OZ
90th Zile 25Z
50Z
OZ
99th Zile 25Z
50Z
Secondary
Infection
Rate
OZ
Mean 25Z
50Z
OZ
90th Zile 25Z
50Z
OZ
99th Zile 25Z
50Z


0.
0.
0.
0.
0.
0.
0.
0.
1.
Removal

OZ
156
208
311
346
461
692
702
936
404
is


0.
0.
0.
0.
0.
1.
1.
1.
2.
Percentage
Removal is

OZ
00
00
00
00
GO
OB
CO
00
00











5Z
GO
00
00
00
00
a
00
00
00
Reduced

5Z
226
301
452
502
669
003
018
358
036
of Time
Reduced










by

1*
10Z
0.
0.
0.
0.
0.
1.
1.
1.
2.
Log
by

10Z
00
00
00
00
00
00
00
00
00
296
395
592
657
877
315
334
779
668
1*









^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------
  ENHANCED SWTR DATA -- FIRST CUSTOMER -• SURFACE:   NO SOFTENING -- TTHMs-25
                         WITH ALTERNATE  DISINFECTION
                          DECEMBER 16, 1991 VERSION

                                                 Percentage of Time Log

Average Number
of Cases per
10,000 per Year






Average Number
of Years for
an Outbreak






Secondary
Infection
Rate
OZ
Mean 25Z
50Z
OZ
90th Zile 2SZ
SOZ
OZ
99th Zile 25Z
SOZ
Secondary
Infection
Rate
OZ
Mean 25Z
SOZ
OZ
90th Zile 2SZ
SOZ
OZ
99th Zile 2SZ
SOZ


0.
0.
0.
0.
0.
0.
0.
0.
1.

OZ
09
CO
CD
OB
GO
09
00
00
OB
Removal is Reduced

OZ 5Z
164 0.237
218 0.316
327 0.474
350 0.508
467 0.677
701 1.016
702 1.018
936 1.358
404 2.036
Percentage of Time
Removal is Reduced

5Z
GO
00
00
00
00
00
00
CO
OB
bv 1*

10Z
0.311
0.414
0.622
0.666
0.888
1.331
1.334
1.779
2.668
Log
bv 1*

10Z
OB
00
00
00
00
00
00
00
00
^Statistical method where distribution mean equal to weighted average of
 resulting finished water levels.

-------