Methodological Considerations in Interpreting Longitudinal Cognitive and Neural Changes Among Older Adults Who Continue Working

To the Editor,

1.

We read with great interest the recent article by Kimura et al. [1], which examined 3‐year changes in cognitive and brain function among community‐dwelling older adults who remained in the workforce. The integration of event‐related potentials (ERPs) with behavioural measures is a valuable methodological step, and the No‐go P3 findings may carry meaningful neurophysiological implications. Nonetheless, several methodological considerations warrant closer attention before these findings are interpreted as strong evidence of cognitive benefit associated with continued employment.

First, the authors administered the same version of the Montreal Cognitive Assessment (MoCA‐J) at both baseline and follow‐up. The observed mean improvement of approximately 1.4 points may, at least in part, reflect practice effects rather than genuine cognitive gains, as such effects are well documented between repeated MoCA administrations, particularly between the first and second exposures to the test [2]. Practice‐related gains of this kind have been shown to account for a meaningful portion of short‐term score change independent of any true improvement in cognitive function, which raises the possibility that the observed change in this study partly reflects test familiarity rather than a genuine cognitive gain. The authors acknowledge this limitation but nonetheless present the MoCA findings as evidence of cognitive improvement. Beyond recommending the use of alternative parallel forms or reliable change index (RCI) corrections in future work, it would strengthen the current findings if the authors were to report RCI‐corrected MoCA‐J change scores using normative practice‐effect data, where such calculations are feasible from their existing dataset. This would allow readers to judge more directly whether the observed change exceeds what would be expected from practice alone.

Second, the single‐arm design introduces a risk of healthy worker survivor bias [3]. Of the original 46 participants, only 23 completed the 3‐year follow‐up, and all remaining participants were still employed. The 23 participants who left the study or their jobs were excluded without a reported comparison of their longitudinal trajectories. It is possible that some participants left the workforce because of declining cognitive function. A clear comparison of baseline characteristics, such as MoCA scores or demographic variables, between completers and non‐completers would help clarify whether attrition was related to baseline cognitive status, and we would encourage the authors to report this comparison explicitly if it has not already been presented. In the absence of such a comparison, together with the lack of a non‐working control group and the differential dropout observed, the findings likely reflect a selected group of cognitively stronger individuals. This limits the strength of any causal or descriptive conclusions regarding the protective role of continued employment.

Third, the authors interpret the significant increase in No‐go P3 amplitude as reflecting improved neural recruitment for inhibitory control. We would suggest this interpretation deserves further consideration rather than being viewed as the only plausible explanation. Research in ageing neuroscience indicates that increased frontal P3 amplitude in older adults does not necessarily imply improved cognitive processing; it may also reflect neural inefficiency or compensatory over‐recruitment, as proposed by the Compensation‐Related Utilization of Neural Circuits Hypothesis (CRUNCH) [4]. In typical ageing, P300 amplitude tends to decrease and latency tends to extend [5]. The stable latency alongside increased amplitude observed in this study is compatible with either enhanced processing or a compensatory neural strategy. Behavioural performance measures, such as reaction time and accuracy on the No‐go task, would be valuable in distinguishing between these possibilities: an improvement in behavioural performance alongside the increased amplitude would argue against a purely compensatory interpretation and lend support to the authors' original conclusion, whereas unchanged or declining performance would make a compensatory account more plausible. We would encourage the authors to discuss their behavioural performance findings explicitly in this context, as doing so would considerably strengthen the interpretation of the ERP results.

Fourth, the sample included 21 men and only two women (mean age: 74.74 years). This marked sex imbalance limits the statistical power available for any sex‐based subgroup comparison and means the reported findings are most directly applicable to older men; conclusions regarding older women cannot be drawn with confidence from this dataset. Separately, the occupations examined (reception, cleaning, and facility management) represent a relatively narrow occupational profile in terms of cognitive demands, which may further limit the generalizability of the findings to older working adults in more cognitively demanding roles. We raise these as two distinct limitations: one concerning statistical power for sex‐based analysis, and the other concerning the broader generalizability of the occupational sample.

In conclusion, the study by Kimura et al. offers valuable insights by incorporating neurophysiological measures within a longitudinal design focused on occupational ageing. However, the potential influence of uncontrolled practice effects on the MoCA‐J, the possibility of differential dropout related to healthy worker survivor bias, the unresolved neurophysiological interpretation of the increased P3 amplitude, and the limited sample size and occupational diversity together warrant caution in interpreting the findings as conclusive evidence of the cognitive benefits of continued employment. We encourage readers to view these findings as preliminary and descriptive, and we hope future studies will incorporate alternative cognitive assessment methods or RCI corrections, an appropriate non‐working control group, larger and more balanced samples, and explicit analyses accounting for dropout.

Funding

The authors have nothing to report.

Ethics Statement

The authors have nothing to report.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

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