Prognostic risk estimates of patients with multiple sclerosis and their physicians: comparison to an online analytical risk counseling tool

Abstract

Background: Prognostic counseling in multiple sclerosis (MS) is difficult because of the high variability of disease progression. Simultaneously, patients and physicians are increasingly confronted with making treatment decisions at an early stage, which requires taking individual prognoses into account to strike a good balance between benefits and harms of treatments. It is therefore important to understand how patients and physicians estimate prognostic risk, and whether and how these estimates can be improved. An online analytical processing (OLAP) tool based on pooled data from placebo cohorts of clinical trials offers short-term prognostic estimates that can be used for individual risk counseling.

Objective: The aim of this study was to clarify if personalized prognostic information as presented by the OLAP tool is considered useful and meaningful by patients. Furthermore, we used the OLAP tool to evaluate patients' and physicians' risk estimates. Within this evaluation process we assessed short-time prognostic risk estimates of patients with MS (final n = 110) and their physicians (n = 6) and compared them with the estimates of OLAP.

Results: Patients rated the OLAP tool as understandable and acceptable, but to be only of moderate interest. It turned out that patients, physicians, and the OLAP tool ranked patients similarly regarding their risk of disease progression. Both patients' and physicians' estimates correlated most strongly with those disease covariates that the OLAP tool's estimates also correlated with most strongly. Exposure to the OLAP tool did not change patients' risk estimates.

Conclusion: While the OLAP tool was rated understandable and acceptable, it was only of modest interest and did not change patients' prognostic estimates. The results suggest, however, that patients had some idea regarding their prognosis and which factors were most important in this regard. Future work with OLAP should assess long-term prognostic estimates and clarify its usefulness for patients and physicians facing treatment decisions.

Figure 1. Perception of the information provided by the OLAP tool.

Boxplots represent ratings on a VAS normalized to 0% to 100% with 50% representing a neutral rating. Extreme poles with pairs of adjectives are labeled to assess understanding (no understanding vs. complete understanding), relevance (not relevant vs. highly relevant), interest (not interesting vs. highly interesting), and threat (threatening vs. reassuring). Data as median and quartiles, outliers represented by plus signs.

Figure 2. Estimates for EDSS ≥6 after 3 years.

Mean estimates by patients at baseline and after exposure to the OLAP tool and by physicians, in relation to the OLAP tool's estimates are given. Median estimates for each of the groups can be read off the y-axis in relation to the predictions made by the OLAP tool (x-axis; binned into 7 categories, therefore not exactly bisecting the angle of x- and y-axis).

Figure 3. Correlations between risk estimates and disease covariates.

Data are shown separately for patients (at baseline and after exposure to the OLAP tool), physicians, and the OLAP tool. Patients' and physicians' estimates correlated most strongly with those factors that also correlated most strongly with the OLAP tool's predictions, both regarding the risk of EDSS 6 or higher after 3 years (upper panel) and relapses after 18 months (lower panel). Note: Course of the disease was given a value of 1 for CIS and RRMS and a value of 2 for SPMS and PPMS.