Stanford Study of the Validity of the Pre-Interventional Preventive Risk Assessment (PIPRA) Tool for Delirium in an At-Risk Elderly Orthopedic Population

Abstract:

Background: Post-operative delirium (POD) occurs more often in older patients undergoing surgical procedures and is associated with multiple comorbidities and mortality. There are currently predictive scales to assess the risk of developing delirium; however, scales with good predictive ability are too cumbersome to be clinically practical. The PIPRA tool eliminates these constraints and provides an accessible approach to evaluate POD risk, with a prior study demonstrating an AUC of 0.74 in orthopedic patients. The algorithm parameters include previously demonstrated risk factors for POD development. In this study, our goal is to determine the clinical utility of PIPRA and its potential to identify patients with high POD risk.

Methods: We enrolled patients who underwent elective orthopedic surgery and tested them before, during, and after for delirium post-operatively using the Confusion Assessment Method (CAM) as applied by a trained research assistant. All patients were assessed for the development of POD during the first three days after surgery. The PIPRA tool, consisting of 9 items, were applied to the subjects' electronic medical record retrospectively resulting in a delirium prediction score.

Results: 250 patients were enrolled in the study over the age of 60 years. Of these patients, PIPRA could be applied to 241 patients. Delirium was diagnosed in only 13 (5.4%) patients. Among the PIPRA parameters collected, the largest differences between delirious and non-delirious patients were history of cognitive impairment and minimum/average PIPRA score. The average PIPRA score difference between non-delirious and delirious patients was essentially doubled. The AUC for this study was 0.748 (95% CI: 0.635-0.861).

Discussion: The PIPRA tool is based on the preoperative delirium predictive model (PIPRA Algorithm) utilizing clinical parameters that are both significantly associated with POD development, practical, and readily available. This predictive model has the potential to identify patients at high-risk of POD and hopefully will aid in reducing associated morbidity and mortality.

This study demonstrated an AUC of 0.748, very similar to previously published results with an AUC 0.74, within the margin of error. The data suggest that certain PIPRA parameters correlate more strongly with delirium risk such as history of cognitive impairment.

Study limitations include small number of delirious patients, being underpowered, variability of PIPRA scores, missing information during chart review, and possible selection bias. These results are not generalizable to other patient cohorts at increased risk for delirium.

The data suggest evidence of PIPRA’s potential to predict POD risk in patients receiving elective orthopedic surgery. Next steps include retroactively applying PIPRA to other surgical populations in more adequately powered studies. After collecting sufficient data, a factor analysis of each PIPRA parameter would likely optimize predictability.

Conclusion: The PIPRA tool introduces an innovative and accessible approach to evaluating the risk of POD in surgical patient populations. Despite the limitations of current studies, this model shows potential to identify patients at high risk of POD with the goal of reducing morbidity and mortality.