Upload
lk-lord
View
215
Download
1
Embed Size (px)
Citation preview
Review
Use of group-randomized trials in pet
population research
L.K. Lord a,*, T.E. Wittum a, J.M. Scarlett b
a Department of Veterinary Preventive Medicine, College of Veterinary Medicine,
The Ohio State University, 1920 Coffey Road, Columbus, OH 43210, United Statesb Department of Population Medicine and Diagnostic Sciences, College of Veterinary Medicine,
Cornell University, Ithaca, NY 14853, United States
Received 10 March 2007; received in revised form 7 June 2007; accepted 10 July 2007
Abstract
Communities invest considerable resources to address the animal welfare and public health concerns
resulting from unwanted pet animals. Traditionally, research in this area has enumerated the pet-owning
population, described pet population dynamics in individual communities, and estimated national eutha-
nasia figures. Recent research has investigated the human–animal bond and explored the community
implications of managed feral cat colonies. These reports have utilized traditional epidemiologic study
designs to generate observational data to describe populations and measure associations. However, rigorous
scientific evaluations of potential interventions at the group level have been lacking. Group-randomized
trials have been used extensively in public health research to evaluate interventions that change a
population’s behavior, not just the behavior of selected individuals. We briefly describe the strengths
and limitations of group-randomized trials as they are used to evaluate interventions that promote social and
behavioral changes in the human public health field. We extend these examples to suggest the appropriate
application of group-randomized trials for pet population dynamics research.
# 2007 Elsevier B.V. All rights reserved.
Keywords: Group-randomized trial; Pet overpopulation; Study methodology
1. Introduction
Unwanted pets are a widely recognized problem in the United States that is associated with
considerable animal welfare and public health concerns. The common means by which
www.elsevier.com/locate/prevetmed
Preventive Veterinary Medicine 82 (2007) 167–175
* Corresponding author. Tel.: +1 614 292 1206; fax: +1 614 292 4142.
E-mail address: [email protected] (L.K. Lord).
0167-5877/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.prevetmed.2007.07.007
communities manage the problem of unwanted pets is with animal shelters. Shelters serve to
reunite owners with lost pets, find unclaimed animals new homes, or euthanize unwanted
animals. Today, there is a major movement among animal sheltering groups to reduce or
eliminate the euthanasia of healthy, adoptable animals in shelters and to identify means to
increase adoptions (Rowan and Williams, 1987). It appears that some success in reducing
euthanasia in shelters has been achieved. Annual euthanasia numbers have declined from an
estimated 13.5–18.6 million dogs and cats in the 1970s to the most recent national estimates of
4–6 million dogs and cats (Rowan and Williams, 1987; Nassar et al., 1992; Arkow, 1994; Irwin,
2001). Despite this progress, a large number of unwanted dogs and cats continue to be
euthanized in shelters each year, demanding further investment of resources and need for better
solutions.
The academic research community has also invested resources to better understand pet
population dynamics. In the 1970s, the major focus of research was on the basics of trying to
enumerate the pet-owning population, describe the pet population dynamics in several
communities, and estimate national euthanasia figures (Schneider and Vaida, 1970; Franti and
Kraus, 1974; Griffiths and Brenner, 1976; Nassar and Mosier, 1980, 1982, 1986; Nassar et al.,
1984). National humane groups such as the American Humane Association and the Humane
Society of the United States also invested resources in trying to estimate the number of animal
shelters and national euthanasia figures (Rowan and Williams, 1987; American Humane
Association, 1992). These early efforts faced many obstacles, primarily the lack of a census for
the pet-owning community and the lack of an accurate list for the number of animal shelters in the
United States. These problems remain today.
Since the 1990s, the focus of pet population dynamic research (called by many ‘‘pet
overpopulation’’ research) has shifted to attempt to better understand reasons for owner
relinquishment of pets resulting from the breakdown of the human–animal bond, factors
associated with successful adoptions, evaluation of the health effects of early age spay-neuter
and the implications of the increasing feral cat population (Patronek et al., 1996a,b; Salman
et al., 1998, 2000; Scarlett et al., 1999; New et al., 2000; Hughes et al., 2002; Hughes and Slater,
2002; Levy et al., 2003; Spain et al., 2004a,b). Much of this research has utilized basic
epidemiological study designs including cross-sectional studies, case–control studies, and
retrospective cohort studies. This era of research has made a tremendous contribution toward a
better understanding of the factors associated with overpopulation and has laid the groundwork
for further advancement.
