Risk factors for postoperative delirium: An umbrella review of systematic reviews

Background: Postoperative delirium (POD) is associated with increased mortality, increased length of hospital stays and increased rates and severity of subsequent cognitive decline including dementia. A wide range of risk factors for POD have been suggested in the literature across multiple surgical specialities. However few are validated and no accurate prognostic models exist. We therefore aimed to map the existing evidence regarding risk factors for POD to help guide future research by undertaking an umbrella review of systematic reviews examining risk factors for POD in any context. Materials and methods: We systematically searched multiple medical databases for systematic reviews examining the risk factors for POD in adults undergoing any surgery. We then selected relevant reviews with minimal overlap in primary studies and extracted information about individual risk factors. Results: Thirty-five relevant reviews were identified of which ten were in trauma and orthopaedic surgery patients (four exclusively examined hip fractures), five were in cardiac surgery patients, and four were in vascular surgery patients. Due to substantial overlap in reviews, eighteen reviews were analysed in detail finding the widely examined and consistent risk factors were increasing age, nursing home residency, pre-existing cognitive impairment, psychiatric disorders, cerebrovascular disease, end stage renal failure, low albumin, higher ASA score, and intra-operative blood transfusion. Many other risk factors were examined, but they were either not studied in multiple systematic reviews, or inconsistent either in results or in categorisation (which for many factors was heterogenous even within systematic reviews). There are also a large number of existing prognostic models, many of which remain unvalidated. Conclusion: Given the wealth of existing literature, future research should avoid simple risk factor evaluation except for novel candidates, validate existing prognostic models where possible, and instead focus on interventional research.