Feline litter box issues associate with cat personality, breed, and age at sterilization

Salla Mikkola Department of Veterinary Biosciences, Faculty of Veterinary Medicine, University of Helsinki, Helsinki, Finland
Department of Medical and Clinical Genetics, Faculty of Medicine, University of Helsinki, Helsinki, Finland
Folkhälsan Research Center, Helsinki, Finland

Search for other papers by Salla Mikkola in
Current site
Google Scholar
PubMed
Close
 MS
,
Milla Salonen Department of Veterinary Biosciences, Faculty of Veterinary Medicine, University of Helsinki, Helsinki, Finland
Department of Medical and Clinical Genetics, Faculty of Medicine, University of Helsinki, Helsinki, Finland
Folkhälsan Research Center, Helsinki, Finland

Search for other papers by Milla Salonen in
Current site
Google Scholar
PubMed
Close
 PhD
,
Emma Hakanen Department of Veterinary Biosciences, Faculty of Veterinary Medicine, University of Helsinki, Helsinki, Finland
Department of Medical and Clinical Genetics, Faculty of Medicine, University of Helsinki, Helsinki, Finland
Folkhälsan Research Center, Helsinki, Finland

Search for other papers by Emma Hakanen in
Current site
Google Scholar
PubMed
Close
 MS
, and
Hannes Lohi Department of Veterinary Biosciences, Faculty of Veterinary Medicine, University of Helsinki, Helsinki, Finland
Department of Medical and Clinical Genetics, Faculty of Medicine, University of Helsinki, Helsinki, Finland
Folkhälsan Research Center, Helsinki, Finland

Search for other papers by Hannes Lohi in
Current site
Google Scholar
PubMed
Close
 PhD

Click on author name to view affiliation information

Abstract

OBJECTIVE

To identify potential risk factors for feline litter box issues (eg, house soiling).

ANIMALS

3,049 privately owned cats.

PROCEDURES

Data were collected using a validated, owner-completed survey with convenience sampling. The feline behavior and personality survey included 138 statements related to cat behavior and questions concerning cat background and health. Statements related to litter box issues were subjected to factor analysis. Associations between the identified factors and personality and background variables were studied using generalized linear models. Strength of these associations (ie, importance) was evaluated by calculating relative and absolute effect sizes.

RESULTS

Factor analysis yielded 2 factors: house soiling and litter box fussiness. This study suggests that fearful cats are more prone to both forms of litter box issues than nonfearful individuals. Other associations we found differed between factors. For example, low sociability toward cats, male sex, and being intact associated only with increased house soiling and older age only with litter box fussiness. The most important variables in the litter box models (ie, sociability toward cats, breed, and activity/playfulness) failed to reach the suggested cutoff for a small effect size.

CLINICAL RELEVANCE

Numerous variables are thought to influence litter box issues, but few studies have examined their relevance. Here, we studied the associations of over 30 background variables and personality traits with 2 litter box issue factors to estimate their importance at the population level. Our results bring new knowledge to this field and may contribute to finding new solutions for these complex issues in the future.

Abstract

OBJECTIVE

To identify potential risk factors for feline litter box issues (eg, house soiling).

ANIMALS

3,049 privately owned cats.

PROCEDURES

Data were collected using a validated, owner-completed survey with convenience sampling. The feline behavior and personality survey included 138 statements related to cat behavior and questions concerning cat background and health. Statements related to litter box issues were subjected to factor analysis. Associations between the identified factors and personality and background variables were studied using generalized linear models. Strength of these associations (ie, importance) was evaluated by calculating relative and absolute effect sizes.

RESULTS

Factor analysis yielded 2 factors: house soiling and litter box fussiness. This study suggests that fearful cats are more prone to both forms of litter box issues than nonfearful individuals. Other associations we found differed between factors. For example, low sociability toward cats, male sex, and being intact associated only with increased house soiling and older age only with litter box fussiness. The most important variables in the litter box models (ie, sociability toward cats, breed, and activity/playfulness) failed to reach the suggested cutoff for a small effect size.

CLINICAL RELEVANCE

Numerous variables are thought to influence litter box issues, but few studies have examined their relevance. Here, we studied the associations of over 30 background variables and personality traits with 2 litter box issue factors to estimate their importance at the population level. Our results bring new knowledge to this field and may contribute to finding new solutions for these complex issues in the future.

Introduction

House soiling is a common problem in cats and includes both urination and defecation outside an owner-designated elimination location such as litter boxes.13 House soiling can indicate that the cat has health issues or, as owners usually consider it problematic, may lead to relinquishment of the cat to an animal shelter.37

Traditionally, house soiling with urine is separated into marking behavior (sometimes called spraying) and latrine behavior (sometimes called inappropriate urination).8 Marking is a normal feline behavior used in communication between individuals,9 and it is more common in intact than sterilized cats.4 Conversely, latrine behavior serves the purpose of emptying the bladder and bowel10 and is sometimes accompanied by inappropriate defecation.8 House soiling is not always a behavioral problem, as latrine behavior especially can be caused by different diseases and health problems, such as urinary tract infections.1,5,6,11 Although the motivations behind marking and latrine behaviors may differ, they are suggested to share risk factors.8 For example, the environment and social conflicts (eg, with other cats) may motivate house soiling.1,9,12,13 Further, previous studies have reported that certain breeds, such as Persians, exhibit more house soiling.2,14 House soiling is often associated with poor litter box management, and cats may avoid the litter box if they find the used substrate unpleasant or the litter box untidy.1,15,16 The tolerance of poor litter box management varies between individuals,1 but variables affecting the tolerance level are not yet identified. In addition, previous studies have not examined possible associations with cat personality, excluding fearfulness, nor have they reported the effect sizes of the found associating variables.

