Quality of the Human–Animal Bond and Mental Wellbeing During a COVID-19 Lockdown

ABSTRACT

Companion animal ownership has been associated with a wide variety of physical and psychological health benefits. The extent to which a person gains any welfare advantages from the animal in their care, however, may be related to a wide variety of factors, one of which is the quality of the human–animal relationship. Thus far, little attention has been devoted to the role of attachment to one's companion animal on psychological wellbeing during a global pandemic, a time when mental health has been shown to be extremely poor. Therefore this study aimed to explore the relationship between the quality of the companion animal–human bond and mental wellbeing during a period of COVID-19-induced national lockdown in the United Kingdom. A purpose-designed online survey that aimed to measure sociodemographic background, companion animal ownership status, attachment level, and various components of mental wellbeing (depression, loneliness, positive experience, stress) was developed and completed by 249 UK-based adults (146 companion animal owners, 103 non-owners). Analysis revealed no significant relationship between companion animal ownership and any of the mental health outcome measures. Attachment to one's companion animal, however, was found to be a strong predictor of mental wellbeing, with higher bonds of attachment associated with higher levels of depression, loneliness, and lower levels of positive experience. Attachment to one's companion animal was not significantly associated with participants’ stress levels. Overall, findings from this study point to emotional vulnerability in people who are highly attached to their companion animal, although limitations must be borne in mind. This is an area worthy of further exploration, particularly considering the pandemic-induced rise in the number of people who have acquired a companion animal and the increment in mental health problems that has been predicted to emerge from COVID-19.

KEYWORDS:

• Attachment
• companion animals
• COVID-19
• human–animal interaction
• mental health

Companion animals are commonplace in today's society. In the UK alone, over 24 million dogs and 24 million cats are kept by the British public (Pet Food Manufacturer’s Association, 2021), with most considered by their caregivers to be an integral part of the family unit. Until recently, it was assumed that the human–animal relationship was a largely unidirectional one, with people caring for their dogs and cats in much the same way as they do their children, but without the same apparent benefits, from a Darwinian perspective (Archer, 1997). Recent research, however, points to a complex relationship, with many people reportedly gaining significant benefits from living with a companion animal.

One of the benefits that has been associated with living with a companion animal is improved health and wellbeing (for reviews see Beetz et al., 2012; Friedmann et al., 2000; Herzog, 2011; Wells, 2007, 2009, 2019). Whilst not entirely conclusive (e.g., Fritz et al., 1995; Miller & Lago, 1990; Siegel et al., 1999; Tower & Nokota, 2006), various studies point to the value of companion animals in improving people's physical and mental welfare. For example, studies have reported a relationship between companion animal “ownership” (a term used hereafter to refer to a person who lives with and/or cares for an animal) and fewer visits to the GP (Headey, 1998), a reduced risk of developing, and greater ability to recover from, heart problems (see Levine et al., 2013), and, if exposed to animals at an early age, a lower occurrence of allergic illness, including rhinitis, eczema, and/or asthma (e.g., Fall et al., 2015; Hesselmar et al., 1999). From a more psychological perspective, companion animal ownership has been associated with decreased loneliness, depression, and improved self-esteem and feelings of self-worth (e.g., Guest et al., 2006; Ko et al., 2016).

Numerous studies now point to a wide array of mental health problems (e.g., depression, anxiety, loneliness) arising from the stressors of living through the COVID-19 global pandemic (e.g., Bueno-Notivol et al., 2021; Salari et al., 2020). This has prompted investigation into the health benefits associated with companion animal ownership at this time. The findings from these studies, although still limited, are mixed. Phillipou and colleagues (2021), for instance, found a poorer quality of life in a group of Australians who cared for a companion animal than those who did not in the middle of their first pandemic-induced lockdown, leading the authors to conclude that companion animal ownership may add to increased burden during stressful times. By contrast, Bohn et al. (2021) found that older adults who lived with a dog (although not a cat or bird) had lower levels of depression during a period of forced lockdown in Brazil, whilst Oliva and Johnston (2021) reported lower levels of loneliness in dog-keeping Australian residents who lived alone during the first pandemic-induced lockdown.

What may be more important than the mere ownership of a companion animal is the quality of the human–animal bond. Attachment theory was first put forward to explain the child–caregiver relationship (Bowlby, 1969), but has also been used successfully to explain human–animal relationships (e.g., Jenkins et al., 2014; Peacock et al., 2012), with studies suggesting that companion animals can serve as important attachment figures (Zilcha-Mano et al., 2012). One might, perhaps intuitively, expect a stronger attachment to one's companion animal to be associated with enhanced wellbeing; indeed, this is supported by some studies. Garrity and colleagues (1989), for example, found lower levels of depression in older adults who reported higher attachment to their companion animals than in individuals who were more weakly bonded. More recently, Teo and Thomas (2019) found that people who were “securely” attached to the animals in their care had lower levels of psychological distress and psychopathology and better quality of life than individuals less securely attached. Interestingly, a small number of studies have reported reduced mental health in people who are more highly attached to their companion animals. For example, Miltiades and Shearer (2011) found that higher levels of attachment to one's companion animal were associated with higher levels of depression in a group of older American adults. More recently, Lass-Hennemann and associates (2020) reported an association between stronger attachment to one's dog and higher levels of psychopathological symptoms.

