Introduction

In the last 30 years there has been a growing literature base about the health benefits to humans, of companion animal ownership. Although a number of studies have considered the physical and psychological health benefits of interaction with a companion animal, the overall results are inconclusive. Nevertheless, there is a popular belief that companion animals are 'good for us'.

This paper briefly reviews some of the claims made by researchers regarding the benefits of companion animals for human physical and mental health. We then comment on the major identifiable gaps in the research evidence, to explain why our understanding is not complete despite many studies having been undertaken. Associated with each gap we offer suggestions for remedies or needed next steps in research. Finally, we discuss a specific area - the use of animal-assisted interventions (AAIs) for the elderly in residential care. We highlight the methodological weaknesses in research that make claims that animals benefit the elderly a dubious conclusion, and one that is lacking in a solid evidence base. This paper is not an exhaustive literature review but rather, a synthesis of knowledge and ideas aimed to stimulate a better quality evidence base for future research. In this paper we focus on the adult literature (and not children). We refer to companion animals as any non-human animal that shares its life with a human caregiver. This is distinct from AAIs, which relates to any therapeutic process that intentionally includes or involves animals as part of the process. AAIs can be grouped as either animal-assisted activities (AAAs) or animal-assisted therapies (AATs). AAAs refer to 'the utilisation of animals that meet specific criteria to provide participants with opportunities for motivational, educational, recreational, and/or therapeutic benefits to enhance quality of life'1 while AATs are 'goal-directed interventions directed and/or delivered by a health/human service professional with specialised expertise, and within the scope of practice of his/her profession'.2

Examples of the claims of recent reviews

Animals have been attributed with positive effects on humans in a number of areas. Cutt et al., in a review of dog ownership, health and physical activity, argue that there is considerable evidence to suggest that living with a dog encourages walking, facilitates health benefits and increases social supports.3 Consistent with this, Wells states in her review that dogs have prophylactic and therapeutic value for human psychological and physical health.4 Barker and Wolen acknowledge that many studies are descriptive, but nevertheless conclude that research supports the health benefits of interacting with companion animals.5 In a review on the benefits of assistance dogs, such as those who help people who have mobility problems or who cannot hear,6 the authors conclude that the literature is so small and the methodologies so flawed, any statements about the real benefits or otherwise of service animals cannot be made.

Filan and Llewellyn-Jones reviewed the literature on AAT for people with dementia and stated that while the duration of beneficial effects has not yet been explored, the presence of dogs, an aquarium and robotic pets may be considered helpful for the behavioural and psychological symptoms of dementia.7 A meta-analysis conducted by Nimer and Lundahl concluded that AAT is associated with moderate effect sizes in improving outcomes in Autism-spectrum symptoms, medical difficulties, behavioural problems and emotional well-being.8

Inconclusive results about the companion animal-human health connection

The conclusions that can be drawn from the present literature on the health effects of owning a companion animal are mixed, and the causal mechanisms are unclear.9-11 Some studies conclude that companion animals are beneficial to health. Other authors have reported that any claims that companion animal ownership is beneficial should be viewed with caution,12 citing the weak methodologies used to investigate the hypothesis and the preponderance of anecdotal and biased research. For example, self-selection of companion animal owners is likely to introduce all kinds of biases that obscure the proper interpretation of results, as would allowing patients in a drug trial to choose whether they tried the new drug or stayed on the old one, or allowing medical practitioners to choose which patients should enter the active treatment group testing a new drug. To date there are no rigorous, randomised double-blind controlled clinical trials to investigate the question of whether companion animals are beneficial for psychological or physical health, as would be expected for any other therapeutic intervention.

Recently the literature on companion animal ownership in the elderly presents data that argue the association between companion animal ownership and health is, in fact, negative. Parslow and Jorm found that companion animal owners did not have reduced systolic blood pressure (as has been reported in previous research), but had higher diastolic blood pressure, higher body mass index and were more likely to smoke cigarettes as compared with those without companion animals.13 In a case review of patients over 75 years of age, Kurrie et al. concluded that companion animals might pose a hazard for the elderly, by increasing the likelihood of falls (interestingly, in their case review one cat fatality was also recorded, when its falling elderly owner crushed it to death).14 Nair and Flynn noted companion animal-related injuries, and some of these (usually through dog attacks) can be serious or even fatal.15 Thompson showed that about half the respondents in a large random survey expressed fear of dog attacks and for half of those, this fear restricted their behaviour.16 Thus other people's companion animals might pose a health hazard.

