MEETING REPORTOpen Access
Key factors influencing canine heartworm,
Dirofilaria immitis, in the United States
Heidi E Brown1, Laura C Harrington2, Phillip E Kaufman3, Tanja McKay4, Dwight D Bowman5*, C Thomas Nelson6,
Dongmei Wang7and Robert Lund7
An examination of the Companion Animal Parasite Council’s (CAPC) canine heartworm data to clarify the spatial
prevalence of heartworm in the United States. Factors thought to influence the spatial risk of disease, as identified
in a recent CAPC workshop, are discussed.
Keywords: Canine heartworm, Dirofilaria immitis, Mosquito vectors, Spatial prevalence
The Companion Animal Parasite Council (CAPC) col-
lected results of 4,769,403 canine heartworm tests dur-
ing the 2011 calendar year from various commercial
testing laboratories in the United States (US), and of this
national sampling of dogs, 56,612 (1.187%) were positive
for heartworm antigen, indicative of an active D. immitis
infection . This rich data set can be used to infer the
national prevalence of heartworm disease and/or verify
the accuracy of predictions made from forecasting mod-
els. Herein, we: 1) present a map of positive heartworm
test prevalence, thereby upgrading existing knowledge;
and 2) construct a list of factors that will be used to
explain the observed rates of heartworm positive tests
over the coming years of similar data collection.
This work stems from a meeting held in Atlanta, GA,
on June 9–10, 2012, during which vector ecologists,
entomologists, and biologists worked with a team of sta-
tisticians to identify risk factors which could be useful
for the development of spatial risk mapping for import-
ant vector-borne canine diseases for which CAPC had
access to collected data with the overarching objective
being to identify the most important factors influencing
Lyme, ehrlichiosis, anaplasmosis, and heartworm infec-
tion rates in the US canine population . The focus of
this paper is on one of these data sets, i.e., the data rela-
tive to canine heartworm (Dirofilaria immitis) infection.
Heartworm infections are a significant health risk to
dogs as even light infections are capable of producing
profound pulmonary vascular and parenchymal disease.
Despite improved diagnostic methods, effective preven-
tives and increasing awareness among veterinary profes-
sionals and pet owners, cases of heartworm infection
continue to be diagnosed in high numbers and are
becoming more prevalent in areas previously considered
to be at a low risk . A survey of veterinary clinics in
2005 reported that over 250,000 dogs tested positive for
heartworms during the 2004 calendar year . When
one considers a 48% response rate to the survey and the
fact that only 30% of the dog population in the US was
tested, the actual numbers of dogs infected are much
higher, probably in the 1 to 1.5 million range.
The test used for antigen detection in dogs has a high
sensitivity (84%, n= 175/208) and high specificity (97%,
n= 30/31) . While concerns about false positives in
areas of low prevalence exist, comparisons with known
prevalence rates and studies that have examined dogs
for the presence of microfilariae indicate that any over-
estimation of the infection rates due to false positives is
not large. A survey for microfilarial presence rather than
antigenemia conducted in Colorado found an overall
prevalence of microfilarial positive dogs in 1990 to be
0.77% for 7,818 dogs tested ; the 2008 data showed a
prevalence in Colorado of 0.4% . In 1981–1982, a sur-
vey of 541 dogs in 12 cities and four counties in North-
ern California found that 31 (5.7%) were positive for
heartworm microfilariae ; the prevalence rates in
these counties by microfilariae were similar to results
from the antigen survey .
* Correspondence: firstname.lastname@example.org
5Department of Microbiology and Immunology, College of Veterinary
Medicine, Cornell University, Ithaca, NY 14853, USA
Full list of author information is available at the end of the article
© 2012 Brown et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Brown et al. Parasites & Vectors 2012, 5:245
An example of the information that the CAPC heart-
worm database contains is represented by a map display-
ing spatially smoothed heartworm positive tests for the
2011 calendar year (Figure 1). The graphic shows that
heartworm is most prominent in the lower Mississippi
River Valley and nearly absent in Northern Montana.
This map was generated using annualized data — no
seasonal features were considered. The results represent
the prevalence of positive tests for circulating heartworm
antigen over a full year amongst a population of dogs
that are visiting a veterinarian and receiving care that
includes at the least a test for heartworm infection.
The proportion of positive heartworm tests was com-
puted for each county in the contiguous 48 states. Pro-
portions are analyzed in preference to total positive test
counts as the number of positive tests can be influenced
by local testing practices. These proportions were then
spatially smoothed via a procedure known as the head-
banging algorithm and grouped into nine probabilistic
categories. The methods account for the varying number
of tests made in different counties; for example, ten
positive tests in a sample of size one hundred statistically
suggests more of a problem than one positive test in ten,
although the sample infection proportions are the same.
The map will be updated in future years as additional
data become available. While this map quantifies the
baseline heartworm infection rates for US counties, it is
desirable to understand what factors explain heartworm
risk. Our immediate goal was to assemble a list of fac-
tors, which are available and quantifiable, that may influ-
ence canine heartworm infection rates.
