Content-balancing strategy in bifactor computerized adaptive patient-reported outcome measurement

Applying multidimensional computerized adaptive testing to the MSQOL-54: a simulation study

BACKGROUND

The Multiple Sclerosis Quality of Life-54 (MSQOL-54) is one of the most commonly-used MS-specific health-related quality of life (HRQOL) measures. It is a multidimensional, MS-specific HRQOL inventory, which includes the generic SF-36 core items, supplemented with 18 MS-targeted items. Availability of an adaptive short version providing immediate item scoring may improve instrument usability and validity. However, multidimensional computerized adaptive testing (MCAT) has not been previously applied to MSQOL-54 items. We thus aimed to apply MCAT to the MSQOL-54 and assess its performance.

METHODS

Responses from a large international sample of 3669 MS patients were assessed. We calibrated 52 (of the 54) items using bifactor graded response model (10 group factors and one general HRQOL factor). Then, eight simulations were run with different termination criteria: standard errors (SE) for the general factor and group factors set to different values, and change in factor estimates from one item to the next set at < 0.01 for both the general and the group factors. Performance of the MCAT was assessed by the number of administered items, root mean square difference (RMSD), and correlation.

RESULTS

Eight items were removed due to local dependency. The simulation with SE set to 0.32 (general factor), and no SE thresholds (group factors) provided satisfactory performance: the median number of administered items was 24, RMSD was 0.32, and correlation was 0.94.

CONCLUSIONS

Compared to the full-length MSQOL-54, the simulated MCAT required fewer items without losing precision for the general HRQOL factor. Further work is needed to add/integrate/revise MSQOL-54 items in order to make the calibration and MCAT performance efficient also on group factors, so that the MCAT version may be used in clinical practice and research.

Patient-reported outcome measures for physical function in cancer patients: content comparison of the EORTC CAT Core, EORTC QLQ-C30, SF-36, FACT-G, and PROMIS measures using the International Classification of Functioning, Disability and Health

BACKGROUND

Patient-reported physical function (PF) is a key endpoint in cancer clinical trials. Using complex statistical methods, common metrics have been developed to compare scores from different patient-reported outcome (PRO) measures, but such methods do not account for possible differences in questionnaire content. Therefore, the aim of our study was a content comparison of frequently used PRO measures for PF in cancer patients.

METHODS

Relying on the framework of the International Classification of Functioning, Disability and Health (ICF) we categorized the item content of the physical domains of the following measures: EORTC CAT Core, EORTC QLQ-C30, SF-36, PROMIS Cancer Item Bank for Physical Function, PROMIS Short Form for Physical Function 20a, and the FACT-G. Item content was linked to ICF categories by two independent reviewers.

RESULTS

The 118 items investigated were assigned to 3 components ('d - Activities and Participation', 'b - Body Functions', and 'e - Environmental Factors') and 11 first-level ICF categories. All PF items of the EORTC measures but one were assigned to the first-level ICF categories 'd4 - Mobility' and 'd5 - Self-care', all within the component 'd - Activities and Participation'. The SF-36 additionally included item content related to 'd9 - Community, social and civic life' and the PROMIS Short Form for Physical Function 20a also included content related to 'd6 - domestic life'. The PROMIS Cancer Item Bank (v1.1) covered, in addition, two first-level categories within the component 'b - Body Functions'. The FACT-G Physical Well-being scale was found to be the most diverse scale with item content partly not covered by the ICF framework.

DISCUSSION

Our results provide information about conceptual differences between common PRO measures for the assessment of PF in cancer patients. Our results complement quantitative information on psychometric characteristics of these measures and provide a better understanding of the possibilities of establishing common metrics.

Key considerations to reduce or address respondent burden in patient-reported outcome (PRO) data collection

Patient-reported outcomes (PROs) are used in clinical trials to provide evidence of the benefits and risks of interventions from a patient perspective and to inform regulatory decisions and health policy. The collection of PROs in routine practice can facilitate monitoring of patient symptoms; identification of unmet needs; prioritisation and/or tailoring of treatment to the needs of individual patients and inform value-based healthcare initiatives. However, respondent burden needs to be carefully considered and addressed to avoid high rates of missing data and poor reporting of PRO results, which may lead to poor quality data for regulatory decision making and/or clinical care.