Distinguishing the 2 urination behaviors from one another can be difficult, especially if the owner has not witnessed the event.8 Therefore, in this study, house soiling included both urination due to marking and latrine behavior and defecation outside the litter box. The term litter box fussiness reflects substrate preference, willingness to use an unclean litter box, and sharing the litter box with 1 or more cats. Litter box issues are used as an umbrella term for both house soiling and litter box fussiness. This study aimed to evaluate the significance of over 30 potential risk factors associated with litter box issues in an owner-completed convenience sample of 3,049 cats.

Materials and Methods

Questionnaire and data

Our study included data from 3,049 cats. Data were collected using a previously validated, owner-completed feline behavior and personality survey.17 The survey was published online and made available for all cat owners over 18 years of age. Thus, data were a convenience sample. In addition to the cats’ basic demographic information, the survey included sections on behavior and personality, background, and health. Data were collected from March 2019 to September 2020 with snowball sampling using social media and has been partly published in our previous studies.17,18 The published data sets included cats’ basic demographic information, a behavior and personality section, and some of the background/environmental variables utilized in this study. The data used in this study are published in Figshare.19

The behavior and personality section of the survey included over 130 statements, and the response options were “strongly disagree,” “somewhat disagree,” “neither agree nor disagree,” “somewhat agree,” “strongly agree,” and “I do not know.”17 A previous study17 used factor analysis to reduce the statements into 7 behavior and personality traits: activity/playfulness, fearfulness, aggression toward people, sociability toward people, sociability toward cats, excessive grooming, and litter box issues. We used the cats’ scores from those factors, excluding litter box issues, as explanatory variables. The activity/playfulness trait included items related to playing, running, and jumping. Fearfulness included a cat’s reactions toward unfamiliar people and sounds in different contexts, and sociability toward cats comprised seeking and enjoying the company of other cats, for example.17

Potential explanatory variables derived from the published data set included age, sex, breed, acquisition place, socialization to humans, socialization to animals, siblings, the main reason for obtaining the cat, food types, feeding style, large scratching trees, small scratching trees, playtime frequency, type of outdoor access, hobby, number of days the cat is left alone during the week, number of other cats in household, owner’s previous cat experience/ownership, and time since last vet visit.18 The extraction of these variables is described in more detail in our previous article18 and is summarized elsewhere (Supplementary Table S1).

Construction of new explanatory variables

For this study, we formed 9 new variables from the background and health sections of the survey: age at sterilization, number of litter boxes, living space, dogs in the family, children in the family, urinary tract disease, diseases causing excessive drinking, intestinal disease, and musculoskeletal disease. The variables age at sterilization, number of litter boxes, living space, dogs in the family, and children in the family were initially continuous variables, but we had to transform them into categorical variables due to high missingness. The age at sterilization variable included the categories “intact,” “less than 4 months,” “4 months to 1 year,” “over 1 year,” and “unknown.” The number of litter boxes variable formed the categories “1,” “2,” “3,” and “4 or more,” and living space formed the categories “under 51 m2,” “52 to 80 m2,” “over 81 m2,” and “unknown.” The dogs in the family and children in the family variables were transformed into binary variables and thus had the categories “yes” and “no.”

The health section of the survey listed various diseases and other health problems, and participants were asked to select all diseases/problems their cat had currently or previously exhibited. Thus, this information was mainly based on each owner’s knowledge of their cat’s current health status. We combined this information into 4 categories: urinary tract disease, disease causing excessive drinking, intestinal disease, and musculoskeletal disease. Cats considered to have urinary tract disease had a urinary tract infection, idiopathic cystitis, and/or bladder stones (urine crystals and uroliths). Diseases causing excessive drinking included kidney failure, hyperthyroidism, and diabetes mellitus, and intestinal disease included repetitive diarrhea, repetitive constipation, and acute or chronic inflammatory bowel disease. Finally, the musculoskeletal disease variable included hip dysplasia or other deformation of the bones/joints and osteoarthritis or other joint pain. The health section also included free-text fields for “other kidney- or urinary tract–related disease/problem,” “other digestive system–related disease/problem,” “other muscular- or skeletal system–related disease/problem,” and “if you did not find your cat’s disease or health problem, describe it below.” We included suitable diseases/problems from these in the described health variables above. All health variables included the categories “yes,” “no,” and “unknown.” Cats that were missing health information were grouped into the “unknown” group.

