- Email: [email protected]
Impact of demographic characteristics in pet ownership: Modeling animal count according to owners income and age
Preventive Veterinary Medicine 109 (2013) 213–218
Contents lists available at SciVerse ScienceDirect
Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed
Impact of demographic characteristics in pet ownership: Modeling animal count according to owners income and age Camila Marinelli Martins a,∗,1 , Ahmed Mohamed b,1 , Ana Marcia Sá Guimarães b , Cristiane da Conceic¸ão de Barros c , Raquel dos Santos Pampuch c , Walfrido Svoboda d , Rita de Cassia Maria Garcia e , Fernando Ferreira a , Alexander Welker Biondo d,2 a b c d e
Department of Preventive Veterinary Medicine, São Paulo University, São Paulo, SP 05508, Brazil Department of Comparative Pathobiology, Purdue University, West Lafayette, IN 47907, USA Department of Health Surveillance, City Secretary of Health, Pinhais, PR 83323, Brazil Departament of Veterinary Medicine, Federal University of Paraná, Curitiba, PR 80035, Brazil Executive Director of Technical Institute for Education and Animal Control, São Paulo, SP 06707, Brazil
a r t i c l e
i n f o
Article history: Received 21 March 2012 Received in revised form 4 October 2012 Accepted 13 October 2012 Keywords: Pet dog population Cat population Pet population management Pet ownership
a b s t r a c t Pet owner characteristics such as age, gender, income/social class, marital status, rural/urban residence and household type have been shown to be associated with the number of owned pets. However, few studies to date have attempted to evaluate these associations in Brazil. Accordingly, the aim of this study was to evaluate the association between age and income of owners and the number of owned dogs and cats in a Brazilian urban center. Pinhais, metropolitan area of Curitiba, Southern Brazil, the seventh largest city in Brazil, was chosen for this study. Questionnaires were administered door-to-door between January and February 2007 and data were analyzed by zero-inflated negative binomial (ZINB) models. A total of 13,555 of 30,380 (44.62%) households were interviewed. The majority (62.43%) of households reported having one or more dogs, with one or two dogs being the most common (29.97% and 19.71%, respectively). Cat ownership per household was much lower (P = 0.001) than dog ownership, with 90% of the households reported having no owned cats. ZINB analyses indicated that income is not associated with the number of both dogs and cats among households that have pets. However, households from higher income categories were more likely to have dogs (but not cats) when compared to the lowest income category (P < 0.05), contradicting a common belief that the poorer the family, the more likely they have pets. Certain age categories were significantly associated with the number of dogs or cats in households that have pets. In addition, most age categories were significantly associated with having dogs and/or cats (P < 0.05). In conclusion, our study has found that age but not household income is associated with the number of dogs or cats in households that have pets; higher income households were more likely to have dogs when compared to low-income households. © 2012 Elsevier B.V. All rights reserved.
∗ Corresponding author at: Departamento de Medicina Veterinária Preventiva e Saúde Animal, Faculdade de Medicina Veterinária e Zootecnia, Universidade de São Paulo – USP, Rua Prof Dr Orlando Marques de Paiva, 87 – Cidade Universitária, São Paulo, SP 05508-270, Brazil. Tel.: +55 11 3091 9937; fax: +55 11 3091 7928. E-mail address: [email protected] (C.M. Martins). 1 These authors equally contributed for this study. 2 He is also a visiting professor in the Department of Veterinary Pathobiology, University of Illinois, Urbana, IL 61802, USA. 0167-5877/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.prevetmed.2012.10.006
214
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218
1. Introduction
2.2. Questionnaires
In recent years, several studies have been conducted in Southern and Southeastern Brazil in an effort to estimate the number of owned dogs and cats in urban centers (Serafini et al., 2008; Nunes et al., 1997; Lima Júnior, 1999; Dias, 2001; Paranhos, 2002; Dias et al., 2004; Alves et al., 2005). These estimates have been used to accurately plan and monitor government investments in public health services, such as rabies and animal control. Besides being lower than estimates of the World Health Organization for developing countries, human:dog ratios vary greatly among different areas of the country (3:1–13:1 for owned dogs) (Serafini et al., 2008; Nunes et al., 1997; Lima Júnior, 1999; Dias, 2001; Paranhos, 2002; Dias et al., 2004; Alves et al., 2005). The same variation has been observed among different owned cat populations (7:1–86:1 for owned cats); however, fewer studies have been undertaken on them cats when compared to dogs (Paranhos, 2002; Dias et al., 2004; Garcia, 2009). It is likely that various demographic and socioeconomic characteristics of the human population in different regions may be associated with and influence the number of owned dogs and cats. Human demographic and socioeconomic characteristics associated with the number of owned pets have been extensively investigated worldwide (Endenburg et al., 1990; Leslie et al., 1994; Downes et al., 2009; Murray et al., 2010). Although the evaluated variables and associations differ greatly among studies, factors such as age, gender, income/social class, marital status, rural/urban residence and household type may be associated with pet ownership (Marx et al., 1988; Endenburg et al., 1990; Leslie et al., 1994; Downes et al., 2009). Similar studies in Brazil are scarce; household type was recently found to be associated with the number of owned dogs and cats in a Brazilian urban center (Serafini et al., 2008). In this previous study, family income was suggested as responsible for differences in human:pet ratios among neighborhoods (Serafini et al., 2008). Since data on income and age of the human population may be routinely accessed by the Brazilian Department of Population Demographics, the objective of this study was to evaluate the association between these variables and the number of owned dogs and cats per household in an urban center.
Questionnaires were administered door-to-door between January and February 2007 by trained health agents of the Center of Zoonosis Control (CCZ) of Pinhais City as part of a non-related study (National Dengue Prevention Program). All houses were visited. Regarding the three apartment building complexes, one apartment in each of the 24 4-story buildings was randomly chosen to be interviewed and counted as a household unit. All visits occurred during work hours (08:00 AM–5:00 PM). Questions asked per household for this study included the age of each household member, household income, and number of dogs and cats. Age information of household members was recorded as 6 independent variables (<1, 1–<10, 10–<20, 20–<30, 30–<40, ≥40 years old). Interviewees were asked the age of every person living in the household. The interviewers then recorded the number of people in each age variable for each household. Based on the Brazilian monthly minimum wage (MW) of R$380.00 (U$182.17) at the time, income was divided in 4 categories (<1, 1–<2, 2–<4 and ≥4 MW). Pet ownership was defined as claiming a dog or cat as one’s or family’s own instead of just having a pet in the household at the time of the study. Pet ownership was assessed irrespective of where the pets were kept (such as house, backyard, barn) or if they were allowed to roam or kept indoor or outdoor. Data was recorded per household and organized in Excel sheets separated by neighborhoods.