The collection of patient-reported outcomes (PROs) may capture patients’ assessments of their health status. Here authors highlight PRO-specific issues that should be considered to minimise respondent burden in clinical trials and routine care.

The patient-reported outcomes measurement information systems (PROMIS®) physical function and its derivative measures in adults: a systematic review of content validity

PURPOSE

This study aims to systematically review and critically appraise the content validity of the adult versions of the Patient-Reported Outcomes Measurement Information System Physical Function (PROMIS-PF) item bank and its derivative measures in any adult population.

METHODS

MEDLINE and EMBASE were searched in October 2021 for studies on measurement properties of PROMIS-PF measures in an adult population. Studies were included if the study described the development of a PROMIS-PF measure or investigated its relevance, comprehensiveness, or comprehensibility. Assessment of the methodological quality of eligible studies, rating of results, and summarizing evidence was performed following the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) methodology for assessing content validity. A modified GRADE approach was used to determine the level of evidence.

RESULTS

Three development studies and eight studies on the content validity of one or more of the PROMIS-PF measures were identified. The methodological quality of most studies was rated doubtful. There was low to high level evidence for sufficient relevance, comprehensiveness, and comprehensibility of most PROMIS-PF measures for healthy seniors and various disease populations. We found low to moderate level evidence for insufficient relevance of PROMIS-PF measures for patients with conditions that affected only one body part, and insufficient comprehensibility of the PROMIS-PF measures for minority elderly.

CONCLUSION

Most PROMIS-PF measures demonstrate sufficient content validity in healthy seniors and various disease populations. However, the quality of this evidence is generally low to moderate, due to limitations in the methodological quality of the studies.

Multidimensional Computerized Adaptive Testing: A Potential Path Toward the Efficient and Precise Assessment of Applied Cognition, Daily Activity, and Mobility for Hospitalized Patients

OBJECTIVE

To develop and evaluate an efficient and precise variable-length functional assessment of applied cognition, daily activity, and mobility to inform mobility preservation and rehabilitation service delivery among hospitalized patients.

DESIGN

A multidimensional item bank tapping into these dimensions was developed, with all items calibrated using a multidimensional graded response model. The items were adaptively selected from the item banks to maximize the test information, and the test ended when a joint stopping rule was satisfied. A simulation study was conducted based on the completed instrument, the Functional Assessment in Acute Care Multidimensional Computerized Adaptive Test (FAMCAT), to compare its measurement precision and efficiency capabilities relative to conventional unidimensional computerized adaptive testing. Precision was measured by the bias and root mean squared error between the estimated and true (ie, simulated) θ estimates, whereas efficiency was measured by average test length. Data were collected by an interviewer reading questions from a tablet computer and entering patients' responses.

SETTING

A large Midwestern hospital.

PARTICIPANTS

A total of 4143 patients hospitalized with medical diagnosis and/or surgical complications, with 2060 in the calibration sample and 2083 in the validation cohort.

INTERVENTION

Not applicable.

RESULTS

Among the 2083 patients in the validation sample, FAMCAT administration required an average of 6 (SD=3.11) minutes. Ninety-six percent had their tests terminated by the standard error rule after responding to an average of 22.05 (SD=7.98) items, whereas 15 were terminated by the change in θ rule, with an average test length of 45.27 (SD=11.49). The remaining 76 responded until reaching the maximum test length of 60 items.

CONCLUSIONS

The FAMCAT has the potential to satisfy the need for structured, frequent, and precise assessment of functional domains among hospitalized patients with medical diagnosis and/or surgical complications. The results are promising and may be informative for others who wish to develop similar instruments when concurrent assessment of correlated domains is required.

Using Machine Learning Methods to Develop a Short Tree-Based Adaptive Classification Test: Case Study With a High-Dimensional Item Pool and Imbalanced Data

This study explores advanced techniques in machine learning to develop a short tree-based adaptive classification test based on an existing lengthy instrument. A case study was carried out for an assessment of risk for juvenile delinquency. Two unique facts of this case are (a) the items in the original instrument measure a large number of distinctive constructs; (b) the target outcomes are of low prevalence, which renders imbalanced training data. Due to the high dimensionality of the items, traditional item response theory (IRT)-based adaptive testing approaches may not work well, whereas decision trees, which are developed in the machine learning discipline, present as a promising alternative solution for adaptive tests. A cross-validation study was carried out to compare eight tree-based adaptive test constructions with five benchmark methods using data from a sample of 3,975 subjects. The findings reveal that the best-performing tree-based adaptive tests yielded better classification accuracy than the benchmark method IRT scoring with optimal cutpoints, and yielded comparable or better classification accuracy than the best benchmark method, random forest with balanced sampling. The competitive classification accuracy of the tree-based adaptive tests also come with an over 30-fold reduction in the length of the instrument, only administering between 3 to 6 items to any individual. This study suggests that tree-based adaptive tests have an enormous potential when used to shorten instruments that measure a large variety of constructs.

