Does linear equating improve prediction in mapping? Crosswalking MacNew onto EQ-5D-5L value sets

Mapping Studies to Estimate Health-State Utilities From Nonpreference-Based Outcome Measures: A Systematic Review on How Repeated Measurements are Taken Into Account

OBJECTIVES

Mapping algorithms are developed using data sets containing patient responses to a preference-based questionnaire and another health-related quality-of-life questionnaire. When data sets include repeated measurements from the same individuals over time, the assumption of observations' independence, required by standard models, is violated, and standard errors are underestimated. This review aimed to identify how studies deal with methodological challenges of repeated measurements, provide an overview of practice to date, and potential implications for future work.

METHODS

We conducted a systematic literature search of MEDLINE, Cumulative Index to Nursing and Allied Health Literature, specialized databases, and previous systematic reviews. A data template was used to extract, among others, start and target instruments if the data set(s) used for estimation and validation had repeated measurements per patient, used regression techniques, and which (if any) adjustments were made for repeated measurements.

RESULTS

We identified 278 publications developing at least 1 mapping algorithm. Of the 278 publications, 121 used a data set with repeated measurements, among which 92 used multiple time points for estimation, and 39 selected specific time points to have 1 observation per participant. A total of 36 studies did not account for repeated measurements. An adjustment was conducted using cluster-robust standard errors (21), random-effects models (30), generalized estimating equations (7), and other methods (7).

CONCLUSIONS

The inconsistent use of methods to account for interdependent observations in the literature indicates that mapping guidelines should include recommendations on how to deal with repeated measurements, and journals should update their guidelines accordingly.

Generating EQ-5D-5L health utility scores from BASDAI and BASFI: a mapping study in patients with axial spondyloarthritis using longitudinal UK registry data

BACKGROUND

Preference-based health-state utility values (HSUVs), such as the EuroQol five-dimensional questionnaire (EQ-5D-5L), are needed to calculate quality-adjusted life-years (QALYs) for cost-effectiveness analyses. However, these are rarely used in clinical trials of interventions in axial spondyloarthritis (axSpA). In these cases, mapping can be used to predict HSUVs.

OBJECTIVE

To develop mapping algorithms to estimate EQ-5D-5L HSUVs from the Bath Ankylosing Disease Activity Index (BASDAI) and the Bath Ankylosing Spondylitis Functional Index (BASFI).

METHODS

Data from the British Society for Rheumatology Biologics Register in Ankylosing Spondylitis (BSRBR-AS) provided 5122 observations with complete BASDAI, BASFI, and EQ-5D-5L responses covering the full range of disease severity. We compared direct mapping using adjusted limited dependent variable mixture models (ALDVMMs) and optional inclusion of the gap between full health and the next feasible value with indirect response mapping using ordered probit (OPROBIT) and generalised ordered probit (GOPROBIT) models. Explanatory variables included BASDAI, BASFI, and age. Metrics to assess model goodness-of-fit and performance/accuracy included Akaike and Bayesian information criteria (AIC/BIC), mean absolute error (MAE) and root mean square error (RMSE), plotting predictive vs. observed estimates across the range of BASDAI/BASFI and comparing simulated data with the original data set for the preferred/best model.

RESULTS

Overall, the ALDVMM models that did not formally include the gap between full health and the next feasible value outperformed those that did. The four-component mixture models (with squared terms included) performed better than the three-component models. Response mapping using GOPROBIT (no squared terms included) or OPROBIT (with squared terms included) offered the next best performing models after the three-component ALDVMM (with squared terms). Simulated data of the preferred model (ALDVMM with four-components) did not significantly underestimate uncertainty across most of the range of EQ-5D-5L values, however the proportion of data at full health was underrepresented, likely due in part to model fitting on a small number of observations at this point in the actual data (4%).

CONCLUSIONS

The mapping algorithms developed in this study enabled the generation of EQ-5D-5L utilities from BASDAI/BASFI. The indirect mapping equations reported for the EQ-5D-5L facilitate the calculation of the EQ-5D-5L utility scores using other UK and country-specific value sets.

Mapping the FACT-G to EQ-5D-3L utility index in cancer with the Chinese values set

OBJECTIVE

The purpose of this research was to create a function for mapping the cancer-specific instrument (FACT-G) to a preference-based measure (EQ-5D-3L) utility index for health-related quality of life, with utility scores generated using the Chinese value set.

