Minimally important differences for interpreting EORTC QLQ-C30 change scores over time: A synthesis across 21 clinical trials involving nine different cancer types

INTRODUCTION

Early guidelines for minimally important differences (MIDs) for the EORTC QLQ-C30 proposed ≥10 points change as clinically meaningful for all scales. Increasing evidence that MIDs can vary by scale, direction of change, cancer type and estimation method has raised doubt about a single global standard. This paper identifies MID patterns for interpreting group-level change in EORTC QLQ-C30 scores across nine cancer types.

METHODS

Data were obtained from 21 published EORTC Phase III trials that enroled 13,015 patients across nine cancer types (brain, colorectal, advanced breast, head/neck, lung, mesothelioma, melanoma, ovarian, and prostate). Anchor-based MIDs for within-group change and between-group differences in change over time were obtained via mean change method and linear regression, respectively. Separate MIDs were estimated for improvements and deteriorations. Distribution-based estimates were derived and compared with anchor-based MIDs.

RESULTS

Anchor-based MIDs mostly ranged from 5 to 10 points. Differences in MIDs for improvement vs deterioration, for both within-group and between-group, were mostly within a 2-points range. Larger differences between within-group and between-group MIDs were observed for several scales in ovarian, lung and head/neck cancer. Most anchor-based MIDs ranged between 0.3 SD and 0.5 SD distribution-based estimates.

CONCLUSIONS

Our results reinforce recent claims that no single MID can be applied to all EORTC QLQ-C30 scales and disease settings. MIDs varied by scale, improvement/deterioration, within/between comparisons and by cancer type. Researchers applying commonly used rules of thumb must be aware of the risk of dismissing changes that are clinically meaningful or underpowering analyses when smaller MIDs apply.

Single-arm studies involving patient-reported outcome data in oncology: a literature review on current practice

Patient-reported outcomes (PROs) are increasingly used in single-arm cancer studies. We reviewed 60 papers published between 2018 and 2021 of single-arm studies of cancer treatment with PRO data for current practice on design, analysis, reporting, and interpretation. We further examined the studies' handling of potential bias and how they informed decision making. Most studies (58; 97%) analysed PROs without stating a predefined research hypothesis. 13 (22%) of the 60 studies used a PRO as a primary or co-primary endpoint. Definitions of PRO objectives, study population, endpoints, and missing data strategies varied widely. 23 studies (38%) compared the PRO data with external information, most often by using a clinically important difference value; one study used a historical control group. Appropriateness of methods to handle missing data and intercurrent events (including death) were seldom discussed. Most studies (51; 85%) concluded that PRO results supported treatment. Conducting and reporting of PROs in cancer single-arm studies need standards and a critical discussion of statistical methods and possible biases. These findings will guide the Setting International Standards in Analysing Patient-Reported Outcomes and Quality of Life Data in Cancer Clinical Trials-Innovative Medicines Initiative (SISAQOL-IMI) in developing recommendations for the use of PRO-measures in single-arm studies.

Minimally important differences for the EORTC QLQ-C30 in prostate cancer clinical trials

Background

The aim of the study was to estimate the minimally important difference (MID) for interpreting group-level change over time, both within a group and between groups, for the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQ-C30) scores in patients with prostate cancer.

Methods

We used data from two published EORTC trials. Clinical anchors were selected by strength of correlations with QLQ-C30 scales. In addition, clinicians’ input was obtained with regard to plausibility of the selected anchors. The mean change method was applied for interpreting change over time within a group of patients and linear regression models were fitted to estimate MIDs for between-group differences in change over time. Distribution-based estimates were also evaluated.

Results

Two clinical anchors were eligible for MID estimation; performance status and the CTCAE diarrhoea domain. MIDs were developed for 7 scales (physical functioning, role functioning, social functioning, pain, fatigue, global quality of life, diarrhoea) and varied by scale and direction (improvement vs deterioration). Within-group MIDs ranged from 4 to 14 points for improvement and − 13 to − 5 points for deterioration and MIDs for between-group differences in change scores ranged from 3 to 13 for improvement and − 10 to − 5 for deterioration.

Conclusions

Our findings aid the meaningful interpretation of changes on a set of EORTC QLQ-C30 scale scores over time, both within and between groups, and for performing more accurate sample size calculations for clinical trials in prostate cancer.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-021-08609-7.

Establishing anchor-based minimally important differences for the EORTC QLQ-C30 in glioma patients

BACKGROUND

Minimally important differences (MIDs) allow interpretation of the clinical relevance of health-related quality of life (HRQOL) results. This study aimed to estimate MIDs for all European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire Core 30 (QLQ-C30) scales for interpreting group-level results in brain tumor patients.