At the June 2012 CAPC meeting, a team of experts
provided up to ten measurable factors, ranked in order
of importance that are most likely to affect heartworm
prevalence and can be used for spatial risk mapping. It
was understood that some of the factors might prove to
be of no value, that some could be of significant value,
that there might be interactions between factors, and
that some important factors could be omitted or non-
measurable. Because heartworm risk varies both spatially
and temporally, the team included factors which could
predict both baseline prevalence rates and inter-annual
variations. Before leaving the meeting, a ranked list was
generated (Table 1).
Figure 1 Spatially smoothed proportions of positive canine heartworm-antigen tests recorded by US veterinarians in 2011. The figure
summarizes 4,769,403 tests performed by veterinarians for circulating heartworm antigen in the US in 2011; of these tests, 56,612 tests (1.187%)
were positive. The population studied is those dogs that are seen by a veterinarian and are tested; about 5% of owned dogs in the US. The map
displays probabilities of a positive test after smoothing by the head-banging algorithm (see text for details). The figure is made by assigning the
smoothed proportions to nine color-coded categories [0.00,0.03], (0.03,0.06], . . .,(0.21,0.24], and (0.24, 1.00]. The colors range from dark green for
the lower proportions to bright red for higher proportions.
Brown et al. Parasites & Vectors 2012, 5:245
Page 2 of 9
Factors and their selection rationale
The factors selected by the heartworm working group
are listed and discussed here, grouped by vector, para-
site, and host factors. Some factors, such as temperature,
are important to multiple factor groups (vector and
parasite). Two criteria in selecting factors were that
1) the factor can be easily measured and 2) data for it
Dog heartworm is a mosquito-transmitted disease with a
cosmopolitan distribution [7,8]. The mosquito becomes
infected when it ingests microfilaria during the act of
blood feeding on an infectious host. To be a competent
vector of D. immitis, mosquitoes must be able to sup-
port nematode development into the infective stage (L3),
and the infective larvae must be able to migrate to the
proboscis of the mosquito . Over 60 species of mos-
quitoes are capable of supporting the development of L3
D. immitis . Summaries of mosquito species naturally
infected with D. immitis in 19 states have appeared in
previous publications [10-12]. In the US, multiple stud-
ies on potential heartworm vectors have been conducted;
however, most studies were restricted to certain loca-
tions. Without known vector distribution data from the
majority of states, estimates are needed to identify areas
where each vector occurs and to what abundance. This
distributional data can then be used to identify areas
with greater risk for heartworm transmission. With a
number of mosquito species likely playing different roles
across different regions of the US  and taking into
account mosquito vector competence data either obtained
naturally and/or experimentally, nine species were identi-
fied as major potential vectors for national forecast model-
ing. These species are Aedes aegypti, Aedes albopictus,
Aedes canadensis, Aedes sierrensis, Aedes trivitattus, Aedes
vexans, Anopheles punctipennis, Anopheles quadrimacu-
latus, and Culex quinquefasciatus.
For most of the nine mosquito species chosen, pub-
lished maps illustrating their geographical home ranges
within North America exist . As many of the mos-
quito species have varying frequency distributions within
even small ranges, the historical literature can serve as
a starting point for fine-tuning the representation of
their ranges. For species, such as Ae. albopictus and
Ae. aegypti, where competition between the species has
led to changes in their historical ranges [13,14], new
geographical distribution data can be made available as
it is accumulated.
Additional vector maps can be generated using sur-
veillance data and habitat modeling, but quality sur-
veillance data are notoriously difficult to acquire.
Consequently, there is a need to use surrogate factors.
Possible surrogates for actual presence/absence or abun-
dance maps of these mosquito species include: 1) vegeta-
tion indices (derived from satellite imagery or classified
land cover maps); 2) urbanization as it relates to rural,
suburban, and urban landscapes; and 3) meteorological
data to capture inter-annual variability. As new mos-
quito distribution maps are generated, they will ideally
be examined by entomologists to verify that they repre-
sent current species ranges.
A synopsis of various mosquito species that breed in
different habitats can be made available and overlaid
on land cover, cropland, and irrigation maps of the US.
For example, previous studies that have focused on
(Ae. sierrensis and An. punctipennis) , rice produc-
tion (An. quadrimaculatus) , and irrigation could
provide data for map generation. These data can be
included both as dominant type of vegetation/cropland
or as a percentage of each type by county.
