Toxicology 313 (2013) 174–184

Contents lists available at ScienceDirect

Toxicology
journal homepage: www.elsevier.com/locate/toxicol

Enhancing the chemical mixture methodology in emergency preparedness and

consequence assessment analysis
Xiao-Ying Yu a,∗ , Clifford S. Glantz a , Juan Yao a , Hua He a , Achille J. Petrocchi b , Douglas K. Craig c ,
John T. Ciolek d , Alexander E. Booth a
a
Pacific Northwest National Laboratory, Richland, WA 99352, USA
b
Petrocchi, LLC, Evergreen, CO 80439, USA
c
Advanced Technologies and Laboratories International, Inc., Germantown, MD 20874-1119, USA
d
AlphaTRAC, Inc., 10385 Westmoor Dr, Suite 310; Westminster, CO 80021-2597, USA

a r t i c l e i n f o a b s t r a c t

Article history: Emergency preparedness personnel at U.S. Department of Energy (DOE) facilities use the chemical mix-
Received 17 April 2012 ture methodology (CMM) to estimate the potential health impacts to workers and the public from the
Received in revised form 9 October 2012 unintended airborne release of chemical mixtures. The CMM uses a Hazard Index (HI) for each chemical
Accepted 28 October 2012
in a mixture to compare a chemical’s concentration at a receptor location to an appropriate concentration
Available online 23 November 2012
limit for that chemical. This limit is typically based on Protection Action Criteria (PAC) values developed
and published by the DOE. As a first cut, the CMM sums the HIs for all the chemicals in a mixture to
Keywords:
conservatively estimate their combined health impact. A cumulative HI > 1.0 represents a concentration
Chemical mixture methodology
Health code numbers
exceeding the concentration limit and indicates the potential for adverse health effects. Next, Health
Weighting factors Code Numbers (HCNs) are used to identify the target organ systems that may be impacted by exposure
Acute health effects to each chemical in a mixture. The sum of the HIs for the maximally impacted target organ system is
Exposure route used to provide a refined, though still conservative, estimate of the potential for adverse health effects
Protective action criteria from exposure to the chemical mixture. This paper explores approaches to enhance the effectiveness of
Priority ranking the CMM by using HCN weighting factors. A series of 24 case studies have been defined to evaluate both
the existing CMM and three new approaches for improving the CMM. The first approach uses a set of
HCN weighting factors that are applied based on the priority ranking of the HCNs for each chemical. The
second approach uses weighting factors based on the priority rankings of the HCNs established for a given
type of concentration limit. The third approach uses weighting factors that are based on the exposure
route used to derive PAC values and a priority ranking of the HCNs (the same ranking as used in the
second approach). Initial testing indicates that applying weighting factors increases the effectiveness of
the CMM in general, though care must be taken to avoid introducing non-conservative results. In the
near future, additional testing and analysis will be conducted that may lead to the adoption of one of the
tested approaches into the CMM.
© 2012 Elsevier Ireland Ltd. All rights reserved.

1. Introduction The challenge of addressing this issue was accepted by

When performing emergency management planning, it is a chal-
lenge to accurately predict the potential health effects to workers
and the public from an unplanned airborne release of a chemical
mixture. Planning for the potential release of a single chemical is
difficult enough, but for a release involving multiple chemicals, the
problem is further complicated by the need to consider additive,
synergistic, and antagonistic effects associated with exposure to
the chemical mixture.
∗ Corresponding author.
E-mail addresses: xiaoying.yu@pnnl.gov (X.-Y. Yu), cliff.glantz@pnnl.gov
(C.S. Glantz).
the American Conference of Governmental Industrial Hygienists
(ACGIH) in the early 1960s and the concept of the threshold and
threshold limit values (TLVs) were introduced for evaluating the
health effects from exposure to chemical mixtures to provide pro-
tection to nearly all workers (Pinto and Bennett, 1963; Stokinger,
1963). This approach assumed chemicals that act on the same tar-
get organ can be treated in an additive manner. In 1971, the ACGIH’s
additive mixture formula was adopted as part of the Occupational
Safety and Health Administration (OSHA) “permissible exposure
limits” (OSHA, 1971). In 2000, the U.S. Environmental Protection
Agency (EPA) issued supplementary guidance for conducting health
risk assessments for chemical mixtures (EPA, 2000). For mixtures
where there is no information on whether exposure to the mixtures
results in either additive or nonadditive effects, the EPA’s default

0300-483X/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tox.2012.10.011
 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184
approach is to combine the health effects of individual chemicals
in an additive manner. The National Academy of Sciences (NAS)
and the Safe Drinking Water Committee of the National Research
Council also made recommendations for a large number of pollut-
ants containing chemical mixtures (ATSDR, 2004 and references
therein).
Building upon this work and those of others as reviewed in
Monosson (2005), the U.S. Department of Energy (DOE) developed
their chemical mixture methodology (CMM) to support emergency
management and planning activities. The CMM is used by DOE, its
contractors, and other governmental and non-governmental orga-
nizations in the U. S. and abroad, to estimate the potential health
impacts to workers and the public from the accidental or malicious
airborne release of chemical mixtures. Developed and maintained
under the sponsorship of the DOE’s Office of Emergency Man-
agement and Policy (DOE/NA-41), use of the CMM for emergency
preparedness and response applications is recommended in DOE G
151.1-2 (DOE, 2007). The CMM is applied using a Microsoft Excel
workbook and includes data for over 3300 chemicals that may be
of potential concern to emergency planners.
Typically, the objective of CMM applications is to evaluate
whether an airborne chemical mixture might inhibit the ability of
exposed individuals to take effective protective actions. The CMM
computes a Hazard Index (HI) for each chemical in the mixture (i)
at a receptor location, as shown in Eq. (1):
Ci
HIi = (1)
Li
where Ci is the concentration of chemical “i” at the receptor (as typ-
ically computed using an appropriate chemical dispersion model)
and Li is the selected concentration limit. It is worth noting that
the definition of HI used in the CMM follows the general guideline
formula recommended by the EPA, that is, the HI is the exposure
level divided by the acceptable exposure level for a chemical.1
The DOE recommends using Protective Action Criteria (PAC)
values2 as airborne concentration limits. PAC values are based on
established exposure limit values including Acute Exposure Guide-
line Level (AEGL, Rusch et al., 2000, 2002), Emergency Response
Planning Guideline (ERPG, Rusch, 1993), and Temporary Emer-
gency Exposure Limit (TEEL, Craig et al., 1995, 1999) values. Three
different types of PACs exist. “PAC-1” is the lowest chemical con-
centration associated with mild, transient health effects. “PAC-2” is
the lowest chemical concentration associated with irreversible or
other serious health effects or an impaired ability to take protec-
tive actions. “PAC-3” is the lowest concentration associated with
life-threatening health effects. PAC-2 values are typically used for
most DOE emergency planning applications, since they preserve a
person’s ability to take protective actions as stated in DOE Order
151.1C (DOE, 2005).
If HI ≤ 1.0, the concentration of a chemical does not exceed its
concentration limit. If a PAC-2 value is used for the chemical’s con-
centration limit, a susceptible individual exposed to just that one
chemical should not experience any health effect that would impair
their ability to take effective protective actions. However, if HI > 1.0,
the chemical concentration would exceed the PAC-2 limit, meaning
a susceptible individual exposed to that chemical may experience
health effects that could impair their ability to take effective pro-
tective actions.
As a first approximation for chemical mixtures, the CMM com-
putes a cumulative HI at a given receptor location by summing the
1 5
Some sources use the term Hazard Quotient (HQ) instead of HI when assessing
individual chemicals.
2
The PAC data set, PAC searchable database, and technical information on the
PAC are available at http://www.atlintl.com/DOE/teels/teel.html.
175
associated HIs for all the individual chemicals in the mixture as in
Eq. (2):