2. Materials and methods 2.1. Study site Pinhais city is part of Curitiba County, capital of Paraná State, southern Brazil, the 7th most populated city in Brazil with a total of 3.16 million people within the metropolitan area. At the time of the sampling, Pinhais had approximately 117,166 inhabitants (IBGE, 2010) distributed in 30,680 households in 15 neighborhoods. There were only three 4-story apartment building complexes (complexes 1, 2 and 3 had nine, ten and five 4-story buildings, respectively); the remaining households were houses (R. Lacerda, personal communication, Department of Urban Planning, Pinhais City Hall).
2.3. Statistical analysis Initial analysis of the data indicated that the distributions of dogs and cats per household were severely right skewed for all neighborhoods and overall (Supplement 1). Therefore, median and ranges of owned dogs and cats in each neighborhood were calculated as descriptive statistics. A non-parametric method (Kruskal–Wallis) was used to test the hypothesis of equal distribution. A post hoc multiple comparison test of the number of pets/household between neighborhoods was performed using the SAS macro KW MC (Elliott and Hynan, 2011). This macro is a convenient way to perform multiple comparisons analysis for non-normally distributed data. It is designed to implement Dunn’s multiple comparison procedure (Dunn, 1964) in cases of unequal sample sizes or when ties are present. Using this procedure, a q statistics was calculated based on the standard errors (SE) for the comparison between two groups (Zar, 2010) and A type 1 error of 0.05 was used. About 38% and 90% of the households had no dogs or cats, respectively. In addition to this excess of zeros, an overdispersed distribution of the data was also observed; mean count and variance were 1.18, 1.83 and 0.17, 0.51 for dogs and cats per household, respectively. Therefore, these distributions are unlikely to fit classical or zeroinflated Poisson distributions (PD). Although ignoring the over-dispersion may not impact the parameters estimates, standard errors may be underestimated and result in inaccurate p-values. On the other hand, assuming PD for a count data with negative binomial distribution may lead to inconsistent parameter estimates. Zero-inflated
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218 Table 1 Count frequency, proportion and central tendency parameters for the owned pet count in the city of Pinhais, Brazil, 2007. Dogs Count 0 1 2 3 4 ≥5 Total
Cats Freq. 5092 4063 2672 1039 382 307 13,555
Percent 37.57 29.97 19.71 7.67 2.82 2.25 100
Count
Freq.
Percent
0 1 2 3 4 ≥5
12,196 878 278 90 37 77
89.97 6.48 2.05 0.66 0.27 0.56
13,556
100
Central tendency parameters among all households 1 0 Median 0 0 Min 16 15 Max Central tendency parameters among dogs, and cat owning households 2 1 Median Min 1 1 16 15 Max
Poisson (ZIP) may account for excess zeroes beyond what is expected by PD but does not account for over-dispersion. Negative binomial model accounts for data overdispersion but does not for the excess zeroes. Based on these criteria and the graphic illustration of prediction obtained from Poisson, zero-inflated negative binomial (ZINB) was determined to be the best analytical approach. All the analyses, except multiple comparison tests, were carried out using Stata 11.2 (StataCorp, College Station, TX, USA), and a pvalue < 0.05 was considered significant. 3. Results All 15 neighborhoods were visited with the exception of one small condominium (300 households), which refused the visits, and an environmentally protected area with no residents (Parque das Águas neighborhood). All the remaining households in the city were visited, resulting in a total of 13,555/30,380 (44.62%) interviews. The other 16,825 households/families either refused to be interviewed or were absent at the time of the visit. The number of owned pets per household ranged from 0 to 16 for dogs and 0 to 15 for cats, with median of 1 and zero, respectively (Table 1). Among pet-owning households, median number of owned dogs was 2 (1–16) and cats 1 (1–15). The majority (62.43%) of households reported having at least one dog, with one or two dogs being the most frequent scenario (29.97% and 19.71% of the households, respectively) (Table 1). Cat ownership per household was much lower (P = 0.0001) than dog ownership, with approximately 90% of the households having no cats (Table 2). Households with one or two cats were also more frequently observed (6.48% and 2.05%, respectively). In contrast to cats, the number of dogs per household was statistically different between neighborhoods (P = 0.0001) (Table 2). The multiple comparisons test for the number of dogs per household showed that 26 different combinations were statistically different (Table S2). Zero-negative binomial (ZINB) model were found to be the appropriate approach for analyzing the count of
215
pet (dogs and cats) ownership per household. The Poisson model poorly fit the data distribution (due to overdispersion) compared to the negative binomial regression. The ZINB model was chosen based on the highly significant Voung statistic (Z = 6.74, p-value < 0.0001 for dogs, Z = 3.18, p-value = 0.0007 for cats) and because the dispersion parameter alpha was significantly different from zero in both models for owned dog 0.25 (0.22–0.29) and owned cat counts 6.30 (5.43–7.39) (Table 3). The effect of neighborhood was considered but was not included in the final model since no statistical significance was found in the final count model (data not shown). The ZINB consisted of two parts: a negative binomial (NB) to model the pet count process among households with owned pets, and a logistic regression (logit) to model the probability of having zero owned pet counts (no dogs or cats). The NB model showed that income categories are not associated with the number of both dogs and cats among households that have pets (P > 0.05). However, the logit indicated that households in higher income categories were more likely to have at least one dog (P < 0.05), but not cats (P > 0.05), when compared with households of <1 MW. In other words, the odds of household within the income category <1 MW having a zero dog count were approximately 2.04, 2.38 and 2.13 times higher than the odds for income categories 1–<2 MW, 2–<4 MW and >4 MW, respectively (Table 3). When considering age, the NB model indicated that the all age variables, except <1 and 20–<30 years old, were associated with the number of dogs per household, whereas age variables of 20 years and older were significantly associated with the number of cats per household (P < 0.05). Or equivalently, the expected number of owned dogs would increase on average by 1.03, 1.09, 1.1 and 1.12 for each additional member of 1–<10, 10–<20, 30–<40 and ≥40 years old, respectively. For owned cats, the expected count per household would increase on average by 1.12, 1.22 and 1.22 cats for each additional member of 20–<30, 30–<40 and ≥40 years old. The logit results showed that all age categories, with the exception of category <1 year old, were significantly associated with having at least one dog (P < 0.05). The odds that a household would have zero owned dogs decrease by 2, 3.12, 4.54, 10, and 12.5 for each additional household member in the age variables of 1–<10 to ≥40 years, respectively. The relationship between age variables and pet count was different for owned cats; only variables 1–<10, 10–<20 and 20–<30 years old were less likely to have zero cats. For each additional household member in these age variables, the odds that this household would have zero cats would decrease by 1.39, 2.27, and 0.75, respectively (all exact pvalues of NB and logit analyses are shown in Table 3). 4. Discussion Our study has shown a much higher ownership preference for dogs compared to cats, as previously observed in Brazil (Paranhos, 2002; Dias et al., 2004; Garcia, 2009). A total of 8463/13,555 (62%) households had at least one dog, whereas only 1360/13,555 (10%) showed at least one cat. This finding is in contrast to developed countries, which
