The platelet variability index tool to determine the likelihood of immune thrombocytopenia (ITP): User testing and tool refinement Free

Introduction

The diagnosis of immune thrombocytopenia (ITP) is challenging due to the lack of a reliable diagnostic laboratory test. The platelet variability index (PVI) is a simple metric that captures the magnitude of the platelet count fluctuations over time and the severity of the thrombocytopenia from 3 or more consecutive platelet count measurements. An elevated PVI score, reflecting a high degree of platelet variability and severe thrombocytopenia, can distinguish between patients with ITP and patients with thrombocytopenia from non-immune causes (Li et al, Blood Advances 2021). We conducted a usability study to refine the PVI application based on user feedback.

Methods

We recruited 5 hematologist users from Canada, the U.S., and Italy to provide feedback on the presentation and utility of the web-based application of the PVI tool. After a virtual presentation of the application to users, we solicited unstructured feedback in person and by email followed by a group discussion. We then delivered a structured electronic survey to users to evaluate 5 reporting options for the PVI score, which varied in the wording of the interpretation summary and in the number of likelihood categories (high, moderate, low +/- very low likelihood of ITP). Options for the interpretation summary included a combination of the expected number of ITP patients (out of 100) in each category based on the previously published derivation cohort (Li et al, Blood Advances 2021) and the overall likelihood category. Respondents rated each option for clarity, clinical utility, interpretability, and workflow using a Likert scale with an opportunity for free text feedback. The final PVI application format was presented back to users for final comment and approval.

Results

During the initial unstructured feedback and discussion, users identified the need for a hierarchical designation of ITP likelihood based on the PVI score rather than a simple dichotomous result (ITP is likely or unlikely). All 5 options presented in the survey were rated as easy to understand, clinically useful, helpful for interpreting the likelihood of ITP, and applicable to clinical workflow by most respondents (80%, n=4). All respondents favored a concise summary of the patient's likelihood of ITP rather than the expected prevalence numbers. Most respondents (60%, n=3) preferred the inclusion of a “very low likelihood of ITP” category to improve interpretability, align with prevalence estimates, and enhance confidence in ruling out the ITP diagnosis for patients with the lowest scores. The final PVI application format that was agreed upon by all users included a likelihood category (from very low likelihood of ITP to high likelihood of ITP) and the expected prevalence numbers for each category presented as a table for reference. For example: “PVI Score = 2. This patient has a very low likelihood of ITP. For PVI score 0 – 2, less than 3 patients per 100 would be expected to have ITP.”

Conclusions

User feedback on the PVI tool focused on clear and concise interpretation summaries for the likelihood of ITP based on the PVI score. Further studies will assess the impact of the PVI tool on clinical outcomes for patients with thrombocytopenia who are suspected of having ITP.