Sex Differences in the Effectiveness of First-Line Tumor Necrosis Factor Inhibitors in Psoriatic Arthritis: Results From the European Spondyloarthritis Research Collaboration Network.

OBJECTIVE: Women with psoriatic arthritis (PsA) may have reduced tumor necrosis factor inhibitor (TNFi) effectiveness compared to men. We examined sex differences in treatment response and retention rates during 24 months of follow-up among patients with PsA initiating their first TNFi. METHODS: Data from patients with PsA across 13 European Spondyloarthritis Research Collaboration Network registries starting their first TNFi were pooled. Logistic regression was used to analyze the association between sex and treatment response using low disease activity (LDA) according to the Disease Activity Score in 28 joints using the C-reactive protein level (DAS28-CRP) (<3.2) at six months as the primary outcome. Analyses were adjusted for age, country, conventional synthetic disease-modifying antirheumatic drug treatment, and TNFi start year. Retention rates were explored using the Kaplan-Meier estimator. RESULTS: We analyzed the treatment response of 7,679 patients with PsA (50% women) with available data on LDA at six months. At baseline, women and men had similar characteristics, including mean DAS28-CRP (women vs men, 4.4 [SD 1.2] vs 4.2 [SD 1.2]), though patient-reported outcome measures were worse in women. At six months, 64% of women and 78% of men had LDA (relative risk [RR] 0.82; 95% confidence interval [CI] 0.80-0.84). This difference was similar after adjustment (RR 0.83; 95% CI 0.81-0.85). TNFi retention rates were evaluated in 17,842 patients with PsA. Women had significantly lower retention rates than men at all time points (women 79%, 64%, and 50% vs men 88%, 77%, and 64% at 6, 12, and 24 months, respectively). CONCLUSION: Despite comparable disease characteristics at baseline, women with PsA have reduced treatment response and retention rates to their first TNFi, highlighting the need to consider sex differences in PsA research and management.

Assessing real-world representativeness of prospective registry cohorts in oncology: insights from patients with esophagogastric cancer.

OBJECTIVES: This study aimed to explore the real-world representativeness of a prospective registry cohort with active accrual in oncology, applying a representativeness metric that is novel to health care. STUDY DESIGN AND SETTING: We used data from the Prospective Observational Cohort Study of Esophageal-Gastric Cancer Patients (POCOP) registry and from the population-based Netherlands Cancer Registry (NCR). We used Representativeness-indicators (R-indicators) and overall survival to investigate the degree to which the POCOP cohort and clinically relevant subgroups were a representative sample compared to the NCR database. Calibration using inverse propensity score weighting was applied to correct differences between POCOP and NCR. RESULTS: The R-indicator of the entire POCOP registry was 0.72 95% confidence interval [0.71, 0.73]. Representativeness of palliative patients was higher than that of potentially curable patients (R-indicator 0.88 [0.85, 0.90] and 0.70 [0.68, 0.71], respectively). Stratification to clinically relevant subgroups based on treatment resulted in higher R-indicators of the respective subgroups. Both after stratification and calibration weighting survival estimates in the POCOP registry were more similar to that in the NCR population. CONCLUSION: This study demonstrated the assessment of real-world representativeness of patients who participated in a prospective registry cohort and showed that real-world representativeness improved when the variability in treatment was accounted for.

Misspecification of confounder-exposure and confounder-outcome associations leads to bias in effect estimates.

