Comment on “Association Between End‐of‐Life Care Practice in Long‐Term Care Facilities and Interprofessional Collaboration With External Facilities: A Questionnaire Survey”

Dear Editor,

1.

We read with great interest the study by Maeda et al. [1], examining collaboration with external facilities and end‐of‐life care practices in long‐term care facilities.

The adjusted logistic regression model entered perceived collaboration with clinics, dental clinics, pharmacies, and home‐visit nursing stations simultaneously, alongside facility‐level covariates, to isolate the independent association between each collaboration type and end‐of‐life care practice. Because all four collaboration scores were measured with the same four‐point instrument and plausibly reflected a shared underlying construct, a facility's general orientation toward external partnerships, mutual adjustment for four correlated exposures in a single model risks redistributing shared variance in ways that can reverse a coefficient's apparent direction [2]. The pharmacy collaboration estimate makes this concrete: the unadjusted odds ratio was 1.51, trending toward the same positive association seen for dental clinics; however, after mutual adjustment, it fell to 0.82, crossing from a positive to an inverse association. A coefficient that changes sign after adjustment for closely related covariates is a recognized signature of collinearity‐driven instability rather than a stable, independently estimated effect, and this pattern deserves direct comment rather than being folded silently into the paper's general call for larger studies [3]. Reporting the pairwise correlations among the four collaboration scores, or presenting a version of the model that adjusts for total collaboration intensity rather than each partner type individually, would allow readers to judge whether the dental clinic finding is a genuinely distinct, isolable effect or whether it is the collaboration type that happened to absorb the largest share of a single underlying construct [4]. The clinic variable follows a comparable, if less dramatic, pattern, moving from an odds ratio near unity unadjusted to a value below one after adjustment, which is consistent with the same shared‐variance explanation rather than four genuinely independent partner‐specific effects.

The paper's concluding recommendation, that strengthening interprofessional collaboration with dental professionals may serve as a practical lever for expanding end‐of‐life care capacity, implicitly assumes that collaboration precedes and enables end‐of‐life care practice. The cross‐sectional design cannot establish this direction, and a clinically plausible alternative runs in the opposite direction [5]. Providing end‐of‐life care within a facility routinely increases the need for palliative oral care and denture management, as residents lose the ability to maintain oral hygiene in the final weeks of life [6]. This would be expected to drive a facility that already practices end‐of‐life care toward closer dental collaboration, rather than the reverse. Under this alternative, dental collaboration is a downstream consequence of end‐of‐life care provision rather than a precondition for it, and the practical implication changes accordingly: rather than building dental partnerships as a strategy to enable end‐of‐life care adoption, facilities already practicing end‐of‐life care may simply need better‐resourced pathways to the dental partnerships that their existing residents already require [7]. Distinguishing these two readings is important for policy, since building a dental liaison infrastructure ahead of end‐of‐life care adoption would be a poor use of resources if the true relationship runs from care practice to collaboration rather than from collaboration to care practice [8]. The finding that group homes for elderly persons with dementia and long‐term care welfare facilities, both settings where residents commonly remain until death, carried the strongest raw association between end‐of‐life care practice and dental collaboration in the subgroup analyses is consistent with either causal direction, but it is at least as compatible with dying‐related oral care needs generating the collaboration as with collaboration enabling the practice, and the study does not weigh these two explanations against one another [9].

Neither concern displaces the paper's central observation that facilities practicing end‐of‐life care report closer collaboration with dental clinics than facilities that do not, an association absent for the other three external partner types and, therefore, unlikely to be a simple artifact of generally better‐resourced facilities reporting more of everything. Presenting the collaboration score correlations to clarify the pharmacy reversal and supplementing the current cross‐sectional data with even a modest longitudinal follow‐up of facilities that newly adopt end‐of‐life care would allow policymakers and facility administrators to judge whether investment in dental liaison arrangements should precede end‐of‐life care rollout or whether it is better positioned as a companion resource introduced alongside it [10]. Given that the underlying survey already captures facility‐level details across four collaboration domains, a follow‐up wave tracking the same facilities would be a comparatively modest extension of the existing infrastructure and would directly resolve the causal ambiguity that this analysis cannot settle on its own.

Author Contributions

S.M.: conceptualization, literature review, critical analysis, manuscript drafting, writing—review and editing, and final approval of the manuscript. T.C.R.: literature review, validation, critical revision, and writing—review and editing. P.P.: conceptualization, supervision, manuscript drafting, writing—original draft, correspondence, project administration, critical revision, and final approval of the manuscript. R.B.: validation, manuscript review, critical revision, and final editing. All authors reviewed and approved the final manuscript and agree to be accountable for all aspects of the work.

Funding

The authors have nothing to report.

Ethics Statement

The authors have nothing to report.

Consent

All authors have read and approved the final manuscript and consent to its publication.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.