Rural, suburban, and urban landscapes
Certain mosquito species are more prevalent in urban
environments due to differences in their preference for
particular oviposition sites. Mosquito species that breed
in artificial containers tend to be more urbanized, such
as Cx. quinquefasciatus and Ae. aegypti. The spatial and
temporal distribution of Cx. quinquefasciatus in residen-
tial areas  and underground storm drains has been
Table 1 Ranked factors initially identified by the working
1 Vector presence (+)
2 HDU (calculated from weather with a 14°C threshold) (±)
3 Lagged weather (7 months to 24 months prior to diagnosis)
a. Precipitation, temperature, and or relative humidity (±)
b. Moisture Index (Precipitation and Evaporation) (±)
4 Presence of coyote and feral dog populations – categorical data (+)
5 Human population density (+)
6 Land cover (Dominant type or Percentage classification) (±)
7 Cropland (Dominant type or Percentage classification) (±)
8 Social economic status (±)
a. Household income (-)
b. Education (-)
c. Foreclosure rates (+)
9 Irrigation (±)
These were the nine factors the working group thought merited testing for
inclusion in the model for the risk of heartworm infection. The factors were
ranked by group consensus based on which would most likely contribute
significantly to the final model. For some factors, e.g., weather, subfactors
were also identified. Finally, for each factor a hypothesis was given as to
where that factor would have a positive (+), negative (−), or unknown (±)
association with heartworm risk.
Brown et al. Parasites & Vectors 2012, 5:245
Page 3 of 9
well documented [18,19], and this type of information
may be useful for modeling. Human population data
from the US Census Bureau and land cover classified
imagery with urban, rural, suburban land use classes
may be helpful inputs for the model to capture measures
Mosquito development rates are temperature- dependent
and completion of the immature stages is dependent on
water availability. In a forecasting model, weather (daily
and monthly, and at various time lags) is expected to influ-
ence intra-annual fluctuations. In a static spatial model,
climate, rather than weather, may better govern the likeli-
hood a vector can complete its lifecycle. In an effort to in-
corporate these influences, we recommend including
temperature (minimum, maximum, mean, and daily vari-
ability) and moisture index (calculated based on precipita-
tion and evaporation rates) . The moisture index M(t)
at day t over the current and previous L days is
M t ð Þ ¼ Σt
i¼t?LP i ð Þ ? D i ð Þ;
where P(i) and E(i) are the precipitation and evaporation
from day i, respectively. Moisture indices may be more
meaningful than rainfall because they take into account
the moisture deficit prior to rainfall and estimate the net
moisture in terms of standing water available to egg-laying
Related indirect factors that could plausibly influ-
ence heartworms include: 1) relative humidity, 2)
elevation, and 3) daily circadian-related events in
mosquito behavior. Relative humidity influences vec-
tor behavior and survival. Elevation may serve as a
surrogate of the climate expected at a given location.
Circadian activity is important to some vector spe-
cies’ behavior [21,22].
Several issues were identified relative to the parasite and
its development within the vector and host. Heartworm
Development Units (HDUs) as they relate to nematode
development in mosquitoes and the lag period between
infection and detection in the data on the current anti-
gen detection maps were considered other important
The D. immitis lifecycle consists of several develop-
mental stages in both the mosquito vector and verte-
brate hosts [9,23-25]. Heartworm transmission is driven
by the ambient temperatures experienced by their
mosquito vectors. As a consequence, ambient air tem-
perature is used to predict the timing of dog heartworm
vector competence [26,27]. The heat requirement for
heartworms to complete incubation to the infective stage
can be expressed in development degree days, or HDUs.
A simple way to calculate HDUs using maximum and
minimum air temperatures is to subtract the heartworm
development threshold (14°C) from the average daily
temperature [9,26]. Accumulated HDUs are summed
from a determined starting day, similar to the moisture
index equation above. If hourly temperatures are uti-
lized, the summation of the difference in hourly
temperature from 14°C can be divided by 24. The num-
ber of accumulated degree days required for development
to infective parasite stage is 130 , although this value
may vary by strain of the parasite. D. immitis larval sur-
vival and development within a mosquito is consequently
dependent on temperature and its fluctuations — larval
heartworms are capable of slowing and recommencing
development in a temperature-dependent fashion. In
addition, D. immitis larval survival is dependent on sur-
vival of its invertebrate vector as the infected mosquito
must survive the parasite burden and ingest another
blood meal from a dog in order to transmit the infective
larval stages. Taken together, these findings suggest that
predicting future temperature variation and accumulated
thermal units may be key in predicting heartworm.
The D. immitis antigen does not appear in the blood
of heartworm infected dogs until 6 to 9 months after in-
fection [9,23,27,28]. Therefore, a positive test for heart-
worm often indicates an infection acquired sometime
during the previous year. The CAPC dataset does not in-
clude information on whether positive cases are regular
or new patients. Therefore, all cases are viewed to repre-
sent infections acquired more than 6 months previously.
As these data are used to develop forecasting models,
it will be critical that factors are considered over an
appropriate history. We recommend examining factors
between 7 months and 2 years prior to the current date.
Heartworm hosts, principally the domestic dog, Canis
familiaris, but wild canids also maintain the parasite and
infect mosquitoes . Effective heartworm prevention
is available for domestic dogs, but compliance is insuf-
ficient to control the disease (only 74%-79% of dogs
visiting the veterinary teaching hospital at the Univer-
sity of Pennsylvania [VHUP] from January 1999 through
December 2006 were being given preventive at any given
time of year) .
