Defining “disaster‐related death” for valid inference: An administrative category in Japan

To the Editor,

We read with interest the descriptive analysis of certified disaster‐related deaths following the 2024 Noto Peninsula earthquake (Kida & Ishikawa, 2025). The tabulation of ICD‐10 categories (Table 2, p. 5) and the co‐occurrence network of narrative descriptors (Figure 2, p. 6) provide useful information for practice and policy.

For readers outside Japan, we should clarify one point of terminology that bears directly on inference. In Japan, disaster‐related death is not a clinical diagnosis or a sociological construct; it is an administrative certification issued by municipalities after a family application and committee review, primarily within the condolence‐grant system. As Kida and Ishikawa (2025) notes, stating this administrative basis at first mention may help align expectations about what such data can demonstrate.

Recent nationwide evidence underscores this context. A cross‐sectional study of 755 certified indirect disaster‐related deaths after the 2011 Great East Japan Earthquake found that only 9.8% of death certificates and 19.2% of medical records explicitly mentioned disaster relevance, while 71.0% relied on bereaved‐family self‐reports. Tsuboi et al. (2025) describes these deaths as certified and recorded by municipalities at family request for condolence grants. These features indicate that the label primarily reflects a legal‐administrative judgment of indirectness rather than a standardized biomedical attribution of causality.

Three issues deserve attention:

Case ascertainment reflects procedures as well as health status. The analyzed population comprises cases that underwent application, certification, and consent for public release. As acknowledged in the Methods and Limitations, fatalities not applied for are excluded; under‐ascertainment of indirectly disaster‐associated deaths is therefore possible.

Administrative certification is not a causal estimate. Pairing ICD‐10 “direct causes of death” (Table 2) with narrative descriptors is informative descriptively; yet without a defined at‐risk denominator, counterfactual comparisons, or standardized medical documentation of “disaster relevance,” causal effects or effect sizes cannot be inferred from these data alone.

Finally, the text network should be interpreted cautiously: it visualizes word co‐appearance rather than causal mechanisms. The co‐occurrence map summarizes word co‐appearance in public narratives under a Jaccard‐based approach and does not encode etiologic direction or magnitude; an explicit note in figure captions could help avoid mechanistic interpretations.

This aligns with broader findings in the literature on classification and interpretation of disaster‐attributed mortality (Combs et al., 1999; Hayakawa, 2016). These points do not detract from the article's practical insights—for example, the prominence of circulatory and respiratory causes among certified cases—but may help readers frame these findings as signals for service planning rather than as established causal pathways. That perspective aligns with work positioning disaster related death as an indicator of medium‐ to long‐term impact with caveats for causal interpretation (Yamamura et al., 2024).

To support valid inference while preserving the descriptive value of the study, we offer concise, neutral wording adjustments that remain close to the design:

Definition at first mention:

Describe disaster‐related death as a municipal administrative certification used in the condolence‐grant system, distinct from clinical cause of death or notions such as “social death.”

Event terminology:

Consider “reported events temporally associated with certified deaths” in place of “causal events.”

Scope of inference:

Note that the analysis describes associations within certified cases and does not estimate causal effects or population‐level risks.

Figure/Results captions:

Clarify that the network shows textual co‐occurrence (no directionality), and that ICD‐10 codes denote immediate/official causes of death distinct from the administrative certification.

We appreciate the careful assembly of municipal public data and view the study as a constructive contribution to disaster nursing. Clear separation among legal certification, clinical cause of death, and hypothesized pathways should facilitate future linkage studies and comparative designs that can quantify risks and mechanisms more directly.

FUNDING INFORMATION

This work was supported by Health Labour Sciences Research, Grant Number 24LA2001.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.