Development of a rapid screener through network analysis to identify central cognitive complaints in haemodialysis patients: A cross-sectional study

Abstract

Objectives Cognitive impairments and cognitive complaints are commonly present in dialysis patients and can affect clinical, functional, occupational, and psychosocial well-being. It is important to screen for patients' cognitive status as it offers a gateway to specialty referral, prevention or rehabilitation programmes, and personalisation of clinical care. The Patient's Assessment of Own Functioning Inventory (PAOFI) is a comprehensive questionnaire that assesses patient-reported difficulties in memory, language, motor/sensory-perceptual skills and higher-level cognitive function. In the current study, we adopted network analysis to identify central cognitive complaints in dialysis patients and derived a PAOFI short form (PAOFI-SF) based on these core symptoms to improve screening efficiency in real-world renal settings. Design Multicentre, cross-sectional study. Setting Participants were recruited from 10 community-based dialysis centres in Singapore, from May to November 2022. Participants A total of 369 eligible haemodialysis patients were invited to join the study, and 268 completed the measures (response rate 72.6%). Outcome measures Cognitive assessment tools including the PAOFI and the Montreal Cognitive Assessment were administered. Results Based on the PAOFI measure, 98 participants (36.6%) endorsed the presence of three or more complaints, indicating clinically significant cognitive complaints. Network analysis identified five central cognitive complaints among dialysis patients: problem-solving difficulty, difficulty following instructions, forgetting how to do tasks, difficulty being understood, and forgetting people known years ago. These core items were combined into a five-item short form of PAOFI, which showed good reliability and validity, and an area under the curve of 83.4% in identifying clinically significant cognitive complaints. The optimal cut-off point of the short form was 11.5 (out of 30), with a specificity of 89.5%, sensitivity of 63.9%, positive predictive value of 77.5% and negative predictive value of 81.4%. This cut-off point also predicted objective cognitive performance even after controlling for sociodemographic and clinical confounders. Conclusions Pending future replication and external validation, the PAOFI-SF may be suitable for use in renal care settings as an initial screening tool to identify patients with cognitive complaints and increased risk of objective cognitive impairments.