METHOD

A cross-sectional study among 243 Chinese patients with cancer was conducted through EQ-5D-3L and FACT-G questionnaires survey. The EQ-5D-3L utility index values wad predicted based on OLS, GLM, CLAD, and Tobit model regression approaches. The performance and predictive power of each model were also evaluated using r² and adj- r², MAE, RMSE, ICC, and MID. Linear equating was used to avoid regression of the OLS model to mean. The model was validated using a 10-fold cross-validation method.

RESULTS

Among all regression models for the FACT-G, the OLS 5 model predicted mean EQ-5D-3L values the best, in terms of model goodness of fit (r²= 0.6230, r²= 58.93%, MAE = 0.0448, RMSE = 0.0624). The OLS model proved to be the most accurate for the mean, and the linear equating scores were much closer to observed scores.

CONCLUSION

Our results suggest that the best algorithm for FACT-G mapping to EQ-5D-3L utility index is OLS model, based on the survey of Chinese patients with cancer.

Mapping the Memorial Anxiety Scale for Prostate Cancer to the SF-6D

PURPOSE

To create a crosswalk that predicts Short Form 6D (SF-6D) utilities from Memorial Anxiety Scale for Prostate Cancer (MAX-PC) scores.

METHODS

The data come from prostate cancer patients enrolled in the North Carolina Prostate Cancer Comparative Effectiveness & Survivorship Study (NC ProCESS, N = 1016). Cross-sectional data from 12- to 24-month follow-up were used as estimation and validation datasets, respectively. Participants' SF-12 scores were used to generate SF-6D utilities in both datasets. Beta regression mixture models were used to evaluate SF-6D utilities as a function of MAX-PC scores, race, education, marital status, income, employment status, having health insurance, year of cancer diagnosis and clinically significant prostate cancer-related anxiety (PCRA) status in the estimation dataset. Models' predictive accuracies (using mean absolute error [MAE], root mean squared error [RMSE], Akaike information criterion [AIC] and Bayesian information criterion [BIC]) were examined in both datasets. The model with the highest prediction accuracy and the lowest prediction errors was selected as the crosswalk.

RESULTS

The crosswalk had modest prediction accuracy (MAE = 0.092, RMSE = 0.114, AIC = - 2708 and BIC = - 2595.6), which are comparable to prediction accuracies of other SF-6D crosswalks in the literature. About 24% and 52% of predictions fell within ± 5% and ± 10% of observed SF-6D, respectively. The observed mean disutility associated with acquiring clinically significant PCRA is 0.168 (standard deviation = 0.179).

CONCLUSION

This study provides a crosswalk that converts MAX-PC scores to SF-6D utilities for economic evaluation of clinically significant PCRA treatment options for prostate cancer survivors.

Predicting EQ-5D-5L crosswalk from the PROMIS-29 profile for the United Kingdom, France, and Germany

Background

EQ-5D health state utilities (HSU) are commonly used in health economics to compute quality-adjusted life years (QALYs). The EQ-5D, which is country-specific, can be derived directly or by mapping from self-reported health-related quality of life (HRQoL) scales such as the PROMIS-29 profile. The PROMIS-29 from the Patient Reported Outcome Measures Information System is a comprehensive assessment of self-reported health with excellent psychometric properties. We sought to find optimal models predicting the EQ-5D-5L crosswalk from the PROMIS-29 in the United Kingdom, France, and Germany and compared the prediction performances with that of a US model.

Methods

We collected EQ-5D-5L and PROMIS-29 profiles and three samples representative of the general populations in the UK (n = 1509), France (n = 1501), and Germany (n = 1502). We used stepwise regression with backward selection to find the best models to predict the EQ-5D-5L crosswalk from all seven PROMIS-29 domains. We investigated the agreement between the observed and predicted EQ-5D-5L crosswalk in all three countries using various indices for the prediction performance, including Bland–Altman plots to examine the performance along the HSU continuum.

Results

The EQ-5D-5L crosswalk was best predicted in France (nRMSE_FRA = 0.075, nMAE_FRA = 0.052), followed by the UK (nRMSE_UK = 0.076, nMAE_UK = 0.053) and Germany (nRMSE_GER = 0.079, nMAE_GER = 0.051). The Bland–Altman plots show that the inclusion of higher-order effects reduced the overprediction of low HSU scores.

Conclusions

Our models provide a valid method to predict the EQ-5D-5L crosswalk from the PROMIS-29 for the UK, France, and Germany.