216
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218
Table 2 Summary statistics for the number of owned cats and dogs in the each neighborhood of Pinhais city, Brazil. Neighborhood
Alto Tarumã Atuba Centro Emiliano Perneta Estância Pinhais Jardim Karla Jd. Amélia Jd. Cláudia Maria Antonieta Parque das Nascentes Pineville Vargem Grande Weissópolis Total
Dogs
Cats
Median
Q1
Q3
Min
Max
Median
Q1
Q3
Min
Max
1 1 1 1 1 1 1 1 1 2 1 1 1 1
0 0 0 0 0 0 0 0 0 1 0 0 0 0
2 2 2 2 2 2 2 2 2 3 2 2 2 2
0 0 0 0 0 0 0 0 0 0 0 0 0 0
10 14 11 14 15 9 9 12 16 13 11 10 11 16
0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0
9 6 10 12 11 10 8 6 7 6 11 15 15 15
Kruskal–Wallis equality-of-populations rank, P-value = 0.0001.
reported similar distributions of dogs and cats, with 22% of the households having dogs and 18% having cats in the UK (PFMA, 2007), and 37% having dogs and 32% having cats in the USA (SAUS, 2011). Although further studies should be conducted to fully establish the pet species preference in Brazil, cultural beliefs and background may play a role in such a scenario (Downes et al., 2009). The number of dogs per household, but not cats, was different between various neighborhoods. A possible
explanation for this disparity is the fact that neighborhoods from Pinhais city vary in respect to human population density and household occupation. Neighborhoods are more or less densely populated (R. Lacerda, personal communication, Department of Urban Planning, Pinhais city), which may result in differences in the number of dogs/household (Serafini et al., 2008). Regardless of such speculations, studies should be performed to evaluate the impact of such demographic differences between neighborhoods on the
Table 3 Results of ZNIB for owned pet count in the city of Pinhais, Brazil, 2007. Negative Binomial Section
Dogs
Variable
Coef.
Cats P>z
[95% Conf.
Coef.
P>z
[95% Conf.
Income (multiples of one minimum salary) 1.00 <1 MW 0.96 1–<2 MW 1.01 2–<4 MW ≥4 MW 1.04
NA 0.421 0.895 0.48
Reference 0.87–1.06 0.91–1.11 0.93–1.16
1.00 1.01 1.08 1.13
NA 0.95 0.68 0.57
Reference 0.71–1.45 0.75–1.57 0.75–1.70
Age (years) <1 1–<10 10–<20 20–<30 30–<40 ≥40
0.99 1.03 1.09 0.99 1.1 1.12
0.704 0.012 <0.001 0.501 <0.001 <0.001
0.91–1.06 1.01–10.5 1.07–1.12 0.97–1.01 1.07–1.12 1.09–1.16
1.03 1.03 1.05 1.12 1.22 1.22
0.83 0.52 0.3 0.01 0.01 <0.001
0.77–1.39 0.94–1.13 0.96–1.16 1.03–1.23 1.06–1.40 1.07–1.39
Logistic section Income (multiples of one MW) <1 1–<2 2–<4 ≥4
1.00 0.49 0.42 0.47
NA 0.006 0.004 0.03
Reference 0.29–0.81 0.23–0.76 0.23–0.93
1.00 1.37 1.27 1.90
NA 0.66 0.75 0.4
Reference 0.34–5.51 0.29–5.48 0.43–8.53
Age (years) <1 1–<10 10–<20 20–<30 30–<40 ≥40 Cons
0.83 0.5 0.32 0.22 0.1 0.08 9.73
0.652 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
0.37–1.87 0.36–0.69 0.23–0.44 0.15–0.31 0.05–0.17 0.04–0.15 5.01–18.90
1.19 0.72 0.44 1.33 0.79 <0.001 0.44
0.62 0.04 0.01 0.02 0.39 0.98 0.33
0.60–2.39 0.53–0.98 0.23–0.84 1.04–1.70 0.46–1.35 0 0.09–2.25
/lnalpha Alpha
−1.38 0.25
<0.001
−1.50–−1.25 0.22–0.29
1.84 6.3
<0.001
1.69–1.99 5.43–7.39
Vuong test
Z = 6.74
<0.0001
Z = 3.18
0.0007
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218
number of dogs/household that may aid in developing targeted educational strategies for pet population control. Our study also found that owner income was not associated with the number of dogs or cats in households that already have pets. However, households within the income category <1 MW were significantly more likely to have zero dogs compared to households higher income categories (1–<2, 2–<4 and ≥4 MW). Therefore, income may not be a good indicator of the number of dogs or cats among pet owners despite its association with having or not a dog. A similar finding was observed in Netherlands, where animal owners tended to have higher incomes than non-owners (Endenburg et al., 1990), but this study did not report the probability of zero pets. Most of the educational initiatives of animal control in this area are concentrated on poor families that already have pets (AWB, personal communication). Our results show that, ideally, other income categories should also be targeted. In contrast to income, our study showed multiple age variables associated with pet ownership/count. A similar report in the United States also found an association between the number of dogs and owner’s age within a category of 25–44 years old (Ramón et al., 2010). The fact that only the age variables between 1 and 30 years old were less likely to have zero cats may imply increasing cat ownership in this age group. Finally, age may aid in modeling pet count in future studies, as different age groups were associated with the number of dogs and cats owned. A potential limitation of this study is that age was recorded as categories. Future studies should explore age as categories and take into consideration the number and age of members in each household. It is noteworthy that the high number of owned dogs and cats may not be necessarily correlated with the fact that owners allow their animals to roam. The roaming of owned dogs may be connected to income and education level and needs to be further explored. Free-roaming owned dogs, as well as loosely owned or community owned animals, may be then responsible for increasing the stray dog population size. A limitation of this study is response bias with respect to the response rate. Although the reason why certain households refused to be interviewed was not registered, interviewers later reported that certain people refused to be interviewed once they were told the expected time to complete the interview (sometimes more than 30 min – the team was responsible for evaluating the conditions of the household with respect to dengue prevention) (R. Lacerda, personal communication, Department of Urban Planning, Pinhais City Hall). The refusal and the absence of people at the time of the visit may imply busy lifestyles, which in turn, may somehow affect the presence and number of pets in a household. Since vertical housing has been associated with a low number of pets in one Brazilian neighborhood (Serafini et al., 2008), the number of animals in the 3 apartment building complexes may be lower than the average for the city. However, the lack of representativeness of these apartment buildings with respect to the overall household population (approximately 1%; data not shown) is unlikely to bias with the results of the present study.
