Barbara Gandek’s research while affiliated with University of Massachusetts Chan Medical School and other places

Background Interpretation of health-related quality of life (QOL) outcomes requires improved methods to control for the effects of multiple chronic conditions (MCC). This study systematically compared legacy and improved method effects of aggregating MCC on the accuracy of predictions of QOL outcomes. Methods Online surveys administered generic physical (PCS) and mental (MCS) QOL outcome measures, the Charlson Comorbidity Index (CCI), an expanded chronic condition checklist (CCC), and individualized QOL Disease-specific Impact Scale (QDIS) ratings in a developmental sample (N = 5490) of US adults. Controlling for sociodemographic variables, regression models compared 12- and 35-condition checklists, mortality vs. population QOL-weighting, and population vs. individualized QOL weighting methods. Analyses were cross-validated in an independent sample (N = 1220) representing the adult general population. Models compared estimates of variance explained (adjusted R ² ) and model fit (AIC) for generic PCS and MCS across aggregation methods at baseline and nine-month follow-up. Results In comparison with sociodemographic-only regression models (MCS R ² = 0.08, PCS = 0.09) and Charlson CCI models (MCS R ² = 0.12, PCS = 0.16), increased variance was accounted for using the 35-item CCC (MCS R ² = 0.22, PCS = 0.31), population MCS/PCS QOL weighting (R ² = 0.31–0.38, respectively) and individualized QDIS weighting (R ² = 0.33 & 0.42). Model R ² and fit were replicated upon cross-validation. Conclusions Physical and mental outcomes were more accurately predicted using an expanded MCC checklist, population QOL rather than mortality CCI weighting, and individualized rather than population QOL weighting for each reported condition. The 3-min combination of CCC and QDIS ratings (QDIS-MCC) warrant further testing for purposes of predicting and interpreting QOL outcomes affected by MCC.

Objective: The objective of this study was to develop the classification system for version of the SF-6D (SF-6Dv2) from the SF-36v2. SF-6Dv2 is an improved version of SF-6D, one of the most widely used generic measures of health for the calculation of quality-adjusted life years. Study design and setting: A 3-step process was undertaken to generate a new classification system: (1) factor analysis to establish dimensionality; (2) Rasch analysis to understand item performance; and (3) tests of differential item function. To evaluate robustness, Rasch analyses were performed in multiple subsets of 2 large cross-sectional datasets from recently discharged hospital patients and online patient samples. Results: On the basis of factor analysis, other psychometric evidence, cross-cultural considerations, and amenability to valuation, the 6-dimension classification used in SF-6D was maintained. SF-6Dv2 resulted in the following modifications to SF-6D: a simpler classification of physical function with clearer separation between levels; a more detailed 5-level description of role limitations; using negative wording to describe vitality; and using pain severity rather than pain interference. Conclusions: The SF-6Dv2 classification system describes more distinct levels of health than SF-6D, changes the descriptions used for a number of dimensions and provides clearer wording for health state valuation. The second stage of the study has developed a utility value set using discrete choice methods so that the measure can be used in health technology assessment. Further work should investigate the psychometric characteristics of the new instrument.

Background: Physical function (PF) is a core patient-reported outcome domain in clinical trials in rheumatic diseases. Frequently used PF measures have ceiling effects, leading to large sample size requirements and low sensitivity to change. In most of these instruments, the response category that indicates the highest PF level is the statement that one is able to perform a given physical activity without any limitations or difficulty. This study investigates whether using an item format with an extended response scale, allowing respondents to state that the performance of an activity is easy or very easy, increases the range of precise measurement of self-reported PF. Methods: Three five-item PF short forms were constructed from the Patient-Reported Outcomes Measurement Information System (PROMIS®) wave 1 data. All forms included the same physical activities but varied in item stem and response scale: format A (“Are you able to …”; “without any difficulty”/“unable to do”); format B (“Does your health now limit you …”; “not at all”/“cannot do”); format C (“How difficult is it for you to …”; “very easy”/“impossible”). Each short-form item was answered by 2217–2835 subjects. We evaluated unidimensionality and estimated a graded response model for the 15 short-form items and remaining 119 items of the PROMIS PF bank to compare item and test information for the short forms along the PF continuum. We then used simulated data for five groups with different PF levels to illustrate differences in scoring precision between the short forms using different item formats. Results: Sufficient unidimensionality of all short-form items and the original PF item bank was supported. Compared to formats A and B, format C increased the range of reliable measurement by about 0.5 standard deviations on the positive side of the PF continuum of the sample, provided more item information, and was more useful in distinguishing known groups with above-average functioning. Conclusions: Using an item format with an extended response scale is an efficient option to increase the measurement range of self-reported physical function without changing the content of the measure or affecting the latent construct of the instrument.

Objective: To investigate whether an extended response scale increases the range of measurement in static PROMIS® Physical Function (PF) measures. Methods: In the course of PROMIS wave 1 data collection, a 5-item PF short form was presented in three different item formats. Two of these formats are used in the PROMIS PF item bank, both utilizing a response scale with five response categories in which the highest PF category indicates performing an activity “without any difficulty” or degree of limitation expressed as “not at all”. In this study, a third item format was presented using six response categories in which the highest levels of function in performing an activity are “easy” and “very easy”. We compared format-specific item information curves between the three versions of the 5-item short form, using a Graded Response Model which included the different versions of the 5-item short form and all remaining items of the PROMIS PF item bank. We additionally simulated response patterns for five groups with different PF levels (“very low” to “very high”; n=10,000/group) to calculate relative validity (RV) coefficients for each format-specific short form indicating the relative power to distinguish between known groups compared to the full item bank (RV=1.0). Results: More than n=8,500 participants responded to a subset of the five items presented in different item formats. Compared to the two standard item formats (difficulty/limitation), the extended six-category format increased the range of measurement by more than 0.5 standard deviations on the latent PF continuum. It also showed higher power to distinguish between groups with “high” and “very high” PF than the difficulty/limitation formats. Conclusions: Using an item format with an extended response scale is a promising and efficient way to increase measurement range in patient-reported PF measures with the potential to reduce ceiling effects.