n
HIi = HI1 + HI2 + · · · + HIn (2)
i=1

n
This cumulative HI, HIi , provides a simple and conserva-
i=1
tive estimate of health effects3 when the chemicals in the mixture
impact different body systems. If the cumulative HI > 1.0 for a chem-
ical mixture, the exposure limit would be exceeded in the event
of an atmospheric release and some kind of mitigating action (e.g.,
implementation of additional protective measures to reduce poten-
tial emissions) would need to be implemented to reduce the health
effects of exposures at the receptor location. The cumulative HI is
often overly conservative because it sums the health effects from
each chemical in the mixture whether or not the chemicals affect
the same or different target organ systems4 in the exposed individ-
ual.
To more accurately gauge health effects from an exposure to
chemical mixtures, the CMM also produces a refined estimate of
health effects by considering the target organ health code numbers
(HCNs) assigned to each chemical in the mixture. HCNs are similar
to medical diagnostic codes in that they are code numbers that
identify specific acute or chronic toxic effects involving individual
target organs (e.g., “acute kidney toxin” or “acute bone marrow
toxin”).5 The HCN-based approach in the CMM assumes that the
HIs from different chemicals are additive only if the chemicals have
HCNs that impact the same target organ effect categories (see Eq.
(3)).

n
HIi(p) = HI1(p) + HI2(p) + · · · + HIn(p) (3)
i=1
Note: “p” represents a target organ effect category.
Further, acute and chronic health effects to the same target
organ set are not considered additive because these impacts occur
over different time scales.
Currently, 60 different HCNs are available for characterizing the
potential target organ effects associated with exposure to a chem-
ical (see Table 1). This includes the addition of 16 new HCNs that
were introduced in the 2008 version of the CMM including 13 new
“acute effect” HCNs that were added to mirror the “chronic effect”
HCNs used in earlier versions of the CMM (Craig et al., 2011; Yu
et al., 2010).
Because the CMM’s HCN approach sums the HI’s only for the
chemicals that target the same or similar target organ systems, an
assumption inherent in this approach is that the impacts of chem-
icals on distinctly different target organs do not influence each
other. As a result, for some chemical mixtures, the use of the HCN-
based approach may produce a summed HI value for the maximum
impacted target organ system that is lower than the value derived
by computing the cumulative HI for all the chemicals in the mix-
ture. In this way, the HCN-based approach provides more realistic,
3
A conservative estimate of health effects involves calculating an HI value that is
high enough to avoid underestimating the potential additive impact of the chemicals
in the mixture.
4
A target organ system (e.g., respiratory system, cardiovascular system) is a set
of biological organs that work together to perform a particular biological function
that may be impacted by exposure to a chemical or chemical mixture.
The CMM assigns HCNs based on both mode of action (i.e., mode of toxicity) and
target organ effect, as discussed in Yu et al. (2010). For simplicity, this paper only
refers to the “target organ health effects” when referring to both mode of action and
target organ effects.
176 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184
Table 1
Health Code Numbers (HCNs) used to classify toxic effects by target organ. The HCNs but still conservative, guidance on the potential health effects from
are listed in the ranking order, from greatest to least impact for emergency prepared- exposure to various chemical mixtures. In recent years, the CMM’s
ness planning, adapted from Craig et al. (1999). Bolded HCNs are those added after
the 2007 version of the CMM.
HCN-based approach has been improved by improving the assign-
ment of HCN values for each of the approximately 3300 chemicals
Rank HCN Target Organ Effect in the CMM data set. More detailed information on how the CMM
1 17.00 Asphyxiants, anoxiants – acute effect works is presented in Yu et al. (2010) and in the supplementary
2 18.00 Explosive, flammable safety hazard (no material that accompanies this paper.
adverse effects with good housekeeping)
Some users of the CMM have reported that the existing HCN-
3 13.00 Blood toxin, methemoglobinemia – acute effect
4 6.00 Cholinesterase toxin – acute effect based approach may be overly conservative at times – not as overly
5 14.01 Eye irritant – severe conservative as computing the cumulative HI for all chemicals in
6 14.00 Severe irritant the mixture – but still more conservative than it needs to be. In
7 15.01 Eye irritant – moderate response, the CMM development team began investigating the per-
8 15.00 Moderate irritant
9 4.01 Eye – acute, other than irritation
formance of the CMM. This study presents the initial results of that
10 11.01 Respiratory irritant – acute severe or moderate investigation and describes the development and testing of sev-
but not mild irritant effects eral potential enhancements to the CMM’s HCN-based approach.
11 14.02 Skin irritant – severe These enhancements involve different ways to employ weighting
12 15.02 Skin irritant – moderate
factor schemes to reduce target organ effects that are unlikely to
13 4.00 Systemic toxin – acute short-term high hazard
effects be significant at the concentration limits of concern.
14 4.08 Heart, Cardiovascular system – acute effects
15 4.05 Brain – acute effects 2. Materials and methods
16 7.01 Central nervous system – acute effects
In this section we present the chemical mixtures used to support CMM testing
17 8.00 Narcotic – acute effect
and describe the different versions of the CMM that are used to examine the past
18 7.00 Nervous system toxin – acute effects
and current performance of the CMM and evaluate potential enhancements in the
19 11.00 Respiratory toxin – acute effects other than
CMM’s approach for calculating HCN-based HIs.
irritation
20 4.02 Nose – acute effects other than irritation
2.1. CMM test cases
21 4.09 Kidney – acute effects
22 4.06 Hematological effects – acute, unspecified
A series of 24 chemical mixtures are examined (summary information on these
23 4.04 Bone marrow – acute blood-forming system
chemical mixtures is presented in Tables 2a–2d). The first 12 of these chemical
and other acute effects
mixtures are based on the reported chemical inventory in laboratories located at
24 4.10 Liver – acute effects
the Pacific Northwest National Laboratory (PNNL). Each of these test mixtures con-
25 4.07 Gastrointestinal tract – acute effects
sists of seven to 15 chemicals. In each of these test cases, the total inventory of
26 4.03 Bladder – acute effects
each chemical mixture is assumed to be released to the atmosphere. This overly
27 4.13 Bone – acute effects
conservative assumption is made to bring the cumulative HI for most of the mix-
28 3.08 Heart, Cardiovascular system – chronic effects
tures into the range between 0.1 and 10. The concentration of each chemical at
29 1.00 OSHA carcinogen (29 CFR 1910.1000) – chronic
the designated receptor is determined through atmospheric dispersion modeling.
effect
EPICode (Homann, 2003), a chemical dispersion model approved by DOE for safety
30 1.01 Bladder carcinogen – chronic effect
assessments, is used in our testing.
31 1.02 Liver carcinogen – chronic effect
The second set of 12 chemical mixtures consists of a random selection6 of chem-
32 2.00 Suspect carcinogen or mutagen – chronic effect
icals from the CMM’s data set. Each of these 12 test cases involves six related pairs
33 2.01 Kidney carcinogen – chronic effect
of chemical mixtures. The odd-numbered test cases each consist of three randomly
34 2.02 Liver carcinogen – chronic effect
selected chemicals. Even numbered test cases feature up to six chemicals; three
35 3.05 Brain – chronic effects
chemicals from the preceding odd-numbered test case, plus two or three additional
36 7.11 Central nervous system – chronic effects
randomly selected chemicals. An identical amount of each chemical was assumed to
37 7.10 Nervous system toxin – chronic effects
be released to the atmosphere in this set of 12 test cases, resulting in a concentration
38 10.00 Respiratory toxin – chronic effects
of 3.5 mg/m3 for each chemical at the designated receptor.
39 9.00 Respiratory sensitizer – chronic effect
In this paper, we use the term “benefit” to describe the reduction in HIs that are
40 3.09 Kidney – chronic effects
achieved by applying the CMM using the HCN-based approach rather than simply
41 3.02 Hematological effects – chronic, unspecified
using cumulative HIs. Equation 4 indicates how this benefit is calculated for a given
42 3.04 Bone marrow – chronic blood-forming system
target organ effect.
and other chronic effects
43 3.10 Liver – chronic effects 
n

n

44 3.07 Gastrointestinal tract – chronic effects HIi − HIi(p)