To date, only a handful of studies have focused specifically on the relationship between the quality of the human–animal bond and people's mental wellbeing during COVID-19, with mixed findings (McDonald et al., 2021; Mueller et al., 2021; Namekata & Yamamoto, 2021; Ratschen et al., 2020). For example, Ratschen and colleagues (2020) found no relationship between the strength of the human–animal bond and changes in the mental health of UK residents from pre-lockdown to lockdown. McDonald and associates (2021), however, found that high levels of attachment to one's companion animal functioned as a protective factor for some American companion animal owners, notably those reporting moderate to high levels of mental health symptoms in the pre-pandemic period. High levels of attachment to one's dog or cat, however, decreased the likelihood of owners moving to a less severe symptomology profile in individuals with severe mental health symptom in the pre-lockdown period; these findings suggest that an overly strong owner–companion animal bond may cause additional mental stress during a global pandemic in individuals with pre-existing psychological health issues.

The research conducted thus far points to a somewhat confusing picture with regards to the role of the human–animal bond on the mental wellbeing of people during a global pandemic, although hints at the importance of relationship quality. Therefore the present study aimed to expand on the limited research in this area by investigating the relationship between attachment to one's companion animal and mental wellbeing. The research was deliberately conducted at the outset of a second period of national lockdown in the UK, a time of uncertainty and social isolation. It was hoped that the study would shed further light on the complex, and relatively understudied, relationship between the quality of the human–companion animal bond and psychological health during a global pandemic and help to elucidate what factors might influence any significant relationships unearthed. Such information could prove useful in helping to identify psychologically vulnerable individuals, an issue of importance given the predicted future rise in pandemic-induced mental health problems (e.g., Han et al., 2020).

Methods

Full ethical approval for the study was granted by the University’s Faculty Ethics Research Committee (EPS 20_259).

Sampling and Participants

UK-based residents over 18 years of age were invited to take part in this study. The online survey (see below) attracted a total of 384 responses. Following screening for inclusion criteria (provision of informed consent, UK resident, aged 18 + years) and quality of data (i.e., failure to complete the survey), 135 individuals were removed; the final dataset comprised 249 eligible participants. As can be seen from Table 1, most of the participants were female, under the age of 30 years, and had no children under the age of 18 to care for. The vast majority of the participants (n = 246, 98.8%) identified as being white/Caucasian; race was therefore not included as a predictor variable in any of the analyses. Most of the respondents lived with at least one other person in their household and engaged in social interactions outside the house on a weekly basis. Just over half of the sample reported living with at least one companion animal, with more of these individuals caring for a dog than a cat. A small number of respondents (n = 13) reported keeping other animals (e.g., lizards, fish, budgerigars) in addition to a dog or cat; given the low subject count for each type of species owned, only the ownership of dogs and/or cats was considered for the purpose of statistical analysis. In keeping with other studies in this area (e.g., Mueller et al., 2021; Saunders et al., 2017), analysis revealed significant sociodemographic differences between participants who owned a companion animal and those who did not for variables of gender (p = 0.03, Fisher exact test), age (χ²₍₄₎ = 11.09, p = 0.03), and parental status (p < 0.001, Fisher exact test). Specifically, compared with non-pet owners, pet owners were more likely to be female (62.1%) aged between 30–39 years old (82.4%), and have children to care for (75.0%).

Table 1. The number and percentage of participants (in total [n = 249] and broken down by companion animal owners [n = 146] and non-owners [n = 103]) according to demographic variables.

Demographic factor	Total sample n (%)	Companion animal owners n (%)	Non-owners n (%)
Gender
Male	54 (21.7)	25 (17.1)	29 (28.2)
Female	195 (78.3)	121 (82.9)	74 (71.8)
Age (in years)
18–29	132 (53.0)	72 (49.3)	60 (58.3)
30–39	34 (13.7)	28 (19.2)	6 (5.8)
40–49	37 (14.9)	22 (15.1)	15 (14.6)
50–59	35 (14.1)	20 (13.7)	15 (14.6)
60+	11 (4.4)	4 (2.7)	7 (6.8)
Residential status
Living alone	13 (5.2)	4 (2.7)	9 (8.7)
Living with 1 other person	50 (20.1)	24 (16.4)	26 (25.2)
Living with 2 other people	61 (24.5)	32 (21.9)	29 (28.2)
Living with 3 other people	47 (26.9)	53 (36.3)	14 (13.6)
Living with 4 + other people	58 (23.3)	33 (22.6)	25 (24.3)
Frequency of interactions
Every day	43 (17.3)	29 (19.9)	14 (13.6)
A few times a week	79 (31.7)	41 (28.1)	38 (36.9)
Once a week	82 (32.9)	50 (34.2)	32 (31.1)
Never	45 (18.1)	26 (17.8)	19 (18.4)
Parental status
Parent of child < 18	72 (28.9)	54 (37.0)	18 (17.5)
Not parent of child < 18	177 (71.1)	92 (63.0)	85 (82.5)
Animal ownership status
Companion animal owner	146 (58.6)
Non-owner	103 (41.4)
Type of companion animal
Dog	91 (62.3)
Cat	55 (37.7)

