WHODAS

Lower cognitive performance was associated with greater difficulties in participants’ mobility, household management, self-care, self-reported cognitive function, participation in their community and social relationships over the previous 30 days. These findings align with previous research showing that cognition is associated with daily functioning, including personal care, lifting and carrying, walking, managing a household, managing medical care and getting around. ^{Reference Elovainio, Kivimäki, Ferrie, Gimeno, De Vogli and Virtanen4–Reference Jacob, Smith, Thoumie, Haro, Stickley and Koyanagi6} This study extends prior research by including younger adults, as 38% of the NCMH and 31% of HealthWise Wales participants were aged 40 or under. The association between cognition and functioning remained significant after controlling for age and when limited to working-age adults, suggesting the findings are not driven solely by age-related decline. While Jacob et al ^{Reference Jacob, Smith, Thoumie, Haro, Stickley and Koyanagi6} included participants as young as 16 years, they relied on the National Adult Reading Test, a measure of crystallised IQ. Our study employed a broader cognitive battery that incorporated both crystallised and fluid intelligence.

Another strength of this study was the use of the same cognitive and outcome measures in both a clinical sample with diagnosed mental health conditions and a population-based sample. In both data-sets, a one s.d. increase in overall ‘g’ was associated with decreases of 17–35% in the odds of reporting impairment across the six domains. The NCMH is a sample of participants with psychiatric and neurodevelopmental diagnoses, including psychotic, mood and anxiety disorders, so cognitive and functional impairments are more likely and, thus, our results are not unexpected. In contrast, the participants recruited from HealthWise Wales were an online population sample with high levels of education and employment (a full comparison of this sample compared with the Welsh population estimates can be found in Lynham et al). ^{Reference Lynham, Jones and Walters19} Despite this, the strengths of these associations were similar across both the clinical and population-derived data-sets, suggesting that the cognition-functioning relationship is not specific to psychiatric populations or older adults.

As a cross-sectional study, we cannot determine the direction of causality between cognition and functioning (for example, by conducting mediation analysis). Functional impairments such as limited motor function could affect cognitive performance. Cognitive ability also influences how well individuals understand and manage their health conditions, as more complex health information increases cognitive burden, reducing the likelihood it is retained. ^{Reference Wilson and Wolf27,Reference Serper, Patzer, Curtis, Smith, O’Conor and Baker28} In studies of diabetes, asthma and chronic obstructive pulmonary disease, processing speed and working memory have been associated with health literacy and subsequent health management. ^{Reference Levinthal, Morrow, Tu, Wu and Murray29–Reference Nguyen, Kirk, Arcury, Ip, Grzywacz and Saldana32} Thus, poorer cognition may lead to functional impairment due to inadequate health management. Longitudinal research has shown that cognitive function can predict later physical function, but not the reverse. ^{Reference Elovainio, Kivimäki, Ferrie, Gimeno, De Vogli and Virtanen4}

In both data-sets, current depression and anxiety symptoms were strongly associated with WHODAS scores. This may partly reflect the commonalities between the HADS and WHODAS, as both assess recent self-reported issues with emotions, self-care and daily activities. Individuals with these symptoms may also be more likely to report difficulties due to negative bias. However, the similar effect of HADS on cognition and employment analysis suggests that the effect extends beyond reporting bias. Cognitive and functional impairments may result from mood and anxiety disorders, ^{Reference Abramovitch, Short and Schweiger7} though it’s also possible that lower cognition increases vulnerability to mood symptoms and functional impairments.