45 3.01 Bladder – chronic effects
i=1 i=1
46 3.03 Bone – chronic effects Benefit(p) = × 100% (4)
47 3.06 Eye – chronic ocular effects 
n

48 12.00 Blood toxin, anemia – chronic effect HIi

49 5.00 Reproductive toxin – acute effects i=1
50 5.10 Reproductive toxin – chronic effects
The overall benefit of the HCN-based approach is determined by taking the
51 4.11 Skin – acute effects other than irritation
smallest benefit7 found in any of the target organ effect categories.
52 3.11 Skin – chronic effects including dermatitis and
sensitization
2.2. Testing objectives and the versions of the CMM used in testing
53 4.12 Skin perforation – acute effects other than skin
absorption
Testing of the CMM is divided into three studies. The first study focuses on
54 3.12 Skin perforation – nasal septum perforation
assessing the effect of updates to the PAC data set on cumulative HI values for the
and other chronic effects other than skin
test mixtures (the HCN-based approach is not used in this first study). The updates
absorption
55 3.00 Systemic toxin – chronic effects
56 16.01 Eye irritant – mild
57 16.00 Mild irritant 6
Each chemical common to all the CMM data sets used in our study was assigned
58 16.02 Skin irritant – mild a unique identification number and a computer program that generates random
59 19.00 Generally low risk health effects – nuisance numbers was then used to randomly select from these chemicals.
particles, vapors or gases 
n
60 20.00 Generally low risk health effects – odor 7
The smallest benefit is associated with the largest HIi(p) value in Eq. (4).
i=1
 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184 177

Table 2a
Chemicals that form the mixtures in the first six test cases.

No. Case 1 Case 2 Case 3 Case 4 Case 5 Case 6

1 Sodium glycolate Calcium chloride Silica, crystalline-quartz Cyclopropane Trichloroacetic acid Tributyl phosphate
dihydrate
2 Potassium cyanide Carbon Lithium bromide Iron(II) chloride Citric acid Phthalic acid
tetrahydrate
3 Dichloroethylene, Sodium oxalate Sodium chloride Ethylene oxide Agar Potassium phosphate,
cis-and trans-1,2- monobasic
4 Ethylene glycol Sodium Sodium phosphate, Acrylic acid polymers Boric acid Nickel sulfate hexahydrate
metabisulfite tribasic
5 Propionic acid Sodium bromide Sulfuric acid Iron Magnesium chloride Ammonium acetate
6 Phenanthrene Magnesium Hydrogen chloride Potassium chloride Potassium chloride Ethylenediaminetetraacetic
carbonate acid, disodium salt
7 Lactic acid Zinc acetate Phosphoric acid Sodium perrhenate Sodium carbonate Cupferron
8 Oxalic acid, dehydrate Ammonium Hexane Palladium Sodium phosphate, Sodium gluconate
thiocyanate tribasic
9 Potassium phosphate, Sodium phosphate, Phthalic acid Sodium hydroxide Ammonium formate
dibasic tribasic
10 Magnesium chloride Sulfur Mercury vapor Sodium bicarbonate Adipic acid
11 Chromic oxide Iron Lithium hydroxide
12 Potassium phosphate, Sodium sulfite Potassium ferricyanide
monobasic
13 Lead chromate Methylene chloride Manganese(II) chloride
(1:2)
14 Sodium carbonate Sodium bromide Molybdenum trioxide
15 Sodium sulfate Hydroxylamine chloride

Table 2b
Chemicals that form the mixtures in the test cases 7–12.

No. Case 7 Case 8 Case 9 Case 10 Case 11 Case 12

1 Barium hydroxide Silica, Strontium nitrate Dimethylacetamide, Sodium Ethyl hexanoic acid, 2-
crystalline-quartz n,n- phosphate,
tribasic
2 Zinc chloride Ascorbic acid Cesium chloride Pentane, n- Sodium Calcium carbide
chloride
3 Silica amorphous Perchloroethylene Ferric ammonium sulfate; Methyl alcohol Sodium nitrite Chlorodifluoromethane
hydrated Sulfuric acid, ammonium
iron(3e+) salt(2:1:1)
4 Sodium hydride Tris- Ferric chloride Copper(II) sulfate Potassium Hexane
hydroxymethylamino pentahydrate bromide
methane
5 Sulfur dioxide Calcium(II) nitrate Sodium bicarbonate Toluene Sodium nitrate Methyl alcohol
tetrahy drate
(1:2:4)
6 Sodium hydroxide Sodium sulfate, Gallium Dimethyl sulfoxide Calcium Acetonitrile
anhydrous chloride
dihydrate
7 Hydrazine hydrate Sodium carbonate Silver nitrate Isopropyl alcohol Potassium Isopropyl alcohol
nitrate
8 Sodium chloride Sodium chloride Acetonitrile Tetrasodium Ethylene glycol
pyrophosphate
9 Potassium chloride Potassium carbonate Methylene chloride Acetone
10 Sodium hydroxide Sulfur Chloroform Xylenes
11 Potassium Diphenyl Tetrahydrofuran Ethyl acetate
permanganate
12 Potassium Sodium acetate Pyridine
hydroxide
13 Ascorbic acid Tris- Ethyl ether
hydroxymethylaminomethane
14 Magnesium Ammonium, Dimethylformamide,
chloride hexadecyltrimethyl-, bromide N,N-
15 Manganese(II) Aluminum oxide Cyclohexane
chloride (1:2)

Table 2c
Chemicals that form the mixtures in test cases 13–18.

No. Case 13 Case 14 Case 15 Case 16 Case 17 Case 18

1 Trichloroacetic acid Trichloroacetic acid Butanenitrile Butanenitrile Bromochlorobenzene, p- Bromochlorobenzene, p-

2 Methylphenol, 4- Methylphenol, 4- Phenylene Phenylene diisocyanate, Ceric ammonium nitrate Ceric ammonium nitrate
diisocyanate, 1,4- 1,4-
3 Mercury hydroxide Mercury hydroxide Benzene Benzene hexachloride Sodium nickel oxide Sodium nickel oxide (Liquid)
hexachloride (Liquid)
4 Sodium nickelate Dichlorobenzene, p- Thorium oxide
5 Glyceryl monostearate Polychlorinated biphenyl Diethylene glycol diacetate
6 Methyl thiocyanate Barium sulfate
178 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184

in question are changes in the PAC-2 values for the tested chemicals. All 24 test cases

Trimethoxysilane
Zirconium silane
are investigated in this study. The two versions of the CMM that are employed in

Silicon(II) oxide
this study are:

Ferric nitrate
Bismuth
Case 24
• The 2007 version of the CMM using PAC Revision 21 concentration limit data. This
is the standard version of the CMM that was used in 2007.
• The 2011 version of the CMM using PAC Revision 26 concentration limit data from
20118 . This is the standard version of the CMM that was used in 2011.