Although basic epidemiologic data have been generated and reported, studies of population
interventions to reduce pet overpopulation are sparse. Few reports of rigorous scientific
evaluation of interventions such as the use of individual randomized trials and group-randomized
trials have been reported. Just as individual randomized trials are considered the ‘‘gold standard’’
for studies to evaluate interventions at the individual level, group-randomized trials (GRT) are
considered the ‘‘gold standard’’ for evaluation of interventions delivered at the group level.
Group-randomized trials have been used extensively in public health research to evaluate
interventions that change a population’s behavior not just the behavior of selected individuals.
The value of group-randomized trials is their effectiveness in evaluating population-level
strategies that seek to change the social norm for a particular behavior. Although increasing the
use of individual randomized trials would be of benefit in pet population research, their
techniques are well described in the veterinary literature and will not be discussed further in this
paper. Despite being used commonly in public health research, group randomized trials have not
been utilized in companion animal research and will be the focus in this manuscript.
L.K. Lord et al. / Preventive Veterinary Medicine 82 (2007) 167–175168
The goals of this report are to: (1) briefly describe how group-randomized trials are used to
evaluate interventions that promote social and behavioral changes in the human public health
field, including a discussion of several of the biases and other issues that affect these trials; (2)
provide examples of how group-randomized trials may be used in pet population dynamics
research. While it is not our intention to provide a comprehensive review of the literature
regarding the use of group-randomized trials, we hope that this manuscript will serve to stimulate
interest and discussion regarding the application of this methodology to the study of pet
overpopulation.
2. Use of group-randomized trials in public health research
2.1. Introduction to group-randomized trials
A group-randomized trial (GRT) is a study designed to evaluate an intervention that operates
at a group level such as a school, a work-site or a community, changes the social or physical
environment, or cannot be delivered at the individual level (Murray, 1998; Donner and Klar,
2000). These studies have been advocated in public health research over the last 30 years to
achieve widespread change in health behaviors. Common groups used in these studies include
communities, worksites and schools. GRTs are considered the gold standard when the allocation
of interventions to groups and not to individuals is necessary.
There are several characteristics, which make group-randomized trials different from the
standard randomized clinical trials used in clinical medicine (Murray, 1998; Murray et al., 2004).
The unit of assignment in a GRT is a group rather than an individual, and the group adds an extra
level of nesting and variability within the study. Groups are also not formed at random but are
based on some common characteristic related to social structure, geography or other common
characteristic. Also, due to the complexity and size of groups in most GRTs, only a small number
of groups can be allocated to the intervention and control conditions, and often the size of the
groups may be relatively large (hundreds/group). Whereas a single drug or treatment is often
evaluated in a clinical trial, interventions in a GRT have traditionally been multi-factorial and
have involved a variety of educational programs, mass media campaigns, and social marketing
tools designed to influence group behavior.
There are numerous examples of group-randomized trials in the public health literature. In the
1970s three large community intervention trials, the Stanford Five-City Project, Minnesota Heart
Health Project, and Pawtucket Heart Health Project, were funded by the National Heart, Lung
and Blood Institute that targeted reducing risk factors for coronary heart disease such as high
blood pressure, elevated cholesterol, and cigarette smoking (Farquhar et al., 1985; Luepker et al.,
1994; Carleton et al., 1995). None of these early studies included randomization of the groups,
and magnitude and level of significance for reduction of risk factors were mixed in the three
studies. In 1989, the National Cancer Institute funded the Community Intervention Trial for
Smoking Cessation (COMMIT) (COMMIT Research Group, 1991; Green et al., 1995). In this
study, 11 matched pairs of communities were randomized to evaluate a 4-year intervention to
encourage smokers to quit smoking. Significant declines in smoking prevalence were seen among
light to moderate smokers, particularly those in the less educated group. In another study,
Communities Mobilizing for Change on Alcohol (CMCA), 15 communities were randomized to
evaluate a community-organizing intervention to limit teen access to alcohol (Wagenaar et al.,
2000). A decline in arrest and traffic crash indicators was found in the intervention communities
as a result. Group-randomized trials have also been used in both the workplace and school setting
L.K. Lord et al. / Preventive Veterinary Medicine 82 (2007) 167–175 169
to promote healthy behaviors such as physical activity and improved eating habits, and to
discourage risky behaviors such as smoking (Secker-Walker et al., 2006; Boutelle et al., 2001;
Stevens et al., 2005).
