The study When Dogs Heal utilized an open cross-sectional Web-based survey to collect data among adult PLHIV. To recruit participants, messages were posted to social media platforms (eg, Facebook and Twitter) by the principal investigator, who posted about the study one time on each platform. Participants were incentivized to participate by offering a chance to win an Apple iPad. Eligibility criteria included was being ≥18 years old and self-reporting an HIV diagnosis; dog ownership was not a requirement to enroll. Interested individuals were directed to a Web link with the study screener, consent statement, and survey. The survey was thoroughly tested by study staff before data collection. Upon clicking the link, volunteers were instructed to complete a series of screening questions and, if eligible, proceeded through the consent process before completing the study questionnaire. The survey questionnaire was deployed using Qualtrics, a Web-based data collection software tool based out of Seattle, WA and Provo, UT. Qualtrics utilizes transport layer security (TLS) encryption for all transmitted data and services are hosted by secure data centers. Participants were able to review and change their previous answers through a back button and were typically shown 1 question per page. Browser cookies and Internal Protocol (IP) address were used to prevent duplicate responses from the same individual. This study was approved by the institutional review board at the study site.

Participants first completed a brief set of demographic questions indicating their primary race (Asian/Pacific Islander, black/African American, white, Latino/Hispanic, other, or multiracial), ethnicity (Spanish/Hispanic/Latino/a), sex assigned at birth (male or female), current gender identity (male, female, or transgender), age, and whether or not they ever owned a pet (yes or no). Participants who reported having ever owned a dog were asked whether they currently owned a dog.

Depression was measured using the shortened Center for Epidemiologic Studies Depression Scale (CES-D10) [58,59].This 10-item self-report measure of depressive symptoms in the prior week uses a 4-point frequency response scale. This measure has demonstrated excellent reliability and validity in prior research [58,60], as well as in this study (alpha=.90). The CES-D10 was scored by summing the 10 items, with higher scores indicating increased depressive symptoms. Per scoring criteria, a cut-off score of 10 or greater was used to determine whether the participant met the criteria for depression.

Resiliency, a potential confounding variable, was measured with the Resilience Research Centre Adult Resilience Measure (RRC-ARM), an adapted version of the Child and Youth Resilience Measure (CYRM-28) [61,62]. The 28-item instrument measures the resources available to individuals that may boost their ability to sustain their well-being through individual resources or capabilities, relationship with primary caregivers, and contextual factors that facilitate a sense of belonging [62]. Cronbach alpha for the 3 subscales of the CYRM-28 range from .794 to .833 and subscale correlations range from .555 to .705 [63]. This sample yielded an alpha of .90. An overall resiliency score was created by averaging across all items to achieve a score 1-5, where higher scores indicate higher levels of resilience.

We used descriptive statistics, including measures of central tendency and dispersion, to characterize the sample. We assessed simple associations between variables using Pearson correlations or chi-square tests, with significance level set at alpha <.05. A multivariable logistic regression analysis assessed the log odds that a participant would meet criteria for depression based on current dog ownership, controlling for age, race, ethnicity, and gender, as well as resilience. Due to low cell numbers, the race variable was dichotomized for analysis (white vs nonwhite). Data were analyzed using SPSS statistics version 24 (IBM Corporation).

In total, 356 data cases were downloaded from Qualtrics, of which 29 cases had a duplicate IP address. The data cases that were most complete were kept in the final dataset, leaving 327 unique volunteers who clicked the survey link; 15 cases that contained no data were deleted (ie, a volunteer clicked the link but did not proceed past the first screen). A total of 28 cases were deleted for not meeting eligibility criteria. Of the participants who were eligible, 32 did not enroll in the study.

A total of 252 adults living with HIV were enrolled into the study in January 2016, yielding a participation rate of 77.1% (252/327). Completion of the study survey took approximately 15 min on average. A total of 51 cases were withdrawn from analysis for largely incomplete responses. The completion rate for the survey was 79.8% (201/252). An additional 2 participants who reported transgender status were removed from the sample because of small cell size for analysis, that is, to minimize the number of variables in analytical models. The final analytic sample included 199 PLHIV.

Table 1 summarizes the characteristics of the sample. The mean age was 48.72 years (standard deviation, SD 10.68; range 21-73) with the mean age at HIV diagnosis of 31.89 years (SD 9.43; range 15-67). The average number of years since HIV diagnosis was 16.87 (SD 10.05; range 0-38). Most participants were male (86.4%, 172/199), white (80.4%, 160/199), not Hispanic/Spanish/Latino/a (91.0%, 181/199), and homosexual or gay (82.9%, 165/199). The vast majority of the sample reported lifetime pet ownership (97.5%, 194/199), with the most commonly reported pet being a dog (92.5%, 184/199). Of the participants who reported ever having owned a dog, 68.9% (126/183) reported they got their dog after their HIV diagnosis, and 91.3% (168/184) indicated it was a household pet as compared with a therapy dog or official service dog. Current dog ownership was prevalent among the sample (68.3%, 136/199).