Because cumulative HI values depend only on the concentration of the chemicals

Trimethoxysilane
and their concentration limit, only changes in the PAC-2 values between these two
versions of the CMM affect the calculation of cumulative HI values.
Ferric nitrate

The second study examines how changes in the characterization and assignment
Bismuth
Case 23

of HCN values for chemicals within the CMM data set affect HCN-based HI values for
the test mixtures. In this case, the same PAC values are used to eliminate the effect
caused by differences in PAC values between different revisions. All 24 test cases are
investigated in this study. The versions of the CMM utilized in this study are:

• The 2007 version of the CMM using PAC Revision 26 concentration limit data.
Tri(2-ethylhexyl) phosphate

• The 2011 version of the CMM using PAC Revision 26 concentration limit data.
Trimethylpyridine, 2,4,6-

Potassium chromate(VI)
Oxalic acid, anhydrous

Oxalic acid, anhydrous

By using the same PAC data set in both versions of the CMM used in this study,
the differences in the computed HCN-based HI results are due to enhancements in
the HCN categories and updates in the HCN assignments to individual chemicals
Chlorthiophos

that were made between 2007 and 2011.

The third study examines how applying different weighting factors to each of the
Case 22

HCN categories affects the performance of the CMM. Under the current HCN-based
approach, the same HI is applied to all of the applicable target organs that may be
impacted by exposure to the chemical under consideration. When using weighting
factors, the HI for each applicable target organ is multiplied by a numeric value
that could range from 1.0 (i.e., no change from the current method) to 0.1 (a 90%
reduction in the HI for that target organ). Weighting reduces the HI for target organs
Tri(2-ethylhexyl) phosphate

that are unlikely to be impacted at the concentration limit assigned to the chemical
Trimethylpyridine, 2,4,6-

and potentially focuses the analysis on target organs that are associated with health
Oxalic acid, anhydrous

effects that would occur at, or around the designated concentration limit.
The 2007 and 2011 versions of the CMM using PAC Rev 26 data are used
to provide the HCN-based HI values for this study. The new weighting factor
approaches tested in this study all use the 2011 version of the HCN and PAC Rev
26 data.
Case 21

2.3. Approaches that use weighting factors

The use of weighting factors is not a new concept in the CMM. Currently, weight-
ing factors of 0.5 and 0.25 are applied in the CMM for moderate and mild irritants
(i.e., HCNs of 15.00, 15.01 and 16.00, 16.02, respectively). Three weighting factor
Tris-hydroxymethylaminomethane

approaches are assessed in our third study. Only the top ten HCNs for each chem-
ical are evaluated in the CMM, HCNs outside the top ten are effectively assigned a
weighting factor of zero. These approaches extend the weighting factor concept to
a broader set of HCN categories.
Approach 1 applies a range of weighting factors to the top ten HCNs for each
Trimethyloctane, 2,4,6-

chemical in a mixture. The weighting factors are based on the priority ranking of the
Sodium cacodylate;
Pentachloroethane

HCNs assigned to each chemical (see Table 1) (Petrocchi et al., 2008; Yu et al., 2010).
Strychnine sulfate

The implementation of Approach 1 involves ranking the HCNs for each chemical
according to the priority order using Table 1. Next, weighting factors are applied
Hexanol, n-;

to each of the top 10 HCNs for a chemical. A number of different weighting factor
Case 20

schemes are candidates for use in Approach 1 (as listed in Table S-4), but only one
scheme is discussed in this paper. This scheme applies a 1.0 weighting factor to
the highest ranking HCN, 0.9 to the second highest ranking, 0.8 to the third highest
ranking, and so on to 0.1 for the tenth highest ranking HCN (see Tables S-7a–c for
a detailed example). This approach uses the existing CMM and PAC data sets and is
Chemicals that form the mixtures in test cases 19–24.

easy to implement because it involves just slight changes in the algorithms used to
Tris-hydroxymethylaminomethane

calculate HCN-based HI values.

Approach 2 is another easy-to-implement approach that applies weighting fac-
tors based on the ranking order of all potential HCNs. Table 3a and b provide two
such weighting schemes, “Alpha” (with percentile weighting factors) and “Beta”
(with stepwise weighting factors). Approach 2-Alpha uses Table 3a (which follows
the same ranking order as Table 1) to assign weighting factors to HCNs. Four different
Sodium cacodylate;
Pentachloroethane

weighting factors are used (1, 0.75, 0.5, and 0.25) with each of these four weight-
ing factors assigned to one quarter of the HCNs in the table. Approach 2-Beta uses
Table 3b to provide an alternative priority ranking of the HCNs based on their poten-
tial impact to an individual’s ability to take protective actions (these HCN rankings
Case 19

8
The PAC Revision 26 was published in the fall of 2010 and the CMM Revision
26 was published in the spring of 2011. When we refer to the “2011 version” of the
Table 2d

CMM we are referring to CMM Revision 26 with PAC Rev 26 data. PAC Revision 27
No.

and CMM Revision 27 were published in 2012; not in time to be used in the testing
1
2
3
4
5
6

reported in this paper.

 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184 179

Table 3
Weighting factors for Approaches 2-Alpha (left) and 2-Beta (right).

a. Approach 2-Alpha b. Approach 2-Beta

Rank HCNc WFd Rank HCN WF Rank HCN WF Rank HCN WF

1 17.00 1 31 1.02 0.5 1 17.00 1 31 10.00 0.4

2 18.00 1 32 2.00 0.5 2 18.00 1 32 9.00 0.4
3 13.00 1 33 2.01 0.5 3 11.01 1 33 7.11 0.4
4 6.00 1 34 2.02 0.5 4 11.00 1 34 7.10 0.4
5 14.01 1 35 3.05 0.5 5 7.01 1 35 12.00 0.4
6 14.00 1 36 7.11 0.5 6 7.00 1 36 3.01 0.4
7 15.01 0.5a 37 7.10 0.5 7 8.00 1 37 3.02 0.4
8 15.00 0.5a 38 10.00 0.5 8 14.01 1 38 3.03 0.4
9 4.01 1 39 9.00 0.5 9 4.08 1 39 3.04 0.4
10 11.01 1 40 3.09 0.5 10 4.05 1 40 3.05 0.4
11 14.02 1 41 3.02 0.5 11 4.01 1 41 3.06 0.4
12 15.02 0.5a 42 3.04 0.5 12 6.00 1 42 3.07 0.4
13 4.00 1 43 3.10 0.5 13 14.00 1 43 3.08 0.4
14 4.08 1 44 3.07 0.5 14 14.02 1 44 3.09 0.4
15 4.05 1 45 3.01 0.5 15 13.00 1 45 3.10 0.4
16 7.01 0.75 46 3.03 0.25 16 15.01 0.8b 46 3.00 0.4
17 8.00 0.75 47 3.06 0.25 17 15.00 0.8b 47 1.00 0.4
18 7.00 0.75 48 12.00 0.25 18 15.02 0.8b 48 1.01 0.4
19 11.00 0.75 49 5.00 0.25 19 4.00 0.8 49 1.02 0.4
20 4.02 0.75 50 5.10 0.25 20 4.02 0.8 50 2.00 0.4
21 4.09 0.75 51 4.11 0.25 21 4.03 0.8 51 2.01 0.4
22 4.06 0.75 52 3.11 0.25 22 4.06 0.8 52 2.02 0.4
23 4.04 0.75 53 4.12 0.25 23 4.07 0.8 53 16.01 0.2
24 4.10 0.75 54 3.12 0.25 24 4.04 0.6 54 16.00 0.2
25 4.07 0.75 55 3.00 0.25 25 4.09 0.6 55 16.02 0.2
26 4.03 0.75 56 16.01 0.25 26 4.10 0.6 56 5.10 0.2
27 4.13 0.75 57 16.00 0.25 27 4.11 0.6 57 3.11 0.2
28 3.08 0.75 58 16.02 0.25 28 4.12 0.6 58 3.12 0.2
29 1.00 0.75 59 19.00 0.25 29 4.13 0.6 59 19.00 0.1
30 1.01 0.75 60 20.00 0.25 30 5.00 0.6 60 20.00 0.1
a
Weighting factors for moderate irritation (15.xy) are set to 0.5 to mirror how these HCNs were weighted in the CMM.
b
Weighting factors for moderate irritation (15.xy) are set to 0.8 in this scheme to reflect an alternative perspective on their potential impact on an individual’s ability to
take protective actions.
c
HCN: Health code number.
d
WF: Weighting factor.

