Validation of a Patient-Completed Caprini Risk Score for Venous Thromboembolism Risk Assessment

Abstract

Introduction Individualized risk assessment for venous thromboembolism (VTE) using the Caprini risk score (CRS), coupled with targeted prophylaxis based on the score, is effective in reducing postoperative VTE. Critics contend that using this tool is time consuming for health care providers. We decided to create a patient-completed CRS and conducted a prospective study to compare the scores calculated by a patient with those calculated by a blinded physician for the same patient. Methods In phase 1, we interviewed patients in our deep vein thrombosis (DVT) support group who had a history of thrombosis and included their family members to determine areas of misunderstanding in the original CRS. We created a patient-completed form based on these interviews. In phase 2, we further optimized the questions after a CRS-trained, blinded physician scored 20 hospitalized patients during the pilot study. In the final (third) phase, we measured the agreement level between the new form filled out by the trained physicians and those filled out by the patients. The study was approved by our local institutional review board. Using PASS version 11, we determined that a sample size of 37 individuals achieves a power of 80%, to detect a 0.1 difference between the null hypothesis correlation of 0.5 and the alternative hypothesis correlation of 0.7 using a two-sided hypothesis test with a significance level of 0.05. We tabulated the individuals' answers and categorized the scores by using SPSS version 23 to estimate the kappa value, linear correlation, and the Bland–Altman test. A kappa value greater than 0.8 indicated an “almost perfect agreement.” Results We tested the first patient-completed CRS version (phase 2) in a 20-patient pilot study. A poor agreement was observed with the body mass index (BMI) responses in multiple iterations, and so we excluded the BMI calculation from the final patient-completed CRS form. We recruited 42 patients with an average age of 55, mostly female (45%), who completed less than college education (62%) to fill out the updated CRS form (phase 3). An almost perfect agreement was found for both the individual questions and the overall score comparing physician and patient answers, resulting in a high correlation (r = 0.95). In Bland–Altman, we did not find any trend for extreme values. Conclusion We created and validated a patient-completed CRS form that has an excellent agreement level with the physician-completed form. From the results, the physician only needs to calculate the BMI. The average time for a patient to complete the form was 5 minutes. The average time for the physician to finalize the score was approximately 6 minutes. Implementation studies are needed to assess the correlation of the aggregated score, derived from this form, with the occurrence of perioperative VTE.