217
5. Conclusion The count data analysis methods used in this study may be applied to data from other Brazilian cities, and probably in urban areas, as it takes into account people that have and do not have pets. As the number of household members of given age categories increase in certain households, one can expect an increase in the number of pets. Therefore, the results of this study can be used to plan public health initiatives and educational programs in Pinhais city based on age stratification. Interventions can target people that still do not have pets and households in higher income categories, so as to help in the decision to acquire or care for a pet. Conflict of interest None. Acknowledgements Funding for Dr. Camila Martins’s research fellowship was provided by the Tesouro Nacional – Universidade Federal do Paraná – UFPR. We kindly thank the help given by the Pinhais Center of Zoonosis Control (CCZ) personnel for the interviews and data collection. 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.prevetmed.2012.10.006. References Alves, M.C.G.P., Matos, M.R., Reichmann, M.L., Dominguez, M.H., 2005. Estimation of the dog and cat population in the state of Sao Paulo. Rev. Saude Publica 39, 1–7. Dias, R.A., 2001. Emprego do sistema de informac¸ão geográfica (SIG) no controle da raiva canina. Dissertac¸ão de Mestrado, Universidade de São Paulo, São Paulo. Dias, R.A., Garcia, R., de, C., Silva, D.F., Amaku, M., Ferreira Neto, J.S., Ferreira, F., 2004. Estimativa de populac¸ões canina e felina domiciliadas em zona urbana do Estado de São Paulo. Rev. Saude Publica 38, 565–570. Downes, M.J., Canty, M.J., More, S.J., 2009. Demography of the pet dog and cat population on the island of Ireland and Human factors influencing pet ownership. Prev. Vet. Med. 92, 140–149. Dunn, O.J., 1964. Multiple contrasts using rank sums. Technometrics 6, 241–252. Elliott, A.C., Hynan, L.S., 2011. A SAS® macro implementation of a multiple comparison post hoc test for a Kruskal–Wallis analysis. Computer methods and programs in biomedicine 102, 75–80. Endenburg, N., Hart, H., de Vries, H.W., 1990. Differences between owners and nonowners of companion animals. Anthozoos IV (2), 120–126. Garcia, R.de C.M., 2009. Estudo da dinâmica populacional canina e felina e avaliac¸ão de ac¸ões para o equilíbrio dessas populac¸ões em área da cidade de São Paulo, SP, Brasil. Tese de Doutorado, Universidade de São Paulo, São Paulo. IBGE, Instituto Brasileiro de Geografia e Estatística, 2010. Disponível em http://www.ibge.com.br/home/estatistica/populacao/censo2010/ tabelas pdf/total populacao parana.pdf (acesso em 04.02.11). Leslie, B.E., Meek, A.H., Kawash, G.F., McKeown, D.B., 1994. An epidemiological investigation of pet ownership in Otario. Can. Vet. J. 35, 218–222. Lima Júnior, A.D., 1999. Dinâmica populacional canina e a persistência da raiva na cidade de Recife (PE), nordeste do Brasil. Dissertac¸ão de Mestrado, Universidade de São Paulo, São Paulo.
218
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218
Marx, M.B., Stallones, L., Garrity, T.F., Johnson, T.P., 1988. Demographics of pet ownership among U.S. adults 21–64 years of age. Anthrozoos II 1, 33–37. Murray, J.K., Browne, W.J., Roberts, M.A., Whitmarsh, A., Gruffydd-Jones, T.J., 2010. Numbers and ownership profiles of cats and dogs in the UK. Vet. Rec. 166, 163–168. Nunes, C.M., Martines, D.A., Fikaris, S., Queiroz, L.H., 1997. Avaliac¸ão da populac¸ão canina da zona urbana do Município de Arac¸atuba, São Paulo, SP, Brasil. Rev. Saude Publica 31, 308–309. Paranhos, N.T., 2002. Estudo das populac¸ões canina e felina em domicílio, município de São Paulo, 2001. Dissertac¸ão de Mestrado, Universidade de São Paulo, São Paulo. Pet Food Manufacturers’ Association (PFMA) 2007. Pet population. Available at: http://www.pfma.org.uk/pet-population-2011/ (accessed 03.05.12).
Ramón, M.E., Slater, M.R., Ward, M.P., 2010. Companion animal knowledge, attachment and pet cat care and their associations with household demographics for residents of a rural Texas town. Prev. Vet. Med. 94, 251–263. Serafini, C.A.V., Rosa, G.A., Guimarães, A.M.S., de Morais, H.A., Biondo, A.W., 2008. Survey of Owned Feline and Canine populations n Apartments from a Neighbourhood in Curitiba, Brazil. Zoonoses Public Health 55, 402–405. Statistical Abstract of the United States - Census Bureau (SAUS), 2011. Available at: http://www.census.gov/prod/2011pubs/ 11statab/arts.pdf (acessed 30.10.12). Zar, J.H., 2010. Biostatistical Analysis, 5th ed. Prentice Hall, p. 241ff.
