Our results provide strong evidence supporting the relevance of self-reported quality of life assessments for patients with cognitive disorders, particularly in patients with severe cognitive dysfunction. It seems that the nature of multiple sclerosis and schizophrenia did not affect this type of assessment.
We examined these two diseases on the basis of the following points: i) the status of chronic illnesses with a high occurrence of reported cognitive deficits, even during the early disease stages [41, 42]; ii) three main composites of cognition were indiscriminately affected [42, 43]; iii) the homogeneous and extensive assessment of cognition, including tests assessing memory, attention, and executive function; iv) the availability of a disease-specific self-reported QoL questionnaire [30, 32]; and v) the surprising similarities related to changes in white matter structure or abnormalities in myelin [24, 25]. Moreover, some studies have suggested that changes in the integrity of white matter can result in impaired cognitive function in MS  and SCZ  patients.
These findings may support for the use of QoL assessment for clinicians who are still perplexed when interpreting the meaning of QoL scores for an individual with cognitive impairment. This present study suggests that cognitively impaired patients, as defined by a global cognitive dysfunction, can reliably and consistently respond to a specific QoL self-reported questionnaire. This assumption is underlined by the suitability indices found in the highly-impaired groups, i.e., 2 or 3 altered functions altered, in both MS and SCZ patients. These indices may be considered satisfactory compared to their respective reference populations. In the highly-impaired groups, factor analysis showed that the structure corresponded with the initial structure of the QoL questionnaires: 8 of the 9 dimensions were clearly identified in the MusiQoL and all the dimensions were identified in the SQoL. Although the IIC values reported in the highly-impaired population of MS individuals were similar to those identified in the reference population, the proportion of dimensions with IIC that exceeded 0.2 compared to the reference population was less satisfactory in SCZ individuals. For MS and SCZ populations, the proportion of dimensions with IDV values greater than the IIC values and the proportion of dimensions with IDV exceeding 0.2 compared to the reference population were less satisfactory, which may be explained by the very restricted definition of the decision rule. Internal consistency coefficients calculated in the highly-impaired groups were close to those of the initial reference populations. The floor and ceiling effects were slightly different compared to those reported in the initial validation publication independent of disease type. In addition, satisfactory INFIT statistics supported the unidimensionality of each of the dimensions.
Regarding external validity, highly-impaired populations showed satisfactory suitability indices. The links between QoL scores and age, severity disease score (EDSS and PANSS), and disease duration were closer to the initial reference populations independent of cognitive status and disease. However, links between QoL scores and gender and educational level were less satisfactory. In summary, the suitability indices of the highly-impaired population may be considered completely acceptable considering the small sample size of the defined populations.
Several previous studies have employed similar approaches to define cognitive dysfunction using a single composite, such as memory [7, 15], attention , and executive functions [17, 46]. It should be acknowledged that a single test of cognitive functioning would never be entirely appropriate to define an impaired cognitive population. One composite cannot be a perfect reflection of global cognitive function because patients suffer from several neuropsychological deficits. It would be unusual to observe one deficit in isolation [20, 47–49], and QoL measurement may be altered depending on the type of cognitive impairment . Thus, it is necessary to report additional information according to other definitions of cognitive dysfunction using a combination of different composites. To the best of our knowledge, this is the first study that uses the definition of cognitive dysfunction, which integrates a combination of different composites (i.e., memory, attention, and concentration).
Several limitations and strengths of this study should be mentioned:
The representativeness of the samples should be discussed. Compared to international and European cohorts, the MS patients in this study exhibited a more severe disability profile [30, 51], and the SCZ patients had a longer illness duration . Thus, an assessment of the reproducibility of our results is needed, using a larger and more diverse group of patients. However, the proportion of cognitively impaired subjects was consistent with the literature for MS [13, 20, 53] and SCZ [54, 55] domains.
One important aspect of this study concerns our definition of cognitive dysfunction because there is little consensus according to Achiron and Barak . We defined cognitive impairment using tests in which the French norms have been previously published [35–37]. This eliminated the need for a control group and enabled a consensus in defining patients as non-impaired or impaired for each test. However, our definition may be questionable because of the absence of a consensus on a ‘global definition’ of a patient with global cognitive dysfunction on the basis of a combination of these tests. However, we are convinced that our findings, independent of the definition of cognitive dysfunction, can help researchers to better understand the relevance of self-reported quality of life assessments for patients with cognitive disorders.
The suitability indices used to define the satisfactory properties relied on arbitrary decision rules, each of which will be discussed. Nevertheless, this approach enabled the determination of the suitability or unsuitability of different structures using the same decision tree independent of the questionnaire and disease. Thus, future studies may be performed to test different decision trees and to discuss the implications of the subsequent results.
Factors previously associated to cognitive performance, such as depression and fatigue [57, 58], and medications  were not considered. However, the aim of this study was to provide evidence supporting the conclusion that cognitively impaired patients reliably answer a self-reported QoL questionnaire regardless of the presence or absence of other factors that could have influenced their performance.