We aimed to investigate stable vs. unstable patterns of subjective cognitive decline (SCD) in regard to progression to mild cognitive impairment (MCI) and dementia in a population-based sample of cognitively unimpaired elderly. Stable SCD, i.e., repeated reports of SCD, yielded an increased risk of progression to MCI and dementia compared to unstable SCD over an average observation period of 4.8 years.

Risk of progression seemed to be driven in part by baseline cognitive functioning in the stable SCD group, but not in the unstable SCD group.

There is a lack of studies that investigated stability of SCD in relation to MCI and dementia progression. One study, based on a memory clinic sample of 230 cognitively normal individuals with SCD over 50 years of age, distinguished three outcomes after eight years, namely non-decline, decline and diagnostically unstable (i.e., a change of diagnosis over time from normal to MCI, then back to normal) [22]. The authors concluded that the presence of SCD was a predictor of future cognitive decline, but it also increased the likelihood of an unstable diagnosis. Such an unstable diagnosis was associated with affective symptoms, particularly anxiety. The authors noted that the associations found in previous studies on cognitive decline and affective symptoms might be explained at least in part by including individuals with an unstable diagnosis who might later revert to normal. Besides, the possibility of reversion to “normal”, i.e., the perception of sufficient memory function with no more experience of SCD, seems much neglected in research as the reversion to normal cognition in MCI [23]. This may suggest, in line with our results, that it might be useful to distinguish between stable and unstable – including reversing – courses of SCD as they might relate to different aetiologies.

That stable SCD might be associated with an increased progression risk to MCI and dementia makes sense in the view of the fact that AD, the most frequent type of dementia, is characterized by a long continuous process of slow neurodegeneration into marked cognitive decline at advanced stages. It seems reasonable to assume that SCD, if a syndrome of preclinical AD, should be constantly experienced over a longer time period in this ongoing process rather than occasionally. Interestingly, this is underlined by our result that an unstable pattern of SCD, by contrast, was not predictive of MCI and dementia. In fact, individuals who did not consistently but only occasionally report SCD throughout the assessments displayed a decreased risk of MCI and dementia compared to individuals with stable SCD and those who never reported SCD. Potentially, this might be explained by temporary clinical conditions out of which SCD might result, e.g., due to a depressive episode, pain, fatigue, medications, or distressing life events [24]. As soon as such temporary conditions resolve, cognitive functioning may not be experienced as impaired anymore.

Otherwise, it has been reported previously that individuals in general struggle with differentiating between a normal and a pathological process of ageing [6] which could also be a reason for inconsistent reports of SCD. From a functional perspective, occasional reports of SCD might also be explained as a “distorted awareness of a present cognitive state compared to a past one” resulting in complaints of inexistent cognitive decline [6]. Individuals may over-estimate subtle changes in cognitive functioning that may only reflect age-related normative cognitive decline without becoming overt dementia later [25]. Over-estimation of cognitive changes could be driven by a fear of anticipatory dementia [26]. Such a fear is also associated with symptom-seeking for the disease [27].

By contrast, if SCD is completely absent, it could reflect either that there are indeed no memory problems or it could indicate poor awareness, i.e., the inability to accurately appraise aspects of cognitive functioning, also referred to as anosognosia [25]. If some of the individuals who did not express SCD had a poor awareness of cognitive changes, and thus, failed to express SCD, then this could explain why individuals who never reported SCD had an intermediate level of progression risk between the stable and unstable SCD group in our study. As there is evidence for variability in the level of awareness in MCI, it is likely to assume different levels of awareness in SCD where changes in cognitive functioning may be more subtle and, therefore, may be even more difficult to evaluate accurately [28]. However, whether differences in awareness are associated with progression risk to MCI and dementia is not known [25, 28].

As hypothesized, individuals with stable SCD exhibited the highest progression risk to MCI and dementia compared to individuals with unstable SCD and no SCD. When adjusted for cognitive functioning (MMSE scores), however, progression risk in individuals with stable SCD was only significantly increased in comparison to individuals with unstable SCD, but not compared to individuals without SCD. We assume, that the averagely slightly lower MMSE scores in individuals with stable SCD could rather reflect early pre-clinical cognitive decline as opposed to in individuals with unstable SCD. It has been previously reported, e.g., by St John and Montgomery [29], that individuals with SCD (regardless of its stability) experience “real” cognitive losses, “which is also apparent as lower [cognitive] scores”. Reisberg et al. [9] added that a lower MMSE score in individuals with SCD is not simply statistically significant, but it may be “real” in that individuals with SCD experience pre-clinical cognitive losses. Our results additionally suggest that this may particularly apply to individuals who have a stable experience of SCD as opposed to an unstable experience of SCD.

As the need for a refinement of the research criteria of SCD has been pointed out [3, 7], our preliminary results suggest that stability of SCD might be a feature that could contribute to better identify individuals at risk for MCI and dementia.

Besides, it has been reported previously that SCD predicts a shorter time to progression compared to individuals without SCD [9, 30]. In our study, it was particularly stable SCD that was associated with a significantly shorter time to progression (median: 6.2 years). By contrast, unstable SCD revealed even a slightly longer progression-free survival time (median: 7.4 years) than controls (median: 7.2 years). However, we cannot exclude that individuals with stable SCD might potentially be more advanced in the pre-clinical phase of cognitive deterioration, thus, having a shorter time until MCI or dementia become overt. SCD might be experienced stable only after some subtle cognitive changes have occurred, perhaps allowing the chance to experience unstable SCD before such changes.

We have to consider some limitations. First, SCD was only measured by asking a simple question concerning memory problems. Even though the question “Do you have problems with your memory?” may be a valid measure for global memory functioning [31], there is a lack of psychometric data concerning its longitudinal use. Future investigations of the stability of SCD with more comprehensive assessments (e.g., assessing other domains but memory or asking for related worries) could shed more light on the potential link to MCI and dementia progression risk.

Second, current research criteria on SCD [3] also include biomarker abnormalities consistent with the AD pathology. In our study, we were not able to evaluate biomarkers in relation to patterns of SCD progression – future studies in this regard might be useful. Third, we otherwise applied those SCD research criteria which lead to the exclusion of a substantial proportion of the LEILA75+ participants. It is questionable whether such strict criteria that exclude major psychiatric, neurological or medical disorders reflect the actual at risk population for progression to MCI and dementia, especially as older adults can often simultaneously have such comorbidity.

Finally, the exclusion of such a substantial part of the cohort may also limit the generalizability of our results, even though the LEILA75+ study does have a population-based design. On the other hand, to the best of our knowledge, we are the first to present results on SCD patterns in community-dwelling elders.

Our present results, though preliminary, suggest an increased risk of progression to MCI and dementia in individuals with stable SCD, i.e., consistent repeated SCD reports, compared to unstable SCD. Baseline cognitive scores, though within a normal range, seem to be a driver of progression in stable SCD. This may have implications for both research and clinical practice. In clinical practice, special attention should be put on repeated reports of SCD. Concerning research, further studies would be useful to establish whether stability, e.g., in form of a time criterion, could hold as an additional SCD research feature.