Contents lists available at SciVerse ScienceDirect
Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed
Impact of demographic characteristics in pet ownership: Modeling animal count according to owners income and age Camila Marinelli Martins a,∗,1 , Ahmed Mohamed b,1 , Ana Marcia Sá Guimarães b , Cristiane da Conceic¸ão de Barros c , Raquel dos Santos Pampuch c , Walfrido Svoboda d , Rita de Cassia Maria Garcia e , Fernando Ferreira a , Alexander Welker Biondo d,2 a b c d e
Department of Preventive Veterinary Medicine, São Paulo University, São Paulo, SP 05508, Brazil Department of Comparative Pathobiology, Purdue University, West Lafayette, IN 47907, USA Department of Health Surveillance, City Secretary of Health, Pinhais, PR 83323, Brazil Departament of Veterinary Medicine, Federal University of Paraná, Curitiba, PR 80035, Brazil Executive Director of Technical Institute for Education and Animal Control, São Paulo, SP 06707, Brazil
a r t i c l e
i n f o
Article history: Received 21 March 2012 Received in revised form 4 October 2012 Accepted 13 October 2012 Keywords: Pet dog population Cat population Pet population management Pet ownership
a b s t r a c t Pet owner characteristics such as age, gender, income/social class, marital status, rural/urban residence and household type have been shown to be associated with the number of owned pets. However, few studies to date have attempted to evaluate these associations in Brazil. Accordingly, the aim of this study was to evaluate the association between age and income of owners and the number of owned dogs and cats in a Brazilian urban center. Pinhais, metropolitan area of Curitiba, Southern Brazil, the seventh largest city in Brazil, was chosen for this study. Questionnaires were administered door-to-door between January and February 2007 and data were analyzed by zero-inflated negative binomial (ZINB) models. A total of 13,555 of 30,380 (44.62%) households were interviewed. The majority (62.43%) of households reported having one or more dogs, with one or two dogs being the most common (29.97% and 19.71%, respectively). Cat ownership per household was much lower (P = 0.001) than dog ownership, with 90% of the households reported having no owned cats. ZINB analyses indicated that income is not associated with the number of both dogs and cats among households that have pets. However, households from higher income categories were more likely to have dogs (but not cats) when compared to the lowest income category (P < 0.05), contradicting a common belief that the poorer the family, the more likely they have pets. Certain age categories were significantly associated with the number of dogs or cats in households that have pets. In addition, most age categories were significantly associated with having dogs and/or cats (P < 0.05). In conclusion, our study has found that age but not household income is associated with the number of dogs or cats in households that have pets; higher income households were more likely to have dogs when compared to low-income households. © 2012 Elsevier B.V. All rights reserved.
∗ Corresponding author at: Departamento de Medicina Veterinária Preventiva e Saúde Animal, Faculdade de Medicina Veterinária e Zootecnia, Universidade de São Paulo – USP, Rua Prof Dr Orlando Marques de Paiva, 87 – Cidade Universitária, São Paulo, SP 05508-270, Brazil. Tel.: +55 11 3091 9937; fax: +55 11 3091 7928. E-mail address: [email protected] (C.M. Martins). 1 These authors equally contributed for this study. 2 He is also a visiting professor in the Department of Veterinary Pathobiology, University of Illinois, Urbana, IL 61802, USA. 0167-5877/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.prevetmed.2012.10.006
214
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218
1. Introduction
2.2. Questionnaires
In recent years, several studies have been conducted in Southern and Southeastern Brazil in an effort to estimate the number of owned dogs and cats in urban centers (Serafini et al., 2008; Nunes et al., 1997; Lima Júnior, 1999; Dias, 2001; Paranhos, 2002; Dias et al., 2004; Alves et al., 2005). These estimates have been used to accurately plan and monitor government investments in public health services, such as rabies and animal control. Besides being lower than estimates of the World Health Organization for developing countries, human:dog ratios vary greatly among different areas of the country (3:1–13:1 for owned dogs) (Serafini et al., 2008; Nunes et al., 1997; Lima Júnior, 1999; Dias, 2001; Paranhos, 2002; Dias et al., 2004; Alves et al., 2005). The same variation has been observed among different owned cat populations (7:1–86:1 for owned cats); however, fewer studies have been undertaken on them cats when compared to dogs (Paranhos, 2002; Dias et al., 2004; Garcia, 2009). It is likely that various demographic and socioeconomic characteristics of the human population in different regions may be associated with and influence the number of owned dogs and cats. Human demographic and socioeconomic characteristics associated with the number of owned pets have been extensively investigated worldwide (Endenburg et al., 1990; Leslie et al., 1994; Downes et al., 2009; Murray et al., 2010). Although the evaluated variables and associations differ greatly among studies, factors such as age, gender, income/social class, marital status, rural/urban residence and household type may be associated with pet ownership (Marx et al., 1988; Endenburg et al., 1990; Leslie et al., 1994; Downes et al., 2009). Similar studies in Brazil are scarce; household type was recently found to be associated with the number of owned dogs and cats in a Brazilian urban center (Serafini et al., 2008). In this previous study, family income was suggested as responsible for differences in human:pet ratios among neighborhoods (Serafini et al., 2008). Since data on income and age of the human population may be routinely accessed by the Brazilian Department of Population Demographics, the objective of this study was to evaluate the association between these variables and the number of owned dogs and cats per household in an urban center.
Questionnaires were administered door-to-door between January and February 2007 by trained health agents of the Center of Zoonosis Control (CCZ) of Pinhais City as part of a non-related study (National Dengue Prevention Program). All houses were visited. Regarding the three apartment building complexes, one apartment in each of the 24 4-story buildings was randomly chosen to be interviewed and counted as a household unit. All visits occurred during work hours (08:00 AM–5:00 PM). Questions asked per household for this study included the age of each household member, household income, and number of dogs and cats. Age information of household members was recorded as 6 independent variables (<1, 1–<10, 10–<20, 20–<30, 30–<40, ≥40 years old). Interviewees were asked the age of every person living in the household. The interviewers then recorded the number of people in each age variable for each household. Based on the Brazilian monthly minimum wage (MW) of R$380.00 (U$182.17) at the time, income was divided in 4 categories (<1, 1–<2, 2–<4 and ≥4 MW). Pet ownership was defined as claiming a dog or cat as one’s or family’s own instead of just having a pet in the household at the time of the study. Pet ownership was assessed irrespective of where the pets were kept (such as house, backyard, barn) or if they were allowed to roam or kept indoor or outdoor. Data was recorded per household and organized in Excel sheets separated by neighborhoods.