BACKGROUND: Confounding is a common issue in epidemiological research. Commonly used confounder-adjustment methods include multivariable regression analysis and propensity score methods. Although it is common practice to assess the linearity assumption for the exposure-outcome effect, most researchers do not assess linearity of the relationship between the confounder and the exposure and between the confounder and the outcome before adjusting for the confounder in the analysis. Failing to take the true non-linear functional form of the confounder-exposure and confounder-outcome associations into account may result in an under- or overestimation of the true exposure effect. Therefore, this paper aims to demonstrate the importance of assessing the linearity assumption for confounder-exposure and confounder-outcome associations and the importance of correctly specifying these associations when the linearity assumption is violated. METHODS: A Monte Carlo simulation study was used to assess and compare the performance of confounder-adjustment methods when the functional form of the confounder-exposure and confounder-outcome associations were misspecified (i.e., linearity was wrongly assumed) and correctly specified (i.e., linearity was rightly assumed) under multiple sample sizes. An empirical data example was used to illustrate that the misspecification of confounder-exposure and confounder-outcome associations leads to bias. RESULTS: The simulation study illustrated that the exposure effect estimate will be biased when for propensity score (PS) methods the confounder-exposure association is misspecified. For methods in which the outcome is regressed on the confounder or the PS, the exposure effect estimate will be biased if the confounder-outcome association is misspecified. In the empirical data example, correct specification of the confounder-exposure and confounder-outcome associations resulted in smaller exposure effect estimates. CONCLUSION: When attempting to remove bias by adjusting for confounding, misspecification of the confounder-exposure and confounder-outcome associations might actually introduce bias. It is therefore important that researchers not only assess the linearity of the exposure-outcome effect, but also of the confounder-exposure or confounder-outcome associations depending on the confounder-adjustment method used.

A progressive three-state model to estimate time to cancer: a likelihood-based approach.

BACKGROUND: To optimize colorectal cancer (CRC) screening and surveillance, information regarding the time-dependent risk of advanced adenomas (AA) to develop into CRC is crucial. However, since AA are removed after diagnosis, the time from AA to CRC cannot be observed in an ethically acceptable manner. We propose a statistical method to indirectly infer this time in a progressive three-state disease model using surveillance data. METHODS: Sixteen models were specified, with and without covariates. Parameters of the parametric time-to-event distributions from the adenoma-free state (AF) to AA and from AA to CRC were estimated simultaneously, by maximizing the likelihood function. Model performance was assessed via simulation. The methodology was applied to a random sample of 878 individuals from a Norwegian adenoma cohort. RESULTS: Estimates of the parameters of the time distributions are consistent and the 95% confidence intervals (CIs) have good coverage. For the Norwegian sample (AF: 78%, AA: 20%, CRC: 2%), a Weibull model for both transition times was selected as the final model based on information criteria. The mean time among those who have made the transition to CRC since AA onset within 50 years was estimated to be 4.80 years (95% CI: 0; 7.61). The 5-year and 10-year cumulative incidence of CRC from AA was 13.8% (95% CI: 7.8%;23.8%) and 15.4% (95% CI: 8.2%;34.0%), respectively. CONCLUSIONS: The time-dependent risk from AA to CRC is crucial to explain differences in the outcomes of microsimulation models used for the optimization of CRC prevention. Our method allows for improving models by the inclusion of data-driven time distributions.

Short-term outcome for high-risk patients after esophagectomy.

Transthoracic esophagectomy (TTE) for esophageal cancer facilitates mediastinal dissection; however, it has a significant impact on cardiopulmonary status. High-risk patients may therefore be better candidates for transhiatal esophagectomy (THE) in order to prevent serious complications. This study addressed short-term outcome following TTE and THE in patients that are considered to have a higher risk of surgery-related morbidity. This population-based study included patients who underwent a curative esophagectomy between 2011 and 2018, registered in the Dutch Upper GI Cancer Audit. The Charlson comorbidity index was used to assign patients to a low-risk (score ≤ 1) and high-risk group (score ≥ 2). Propensity score matching was applied to produce comparable groups between high-risk patients receiving TTE and THE. Primary endpoint was mortality (in-hospital/30-day mortality), secondary endpoints included morbidity and oncological outcomes. Additionally, a matched subgroup analysis was performed, including only cervical reconstructions. Of 5,438 patients, 945 and 431 high-risk patients underwent TTE and THE, respectively. After propensity score matching, mortality (6.3 vs 3.3%, P = 0.050), overall morbidity, Clavien-Dindo ≥ 3 complications, pulmonary complications, cardiac complications and re-interventions were significantly more observed after TTE compared to THE. A significantly higher mortality after TTE with a cervical reconstruction was found compared to THE (7.0 vs. 2.2%, P = 0.020). Patients with a high Charlson comorbidity index predispose for a complicated postoperative course after esophagectomy, this was more outspoken after TTE compared to THE. In daily practice, these outcomes should be balanced with the lower lymph node yield, but comparable positive node count and radicality after THE.

Some Patients Are More Equal Than Others: Variation in Ventilator Settings for Coronavirus Disease 2019 Acute Respiratory Distress Syndrome.