Measures of the size and infection status of these
susceptible and at-risk populations will be important to
estimating the spatial risk of disease. Although dogs
evacuated from the Gulf coast states following the 2005
hurricanes presented as 48.8% dirofilariasis-positive ,
the percentage positive rate among the general popula-
tion of owned dogs is expected to be much less. Ideally,
data would be available on the susceptible population
Brown et al. Parasites & Vectors 2012, 5:245
Page 4 of 9
of companion animals, coyotes, and feral dogs. However,
such data are likely difficult to acquire. Surrogates for
assessing the susceptible population include estimates of
feral /wild populations, sales data on HW prevention for
companion animals, and socioeconomic factors.
Feral dogs and coyotes are perhaps the most significant
heartworm reservoirs in North America as these compe-
tent reservoir populations are not covered through a
prophylactic program. Indeed, many infections in non-
protected domestic dogs (or dogs receiving inadequate
protection) are probably a major source of infection via
mosquitoes for other domestic dogs. These populations
support parasite development and routinely have cir-
culating microfilariae infective to mosquitoes [31-33].
Unfortunately, accurate data on their population size
and the prevalence of heartworm within a given popula-
tion is difficult to obtain, however, the working group
considered this information to be valuable. As such,
obtaining population size data for coyotes from wildlife
authorities and on stray dogs from animal control agen-
cies could be achieved through telephone inquiries.
Methods have been developed that provide for coyote
population estimates and their relative reliability .
Background heartworm prevalence in reservoir popu-
lations could be estimated from published coyote survey
results, as well data from dogs in shelters. Many reports
on coyote and other wild canid species’ Dirofilaria infec-
tions have been published with report rates as high
as 71% ; however, most surveys occurred in the
early 1980’s . More recent surveys conducted in
Oklahoma/Texas, Illinois, Florida and California have
reported 6.5, 16, 40 and 42-44% of coyotes being positive
[36-40]. In contrast, studies in Arizona and eastern
Washington state reported no coyote infections [41,42],
while a separate recent survey of Arizona wild canids
(feral dogs and coyotes) reported that 14% were positive
for heartworm . Such divergent results emphasize the
care needed to utilize a sentinel survey of feral canines.
Regional data on the percentage of dogs on prevention
or doses of product sold
There are two populations of canids in any heartworm
infected area: canids on preventive therapy and canids
that are not. Many of the dogs being tested are likely
from the protected population. As the protection status
of each tested dog is not available in the CAPC data,
surrogate factors might prove useful. For example, data
on the total dog-equivalent doses sold in a given area
could serve as an indicator for number of animals in an
area on heartworm prevention. A few studies have
attempted to estimate the percentage of dogs on preven-
tatives through owner surveys , but it is difficult to
generalize nationally from geographically localized small-
scale surveys. Such data, therefore, are not necessarily a
good representation of any certain clinic’s patient demo-
graphics. Other sources of potential heartworm pre-
ventative use would involve pharmaceutical companies’
Social economic status
In a study conducted of patients presented at a Pennsyl-
vania teaching hospital, patient age, owner household
income, and being neutered were factors that were asso-
ciated with an increased likelihood of heartworm pre-
ventative compliance . At the highest income levels
in the study, compliance began to drop, sometimes con-
siderably, thereby adding a level of caution when using a
linear association with socioeconomic status. In the case
of the dogs shipped out of Louisiana after Hurricane
Katrina, intact dogs were 1.6 times more likely to have
dirofilariasis than neutered dogs .
A relationship between pet relinquishment and fore-
closures has been established in California . Areas
with concentrated foreclosures had greater concentra-
tions of pet relinquishments. Furthermore, it was
reported that residents in low-valued homes are more
likely to have un-neutered dogs and were more likely
to relinquish these animals, and that neuter status and
relinquishment shifted as home values increased .
Taken with the lower compliance by owners of reported
unneutered dogs , these factors may be valuable in
the model development.
In addition to the factors discussed above, a list of
covariates, secondary in importance to the ones above,
were discussed. These secondary covariates include:
While the effect of latitude may depend on the location,
latitude could serve as a surrogate for general weather,
which influences the presence and abundance of mos-
quito species . However, because the more nuanced
and finer resolution meteorological data are available for
use as potential factors, latitude may be duplicative to
these other factors.
Geographic mobility or migration behavior of humans
(clients) within a given area
As pet owners with unprotected pets move between high
and low transmission regions, they may be facilitating
heartworm spread to new areas (see host factors above).
Bringing infectious or susceptible dogs into new areas
may influence the spread and inter-annual incidence of
heartworm. Geographic mobility data may be important
in identifying heartworm outbreaks. This said, the factor
Brown et al. Parasites & Vectors 2012, 5:245
Page 5 of 9
may be captured in demographic data already described
above, and is hence considered secondary.
In regions where precipitation is low (the Southwest for
example) or where irrigation is common, pollen may
serve as a surrogate for vector habitat. Pollen counts
may be a better measure in regions where precipitation
underestimates the availability of immature habitat.