2. Materials and methods 2.1. Study site Pinhais city is part of Curitiba County, capital of Paraná State, southern Brazil, the 7th most populated city in Brazil with a total of 3.16 million people within the metropolitan area. At the time of the sampling, Pinhais had approximately 117,166 inhabitants (IBGE, 2010) distributed in 30,680 households in 15 neighborhoods. There were only three 4-story apartment building complexes (complexes 1, 2 and 3 had nine, ten and five 4-story buildings, respectively); the remaining households were houses (R. Lacerda, personal communication, Department of Urban Planning, Pinhais City Hall).
2.3. Statistical analysis Initial analysis of the data indicated that the distributions of dogs and cats per household were severely right skewed for all neighborhoods and overall (Supplement 1). Therefore, median and ranges of owned dogs and cats in each neighborhood were calculated as descriptive statistics. A non-parametric method (Kruskal–Wallis) was used to test the hypothesis of equal distribution. A post hoc multiple comparison test of the number of pets/household between neighborhoods was performed using the SAS macro KW MC (Elliott and Hynan, 2011). This macro is a convenient way to perform multiple comparisons analysis for non-normally distributed data. It is designed to implement Dunn’s multiple comparison procedure (Dunn, 1964) in cases of unequal sample sizes or when ties are present. Using this procedure, a q statistics was calculated based on the standard errors (SE) for the comparison between two groups (Zar, 2010) and A type 1 error of 0.05 was used. About 38% and 90% of the households had no dogs or cats, respectively. In addition to this excess of zeros, an overdispersed distribution of the data was also observed; mean count and variance were 1.18, 1.83 and 0.17, 0.51 for dogs and cats per household, respectively. Therefore, these distributions are unlikely to fit classical or zeroinflated Poisson distributions (PD). Although ignoring the over-dispersion may not impact the parameters estimates, standard errors may be underestimated and result in inaccurate p-values. On the other hand, assuming PD for a count data with negative binomial distribution may lead to inconsistent parameter estimates. Zero-inflated
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218 Table 1 Count frequency, proportion and central tendency parameters for the owned pet count in the city of Pinhais, Brazil, 2007. Dogs Count 0 1 2 3 4 ≥5 Total
Cats Freq. 5092 4063 2672 1039 382 307 13,555
Percent 37.57 29.97 19.71 7.67 2.82 2.25 100
Count
Freq.
Percent
0 1 2 3 4 ≥5
12,196 878 278 90 37 77
89.97 6.48 2.05 0.66 0.27 0.56
13,556
100
Central tendency parameters among all households 1 0 Median 0 0 Min 16 15 Max Central tendency parameters among dogs, and cat owning households 2 1 Median Min 1 1 16 15 Max
Poisson (ZIP) may account for excess zeroes beyond what is expected by PD but does not account for over-dispersion. Negative binomial model accounts for data overdispersion but does not for the excess zeroes. Based on these criteria and the graphic illustration of prediction obtained from Poisson, zero-inflated negative binomial (ZINB) was determined to be the best analytical approach. All the analyses, except multiple comparison tests, were carried out using Stata 11.2 (StataCorp, College Station, TX, USA), and a pvalue < 0.05 was considered significant. 3. Results All 15 neighborhoods were visited with the exception of one small condominium (300 households), which refused the visits, and an environmentally protected area with no residents (Parque das Águas neighborhood). All the remaining households in the city were visited, resulting in a total of 13,555/30,380 (44.62%) interviews. The other 16,825 households/families either refused to be interviewed or were absent at the time of the visit. The number of owned pets per household ranged from 0 to 16 for dogs and 0 to 15 for cats, with median of 1 and zero, respectively (Table 1). Among pet-owning households, median number of owned dogs was 2 (1–16) and cats 1 (1–15). The majority (62.43%) of households reported having at least one dog, with one or two dogs being the most frequent scenario (29.97% and 19.71% of the households, respectively) (Table 1). Cat ownership per household was much lower (P = 0.0001) than dog ownership, with approximately 90% of the households having no cats (Table 2). Households with one or two cats were also more frequently observed (6.48% and 2.05%, respectively). In contrast to cats, the number of dogs per household was statistically different between neighborhoods (P = 0.0001) (Table 2). The multiple comparisons test for the number of dogs per household showed that 26 different combinations were statistically different (Table S2). Zero-negative binomial (ZINB) model were found to be the appropriate approach for analyzing the count of
215
pet (dogs and cats) ownership per household. The Poisson model poorly fit the data distribution (due to overdispersion) compared to the negative binomial regression. The ZINB model was chosen based on the highly significant Voung statistic (Z = 6.74, p-value < 0.0001 for dogs, Z = 3.18, p-value = 0.0007 for cats) and because the dispersion parameter alpha was significantly different from zero in both models for owned dog 0.25 (0.22–0.29) and owned cat counts 6.30 (5.43–7.39) (Table 3). The effect of neighborhood was considered but was not included in the final model since no statistical significance was found in the final count model (data not shown). The ZINB consisted of two parts: a negative binomial (NB) to model the pet count process among households with owned pets, and a logistic regression (logit) to model the probability of having zero owned pet counts (no dogs or cats). The NB model showed that income categories are not associated with the number of both dogs and cats among households that have pets (P > 0.05). However, the logit indicated that households in higher income categories were more likely to have at least one dog (P < 0.05), but not cats (P > 0.05), when compared with households of <1 MW. In other words, the odds of household within the income category <1 MW having a zero dog count were approximately 2.04, 2.38 and 2.13 times higher than the odds for income categories 1–<2 MW, 2–<4 MW and >4 MW, respectively (Table 3). When considering age, the NB model indicated that the all age variables, except <1 and 20–<30 years old, were associated with the number of dogs per household, whereas age variables of 20 years and older were significantly associated with the number of cats per household (P < 0.05). Or equivalently, the expected number of owned dogs would increase on average by 1.03, 1.09, 1.1 and 1.12 for each additional member of 1–<10, 10–<20, 30–<40 and ≥40 years old, respectively. For owned cats, the expected count per household would increase on average by 1.12, 1.22 and 1.22 cats for each additional member of 20–<30, 30–<40 and ≥40 years old. The logit results showed that all age categories, with the exception of category <1 year old, were significantly associated with having at least one dog (P < 0.05). The odds that a household would have zero owned dogs decrease by 2, 3.12, 4.54, 10, and 12.5 for each additional household member in the age variables of 1–<10 to ≥40 years, respectively. The relationship between age variables and pet count was different for owned cats; only variables 1–<10, 10–<20 and 20–<30 years old were less likely to have zero cats. For each additional household member in these age variables, the odds that this household would have zero cats would decrease by 1.39, 2.27, and 0.75, respectively (all exact pvalues of NB and logit analyses are shown in Table 3). 4. Discussion Our study has shown a much higher ownership preference for dogs compared to cats, as previously observed in Brazil (Paranhos, 2002; Dias et al., 2004; Garcia, 2009). A total of 8463/13,555 (62%) households had at least one dog, whereas only 1360/13,555 (10%) showed at least one cat. This finding is in contrast to developed countries, which