OBJECTIVES: As coronavirus disease 2019 is a novel disease, treatment strategies continue to be debated. This provides the intensive care community with a unique opportunity as the population of coronavirus disease 2019 patients requiring invasive mechanical ventilation is relatively homogeneous compared with other ICU populations. We hypothesize that the novelty of coronavirus disease 2019 and the uncertainty over its similarity with noncoronavirus disease 2019 acute respiratory distress syndrome resulted in substantial practice variation between hospitals during the first and second waves of coronavirus disease 2019 patients. DESIGN: Multicenter retrospective cohort study. SETTING: Twenty-five hospitals in the Netherlands from February 2020 to July 2020, and 14 hospitals from August 2020 to December 2020. PATIENTS: One thousand two hundred ninety-four critically ill intubated adult ICU patients with coronavirus disease 2019 were selected from the Dutch Data Warehouse. Patients intubated for less than 24 hours, transferred patients, and patients still admitted at the time of data extraction were excluded. MEASUREMENTS AND MAIN RESULTS: We aimed to estimate between-ICU practice variation in selected ventilation parameters (positive end-expiratory pressure, Fio2, set respiratory rate, tidal volume, minute volume, and percentage of time spent in a prone position) on days 1, 2, 3, and 7 of intubation, adjusted for patient characteristics as well as severity of illness based on Pao2/Fio2 ratio, pH, ventilatory ratio, and dynamic respiratory system compliance during controlled ventilation. Using multilevel linear mixed-effects modeling, we found significant (p ≤ 0.001) variation between ICUs in all ventilation parameters on days 1, 2, 3, and 7 of intubation for both waves. CONCLUSIONS: This is the first study to clearly demonstrate significant practice variation between ICUs related to mechanical ventilation parameters that are under direct control by intensivists. Their effect on clinical outcomes for both coronavirus disease 2019 and other critically ill mechanically ventilated patients could have widespread implications for the practice of intensive care medicine and should be investigated further by causal inference models and clinical trials.

The important role of cisplatin in the treatment of HPV-positive oropharyngeal cancer assessed by real-world data analysis.

OBJECTIVES: The prognostic advantage of human papillomavirus (HPV)-positive oropharyngeal squamous cell carcinoma (OPSCC) resulted in the initiation of treatment de-intensification studies. Two randomized controlled trials (RCTs) reported inferior survival of HPV-positive OPSCC treated with radiotherapy plus cetuximab compared to standard of care radiotherapy plus cisplatin. In this study we investigated whether the important role of cisplatin in the treatment of HPV-positive OPSCCs would also emerge from causal inference analyses of real-world data. MATERIAL AND METHODS: A retrospective cohort of 263 advanced-stage OPSCC-patients from 5 European clinics was studied, treated with radiotherapy (RT) alone or cisplatin-based chemoradiotherapy (CRT) based on standard clinical indications. Causal inference was applied to adjust for treatment assignment, thereby simulating a randomized setting. Average treatment effect of concurrent cisplatin on overall survival (OS) probability was estimated using Bayesian Additive Regression Trees (BART) and Bayesian logistic regression. RESULTS: Significantly better survival probabilities were found for HPV-positive OPSCC treated with CRT compared to RT alone (3-year OS probability 0.961 versus 0.798, p = 0.008). CONCLUSION: This study using causal inference of retrospective patient data confirms the important role of cisplatin in the treatment of HPV-positive OPSCC. Causal inference analyses of real-world data complements the evidence from the published RCTs.

Association of Cognitive Function Trajectories in Centenarians With Postmortem Neuropathology, Physical Health, and Other Risk Factors for Cognitive Decline.

