Subjective memory complaints and informant-reports of memory decline are considered as important predicting factors of cognitive disorders including mild cognitive impairment and dementia,¹²³⁴ and especially useful in predicting Alzheimer's dementia which accounts for 70% of overall dementia.³⁴

There have been many studies reporting that subjective memory complaints predict conversion to dementia in older population.^34567891011 Also, a study of community-dwelling elderly aged 65 years and older reported that the group with subjective memory complaints showed greater cognitive impairment in neuropsychological assessment compared to the group without subjective memory complaints,¹² suggesting subjective memory complaints might be capable of screening cognitive disorders.

Recent studies support the diagnostic prediction ability of subjective memory complaints with biological evidences.¹³¹⁴¹⁵ Laws et al.¹⁴ reported that elderly people with subject memory complaints showed significantly greater atrophy of hippocampus compared to the ones without subjective memory complaints. Jorm et al.¹⁵ reported that elderly population with e4 allele of APOE gene showed significantly more subjective memory complaints.

However, subjective memory complaints have been considered to be less reliable than informant-reports of memory decline in discriminating cognitive disorders, because it can be influenced by factors such as personality trait, mood and anxiety at the time of the complaints.¹³¹⁴ A number of studies reported that it did not predict cognitive disorders.¹⁶

Meanwhile, numerous studies have been constantly reporting that informant-reports of memory decline discriminates cognitive disorders with clear validity. Recent studies reported that informant-reports of memory decline predicted cognitive disorders, even before clinical diagnosis by standard assessments,¹⁷ and also it detected cognitive disorders even in earlier stage of disease process than MCI.¹⁸ A recent study subjected to community-dwelling elderly population reported that for the group with objective cognitive impairment, informant-reports of memory decline screened cognitive disorders more accurately compared to subjective memory complaints.¹⁹ On the basis of results reported by numerous studies, a variety of informant questionnaires with reliability and validity in screening MCI and dementia have been developed and utilized in clinical practices and researches.^15202122

As described above, subjective memory complaints and informant-reports of memory decline both seem to be capable of screening cognitive disorders. However, few studies compared the screening ability of subjective memory complaints, informant-reports of memory decline, and the combination of the two.

This study aimed to compare the screening accuracy of subjective memory complaints, informant-reports of memory decline and the combination of these two for discriminating cognitive disorders including MCI and dementia in memory clinic setting.

Logistic regression analyses showed that all three models including SMCQ, SIRQD and SMCQ+SIRQD, respectively, significantly discriminated MCI, D and CDall from CN. It is consistent with previous researches in supporting that both subjective memory complaints and informant-report of memory decline not only predict conversion to cognitive disorders of various levels.³¹¹²⁹ but also screen cognitive disorders of various levels.³⁰³¹³²

There have been a few researches exploring combination models of various screening tools to improve screening ability of cognitive disorders. One study explored verbal fluency-supplemented-MMSE model which has better assessment of frontal lobe function, to improve screening ability.³³ Another study explored the usefulness of neurodegenerative quantification using models of various biological markers such as cortical thickness, glucose metabolism, hippocampal volume of brain.³⁴ There were also a study exploring diagnostic predicting ability of models of different combinations of MMSE, brain MRI and PET,³⁵ and a study verifying screening ability of combination models of certain neuroimaging markers.³⁶ Results of these studies proposed that combination of various screening tools and biological markers improved the screening ability for cognitive disorders, which are consistent with the result of this study.³⁴³⁵³⁶ While our study has a consistent result with previous studies, it differs from other studies in that the screening tools included both subjective and objective tools which are SMCQ, self-reporting questionnaire for memory complaints and SIRQD, informant-report of participants' memory decline.