2.2. Potential sources of bias in group-randomized trials
GRTs are particularly susceptible to several forms of bias that must be controlled for as much
as possible (Murray, 1998, 2001). The first type of bias is selection bias, meaning that differences
exist between groups before the intervention is carried out and these differences are not evenly
distributed between treatment and control groups. In most individual randomized trials this bias
is handled through randomization. In GRTs, however, with the small number of groups it may not
be possible to completely eliminate selection bias even with randomization. Selection bias is part
of the reason that having as many groups as possible is critical in GRTs.
Group-randomized trials occur in the ‘‘real’’ world, and schools, worksites and communities are
potentially exposed to influences other than the intervention. If these other external influences have
an impact on the entire population in the study, they are referred to as bias due to history. If these
influences equally affect both treatment and control groups, they are less of a concern than those that
only affect groups in one arm of the study, the latter being called bias due to differential history. For
example, if the intervention involves implementing a smoking cessation program in the treatment
communities, bias due to differential history would occur if during the study period the local health
departments in some of the control communities also implemented a smoking cessation campaign.
This could erroneously cause the investigators to conclude the intervention had less of an effect in
reducing smoking in the treatment communities than was actually the case.
Another type of bias that is of particular concern in GRTs is maturation bias, also known as
secular trends. This has been a concern in some of the more recent community trials such as the
coronary heart disease trials and COMMIT discussed above. This type of bias occurs when the
natural development or long-term trends of a behavior affect the results of the study, and can be
non-differential or differential. For example, in the COMMIT study it was believed that the
secular trend of a decline in smoking among the control groups masked part of the intervention
effect found in the intervention groups. Although secular trends cannot be prevented, study
design techniques can be implemented to help evaluate and control for their effects.
The other major type of bias that is of particular concern in GRTs is that of contamination.
This occurs when the program that is implemented in the intervention groups leaks or crosses
over into the control groups. For example, this can occur when community leaders in one or
more of the intervention groups share intervention information with their counterparts in
control groups, and programs suddenly appear in the control groups. Researchers must take
particular care in monitoring the control groups for contamination and making sure that
participants in the intervention groups understand the need to keep program information in their
own groups.
2.3. Other methodological issues in group-randomized trials
One major issue with group-randomized trials that is difficult to address despite an adequate
design is allowing enough time for the intervention to take effect. In many cases, an intervention
that involves changing behavior can take years to have its full impact, and depending on the
amount of time and funding available, evaluation of the intervention may occur prior to its
reaching its full impact.
L.K. Lord et al. / Preventive Veterinary Medicine 82 (2007) 167–175170
Several advances have been made in the design of these studies, particularly in community-
based trials (Sorensen et al., 1998; Thompson et al., 2003). One of the areas that researchers have
tried to address is community involvement. In early intervention trials, the intervention was
chosen by the researchers with no input from the community, while today most studies include a
community advisory board (Sorensen et al., 1998; Thompson et al., 2003). Most community-
based trials also include some type of process evaluation to understand how the components of
the intervention are delivered and to make sure they are received by individuals. Finally, more
emphasis has been put on maintenance of intervention programs, particularly if they are
successful. Researchers, groups and funding organizations need to consider what is required to
maintain a successful intervention program at some level in order to both monitor long-term
trends beyond the initial study and to help ensure community buy-in of the program.
2.4. Statistical issues in group-randomized trials
There are several issues that must be accounted for in GRTs that are not of issue in individual
randomized trials. First, members of groups have something in common, and this level of
correlation (called the intraclass correlation coefficient or ICC) must be taken into account (Kish,
1965). The ICC is almost always positive in group-randomized trials and leads to an increase in
the variance of the group means. This ICC lowers the power of a study and if not accounted for in
the analysis, will inflate the type I error rate (Feng et al., 2001; Murray, 1998). Second, because
the unit of assignment is the group and not the individual, testing for intervention effects depend
on the degrees of freedom based on the number of groups. Because the number of groups is small,
the power of these tests for intervention effects is further reduced and if the incorrect degrees of
freedom are used, the type I error rate will also be inflated. This can be particularly problematic in
detecting statistically significant intervention effects in group-randomized trials. It has also been
shown that by modeling time explicitly in the analysis and adjusting for individual level
regression covariates, the operative ICC in group-randomized trials can be reduced (Murray and
Blitstein, 2003).