Importance: Understanding mechanisms associated with prolonged cognitive health in combination with exceptional longevity might lead to approaches to enable successful aging. Objective: To investigate trajectories of cognitive functioning in centenarians across domains, and to examine the association of these trajectories with factors underlying cognitive reserve, physical health, and postmortem levels of Alzheimer disease (AD)-associated neuropathology. Design, Setting, and Participants: This cohort study used neuropsychological test data and postmortem neuropathological reports from Dutch centenarians who were drawn from the 100-plus Study between January 2013 and April 2019. Eligible participants self-reported being cognitively healthy, which was confirmed by a proxy. Data analysis was performed between June 2019 and June 2020. Exposures: Age, sex, APOE ε genotype, factors of cognitive reserve, physical health, and AD-associated neuropathology (ie, amyloid-ß, neurofibrillary tangles, and neuritic plaques). Main Outcomes and Measures: In annual visits (until death or until participation was no longer possible), centenarians underwent an extensive neuropsychological test battery, from which an mean z score of global cognition, memory, executive functions, verbal fluency, visuospatial functions, and attention/processing speed was calculated. Linear mixed models with a random intercept and time as independent variable were used to investigate cognitive trajectories, adjusted for sex, age, education, and vision and hearing capacities. In a second step, linear mixed models were used to associate cognitive trajectories with factors underlying cognitive reserve, physical health at baseline, and AD-associated neuropathology. Results: Of the 1023 centenarians approached, 340 were included in the study. We analyzed 330 centenarians for whom cognitive tests were available at baseline (239 [72.4%] women; median [interquartile range] age of 100.5 [100.2-101.7] years), with a mean (SD) follow-up duration of 1.6 (0.8) years. We observed no decline across investigated cognitive domains, with the exception of a slight decline in memory function (ß, -0.10 SD per year; 95% CI, -0.14 to -0.05 SD; P < .001). Cognitive performance was associated with factors of physical health (eg, higher Barthel index: ß, 0.37 SD per year; 95% CI, 0.24-0.49; P < .001) and cognitive reserve (eg, higher education: ß, 0.41 SD per year; 95% CI, 0.29-0.53; P < .001), but none of these factors were associated with the rate of decline. Neuropathological reports were available for 44 participants. While centenarian brains revealed varying loads of postmortem neuropathological hallmarks of AD, this was not associated with cognitive performance or rate of decline. Conclusions and Relevance: While we observed a slight vulnerability for decline in memory function, centenarians maintained high levels of performance in all other investigated cognitive domains for up to 4 years despite the presence of risk factors of cognitive decline. These findings suggest that mechanisms of resilience may underlie the prolongation of cognitive health until exceptional ages.

Development of a multiomics database for personalized prognostic forecasting in head and neck cancer: The Big Data to Decide EU Project.

BACKGROUND: Despite advances in treatments, 30% to 50% of stage III-IV head and neck squamous cell carcinoma (HNSCC) patients relapse within 2 years after treatment. The Big Data to Decide (BD2Decide) project aimed to build a database for prognostic prediction modeling. METHODS: Stage III-IV HNSCC patients with locoregionally advanced HNSCC treated with curative intent (1537) were included. Whole transcriptomics and radiomics analyses were performed using pretreatment tumor samples and computed tomography/magnetic resonance imaging scans, respectively. RESULTS: The entire cohort was composed of 71% male (1097)and 29% female (440): oral cavity (429, 28%), oropharynx (624, 41%), larynx (314, 20%), and hypopharynx (170, 11%); median follow-up 50.5 months. Transcriptomics and imaging data were available for 1284 (83%) and 1239 (80%) cases, respectively; 1047 (68%) patients shared both. CONCLUSIONS: This annotated database represents the HNSCC largest available repository and will enable to develop/validate a decision support system integrating multiscale data to explore through classical and machine learning models their prognostic role.

Machine learning for the prediction of sepsis: a systematic review and meta-analysis of diagnostic test accuracy.

PURPOSE: Early clinical recognition of sepsis can be challenging. With the advancement of machine learning, promising real-time models to predict sepsis have emerged. We assessed their performance by carrying out a systematic review and meta-analysis. METHODS: A systematic search was performed in PubMed, Embase.com and Scopus. Studies targeting sepsis, severe sepsis or septic shock in any hospital setting were eligible for inclusion. The index test was any supervised machine learning model for real-time prediction of these conditions. Quality of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology, with a tailored Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) checklist to evaluate risk of bias. Models with a reported area under the curve of the receiver operating characteristic (AUROC) metric were meta-analyzed to identify strongest contributors to model performance. RESULTS: After screening, a total of 28 papers were eligible for synthesis, from which 130 models were extracted. The majority of papers were developed in the intensive care unit (ICU, n = 15; 54%), followed by hospital wards (n = 7; 25%), the emergency department (ED, n = 4; 14%) and all of these settings (n = 2; 7%). For the prediction of sepsis, diagnostic test accuracy assessed by the AUROC ranged from 0.68-0.99 in the ICU, to 0.96-0.98 in-hospital and 0.87 to 0.97 in the ED. Varying sepsis definitions limit pooling of the performance across studies. Only three papers clinically implemented models with mixed results. In the multivariate analysis, temperature, lab values, and model type contributed most to model performance. CONCLUSION: This systematic review and meta-analysis show that on retrospective data, individual machine learning models can accurately predict sepsis onset ahead of time. Although they present alternatives to traditional scoring systems, between-study heterogeneity limits the assessment of pooled results. Systematic reporting and clinical implementation studies are needed to bridge the gap between bytes and bedside.

