Response by Kramer et al to Letters Regarding Article, “Dog Ownership and Survival: A Systematic Review and Meta-Analysis”
In Response:
We thank Simonato et al and Owen et al for their interest in our recent meta-analysis of observational studies evaluating the association of dog ownership with all-cause and cardiovascular mortality.1 This meta-analysis involved 10 studies and included data from 3 837 005 participants across 8 countries. Our results demonstrated that owning a dog was associated with 24% reduction in all-cause mortality and 31% reduction in cardiovascular mortality as compared to individuals who did not have a dog. In this analysis, the positive impact of dog ownership on all-cause mortality was evident with both random-effect (RR, 0.76; 95% CI, 0.67–0.86) or fixed-effect modelling (RR, 0.72; 95% CI, 0.72–0.73). We fully concur with Simonato and Owen that the observed association may be partially explained by confounders such as younger age, better physical fitness, and higher socioeconomic status in the population who own a dog as compared to non-dog owners. As discussed in our article, the results were based on data obtained from observational studies given the lack of randomized trials evaluating dog ownership. In this context, the inherent limitation of such data with potential impact of residual confounders cannot be dismissed.
In an attempt to address the potential impact of confounders on our findings, we have conducted a sensitivity analysis in which the meta-analysis was repeated, including only the studies that reported adjusted hazard ratios (HRs), using the fully adjusted estimates reported in each study. This analysis revealed that the findings remained unchanged. Including the 5 studies that reported adjusted estimates, there was a 17% reduction in all-cause mortality (HR, 0.83 [95% CI, 0.82–0.84]; Figure [A]) and 20% reduction in cardiovascular mortality (HR, 0.80 [95% CI, 0.76–0.83]; Figure [B]) in participants who own a dog as compared to non-dog owners. Notably, given the smaller number of studies in this sensitivity analyses, the largest study which included >3 million participants2 had the main impact on the overall estimate.
Figure. Meta-analysis of the association between dog ownership and the adjusted hazard ratio for (A) all-cause mortality and (B) cardiovascular mortality. Isquared is a measure of heterogeneity between studies, where I-squared >50% indicates significant heterogeneity. HR indicates hazard ratio; and CI, confidence interval.
The results of our adjusted meta-analyses may be contradictory to those of other adjusted analyses due to methodologic discrepancies. Given that our meta-analysis of adjusted HR included few studies (5 studies as opposed to 9 studies in our original analyses), the use of fixed-effect model (instead of random-effect) is recommended to avoid a “small-study effect”. This potential bias is relevant for a meta-analysis in which high between-study heterogeneity is observed in the context of few studies representing large differences in the size of the study population. Specifically, in such as setting, the comparatively equal weighting of small and large studies in a random-effects analysis can result in inappropriate weighting of the small studies, as compared to an analysis using fixed-effects3,4. This aspect is particularly relevant to the adjusted meta-analysis of observational studies evaluating dog ownership as one study included >3 million participants, two studies had more than 50 000 participants, and two studies had less than 10 000 participants. If random-effect model is used for such analysis, a non-significant impact of dog ownership for all-cause mortality is observed because the weight of the largest study (n=3 432 144 participants) for the overall estimate would be 23.9%, which is close to the 17.3% weight of the smaller study (n=4039 participants) (HR, 0.96 [95% CI, 0.82–1.09]). This scenario is a typical example of a biased meta-analysis demonstrating a small-study effect wherein small studies have contributed to the final result with similar weighting to that of much larger studies, thereby obscuring the most reliable overall estimate3,4. Thus, as recommended by Cochrane Collaboration and statistical literature3,4. our adjusted analyses of 5 studies using fixed-effect model confirmed a positive impact of dog ownership in reducing all-cause mortality.