Survey

A purpose-designed questionnaire was developed that aimed to collect information on the sociodemographic background, companion animal ownership status, and psychological health of participants. The initial part of the survey collected demographic information, including details on respondents’ gender (male, female, other), race (open-ended), age (18–29, 30–39, 40–49, 50–59, 60 + years), and parental status (parent of child under 18, not parent of child under 18). As the study was conducted during a period of lockdown, but permitted meetings with others for exercise/wellbeing purposes, information was also collected on participants’ residential status (living alone, living with 1 other person, living with 2 other people, living with 3 other people, living with 4 + other people) and the frequency of interactions that took place outside the house (every day, a few days a week, once a week, never). Finally, information was collected on the companion animal ownership status of the respondents (companion animal owner, non-owner). People who lived with a companion animal were required to indicate what type of animal they owned (dog, cat, other [to be specified]) and to complete the Lexington Attachment to Pets Scale (LAPS, Johnson et al., 1992) to explore the strength of the animal–owner bond. The LAPS requires owners to assess their degree of agreement with 23 statements (e.g., “I consider my pet to be a friend”) on a 4-point Likert scale. The scale has been shown to have good internal consistency (coefficient alpha = 0.928) and examines emotional attachment to both dogs and cats. The survey is the most commonly used indicator of owner–pet attachment in studies of the human–animal bond (e.g., Bagley & Gonsman, 2005). Participants who owned more than one companion animal were asked to complete the LAPS for the animal they felt the most strongly connected to.

A number of scales were used to measure the psychological health of the respondents, with outcome measures chosen for focus because of their high prevalence during the COVID-19 pandemic (e.g., Bueno-Notivol et al., 2021; Salari et al., 2020):

Depression

The severity of depression was measured using the Patient Health Questionnaire (PHQ-9). This tool requires individuals to respond to nine questions (e.g., “over the last 2 weeks, how often have you been bothered by feeling down, depressed or hopeless?”) using a 4-point Likert scale, ranging from 0 (not at all) to 3 (nearly every day). Overall scores range from 0 to 27. The PHQ-9 is both a reliable and valid measure of depression severity and a useful tool in both clinical and research settings (Kroenke et al., 2001).

Positive Experience

Positive experience was explored using the SPANE-P (Diener et al., 2010), a 6-item scale that measures the frequency of positive emotions that people experience. The scale consists of three adjectives that describe general feelings (e.g., “pleasant”) and three that describe more specific emotions (e.g., “joyful”). Participants are required to indicate how often they have felt each of these during the last month, using a 5-point Likert scale which ranges from 1 (very rarely/never) to 5 (very often/always). Total scores range from 6 (very low positive experience) to 30 (very high positive experience). The SPANE has shown favorable psychometric properties in numerous studies (e.g., Jovanovic, 2015).

Loneliness

Loneliness was measured using the 3-item UCLA Loneliness Scale (Russell, 1996), a tool that shows strong reliability and validity (Russell, 1996). Individuals are required to respond to three questions (e.g., “how often do you feel you lack companionship?”) that are scored using a 3-point scale, ranging from 1 (hardly ever) to 3 (often). Overall scores range from low (3) to high (9) levels of loneliness.

Stress

Stress was assessed using the Perceived Stress Scale (PSS, Cohen et al., 1983), a tool that measures how unpredictable, uncontrollable, and/or overloaded people consider their lives. Several versions of the measure exist (PSS-14; PSS-10, PSS-4); the latter was employed in the present investigation to decrease participant burden, with individuals required to respond to four questions (e.g., “in the last month, how often have you felt that you were unable to control the important things in your life?”) using a 5-point Likert scale (0 = never through to 4 = often). Total scores range from low (0) to high (40) perceived stress. The psychometric properties of the scale are well documented (e.g., Karam et al., 2012) and it is the most widely used instrument for measuring perception of stress (Andreou et al., 2011).

Procedure

The survey was administered using the online platform Qualtrics and was advertised to the general public at the start of January 2021 using a variety of approaches, including word of mouth, social media (Facebook, Twitter), and social aggregation websites (e.g., Reddit). The advertisement indicated that the survey was concerned with exploring general health and wellbeing during a pandemic, rather than drawing specific attention to the focus on companion animal ownership or attachment. Interested individuals followed a link to the survey and, if still keen to take part, indicated their consent by checking a box. Individuals who did not meet the necessary inclusion criteria (i.e., because they were under the age of 18 years and/or were not a UK-based resident) were not allowed to complete the consent form or go any further with the study. The study remained open until January 31, 2021.

Data Analysis

Simple descriptive statistics were carried out to explore the frequency and percentage of responses to the sociodemographic information. A linear regression analysis was subsequently conducted to explore what factors were related to companion animal owners’ attachment levels. Overall LAPS score was set as the criterion variable, while factors of animal type (dog, cat), owner gender (male, female [none of the participants checked the “other” category]), age (18–29, 30–39, 40–49, 50–59, 60 + years), parental status (parent of child under 18, not parent of child under 18), residential status (living alone, lives with 1 other person, lives with 2 other people, lives with 3 other people, lives with 4 + other people) and the frequency of interactions that took place outside the house (every day, a few days a week, once a week, never) were set as the predictor variables. To assess the association between companion animal ownership (companion animal owner, non-owner) and mental wellbeing, a multivariable linear regression was conducted for each of the outcome measures (e.g., depression, loneliness), controlling for gender, age, residential status, parental status, and frequency of social interactions. The same analysis was carried out for the companion animal-owning cohort of the sample only, to explore whether the type of animal owned (dog, cat) and/or strength of the human–animal bond (overall LAPS score) was associated with any of the mental health outcome measures, adjusting again for the same variables outlined earlier (e.g., gender, age). Finally, Pearson's moment correlations were carried out to further explore any significant relationships between attachment to one's companion animal and mental health outcome measures. Data met the necessary assumptions for all statistical analyses.