3. Using group-randomized trials in pet population research
Many proposed programs to reduce pet overpopulation are targeted at groups and cannot be
effectively evaluated at the individual animal or owner level. These programs are implemented at
a larger level such as communities and are often designed to initiate societal change. Thus the use
of group-randomized trials is warranted to provide rigorous scientific evaluation of their effect.
Below are examples of how group-randomized trials could be used in pet population research.
3.1. Evaluation of feral cat trap-neuter-return (TNR) programs
An issue of importance that lends itself to evaluation with group-randomized trials is the use
of trap-neuter-return (TNR) programs to reduce the feral cat population in communities. The
problem of free-roaming cats has been well documented, and is one of the largest challenges
facing the humane movement in the 21st century (Patronek and Rowan, 1995; Kirkwood, 1998;
APPMA, 2006). TNR programs are largely promoted to help address this growing problem, but
few rigorous studies have been done to evaluate their effectiveness. One descriptive study showed
a decline in the percentage of impounded cats euthanized and an increase in the percentage of
impounded cats adopted after a TNR program was implemented in a community (Hughes et al.,
L.K. Lord et al. / Preventive Veterinary Medicine 82 (2007) 167–175 171
2002). The authors recognized, however, that this effect could not be attributed directly to the
TNR program, as there were other changes implemented during the same time in the community
that could also have explained the observed effects. Two studies conducted on university
campuses showed a decline in the number of cats and kittens caught when a comprehensive TNR
and adoption program was implemented (Hughes and Slater, 2002; Levy et al., 2003). These
latter two studies demonstrate that in smaller, controlled environments with active caregivers,
TNR could be an effective means to control and even reduce feral cat populations. In a more
recent study using mathematical population modeling, TNR was not shown to reduce the per
capita growth in cat populations, and any population effects were minimal for two large county
TNR programs (Foley et al., 2005). In another study that employed matrix population models, it
was shown that annual neutering of greater than 75% of the fertile cat population would be
needed to control cat populations, which may be unrealistic for many larger scale TNR programs
(Andersen et al., 2004). These types of studies using mathematical modeling rely on certain
assumptions such as controlled populations and can make their findings somewhat difficult to
interpret or apply in a real-life setting. All of these studies, however, underscore the need for
further investigation to access the community impact of TNR programs, particularly as they
relate to reducing the numbers of cats entering shelters.
A group-randomized trial could be used to evaluate the effectiveness of TNR programs at the
community level. To set up a study, neighborhoods serviced by a particular humane society could
be defined by geographical boundaries using U.S. Census Bureau census tracts. These
neighborhoods could then be randomized into treatment and control areas, where treatment
neighborhoods would implement TNR programs for a specified time period and control
neighborhoods would not. Assessments of the number of feral cats and the number of complaint
calls at the humane society before and after the start of the intervention would be measured. The
humane society would track the addresses of all incoming feral cats to attribute them to their
respective neighborhoods. This approach would allow an area to be divided into neighborhoods
that were small enough for the intervention to have a realistic opportunity to have an impact,
instead of trying to measure the effect for an overall county as has been done in prior research.
3.2. Evaluation of programs to improve pet reunification
Reunification of pets is an important means to expeditiously move animals out of shelters.
Three epidemiological studies were conducted recently in Dayton, Ohio to describe the lost and
found search process (Lord et al., 2007a,b,c). Two initial studies used a cohort of lost dogs and
cats, respectively, that were identified by owner reports of losing a pet to local animal agencies or
placement of a lost pet advertisement in the local newspaper. These studies showed that only 48%
of dogs and 19% of cats were wearing some type of identification or had a microchip. Further, the
studies showed that for dogs, utilization of the animal sheltering system through calling and/or
visiting an animal agency, the pet wearing a dog license tag, and the hanging of neighborhood
signs all increased the probability of reunification. For cats, two-thirds returned home on their
own and the most successful method of reunification was posting of neighborhood signs. These
findings provide insight into the lost and found pet search process and point out many areas where
interventions through education on search methods and identification could be evaluated with
more sophisticated study designs. If these interventions were offered in communities or other
sites such as veterinary clinics, then the GRT design could be used.