Using subdomain-specific item sets affected PROMIS physical function scores differently in cardiology and rheumatology patients

OBJECTIVES

The Patient-Reported Outcomes Measurement Information System (PROMIS) physical function (PF) item bank has been developed to standardize patient-reported PF across medical fields. However, evidence of scoring equivalence across cardiology and rheumatology patients is still missing. Therefore, this study aims to investigate both (1) the extent of disease-related differential item functioning (DIF) and (2) the impact of the disease group on using subdomain-specific item sets for generating PROMIS PF scores in cardiology and rheumatology patients.

STUDY DESIGN AND SETTING

Ordinal regression was used to evaluate DIF between cardiology (n = 201) and rheumatology (n = 200) inpatients. To explore the disease-specific impact of PF subdomains on scoring, we compared scores derived from the full item bank with scores derived from subdomain-specific item sets for each disease group.

RESULTS

DIF was detected in 18 items, predominately from the upper extremity subdomain. When upper extremity items were used, cardiology patients reached systematically higher scores than using the full item bank. Rheumatology patients scored substantially higher when mobility items were used.

CONCLUSION

Applying the PROMIS PF metric to disease-specific item sets including items from differing subdomains may lead to biased comparisons of PF levels across disease groups. Disease-specific item parameters should be provided for items showing DIF, and subdomain-related content balancing is recommended for scoring the generic PROMIS PF construct.

Application of Dimension Reduction to CAT Item Selection Under the Bifactor Model

Multidimensional computerized adaptive testing (MCAT) based on the bifactor model is suitable for tests with multidimensional bifactor measurement structures. Several item selection methods that proved to be more advantageous than the maximum Fisher information method are not practical for bifactor MCAT due to time-consuming computations resulting from high dimensionality. To make them applicable in bifactor MCAT, dimension reduction is applied to four item selection methods, which are the posterior-weighted Fisher D-optimality (PDO) and three non-Fisher information-based methods-posterior expected Kullback-Leibler information (PKL), continuous entropy (CE), and mutual information (MI). They were compared with the Bayesian D-optimality (BDO) method in terms of estimation precision. When both the general and group factors are the measurement objectives, BDO, PDO, CE, and MI perform equally well and better than PKL. When the group factors represent nuisance dimensions, MI and CE perform the best in estimating the general factor, followed by the BDO, PDO, and PKL. How the bifactor pattern and test length affect estimation accuracy was also discussed.

Streamlining the Assessment of Patient-Reported Outcomes in Weight Loss and Body Contouring Patients: Applying Computerized Adaptive Testing to the BODY-Q

BACKGROUND

The BODY-Q is a widely used patient-reported outcome measure of surgical outcomes in weight loss and body contouring patients. Reducing the length of the BODY-Q assessment could overcome implementation barriers in busy clinics. A shorter BODY-Q could be achieved by using computerized adaptive testing, a method to shorten and tailor assessments while maintaining reliability and accuracy. In this study, the authors apply computerized adaptive testing to the BODY-Q and assess computerized adaptive testing performance in terms of item reduction and accuracy.

METHODS

Parameters describing the psychometric properties of 138 BODY-Q items (i.e., questions) were derived from the original validation sample (n = 734). The 138 items are arranged into 18 scales reflecting Appearance, Quality of Life, and Experience of Care domains. The authors simulated 1000 administrations of the computerized adaptive testing until a stopping rule, reflecting assessment accuracy of standard error less than 0.55, was met. The authors describe the reduction of assessment length in terms of the mean and range of items administered. The authors assessed accuracy by determining correlation between full test and computerized adaptive testing scores.

RESULTS

The authors ran 54 simulations. Mean item reduction was 36.9 percent (51 items; range, 48 to 138 items). Highest item reduction was achieved for the Experience of Care domain (56.2 percent, 22.5 items). Correlation between full test scores and the BODY-Q computerized adaptive test scores averaged 0.99.