Factor analysis of items related to litter box issues

We first selected only the following behavior statements that related to house soiling/litter box: cat urinates (crouching position) in inappropriate places, sprays (standing position with tail raised vertically and vibrating) indoors (eg, on furniture or walls), defecates in inappropriate places indoors, refuses to use the litter box if it is dirty, does not want to use the same litter box as other cats in the household, prefers specific types of cat litter, and shows substrate preference when urinating or defecating in inappropriate places (eg, laundry, bedding, carpets, or flower pots). We discarded the last statement due to high missingness (approx 77%) and after that, excluded individuals that had over 20% missingness in total in the remaining selected statements. We decided to conduct an explanatory factor analysis (EFA) to reduce these 6 remaining statements (items) into a smaller set of continuous factors, which in EFA are assumed to underlie these measured items.20

Before factor analysis, we tested the appropriateness of our data set for EFA using the Kaiser-Meyer-Olkin test for sampling adequacy (Kaiser-Meyer-Olkin = 0.69) and Bartlett test of sphericity (P < .0001), which indicate whether it is reasonable to conduct factor analysis.20 All statements were on ordinal scale (ordered but the distance between categories is not known), and therefore we used a polychoric correlation matrix in the factor analysis.21 We utilized mean imputation for missing data, meaning that missing values were replaced with the study population’s mean score of the item. We did not use rotation in factor analysis, as it complicates further use of the factor scores. We chose the optimal number of factors on the basis of the Scree test (ie, visually inspecting the scree plot of factor eigenvalues), parallel analysis, and Velicer minimum average partial test.20 We also extracted structures with 1 to 3 factors to evaluate the meaningfulness of different factor structures. In addition, we also compared the root mean square error of approximation and the Tucker-Lewis index, which both measure the model fit,22 between possible structures.

The structure with 2 factors was the most suitable and biologically the most meaningful (for item loadings, see Supplementary Table S2). The first factor, which we named “house soiling,” included the following items: cat urinates in inappropriate places, sprays indoors, and defecates in inappropriate places indoors. The second factor, named “litter box fussiness,” included the following items: cat refuses to use the litter box if it is dirty, does not want to use the same litter box as the other cats in the household, and prefers specific types of cat litter. We evaluated the factors’ internal consistency reliability with Cronbach α and Guttman λ 6 reliability coefficients.23 The α and λ for house soiling were 0.57 and 0.51, respectively, and for litter box fussiness were 0.53 and 0.47, respectively. We extracted factor scores for all cats using mean imputation and the correlation-preserving factor score estimation method (method = “TenBerge” in package psych23) and used them (ie, house soiling and litter box fussiness) as the dependent variables in our study (see next sections). The psych23 package was used in all analyses above.

Preselection of variables with relative influences

To reduce the number of explanatory variables (n = 34), we first calculated their relative influences for all 6 statements related to litter box issues and for the 2 newly formed factors (ie, house soiling and litter box fussiness) using generalized boosted regression models fitted with the gbm package.24 The number of trees was set to 10,000, interaction depth to 1, shrinkage to 0.01, and bag fraction to 0.5 in all 8 models.24 The variables with a relative influence of less than 1 in any of the models were discarded before the next model selection phase (Supplementary Table S3).

Model selection and fit

We used generalized linear models to identify the explanatory variables that had the strongest association with house soiling and litter box fussiness. As both factors were nonnormally distributed, we chose more suitable distributions using the rcompanion25 and boot packages.26 The γ distribution with the log-link function was the best-fitting distribution.

To select the best models for house soiling and litter box fussiness, we used the airGLMs package27 to perform a forward stepwise Akaike information criterion (AIC) model selection. AIC is a mathematical method for evaluating the model fit, enabling the comparison of different candidate models.28 Age and sex were included in the starting model of both traits. The other 10 explanatory variables we included in the model selection were personality/behavior traits, breed, age at sterilization, living space, number of litter boxes, playtime frequency, type of outdoor access, owner’s previous cat experience/ownership, other cats in household, children in the family, and urinary tract disease. In addition, we included the following possible interactions between variables: sex and age at sterilization, age and urinary tract disease, living space and other cats in household, and number of litter boxes and other cats in household.

Model selection with the AIC selection approach tends to overfit models28 by also selecting less relevant explanatory variables into the models to maximize the model fit. Therefore, we divided the data into 10 smaller data sets (n = 1,524 to 1,525) with repeated train-test validation by randomly splitting the data 5 times into training (50% of data) and testing (50% of data) data sets with the caret package.29 We performed the forward stepwise model selection individually for all data sets and ended up with 10 models for house soiling and 10 models for litter box fussiness. We then calculated the number of times each variable was included in the models. We formed the final models for both traits by only including variables that were included in at least 90% of the models (Supplementary Table S4).

After model selection, we evaluated model fit. First, we visually inspected the residuals and evaluated the presence of heteroscedasticity with the rcompanion25 and boot packages.26 Model fit was moderate for house soiling and good for litter box fussiness. Second, we inspected the linearity assumption of the continuous variables with generalized additive models using the gam package.30 In house soiling, age and sociability toward cats did not meet the linearity assumption. We therefore added age as both a linear and quadratic variable (age and age2) into the model, but we did not add sociability toward cats as a quadratic variable because it did not improve its linearity. In litter box fussiness, the variables age, fearfulness, and activity/playfulness did not meet the linearity assumption, and we added each into the model as both linear and quadratic variables. Third, we inspected possible outliers with the broom,31 dplyr,32 and ggplot2 packages.33 We found several outliers and compared the results of the data sets that included or excluded them. As removing these outliers did not affect the results and they were real observations, we kept them in the data. Finally, we evaluated multicollinearity with the car package34 and inspected the general fit of the models with the Durbin-Watson test, which detect the possible autocorrelation in the residuals, from the lmtest package.35