However, with the inclusion of crop types, forestation
coverage, and meteorological data, pollen counts seem
Factors with insufficient data for inclusion
A number of factors considered important seemed
unlikely to be sufficiently quantified for use in modeling.
These factors can be divided into two groups: those im-
portant to vector and parasite development and those
important to host susceptibility. Some of the vector and
parasite factors excluded were: vector infection rates,
detailed reservoir infection rates, vector abundance and
flight range, vector competence (vector efficiency index),
vector survival, temperature-dependent development
rates of vectors (under natural fluctuating temperature
regimes), and variations in heartworm development
thresholds. These factors are important for accurate
forecasting models because they dictate the rates at
which vectors emerge in the population and become
infectious. However, due to lack of data availability, we
excluded them from the model factors list.
The number of susceptible hosts in a population is
also important. Direct measures of susceptible popula-
tions are not available. Potential surrogates for these
data are presented above.
Concern was expressed that not all of the above factors
are easily obtainable or recorded to a fine geographical
resolution, which may bias the model predictions.
County-level data might be misleading in some locales —
a single county can encompass multiple climate zones in
western parts of the US. County size and homogeneity of
the above factors (e.g., human population density, crop
types, and climates) will vary geographically. There also
was concernover what the
Differences in testing procedures, which may vary
geographically, may influence the model predictions. In
addition, testing rates are likely to not be evenly distribu-
ted across populations such that certain populations,
which may have a considerable role in disease trans-
mission, may not be adequately captured. Aspects of these
limitations can be addressed statistically; others will be
discussed with respect to interpretation of the model
data truly represent.
Data are available for many of the key factors. Geo-
graphic data are readily available from the US Geologic
Survey Land Cover Institute and the National Atlas
of the US [46,47]. Additional land cover data can be
acquired from the US Department of Agriculture Eco-
nomic Research Service and the University of Nebraska –
Lincoln National Drought Mitigation Center [48,49].
Demographic data are available from the US Census
A spatial smoothing procedure based on the head-
banging algorithm method described by Hansen is used
to create a first baseline map (Figure 1) . The head-
banging algorithm is particularly useful in describing
data with high local variations as it is median polished
(not easily influenced by outliers). This algorithm is
named from a child’s game where a face is pressed
against a board of pins protruding at various lengths,
leaving a general impression of the child’s face while
smoothing away any excessively-varying local features
that are more attributable to random chance. We prefer
head-banging to classical Kriging smoothing techniques
since the latter could be unduly influenced by a few
counties with high heartworm prevalence.
A complication is that varying numbers of dogs were
tested in distinct counties. To handle this aspect, the
county data is converted to a common basis via standard
normal Z-scores. Here, p(s) is the probability that a sin-
gle tested dog is heartworm positive at location (county),
s. If N(s) tests were conducted in this county, k(s) of
which are positive, then p(s) is simply estimated as k(s)/
N(s). The standard normal Z-score is this estimated
probability divided by its standard error:
Z s ð Þ ¼
k s ð Þ=N s ð Þ
N s ð Þ
k s ð Þ=N s ð Þ 1?k s ð Þ=N s ð ÞÞ
2 f g:
Counties where no tests are performed do not influ-
ence the analysis. Our conventions take Z(s)=10,000
when all tests in the county are positive, and Z(s)=0
when all tests in the county are negative (these some-
what arbitrary conventions are needed to prevent div-
ision by zero).
After Z-scores are computed for each county, the head-
banging algorithm is applied to spatially smooth them.
From the smoothed Z-scores and the county-by-county
values of N, one can then convert back to a smoothed
probability, representing the probability of a positive test
in each county. Figure 1 is a geographic display of these
smoothed probabilities using 20 nearest neighbors (these
are viewed as adjacent counties).
Brown et al. Parasites & Vectors 2012, 5:245
Page 6 of 9
The smoothed county, s, estimate of p(s), denoted
by p*(s), will be key in our factor identification task.