216
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218
Table 2 Summary statistics for the number of owned cats and dogs in the each neighborhood of Pinhais city, Brazil. Neighborhood
Alto Tarumã Atuba Centro Emiliano Perneta Estância Pinhais Jardim Karla Jd. Amélia Jd. Cláudia Maria Antonieta Parque das Nascentes Pineville Vargem Grande Weissópolis Total
Dogs
Cats
Median
Q1
Q3
Min
Max
Median
Q1
Q3
Min
Max
1 1 1 1 1 1 1 1 1 2 1 1 1 1
0 0 0 0 0 0 0 0 0 1 0 0 0 0
2 2 2 2 2 2 2 2 2 3 2 2 2 2
0 0 0 0 0 0 0 0 0 0 0 0 0 0
10 14 11 14 15 9 9 12 16 13 11 10 11 16
0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0
9 6 10 12 11 10 8 6 7 6 11 15 15 15
Kruskal–Wallis equality-of-populations rank, P-value = 0.0001.
reported similar distributions of dogs and cats, with 22% of the households having dogs and 18% having cats in the UK (PFMA, 2007), and 37% having dogs and 32% having cats in the USA (SAUS, 2011). Although further studies should be conducted to fully establish the pet species preference in Brazil, cultural beliefs and background may play a role in such a scenario (Downes et al., 2009). The number of dogs per household, but not cats, was different between various neighborhoods. A possible
explanation for this disparity is the fact that neighborhoods from Pinhais city vary in respect to human population density and household occupation. Neighborhoods are more or less densely populated (R. Lacerda, personal communication, Department of Urban Planning, Pinhais city), which may result in differences in the number of dogs/household (Serafini et al., 2008). Regardless of such speculations, studies should be performed to evaluate the impact of such demographic differences between neighborhoods on the
Table 3 Results of ZNIB for owned pet count in the city of Pinhais, Brazil, 2007. Negative Binomial Section
Dogs
Variable
Coef.
Cats P>z
[95% Conf.
Coef.
P>z
[95% Conf.
Income (multiples of one minimum salary) 1.00 <1 MW 0.96 1–<2 MW 1.01 2–<4 MW ≥4 MW 1.04
NA 0.421 0.895 0.48
Reference 0.87–1.06 0.91–1.11 0.93–1.16
1.00 1.01 1.08 1.13
NA 0.95 0.68 0.57
Reference 0.71–1.45 0.75–1.57 0.75–1.70
Age (years) <1 1–<10 10–<20 20–<30 30–<40 ≥40
0.99 1.03 1.09 0.99 1.1 1.12
0.704 0.012 <0.001 0.501 <0.001 <0.001
0.91–1.06 1.01–10.5 1.07–1.12 0.97–1.01 1.07–1.12 1.09–1.16
1.03 1.03 1.05 1.12 1.22 1.22
0.83 0.52 0.3 0.01 0.01 <0.001
0.77–1.39 0.94–1.13 0.96–1.16 1.03–1.23 1.06–1.40 1.07–1.39
Logistic section Income (multiples of one MW) <1 1–<2 2–<4 ≥4
1.00 0.49 0.42 0.47
NA 0.006 0.004 0.03
Reference 0.29–0.81 0.23–0.76 0.23–0.93
1.00 1.37 1.27 1.90
NA 0.66 0.75 0.4
Reference 0.34–5.51 0.29–5.48 0.43–8.53
Age (years) <1 1–<10 10–<20 20–<30 30–<40 ≥40 Cons
0.83 0.5 0.32 0.22 0.1 0.08 9.73
0.652 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
0.37–1.87 0.36–0.69 0.23–0.44 0.15–0.31 0.05–0.17 0.04–0.15 5.01–18.90
1.19 0.72 0.44 1.33 0.79 <0.001 0.44
0.62 0.04 0.01 0.02 0.39 0.98 0.33
0.60–2.39 0.53–0.98 0.23–0.84 1.04–1.70 0.46–1.35 0 0.09–2.25
/lnalpha Alpha
−1.38 0.25
<0.001
−1.50–−1.25 0.22–0.29
1.84 6.3
<0.001
1.69–1.99 5.43–7.39
Vuong test
Z = 6.74
<0.0001
Z = 3.18
0.0007
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218
number of dogs/household that may aid in developing targeted educational strategies for pet population control. Our study also found that owner income was not associated with the number of dogs or cats in households that already have pets. However, households within the income category <1 MW were significantly more likely to have zero dogs compared to households higher income categories (1–<2, 2–<4 and ≥4 MW). Therefore, income may not be a good indicator of the number of dogs or cats among pet owners despite its association with having or not a dog. A similar finding was observed in Netherlands, where animal owners tended to have higher incomes than non-owners (Endenburg et al., 1990), but this study did not report the probability of zero pets. Most of the educational initiatives of animal control in this area are concentrated on poor families that already have pets (AWB, personal communication). Our results show that, ideally, other income categories should also be targeted. In contrast to income, our study showed multiple age variables associated with pet ownership/count. A similar report in the United States also found an association between the number of dogs and owner’s age within a category of 25–44 years old (Ramón et al., 2010). The fact that only the age variables between 1 and 30 years old were less likely to have zero cats may imply increasing cat ownership in this age group. Finally, age may aid in modeling pet count in future studies, as different age groups were associated with the number of dogs and cats owned. A potential limitation of this study is that age was recorded as categories. Future studies should explore age as categories and take into consideration the number and age of members in each household. It is noteworthy that the high number of owned dogs and cats may not be necessarily correlated with the fact that owners allow their animals to roam. The roaming of owned dogs may be connected to income and education level and needs to be further explored. Free-roaming owned dogs, as well as loosely owned or community owned animals, may be then responsible for increasing the stray dog population size. A limitation of this study is response bias with respect to the response rate. Although the reason why certain households refused to be interviewed was not registered, interviewers later reported that certain people refused to be interviewed once they were told the expected time to complete the interview (sometimes more than 30 min – the team was responsible for evaluating the conditions of the household with respect to dengue prevention) (R. Lacerda, personal communication, Department of Urban Planning, Pinhais City Hall). The refusal and the absence of people at the time of the visit may imply busy lifestyles, which in turn, may somehow affect the presence and number of pets in a household. Since vertical housing has been associated with a low number of pets in one Brazilian neighborhood (Serafini et al., 2008), the number of animals in the 3 apartment building complexes may be lower than the average for the city. However, the lack of representativeness of these apartment buildings with respect to the overall household population (approximately 1%; data not shown) is unlikely to bias with the results of the present study.