Results

Companion Animal Attachment

Descriptive statistics revealed a mean LAPS score of 48.36 (SE = 0.78, range = 16–66). A total of 145 cases were analyzed for the logistic regression model concerned with attachment level, which was found to be significantly reliable (R² = 0.25, F_(6,145) = 6.96, p < 0.001). Gender was significantly associated with participants’ overall LAPS scores (see Table 2), with females showing higher levels of attachment to their companion animal (M = 55.36, SE = 0.82) than males (M = 47.68, SE = 2.03). Parental status also significantly predicted participants’ LAPS scores, with carers of children under the age of 18 having lower levels of attachment to their companion animals (M = 51.85, SE = 1.45) than non-carers (M = 55.33, SE = 0.91). Type of animal owned was also significantly associated with people's LAPS scores, with dog owners having higher levels of attachment (M = 56.08, SE = 0.93) than cat owners (M = 50.67, SE = 1.34).

Table 2. Results of the linear regression analysis for Lexington Attachment to Pets Scale scores involving predictor variables of participant gender, age, parental status, residential status, frequency of social interactions, and type of companion animal owned (n = 145).

Predictor	B	95% CI	t	p
Gender	8.18	4.43, 1.92	4.32	< 0.001
Age	−1.04	−2.30, 0.22	−1.63	0.10
Parental status	4.79	1.62, 7.96	3.00	0.003
Residential status	0.59	−0.86, 2.03	0.80	0.42
Social interaction	−1.18	−2.58, 0.22	−1.67	0.10
Type of animal	−5.07	−8.02, −2.11	−3.39	0.001

Residential status, age, and frequency of social interactions did not serve as reliable predictors of participants’ LAPS scores.

Depression

A multiple linear regression analysis was carried out to explore the association between companion animal ownership and depression, adjusting for covariates. A significant regression equation was found (F_(6,248) = 6.34, p < 0.001) with an R² of 0.14. Gender, age, parental status, and frequency of social interactions were all found to be significant predictors of depression scores (Table 3). Females had higher PHQ scores (M = 10.90, SE = 0.54) than males (M = 7.67, SE = 0.86), people with children to care for had higher depression scores (M = 11.64, SE = 0.96) than those without (M = 9.61, SE = 0.53), and younger participants had higher depression scores (18–29 years: M = 11.20, SE = 0.58; 30–39 years: M = 10.21, SE = 1.31; 40–49 years: M = 11.24, SE = 1.45) than older individuals (50–59 years: M = 6.80, SE = 1.31; 60 + years: M = 5.45, SE = 1.25). In addition, people who met up with others more frequently had lower depression scores (every day: M = 7.93, SE = 6.52; a few times a week: M = 9.59, SE = 7.04) than those who met with others less often (once a week: M = 10.29, SE = 7.63; never: M = 13.24, SE = 7.68).

Table 3. Results of the linear regression analysis for depression (PHQ) scores involving predictor variables of participant gender, age, parental status, residential status, frequency of social interactions, and companion animal ownership status (n = 249).

Predictor	B	95% CI	t	p
Gender	3.03	0.90, 5.19	2.77	0.006
Age	−1.28	−2.01, −0.54	−3.42	0.001
Parental status	−2.29	−4.47, 0.13	−2.08	0.04
Residential status	−0.11	−0.95, 0.72	−0.27	0.79
Social interaction	1.50	0.61, 2.40	3.30	0.001
Animal ownership	−0.42	−2.25, 1.41	−0.45	0.65

Residential status and companion animal ownership were not found to be useful predictors of respondents’ depression scores.

A further regression analysis was carried out for the companion animal-owning cohort of participants to explore whether the type of animal owned and/or attachment level predicted depression (adjusting for covariates). The full model was found to be significantly reliable (F_(7,145) = 9.14, p < 0.001; R² = 0.32). Attachment level was found to be the strongest predictor of participants’ PHQ scores (Table 4). Higher levels of attachment to one's companion animal were associated with significantly (r = 0.10, n = 146, p < 0.001) higher levels of depression.

Table 4. Results of the linear regression analysis for depression (PHQ) scores for pet owners involving predictor variables of gender, age, residential status, parental status, frequency of social interactions, type of companion animal owned, and LAPS scores (n = 145).