Estimating adjuvant treatment effects in Stage II colon cancer: Comparing the synthesis of randomized clinical trial data to real-world data.

There is an ongoing discussion regarding the impact of adjuvant chemotherapy in Stage II colon cancer. We therefore estimated adjuvant treatment effect in Stage II colon cancer using pooled disease-free survival (DFS) data from randomized clinical trials (RCT approach) and compared this to real-world data (RWD approach) estimates. First, we estimated the treatment effect in RCTs by (i) searching relevant trials reporting DFS data, (ii) generating patient-level data from reported DFS data and (iii) estimating treatment effect in the patient-level data. Second, the treatment effect was estimated in an observational cohort of 1,947 patients provided by the Netherlands Cancer Registry using three propensity score methods; matching, weighting and stratification. In the RCT approach, patient-level data of 4,489 patients (events: 853) were generated from seven trials which compared two of the following treatment arms: control, 5FU/LV or FOLFOX. A Cox model was used to estimate a hazard ratio (HR) of 0.77 (0.43;1.10) for 5FU/LV vs. control and 0.93 (0.72;1.15) for FOLFOX vs. 5FU/LV. In the RWD approach, HRs for any adjuvant treatment vs. control were 0.95 (0.50;1.80), 0.88 (0.24;3.21) and 1.05 (0.04;2.06) using matching, weighting and stratification, respectively. There was no significant difference with the estimates from the RCT approach (interaction test, p > 0.10). The RCT data suggest a clinically relevant benefit of adjuvant chemotherapy in terms of DFS, but the estimate did not reach statistical significance. Stratified analyses are required to evaluate whether treatment effect differs in specific subgroups.

Right Dose Right Now: bedside data-driven personalized antibiotic dosing in severe sepsis and septic shock - rationale and design of a multicenter randomized controlled superiority trial.

BACKGROUND: Antibiotic exposure is often inadequate in critically ill patients with severe sepsis or septic shock and this is associated with worse outcomes. Despite markedly altered and rapidly changing pharmacokinetics in these patients, guidelines and clinicians continue to rely on standard dosing schemes. To address this challenge, we developed AutoKinetics, a clinical decision support system for antibiotic dosing. By feeding large amounts of electronic health record patient data into pharmacokinetic models, patient-specific predicted future plasma concentrations are displayed graphically. In addition, a tailored dosing advice is provided at the bedside in real time. To evaluate the effect of AutoKinetics on pharmacometric and clinical endpoints, we are conducting the Right Dose Right Now multicenter, randomized controlled, two-arm, parallel-group, non-blinded, superiority trial. METHODS: All adult intensive care patients with a suspected or proven infection and having either lactatemia or receiving vasopressor support are eligible for inclusion. Randomization to the AutoKinetics or control group is initiated at the bedside when prescribing at least one of four commonly administered antibiotics: ceftriaxone, ciprofloxacin, meropenem and vancomycin. Dosing advice is available for patients in the AutoKinetics group, whereas patients in the control group receive standard dosing. The primary outcome of the study is pharmacometric target attainment during the first 24 h. Power analysis revealed the need for inclusion of 42 patients per group per antibiotic. Thus, a total of 336 patients will be included, 168 in each group. Secondary pharmacometric endpoints include time to target attainment and fraction of target attainment during an entire antibiotic course. Secondary clinical endpoints include mortality, clinical cure and days free from organ support. Several other exploratory and subgroup analyses are planned. DISCUSSION: This is the first randomized controlled trial to assess the effectiveness and safety of bedside data-driven automated antibiotic dosing advice. This is important as adequate antibiotic exposure may be crucial to treat severe sepsis and septic shock. In addition, the trial could prove to be a significant contribution to clinical pharmacometrics and serve as a stepping stone for the use of big data and artificial intelligence in the field. TRIAL REGISTRATION: Netherlands Trial Register (NTR), NL6501/NTR6689. Registered on 25 August 2017. European Clinical Trials Database (EudraCT), 2017-002478-37. Registered on 6 November 2017.