Parslow et al. concluded from a large cross-sectional survey that companion animal owners reported more depressive symptoms, that married female companion animal owners had poorer physical health, and that caring for a companion animal was associated with symptoms of depression, poorer physical health, higher rates of pain relief medication and higher levels of psychoticism as assessed by the Eysenck Personality Questionnaire (usually taken as an indicator of impulsivity, autonomy and aggression).17

Thus, with the important question 'Are companion animals beneficial for health?' as yet unanswered conclusively, there is scope, and a need, for further research in the area that addresses the weaknesses.

Most of the literature has been concerned with the relationship between companion animal ownership and chronic conditions such as cardiovascular disease, and risk factors such as exercise and blood pressure levels. Depression and social isolation are risk factors for heart disease and also reduce the quality of life of those struggling with chronic illnesses of every type.18 There is high comorbidity of physical and emotional distress. The Australian Longitudinal Study of Aging found strong interconnections between physical, psychological and social functioning, in their large sample of 1403 community-living adults aged over 70 years.19 Two mechanisms by which companion animal ownership may reduce the burden of illness, which need to be examined in a rigorous method, are increased exercise and decreased depression. Both may mediate benefits of companion animal ownership in people with inadequate previous levels of physical activity, social support, and sense of personal value and worth.

First gap in the evidence: weak research designs

The companion animal-health literature has been fairly criticised for its preponderance of descriptive and cross-sectional research designs. These, even with a longitudinal element, do not allow conclusions to be confidently drawn about whether or for whom companion animal ownership might be recommended as a health-promoting measure. Studies are needed that are based upon the quantitative methodologies used to assess other healthcare strategies, namely randomised double-blind controlled intervention trials. The populations employed would vary, but could include those people living in the community, in psychiatric facilities and in residential aged care. Such research would be an important contribution to the literature. The focus of research would be an investigation of psychological health, along with physical and physiological parameters. Because of the increased incidence of chronic illness in older people, research could focus on those aged over 65 years. However, randomised double-blind controlled intervention trials for individuals with any illness are plausible and would be valuable additions to knowledge.

Before funding is likely to be made available for ambitious projects such as these, more preparatory studies are needed. For example, if conducting research into elderly peoples' companion animals and health, we could expect that about half of households of 65+ year olds will already include a companion animal,17 but we need first to discover how many of those might agree to accept one if offered recompense (in the form of companion animal food, money or vet bill vouchers for example) in return for completion of research measures. As suggested by Furber,20 we need to identify elderly people's experiences and interests regarding companion animal ownership, the problems associated with owning a companion animal, and reasons why a companion animal may not be wanted, or had to be relinquished.

While a large randomised trial with placebo controls and double-blind assessments of outcome is difficult to conduct,21 much useful information could be gained from an intervention study where companion animals are given to elderly people who do not have one, with adequate longitudinal follow ups. One of the only intervention studies was conducted by Serpell.22 He compared the health and mental health of new companion animal owners with non-owners, over 10 months. Unfortunately his report lacked any information about the ages of participants or the method of recruitment of non-owners. Worse, there was no randomisation, so the mild benefits he found especially for dog ownership may be based on self-selection into groups. Allen et al. did conduct a randomised trial of companion animal ownership effects on hypertension, by telling half their sample of living-alone stockbrokers with pathologically high blood pressure to adopt a dog or cat as a companion animal.23 All also started medication, which succeeded in lowering their blood pressure; however, those who received a companion animal responded to mental stressors such as arithmetic tasks with only half as much reactivity in terms of blood pressure elevations. The authors attributed the benefit of companion animal ownership to the mechanism of non-judgmental social support provided by the companion animals. Replication with participants of more diverse living arrangements and socio-economic status is desirable.

For an intervention study, randomly selected consenting older people could be invited to choose a companion animal cat or dog from a shelter, and their health would be re-assessed after 6 and 12 months. Waiting-list controls could be used. To increase ecological validity only some of the financial costs of companion animal ownership should be defrayed by the researchers, who would also need to plan for contingencies such as owner or companion animal ill-health, or owner-companion animal rejection or temperamental mismatch. It would not of course be possible to 'blind' the participants as to whether they are receiving the active treatment (companion animal) or a credible placebo (whatever the researchers might plan that to be). However, outcome assessment by workers blind to the treatment is very desirable, for example to reduce the risks of inadvertent bias in reported health and well-being, according to the pre-existing expectations of either the research participants or the data collector.