are intended to be used in conjunction with PAC-2 concentration limits). This prior-
ity ranking places an emphasis on acute effects, particularly those associated with
the nervous, cardiovascular, respiratory, and vision systems. Chronic effects receive
a comparatively lower ranking under this approach because the onset of these health
effects would occur only after an individual completes taking recommended protec-
tive actions. Under Approach 2-Beta, the weighing factors assigned are 1.0, 0.8, 0.6,
0.4, 0.2, and 0.1. If a given chemical does not have at least one HCN with a weighting
factor of 1.0, the chemical’s largest weighting factor is increased to 1.0. This ensures
that at least one or more HCNs are associated with the fully weighted HI value for
the chemical and this maintains conservatism within this approach.
Approach 3 assigns weighting factors to each chemical in a mixture based in
part upon the route of exposure used to calculate the PAC values for each chem-
ical. Approach 3 uses two sets of weighting factors instead of one. The first set of
weighting factors are based on the routes of exposure used in the toxicity studies
that are documented in the scientific literature for the chemical. Table 4a displays
the references and the corresponding priority of these references used in identifying
routes of exposure for each chemical. It is worth noting that the priority order for
using these references is almost identical to the order used to derive TEEL values
(DOE, 2008). If the relevant literature lists inhalation, skin contact, or eye contact
as a tested route of exposure, initial weighting factors of 1.0 are assigned to all
of the ten HCNs that may involve one of these routes of exposure. If other routes
of exposure are involved (e.g., oral, intravenous) the appropriate weighting factor
from Table 4b is assigned to the HCN. If multiple routes of exposure are indicated,
the highest applicable weighting factor is used. The second set of weighting fac-
tors used in this approach are the stepwise factors developed for Approach 2-Beta.
Approach 3 requires more data than what is currently collected in the existing HCN
development procedure (Petrocchi et al., 2008; Yu et al., 2010). This would make it
more time-consuming to implement Approach 3 than Approaches 1 or 2.
Table 5 provides an example of the weighting factors used by each of the new
HCN-based approaches for a chemical. Weighting factors are listed for each of the
HCNs associated with tri(2-ethylhexyl) phosphate (CASRN: 78-42-2), a chemical in
Test Case 22.
3. Results
Table 4a
Priority order of the literature used to identify exposure routes in Approach 3.
Order References
1 AEGLa : toxicity data
2 ERPGb : toxicity data
3 HSDBc : toxicity data
4 RTECSd : toxicity data
TLVs and BEIse : toxicity data
5 NIOSHf : toxicity data
6 CHRISg : symptoms
7 SAXh : safety profile
8 MSDSi : toxicity data
The following are brief descriptions of each literature resource:
a
AEGL: Acute Exposure Guideline Levels, National Academy of Sciences (NAS)
Documentation [NAS, 2000–2011].
b
ERPG: Emergency Response Planning Guidelines (ERPGs) Documentation,
American Industrial Hygiene Association (AIHA) [AIHA, 2001–2011].
c
HSDB: Hazardous Substances Data Bank (HSDB), National Library of Medicine.
d
RTECS: Registry of Toxic Effects of Chemical Substances (RTECS), National Insti-
tute for Occupational Safety and Health (NIOSH). In this table, HSDB is placed ahead
of RTECS because the former provides more summary data on human and animal
toxicity.
e
TLVs and BEIs: Threshold Limit Value (TLV) occupational exposure guidelines
and Biological Exposure Indices (BEIs) by the American Conference of Governmental
Industrial Hygienists (ACGIH® ) [ACGIH, 2001–2011].
f
NIOSH: Pocket Guide to Chemical Hazards and ICSC International Chemical
Safety Cards, NIOSH, National Institute for Occupational Safety and Health.
g
CHRIS: Chemical Hazards Response Information System (CHRIS), U.S. Coast
Guard.
h
SAX: Sax’s Dangerous Properties of Industrial Materials, 11th Edition (SAX) CD
ROM [SAX, 1998–2004].
i
MSDS: Sigma Aldrich Material Safety Data Sheets (MSDSs) and other references.

In this section we present the results of Studies 1, 2, and 3.

180 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184

Table 4b
Weighting factors used in the Approach 3 evaluation of studied exposure routes.
Exposure routes Weighting factor
Inhalation 1 test cases. The HCN-based approach should provide benefits only
Skin and/or eye contact 1 if there are two or more chemicals in a mixture that are significant
Oral 0.75
Other exposure routes but primary target organ 0.5
Other unspecified routes but not primary target organ 0.25
3.1. Study 1 — impact of updates to PAC values on cumulative HI
values
Changes in a concentration limit for a given chemical (i.e.,
changes in PAC values) can have a substantial impact on the HI value
calculated at a designated receptor, because the HI is inversely pro-
portional to the concentration limit. The introduction of new health
effects data may cause a substantial change in a chemical’s PAC
value from one data set revision to the next. This effect was assessed
in this study. Changes in the PAC derivation methodology can also
cause substantial changes in the PAC data set. However, significant
changes in the PAC derivation methodology do not exist in the data
sets used in our testing. A major change in the derivation method-
ology was made in 2012 and is incorporated in PAC Rev 27. The
impact on the CMM of these changes will be examined in future
work.
Table 6 presents summary results from this study. In nine of the
24 test cases, there is a >50% difference between the cumulative HI
values obtained using the 2007 and 2011 PAC data sets. Percent-
age differences for the test cases range from a 76% decrease to an
800% increase in the cumulative HI between 2007 and 2011. These
changes are due to changes in the concentration limit for one or
more chemicals in the mixture. This study illustrates the sensitiv-
ity of cumulative HI values to the concentration limits assigned to
the chemicals in the mixture.
3.2. Study 2 — the benefit of using HCN-based HI values over
simple cumulative HI values
The benefit from using the HCN-based approach, rather than the
overly conservative cumulative HI, was evaluated for each of the
24 case studies. This “benefit” is calculated using Equation 4. Fig. 1
presents the benefit obtained by using the HCN-based approach
using the 2007 and 2011 versions of the CMM, with both versions
using the PAC Rev 26 data set.
Using the 2007 version of the CMM, the benefit from using the
HCN-based approach is larger than 5% in 6 of the 24 test cases. Using
the 2011 version of the CMM, a benefit >5% occurs in only 4 of the 24
contributors to the cumulative HI and the chemicals do not have an
overlap in their HCN assignments. The addition of new acute HCN
categories and the more comprehensive assignment of HCNs to all
the chemicals for the 2011 CMM data set have reduced the number
of test cases showing an appreciable benefit. This is in part due to
the less comprehensive characterization in 2007 of the HCNs for
acute toxic effects.
While the application of the new HCN categories and data is
appropriate, it highlighted several questions being raised by CMM
users. Why is the CMM’s HCN-based approach providing signifi-
cant benefits for a relatively low percentage of chemical mixtures?
Is it being over-conservative by providing an equal weighting to
all target organ systems that a chemical may affect, even though
only a small number of target organs are likely to produce an
adverse health effect at the identified concentration limit (e.g., PAC-
2 value)?
As a result of these questions, the CMM development team has
started to investigate whether to focus the CMM on those target
organ systems that can produce pertinent health effects above the
specified concentration limit, and if so, how to do it while still
maintaining an appropriate level of conservatism.
3.3. Study 3 — testing of candidate approaches that use HCN
weighting factors
Approaches 1, 2-Alpha, and 2-Beta were tested using 12 test
mixtures and Approach 3 was tested using 6 test mixtures (the
smaller number of test cases for Approach 3 is owing to a limita-
tion on the time available to characterize the routes of exposure
for all the chemicals in each test mixture). We focus our presen-
tation of results on four case studies (i.e., Test Cases 18, 20, 22,
and 24; see Tables 2c and 2d for a list of the chemicals in these
test cases), because they provide a representative cross-section of
mixture characteristics and testing results.
In Test Case 18, one chemical in this six chemical mixture,
mercury hydroxide, accounts for over 93% of the cumulative HI
for the mixture. In a case where only one chemical in the mixture
overwhelmingly dominates the health impacts, the HCN-based
approach, either with or without weighting factors, provides little