Specifically, after the smoothed probability estimates are
computed, we will consider logistic regression models
of the form
Logit p?sÞÞ ¼ μ þ β1f1s ð Þ þ ::: þ βLfLs ð Þ þ α s ð Þ;
where L is the number of factors β1. . . , βLare regres-
sion coefficients, f1(s),...,fL(s) are values of the observed
regression factors for county s, μ is an overall location
parameter, and α(s) is zero mean random error for
county, s. Here, logit is defined for values in [0,1] (that
is, probabilities) via
Logit p?s ð Þ
Factor j is significant in the prediction of a positive
heartworm test if βj≠0. Standard forward and backwards
regression model factor selection routines can be used
to determine which of the L factors are significant
and how significant is each factor. The interested reader
is referred to Casella and Berger (2002) for further
Forecasts of future prevalence can be obtained from
the above logistic regression model as various predictors
and data from future years are considered. By invert-
ing the inverting the logit transform of the regression
model utilizing forecasted predictor factors, can be
an estimated value of p*(s) is obtained. For example, if
annual temperature is an important factor, one could
use historical temperature data to forecast next year’s
annual average temperature. This forecasted factor is
then used in the regression equation along with its
accompanying estimated value of β. Right now, any
such forecasts are annual in nature as no seasonality has
been considered. However, after a few years of data
are collected, it may be possible to quantify seasonal
effects and make monthly forecasts. The CAPC data is
ð Þ ¼ log p?s ð Þð Þ ? log 1 ? p?s ð ÞðÞ:
Identifying factors involved in heartworm disease is not
a new endeavor. Initial work in this area occurred in
Canada and the US [26,27], and this has since led to
studies on the prediction of the different transmis-
sion seasons in Europe , the United Kingdom 
and Argentina . These studies predict the beginning
and end of seasonal transmission, and some consider the
impact of climate change . Utilizing the rich CAPC
data base we have access to, we expect to obtain a
clearer understanding of canine heartworm transmission
in the US. Our working
eral factors that could be used in model development.
These factors, combined with the modeling approaches
outlined above, will be fitted to the comprehensive
CAPC data set in the future to generate detailed estima-
tion of canine heartworm risk at a county-by-county
resolution in the US.
Ae: Aedes; An: Anopheles; CAPC: Companion Animal Parasite Council;
Cx: Culex; D: Dirofilaria; HDU: Heartworm Degree Unit; VHUP: Veterinary
Hospital at the University of Pennsylvania.
The authors have no competing interests relative to the work presented in
DW and RL were responsible for the statistical presentation and the
production of Figure 1. HEB, LCH, PK, and TM were the group participants
who identified the risk factors, built the framework for their inclusion into
the manuscript, and provided the rationale in the paper for their inclusion.
DDB and CTN were responsible for generating the initial draft document
from the minutes of the meeting for circulation to the group, the
compilation of the later drafts of the document, incorporation of all
comments, and assistance with formatting for final submission. All authors
read and approved the final version of the manuscript.
The authors would like to thank the Companion Animal Parasite Council for
its facilitation of the meeting through its organizational efforts and its
financial support for travel, food, and lodging to make the meeting a
possibility. Robert Lund’s work was supported by National Science
Foundation Grant DMS 0905570.
1School of Geography and Development, University of Arizona, Tucson, AZ
85721, USA.2Department of Entomology, Cornell University, Ithaca, NY
14853, USA.3Entomology and Nematology Department, University of Florida,
Gainesville, FL 32611, USA.4Department of Biological Sciences, Arkansas
State University, State University, AR 72467, USA.5Department of
Microbiology and Immunology, College of Veterinary Medicine, Cornell
University, Ithaca, NY 14853, USA.6Animal Medical Center, Anniston, AL
36201, USA.7Department of Mathematical Sciences, Clemson University,
Clemson, SC 29634-0975, USA.
Received: 10 October 2012 Accepted: 22 October 2012
Published: 30 October 2012
CAPC Parasite Prevalence Maps Heartworm. 2011. http://www.capcvet.org/
2.Bowman DD, Little SE, Lorentzen L, Shields J, Sullivan MP, Carlin EP:
Prevalence and geographic distribution of Dirofilaria immitis, Borrelia
burgdorferi, Ehrlichia canis, and Anaplasma phagocytophilum in dogs in
the United States: results of a national clinic-based serologic survey.
Vet Parasitol 2009, 160:138–48.
3. Guerrero J, Nelson CT, Carithers DS: Results and realistic implications of
the 2004 AHS-Merial heartworm survey. In Proceedings of the 51st Annual
Meeting of the American Association of Veterinary Parasitologists. Honolulu:
2006. July 15–18, 2006, Abstract 63.
4.Atkins CE: Comparison of results of three commercial heartworm antigen
test kits in dogs with low heartworm burdens. J Am Vet Med Assoc 2003,
5.Frank GR, Grieve RB, Mok M, Smart DJ, Salman MD: Survey of heartworm
(Dirofilaria immitis) infection in Colorado dogs: a model for surveying
prevalence in low-endemic areas. In Proceedings of the heartworm
symposium '92, Austin, Texas, USA, 27–29 March, 1992. Edited by Soll MD,
Batavia IL. 1992:5–10.
6.Theis JH, Franti C, Lambert L, Giammattei V, Parker V, Lee G: Risk factors for
heartworm infection in dogs living in Sierra Nevada foothills and
Sacramento Valley counties; public health implications. California
Veterinarian 1984, 38:13–17.
Brown et al. Parasites & Vectors 2012, 5:245
Page 7 of 9
7.Ludlam KW, Jachowski LA Jr, Otto GF: Potential vectors of Dirofilaria
immitis. J Am Vet Med Assoc 1970, 157:1354–59.
Comiskey N, Wesson DM: Dirofilaria (Filarioidea: Onchocercidae) infection
in Aedes albopictus (Diptera: Culicidae) collected in Louisiana. J Med
Entomol 1995, 32:734–37.