CONCLUSIONS

Substantial item reduction is possible by using BODY-Q computerized adaptive testing. Reduced assessment length using BODY-Q computerized adaptive testing could reduce patient burden while preserving the accuracy of clinical patient-reported outcomes for patients undergoing weight loss and body contouring operations.

Measurement versus prediction in the construction of patient-reported outcome questionnaires: can we have our cake and eat it?

BACKGROUND

Two important goals when using questionnaires are (a) measurement: the questionnaire is constructed to assign numerical values that accurately represent the test taker's attribute, and (b) prediction: the questionnaire is constructed to give an accurate forecast of an external criterion. Construction methods aimed at measurement prescribe that items should be reliable. In practice, this leads to questionnaires with high inter-item correlations. By contrast, construction methods aimed at prediction typically prescribe that items have a high correlation with the criterion and low inter-item correlations. The latter approach has often been said to produce a paradox concerning the relation between reliability and validity [1-3], because it is often assumed that good measurement is a prerequisite of good prediction.

OBJECTIVE

To answer four questions: (1) Why are measurement-based methods suboptimal for questionnaires that are used for prediction? (2) How should one construct a questionnaire that is used for prediction? (3) Do questionnaire-construction methods that optimize measurement and prediction lead to the selection of different items in the questionnaire? (4) Is it possible to construct a questionnaire that can be used for both measurement and prediction?

ILLUSTRATIVE EXAMPLE

An empirical data set consisting of scores of 242 respondents on questionnaire items measuring mental health is used to select items by means of two methods: a method that optimizes the predictive value of the scale (i.e., forecast a clinical diagnosis), and a method that optimizes the reliability of the scale. We show that for the two scales different sets of items are selected and that a scale constructed to meet the one goal does not show optimal performance with reference to the other goal.

DISCUSSION

The answers are as follows: (1) Because measurement-based methods tend to maximize inter-item correlations by which predictive validity reduces. (2) Through selecting items that correlate highly with the criterion and lowly with the remaining items. (3) Yes, these methods may lead to different item selections. (4) For a single questionnaire: Yes, but it is problematic because reliability cannot be estimated accurately. For a test battery: Yes, but it is very costly. Implications for the construction of patient-reported outcome questionnaires are discussed.

Some recommendations for developing multidimensional computerized adaptive tests for patient-reported outcomes

PURPOSE

Multidimensional item response theory and computerized adaptive testing (CAT) are increasingly used in mental health, quality of life (QoL), and patient-reported outcome measurement. Although multidimensional assessment techniques hold promises, they are more challenging in their application than unidimensional ones. The authors comment on minimal standards when developing multidimensional CATs.

METHODS

Prompted by pioneering papers published in QLR, the authors reflect on existing guidance and discussions from different psychometric communities, including guidelines developed for unidimensional CATs in the PROMIS project.

RESULTS

The commentary focuses on two key topics: (1) the design, evaluation, and calibration of multidimensional item banks and (2) how to study the efficiency and precision of a multidimensional item bank. The authors suggest that the development of a carefully designed and calibrated item bank encompasses a construction phase and a psychometric phase. With respect to efficiency and precision, item banks should be large enough to provide adequate precision over the full range of the latent constructs. Therefore CAT performance should be studied as a function of the latent constructs and with reference to relevant benchmarks. Solutions are also suggested for simulation studies using real data, which often result in too optimistic evaluations of an item bank's efficiency and precision.

DISCUSSION

Multidimensional CAT applications are promising but complex statistical assessment tools which necessitate detailed theoretical frameworks and methodological scrutiny when testing their appropriateness for practical applications. The authors advise researchers to evaluate item banks with a broad set of methods, describe their choices in detail, and substantiate their approach for validation.