Wound dehiscences following pre-implant bone augmentation with autogenous iliac crest bone grafts: A retrospective cohort study.

PURPOSE: To evaluate possible risk factors associated with wound dehiscences following pre-implant alveolar bone augmentation with autologous anterior iliac crest bone grafts covered with resorbable collagen membranes or human demineralised bone laminae. MATERIALS AND METHODS: Data of 161 patients who underwent bone augmentation prior to the insertion of dental implants were analysed. The preoperative dental status, locations of alveolar bone augmentation sites and location of wound dehiscences were recorded. Gender, age, smoking, alcohol exposure, and dental and medical histories were reviewed. Information was also collected on the surgeons, augmentation technique, application of a collagen membrane, fixation screw type and suture material. Univariate logistic regression analysis was used to evaluate pre- and perioperative variables as predictors of dehiscences. RESULTS: A total of 42 (26.1%) of the 161 augmented patients developed a wound dehiscence following surgery. Most commonly affected sites were the anterior maxilla, followed by the anterior mandible. Males developed wound dehiscences with higher probability than females (odds ratio female = 0.444; P = 0.025; 95% CI: 0.214 to 0.903). Furthermore, marginal associations (P < 0.10) are found for smoking and an anterior location of the augmentation. Smokers were found to have higher probability of a wound dehiscence (odds ratio 2.089; P = 0.064; 95% CI: 0.957 to 4.500) compared to non-smokers. A posterior location of the augmentation was associated with lower probability of a wound dehiscence (odds ratio 0.188; P = 0.076; 95% CI: 0.035 to 0.802) compared to an anterior location. CONCLUSIONS: Based on this study population, smoking in males seems to be the most important risk factor for the development of wound dehiscences after pre-implant alveolar bone augmentation procedures.

Psychometric properties of the Dutch-Flemish Patient-Reported Outcomes Measurement Information System Pain Behavior item bank in patients with musculoskeletal complaints.

We studied the psychometric properties of the 39-item v1.1 Dutch-Flemish Patient-Reported Outcomes Measurement Information System Pain Behavior item bank in a sample of 1,602 patients with musculoskeletal complaints. We evaluated the assumptions of the underlying item response theory (IRT) model (unidimensionality and local dependency with confirmatory factor analyses), and monotonicity with scalability coefficients). We studied the IRT model fit of all items and estimated the item parameters of the IRT model. Differential item functioning (DIF) was studied for age and gender, and DIF for language was studied as a measure of cross-cultural validity. Confirmatory factor analyses showed suboptimal fit of a unidimensional model, but a bifactor model showed low risk of bias when a unidimensional model was assumed (Omega H = .92, explained common variance of .70). Fifteen item pairs (2%) were locally dependent. Five items showed poor scalability. All items fitted the IRT model; slope parameters ranged from .60 to 2.00, and threshold parameters from -2.05 to 6.80. One item showed DIF for age, 1 item DIF for gender, and 5 items showed DIF for language, but the impact on total scores was low. Our study shows limitations of the Dutch-Flemish Patient-Reported Outcomes Measurement Information System Pain Behavior item bank when used in a primary care population with musculoskeletal complaints. PERSPECTIVE: We studied the psychometric properties of the Dutch-Flemish Patient-Reported Outcomes Measurement Information System Pain Behavior item bank in a large primary care population of patients with musculoskeletal complaints. It showed that the Pain Behavior item bank has limitations when used in this population.

Validation of the Dutch-Flemish PROMIS Pain Interference Item Bank in Patients With Musculoskeletal Complaints.