Effect size

We measured the strength of the relationship of explanatory variables with outcome variables (house soiling and litter box fussiness) using 2 different methods. We inspected the relative effect size of the variables with variance-based variable importance using a feature importance ranking measure approach from the vip package.36 The variable importance was scaled, and the most important variables therefore reached a score of 100. In addition, we examined absolute effect sizes by calculating the partial Cohen f2 with the effectsize package.37

Extraction of results

We first conducted an ANOVA to obtain the overall effect and statistical significance of the variables with the car package.34 Then we used the emmeans package38 to extract the estimated marginal means and calculate the adjusted means, confidence limits, and pairwise comparisons for the levels of categorical variables. For continuous variables, we used the effects package.34,39 Finally, we corrected the obtained P values for the false discovery rate (FDR). We set the significance cutoff at P < .05. All analyses were conducted using R version 4.1.2.40

Ethics statement

The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the University of Helsinki Viikki Campus Research Ethics Committee (February 11, 2019). Informed consent was obtained from all respondents.

Results

Study subjects

We studied the environmental and demographic variables associated with litter box issues in a cohort of 3,049 cats. Cat age varied between 0.3 and 22.7 years, and mean age was 6 years. Most cats (90%) were neutered, sterilized, or being administered medical contraception, and 49% were female. The number of cats within the 26 breed groups ranged from 34 Turkish Vans to 740 Landrace cat shorthairs.

Variables associated with house soiling

The final generalized linear model for house soiling included the explanatory variables age, sex, age at sterilization, breed, fearfulness, sociability toward cats, number of litter boxes, playtime frequency, and urinary tract disease (n = 3,036). After FDR correction, only sex, age at sterilization, breed, fearfulness, sociability toward cats, and urinary tract disease were statistically significant (Table 1).

Table 1

Associations between the explanatory variables in the house soiling model (n = 3,036) and their scaled variance-based variable importance (relative effect size) and Cohen f2 (absolute effect size). Age did not meet the linearity assumption in the model, so it was included also as quadratic variable (age2). DF = degrees of freedom. F = F value. All P values are controlled for false discovery rate.

Explanatory variable DF F P Variable importance Cohen f2
Sociability toward cats 1 24.269 < .001* 100 0.008
Breed 25 2.292 .009* 82 0.019
Fearfulness 1 13.526 .009* 58 0.005
Age at sterilization 4 4.334 .039* 43 0.006
Age 1 3.046 .306 41 0.001
Urinary tract disease 2 14.041 < .001* 38 0.009
Playtime frequency 4 3.138 .122 28 0.004
No. of litter boxes 3 3.403 .127 21 0.003
Sex 1 9.325 .044* 15 0.003
Age2 1 2.504 .363 NA 0.001

*Significant (P < .05) association between variables.

A house soiling score correlated positively with fearfulness and negatively with sociability toward cats (Table 1; Supplementary Figure S1). Male cats had a higher mean score than females (P = .044). On average, intact cats scored higher than cats sterilized at under 4 months of age (P = .006) and cats sterilized between 4 months to 1 year of age (P = .039; Figure 1). Cats with 1 or more owner-reported urinary tract diseases had a higher mean score than cats with no urinary tract disease (P < .001). The highest mean score was observed in Bengal cats and the lowest in Siberian and Neva Masquerade cats (P < .001). The rest of the pairwise comparisons can be found elsewhere (Supplementary Tables S5 and S6).

Figure 1
Figure 1

Associations of breed (A), age at sterilization (B), and urinary tract disease (C) with inappropriate elimination in the generalized linear model (n = 3,036). Factor scores are normalized, the score mean is 0, and SD is 1. Error bars indicate 95% confidence limits.

Citation: Journal of the American Veterinary Medical Association 261, 5; 10.2460/javma.22.10.0441

The relative importance of variables varied between 15 and 100 (Table 1). Sociability toward cats was the most important variable, followed by breed and fearfulness. Cohen f2 estimates varied between < 0.001 and 0.019. Therefore, on the basis of the suggested Cohen cutoff values (small = 0.02, medium = 0.15, and large = 0.35), all variables had less than small effect sizes.41 Breed had the largest importance and almost a small effect size (f2 = 0.019).

Variables associated with litter box fussiness

The final generalized linear model for litter box fussiness included the explanatory variables age, sex, fearfulness, activity/playfulness, and children in the family (n = 3,049). After FDR correction, all variables, except sex, remained statistically significant (Table 2).

Table 2

Associations between the explanatory variables in the litter box fussiness model (n = 3,049) and their scaled variance-based variable importance (relative effect size) and Cohen f2 (absolute effect size). Age, activity/playfulness and fearfulness did not meet the linearity assumption in the model, so they were included also as quadratic variables. DF = degrees of freedom. F = F value. All P values are controlled for false discovery rate.

Explanatory variable DF F P Variable importance Cohen f2
Activity/playfulness 1 45.198 < .001* 100 0.015
Age 1 23.468 < .001* 83 0.008
Fearfulness 1 26.429 < .001* 67 0.009
Children in the family 1 21.175 < .001* 24 0.007
Sex 1 0.862 .353 4 < 0.001
Age2 1 5.325 .028* NA 0.002
Activity/playfulness2 1 11.219 .001* NA 0.004
Fearfulness2 1 0.955 .353 NA < 0.001

*Significant (P < .05) association between variables.