Predictor	B	95% CI	t	p
Gender	−0.85	−3.84, 2.14	−0.56	0.57
Age	−0.16	−1.11, 0.79	−0.33	0.74
Parental status	−3.36	−5.81, −0.91	−2.71	0.008
Residential status	−0.26	−1.35, 0.82	−0.48	0.63
Social interaction	2.57	1.51, 3.63	4.79	< 0.001
Type of animal	1.98	−0.32, 4.29	1.70	0.09
LAPS score	0.40	0.28, 0.53	6.36	< 0.001

Parental status was also found to be a significant predictor of depression scores, with people who had children under the age of 18 to look after having higher scores for this outcome measure (M = 11.50, SE = 1.14) than individuals without this responsibility (M = 10.18, SE = 0.73). The frequency of social interactions also significantly predicted depression scores. People who met up with others every day had significantly lower PHQ scores (M = 7.17, SE = 6.50) than those who met up with others less frequently (a few times a week: M = 10.63, SE = 1.14; once a week: M = 10.46, SE = 1.05; never: M = 15.04, SE = 1.44).

Participant age, gender, residential status, and type of companion animal owned did not significantly predict people's PHQ scores.

Positive Experience

A multiple linear regression was carried out to predict positive SPANE scores based on a number of predictor variables and adjusting for covariates. A significant regression equation was found (F_(6,248) = 4.61, p < 0.001) with an R² of 0.10. Gender, age, and frequency of social interactions were found to be significant predictors of SPANE scores (Table 5). Males had higher positive SPANE scores (M = 20.35, SE = 0.58) than females (M = 18.03, SE = 0.38), while older people had higher positive SPANE scores (60 + years: M = 21.91, SE = 1.69; 50–59 years: M = 20.20, SE = 0.99) than younger individuals (18–29 years: M = 17.86, SE = 0.44; 30–39 years: M = 19.26, SE = 0.82; 40–49 years: M = 17.70, SE = 0.82). People who met up with others every day had significantly higher positive SPANE scores (M = 20.44, SE = 0.69) than people who met with others a few times a week (M = 19.10, SE = 0.55), once a week (M = 18.04, SE = 0.62), or never (M = 16.62, SE = 5.26).

Table 5. Results of the linear regression analysis for positive experience (SPANE) scores involving predictor variables of participant gender, age, residential status, parental status, frequency of social interactions, and companion animal ownership status (n = 249).

Predictor	B	95% CI	t	p
Gender	−2.34	−3.88, −0.80	−2.99	0.003
Age	0.61	0.08, 1.13	2.28	0.02
Parental status	−0.09	−1.64, 1.45	−0.12	0.90
Residential status	−0.18	−0.78, 0.41	−0.61	0.54
Social interaction	−1.17	−1.81, −0.53	−3.60	< 0.001
Animal ownership	0.57	−0.73, 1.87	0.86	0.39

Companion animal ownership, residential status, and parental status were not found to significantly predict SPANE scores.

The same analysis was carried out on the companion animal-owning group of participants, yielding a significant regression model (F_(7,145) = 4.72, p < 0.001) with an R² of 0.19. The level of attachment to one's companion animal significantly predicted positive experience (Table 6). A Pearson's moment correlation yielded a significant (r = –0.27, n = 146, p = 0.001) negative relationship between the two variables, with a higher degree of attachment to one's companion animal being associated with lower positive SPANE scores. The frequency of social interactions was also found to be a significant predictor of positive experience. Again, companion animal owners who met up with others every day had significantly higher positive SPANE scores (M = 20.44, SE = 0.69) than those who met up with others less frequently (a few times a week: M = 19.10, SE = 0.55; once a week: M = 18.04, SE = 0.62; never: M = 16.62, SE = 0.78). None of the other variables were found to significantly predict SPANE scores.

Table 6. Results of the linear regression analysis for positive experience (SPANE) scores for pet owners involving predictor variables of gender, age, residential status, parental status, frequency of social interactions, type of companion animal owned, and LAPS scores (n = 145).

Predictor	B	95% CI	t	p
Gender	−0.76	−3.09, 1.57	−0.64	0.52
Age	−0.05	−0.80, 0.69	−0.14	0.89
Parental status	0.04	−1.87, 1.95	0.04	0.97
Residential status	−0.59	−1.43, 0.26	−1.37	0.17
Social interaction	−1.69	−2.51, 0.86	−4.04	< 0.001
Type of animal	−0.76	−2.55, 1.04	−0.83	0.41
LAPS score	−0.16	−0.26, 0.07	−3.32	0.001

Loneliness

The full regression model was found to be significantly reliable for the dependent variable of loneliness as measured by the UCLA (F_(6,248) = 4.61, p < 0.001), with an R² of 0.10. Gender, age, and frequency of social interactions were significantly associated with participants’ UCLA scores (Table 7). Female participants reported higher loneliness scores (M = 6.32, SE = 0.14) than males (M = 5.70, SE = 0.23), while younger individuals had higher loneliness scores (18–29 years: M = 6.42, SE = 0.15; 30–30 years: M = 6.38, SE = 0.33; 40–49 years: M = 6.30, SE = 0.32) than those aged 50–59 (M = 5.26, SE = 0.38) and 60+ (M = 5.27, SE = 0.56). People who met with others every day had lower loneliness scores (M = 5.42, SE = 0.26) than those who met a few times a week (M = 6.10, SE = 0.20), once a week (M = 6.35, SE = 0.22), or never (M = 6.76, SE = 0.29).

Table 7. Results of the linear regression analysis for loneliness (UCLA-3 item) scores involving predictor variables of participant gender, age, parental status, residential status, frequency of social interactions, and companion animal ownership status (n = 249).

