Individualized decision aid for diverse women with lupus nephritis (IDEA-WON): A randomized controlled trial

Abstract

Background: Treatment decision-making regarding immunosuppressive therapy is challenging for individuals with lupus. We assessed the effectiveness of a decision aid for immunosuppressive therapy in lupus nephritis.

Methods and findings: In a United States multicenter, open-label, randomized controlled trial (RCT), adult women with lupus nephritis, mostly from racial/ethnic minority backgrounds with low socioeconomic status (SES), seen in in- or outpatient settings, were randomized to an individualized, culturally tailored, computerized decision aid versus American College of Rheumatology (ACR) lupus pamphlet (1:1 ratio), using computer-generated randomization. We hypothesized that the co-primary outcomes of decisional conflict and informed choice regarding immunosuppressive medications would improve more in the decision aid group. Of 301 randomized women, 298 were analyzed; 47% were African-American, 26% Hispanic, and 15% white. Mean age (standard deviation [SD]) was 37 (12) years, 57% had annual income of <$40,000, and 36% had a high school education or less. Compared with the provision of the ACR lupus pamphlet (n = 147), participants randomized to the decision aid (n = 151) had (1) a clinically meaningful and statistically significant reduction in decisional conflict, 21.8 (standard error [SE], 2.5) versus 12.7 (SE, 2.0; p = 0.005) and (2) no difference in informed choice in the main analysis, 41% versus 31% (p = 0.08), but clinically meaningful and statistically significant difference in sensitivity analysis (net values for immunosuppressives positive [in favor] versus negative [against]), 50% versus 35% (p = 0.006). Unresolved decisional conflict was lower in the decision aid versus pamphlet groups, 22% versus 44% (p < 0.001). Significantly more patients in the decision aid versus pamphlet group rated information to be excellent for understanding lupus nephritis (49% versus 33%), risk factors (43% versus 27%), medication options (50% versus 33%; p ≤ 0.003 for all); and the ease of use of materials was higher in the decision aid versus pamphlet groups (51% versus 38%; p = 0.006). Key study limitations were the exclusion of men, short follow-up, and the lack of clinical outcomes, including medication adherence.

Conclusions: An individualized decision aid was more effective than usual care in reducing decisional conflict for choice of immunosuppressive medications in women with lupus nephritis.

Trial registration: Clinicaltrials.gov, NCT02319525.

Fig 1. CONSORT diagram for patient selection.

The figure includes 1,005 patients prescreened using electronic medical records and/or in-person screening. Of these, 704 were excluded, 19 declined to participate (12 had no time; 5 were not interested in participating in a research study, and 2 did not want any more information about medications for lupus), and 1 was excluded due to psychosis; An asterisk (*) indicates that the remaining 684 did not meet eligibility, and the reasons were as follows: no nephritis (n = 388); candidate for or already had renal transplant (n = 162); dialysis (n = 99); potential but no clinic appointments (n = 29); and miscellaneous (n = 6): needed biopsy (n = 1); kidney disease not due to lupus (n = 1); lupus complications, nonrenal (n = 1); none of the scenario matches any treatment option for the patient (n = 1); and not sufficient evidence from clinic notes (n = 2). ACR, American College of Rheumatology.

Fig 2. Pre and Post-intervention decisional conflict (0–100; higher score indicates more conflict).

Dashed line represents the threshold for unresolved clinically significant decisional conflict (≥25) on the DCS. The box plot shows the median, indicated by the solid line, and 25th and 75th percentiles as the lower and upper bounds of the box. The whiskers represent the minimum and maximum values. DCS, Decisional Conflict Scale.

Fig 3. Subgroup analyses for decisional conflict scale (DCS) score.

Differences that are statistically significant after correcting for multiple comparisons are represented with a filled square; others are presented with an open square. The hashed vertical line represents a difference of zero in DCS scores post-intervention, i.e., no difference between the decision aid and pamphlet groups. An asterisk (*) indicates statistically significant Bonferroni-corrected p-value (p < 0.0008). The categorization for subgroups were as follows: graphical literacy: low, 0–2, High, 3–4; SAHL: low, 0–14, high, >14; numeracy: low, 0–3, high, 4–6; trust in physicians: low, <44, high, 44–55. DCS, Decisional Conflict Scale; SAHL, Short Assessment of Health Literacy.

Fig 4. Subgroup analyses for unresolved decisional conflict scale (DCS) score of ≥ 25.

Differences that are statistically significant after correcting for multiple comparisons are represented with a filled square; others are presented with an open square. The hashed vertical line represents an odds ratio of one for unresolved DCS scores post-intervention, i.e., no difference in the proportion of people with unresolved decisional conflict between the decision aid and the pamphlet groups. An asterisk (*) indicates statistically significant Bonferroni-corrected p-value (p < 0.0008). The categorizations for subgroups were as follows: graphical literacy: low, 0–2, high, 3–4; SAHL: low, 0–14; high, >14; numeracy: low, 0–3; high, 4–6; trust in physicians: low, <44, high, 44–55. DCS, Decisional Conflict Scale; SAHL, Short Assessment of Health Literacy.