Litter box fussiness score correlated positively with age, fearfulness, and activity/playfulness (Table 2; Figure 2). Cats living in families with children had higher mean scores than cats living in childless families (P < .001; Supplementary Table S5).

Figure 2
Figure 2

Associations of children in the family (A), age (B), activity/playfulness (C), and fearfulness (D) with litter box fussiness in the generalized linear model (n = 3,049). Factor scores are normalized, the score mean is 0, and SD is 1. Error bars and gray areas indicate 95% confidence limits.

Citation: Journal of the American Veterinary Medical Association 261, 5; 10.2460/javma.22.10.0441

The relative importance of the variables varied between 4 and 100 (Table 2). The most important variable was activity/playfulness, followed by age and fearfulness. Cohen f2 estimates varied between < 0.001 and 0.015. The variable with the highest Cohen f2 estimate was activity/playfulness. Thus, none of the variables reached the suggested cutoff for a small effect size.

Discussion

We identified associations between several environmental, personality, and demographic variables with feline litter box issues utilizing validated survey data of 3,049 privately owned cats. Sociability toward cats, breed, and fearfulness were the most important risk factors for house soiling, while activity/playfulness, age, and fearfulness were the most important risk factors for litter box fussiness. Other factors that had statistically significant associations with house soiling were age at sterilization, urinary tract disease, and sex, while living in the family with children associated significantly with litter box fussiness. The correlation between house soiling and litter box fussiness was 0.51; thus, a high score in one trait usually corresponded to a high score in the other but not necessarily.

Personality traits were associated with both factors. Fearful cats had higher scores for both house soiling and litter box fussiness. The association between a cat’s personality and house soiling has rarely been studied and the association with litter box fussiness not at all. Barcelos et al8 noticed that cats with a relaxed personality had a lower risk of marking behavior than other cats, and Porters et al42 found that kittens that were shy or frightened of strangers were more likely to show house soiling as adults compared with cats who behaved friendly as kittens. Cats that scored high in sociability toward cats had lower scores for house soiling. This is interesting, as the number of other cats in a household did not associate significantly with either form of litter box issues in this study. In litter box fussiness, activity/playfulness had a nonlinear effect. This is a novel finding that requires more research.

Cat age only associated with litter box fussiness. On average, older cats were stricter about the hygiene of their litter boxes than younger cats. One hypothesis is that older, more experienced cats may have learned to associate certain substrate types with negative experiences, such as pain during urination. Barcelos et al8 reported that the marking cats were older than the cats in the latrine or control groups. Another study13 reported that house soiling (urinating or defecating in inappropriate areas indoors) increased with age but did not examine the possible effect of age on marking behavior. Although the studied behaviors differ, we expected to also find age as a significant factor in our house soiling variable.

The mean house soiling score was slightly higher in male cats than in females. Some previous studies agree with this finding, as male cats reportedly exhibit more inappropriate urination3,43,44 and defecation45 than females. Fifty percent of veterinarians in a study by Takeuchi and Mori46 rated female cats as exhibiting house soiling more commonly than males, while the other half stated that no sex difference exists. However, house soiling did not include marking behavior in their study.46

Cats sterilized at the age of 1 year or less had a lower mean score in house soiling compared with intact cats. On the other hand, comparing intact cats with cats sterilized later than this showed no statistically significant difference. Neutering has previously been used successfully to reduce or prevent urine marking in free-roaming cats (both kittens and adults).47 In addition, another study48 reported that the castration of males before 5.5 months of age associated with decreased spraying compared with cats castrated at an older age. In that study,48 they also suggested that veterinarians should consider recommending routine sterilization for client-owned cats before the age of 6 to 8 months. Another study42 did not find a difference between cats neutered at age 8 to 12 weeks or 6 to 8 months. Thus, we might suggest that sterilization before adulthood and before marking behavior begins may reduce the probability of house soiling but cannot state the best age.

Breed associated with house soiling but not with litter box fussiness. The Bengal scored the highest, followed by breeds such as the Landrace cat, Oriental, Norwegian Forest cat, and Persian and Exotic. The Siberian and Neva Masquerade scored the lowest. Persians have also exhibited more house soiling in previous studies.2,14,44 In contrast, no differences between breeds were observed in other studies.45,46 In our study, most breeds did not differ in the pairwise comparisons and individual differences within a breed were large.

A history of urinary tract disease associated with house soiling. Cats that currently have or have previously had a urinary tract disease scored higher in house soiling than cats without owner-reported urinary tract diseases. Cats exhibiting house soiling were more likely to have had a urinary tract disease also in the study by Horwitz,49 but none of the disease-related factors were significant in the study by Barcelos et al.8 The other studied disease groups, including diseases causing excessive drinking, intestinal disorders, and musculoskeletal diseases, had low relative influences and were not included in the model selection phase.

Of the studied environmental variables, only living in a family with children had a statistically significant association with litter box issues. Cats living in a family with children had a higher mean score in litter box fussiness than cats living in homes without children. Previously, studies have noted that children can be stressors for cats, as they may pursue the cat and the home may lack places where the cat can avoid meeting children, for example.1 Families with children may also have less time for litter box cleaning. Litter boxes may therefore be cleaned less frequently in families with children than litter boxes in families with adults only.