In comparisons of screening ability among the three different models by the differences of -2LL, SMCQ+SIRQD model showed significantly higher screening accuracy compared to SMCQ model in any cognitive disorders, which is consistent with previous studies that reported combination of certain screening tools improved the screening ability of cognitive disorders.³³³⁷ On the other hand, SMCQ+SIRQD model showed significantly higher accuracy only in screening MCI and CDall compared to SIRQD model, while it showed significantly lower screening accuracy for dementia. As for dementia, SIRQD model showed the highest screening accuracy of 94.6%. These results are consistent with previous studies which reported that subjective memory complaints did not show consistent screening ability in all levels of cognitive disorders.³⁸³⁹ The results of ROC curve analysis also showed that while SMCQ+SIRQD had significantly higher accuracy than SMCQ in screening MCI, D and CDall, the combination had no differences compared to SIRQD. These results indicate that the screening accuracy of the combination model was not always superior to the individual tools, as shown in the results of the regression analyses.

Subjective memory complaints have been reported to predict conversion of the cognitively normal without objective cognitive impairment in neuropsychological assessments to MCI or dementia in several longitudinal studies,³⁰⁴⁰ but there was also a cross-sectional study reporting that it had a limited screening ability for cognitive disorders.⁴¹ There are also studies reporting that subjective memory complaints in cognitively normal and MCI groups of elderly, showed statistically significant association with cognitive impairment assessed by neuropsychological batteries, and the higher the cognitive impairment, the higher the level of subjective memory impairment were.^12324243 However, a number of studies reported that in a group who has progressed to dementia, subjective memory complaints did not have any association with objective cognitive impairment, but the tendency of anosognosia, which is deficit of self-awareness of diseases, increases.⁴⁴⁴⁵⁴⁶ Anosognosia in healthy volunteers has been reported to be involved with prefrontal cortex, temporoparietal junction and posterior dorsomedial regions of parietal lobe including the precuneus,⁴⁷⁴⁸ and decline of regional cerebral blood flow in those regions has been reported in early stage of AD dementia.⁴⁹⁵⁰ Anosognosia in dementia group could be considered to be a major cause of the result that SMCQ+SIRQD model did not show significantly higher screening ability for dementia compared to SIRQD.

Therefore, it may not be the best to simply summing the scores of tools assessing subjective memory complaints and informant-reports of cognitive decline in all cognitive disorders to obtain the score for the combination model of the tools, because it does not consider severity of cognitive impairment which could cause discrepancy between the results of assessments. Rather, a detailed screening model which considers the discrepancy between subjective memory complaints and informant-reports of memory decline depending on the level of actual cognitive impairment could be needed.

Limitations of this study are as follows: 1) this study did not control personality traits and mood states which could affect participants' subjective memory complaints and informant-reports of memory decline. Mood states such as depression and anxiety and personality traits could exaggerate subject memory complaints and also allow informants to misinterpret cognitive disorders as personality or mood problems themselves. If these variables were controlled, the reliability of the results could be increased. 2) this study did not subdivide groups of MCI and dementia by severity and types of cognitive decline. If detailed subdivision within each group of cognitive disorder was executed, it could have been possible to explore the predicting ability of screening models depending on the severity or type of cognitive decline. 3) as this is a cross-sectional study, it was able to explore the cross-sectional screening ability of different models, but was not able to explore the ability of the models in predicting conversion to further level of cognitive disorders.

Strengths of this study are not only a large sample size, but also a panel of psychiatrists with expertise in dementia research confirmed diagnosis through comprehensive examination and discussion of all the available raw data, therefore in comparisons of groups results of higher reliability could be derived. Also, this study is unique and new in that it explored a combination model of subjective and objective tools, compared to previous studies which explored only the objective tools such as biological markers, neuropsycological assessment and questionnaires or exams completed by observers.

In conclusion, our findings suggest that each of subjective memory complaints reported by patients themselves, reports of cognitive decline by informants, and the combination of both are all useful for screening of MCI and dementia. The combined information of both subjective memory complaints and informant-reports of memory decline can also improve MCI screening ability of each individual information. However, such combination appears not better than informant-reports alone in regard of dementia screening.