Table 5
Weighting factors applied using Approaches 1 through 3 using tri(2-ethylhexyl) phosphate as an example.

Weighting factors for Tri(2-ethylhexyl) phosphate

HCN priority HCNsa Approach 1 generic Approach 2-alpha Approach 2-beta Approach 3
ranking weighting factor percentile weighting stepwise weighting
factor factor Exposure weighting Stepwise weighting
factor factor

1 15.01 0.5b 0.5 1.0c 1 × 0.8

2 15.02 0.5b 0.5 0.8 1 × 0.8
3 7.11 0.8 0.5 0.4 1 × 0.4
4 10.00 0.7 0.5 0.4 1 × 0.4
5 3.09 0.6 0.5 0.4 1 × 0.4
6 3.02 0.5 0.5 0.4 0.75 × 0.4
7 3.10 0.4 0.5 0.4 0.75 × 0.4
8 3.07 0.3 0.5 0.4 0.75 × 0.4
9 3.01 0.2 0.5 0.4 1 × 0.4
10 3.11 0.1 0.25 0.2 0.75 × 0.2
a
HCN is health code number.
b
In Approach 1, the weighting factors assigned to the HCNs ranked in the top two positions are 1.0 and 0.9, respectively. However, an exception is illustrated in this
example; both the current HCN-based approach and Approach 1 apply weighting factors of 0.5 to moderate irritants.
c
Under Approach 2-Beta, if a given chemical does not have at least one HCN with a weighting factor of 1.0, the chemical’s largest weighting factor is increased to 1.0. In
this case, the weighting factor is increased from 0.8 to 1.0.
 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184 181

Table 6
Cumulate HI Values Computed for the 24 Test Cases. Values are computed using the 2007 version of the CMM with PAC Rev 21 data (HI2007 ) and the 2011 Version of the CMM
with PAC Rev 26 data (HI2011 ).

Test case no. Cumulative HI2007 a Cumulative HI2011 b Performance ratioc Percentage differenced

1 7.0 2.8 2.6 −60%

2 2.1 1.7 1.3 −19%
3 40.9 22.9 1.8 −44%
4 4.3 4.2 1.0 −2%
5 1.8 2.1 1.2 17%
6 6.5 6.0 1.1 −8%
7 7.5 48.6 6.5 548%
8 1.8 1.7 1.0 −6%
9 2.0 1.4 1.5 −30%
10 1.9 4.1 1.0 116%
11 2.6 2.8 1.1 8%
12 6.4 6.3 1.0 −2%
13 30.2 30.2 1.0 0%
14 31.5 32.8 1.0 4%
15 1.0 1.0 1.0 0%
16 4.5 1.1 4.1 −76%
17 1.6 3.2 2.0 100%
18 1.6 3.2 2.0 100%
19 0.1 0.9 8.9 800%
20 0.9 1.7 1.9 89%
21 0.3 0.4 1.3 33%
22 1.3 1.1 1.2 −15%
23 0.6 1.0 1.7 67%
24 0.9 1.2 1.4 33%

Avge 6.6 7.6 1.9 15%

a
The Cumulative hazard index (HI) computed using the CMM with PAC Rev 21 from 2007.
b
The Cumulative HI computed using the CMM with PAC Rev 26 from 2011.
c
The performance ratio for each test case is the greater of HI2011 /HI2007 and HI2007 /HI2011 .
d
The percentage difference is calculated as (HI2011 − HI2007 )/HI2007 × 100%.
e
Average is abbreviated as avg.

if any benefit over using the cumulative HI value for the mixture.
Fig. 2 shows that as expected, none of the tested approaches
provides any substantial benefit in this test case.
In Test Case 20, two of the six chemicals in the mixture,
strychnine sulfate and sodium cacodylate (also called sodium
dimethylarsinate), account for over 95% of the cumulative HI for the
mixture. Because these two dominant chemicals share the HCN for
acute cardiovascular system effects (“4.08”), the application of any
non-weighted HCN-based approach should provide little or no ben-
efit over using the cumulative HI value (see Table S-9a for all of the
HCN assignments for the chemicals in this test case). Fig. 3 displays
the benefit obtained using various HCN-based approaches for Test
Case 20, and as predicted, the non-weighted approaches show neg-
ligible benefits. Comparing the weighted HCN-based approaches,
Approach 1 produces a benefit of just over 15%. This benefit is due
to the HCN for acute cardiovascular effects being ranked second
and third, respectively, for the two dominant chemicals in the mix-
ture. The associated 0.9 and 0.8 weighting factors assigned to this
HCN produces a net 15% benefit for the mixture. Neither Approach
2-Alpha or 2-Beta provides any substantial benefit because of the

