Short communication The development and validation of a carer questionnaire to assess cognitive function in neuropsychiatric patients Sharan Randhawa a , Mark Walterfang a,b, 4 , Kathryn Miller a , Amelia Scholes a , Ramon Mocellin a , Dennis Velakoulis a,b a Neuropsychiatry Unit, Royal Melbourne Hospital, Melbourne, Australia b Melbourne Neuropsychiatry Centre, University of Melbourne, Melbourne, Australia Received 30 October 2006 Abstract Objectives: The carer history is an integral part of the assessment of patients with cognitive impairment. We aimed to develop a comprehensive yet concise carer questionnaire, the CogRisk, which captures actuarial risk variables for cognitive impairment in addition to key symptoms suggestive of cognitive decline in a number of cognitive domains, and to then assess its validity and reliability in a neuropsychiatric population. Method: Carers of patients assessed for cognitive impairment completed the CogRisk, and patients were clinically assessed using the Mini- Mental State Examination (MMSE) and Neuropsychiatry Unit COGnitive assessment tool (NUCOG). Reliability was assessed using test–retest and interrater measures and measures of internal consistency. Construct and concurrent validity was assessed using correlation between total and subscale scores on the CogRisk, total scores on the NUCOG and MMSE, and subscale scores on the NUCOG. Predictive validity was determined using measures of sensitivity and specificity and using receiver operating character- istic (ROC) methods. Results: The CogRisk was completed by all carers in less than 10 min. The total CogRisk score correlated significantly with total MMSE and NUCOG scores (r =À0.511 and À0.563, respectively) and remained highly significant when age and education were controlled for. Internal consistency of CogRisk items was high (a =0.943). Intrarater reliability of the CogRisk was high with an intraclass correlation coefficient of .978 ( Pb.001), and interrater reliability between carers was also high at 0.868 ( Pb.05). Sensitivity and specificity for the detection of dementia were .70 and .73, respectively, with area under the ROC curve not significantly different from that of the MMSE or NUCOG. Conclusion: The CogRisk is a brief carer-rated tool of a patient’s cognitive functioning developed for use within a neuropsychiatric setting. It exhibited good concurrent validity, internal consistency, and interrater and intrarater reliability. The CogRisk also demon- strated good sensitivity and specificity for dementia. The CogRisk provides carer information, which complements the clinical assessment and can be used to focus on direct carer interview. D 2007 Elsevier Inc. All rights reserved. Keywords: Cognitive; Informant; Neurocognitive; Dementia; Neuropsychiatric Introduction An important component of the neurocognitive assess- ment in psychiatry is the bcognitive history,Q which can be provided by both patient and carer. The veracity of a patient account of cognitive symptoms may be limited, due to communication or language difficulties following stroke, a lack of insight or awareness of deficits in dementia, or a lack of cooperation or capacity to attend to the interview process in major mental illness [1–3]. As a result, patients with significant cognitive impairment will often under-report their degree of symptomatology [4,5]. In this circumstance, clinicians will generally rely on a corroborative history, usually from carers who have a significant degree of ongoing contact with the patient, as carer accounts of cognitive impairment are strongly predictive of clinical diagnoses of dementia when multiple domains of cognitive function are enquired about [1,6,7]—although poorly 0022-3999/07/$ – see front matter D 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jpsychores.2007.02.004 4 Corresponding author. Neuropsychiatry Unit, Royal Melbourne Hospital, Level 2, John Cade Building Royal Melbourn, Parkville, Vic 3050, Australia. Tel.: +61 393428750; fax: +61 393428483. E-mail address: mark.walterfang@mh.org.au (M. Walterfang). Journal of Psychosomatic Research 63 (2007) 93 – 98