Strategies to improve patient-reported outcome completion rates in longitudinal studies

Multiple myeloma (MM) is an incurable malignancy of plasma cells in the bone marrow. MM is associated with severe morbidity, specifically caused by bone destruction and pathological bone fractures, renal dysfunction, high infection rate and potential physical disability [1, 2]. The prognosis of MM has improved markedly over the past 20 years, and the median survival of MM patients under the age of 70 has increased from 3 years [3] to 6–7 years [4, 5]. The improved prognosis is mediated by the introduction of high dose chemotherapy with autologous stem cell support (HDT) in the 1990s, new treatment options with immunomodulatory drugs (IMiDs), such as thalidomide, lenalidomide and pomalidomide [6‐8], and the proteasome inhibitors, bortezomib, carfilzomib and ixazomib [9‐11]. Most recently, the monoclonal antibodies elotuzumab and daratumumab [12, 13] have been introduced, and therefore the prognosis is expected to improve even further in coming years [14].

Treatment choice in MM depends on the patients’ age, disease complications, existing comorbidity and whether the patient is judged fit for specific regimens, such as HDT. Treatment usually involves repeated cycles of a 2–3 drug combination therapy with a proteasome inhibitor, IMiD, cytostatic agent and/or steroid. Treatment implies a risk of both acute adverse events, such as infections, as well as late effects, such as peripheral neuropathy and fatigue [15‐18].

Patients with MM report a high symptom burden. Common symptoms include fatigue, pain, constipation, insomnia and tingling in the hands/feet, with a consequent decrease in physical and cognitive functioning [19‐21]. Compared to patients with other haematological malignancies, patients with MM report low health-related quality of life (HRQoL) [21, 22]. Longitudinal HRQoL studies of patients with MM suggest that clinically beneficial improvements in HRQoL are more likely during primary treatments than during treatment for relapse [23].

The patients’ experience of symptoms and HRQoL can be validly and reliably captured with patient-reported outcome (PRO) questionnaires [24]. Integration of PROs into routine care management of patients with advanced cancer is associated with increased overall survival and reduction of the symptom burden [25, 26]. Typically, PRO assessments are scheduled at key time points and when a questionnaire is not completed at a scheduled time, it is referred to as a non-response (NR). Consequently, missing data can lead to a variety of problems, more so as NR rates increase, including loss of study power and precision [27, 28]. Further, there is risk of bias if the reason for NR is related to the patient’s health status and appropriate analysis methods are not applied. For example, if only complete case analysis methods are used, there is a risk of overestimated HRQoL and underestimated toxicity [27‐29]. This is due to the analysis being based on available data from patients with completed PRO assessments and who presumably have better HRQoL, while patients who drop out might have more toxicity and a worse HRQoL [27, 30‐34]. Thus, NR represents a threat to internal and external validity and is one of the inherent barriers in establishing high quality PRO data for use in patient-centred care [35‐37]. Several strategies designed to minimize NR have been proposed, and these can be integrated into the study design, protocol and implementation procedures for the PRO study [29]. These include ensuring that staff are aware of the importance of reducing NR and have access to written study procedure guidelines and support [29, 31]. Also, given the time-sensitive nature of PRO data, real-time monitoring of PRO completion rates during the study is recommended [38]. However, we are not aware of any studies that have assessed and documented the effectiveness of such strategies in reducing NR.

The study “Quality of life in Danish multiple myeloma patients” (QoL-MM) is a national multicentre, prospective, observational and primarily web-based survey with real-time monitoring of NRs. Several strategies to reduce NR have been implemented including education of study nurses, electronic reminders and real-time monitoring of PRO completion. The aim of this analysis was to investigate whether electronic reminders and real-time monitoring of NRs improve PRO completion rate.

The QoL-MM study includes newly diagnosed or relapsed, treatment-demanding patients with MM who, according to International Myeloma Working Group (IMWG) criteria, are eligible for inclusion [2]. Broad inclusion criteria ensure inclusion of a population-based, representative cohort of MM patients. Only, patients who are not able to understand the Danish language or who are diagnosed with a psychiatric condition are ineligible. Study sites include all 10 Danish departments of haematology. The goal is to recruit 800 patients, and each patient is followed for 24 months or until early drop-out due to withdrawal of consent, death or permanent lack of ability to fill out the questionnaires. The patients are introduced to the study by their treating physician or nurse, and written informed consent is obtained prior to inclusion. Demographic data are collected as part of the inclusion interview by a local study nurse. Moreover, the patients provide information related to activity of daily living (ADL), instrumental ADL and self-reported diseases, summarized into the Charlson Comorbidity Index [39‐41]. This information is used to calculate the Freiburger Comorbidity Index of 0-3 and the IMWG myeloma frailty score which divides the patients into three categories of “Frail”, “Intermediate Fitness” or “Fit” [42, 43]. The patient’s Karnofsky Performance status was assessed by the local study nurse [44]. Clinical data, e.g. date of diagnosis, MM subtype and the prognostic score, International Staging System, are collected from The Danish Multiple Myeloma Registry [45, 46]. Data on admissions, discharges and other hospital procedures are captured from The National Registry of Patients [47].