METHODS

Clinical and HRQOL data from three glioma trials were used. Clinical anchors were selected for each EORTC QLQ-C30 scale, based on correlation (>0.30) and clinical plausibility of association. Changes in both HRQOL and the anchors were calculated, and for each scale and time period, patients were categorized into one of the three clinical change groups: deteriorated by one anchor category, no change, or improved by one anchor category. Mean change method and linear regression were applied to estimate MIDs for interpreting within-group change and between-group differences in change over time, respectively. Distribution-based methods were applied to generate supportive evidence.

RESULTS

A total of 1687 patients were enrolled in the three trials. The retained anchors were performance status and eight Common Terminology Criteria for Adverse Events (CTCAE) scales. MIDs for interpreting within-group change ranged from 4 to 12 points for improvement and -4 to -14 points for deterioration. MIDs for between-group difference in change ranged from 4 to 9 for improvement and -4 to -16 for deterioration. Most anchor-based MIDs were closest to the 0.3 SD distribution-based estimates (range: 3-10).

CONCLUSIONS

MIDs for the EORTC QLQ-C30 scales generally ranged between 4 and 11 points for both within-group mean change and between-group mean difference in change. These results can be used to interpret QLQ-C30 results from glioma trials.

Minimally important differences for interpreting the EORTC QLQ-C30 in patients with advanced colorectal cancer treated with chemotherapy

AIM

The European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQ-C30) assesses the health-related quality of life of patients in cancer trials. There are currently no minimally important difference (MID) guidelines for the EORTC QLQ-C30 for colorectal cancer (CRC). This study aims to estimate MIDs for the EORTC QLQ-C30 scales in patients with advanced CRC treated with chemotherapy and enrolled in clinical trials.

METHOD

The data were obtained from three published EORTC trials that treated CRC patients using chemotherapy. Potential anchors were selected from clinical variables based on their correlation with EORTC QLQ-C30 scales. Anchor-based MIDs for within-group change and between-group change were estimated via the mean change method and linear regression, respectively, and summarized using weighted correlation. Distribution-based MIDs were also examined.

RESULTS

Anchor-based MIDs were determined for deterioration in 8 of the 14 EORTC QLQ-C30 scales and in 9 scales for improvement, and varied by scale, direction of change and anchor. MIDs for improvement (deterioration) ranged from 6 to 18 (-11 to -5) points for within-group change and 5 to 15 (-10 to -4) for between-group change. Summarized MIDs (in absolute values) per scale mostly ranged from 5 to 10 points.

CONCLUSIONS

These findings have clinical relevance for the interpretation of treatment efficacy and the design of clinical trials by informing sample size requirements.

Minimally important differences for interpreting European Organisation for Research and Treatment of Cancer (EORTC) Quality of life Questionnaire core 30 scores in patients with ovarian cancer

INTRODUCTION

Minimal important differences (MIDs) are useful for interpreting changes or differences in health-related quality of life scores in terms of clinical importance. There are currently no MID guidelines for the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire core 30 (EORTC QLQ-C30) specific to ovarian cancer. This study aims to estimate MIDs for interpreting group-level change of EORTC QLQ-C30 scores in ovarian cancer.

METHODS

Data were derived from four EORTC published trials. Clinical anchors for each EORTC QLQ-C30 scale were selected using correlation strength and clinical plausibility. MIDs for within-group change and between-group differences in change over time were estimated via mean change method and linear regression respectively. For each EORTC QLQ-C30 scale, MID estimates from multiple anchors were summarized via weighted-correlation. Distribution-based MIDs were also examined as supportive evidence.

RESULTS

Anchor-based MIDs were determined for deterioration in 7 of the 14 EORTC QLQ-C30 scales assessed, and in 11 scales for improvement. Anchor-based MIDs for within-group change ranged from 4 to 19 (improvement) and - 9 to -4 (deterioration). Between-group MIDs ranged from 3 to 13 (improvement) and - 11 to -4 (deterioration). Generally, absolute anchor-based MIDs for most scales ranged from 4 to 10 points.

CONCLUSIONS

Our findings will aid interpretation of EORTC QLQ-C30 scores in ovarian cancer and inform sample size calculations in future ovarian cancer trials with endpoints that are based on EORTC QLQ-C30 scales.

Minimally important differences for interpreting European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 scores in patients with head and neck cancer

BACKGROUND

We aimed to estimate minimally important difference (MID) for interpreting group-level change over time for European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire core 30 (EORTC QLQ-C30) scores in head and neck cancer.

METHODS

Data were derived retrospectively from two published EORTC trials. Clinical anchors were selected using correlation strength and clinical plausibility of the given anchor/QLQ-C30 scale pair. MIDs for within-group and between-group change were estimated via the mean change method and linear regression, respectively. Distribution-based MIDs were also examined. MIDs for two of the scales, dyspnea and nausea/vomiting, are more uncertain considering their low correlations with the anchors.

