Laboratory rodent housing often fails to meet rodents' behavioral and physiological needs. We previously found that compared to well-resourced (often called 'enriched') housing, conventional cages increase mortality rates and the morbidity of stress-sensitive experimentally-induced diseases (anxiety, cancer, cardiovascular disease, depression, stroke). This systematic review and meta-analysis updates and re-analyzes this dataset, and supplements it with an author survey (via protocol https://hdl.handle.net/10214/26983), to test the hypothesis that cages meeting more needs are better for rodent health. This hypothesis predicts that providing more types of resources ('enrichments' meeting different needs) will result in dose-dependent health benefits. We also explored whether this relationship was linear (such that each additional resource has equivalent value), or instead reflected diminishing welfare returns (perhaps even reaching a plateau), as a plausible alternative. Updating previous searches (May 24, 2020, updated May 6, 2022, via Ovid, CABI, Web of Science, Proquest, SCOPUS) yielded 1589 further publications. After screening for inclusion criteria (published in English, using mice or rats, and providing resources in long-term housing), this yielded 48 new articles, totaling 233 unique articles in the combined dataset (using 4953 mice, 2611 rats). Each beneficial resource type (additional space, burrowing substrates, chewing/gnawing materials, environmental complexity, foraging opportunities, fresh plant material or its odours, nesting material, shelters, sweet or high fat food, wheels) was given one point if added to well-resourced cages, up to a potential maximum of ten (with well-resourced cages in practice supplying 1–5 additional resources over control conditions). The prediction of dose-dependent benefits was met for disease morbidity: as more resource-types were supplied (compared to controls), health benefits linearly increased (F1,164= 9.12, p = 0.0029) Each additional resource increased the standardized mean difference by 0.11 (0.04–0.19). No such effect occurred for mortality (F1,13 = 0.59, p = 0.4565), but power here was low. Risk of bias (assessed using the Systematic Review Center for Laboratory animal Experimentation 'SYRCLE' tool) in included studies was high; however, overall effects were large and confidence in the analysis was considered high. Providing multiple resources is thus important for rodent health: here, providing up to five additional resource-types (the maximum we could assess) steadily reduced morbidity. However, there was no evidence of diminishing returns (let alone plateau effects) over this range, and so additional resources should be supplied to further improve rodent health and welfare (and perhaps even reach asymptotic levels). This research was funded by NSERC and UFAW.
Read full abstract