Testing mapping algorithms of the cancer-specific EORTC QLQ-C30 onto EQ-5D in malignant mesothelioma

Our plain-English summary of Testing mapping algorithms of the cancer-specific EORTC QLQ-C30 onto EQ-5D in malignant mesothelioma, by D.T. Arnold, D. Rowen, M.M. Versteegh, et al (Health and Quality of Life Outcomes, 2015).

The study

• Authors: D.T. Arnold, D. Rowen, M.M. Versteegh, A. Morley, C.E. Hooper, N.A. Maskell
• Journal: Health and Quality of Life Outcomes
• Published: 2015;13:6
• Cited by: 19 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Background: In order to estimate utilities for cancer studies where the EQ-5D was not used, the EORTC QLQ-C30 can be used to estimate EQ-5D using existing mapping algorithms. Several mapping algorithms exist for this transformation, however, algorithms tend to lose accuracy in patients in poor health states. The aim of this study was to test all existing mapping algorithms of QLQ-C30 onto EQ-5D, in a dataset of patients with malignant pleural mesothelioma, an invariably fatal malignancy where no previous mapping estimation has been published.

Methods: Health related quality of life (HRQoL) data where both the EQ-5D and QLQ-C30 were used simultaneously was obtained from the UK-based prospective observational SWAMP (South West Area Mesothelioma and Pemetrexed) trial. In the original trial 73 patients with pleural mesothelioma were offered palliative chemotherapy and their HRQoL was assessed across five time points. This data was used to test the nine available mapping algorithms found in the literature, comparing predicted against observed EQ-5D values. The ability of algorithms to predict the mean, minimise error and detect clinically significant differences was assessed.

Results: The dataset had a total of 250 observations across 5 timepoints. The linear regression mapping algorithms tested generally performed poorly, over-estimating the predicted compared to observed EQ-5D values, especially when observed EQ-5D was below 0.5. The best performing algorithm used a response mapping method and predicted the mean EQ-5D with accuracy with an average root mean squared error of 0.17 (Standard Deviation; 0.22). This algorithm reliably discriminated between clinically distinct subgroups seen in the primary dataset.

Conclusions: This study tested mapping algorithms in a population with poor health states, where they have been previously shown to perform poorly. Further research into EQ-5D estimation should be directed at response mapping methods given its superior performance in this study.

In plain English

This study tested published formulas, called mapping algorithms, that estimate scores on the EQ-5D from answers to a cancer questionnaire, the EORTC QLQ-C30. The EQ-5D is a health questionnaire used in cost-effectiveness assessments, scored so that 1 means full health and 0 means dead. The authors used data from the UK SWAMP study. In it, 73 patients with newly diagnosed malignant pleural mesothelioma who were unsuitable for surgery were offered chemotherapy and completed both questionnaires at up to five time points, giving 250 paired observations. The study did not examine asbestos exposure. Nine algorithms were compared. All eight linear regression algorithms predicted average scores that differed significantly from the observed average. All algorithms over-estimated scores in patients whose observed EQ-5D was below 0.5. The Longworth response mapping algorithm predicted an average of 0.6425 against an observed 0.6572, a difference that was not statistically significant. It also picked up all three clinical subgroup differences seen in the original study, two of them significantly. The authors concluded that it performed best. Limitations they noted included the small dataset and the mostly male population.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Methodological study testing nine published mapping algorithms (QLQ-C30 to EQ-5D) against data from a prospective observational study
• Who was studied: 73 patients with newly diagnosed malignant pleural mesothelioma deemed unsuitable for surgery, from the UK multicentre SWAMP study; 250 paired questionnaire observations across 5 time points; 86% male, mean age 70 (range 49 to 89)
• Exposure: Not an asbestos exposure study. Patients were offered pemetrexed and cisplatin chemotherapy. The study tested nine algorithms for converting QLQ-C30 scores into EQ-5D scores.
• Main results:
  • The patients were on average in poor health: mean EQ-5D score of 0.66, with 18% of observations below 0.5 using the UK tariff
  • Across all 250 observations, every algorithm except Longworth predicted a mean EQ-5D significantly different from the observed mean. For Longworth, the predicted mean was 0.6425 against an observed 0.6572, with p = 0.192 (not significant).
  • All algorithms performed worse in poor health states, over-predicting scores and showing larger errors
  • In the 45 observations with observed EQ-5D below 0.5, the Jang, Longworth and McKenzie algorithms were the most accurate but still over-predicted, by 0.09, 0.17 and 0.13 respectively
  • All 8 linear regression algorithms failed to detect 2 of the 3 clinical subgroup differences seen in the original study. The Longworth algorithm detected all 3, with statistical significance in 2.
  • Both the Longworth and McKenzie algorithms produced score distributions significantly different from the observed distribution: McKenzie p less than 0.001, Longworth p = 0.026
• The authors' conclusion: The authors concluded that the Longworth response mapping algorithm performed best of the 9 tested in this mesothelioma population. They also concluded that further research should be directed at response mapping, while noting that mapping is a second-best solution to collecting EQ-5D directly.
• Limitations the authors state:
  • The dataset had only 250 observations, so the authors did not build their own algorithm
  • The population was mostly male, which could have affected algorithms derived from mainly female breast cancer populations
  • Results were based only on version 3 of the QLQ-C30
  • The choice of best-performing model may be considered arbitrary, because there are no accepted criteria for choosing one
  • Testing each algorithm against its own country's tariff may not reflect real use, where the country tariff is typically fixed

Topics

The National Library of Medicine indexes this study under subjects our site covers: Lung cancer, Mesothelioma.

Medical terms in this study, explained in our Lexicon: Mesothelioma, Pleural Mesothelioma, Quality of Life, Pemetrexed (Alimta), Palliative Care, Chemotherapy.

Related studies

• Health-related quality of life in patients with lung cancer: validation of the Mexican-Spanish version and association with prognosis of the EORTC QLQ-LC13 questionnaire — Ó. Arrieta, C. Núñez-Valencia, et al. (2012)
• Comparison of exome-based HLA class I genotyping tools: identification of platform-specific genotyping errors — K. Kiyotani, T.H. Mai, et al. (2017)
• Identifying patients with suspected lung cancer in primary care: derivation and validation of an algorithm — J. Hippisley-Cox, C. Coupland (2011)
• Comparison of the ultra-low-dose Veo algorithm with the gold standard filtered back projection for detecting pulmonary asbestos-related conditions: a clinical observational study — M. Tekath, F. Dutheil, et al. (2014)
• Current treatment strategies in malignant pleural mesothelioma with a treatment algorithm — M. Sayan, M.F. Eren, et al. (2019)
• Combining a gravitational search algorithm, particle swarm optimization, and fuzzy rules to improve the classification performance of a feed-forward neural network — M.L. Huang, Y.C. Chou (2019)

Browse return to medical research.