Background To document the development and evaluation of the Quality of life Disease Impact Scale (QDIS®), a measure that standardizes item content and scoring across chronic conditions and provides a summary, norm-based QOL impact score for each disease. MethodsA bank of 49 disease impact items was constructed from previously-used descriptions of health impact to represent ten frequently-measured quality of life (QOL) content areas and operational definitions successfully utilized in generic QOL surveys. In contrast to health in general, all items were administered with attribution to a specific disease (osteoarthritis, rheumatoid arthritis, angina, myocardial infarction, congestive heart failure, chronic kidney disease (CKD), diabetes, asthma, or COPD). Responses from 5418 adults were analyzed as five disease groups: arthritis, cardiovascular, CKD, diabetes, and respiratory. Unidimensionality, item parameter and scale-level invariance, reliability, validity and responsiveness to change during 9-month follow-up were evaluated by disease group and for all groups combined using multi-group confirmatory factor analysis (MGCFA), item response theory (IRT) and analysis of variance methods. QDIS was normed in an independent chronically ill US population sample (N = 4120). ResultsMGCFA confirmed a 1-factor model, justifying a summary score estimated using equal parameters for each item across disease groups. In support of standardized IRT-based scoring, correlations were very high between disease-specific and standardized IRT item slopes (r = 0.88–0.96), thresholds (r = 0.93–0.99) and person-level scores (r ≥ 0.99). Internal consistency, test-retest and person-level IRT reliability were consistently satisfactory across groups. In support of interpreting QDIS as a disease-specific measure, in comparison with generic measures, QDIS consistently discriminated markedly better across disease severity levels, correlated higher with other disease-specific measures in cross-sectional tests, and was more responsive in comparisons of groups with better, same or worse evaluations of disease-specific outcomes at the 9-month follow-up. Conclusions Standardization of content and scoring across diseases was shown to be justified psychometrically and enabled the first summary measure of disease-specific QOL impact normed in the chronically ill population. This disease-specific approach substantially improves discriminant validity and responsiveness over generic measures and provides a basis for better understanding the relative QOL impact of multiple chronic conditions in research and clinical practice.

Aim: An essential step in the development of patient-reported outcomes (PROs) is the determination of the specific item format, including item content and wording. Using the example of measuring physical function (PF), respective item content can focus on either a person’s capability or on a person’s performance, while respective item wording may either be deficit- or resource-oriented. The aim of this study was to investigate if these aspects of item structure affect item information and resulting person parameters (thetas). Methods: In the course of the PROMIS wave 1 data collection, participants rated their PF status based on five daily activities presented in four different item formats, resulting in a set of 20 items. Three formats aimed at capability (“Are you able to…?”, “How difficult is it for you to…?”, “Does your health now limit you to…?”); one asked for performance (“Over the last 7 days, did you…?”). Using Item-Response Theory (IRT), we applied a Graded Response Model (GRM) and a Generalized Partial Credit Model (GPCM) and compared the format-specific item information curves for each of the five activities. By fitting an IRT-based mixed-effect model, we examined the impact of the item format on the theta estimates. Results: Using an incomplete blocked design, data were used from n=15,721 participants who responded to a subset of the PROMIS PF item bank. Item information was similar for capability item formats with a maximum information between 4 and 9, depending on rated activity and the underlying IRT model. The performance item format led to considerably lower item information with a maximum information of <0.5. Capability formats did not differ in resulting theta estimates, while the performance format systematically underestimated person parameters (p<.0001). The range of high item information along the theta continuum was broader for the GRM, while the maximum information was higher for the GPCM. Conclusions: Different capability item formats used for the assessment of physical function lead to similarly high item information and to equal theta estimations. In contrast, a performance item format does not seem to be appropriate for measuring the PROMIS PF construct.

Assessment of health-related quality of life (HRQoL) is well established in clinical research, but ceiling effects in validated tools might prevent detection of changes in well respondents. Tobacco Quality of Life Impact Tool (TQOLITv1) uses conceptual and psychometric advances to enhance detection of HRQoL changes. In a 6-month, forced-switch study, the German TQOLITv1 was assessed in healthy adult (age 23-55 years) current and matched former-smokers. At baseline, smokers were switched to reduced toxicant prototype (RTP) or conventional cigarette for 6 months. TQOLITv1 responses were collected at baseline, 3 and 6 months from current smokers whilst former smokers completed it at the latter two time points. TQOLITv1 includes SF-36v2 and new smoking-specific, physical and general-health measures. Reliability at baseline was good (Cronbach's coefficient alpha > 0.70) for all measures. The baseline percentage with the best possible score (ceiling effect) for former and current smokers was substantially better for the new physical function than SF-36 physical function measure (35% vs. 59% at ceiling, respectively). New smoking-specific measures discriminated current from former smokers better than general health measures. Smoking-specific symptoms (r = 0.73) were more stable from baseline to 6 months than other measures (r = 0.38-0.54) particularly more than the SF-36 mental component score (r = 0.24). Although both product smoking groups worsened in most HRQoL measures, changes in general and smoking-specific HRQoL impact measures favored RTP smokers. The German TQOLITv1 is sufficiently reliable and valid to assess HRQoL and may be more useful than SF-36v2 in evaluation of interventions in well smoking populations including those consuming RTPs. © The Author 2015. Published by Oxford University Press on behalf of the Society for Research on Nicotine and Tobacco.