Fig. 1. Benefits from using the HCN-based approach for all 24 test cases.
182 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184
Fig. 2. Comparison of benefits between weighted and non-weighted results for test
case 18.
Fig. 3. Comparison of benefits between weighted and non-weighted results for test
case 20.
unit weighting factor (i.e., “1.0”) assigned to acute cardiovascular
effects under the percentile and stepwise approaches for the two
dominant chemicals. The benefit from using Approach 3 is not sig-
nificant because of the unit weighting factors assigned to both the
exposure route and stepwise weighting for acute cardiovascular
effects for the two dominant chemicals.
In Test Case 22, three of the five chemicals in the mixture,
tri(2-ethylhexyl) phosphate, chlorthiophos, and potassium chro-
mate (VI) account for over 90% of the cumulative HI value (see
Table S-8). The only overlap in HCN values between these three
chemicals is in the areas of eye irritation and nervous system
effects. As shown in Fig. 4, the HCN-based HIs associated with
Fig. 4. Comparison of benefits between weighted and non-weighted results for test
case 22.
Fig. 5. Comparison of benefits between weighted and non-weighted results for test
case 24.
the non-weighted versions of the CMM show no significant ben-
efit over the cumulative HI. The nervous system is the dominant
target organ system for health effects using the non-weighted
versions of the CMM. In contrast, all of the weighted HCN-based
approaches provide benefits. Approach 1 provides the highest ben-
efit value (34%), with Approaches 2-Alpha, 2-Beta, and 3 providing
a benefit of 25, 8, and 16%, respectively. One common factor con-
tributing to these benefits is the application of weighting factors
to reduce the emphasis placed on chronic nervous system effects
when calculating HCN-based HIs. The weighting factor approaches
all account for this important consideration, the non-weighted
approaches do not. As a result, the eye is the dominant target
organ under Approaches 1, 2-Beta, and 3 (owing to severe and
moderate eye irritation) while the respiratory system is the dom-
inant target organ system under Approach 2-Alpha. The higher
benefits for Approaches 1 and 2-Alpha, compared to the other
weighted HCN-based approaches, are linked to the low weight-
ing factor of 0.5 currently assigned to moderate eye irritation in
Table 3a. The higher benefit for Approach 3 compared to Approach
2-Beta is linked to the use of an oral exposure route, with a
0.75 weighting factor, used to develop the concentration limit for
chlorthiophos.
In Test Case 24, four of the six chemicals in the mixture,
trimethoxysilane, zirconium silane, bismuth, and silicon oxide
account for 68, 18, 9, and 6% of the cumulative HI value (see Table
S-9a). In this mixture, there is no overlap in any HCN category
for all of the key chemicals. As shown in Fig. 5, the HCN-based
approaches using 2007 and 2011 versions of the CMM both pro-
duce benefits of about 7%. Approaches 1, 2-Alpha, 2-Beta, and 3
produce benefits of 17, 24, 7, and 7%, respectively. For all the
approaches, the respiratory system is the driving target organ
system for health effects. The higher benefit for Approach 1 is
owing to the placement of moderate eye irritation ahead of res-
piratory health effects in the ranking of weighting factors provided
in Table 3a. The ranking provided in Table 3b appears more appro-
priate for emergency response situations than those presented in
Table 3a, because the former assigned higher weighting factors for
acute toxic effects such as irritants (e.g., weighting factor for HCN
15.01 is 0.8 instead of 0.5) and lower weighting factors to chronic
toxic effects (e.g., weighting factor for HCN 3.08 is 0.4 instead of
0.75).
4. Discussion
In this section we discuss the differences and similarities among
the HCN-based weighting approaches, the benefits of using the new
approaches, and the direction of future work.
 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184
4.1. Differences and similarities among the weighting factor
Approaches
A key difference between Approaches 1, 2, and 3 is that Approach
1 only assigns the top weighting factor (i.e., 1.0) to its highest ranked
HCN value. In Approach 2-Alpha, a chemical that does not have any
high ranking HCNs may have its top HCN assigned a weighting fac-
tor that is <1.0. This may provide non-conservative results in some
instances, because it is assumed that the concentration limit for a
chemical is based on the health effects associated with at least one
HCN category. The other approaches avoid this problem by always
(or almost always) assigning a chemical’s highest ranking HCN a
weighting factor of 1.0, even if a lower value is specified in the
ranking tables.
An interesting feature of Approaches 2 and 3 is that each HCN
is assigned the same weighting factor for all the chemicals in the
mixture (except for instances where the highest weighting factor in
Approach 2-Beta or 3 is increased to 1.0), regardless of the ranking
order of the HCNs assigned to each chemical. In contrast, chemicals
that share the same HCNs may find that these common HCNs are
assigned quite different weighting factors in Approach 1.
Because Approach 3 uses the Approach 2-Beta weighting fac-
tors in addition to its exposure route factors, Approach 2-Beta will
always be as conservative as, or more conservative in its HI projec-
tions than Approach 3.
4.2. Benefits obtained from using the new approaches
The use of weighting factors in Approaches 1 and 2 often provide
an increase in the benefit of using the HCN-based approach, rela-
tive to the benefit derived from using the current non-weighted
approach. An increase in benefit is seen in about 90% of the exam-
ined test cases using Approach 1, 33% of the examined test cases
using Approach 2-Alpha, and 17% of the examined test cases using
Approach 2-Beta.
The high number of test cases with significant benefits under
Approach 1 is easy to understand. In Approach 1, only one HCN
category is assigned a weighting factor of “1.0” and all the HCN
categories are assigned somewhat lower weighting factors. As a
result, the HI contribution from important health effect contrib-
utors may be reduced by 10% or more. As a result, there may be
instances where the frequent application of lower weighting factors
may undervalue the results in certain HCN categories and therefore
underestimate the HIs for the target organ effect. This potential
under-conservatism may be eliminated through slight modifica-
tions to this approach. For example, this might involve assigning
a weighting factor of 1.0 to more of the higher ranking HCN cate-
gories. A preliminary examination of this sort of modified version
of Approach 1 hints that this may produce results comparable to
Approaches 2-Alpha and -Beta.
Approach 2-Alpha also appears to provide under-conservative
results in some test cases. It seems to undervalue some poten-
tially important HCNs. Approach 2-Beta is the most promising of
the easy-to-implement enhanced approaches to the CMM tested
in this study, although the numerical values of the benefits are not
the highest. In searching for a way to improve the CMM, one of
the most important factors is to maintain suitable conservatism.
Thus, we are not looking for an approach that provides the high-
est numerical values in terms of the benefit, but one that provides
some benefit when appropriate while maintaining a level of conser-
vatism that will safeguard workers and the public in an emergency
response situation.
Approach 2-Beta provides an enhanced benefit over the non-
weighted version of the CMM in cases where such a benefit appears
to be warranted. In cases where a benefit does not seem to be
warranted, Approach 2-Beta provides nearly identical results as
183
the non-weighted version of the CMM. If further testing shows
that Approach 2-Beta continues to provide appropriate and tech-
nically defensible results, while still maintaining an appropriate
level of conservatism, it may prove to be an approach worthy of
implementation in the CMM in the near future.
Further testing of Approach 3 is underway. Its incorporation
of Approach 2-Beta’s weighting factors looks promising, but the
impact of its additional weighting factor based on the exposure
routes used in published laboratory testing needs further assess-
ment. Although the notion to use the route of exposure to weight
HCN values is an interesting concept, it might not be conservative