Ledesma N, Harrington L: Mosquito vectors of dog heartworm in the
United States: vector status and factors influencing transmission
efficiency. Top Comp Anim Med 2011, 26:178–85.
Scoles GA: Vectors of canine heartworm in the United States: a review of
the literature including new data from Indiana, Florida, and Louisiana. In
Recent Advances in Heartworm Disease: Symposium’98: Proceedings of the
American Heartworm Society: 1–3 May 1998. Edited by Seaward RL, Courtney
CH. Tampa: 1998:21–36.
Bowman DD, Atkins CE: Heartworm biology, treatment, and control. Vet
Clin Small Anim 2009, 39:1127–58.
Darsie RF Jr, Ward RA: Identification and Geographical Distribution of the
Mosquitoes of North America, North of Mexico. Gainesville: University Press of
Nasci RS, Hare SG, Willis FS: Interspecific mating between Louisiana strains
of Aedes albopictus and Aedes aegypti in field and laboratory. J Am Mosq
Control Assoc 1989, 5:416–21.
O’Meara GF, Evans LF Jr, Gettman AD, Cuda JP: Spread of Aedes albopictus
and decline of Ae. aegypti (Diptera: Culicidae) in Florida. J Med Entomol
Petersen WH, Zack RS, Dykstra EA, Owen JP: New distribution records of
mosquitoes in eastern Washington state. J Am Mosq Control Assoc 2010,
Jamieson DH, Olson LA, Wilhide JD: A larval mosquito survey in
northeastern Arkansas including a new record for Aedes albopictus. J Am
Mosq Control Assoc 1994, 10:236–39.
Schreiber ET, Webb JP Jr, Hazelrigg JE, Mulla MS: Bionomics of adult
mosquitoes associated with urban residential areas in the Los Angeles
Basin, California. Bull Soc Vector Ecol 1989, 14:301–18.
Su T, Webb JP, Meyer RP, Mulla MS: Spatial and temporal distribution of
mosquitoes in underground storm drain systems in Orange County,
California. J Vec Ecol 2003, 28:79–89.
Müller GC, Junnila A, Qualls W, Revay EE, Kline DL, Allan S, Schlein Y, Xue
RD: Control of Culex quinquefasciatus in a storm drain system in Florida
using attractive toxic sugar baits. Med Vet Entomol 2010,
Gong H, DeGaetano AT, Harrington LC: Climate-based Models for West
Nile Culex Mosquito Vectors in the Northeastern USA. Int J Biometeorol
Spielman A: Structure and seasonality of Neartic Culex pipiens
populations. Proc NY Acad Sci 2001, 951:220–234.
Eldridge BF: Environmental control of ovarian development in
mosquitoes of the Culex pipiens complex. Science 1966,
Grieve RB, Lok JB, Glickman JT: Epidemiology of canine heartworm
infection. Epidemiol Rev 1983, 5:220–46.
Lok JB, Knight DH: Laboratory verification of a seasonal heartworm
transmission model. In Recent advances in heartworm disease: Symposium
'98, Tampa, Florida, USA, 1–3 May, 1998. Edited by Seward RL, Knight DH.
Batavia: American Heartworm Society; 1998:15–20.
Nayar JK, Bradley TJ: Effects of infection with Dirofilaria immitis on diuresis
and oocyte development in Aedes taeniorhynchus and Anopheles
quadrimaculatus (Diptera: Culicidae). J Med Entomol 1987,
Slocombe JOD, Surgeoner GA, Srivastava B: Determination of the
heartworm transmission period and its used in diagnosis and control. In
Proceedings of the Heartworm Symposium '89, Charleston, South Carolina,
USA, 17–19 March, 1989. Edited by Otto GF, Jackson RF, Knight DH,
Campbell WC, Courtney CH, Dillon R, Hite SC, Jackson RI, Levine BG, Lewis
RE, Noyes JD. Washington, DC: American Heartworm Society;
Knight DH, Lok JBL: Seasonality of heartworm infection and implications
for chemoprophylaxis. Clin Tech Sm Anim Pract 1998, 13:77–82.
Frank GR, Pace PM, Donoghue AR: Antigenemia and microfilaremia in
canine experimental dirofilariasis. In Recent advances in heartworm disease:
Symposium 01, San Antonio, Texas, USA, 20–22 April, 2001. Edited by Seward
RL, Knight DH. Batavia: American Heartworm Society; 2001:211–14.
29.Gates MC, Nolan TJ: Factors influencing heartworm, flea, and tick
preventative use in patients presenting to a veterinary teaching hospital.
Preventative Vet Med 2010, 93:193–200.
Levy JK, Edinboro CH, Glotfelty C-S, Dingman PA, West AL, Kirkland-Cady
DK: Seroprevalence of Dirofilaria immitis, feline leukemia virus, and feline
immunodeficiency virus infection among dogs and cats exported from
the 2005 Gulf Coast hurricane disaster area. J Amer Vet Med Assoc 2007,
Sacks BN: Increasing prevalence of canine heartworm in coyotes from
California. J Wildlife Dis 1998, 34:386–9.