A Multidimensional Computerized Adaptive Short-Form Quality of Life Questionnaire Developed and Validated for Multiple Sclerosis: The MusiQoL-MCAT

The aim was to develop a multidimensional computerized adaptive short-form questionnaire, the MusiQoL-MCAT, from a fixed-length QoL questionnaire for multiple sclerosis.A total of 1992 patients were enrolled in this international cross-sectional study. The development of the MusiQoL-MCAT was based on the assessment of between-items MIRT model fit followed by real-data simulations. The MCAT algorithm was based on Bayesian maximum a posteriori estimation of latent traits and Kullback-Leibler information item selection. We examined several simulations based on a fixed number of items. Accuracy was assessed using correlations (r) between initial IRT scores and MCAT scores. Precision was assessed using the standard error measurement (SEM) and the root mean square error (RMSE).The multidimensional graded response model was used to estimate item parameters and IRT scores. Among the MCAT simulations, the 16-item version of the MusiQoL-MCAT was selected because the accuracy and precision became stable with 16 items with satisfactory levels (r ≥ 0.9, SEM ≤ 0.55, and RMSE ≤ 0.3). External validity of the MusiQoL-MCAT was satisfactory.The MusiQoL-MCAT presents satisfactory properties and can individually tailor QoL assessment to each patient, making it less burdensome to patients and better adapted for use in clinical practice.

Item exposure control for multidimensional computer adaptive testing under maximum likelihood and expected a posteriori estimation

Item bank stratification has been shown to be an effective method for combating item overexposure in both uni- and multidimensional computer adaptive testing. However, item bank stratification cannot guarantee that items will not be overexposed-that is, exposed at a rate exceeding some prespecified threshold. In this article, we propose enhancing stratification for multidimensional computer adaptive tests by combining it with the item eligibility method, a technique for controlling the maximum exposure rate in computerized tests. The performance of the method was examined via a simulation study and compared to existing methods of item selection and exposure control. Also, for the first time, maximum likelihood (MLE) and expected a posteriori (EAP) estimation of examinee ability were compared side by side in a multidimensional computer adaptive test. The simulation suggested that the proposed method is effective in suppressing the maximum item exposure rate with very little loss of measurement accuracy and precision. As compared to MLE, EAP generates smaller mean squared errors of the ability estimates in all simulation conditions.

Psychometrics behind Computerized Adaptive Testing

The paper provides a survey of 18 years' progress that my colleagues, students (both former and current) and I made in a prominent research area in Psychometrics-Computerized Adaptive Testing (CAT). We start with a historical review of the establishment of a large sample foundation for CAT. It is worth noting that the asymptotic results were derived under the framework of Martingale Theory, a very theoretical perspective of Probability Theory, which may seem unrelated to educational and psychological testing. In addition, we address a number of issues that emerged from large scale implementation and show that how theoretical works can be helpful to solve the problems. Finally, we propose that CAT technology can be very useful to support individualized instruction on a mass scale. We show that even paper and pencil based tests can be made adaptive to support classroom teaching.

Development of a cross-cultural item bank for measuring quality of life related to mental health in multiple sclerosis patients

OBJECTIVE

Quality of life (QoL) measurements are considered important outcome measures both for research on multiple sclerosis (MS) and in clinical practice. Computerized adaptive testing (CAT) can improve the precision of measurements made using QoL instruments while reducing the burden of testing on patients. Moreover, a cross-cultural approach is also necessary to guarantee the wide applicability of CAT. The aim of this preliminary study was to develop a calibrated item bank that is available in multiple languages and measures QoL related to mental health by combining one generic (SF-36) and one disease-specific questionnaire (MusiQoL).

METHODS

Patients with MS were enrolled in this international, multicenter, cross-sectional study. The psychometric properties of the item bank were based on classical test and item response theories and approaches, including the evaluation of unidimensionality, item response theory model fitting, and analyses of differential item functioning (DIF). Convergent and discriminant validities of the item bank were examined according to socio-demographic, clinical, and QoL features.

RESULTS

A total of 1992 patients with MS and from 15 countries were enrolled in this study to calibrate the 22-item bank developed in this study. The strict monotonicity of the Cronbach's alpha curve, the high eigenvalue ratio estimator (5.50), and the adequate CFA model fit (RMSEA = 0.07 and CFI = 0.95) indicated that a strong assumption of unidimensionality was warranted. The infit mean square statistic ranged from 0.76 to 1.27, indicating a satisfactory item fit. DIF analyses revealed no item biases across geographical areas, confirming the cross-cultural equivalence of the item bank. External validity testing revealed that the item bank scores correlated significantly with QoL scores but also showed discriminant validity for socio-demographic and clinical characteristics.

CONCLUSION

This work demonstrated satisfactory psychometric characteristics for a QoL item bank for MS in multiple languages. This work may offer a common measure for the assessment of QoL in different cultural contexts and for international studies conducted on MS.