Adjusted sensitivity analysis including only participants with previous cardiovascular disease was not possible. However, in a recent study by Mubanga et al,5 dog owners had a lower risk of death after a hospitalization for acute myocardial infarction in models adjusted for age, sex, socioeconomic status, and comorbidities. A positive effect was observed for individuals who live alone (HR, 0.67 [95% CI, 0.61–0.75]) and for those who lived with a partner or a child (HR, 0.85 [95% CI, 0.80–0.90]). Similarly, Friedmann et al6 demonstrated that owning a dog was a significant predictor of survival 1 year following an acute myocardial infarction in an analysis matched for not only participant’s age and gender but also prevalence of diabetes mellitus and left ventricular ejection fraction.
The results of our adjusted analyses coupled with previous studies demonstrating positive impact of dog ownership on cardiovascular risk factors, reinforces the concept that dog ownership may be associated with a positive impact on human health, an effect that holds biological rationale. Specifically, dog ownership has been shown to increase owner’s physical activity, lower blood pressure, reduce physiological indices in response to stress, and improve mental health among others positive effects beyond the cardiovascular system. In addition, the existence of a special human-dog bond was demonstrated in an experiment published in Science7 in which an oxytocin-gaze positive loop was observed between dogs and their owners that was similar to the hormonal response observed when mothers interact with their infants.
In conclusion, while we recognize that the current evidence consisting of mostly observational data cannot establish any causal association between dog ownership and survival, we disagree that dog ownership is solely a marker of increased physical activity. Given the growing body of evidence, we speculate that such positive interspecies interaction promotes greater adherence to a healthy lifestyle, improves cardiophysiological indices, improves overall well-being, and promotes the activation of a hormonal loop associated with positive emotions that are likely beneficial to human health.
Disclosures
Dr Kramer reports grants from Boehringer Ingelheim, outside the submitted work, and is a proud dog owner of Romeo Kramer. The other authors report no conflicts.
Footnotes
References
- 1.
Kramer CK, Mehmood S, Suen RS . Dog ownership and survival: a systematic review and meta-analysis.Circ Cardiovasc Qual Outcomes. 2019; 12:e005554. doi: 10.1161/CIRCOUTCOMES.119.005554LinkGoogle Scholar - 2.
Mubanga M, Byberg L, Nowak C, Egenvall A, Magnusson PK, Ingelsson E, Fall T . Dog ownership and the risk of cardiovascular disease and death – a nationwide cohort study.Sci Rep. 2017; 7:15821. doi: 10.1038/s41598-017-16118-6CrossrefMedlineGoogle Scholar - 3.
Higgins JPT, Green S (editors) . Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0. The Cochrane Collaboration, updated 2011. www.handbook.cochrane.org.Comparing fixed and random effects estimates: https://handbook-5-.cochrane.org/chapter_10/10_4_4_1_comparing_fixed_and_random_effects_estimates.htmGoogle Scholar - 4.
Lin E, Tong T, Chen Y, Wang Y . Fixed-effects model: the most convincing model for meta-analysis with few studies.ArXiv. https://arxiv.org/abs/2002.04211v1Google Scholar - 5.
Mubanga M, Byberg L, Egenvall A, Ingelsson E, Fall T . Dog ownership and survival after a major cardiovascular event: a Register-Based Prospective Study.Circ Cardiovasc Qual Outcomes. 2019; 12:e005342. doi: 10.1161/CIRCOUTCOMES.118.005342LinkGoogle Scholar - 6.
Friedmann E, Thomas SA . Pet ownership, social support, and one-year survival after acute myocardial infarction in the Cardiac Arrhythmia Suppression Trial (CAST).Am J Cardiol. 1995; 76:1213–1217. doi: 10.1016/s0002-9149(99)80343-9CrossrefMedlineGoogle Scholar - 7.
Nagasawa M, Mitsui S, En S, Ohtani N, Ohta M, Sakuma Y, Onaka T, Mogi K, Kikusui T . Social evolution. Oxytocin-gaze positive loop and the coevolution of human-dog bonds.Science. 2015; 348:333–336. doi: 10.1126/science.1261022CrossrefMedlineGoogle Scholar