As of August 16th 2018, 481 patients were found eligible for the QoL-MM study, and hereof 292 provided written consent for participation and inclusion in the study. Of the 292 patients included, 271 had reached at least the first follow-up PRO assessment time point at 4 weeks and were included in the analyses. Patient and disease characteristics are presented in Table 1. Of all patients, 55% had newly diagnosed symptomatic MM of which 52% started an induction regimen with planned HDT at study entry. A daratumumab containing regimen was the most frequently used regimen in treatment of relapse MM and 24% of the patients with relapse started fifth or later line of therapy. Twenty-four percent of both newly diagnosed and relapse MM were 76 years or older, and 17% of the NDMM patients and 12% of the RMM were characterized as “Frail” according to the IMWG myeloma frailty score.

Electronic completion of follow-up questionnaires was chosen by 85% of the 271 patients, and 15% chose the paper-and-pencil method. Three patients changed mode of answering during follow-up, two from electronic method to paper method, as the electronic method was found to be too complicated.

The aim of the analysis was to investigate whether the strategies used to minimize NRs to scheduled questionnaires could increase PRO completion rate in a longitudinal study of MM patients receiving anti-myeloma treatment. Our initiatives included education of study nurses, electronic reminders and real-time monitoring of NR. A significant higher proportion of uncompleted questionnaires were completed after the patients had received the electronic reminder and after real-time monitoring. Those strategies resulted in a very high PRO completion rate of 95%, with just 5% non-response.

PRO completion rates in longitudinal studies of patients with MM are often not reported, and of the studies, where this information was reported, the PRO completion rate was 78–98% [53]. Only one study has reported a higher PRO completion rate compared to our findings. This was the NMSG 4/90 study by Gulbrandsen et al. [54], where the PRO completion rate of the historical control group of newly diagnosed patients with MM was 98%. This historical control group originates from the EORTC QLQ-C30 validation study of patients with MM [55]. One of the aims in the validation study was to evaluate the applicability of the questionnaire in a cohort of patients with MM and included sampling of data concerning the patient’s need for assistance in completing the questionnaires. The authors found that up to 30% of the MM patients reported need of assistance in completing the questionnaire. Other strategies of how this high PRO completion rate was achieved are not described in the paper [55].

When the QoL-MM study was designed, there had been particular focus on how we could minimize NR. We introduced real-time monitoring of NR and provided the patients with reminders. Staff resources were dedicated for this purpose, software as well as a high proportion of the patients choosing the web-based questionnaire method made it possible, and we succeeded in reaching a high PRO completion rate. Still, some NR could not be avoided. We found that low performance status at the time of inclusion and choosing paper-and-pencil method were predictors for not completing the questionnaires on-time. Risk of never response was higher for females, un-fit patients at baseline and choosing paper method for questionnaire completion. Special attention and guidance should be provided from the study staff to avoid NR from those patients. Information about the clinical status of patients when they fail to complete a scheduled questionnaire and the reason for NR might assist the PRO researchers in making the correct assumption for the underlying missing data mechanisms [27, 28, 56]. The link between the documented reasons, the predictors for NR and the missing data mechanisms as well as estimation of the impact of NR and patient drop-out on HRQoL results will be investigated in future analyses of the QoL-MM study.

The overall aim of the QoL-MM study is to describe the quality of life of the general population of patients with MM from diagnosis to late, advanced disease throughout different kinds of anti-myeloma therapies [57]. Methodological considerations concerning PRO assessment time points were included as part of the study planning. Clinical visits are frequently chosen time points for PRO assessment in clinical trials of MM patients [53]. This decision has the advantage of reducing the risk of NR, since patients have the opportunity to complete the questionnaires at the hospital with assistance from the study nurse. Disadvantages in using day 1 of treatment cycles include a potential risk of underestimation of toxicities that occur after day 1 and not capturing periods with temporary decline in HRQoL, resulting in rescheduling of chemotherapy [58]. We chose to collect the PRO data in QoL-MM at predefined calendar time points to meet the overall study aim and thereby capture HRQoL at regular non-clinic time points throughout the MM patients´ diverse disease trajectories. This decision could have made the study vulnerable for low PRO completion rates, since the general population of patient with MM can be frail and are at risk of adverse events, hospital admissions, as well as risk of physical and mental disabilities caused by the disease and therapy. Therefore, we implemented the educational and procedural strategies to reduce NRs. The same strategies might be useful to ensure high PRO completion rate in routine assessment and clinical utilization of PRO measures in the broad group of cancer patients [59].

Another strategie we chose was to use a pre-planned time frame of 7 days for each scheduled questionnaire. This allowed the study nurses to clearly communicate the expectations to the patients participating in the study and systematically capture reasons for NR from every patient who failed to complete the questionnaire within the 7-day response window [60, 61]. Whether patients who were not able to complete the questionnaires within the expected time frame have a poorer HRQoL will be investigated as part of the QoL-MM study.

We evaluated strategies to maximize PRO completion in a longitudinal cohort study of patients with MM receiving anti-myeloma treatment. Real-time monitoring of NR, electronic reminders and education of study nurses are effective strategies that resulted in a questionnaire completion rate of 95%. To our knowledge, the QoL-MM study is the first study to provide insight into how to ensure high PRO completion rates in a cohort of cancer patients receiving chemotherapy. We propose our applied strategies as a model for improving PRO completion rates in clinical trials, registries and routine care management to increase the quality and value of PRO data in patient-centred care.