Positive relationships between companion animal ownership and both physical and psychological health may prove to be mediated by baseline levels of physical activity, social supports and feelings of self-worth. If dog owners engage in more exercise than people with other companion animals, might people whose activity level is low show health benefits from acquiring a dog? If owning a dog increases exercise in the able elderly, it could be promoted through public health campaigns and face-to-face consultations with healthcare professionals.24

Alternatively, the main benefit of companion animal ownership might be found to be facilitation of increased social supports, in people whose supports were previously inadequate. For example, they might begin to attend companion animal clubs or classes, or to interact socially with other companion animal owners. The non-judgmental nature of companion animal support might be particularly valuable for people who lack social confidence, or whose recent social experiences have been unsatisfying. The unconditional, non-evaluative nature of a companion animal's emotional support may make their company less stressful than that of a human peer, as was suggested by Allen et al.25

Longitudinal qualitative research would also be valuable to help us understand how companion animal ownership might change in its impact upon an individual over time. The research evidence base lacks in-depth information from qualitative research conducted without prior assumptions. Qualitative research has the advantage of being open-ended; themes may be identified that have not previously been considered as important, and these may be pivotal in helping to understand the mechanisms at work in the relationship of companion animal ownership to health.

To overcome the problem of researchers finding only what they expect, open-ended and deep interviews of companion animal owners and companion animal non-owners are desirable. Qualitative research into the health benefits of companion animals certainly does exist. However, there is a tendency for these studies to be descriptive, rather than generating new hypotheses or theories that can be further explored in subsequent research. The trustworthiness and defensibility required in qualitative methodologies is lacking in some of the current literature:26 thus there is considerable scope for worthwhile contributions to our knowledge through high-quality qualitative research. The specific methodology chosen is not as important as the rigour of the method. Thus, new research could include ethnographic studies (drawn from the discipline of anthropology), around companion animals and health. This would involve carrying out fieldwork based on participant observation. For example, a researcher might choose to spend a year following a specific group of people and their companion animals: fertile material would be found in nursing homes with a shared companion animal, or in the homes and lives of people who rely upon a guide dog, to give only two examples. Another avenue of qualitative research could involve collecting life histories and narratives from people who reflect on the possible relationships between companion animals and their psychological and physical health. Such data may serve as valuable avenues of enquiry and sources of plausible hypotheses that can be systematically explored in subsequent quantitative research such as surveys. A combination of semistructured interviews with standardised questionnaires measuring physical and mental health may also be pursued; this type of mixed methodology is particularly appropriate for applied health research.26

Groups with vested interests have funded some of the companion animal-health research, such as the companion animal food or companion animal care industry. Such potential conflicts of interest risk being perceived as problematic, in an era of growing awareness of the influence on drug companies on medical research.27 Solutions may be to ask researchers for declarations of any conflict of interest (as is now usually the case when submitting to peer-reviewed journals), to insist on transparency of study design and findings, and where possible, conducting research independently of bodies who may favour one outcome over another.

Second gap in the evidence: failure to control for other influences

Pachana et al. found confounding of companion animal ownership with sociodemographic factors such as income, household size, area of residence and usual activity levels in a longitudinal survey of elderly women.28 As they point out it is difficult to disentangle the direction of causality, in the association between better health and companion animal ownership. Research studies are needed that take into account variables that have been previously less well considered, including the type of social interactions available to the companion animal owner, their leisure time, their financial resources,29 and the important but often ignored variable of attachment, the emotional bond between the owner and the animal.30

Some parts of the research evidence base suffer from a lack of standardised measures. Reported health habits such as exercise and smoking can in the right context be reasonably reliable if based on self-report, but more objective measures are potentially available, such as pedometers (for steps walked), salivary cotinine (for smoking), and structured observations of mobility and fitness. Health and psychological well-being measures have been standardised within the health psychology and public health domains and are now readily available and of known reliability and validity; these include both self-report and physiological forms such as blood pressure, salivary cortisol and body mass index. As noted by McNicholas et al.,11 quality of life is an important dimension of health in addition to the more traditional biomedical and risk factor assessments.