In HealthWise Wales, both history of depression/anxiety and current symptoms were associated with poorer functioning. In the NCMH, history of depression or anxiety was not included in the final model, as neither diagnosis was associated with WHODAS scores. There are three key differences between the NCMH and HealthWise Wales that may explain this finding. First, the NCMH is a clinical sample, so those without depression or anxiety will have other (possibly more severe) mental health conditions. In contrast, HealthWise Wales is a population-based sample, so we were comparing participants with a diagnosis of depression/anxiety against predominantly those without a history of psychiatric disorders. As such, the analyses across the two samples were addressing slightly different questions. Second, diagnoses were derived from different sources in the NCMH (self-report of history of diagnosis by a health professional) and HealthWise Wales (health records). Few studies have compared agreement between self-report and health record diagnoses of depression and anxiety. A UK Biobank study found low agreement between self-report diagnoses of depression and anxiety and those obtained through medical records of hospital admissions, which likely reflects the low rate of hospital admissions among affected individuals. ^{Reference Davis, Cullen, Adams, Brailean, Breen and Coleman33} Nevertheless, self-reported diagnoses are generally considered a valid measure, as studies comparing self-reported clinical diagnoses with validated questionnaires or structured interviews show that most participants who report being diagnosed with depression or anxiety would meet criteria for those disorders. ^{Reference Davis, Cullen, Adams, Brailean, Breen and Coleman33–Reference Davies, Buckman, Adey, Armour, Bradley and Curzons35} Third, definitions of diagnoses differed between cohorts. The NCMH used lifetime history of depression or anxiety diagnosis, while diagnoses in HealthWise Wales were derived using a validated algorithm that combines depression and anxiety and requires both record of diagnosis ever and evidence of treatment and symptoms or diagnosis within 12 months of the CONCA assessment. To improve comparability, we created a combined depression/anxiety variableam restricted to participants reporting current treatment, but this was not associated with cognition so this variable was also not included in the final models for the NCMH. The only diagnosis associated with both cognitive performance and WHODAS scores in the NCMH was bipolar disorder, which is consistent with studies showing poor cognitive and functional outcomes in bipolar disorder. ^{Reference Depp, Mausbach, Harmell, Savla, Bowie and Harvey8}

Lifestyle factors that may underlie the association between cognitive function and self-reported functional impairment were also investigated in the current study, including alcohol use, smoking, BMI and physical activity (the latter two in HealthWise Wales only). In HealthWise Wales, physical activity was the only factor associated with better overall functioning. In the NCMH, neither smoking nor alcohol use was significantly associated with functioning in the final model.

Finally, it is worth noting that the association between cognition and WHODAS scores remained significant after controlling for these covariates. This may suggest a direct association between cognition and functioning, or that other explanatory variables were not included in this study. The focus of this study was on the role of mental health, but future studies should evaluate the role of physical health, particularly given evidence that some common physical health problems have an impact on cognition. ^{Reference Xue, Xu, Ou, Cao, Tan and Tan36,Reference Ou, Tan, Shen, Xu, Hou and Dong37}

Real-world indicators of functioning

Two real-world functioning indicators were associated with cognition: living with a partner or spouse and being in employment or education. In HealthWise Wales, an increase of 1 s.d. in ‘g’ was associated with a 19% increase in the likelihood of cohabiting. Few studies have examined the relationship between cohabitation/marriage and current cognitive function; however, one study of healthy, middle-aged adults found that married individuals had better episodic memory than single individuals. ^{Reference Mousavi-Nasab, Kormi-Nouri, Sundstrom and Nilsson38} The authors proposed that the social interactions involved in marriage may stimulate cognitive function by facilitating neuronal growth and maintenance, as well as protecting against degeneration. ^{Reference Mousavi-Nasab, Kormi-Nouri, Sundstrom and Nilsson38} An alternative explanation is that cognitive function may play a role in the ability to develop and maintain close interpersonal relationships, as this requires skills such as communication, planning and processing of others’ emotions. It should be noted that cohabitation is an imperfect measure of social functioning. We sought to include a real-world measure to supplement the self-reported WHODAS data. Cohabitation was selected from the existing data as the most appropriate measure of social functioning available in HealthWise Wales, but it does not fully reflect the range of social functioning assessed by the WHODAS, which evaluates community participation, dealing with strangers and maintaining friendships. Nevertheless, we did find an association between cognition and cohabitation, and this remained significant after accounting for current mood symptoms, smoking history, physical activity and education. This suggests that cognition may be a potential determinant of close interpersonal relationships in the general population and warrants further investigation.