For example, we know that in a population of dogs and cats that were lost, only 8% of dogs and
7% of cats had a microchip, and only 1.5% of the dogs and no cats that were reunited were done
L.K. Lord et al. / Preventive Veterinary Medicine 82 (2007) 167–175172
so by a microchip (Lord et al., 2007a,b). Given the emphasis placed on microchips as a recovery
method for lost pets by the humane community, particularly with the recent difficulties in
reunification during the response to hurricane Katrina, evaluation of methods to increase the use
of microchips is warranted. A community based group randomized trial could be used to evaluate
promotional programs in communities (intervention group) compared to control communities by
using pre- and post-intervention surveys of residents to evaluate if microchip promotion
increased their prevalence at the community level. Microchip and identification campaigns could
also be used in a similar approach to assess the shelter return to owner rate in those communities
where programs were implemented. Because these microchip and identification campaigns
would be offered at the community level and not to particular individuals, a group-randomized
trial would be required.
4. Conclusions
Pet population research would benefit from the use of group-randomized trials. Funding
agencies have invested considerable resources in programs to reduce pet overpopulation, but in
most cases scientific evaluation of the impact of these programs has not been conducted. Because
much of the funding is implemented through community groups such as humane societies,
funding agencies are increasingly asking what impact their money is having at the community
level. Funding groups, communities, owners and most importantly pets, will be better served if
community programs are scientifically evaluated using GRTs.
These types of interventional studies are not without challenges. They require a change in
approach and a redirection of research dollars. They will be expensive and time-consuming
and will require pilot studies to evaluate their feasibility and to estimate intraclass correlation
coefficients so that larger studies have sufficient statistical power. Traditional epidemiologists
will have to involve researchers with expertise in the field of health behavior and promotion
and be willing to devote resources to establishing community involvement. This approach
may appear daunting and require more resources than traditional funding agencies are willing
to provide. As a society, however, the continued euthanasia of millions of pets in animal
shelters in the United States each year is morally unacceptable and as researchers, we have a
responsibility to advance our knowledge to impact the problem to the greatest extent
possible. We ask that the research community and funding agencies consider the need for
better evaluation of interventions and explore the use of group-randomized trials as a method
to do so.
References
American Humane Association, 1992. Animal shelter reporting study. Denver, CO: American Humane Association.
Andersen, M.C., Martin, B.J., Roemer, G.W., 2004. Use of matrix population models to estimate the efficacy of euthanasia
versus trap-neuter-return for management of free-roaming cats. J. Am. Vet. Med. Assoc. 225, 1871–1876.
APPMA, 2006. 2005–2006 National pet owner survey. American Pet Prodcts Manufacturing Association, Greenwich, CT.
Page 80–81.
Arkow, P., 1994. A new look at pet overpopulation. Anthrozoos 7, 202–205.
Boutelle, K.N., Jeffery, R.W., Murray, D.M., Schmitz, M.K., 2001. Using signs, artwork, and music to promote stair use in
a public building. Am. J. Public Health 91, 2004–2006.
Carleton, R.A., Lasater, T.M., Assaf, A.R., Feldman, H.A., McKinlay, S., 1995. The Pawtucket heart health program:
community changes in cardiovascular risk factors and projected disease risk. Am. J. Public Health 85, 777–785.
COMMIT. Research Group, 1991. Community Intervention Trial for Smoking Cessation (COMMIT): summary of design
and intervention. J. Natl. Cancer Inst. 83, 1620–1628.
L.K. Lord et al. / Preventive Veterinary Medicine 82 (2007) 167–175 173
Donner, A., Klar, N., 2000. Design and Analysis of Cluster Randomization Trials in Health Research. Oxford University
Press Inc., New York.
Farquhar, J.W., Fortmann, S.P., Maccoby, N., Haskell, W.L., Williams, P.T., Flora, J.A., Taylor, C.B., Brown Jr., B.W.,
Solomon, D.S., Hulley, S.B., 1985. The Stanford Five-City Project: design and methods. Am. J. Epidemiol. 122, 323–
334.