Predictor	B	95% CI	t	p
Gender	0.64	0.08, 1.20	2.25	0.03
Age	−0.29	−0.48, −0.10	−2.99	0.003
Parental status	−0.10	−0.66, 0.46	−0.35	0.72
Residential status	−0.02	−0.22, 0.21	−0.02	0.98
Social interaction	0.38	0.15, 0.61	3.20	0.002
Animal ownership	−0.11	−0.59, 0.37	−0.47	0.64

A further linear regression was carried out for the companion animal-owning group of participants. A significant regression equation was found (F_(7,145) = 5.57, p < 0.001) with an R² of 0.22. The level of attachment to one's companion animal, again, significantly predicted loneliness (Table 8). A Pearson's moment correlation yielded a significant (r = 0.32, n = 146, p < 0.001) positive relationship between the two variables, with increasing levels of attachment to one's companion animal associated with increasing loneliness. The frequency of social interactions was also found to significantly predict UCLA scores. People who met up with others every day had lower UCLA scores (M = 5.42, SE = 0.26) than those who met up with others less frequently (a few times a week: M = 6.10, SE = 0.20; once a week: M = 6.35, SE = 0.22; never: M = 6.76, SE = 0.29). None of the other variables served as significant predictors of loneliness.

Table 8. Results of the linear regression analysis for loneliness (UCLA-3 item) scores for pet owners involving predictor variables of gender, age, residential status, parental status, frequency of social interactions, type of companion animal owned, and LAPS scores (n = 145).

Predictor	B	95% CI	t	p
Gender	−0.13	−0.94, 0.68	−0.32	0.75
Age	−0.12	−0.38, 0.14	−0.92	0.36
Parental status	−0.37	−1.03, 0.30	−1.09	0.28
Residential status	0.02	−0.27, 0.32	0.17	0.87
Social interaction	0.60	0.31, 0.88	4.10	< 0.001
Type of animal	0.33	−0.30, 0.95	1.04	0.30
LAPS score	0.08	0.04, 0.11	4.39	< 0.001

Stress

A multiple linear regression was carried out to predict PSS scores based on a number of predictor variables and adjusting for covariates. The regression equation was found to be non-significant (F_(6,248) = 1.01, p = 0.42) with an R² of 0.02. None of the independent variables significantly (p > 0.05) predicted the participants’ stress scores (Table 9). The same analysis was carried out on the companion animal-owning cohort of participants only, and, again, showed a non-significant model (F_(7,145) = 1.91, p = 0.07) with an R² of 0.09 (Table 10).

Table 9. Results of the linear regression analysis for stress (PSS) scores involving predictor variables of participant gender, age, parental status, residential status, frequency of social interactions, and companion animal ownership status (n = 249).

Predictor	B	95% CI	t	p
Gender	0.10	−0.46, 0.66	0.34	0.73
Age	−0.07	−0.26, 0.12	−0.68	0.49
Parental status	−0.40	−0.97, 0.16	−1.41	0.16
Residential status	0.02	−0.20, 0.24	0.19	0.85
Social interaction	0.20	−0.04, 0.43	1.66	0.10
Animal ownership	0.01	−0.46, 0.49	0.05	0.96

Table 10. Results of the linear regression analysis for stress (PSS) scores for pet owners involving predictor variables of gender, age, residential status, parental status, frequency of social interactions, type of companion animal owned, and LAPS scores (n = 145).

Predictor	B	95% CI	t	p
Gender	−0.54	−1.36, 0.28	−1.30	0.19
Age	0.18	−0.08, 0.44	1.33	0.18
Parental status	−0.48	−1.15, 0.18	−1.44	0.15
Residential status	0.17	−0.13, 0.46	0.10	0.27
Social interaction	0.38	0.09, 0.67	2.57	0.01
Type of animal	0.08	−0.54, 0.71	0.26	0.79
LAPS score	0.03	−0.01, 0.06	1.58	0.12

Discussion

This study explored the relationship between the quality of the human–animal bond and mental wellbeing during the second period of COVID-19-induced lockdown (January 2021) in the United Kingdom. The results from this study showed no significant effect of living with a companion animal (i.e., companion animal “ownership”) on participants’ mental health, as assessed by validated scales designed to measure depression, loneliness, stress, and perceived positive emotions. The literature on the so-called “pet effect” is somewhat conflicting in nature (for a recent review see Wells, 2019), but the findings from this study are in line with many investigations published more recently (e.g., Mueller et al., 2021; Teo & Thomas, 2019). Interestingly, Ratschen and associates (2020) found that companion animal ownership, regardless of species kept, was associated with less deterioration in mental health and smaller increases in loneliness during the first period of national lockdown in the UK, although this study required retrospective data collection and reported upon changes in mental wellbeing. Differences in methodological approach, sample size, and populations under scrutiny are likely to explain some of the variation in results published in this field (for a discussion of these issues, see Herzog, 2011; Wells, 2009).