RESULTS

Anchor-based MIDs could be determined for deterioration in 7 of the 14 QLQ-C30 scales assessed, and in 3 scales for improvement. MIDs varied by scale, direction of change, and anchor. Absolute MID values ranged from 5 to 15 points for within-group change and 4 to 12 for between-group change. Most MIDs were within 4 to 10 points.

CONCLUSIONS

Our findings, if confirmed, will aid interpreting changes in selected QLQ-C30 scale scores over time and inform sample size calculations in future clinical trials in head and neck cancer.

International standards for the analysis of quality-of-life and patient-reported outcome endpoints in cancer randomised controlled trials: recommendations of the SISAQOL Consortium

Patient-reported outcomes (PROs), such as symptoms, function, and other health-related quality-of-life aspects, are increasingly evaluated in cancer randomised controlled trials (RCTs) to provide information about treatment risks, benefits, and tolerability. However, expert opinion and critical review of the literature showed no consensus on optimal methods of PRO analysis in cancer RCTs, hindering interpretation of results. The Setting International Standards in Analyzing Patient-Reported Outcomes and Quality of Life Endpoints Data Consortium was formed to establish PRO analysis recommendations. Four issues were prioritised: developing a taxonomy of research objectives that can be matched with appropriate statistical methods, identifying appropriate statistical methods for PRO analysis, standardising statistical terminology related to missing data, and determining appropriate ways to manage missing data. This Policy Review presents recommendations for PRO analysis developed through critical literature reviews and a structured collaborative process with diverse international stakeholders, which provides a foundation for endorsement; ongoing developments of these recommendations are also discussed.

Minimally Important Differences for Interpreting EORTC QLQ-C30 Scores in Patients With Advanced Breast Cancer

Background

We aimed to estimate the minimally important difference (MID) for interpreting group-level change over time, both within a group and between groups, for the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire core 30 (EORTC QLQ-C30) scores in patients with advanced breast cancer.

Methods

Data were derived from two published EORTC trials. Clinical anchors (eg, performance status [PS]) were selected using correlation strength and clinical plausibility of their association with a particular QLQ-C30 scale. Three change status groups were formed: deteriorated by one anchor category, improved by one anchor category, and no change. Patients with greater anchor changes were excluded. The mean change method was used to estimate MIDs for within-group change, and linear regression was used to estimate MIDs for between-group differences in change over time. For a given QLQ-C30 scale, MID estimates from multiple anchors were triangulated to a single value via a correlation-based weighted average.

Results

MIDs varied by QLQ-C30 scale, direction (improvement vs deterioration), and anchor. MIDs for within-group change ranged from 5 to 14 points (improvement) and −14 to −4 points (deterioration), and MIDs for between-group change over time ranged from 4 to 11 points and from −18 to −4 points. Correlation-weighted MIDs for most QLQ-C30 scales ranged from 4 to 10 points in absolute values.

Conclusions

Our findings aid interpretation of changes in EORTC QLQ-C30 scores over time, both within and between groups, and for performing more accurate sample size calculations for clinical trials in advanced breast cancer.

Moving forward toward standardizing analysis of quality of life data in randomized cancer clinical trials

Background

There is currently a lack of consensus on how health-related quality of life and other patient-reported outcome measures in cancer randomized clinical trials are analyzed and interpreted. This makes it difficult to compare results across randomized controlled trials (RCTs) synthesize scientific research, and use that evidence to inform product labeling, clinical guidelines, and health policy. The Setting International Standards in Analyzing Patient-Reported Outcomes and Quality of Life Endpoints Data for Cancer Clinical Trials (SISAQOL) Consortium aims to develop guidelines and recommendations to standardize analyses of patient-reported outcome data in cancer RCTs.

Methods and Results

Members from the SISAQOL Consortium met in January 2017 to discuss relevant issues. Data from systematic reviews of the current state of published research in patient-reported outcomes in cancer RCTs indicated a lack of clear reporting of research hypothesis and analytic strategies, and inconsistency in definitions of terms, including “missing data,”“health-related quality of life,” and “patient-reported outcome.” Based on the meeting proceedings, the Consortium will focus on three key priorities in the coming year: developing a taxonomy of research objectives, identifying appropriate statistical methods to analyze patient-reported outcome data, and determining best practices to evaluate and deal with missing data.

Conclusion

The quality of the Consortium guidelines and recommendations are informed and enhanced by the broad Consortium membership which includes regulators, patients, clinicians, and academics.