Feng, Z., Diehr, P., Peterson, A., McLerran, D., 2001. Selected statistical issues in group randomized trials. Annu. Rev.
Public Health 22, 167–187.
Foley, P., Foley, J.E., Levy, J.K., Paik, T., 2005. Analysis of the impact of trap-neuter-return programs on populations of
feral cats. J. Am. Vet. Med. Assoc. 227, 1775–1781.
Franti, C.E., Kraus, J.F., 1974. Aspects of pet ownership in Yoko county, California. J. Am. Vet. Med. Assoc. 164, 166–
171.
Green, S., Corle, D., Gail, M., 1995. Interplay between design and analysis for behavioral intervention trials with
community as the unit of randomization. Am. J. Epidemiol. 14, 587–593.
Griffiths, A.O., Brenner, A., 1976. Survey of cat and dog ownership in Champaign county, Illinois. J. Am. Vet. Med.
Assoc. 170, 1333–1340.
Hughes, K.L., Slater, M.R., 2002. Implementation of a feral cat management program on a university campus. J. Appl.
Anim. Welf. Sci. 5, 15–28.
Hughes, K.L., Slater, M.R., Haller, L., 2002. The effects of implementing a feral cat spay/neuter program in a Florida
county animal control service. J. Appl. Anim. Welf. Sci. 5, 285–298.
Irwin, P.G., 2001. Overview: the state of animals in 2001. In: Salem, D.J., Rowan, A.N. (Eds.), The State of the Animals
2001. Humane Society Press, Washington, DC, pp. 1–19.
Kirkwood, S., 1998. Free-roaming cats: in search of new approaches. Anim. Shelter 21, 5–6.
Kish, L., 1965. Survey Sampling. John Wiley & Sons, New York.
Levy, J.K., Gale, D.W., Gale, L.A., 2003. Evaluation of the effect of a long-term trap-neuter-return and adoption program
on a free-roaming cat population. J. Am. Vet. Med. Assoc. 222, 42–46.
Lord, L.K., Wittum, T.E., Ferketich, A.K., Funk, J.A., Rajala-Schultz, P.J., 2007a. Search and identification methods that
owners use to find a lost dog. J. Am. Vet. Med. Assoc. 230, 211–216.
Lord, L.K., Wittum, T.E., Ferketich, A.K., Funk, J.A., Rajala-Schultz, P.J., 2007b. Search and identification methods that
owners use to find a lost cat. J. Am. Vet. Med. Assoc. 230, 217–220.
Lord, L.K., Wittum, T.E., Ferketich, A.K., Funk, J.A., Rajala-Schultz, P.J., 2007c. Search methods that people use to find
owners of lost pets. J. Am. Vet. Med. Assoc. 230, 1835–1840.
Luepker, R.V., Murray, D.M., Jacobs Jr., D.R., Mittelmark, M.B., Bracht, N., Carlaw, R., Crow, R., Elmer, P., Finnegan, J.,
Folsom, A.R., 1994. Community education for cardiovascular disease prevention: risk factor changes in the
Minnesota Heart Health Program. Am. J. Public Health 84, 1383–1393.
Murray, D.M., 1998. Design and Analysis of Group Randomized Trials. Oxford University Press Inc., New York.
Murray, D.M., 2001. Statistical models appropriate for designs often used in group-randomized trials. Stat. Med. 20,
1373–1385.
Murray, D.M., Blitstein, J.L., 2003. Methods to reduce the impact of intraclass correlation in group-randomized trials.
Eval. Rev. 27, 79–103.
Murray, D.M., Varnell, S.P., Blitstein, J.L., 2004. Design and analysis of group-randomized trials: a review of recent
methodological developments. Am. J. Public Health 94, 423–432.
Nassar, R., Mosier, J.E., 1980. Canine population dynamics: a study of the Manhattan, Kansas, canine population. Am. J.
Vet. Res. 41, 1798–1803.
Nassar, R., Mosier, J.E., 1982. Feline population dynamics: a study of the Manhattan, Kansas, feline population. Am. J.
Vet. Res. 43, 167–170.