Interestingly, the number of litter boxes and living with other cats did not associate with house soiling or litter box fussiness in our study. Similarly, Barcelos et al8 did not find significant factors related to litter box attributes. Previous literature, however, has advised that the number of litter boxes in multicat households should equal the number of cats plus 1 additional box so that a litter box is available at any given time.6,11 This advice, however, lacks experimental testing.

Other cats and dogs can also act as stressors,1 and some studies have reported an increased risk of house soiling in multicat households.8,43 However, the owner cannot always identify the individual exhibiting soiling behavior in multicat households,6 which could explain the lack of this association,13 as these individuals may be omitted from the data due to missing answers. In general, literature has reported contradictory results concerning multicat households and problematic behaviors.50

In house soiling, sociability toward cats had the highest relative importance and breed had the highest absolute importance. In litter box fussiness, activity/playfulness reached both the highest relative and absolute importance. However, Cohen f2 values indicated less than small effect sizes even for these variables. In general, effect size estimates in individual differences are low, as personality traits are very complex.51 Therefore, none of the studied variables explaining a large portion of the variance in litter box issues is understandable. In fact, a recent meta-analytic study recommended lowering suggested cutoff values for correlations, which would also reflect to the Cohen f2 values.41,51

This study had some limitations. Our survey design was cross-sectional, so identifying variable causalities was impossible. In addition, not all owners had the same possibility or willingness to participate in the study because it was based on a convenience sample collected using the internet platform. The health section was filled out by the current owner of each cat, and some cats may therefore have had unidentified/reported diseases or other health problems. Statements related to the litter box fussiness factor, such as willingness to use an unclean litter box, involve participant interpretations and are therefore somewhat subjective. We lacked information concerning punishment and free-ranging cats in the area, for example, which may also be significant reasons for house soiling.6,42,43 In addition, we lacked further information on litter box specifications (eg, size of the litter box), which may be as important as the number of litter boxes.11 Finally, as the distributions of some of the variables were far from normal, we had to make compromises when choosing suitable distributions (eg, house soiling and cat sociability).

In conclusion, our study suggested that fearful cats are more prone to both house soiling and litter box fussiness than their nonfearful conspecifics. The rest of the variables were trait specific. Low sociability toward other cats, male sex, being intact, and breed differences were associated only with house soiling. Existing breed differences may suggest that house soiling also has a heritable aspect. Older age and living in a family with children were associated only with increased litter box fussiness. The effect sizes of the variables were small in both traits, reflecting the complex nature of litter box issues.

Supplementary Materials

Supplementary materials are posted online at the journal website: avmajournals.avma.org

Acknowledgments

This research was funded by a Finnish cat association Suomen Kissaliitto ry, the Agria and Svenska Kennel Club Research Fund (N2019-0005), and the Academy of Finland (308887). Funding sources did not have any involvement in the study design, data analysis and interpretation, or writing and publication of the manuscript.

Hannes Lohi was a co-founder and Emma Hakanen an employee of Petsofi Ltd, which provided the survey platform for data acquisition. The authors declare that there were no other competing interests.

We would like to thank all the cat owners who took the time to fill out the survey and the cat and breed associations who advertised the survey. In addition, we want to acknowledge Julia Niskanen for developing the airGLMs, which boosted the model selection process considerably.

References

  • 1.

    Herron ME. Advances in understanding and treatment of feline inappropriate elimination. Top Companion Anim Med. 2010;25(4):195202. doi:10.1053/j.tcam.2010.09.005

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2.

    Amat M, de la Torre JLR, Fatjó J, Mariotti VM, Van Wijk S, Manteca X. Potential risk factors associated with feline behaviour problems. Appl Anim Behav Sci. 2009;121(2):134139. doi:10.1016/j.applanim.2009.09.012

    • Search Google Scholar
    • Export Citation
  • 3.

    Strickler BL, Shull EA. An owner survey of toys, activities, and behavior problems in indoor cats. J Vet Behav. 2013;9(5):207214. doi:10.1016/j.jveb.2014.06.005

    • Search Google Scholar
    • Export Citation
  • 4.

    Patronek GJ, Glickman LT, Beck AM, McCabe GP, Ecker C. Risk factors for relinquishment of cats to an animal shelter. J Am Vet Med Assoc. 1996;209(3):582588. doi:10.1016/j.foar.2012.09.001

    • Search Google Scholar
    • Export Citation
  • 5.

    Neilson J. Thinking outside the box: feline elimination. J Feline Med Surg. 2004;6(1):511. doi:10.1016/j.jfms.2003.09.008

  • 6.

    Neilson JC. Feline house soiling: elimination and marking behaviors. Clin Tech Small Anim Pract. 2004;19(4):216224. doi:10.1053/j.ctsap.2004.10.003

  • 7.

    Casey RA, Vandenbussche S, Bradshaw JWS, Roberts MA. Reasons for relinquishment and return of domestic cats (Felis Silvestris Catus) to rescue shelters in the UK. Anthrozoos. 2009;22(4):347358. doi:10.2752/089279309X12538695316185

    • Search Google Scholar
    • Export Citation
  • 8.