Chi-square tests (with Fisher exact test as needed for low cell sizes) and Pearson correlations indicated that there was no significant relationship between depression and demographic characteristics (N=199): age (r=−.08, P=.28), race (χ²₁=1.8, P=.21), ethnicity (χ²₁<0.1, P>.99), gender (χ²₁=0.5, P=.54), and sexual orientation (χ²₁=0.6, P=.51). Resilience was moderately and negatively correlated with depression (r=−0.31, P<.001) (data not shown).

Table 2 presents the results of the multivariable logistic regression model. Although not significant in bivariate analysis, we retained age, race, ethnicity, and gender in the model for substantive reasons. The logistic regression model was statistically significant, χ²₆=37.206, P<.001. Of the 6 predictor variables, only 2 were statistically significant: dog ownership and resilience. Noncurrent dog owners had 3 times higher odds of depression in comparison with current dog owners (OR 3.01, 95% CI 1.54-6.21), controlling for demographic factors and resilience.

This study explored the impact of current dog ownership on depression for a sample of adult PLHIV. It was hypothesized that among PLHIV, current dog ownership would be associated with decreased likelihood of meeting depressive criteria. Our hypothesis was supported in that noncurrent dog ownership was significantly and positively associated with depression, controlling for the potential influence of resilience and other demographic factors. Individuals who were not current dog owners had 3 times higher odds of depression compared with current dog owners. This finding adds to and updates prior findings regarding the potential positive impact of dog ownership on depression in PLHIV.

This is a notable finding as depression is the most common psychiatric disorder associated with HIV disease [64]. Depression influences not only the psychological health of PLHIV but is a correlate of overall HIV disease progression as well. Support for or promotion of dog adoption and ownership may be a novel intervention to positively impact depression and, in turn, positively affect other HIV-related health outcomes. The growing adoption of dogs for diabetes-specific intervention and related research will inform feasibility and potential efficacy for PLHIV. The high satisfaction reported by dog owners in DAD programs provides preliminary evidence that full-time dog adoption may be feasible for other populations with chronic illnesses, such as PLHIV. Although research to date suggests benefits of dog ownership among PLHIV, longitudinal and intervention research is needed to establish whether or not dog adoption is effective in reducing psychological symptoms and promoting overall health.

Although little research has been done on the potential mechanisms of action, dog ownership may impact depression and overall psychological health among PLHIV through bolstering social support networks. Pet dogs may act as a source of social support that can supplement existing support networks or fill network voids for vulnerable populations [24,30,35-39,44,52,65]. Allen and colleagues found evidence that pets, specifically dogs, act as sources of nonevaluative social support when faced with a stressor. This nonevaluative social support may facilitate better and faster physiological responses to mental stress over and above spousal and friend support [24,35,36]. The numerous positive associations of social support for PLHIV, including reduced depression, have been well documented [66,67]. Dog ownership may provide a unique type of social support for PLHIV that positively impacts depression over and above their existing social support networks.

Beyond ameliorating depressive symptoms, dog ownership may also have an effect on other important factors related to health outcomes for PLHIV, including ART adherence. As noted earlier, depression is consistently associated with suboptimal ART adherence. Inadequate social support is also routinely found to be correlated with poor ART adherence [68,69]. One additional mechanism through which dog ownership may influence ART adherence is through the routinization associated with pet care, which reflects the extent to which one’s life is organized and follows a predictable pattern [52-54,70]. Routinization has been found to be an important factor in facilitating ART adherence [71-75]. Preliminary evidence suggests that pets may facilitate the adoption of daily routines into which ART regimens can be incorporated [52-56]. Daily and continuous care for the dog through walking and feeding can provide a source of social interaction and give PLHIV a sense of purpose while establishing a healthy routine for themselves. One qualitative study on routinization found that even when participants were experiencing homelessness, having at least one recurring daily activity was associated with >70% adherence to ART; among the daily activities discussed was walking their dog [56]. Although ART adherence and routinization were not measured in this study, future studies should explore these factors, in addition to social support, as potential mediators of dog ownership on health outcomes for PLHIV.

Several limitations of this study should be noted. The data are cross-sectional in nature and therefore causality cannot be inferred. For example, we measured depressive symptoms over the past 7 days, meaning we do not have data regarding onset of depression or whether depressive symptoms preceded or followed dog companionship. Additionally, only participants with access to the Internet were able to enroll. Although the vast majority of Americans report having access to the Internet, Internet adoption is still closely dependent on income and education [76]. This sample comprised mostly middle-aged, nonheterosexual white men who had been infected for more than 10 years (on average), which is not reflective of the current trends in HIV incidence among younger racial and ethnic minorities. Therefore, findings may not generalize to younger and recently infected PLHIV.

In conclusion, growing evidence suggests that dog ownership reduces the likelihood of depression and, therefore, may confer long-term health benefits on PLHIV. Given this evidence, testing of dog-specific AAIs to reduce depressive symptoms as well as to improve HIV-related health outcomes is a logical next step. Given the relatively low rate of viral suppression in PLHIV in comparison with national targets, innovative interventions are needed to reach goals established under the national HIV/AIDS strategy.