Nassar, R., Mosier, J.E., 1986. Understanding the dynamics of your community’s pet population. Vet. Med. 81, 1120–
1126.
Nassar, R., Mosier, J.E., Williams, L.W., 1984. Study of the feline and canine populations in the greater Las Vegas area.
Am. J. Vet. Res. 45, 282–287.
Nassar, R., Talboy, J., Moulton, C., 1992. Animal Shelter Reporting Study—1990. American Humane Association,
Englewood, Colo.
New Jr., J.C., Salman, M.D., King, M., Scarlett, J.M., Kass, P.H., Hutchison, J.M., 2000. Characteristics of shelter-
relinquished animals and their owners compared with animals and their owners in U.S. pet-owning households. J.
Appl. Anim. Welf. Sci. 3, 179–201.
Patronek, G.J., Rowan, A.N., 1995. Editorial: determining dog and cat numbers and population dynamics. Anthrozoos 8,
199–204.
L.K. Lord et al. / Preventive Veterinary Medicine 82 (2007) 167–175174
Patronek, G.J., Glickman, L.T., Beck, A.M., McCabe, G.P., Ecker, C., 1996a. Risk factors for relinquishment of dogs to an
animal shelter. J. Am. Vet. Med. Assoc. 209, 572–581.
Patronek, G.J., Glickman, L.T., Beck, A.M., McCabe, G.P., Ecker, C., 1996b. Risk factors for relinquishment of cats to an
animal shelter. J. Am. Vet. Med. Assoc. 209, 582–588.
Rowan, A.N., Williams, J., 1987. The success of companion animal management programs: a review. Anthrozoos 1, 110–
122.
Salman, M.D., New Jr., J.G., Scarlett, J.M., Kass, P.H., 1998. Human and animal factors related to the relinquishment of
dogs and cats in 12 selected animal shelters in the United States. J. Appl. Anim. Welf. Sci. 1, 207–226.
Salman, M.D., Hutchison, J., Ruch-Gallie, R., Kogan, L., New Jr., J.G., Kass, P.H., Scarlett, J.M., 2000. Behavioral
reasons for relinquishment of dogs and cats to 12 shelters. J. Appl. Anim. Welf. Sci. 3, 93–106.
Scarlett, J.M., Salman, M.D., New Jr., J.G., Kass, P.H., 1999. Reasons for relinquishment of companion animals in U.S.
animal shelters: selected health and personal issues. J. Appl. Anim. Welf. Sci. 2, 41–57.
Schneider, R., Vaida, M.L., 1970. Survey of canine and feline populations: Alameda and Contra Costa counties,
California, 1970. J. Am. Vet. Med. Assoc. 166, 481–486.
Secker-Walker, R.H., Flynn, B.S., Solomon, L.J., 2006. Helping women quit smoking: results of a community
intervention program. Am. J. Public Health 90, 940–946.
Sorensen, G., Emmons, K., Hunt, M.K., Johnston, D., 1998. Implications of the results of community intervention trials.
Annu. Rev. Public Health 19, 379–416.
Spain, C.V., Scarlett, J.M., Houpt, K.A., 2004a. Long-term risks and benefits of early-age gonadectomy in cats. J. Am. Vet.
Med. Assoc. 224, 372–379.
Spain, C.V., Scarlett, J.M., Houpt, K.A., 2004b. Long-term risks and benefits of early-age gonadectomy in dogs. J. Am.
Vet. Med. Assoc. 224, 380–387.
Stevens, J., Murran, D.M., Catellier, D.J., Hannan, P.J., Lytele, L.A., Elder, J.P., Young, D.R., Simons-Morton, D.G.,
Webber, L.S., 2005. Design of the trial of activity in adolescent girls (TAAG). Cont. Clin. Trials 26, 223–233.
Thompson, B., Coronado, G., Snipes, S.A., Puschel, K., 2003. Methodologic advances and ongoing challenges in
designing community-based health promotion programs. Annu. Rev. Public Health 24, 315–340.
Wagenaar, A.C., Murray, D.M., Toomey, T.L., 2000. Communities Mobilizing for Change on Alcohol (CMCA): effects of
a randomized trial on arrest and traffic crashes. Addiction 95, 209–217.
L.K. Lord et al. / Preventive Veterinary Medicine 82 (2007) 167–175 175