No significant difference in the mental health status of dog versus cat owners was observed in this study. One might have expected dog owners to have reported greater psychological wellbeing than cat owners, given the reported merits of dogs in helping to enhance their owners’ self-esteem, facilitate social contacts with others, and ward off feelings of depression and loneliness (e.g., Siegel, 1990; Zasloff & Kidd, 1994). That said, this study was conducted during a period of lockdown, when people were encouraged to socially distance; dog walking in this context, whilst normally serving as a strong social catalyst (e.g., McNicholas & Collis, 2000; Messent, 1983; Wells, 2004), may therefore have been perceived as a stressful activity by some owners. Other issues relevant to this particular timeframe, such as possible challenges in seeking veterinary care, accessing pet food, and even the expense involved in caring for an animal at a time when people's livelihoods were more insecure, may have played a role in negating any potential group-level health benefits for this specific cohort (see Applebaum et al., 2020; Shoesmith et al., 2021).

Dog owners were found to be more strongly attached to their companion animals than cat owners, a result that concurs with other studies (e.g., Lass-Hennemann et al., 2020; Le Roux & Wright, 2020; Mueller et al., 2021; Muldoon et al., 2019; Winefield et al., 2008). This discrepancy in attachment is most likely related to the social nature of these animals. The domestic cat is a solitary creature, while the dog is a social species with a proclivity for developing strong bonds of attachment, particularly to humans (Topal et al., 2005; Udell & Brubaker, 2016). Dogs are also more likely than other companion animals to respond to human emotions, even adapting their behavior in response to their carers’ emotional cues, thereby encouraging closer bonds of attachment (Payne et al., 2015). Interestingly, cats have been shown to display important indicators of human-directed attachment, including proximity seeking, separation anxiety, and reunion behavior (Vitale et al., 2019); however, cats do not appear to attach to their owners as a focus of security or safety in the same manner as dogs, which may have implications for how strongly owners become attached to them.

Whilst this study found no significant effect of companion animal ownership on mental wellbeing, the findings point to a very significant influence of the quality of the human–animal relationship on psychological health, with a stronger attachment to one's companion animal associated with higher levels of depression and loneliness and lower levels of perceived positive experience. Some of the sociodemographic variables, notably gender, age, and parental status, also served as significant predictors of psychological ill-health. In keeping with other studies (e.g., Evenson & Simon, 2005; Kessler et al., 2007; Riecher-Rossler, 2016), females, younger individuals, and those with parenting duties were found to be particularly prone to mental health problems. Importantly, and in line with previous work in this area (Peacock et al., 2012), albeit prior to COVID-19, attachment to one's companion animal was found to be a stronger predictor of the outcome measures than any of the other independent variables.

The literature in this area is conflicting in nature, with some studies pointing to a buffering influence of strong companion animal attachment on owners’ mental wellbeing (e.g., Garrity et al., 1989; Teo & Thomas, 2019; Wen Li et al., 2017). Other studies, however, are more in line with the results of the present investigation. For example, Miltiades and Shearer (2011) reported higher levels of depression in older adults with higher pet attachment than in individuals less well attached to their pets, while Hartwig and Signal (2020) reported a trend for increasing levels of loneliness with increasing pet attachment in a group of Australian adolescents. Interestingly, within the context of COVID-19, Ratschen and colleagues (2020) reported no significant relationship between companion animal attachment and the mental health of people during the first UK-based national lockdown, whilst McDonald and others (2021) found that the protective benefits of being highly attached to one's companion animal were correlated with the mental health status of the owner prior to lockdown; people with moderate to high levels of mental health symptoms pre-pandemic seemed to gain more health benefits from their companion animal than individuals reporting severe mental health symptoms. Although not carried out during a strict lockdown, Namekata and Yamamoto (2021) reported a relationship between high attachment to one's pet and positive mood in a group of university students majoring in animal science during the COVID-19 pandemic in Japan. Interestingly, Kogan and associates (2021) found that women under the age of 40 who were highly bonded to their companion animal gained the greatest mental health benefits during the initial months of the COVID-19 pandemic, although validated mental health outcomes and companion animal attachment scales were not used in this study, making it difficult to compare the results with those of the present investigation.

Stress was the only variable not found to be significantly associated with companion animal attachment in this study. This is an interesting finding given suggestions that a strong owner–companion animal bond may cause additional mental stress during a global pandemic (see McDonald et al., 2021). Outside the timeframe of COVID-19, Le Roux and Wright (2020), using the same scale to measure stress as employed in the current investigation (PSS), found no relationship between owners’ perceived stress and pet attachment. However, a Chinese study reported a negative correlation between pet attachment and stress (Wu et al., 2018).

The findings from this study beg the question as to why higher levels of attachment should be associated with poorer psychological wellbeing. One might intuitively have expected the opposite result when considered in relation to Bowlby's theory of attachment (1969) and the known merits of secure attachment and positive mental health (for a review, see Mikulincer & Shaver, 2012). A range of theories may be worth considering. It has been suggested that people predisposed to psychological problems might be more likely to obtain an animal and bond strongly to it as a sort of self-help strategy (Mullersdorf et al., 2012). Alternatively, or equally, it may be the case that people who are strongly attached to their companion animals lack quality relationships with others, thereby indirectly leading to, for example, increased loneliness and depression (e.g., Chur-Hansen et al., 2009). This hypothesis does not seem to fit with the data arising from the present investigation, however, as there was no evidence of any significant relationship between attachment level and social support, either in terms of the number of people who participants lived with or how often they engaged in social interactions outside the house. That said, living, and/or meeting up, with others does not necessarily equate to social support satisfaction or the presence of well-functioning relationships, although meeting up with people on a regular basis (regardless of relationship strength) certainly seems to serve as a useful predictor of mental health outcomes, at least in this study, with more frequent social interactions associated with better psychological wellbeing.