STUDY DESIGN: A cross-sectional study. OBJECTIVE: The aim of this study was to validate the Dutch-Flemish PROMIS Pain Interference item bank in patients with musculoskeletal complaints. SUMMARY OF BACKGROUND DATA: PROMIS item banks have been developed and validated in the US. They need to be further validated in various patient populations and in different languages. METHODS: One thousand six hundred seventy-seven patients answered the full item bank. A Graded Response Model (GRM) was used to study dimensionality with confirmatory factor analyses and by assessing local independency. Monotonicity was evaluated with Mokken scaling. An Item Response Theory (IRT) model was used to study item fit and to estimate slope and threshold parameters. Differential item functioning (DIF) for language, age, and gender was assessed using ordinal logistic regression analyses. DIF for language was evaluated by comparing our data with a similar US sample. Hypotheses concerning construct validity were tested by correlating item bank-scores with scores on several legacy instruments. RESULTS: The GRM showed suboptimal evidence of unidimensionality in confirmatory factor analysis [Comparative Fit Index (CFI): 0.903, Tucker-Lewis Index (TLI): 0.897, Root Mean Square Error of Approximation (RMSEA): 0.144], and 99 item pairs with local dependence. A bifactor model showed good fit (CFI: 0.964, TLI: 0.961, RMSEA: 0.089), with a high Omega-H (0.97), a high explained common variance (ECV: 0.81), and no local dependence. Sufficient monotonicity was shown for all items (Mokken H(i): 0.367-0.686). The unidimensional IRT model showed good fit (only two items with S-X < 0.001), with slope parameters ranging from 1.00 to 4.27, and threshold parameters ranging from -1.77 to 3.66. None of the items showed DIF for age or gender. One item showed DIF for language. Correlations with legacy instruments were high (Pearson R: 0.53-0.75), supporting construct validity. CONCLUSION: The high omega-H and the high ECV indicate that the item bank could be considered essentially unidimensional. The item bank showed good item fit, good coverage of the pain interference trait, and good construct validity. LEVEL OF EVIDENCE: N/A.

Tree-based Machine Learning Methods for Survey Research.

Predictive modeling methods from the field of machine learning have become a popular tool across various disciplines for exploring and analyzing diverse data. These methods often do not require specific prior knowledge about the functional form of the relationship under study and are able to adapt to complex non-linear and non-additive interrelations between the outcome and its predictors while focusing specifically on prediction performance. This modeling perspective is beginning to be adopted by survey researchers in order to adjust or improve various aspects of data collection and/or survey management. To facilitate this strand of research, this paper (1) provides an introduction to prominent tree-based machine learning methods, (2) reviews and discusses previous and (potential) prospective applications of tree-based supervised learning in survey research, and (3) exemplifies the usage of these techniques in the context of modeling and predicting nonresponse in panel surveys.

Psychometric properties of the PROMIS Physical Function item bank in patients receiving physical therapy.

OBJECTIVES: The Patient-Reported Outcomes Measurement Information System (PROMIS) is a universally applicable set of instruments, including item banks, short forms and computer adaptive tests (CATs), measuring patient-reported health across different patient populations. PROMIS CATs are highly efficient and the use in practice is considered feasible with little administration time, offering standardized and routine patient monitoring. Before an item bank can be used as CAT, the psychometric properties of the item bank have to be examined. Therefore, the objective was to assess the psychometric properties of the Dutch-Flemish PROMIS Physical Function item bank (DF-PROMIS-PF) in Dutch patients receiving physical therapy. DESIGN: Cross-sectional study. SETTING AND PARTICIPANTS: 805 patients >18 years, who received any kind of physical therapy in primary care in the past year, completed the full DF-PROMIS-PF (121 items). METHODS: Unidimensionality was examined by Confirmatory Factor Analysis and local dependence and monotonicity were evaluated. A Graded Response Model was fitted. Construct validity was examined with correlations between DF-PROMIS-PF T-scores and scores on two legacy instruments (SF-36 Health Survey Physical Functioning scale [SF36-PF10] and the Health Assessment Questionnaire Disability-Index [HAQ-DI]). Reliability (standard errors of theta) was assessed. RESULTS: The results for unidimensionality were mixed (scaled CFI = 0.924, TLI = 0.923, RMSEA = 0.045, 1th factor explained 61.5% of variance). Some local dependence was found (8.2% of item pairs). The item bank showed a broad coverage of the physical function construct (threshold-parameters range: -4.28-2.33) and good construct validity (correlation with SF36-PF10 = 0.84 and HAQ-DI = -0.85). Furthermore, the DF-PROMIS-PF showed greater reliability over a broader score-range than the SF36-PF10 and HAQ-DI. CONCLUSIONS: The psychometric properties of the DF-PROMIS-PF item bank are sufficient. The DF-PROMIS-PF can now be used as short forms or CAT to measure the level of physical function of physiotherapy patients.