Our finding that cognitive function is associated with employment status is consistent with previous research, which demonstrates that employed participants have higher cognitive scores than unemployed participants. ^{Reference Heaton, Akshoomoff, Tulsky, Mungas, Weintraub and Dikmen39} An increase of 1 s.d. in ‘g’ was associated with a 22% increase in the likelihood of employment in HealthWise Wales and a 31% increase in the NCMH. In both data-sets, the association between cognition and employment was not significant after adjusting for covariates. In the NCMH, higher HADS scores and diagnosis of bipolar disorder or psychosis were associated with a reduced likelihood of employment, while current alcohol use and having a degree were associated with a higher likelihood of employment. Thus, the relationship between lower cognition and reduced employment appears to be explained by education and illness-related factors, such as current depressive and anxiety symptoms and diagnosis of severe mental illness. This also seemed to be the case in HealthWise Wales, where the association was not significant after inclusion of current depression/anxiety symptoms and physical activity.

The use of real-world measures was important given evidence from psychiatric studies that self-reported disability may not fully align with objective functional outcomes. ^{Reference Strassnig, Kotov, Fochtmann, Kalin, Bromet and Harvey40} Our findings that cognitive function is associated with WHODAS scores and cohabitation but not employment status (after accounting for covariates) may suggest that cognition is more closely related to perceived disability and social functioning than occupational status. However, a limitation of this study is that we did not have data on other functional measures, such as living independently, other social contact (family, friendships) or performance-based measures to fully explore this.

Limitations

This study has several limitations beyond its cross-sectional design. Online data collection can lead to recruitment bias, as we have previously demonstrated. ^{Reference Lynham, Jones and Walters18,Reference Lynham, Jones and Walters19} In the HealthWise Wales cohort, the WHODAS and HADS were optional, which limited the sample size and generalisability. The subsample who completed the WHODAS was younger, more likely to be women and had a higher degree of education, and was less likely to smoke or use a tablet computer. This may mean they were less likely to report functional impairments than the wider cohort. Nonetheless, we obtained WHODAS data on over 1000 participants, 22–67% of whom reported impairment across domains, and found an association between cognition and WHODAS scores.

Data from CONCA were integrated with existing HealthWise Wales data, resulting in some variables being collected at different times. To address this, we included only data collected within 1 year of CONCA participation. Importantly, variables such as BMI, alcohol use and physical activity are generally stable over time. ^{Reference Mertens, Clarys, Mullie, Lefevre, Charlier and Knaeps41,Reference Burgard, Lin, Segal, Elliott and Seelye42} In addition, there were some differences in the measurement of some variables, such as diagnosis and alcohol use, as the analyses were constrained by the availability of existing data. However, our main variables of interest (cognition, functioning and mood symptoms) were comparable across both cohorts.

While the HADS aims to distinguish between depression and anxiety, the high correlation between subscales and frequent co-occurrence of both disorders meant we could not isolate their effects. However, given the high comorbidity seen in the general population, focusing on these symptoms together may better reflect real-world experiences. ^{Reference Lamers, van Oppen, Comijs, Smit, Spinhoven and van Balkom43}

Both HealthWise Wales and the NCMH recruited in the same regions simultaneously, so some sample overlap is possible. We were unable to determine the extent of this due to data-sharing restrictions. Any overlap is likely to be minimal, as this would require participation in four separate studies (HealthWise Wales, the NCMH and two CONCA follow-ups).

Implications and future directions

In summary, we report that cognitive performance was associated with measures of functioning in a population sample and a psychiatric sample and the strength of these associations were similar in both samples. A measure of current depression and anxiety symptoms was identified as a key factor in this relationship, highlighting the important contribution of common mental health symptoms towards functioning. However, cognition remained associated with self-reported functional difficulties after controlling for these symptoms. Taken together, our results suggest that the cognition-functioning relationship is not specific to psychiatric populations and cannot be fully explained by current symptoms or psychiatric history in the general population. This strengthens the existing evidence that cognitive function is an important marker of functional difficulties and suggests both cognitive function and depression/anxiety symptoms may be valuable targets for interventions designed to improve functional outcomes, not just in populations of people with psychiatric disorders but more widely. Longitudinal and mediation approaches are needed to determine the causal direction of the associations between cognition, mental health and functioning.