Sacks BN, Woodward DL, Colwell AE: A long-term study of non-native-
heartworm transmission among coyotes in a Mediterranean ecosystem.
Oikos 2003, 102:478–90.
Holzman S, Conroy MJ, Davidson WR: Diseases, parasites and survival of
coyotes in south-central Georgia. J Wildlife Dis 1992,
Henke SE, Knowlton FF: Techniques for estimating coyote abundance. In
Coyotes in the Southwest: A Compendium of Our Knowledge. Symposium
Proceedings, December 13–14, 1995. Edited by Rollins D, Richardson C,
Blankenship T, Canon K, Henke S. San Angelo: ICWDM website,
DigitalCommons@Univeristy of Nebraska Lincoln; 1996:71–78. http://
Custer JW, Pence DB: Dirofilariasis in wild canids from the gulf coastal
prairies of Texas and Louisiana, U.S.A. Vet Parasitol 1981,
Anderson RC: Filarioid Nematodes. In Parasitic Diseases of Wild Mammals.
Edited by Samuel WM, Pybus MJ, Kocan AA. Ames, IA: Iowa State University
Foster GW, Main MB, Kinsella JM, Dixon LM, Terrell SP, Forrester DJ: Parasitic
helminthes and arthropods of coyotes (Canis latrans) from Florida, U.S.A.
Comp Parasitol 2003, 70:162–6.
Nelson TA, Gregory DG, Laursen JR: Canine heartworms in coyotes in
Illinois. J Wildlife Dis 2003, 39:593–9.
Sacks BN, Caswell-Chen EP: Reconstructing the spread of Dirofilaria
immitis in California coyotes. J Parasitol 2003, 89:319–23.
Paras KL, Little SE, Reichard MV, Reiskind MH: Detection of Dirofilaria
immitis and Ehrlichia species in coyotes (Canis latrans), from rural
Oklahoma and Texas. Vector Borne Zoonotic Dis 2012, 1
Grinder M, Krausman PR: Morbidity-mortality factors and survival of an
urban coyote population in Arizona. J Wildlife Dis 2001,
Foreyt WJ: Prevalence of heartworm (Dirofilaria immitis) in coyotes (Canis
latrans) in Washington state. Northwest Sci 2008,
Olsen-Mikitowicz VM: Prevalence of Dirofilaria immitis, the causative agent in
canine heartworm disease, in Arizona from 2008 to 2011 using feral canines as
a sentinel species. B.S. Thesis. USA: University of Arizona; 2011.
Morris GD, Steffler J: Was pet relinquishment related to foreclosure?: a
spatial research note from California during the height of foreclosure.
Social Sci J 2011, 48:739–45.
Edillo F, Kiszewski A, Manjourides J, Pagano M, Hutchinson M, Kyle A, Arias J,
Gaines D, Lampman R, Novak R, Foppa I, Lebelcyzk C, Simth R, Moncayo A,
Spielman A: Effects of latitude and longitude on the population structure
of Culex pipiens s.l., vectors of West Nile virus in North America. Am J
Trop Med Hyg 2009, 81:842–48.
The USGS Land Cover Institute (LCI): http://landcover.usgs.gov/
USDA Economic Research Service Major Land Uses; http://www.ers.usda.gov/
National Drought Mitigation Center: U.S. Drought Monitor.: ; http://drought.
United States Census Bureau: http://www.census.gov/#.
Hansen KM: Head-banging: robust smoothing in the plane. IEEE Trans on
GeosciRemote Sens 1991, 29:369–78.
Casella G, Berger RL: Statistical Inference. Secondth edition. Pacific Grove, CA
USA: Duxbury Press; 2002:700.
Genchi C, Rinaldi L, Cascone C, Mortarino M, Cringoli G: Is heartworm
disease really spreading in Europe? Vet Parasitol 2005, 133:137–48.
Brown et al. Parasites & Vectors 2012, 5:245
Page 8 of 9
54.Medlock JM, Barrass I, Kerrod E, Taylor MA, Leach S: Analysis of climatic Download full-text
predictions for extrinsic incubation of Dirofilaria in the United Kingdom.
Vector Borne Zoonotic Dis 2007, 7:4–14.
Vezzani D, Carbajo AE: Spatial and temporal transmission risk of Dirofilaria
immitis in Argentina. Int J Parasitol 2006, 36:1463–72.
Genchi C, Kramer LH, Rivasi F: Dirofilarial Infections in Europe. Vector Borne
Zoonotic Dis 2011, 11:1307–17.
Cite this article as: Brown et al.: Key factors influencing canine
heartworm, Dirofilaria immitis, in the United States. Parasites & Vectors
Submit your next manuscript to BioMed Central
and take full advantage of:
• Convenient online submission
• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at
Brown et al. Parasites & Vectors 2012, 5:245
Page 9 of 9