To gauge the owner's emotional relationship with the companion animal, researchers have tried looking at whether the owner has sole, shared or no responsibility for companion animal care. However, that information may not adequately capture the psychologically important aspects of their relationship, for example the undemanding/unconditional nature of the animal's 'affection', or the sense of security and self-worth associated with the reciprocal emotional bond known to psychologists as attachment.31 A possible downside of intense owner-companion animal attachment is the owner's isolation from human contacts because of companion animal care responsibilities. Thus, there is a need to examine owners' human social supports, and more fully explore owners who are fiercely attached to their companion animal to the exclusion of human relationships, including the reasons behind such strong attachments. Stallones et al. found that for participants aged 45-64 years in a large national survey, individuals with high attachment to a companion animal had fewer human social supports.32 Some people with, for example, chronic psychological illness might find human relationships too challenging and for them a companion animal might be a perfect companion; however, such people might be particularly vulnerable to pathological grief when the companion animal dies.33

Another need is the refinement of psychometric scales to measure companion animal attachment. A number of authors have acknowledged that a flaw in their research has been the lack of attachment measures, a consideration of which may help to explain the contradictory results that characterise the current literature. Although two existing scales have high internal reliability,32,34 their theoretical and pragmatic origins are unclear and from a psychological viewpoint, some crucial additional items may prove to increase validity. Attachment to companion animals (especially for relatively socially isolated individuals or those with few sources of a sense of value or purpose) may predict well-being but not necessarily in a linear fashion: the relationship between companion animal attachment and health might, for example, follow an inverted U curve, with very low and very high companion animal attachment both associated with poorer health than moderate attachment.35

Another psychological benefit of companion animal ownership, especially for sick, elderly or disabled people, may be the sense of self-worth and purpose generated by caring responsibilities. A widely used model of psychological well-being includes the variables Self-acceptance, Positive relations with others, Autonomy, Environmental Mastery, Purpose in life and Personal growth.36 While all are correlated with life satisfaction and inversely with depression, they show different patterns of association and of change at different age periods. Personal growth and sense of purpose in life are particularly likely to decrease with ageing. We hypothesise that having a companion animal to care for and to provide companionship and stimulation may ward off these adverse changes. The potential role of companion animals to stimulate activity in their owners and reduce depression, by increasing their social contacts and sense of being needed, and to provide attachment figures, give rigorous research in this area a strong rationale.

Positive relationships between companion animal ownership and both physical and psychological health may prove to be mediated by baseline levels of physical activity, social supports and feelings of self-worth. Human social supports can be relatively easily measured using several brief standardised scales. Those with impoverished human social supports and few reasons to feel valued by others could be predicted to show greater mental health benefits from companion animal ownership than those who have adequate social supports, if their attachment to the companion animal is at least moderate.

AAIs for the elderly in residential aged care: one example

Recently some as yet unpublished work has been undertaken by the second author to synthesise the best available evidence on the role of AAIs for the elderly in residential aged care; this has focused exclusively on the use of dogs. As mentioned throughout this paper, the elderly are one such population that has the potential to benefit from human-animal interactions because of declines in physical, social and cognitive ability commonly associated with aging. Reported benefits of AAIs for the elderly include enhanced sensory stimulation, facilitated social interaction, stress reduction, companionship, increased resident-therapist interaction, muscle strength, range of motion and pain management and reduced blood pressure and heart rate.37 Most residents of long-term care facilities do not choose to live in such facilities but are there because they can no longer look after themselves because of their often complex morbidities. Opportunities to interact with animals may enhance the physical, emotional and social health of some individuals because of the interaction between human and animal not needing to be dependent on a high level of cognitive function.38

A systematic review was undertaken to evaluate if such benefits exist for residents of aged care facilities. Only randomised controlled trials were eligible for inclusion into the review and after an exhaustive search of the literature eight studies met the inclusion criteria. Findings were based on methodologically flawed papers with preliminary evidence suggesting that in the majority of outcomes measured, AAAs were beneficial to residents in the short term following implementation; however, they were not superior to control or alternative interventions such as visits from humans or interactions with inanimate objects.

Like the literature available on animal ownership, the methodological quality of studies in this area presents similar challenges in producing solid conclusions. The issue of weak design frequently arose during the search for papers. The majority of literature was anecdotal or descriptive in nature. Half of the studies included in the review were doctoral theses. Although this systematic review limited inclusion to randomised controlled trials, only three of the eight studies adequately described the method of randomisation. An assumption was made for those papers that did not describe the randomisation process that it had adequately been conducted; however, this may not have been the case and lead to selection bias. It was also not clear for the majority of studies whether allocation to treatment groups was concealed from the allocator as most did not clearly identify who the allocator was and the method that was used. Other studies had to be excluded from the review as despite the inclusion of a control group, there was no explanation as to how the control group was constituted.

Some papers did not define who was measuring outcomes and whether they were blinded to treatment allocation. It was obvious in some instances that blinding was not possible, when outcomes such as smile, and eye contact were measured during the intervention period.