It may be the case that attachment to one's pet is not a linear construct but, rather, that extreme levels of attachment (either at the high or low end) signal problems with psychological health. Indeed, Chur-Hansen and colleagues (2009) proposed that the relationship between attachment to one's companion animal and any health benefits incurred may follow an inverted U-shaped curve, with optimal health benefits lying at the moderate points on the curve and extreme levels of attachment (both low and high) being detrimental to a person's health (see also Hill et al., 2020). Again, our data did not support this hypothesis, with no evidence of low levels of attachment having as detrimental an impact on psychological wellbeing as high levels of attachment; perhaps a larger dataset, with a wider spread of attachment levels, would yield different results.

Owner personality may be a variable worthy of exploring in the context of companion animal attachment and wellbeing. One might expect high attachment levels to be associated with personality traits that are closely aligned to psychological ill-health. The link between personality and attachment style/levels has received surprisingly little attention, but some studies do seem to point to a relationship. For example, a positive correlation between attachment to one's pet and neuroticism was reported (Reevy & Delgado, 2015, 2020). This is interesting given the link between this specific personality trait and psychological health disorders, notably depression and anxiety (e.g., Angst & Clayton, 1986; Hirschfeld et al., 1989; Xia et al., 2011).

Limitations

There are limitations to this investigation that must be acknowledged. Given the online nature of the study (a methodological approach deliberately adopted to facilitate data collection during a global pandemic), it is possible that recruitment attracted a certain cohort of people. As with most studies in this area (for a discussion of this issue, see Wilson & Barker, 2003), the majority of participants were female, a variable found to be associated with both companion animal attachment and mental wellbeing. Although challenging, it would be useful for future studies to focus more specifically on male companion animal owners, particularly in light of the difference in attachment styles that exist between the sexes (e.g., Barry et al., 2015; Scharfe, 2017). That said, gender differences have not been reported for various human–animal-related issues, including intimacy of the owner–companion animal bond (Evans-Wilday et al., 2018); our female-strong sample, therefore, may not have had a detrimental impact on our findings.

Gender aside, the sample recruited for this study was relatively young (with roughly half of the participants under the age of 30) and extremely limited in terms of its racial diversity; whether findings can be generalized to older individuals or those of other racial and ethnic identities is unknown. Future research should consider the role of these variables, in addition to the myriad of other factors that may impact upon health outcomes: for example, socioeconomic status, income, and education level. Indeed, it is considered a necessary exercise to widen the sampling pool more generally, in order to allow for greater generalizability of the results than are allowed for by the present investigation, particularly considering the relatively small sample size (n = 249) and somewhat “niche,” self-selected sample.

Whilst the scale used to assess pet attachment (LAPS) in the present study is the most commonly employed one in this area, a response bias leaning towards higher attachment cannot be ruled out (in this study, 60% of the participants had attachment scores higher than the mean). Future studies may be able to address this by including additional, perhaps more objective, measures of pet attachment (e.g., recording frequency of physical contacts between owners and their pets; measuring oxytocin levels) and exploring the relationship between these types of attachment tool. Recruiting participants with low levels of attachment to their pets, although potentially challenging, may yield some interesting information. Also, some consideration is required regarding the completion of the LAPS by participants with multiple pets. Although not problematic for the current investigation (given the lack of relationship between animal species and any outcome measures), participants with more than one companion animal might perhaps be encouraged to complete the survey on behalf of their favorite animal and to indicate what species this is.

Finally, research in this area, including the present investigation, has been dominated by cross-sectional studies; longitudinal research is still sorely lacking and, although challenging to undertake, would be useful in helping to resolve some of the issues that are inherent in studies of this nature (for a discussion, see Herzog, 2011; Wells, 2019).

Conclusions

Overall, the findings from this study point to stronger companion animal attachment being associated with poorer mental health during the second period of national lockdown in the United Kingdom. The research highlights the emotional vulnerability of people who develop a strong bond of attachment to the animals in their care, although the limitations highlighted above must be taken on board. Whether the quality of the human–animal bond harbors potential as a tool for predicting emotional vulnerability and/or mental health issues is still unknown, but the findings from this study certainly point towards an association. As acknowledged by others (e.g., Johnson et al., 1992), the direction of the companion animal–owner relationship is complex and, at times, contradictory. It is clear from the present investigation, and other studies conducted during the pandemic (e.g., McDonald et al., 2021), that companion animals should not necessarily be seen as a solution to owners’ health problems. Further research is recommended to explore the variables that impact upon the human–animal bond, taking into consideration additional individual differences (e.g., owner personality, resilience) that may moderate the physical, mental, and emotional health of the person and determine the degree to which one gains any health benefits from their companion animal. This is an area worthy of further exploration considering the recent pandemic-induced rise in dog and cat ownership and the increment in mental health problems that has been predicted to emerge from COVID-19.

Acknowledgements

The authors thank all of those individuals who took part in this study.

Disclosure Statement

No potential conflict of interest was reported by the authors.