    Barcelos AM, McPeake K, Affenzeller N, Mills DS. Common risk factors for urinary house soiling (periuria) in cats and its differentiation: the sensitivity and specificity of common diagnostic signs. Front Vet Sci. 2018;5(108):108. doi:10.3389/fvets.2018.00108

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9.

    de Souza Machado D, Oliveira PMB, Machado JC, Ceballos MC, Sant’Anna AC. Identification of separation-related problems in domestic cats: a questionnaire survey. PLoS One. 2020;15(4):e0230999. doi:10.1371/journal.pone.0230999

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10.

    Hart BL, Hart LA. Feline behavioural problems and solutions. In: Turner DC, Bateson P, eds. The Domestic Cat: The Biology of Its Behaviour. 3rd ed. Cambridge University Press; 2013:201212. doi:10.1017/CBO9781139177177.019

    • Search Google Scholar
    • Export Citation
  • 11.

    Carney HC, Sadek TP, Curtis TM, et al; American Association of Feline Practitioners; International Society of Feline Medicine. AAFP and ISFM Guidelines for diagnosing and solving house-soiling behavior in cats. J Feline Med Surg. 2014;16(7):579598. doi:10.1177/1098612X14539092

    • Search Google Scholar
    • Export Citation
  • 12.

    Amat M, Camps T, Manteca X. Stress in owned cats: behavioural changes and welfare implications. J Feline Med Surg. 2016;18(8):577586. doi:10.1177/1098612X15590867

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13.

    Duffy DL, de Moura RTD, Serpell JA. Development and evaluation of the Fe-BARQ: a new survey instrument for measuring behavior in domestic cats (Felis s. catus). Behav Processes. 2017;141(pt 3):329341. doi:10.1016/j.beproc.2017.02.010

    • Search Google Scholar
    • Export Citation
  • 14.

    Wassink-van der Schot AA, Day C, Morton JM, Rand J, Phillips CJC. Risk factors for behavior problems in cats presented to an Australian companion animal behavior clinic. J Vet Behav. 2016;14:3440. doi:10.1016/j.jveb.2016.06.010

    • Search Google Scholar
    • Export Citation
  • 15.

    Borchelt PL. Cat elimination behavior problems. Vet Clin North Am Small Anim Pract. 1991;21(2):257264. doi:10.1016/s0195-5616(91)50031-0

  • 16.

    Ellis JJ, McGowan RTS, Martin F. Does previous use affect litter box appeal in multi-cat households? Behav Processes. 2017;141(pt 3):284290. doi:10.1016/j.beproc.2017.02.008

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17.

    Mikkola S, Salonen M, Hakanen E, Sulkama S, Lohi H. Reliability and validity of seven feline behavior and personality traits. Animals (Basel). 2021;11(7):1991. doi:10.3390/ani11071991

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18.

    Mikkola S, Salonen M, Hakanen E, Lohi H. Fearfulness associates with problematic behaviors and poor socialization in cats. iScience. 2022;25(10):105265. doi:10.1016/j.isci.2022.105265

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19.

    Mikkola S, Salonen M, Hakanen E, Sulkama S, Lohi H. Feline litter box issues data. Figshare. February 3, 2023. Accessed February 3, 2023. https://figshare.com/articles/dataset/Feline_litterbox_issues_data/22003823

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 20.

    Williams B, Onsman A, Brown T. Exploratory factor analysis: a five-step guide for novices. Australas J Paramed. 2010;8(3):113. doi:10.33151/ajp.8.3.93

    • Search Google Scholar
    • Export Citation
  • 21.

    Kaiser HF. A second generation little jiffy. Psychometrika. 1970;35(4):401415.

  • 22.

    Xia Y, Yang Y. RMSEA, CFI, and TLI in structural equation modeling with ordered categorical data: the story they tell depends on the estimation methods. Behav Res Methods. 2019;51(1):409428. doi:10.3758/s13428-018-1055-2

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 23.

    Revelle W. psych: Procedures for Psychological, Psychometric, and Personality Research. The R Foundation; 2021. Accessed September 28, 2022. https://cran.r-project.org/package=psych

    • Search Google Scholar
    • Export Citation
  • 24.

    Greenwell B, Boehmke B, Cunningham J, Developers G. gbm: Generalized Boosted Regression Models. The R Foundation; 2020. Accessed September 28, 2022. https://cran.r-project.org/package=gbm

    • Search Google Scholar
    • Export Citation
  • 25.

    Mangiafico S. rcompanion: Functions to Support Extension Education Program Evaluation. The R Foundation; 2019. Accessed September 28, 2022. https://cran.r-project.org/package=rcompanion

    • Search Google Scholar
    • Export Citation
  • 26.

    Canty A, Ripley B. boot: Bootstrap R (S-Plus) Functions. The R Foundation; 2021. Accessed September 28, 2022. https://cran.r-project.org/web/packages/boot/index.html

    • Search Google Scholar
    • Export Citation
  • 27.

    Niskanen J, Salonen MK, Puurunen J. airGLMs. GitHub Inc; 2021. Accessed September 28, 2022. https://github.com/JNisk/airGLMs

  • 28.

    Vrieze SI. Model selection and psychological theory: a discussion of the differences between the Akaike information criterion (AIC) and the Bayesian information criterion (BIC). Psychol Methods. 2012;17(2):228243. doi:10.1037/a0027127

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 29.