The sample sizes were small, ranging from 36 to 80 participants. The length of the interventions (i.e. the interaction with the animal) varied from anywhere between 6 and 30 minutes per session. The durations on the shorter side of the scale in particular seem extremely small to be able to establish any level of attachment and subsequent benefit, suggesting the literature is unclear on the optimum interaction time required.

The follow-up time for measuring outcomes was quite short, varying from 9 days to 14 weeks. It is impossible to determine whether the benefits reported would remain in the longer term. It would seem imperative to conduct research that measured outcomes on a longer scale, at least at 6 and 12 months to determine if the effects were not based on the 'novelty' factor of such an intervention.

The issues surrounding the failure to control for other influences was also prevalent among this literature. Utilising a complex population such as the elderly in long-term care with multiple comorbidities would warrant comprehensive collection of baseline characteristics in order to be able to accurately compare intervention and control groups. Some papers reported basic characteristics such as age and sex. Others were more comprehensive and measured factors such as level of care, past companion animal ownership and time in residence. Many factors that may impact on interactions were not captured in studies such as medication usage, hearing, vision and mobility impairments, and attitudes to animals. Studies that measured across more than one facility did not describe the care and services provided to residents. Differences such as other types of therapies offered or the staffing levels and mix for example could impact on resident outcomes.

One of the prominent limitations found in this area related to the presence and level of interaction of the animal handler or the researcher (which in some cases were one and the same). Studies were unclear in describing whether a person/people (besides the participant) were involved in the intervention. In some cases it was clear that the researcher/handler was present during the interaction, in others it was not clear who was involved. Some of the interventions involved the researcher/handler interacting with the participant freely,39 while others used predeveloped scripts in an attempt to limit the interaction between the researcher/handler and the participants.40 This suggests that the presence/interaction level of the researcher/handler was not adequately controlled for implying that the outcomes produced may have in fact been related to the interaction of the researcher/handler and not the animal. A few studies did however utilise multiple treatment arms to control for this interaction.41,42 A treatment condition (researcher/handler and the dog), a control condition and another treatment arm were some examples used.

Lastly in regards to the dogs used in the interventions, most papers did not describe the characteristics of the animal (e.g. breed, age). Some dogs were allowed to wander freely during the intervention while others were leashed during the entire intervention period. Some papers failed to provide this level of detail. It was not always clear what level of interaction the participants had with the animal. It was noted that staff members owned some of the dogs used for the interventions, with one paper43 providing details that the dog would become distracted during the intervention and wanted to interact with its owner. Controlling for such factors would seem critical to obtaining methodologically sounds results.

In conclusion, the literature base of one particular area of human-animal interaction (i.e. AAIs in residential aged care) has been reviewed. A sample of methodological challenges has been presented. It is probable that these challenges exist across many areas of human-animal research.

Conclusion

This paper shows that sound, empirically based evidence of whether and how human psychological and physical health benefits arise from human-companion animal interactions is needed. There is a strong tendency in the literature to assume that human-companion animal interactions are beneficial, and while this may well be the case in many instances, hard evidence is lacking. Claims in favour of the efficacy of human-animal interactions to improve and promote health are poorly supported by well-controlled research studies, and many claims are founded on anecdotal evidence, descriptive research and qualitative and quantitative data collected in research with weak designs. One consequence of flawed research is that the mechanisms by which humans might benefit are not clearly understood. For example, the characteristics of both the humans and the animals concerned that may predict positive consequences for the humans cannot be stated definitively. We do not know, for instance, whether people of certain age groups, health status, personality or social circumstances are more likely to benefit than others or conversely, whether some people may perhaps suffer adverse consequences from interactions with animals. Further, it is not clear whether ownership and shared living arrangements with the animal are important, or if other kinds of contact with companion animals are sufficient for the human to benefit, such as in the case of brief interventions during animal-assisted interventions and therapies. Explorations of the effects of animal-assisted interventions and therapies are particularly weakened by the confounding of the effects of interaction with the handler or therapist against interactions with the animal. High-quality, rigorous research that addresses the questions of how, why and under what conditions humans benefit or do not benefit from interactions with animals would illuminate our understanding of how to increase well-being in many different healthcare contexts. Clearly there are many methodological challenges to overcome, and these are considerable - otherwise the literature would be characterised by research of a higher quality. However, efforts to improve the evidence base in this area are to be commended and encouraged: an increase in knowledge is highly desirable, so that human-companion animal relationships, interventions and therapies can be promoted where appropriate, and evaluated, for the benefit of the health of the wider community.

References