under some circumstances. For example, our experience shows oral
exposure routes are commonly found in the health effects litera-
ture and are associated with derived concentration limits. In some
cases, oral exposure studies appear to be used as surrogates for
more resource-intensive inhalation studies. In such instances, the
use of a weighting factor (0.75) for the oral exposure route may
not be appropriate because the inhalation route of exposure (with
a weighting factor of 1.0) may be applicable even if is not tested in
the laboratory.
Another interesting observation is the high incidence in which
HCNs for irritation are assigned to chemicals. This may be because
rabbit eye exposures are often the first tests carried out with a new
chemical. Additionally, irritation effects are often described in a
qualitative manner (i.e., severe, moderate, or mild) rather than by
quantitative data. Thus ranking HCNs according to the significance
of the effect for a particular chemical and then assigning a weight-
ing factor to each HCN based on its significance in an emergency
appears to be an appealing and simple solution.
4.3. Future work
All of the candidate approaches assessed in this paper use
weighting factors to reduce the HI for target organ systems that may
not be associated with health effects at a chemical’s concentration
limit. A potential limitation of these approaches is that they use
weighting factors that are based on a relative ranking of the HCNs
and are not based on the specific observed health effects reported
in the scientific literature. For future consideration, an alternate
approach, the Primary/Secondary/Tertiary (PST) approach, is pro-
posed. It involves identifying the effects on key target organs
observed in health effects studies for individual chemicals. The
HCNs associated with these reported health effects applicable for
determining a concentration limit would be considered primary
HCNs and be given a weighting factor of “1.0”. The HCNs not asso-
ciated with observed health effects at those concentrations would
be considered secondary or tertiary HCNs and assigned appropri-
ately lower weighting factors. It is assumed that the secondary
and tertiary HCNs would be associated with health effects at con-
centrations greater, or perhaps much greater, than the applicable
concentration limit. Implementing this approach would require a
careful review of the literature used to appropriately characterize
the HCN values for each of the over 3300 chemicals in the CMM
data set. An initial scoping examination of this new approach is
proposed to determine the feasibility and resource requirements
associated with its further investigation.
5. Conclusion
Enhancements to the CMM’s HCN-based approach can provide
more accurate estimates of health effects that can be used to make
appropriate and cost effective emergency planning decisions. Using
diverse sets of chemical mixtures, the benefits and drawbacks of
several easy-to-implement approaches for enhancing the CMM
were studied. Each of these new approaches explored different
184 X.-Y. Yu et al. / Toxicology 313 (2013) 174–184
ways of implementing HCN weighting factors to achieve a balance
between an enhanced benefit and an appropriate level of conser-
vatism. Approach 2-Beta, employing a priority ranking of the HCNs
based on their potential impact to an individual’s ability to take
protective actions, seemed to provide the most appropriate bal-
ance during initial testing and it warrants further study. Approach
3, incorporating exposure route information, underwent limited
testing and it also warrants further study. Our short term objec-
tive is to conduct additional testing and make a recommendation
regarding the enhancements that warrant incorporation into the
CMM. A more comprehensive and technically rigorous approach
for improving the CMM, such as one that uses the specific target
organ effects reported in published studies, is a potential long term
goal.
Conflict of interest
No conflict of interest.
Acknowledgments
This work would not be possible without the support
and encouragement of James Fairobent, David Freshwater, and
other members of the Department of Energy Office of Emer-
gency Management and Policy. Additional thanks go to our
fellow members of the expanded TEEL Advisory Group, includ-
ing Jayne-Anne Bond, Po-Yung Lu, Tom Tuccinardi, and Eva
Hickey. The content and conclusions expressed in this article
are solely those of the authors and do not necessarily reflect
the views of DOE or their employers. PNNL is operated by Bat-
telle for the U.S. Department of Energy (DOE) under Contract
DE-AC05-76RL01830.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.tox.2012.10.011.
References
ATSDR (U.S. Agency for Toxic Substances and Disease Registry). 2004, Guidance man-
ual for the assessment of joint toxic action of chemical mixtures, U.S. Department
of Health and Human Services, Agency for Toxic Substances and Disease Registry,
Division of Toxicology, http://www.atsdr.cdc.gov/interactionprofiles/ipga.html,
(last accessed: 09.12).
Craig, D.K., Davis, J.S., DeVore, R., Hansen, D.J., Petrocchi, A.J., Powell, T.J., 1995. Alter-
native guideline limits for chemicals without environmental response planning
guidelines. Am. Ind. Hyg. Assoc. J. 56, 919–925.
Craig, D.K., Baskett, R.L., Davis, J.S., Dukes, L., Hansen, D.J., Petrocchi, A.J., Powell, T.J.,
Sutherland, P.J., Tuccinardi Jr., T.E., 1999. Recommended default methodology
for analysis of airborne exposures to mixtures of chemicals in emergencies. Appl.
Occup. Environ. Hygiene 14 (9), 609–617.
Craig, D.K., Petrocchi, A.J., Glantz, C.S., Yu, X.-Y. Bond, J.-A., Bouslaugh,
P.R., He, H., Schutte, K.A. 2011. The User’s Guide for the Chemical
Mixture Methodology (Compatible with PAC Rev. 26, February 2011).
PNNL-SA-76932, PNNL, Richland, WA. Available at: http://orise.orau.gov/
emi/scapa/chem-mixture-methodolgy/default.htm
DOE (U.S. Department of Energy). 2005. DOE O 151.1C. Comprehensive emergency
management system November 2, 2005 ed. p 90. Available at: https://www.
directives.doe.gov/directives/current-directives/151.1-BOrder-c/view
DOE (U.S. Department of Energy). 2007. DOE G 151.-2. Technical planning basis,
Emergency management guide, July 11, 2007, ed. p 90. Avaiable at: http://doe.
test.doxcelerate.com/directives/current-directives/151.1-EGuide-2/view
DOE (U.S. Department of Energy). 2008. DOE Handbook Temporary Emer-
gency Exposure Limits for Chemicals: Methods and Practice (DOE-HDBK-
1046-2008) August 2008 ed. p. 52. Available at: http://orise.orau.gov/
emi/scapa/chem-pacs-teels/reference.htm
EPA (U.S. Environmental Protection Agency). 2000, Supplementary guidance for
conducting health risk assessment of chemical mixtures, EPA/630/R-00/002,
www.epa.gov/ncea/pdfs/chem mix/chem mix 08 2001.pdf, (last accessed:
09.12).
Homann, S. 2003. EPIcode. Version 7.0. Available at: http://www.hss.energy.gov/
nuclearsafety/qa/sqa/central registry/EPIcode/EPI.htm
Monosson, E., 2005. Chemical mixtures: considering evolution of tox-
icology and chemical assessment. Environ. Health Perspect. 113
(4), 383–390, Also available at: http://www.ncbi.nlm.nih.gov/pmc/
articles/PMC1278475/#b68-ehp0113-000383
OSHA (Occupational Safety and Health Administration). 1971. Occupational safety
and health act. Standards: Permissible exposure limits for air contaminants. 29
CRF 1910.1000.
Petrocchi, A.J, Craig, D.K., Bond, J.-A., Trott, D.M., Yu, X.-Y. 2008. Health Code Number
(HCN) development procedure. URS-WD Report 2008–2009.03-A.
Pinto, S.S., Bennett, B.M., 1963. Threshold limit values for 1963. Arch. Environ. Health
7, 592–599.
Rusch, G.M., Garrett, R., Tobin, P., Falke, E., Lu, P.Y., 2000. The development of acute
exposure guideline levels for hazardous substances. Process Saf. Prog. 19 (2),
98–102.
Rusch, G.M., Garrett, R., Tobin, P., Falke, E., Lu, P.Y., 2002. The development of acute
exposure guideline levels for hazardous substances. Drug Chem. Toxicol. 25 (4),
339–348.
Rusch, G.M., 1993. The history and development of emergency response planning
guidelines. J. Hazard. Mater. 33 (2), 193–202.
Stokinger, H.E., 1963. International threshold limits values – 1963. Am. Ind. Hyg.
Assoc. J. 24 (5), 469–474.
Yu, X.-Y., Petrocchi, A.J., Craig, D.K., Glantz, C.S., Trott, D.M., Ciolek, J.T., Lu, P.-Y., Bond,
J.A., Tuccinardi Jr., T.E., Bouslaugh, P.R., 2010. The development and application
of the Chemical Mixture Methodology in analysis of potential health impacts
from airborne release in emergencies. J. Appl. Toxicol. 30 (6), 513–524.