217
5. Conclusion The count data analysis methods used in this study may be applied to data from other Brazilian cities, and probably in urban areas, as it takes into account people that have and do not have pets. As the number of household members of given age categories increase in certain households, one can expect an increase in the number of pets. Therefore, the results of this study can be used to plan public health initiatives and educational programs in Pinhais city based on age stratification. Interventions can target people that still do not have pets and households in higher income categories, so as to help in the decision to acquire or care for a pet. Conflict of interest None. Acknowledgements Funding for Dr. Camila Martins’s research fellowship was provided by the Tesouro Nacional – Universidade Federal do Paraná – UFPR. We kindly thank the help given by the Pinhais Center of Zoonosis Control (CCZ) personnel for the interviews and data collection. 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.prevetmed.2012.10.006. References Alves, M.C.G.P., Matos, M.R., Reichmann, M.L., Dominguez, M.H., 2005. Estimation of the dog and cat population in the state of Sao Paulo. Rev. Saude Publica 39, 1–7. Dias, R.A., 2001. Emprego do sistema de informac¸ão geográfica (SIG) no controle da raiva canina. Dissertac¸ão de Mestrado, Universidade de São Paulo, São Paulo. Dias, R.A., Garcia, R., de, C., Silva, D.F., Amaku, M., Ferreira Neto, J.S., Ferreira, F., 2004. Estimativa de populac¸ões canina e felina domiciliadas em zona urbana do Estado de São Paulo. Rev. Saude Publica 38, 565–570. Downes, M.J., Canty, M.J., More, S.J., 2009. Demography of the pet dog and cat population on the island of Ireland and Human factors influencing pet ownership. Prev. Vet. Med. 92, 140–149. Dunn, O.J., 1964. Multiple contrasts using rank sums. Technometrics 6, 241–252. Elliott, A.C., Hynan, L.S., 2011. A SAS® macro implementation of a multiple comparison post hoc test for a Kruskal–Wallis analysis. Computer methods and programs in biomedicine 102, 75–80. Endenburg, N., Hart, H., de Vries, H.W., 1990. Differences between owners and nonowners of companion animals. Anthozoos IV (2), 120–126. Garcia, R.de C.M., 2009. Estudo da dinâmica populacional canina e felina e avaliac¸ão de ac¸ões para o equilíbrio dessas populac¸ões em área da cidade de São Paulo, SP, Brasil. Tese de Doutorado, Universidade de São Paulo, São Paulo. IBGE, Instituto Brasileiro de Geografia e Estatística, 2010. Disponível em http://www.ibge.com.br/home/estatistica/populacao/censo2010/ tabelas pdf/total populacao parana.pdf (acesso em 04.02.11). Leslie, B.E., Meek, A.H., Kawash, G.F., McKeown, D.B., 1994. An epidemiological investigation of pet ownership in Otario. Can. Vet. J. 35, 218–222. Lima Júnior, A.D., 1999. Dinâmica populacional canina e a persistência da raiva na cidade de Recife (PE), nordeste do Brasil. Dissertac¸ão de Mestrado, Universidade de São Paulo, São Paulo.
218
C.M. Martins et al. / Preventive Veterinary Medicine 109 (2013) 213–218
Marx, M.B., Stallones, L., Garrity, T.F., Johnson, T.P., 1988. Demographics of pet ownership among U.S. adults 21–64 years of age. Anthrozoos II 1, 33–37. Murray, J.K., Browne, W.J., Roberts, M.A., Whitmarsh, A., Gruffydd-Jones, T.J., 2010. Numbers and ownership profiles of cats and dogs in the UK. Vet. Rec. 166, 163–168. Nunes, C.M., Martines, D.A., Fikaris, S., Queiroz, L.H., 1997. Avaliac¸ão da populac¸ão canina da zona urbana do Município de Arac¸atuba, São Paulo, SP, Brasil. Rev. Saude Publica 31, 308–309. Paranhos, N.T., 2002. Estudo das populac¸ões canina e felina em domicílio, município de São Paulo, 2001. Dissertac¸ão de Mestrado, Universidade de São Paulo, São Paulo. Pet Food Manufacturers’ Association (PFMA) 2007. Pet population. Available at: http://www.pfma.org.uk/pet-population-2011/ (accessed 03.05.12).
Ramón, M.E., Slater, M.R., Ward, M.P., 2010. Companion animal knowledge, attachment and pet cat care and their associations with household demographics for residents of a rural Texas town. Prev. Vet. Med. 94, 251–263. Serafini, C.A.V., Rosa, G.A., Guimarães, A.M.S., de Morais, H.A., Biondo, A.W., 2008. Survey of Owned Feline and Canine populations n Apartments from a Neighbourhood in Curitiba, Brazil. Zoonoses Public Health 55, 402–405. Statistical Abstract of the United States - Census Bureau (SAUS), 2011. Available at: http://www.census.gov/prod/2011pubs/ 11statab/arts.pdf (acessed 30.10.12). Zar, J.H., 2010. Biostatistical Analysis, 5th ed. Prentice Hall, p. 241ff.