Quality of Patient-Reported Outcomes in Multiple Myeloma Randomized Controlled Trial Protocols and Publications...

Patient-reported outcomes (PROs) are now frequently incorporated into multiple myeloma (MM) randomized controlled trials (RCTs) to help inform clinical decision making. Although the quality of PRO components in cancer RCT protocols is generally recognized as suboptimal, there are limited data on adherence to international quality standards in MM RCT protocols.

We performed a systematic review to identify MM RCTs published between January 2014 and June 2023 that utilized the EORTC QLQ-C30 questionnaire. The quality of PRO-specific protocol content was evaluated using the SPIRIT-PRO guidelines, which establish key requirements for PRO inclusion in protocols. The framework consists of 16 items: 5 elaborations of the existing SPIRIT checklist and 11 PRO-specific extensions items. The quality of PRO reporting was evaluated using CONSORT-PRO Extension, which comprises 14 items to improve the reporting of PRO data in clinical trials.

Our systematic review identified 35 RCTs encompassing 20,612 patients, with 24 trials (69%) having publicly accessible protocols. The median protocol compliance was 7.5 SPIRIT-PRO items. Analysis of 24 protocols showed strong compliance in assessment schedules (96%), PRO domains/instrument justification (79%), and PRO objectives (71%). Intervention deviation procedures were described in 62% of protocols. Half included comprehensive data collection plans, while 58% addressed missing data methods and 42% outlined PRO research questions or multiplicity controls. Notably weak areas included strategies for preventing missing data (38%), and only 8% of protocols detailed PRO monitoring plans, personnel specifications, or eligibility criteria. CONSORT-PRO scores varied across all 35 RCTs (median 11, range 0.5-14). Among the 35 RCTs, while trials showed strong reporting of statistical elements (94% precision estimates, 91% subgroup analyses, 89% intention-to-treat, 86% timepoint completion) and results (80% hypothesized domains, 71% PRO validity, 69% clinical implications, 66% limitations), key PRO-specific components were reported by less than two-thirds of RCTs: only 11% stated PRO hypotheses, 31% specified administration mode, 43% identified domains, 60% addressed missing data, and 63% provided PRO rationale. Two factors were associated with higher reporting quality, both potentially reflecting publication bias: having a secondary PRO-focused paper (p < 0.001) and finding statistically significant EORTC QLQ-C30 differences (p = 0.024). Our multivariable analysis showed no significant association between SPIRIT-PRO scores and CONSORT-PRO scores.

Despite some foundational strengths of existing MM RCT protocols, gaps exist in PRO methodological specifications, statistical analysis, and clinical interpretation. Our findings may help to better inform PRO implementation in future MM RCT protocols. For example, since missing PRO data can be due to informative censoring, there should be increased attention on considering how to minimize missing data, already at the stage of protocol writing. Our findings regarding predictors of higher reporting quality suggest that trials with significant PRO differences receive systematically better reporting, possibly introducing bias in the available evidence base.