    Kuhn M. caret: Classification and Regression Training. The R Foundation; 2021. Accessed September 28, 2022. https://cran.r-project.org/package=caret

    • Search Google Scholar
    • Export Citation
  • 30.

    Hastie T. gam: Generalized Additive Models. The R Foundation; 2020. Accessed September 28, 2022. https://cran.r-project.org/package=gam

  • 31.

    Robinson D, Hayes A, Couch S. broom: Convert Statistical Analysis Objects into Tidy Tibbles. The R Foundation; 2021. Accessed September 28, 2022. https://cran.r project.org/package=broom

    • Search Google Scholar
    • Export Citation
  • 32.

    Wickham H, François R, Lionel H, Müller K. dplyr: A Grammar of Data Manipulation. The R Foundation; 2021. Accessed September 28, 2022. https://cran.r-project.org/package=dplyr

    • Search Google Scholar
    • Export Citation
  • 33.

    Wickham H. ggplot2: Elegant Graphics for Data Analysis. The R Foundation; 2016. Accessed September 28, 2022. https://ggplot2.tidyverse.org

    • Search Google Scholar
    • Export Citation
  • 34.

    Fox J, Weisberg S. An R Companion to Applied Regression. 3rd ed. Sage Publications Inc; 2019.

  • 35.

    Zeileis A, Hothorn T. Diagnostic checking in regression relationships. R News. 2002;2(3):710. https://cran.r-project.org/doc/Rnews/ Accessed Sep 28, 2022.

    • Search Google Scholar
    • Export Citation
  • 36.

    Greenwell BM, Boehmke BC. Variable importance plots—an introduction to the vip package. R J. 2020;12(1):343366. doi:10.32614/RJ-2020-013

    • Search Google Scholar
    • Export Citation
  • 37.

    Ben-Shachar M, Lüdecke D, Makowski D. effectsize: estimation of effect size indices and standardized parameters. J Open Source Softw. 2020;5(56):2815. doi:10.21105/joss.02815

    • Search Google Scholar
    • Export Citation
  • 38.

    Lenth R. emmeans: Estimated Marginal Means, aka Least-Squares Means. The R Foundation; 2021. Accessed September 28, 2022. https://cran.r-project.org/package=emmeans

    • Search Google Scholar
    • Export Citation
  • 39.

    Fox J. Effect displays in R for generalised linear models. J Stat Softw. 2003;8(15):127. doi:10.18637/jss.v008.i15

  • 40.

    A language and environment for statistical computing. Version 4.1.2. The R Foundation. Accessed September 28, 2022. https://www.r-project.org/

    • Search Google Scholar
    • Export Citation
  • 41.

    Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Lawrence Erlbaum Associates; 1988.

  • 42.

    Porters N, de Rooster H, Verschueren K, Polis I, Moons CPH. Development of behavior in adopted shelter kittens after gonadectomy performed at an early age or at a traditional age. J Vet Behav. 2014;9(5):196206. doi:10.1016/j.jveb.2014.05.003

    • Search Google Scholar
    • Export Citation
  • 43.

    Pryor PA, Hart BL, Bain MJ, Cliff KD. Causes of urine marking in cats and effects of environmental management on frequency of marking. J Am Vet Med Assoc. 2001;219(12):17091713. doi:10.2460/javma.2001.219.1709

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 44.

    Hart BL, Hart LA. Your Ideal Cat: Insights into Breed and Gender Differences in Cat Behaviour. Purdue University Press; 2013. doi:10.2307/j.ctt6wq4zv

    • Search Google Scholar
    • Export Citation
  • 45.

    Naderi MM, Rafiei SM, Sattari B, Ale-Davoudl SJ, Seif AA, Bokaei S. The first study on classification of Iranian domestic cats’ behavior problems and their associated risk factors. Glob Vet. 2011;6(3):339345.

    • Search Google Scholar
    • Export Citation
  • 46.

    Takeuchi Y, Mori Y. Behavioral profiles of feline breeds in Japan. J Vet Med Sci. 2009;71(8):10531057. doi:10.1292/jvms.71.1053

  • 47.

    Cafazzo S, Bonanni R, Natoli E. Neutering effects on social behaviour of urban unowned free-roaming domestic cats. Animals (Basel). 2019;9(12):1105. doi:10.3390/ani9121105

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 48.

    Spain CV, Scarlett JM, Houpt KA. Long-term risks and benefits of early-age gonadectomy in cats. J Am Vet Med Assoc. 2004;224(3):372379. doi:10.2460/javma.2004.224.372

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 49.

    Horwitz DF. Behavioral and environmental factors associated with elimination behavior problems in cats: a retrospective study. Appl Anim Behav Sci. 1997;52(1-2):129137. doi:10.1016/S0168-1591(96)01073-8

    • Search Google Scholar
    • Export Citation
  • 50.

    Finka LR, Foreman-Worsley R. Are multi-cat homes more stressful? A critical review of the evidence associated with cat group size and wellbeing. J Feline Med Surg. 2022;24(2):6576. doi:10.1177/1098612X211013741

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 51.

    Gignac GE, Szodorai ET. Effect size guidelines for individual differences researchers. Pers Individ Dif. 2016;102:7478. doi:10.1016/j.paid.2016.06.069

    • Search Google Scholar
    • Export Citation
All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 5064 2059 158
PDF Downloads 4329 1516 90
Advertisement