Dynamic prediction of recurrent events data by landmarking with application to a follow-up study of patients after kidney transplant

This paper extends dynamic prediction by landmarking to recurrent event data. The motivating data comprised post-kidney transplantation records of repeated infections and repeated measurements of multiple markers. At each landmark time point ts, a Cox proportional hazards model with a frailty term was fitted using data of individuals who were at risk at landmark s. This model included the time-updated marker values at ts as time-fixed covariates. Based on a stacked data set that merged all landmark data sets, we considered supermodels that allow parameters to depend on the landmarks in a smooth fashion. We described and evaluated four ways to parameterize the supermodels for recurrent event data. With both the study data and simulated data sets, we compared supermodels that were fitted on stacked data sets that consisted of either overlapping or non-overlapping landmark periods. We observed that for recurrent event data, the supermodels may yield biased estimates when overlapping landmark periods are used for stacking. Using the best supermodel amongst the ones considered, we dynamically estimated the probability to remain infection free between ts and a prediction horizon thor, conditional on the information available at ts.

Validation of prediction models based on lasso regression with multiply imputed data

Background

In prognostic studies, the lasso technique is attractive since it improves the quality of predictions by shrinking regression coefficients, compared to predictions based on a model fitted via unpenalized maximum likelihood. Since some coefficients are set to zero, parsimony is achieved as well. It is unclear whether the performance of a model fitted using the lasso still shows some optimism. Bootstrap methods have been advocated to quantify optimism and generalize model performance to new subjects. It is unclear how resampling should be performed in the presence of multiply imputed data.

Method

The data were based on a cohort of Chronic Obstructive Pulmonary Disease patients. We constructed models to predict Chronic Respiratory Questionnaire dyspnea 6 months ahead. Optimism of the lasso model was investigated by comparing 4 approaches of handling multiply imputed data in the bootstrap procedure, using the study data and simulated data sets. In the first 3 approaches, data sets that had been completed via multiple imputation (MI) were resampled, while the fourth approach resampled the incomplete data set and then performed MI.

Results

The discriminative model performance of the lasso was optimistic. There was suboptimal calibration due to over-shrinkage. The estimate of optimism was sensitive to the choice of handling imputed data in the bootstrap resampling procedure. Resampling the completed data sets underestimates optimism, especially if, within a bootstrap step, selected individuals differ over the imputed data sets. Incorporating the MI procedure in the validation yields estimates of optimism that are closer to the true value, albeit slightly too larger.

Conclusion

Performance of prognostic models constructed using the lasso technique can be optimistic as well. Results of the internal validation are sensitive to how bootstrap resampling is performed.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2288-14-116) contains supplementary material, which is available to authorized users.

A joint model for repeated events of different types and multiple longitudinal outcomes with application to a follow-up study of patients after kidney transplant

This paper presents an extension of the joint modeling strategy for the case of multiple longitudinal outcomes and repeated infections of different types over time, motivated by postkidney transplantation data. Our model comprises two parts linked by shared latent terms. On the one hand is a multivariate mixed linear model with random effects, where a low-rank thin-plate spline function is incorporated to collect the nonlinear behavior of the different profiles over time. On the other hand is an infection-specific Cox model, where the dependence between different types of infections and the related times of infection is through a random effect associated with each infection type to catch the within dependence and a shared frailty parameter to capture the dependence between infection types. We implemented the parameterization used in joint models which uses the fitted longitudinal measurements as time-dependent covariates in a relative risk model. Our proposed model was implemented in OpenBUGS using the MCMC approach.

Living on three time scales: the dynamics of plasma cell and antibody populations illustrated for hepatitis a virus

Understanding the mechanisms involved in long-term persistence of humoral immunity after natural infection or vaccination is challenging and crucial for further research in immunology, vaccine development as well as health policy. Long-lived plasma cells, which have recently been shown to reside in survival niches in the bone marrow, are instrumental in the process of immunity induction and persistence. We developed a mathematical model, assuming two antibody-secreting cell subpopulations (short- and long-lived plasma cells), to analyze the antibody kinetics after HAV-vaccination using data from two long-term follow-up studies. Model parameters were estimated through a hierarchical nonlinear mixed-effects model analysis. Long-term individual predictions were derived from the individual empirical parameters and were used to estimate the mean time to immunity waning. We show that three life spans are essential to explain the observed antibody kinetics: that of the antibodies (around one month), the short-lived plasma cells (several months) and the long-lived plasma cells (decades). Although our model is a simplified representation of the actual mechanisms that govern individual immune responses, the level of agreement between long-term individual predictions and observed kinetics is reassuringly close. The quantitative assessment of the time scales over which plasma cells and antibodies live and interact provides a basis for further quantitative research on immunology, with direct consequences for understanding the epidemiology of infectious diseases, and for timing serum sampling in clinical trials of vaccines.