Harm, benefit and costs associated with low-dose glucocorticoids added to the treatment strategies for rheumatoid arthritis in elderly patients (GLORIA trial): study protocol for a randomised controlled trial.

BACKGROUND: Rheumatoid arthritis (RA) is a chronic inflammatory disease of the joints affecting 1% of the world population. It has major impact on patients through disability and associated comorbidities. Current treatment strategies have considerably improved the prognosis, but recent innovations (especially biologic drugs and the new class of so-called "JAK/STAT inhibitors") have important safety issues and are very costly. Glucocorticoids (GCs) are highly effective in RA, and could reduce the need for expensive treatment with biologic agents. However, despite more than 65 years of clinical experience, there is a lack of studies large enough to adequately document the benefit/harm balance. The result is inappropriate treatment strategies, i.e. both under-use and over-use of GCs, and consequently suboptimal treatment of RA. METHODS: The GLORIA study is a pragmatic multicentre, 2-year, randomised, double-blind, clinical trial to assess the safety and effectiveness of a daily dose of 5 mg prednisolone or matching placebo added to standard of care in elderly patients with RA. Eligible participants are diagnosed with RA, have inadequate disease control (disease activity score, DAS28 ≥ 2.6), and are ≥ 65 years. The primary outcome measures are the time-averaged mean value of the DAS28 and the occurrence of serious adverse events or adverse events of special interest. During the trial, change in antirheumatic therapy is permitted as clinically indicated, except for GCs. Cost-effectiveness and cost-utility are secondary outcomes. The main challenge is the interpretation of the trial result with two primary endpoints and the pragmatic trial design that allows co-interventions. Another challenge is the definition of safety and the relative lack of power to detect differences between treatment groups. We have chosen to define safety as the number of patients experiencing at least one serious adverse event. We also specify a decision tree to guide our conclusion on the balance of benefit and harm, and our methodology to combat potential confounding caused by co-interventions. DISCUSSION: Pragmatic trials minimise impact on daily practice and maximise clinical relevance of the results, but analysis and interpretation of the results is challenging. We expect that the results of this trial are of importance for all rheumatologists who treat elderly patients with RA. TRIAL REGISTRATION: ClinicalTrials.gov, NCT02585258 . Registered on 20 October 2015.

The Dutch-Flemish PROMIS Physical Function item bank exhibited strong psychometric properties in patients with chronic pain.

OBJECTIVE: The objective of this study was to assess the psychometric properties of the Dutch-Flemish Patient-Reported Outcomes Measurement Information System (PROMIS) Physical Function item bank in Dutch patients with chronic pain. STUDY DESIGN AND SETTING: A bank of 121 items was administered to 1,247 Dutch patients with chronic pain. Unidimensionality was assessed by fitting a one-factor confirmatory factor analysis and evaluating resulting fit statistics. Items were calibrated with the graded response model and its fit was evaluated. Cross-cultural validity was assessed by testing items for differential item functioning (DIF) based on language (Dutch vs. English). Construct validity was evaluated by calculation correlations between scores on the Dutch-Flemish PROMIS Physical Function measure and scores on generic and disease-specific measures. RESULTS: Results supported the Dutch-Flemish PROMIS Physical Function item bank's unidimensionality (Comparative Fit Index = 0.976, Tucker Lewis Index = 0.976) and model fit. Item thresholds targeted a wide range of physical function construct (threshold-parameters range: -4.2 to 5.6). Cross-cultural validity was good as four items only showed DIF for language and their impact on item scores was minimal. Physical Function scores were strongly associated with scores on all other measures (all correlations ≤ -0.60 as expected). CONCLUSION: The Dutch-Flemish PROMIS Physical Function item bank exhibited good psychometric properties. Development of a computer adaptive test based on the large bank is warranted.

Disentangling mode-specific selection and measurement bias in social surveys.

A large-scale mixed-mode experiment linked to the Dutch Crime Victimization Survey was conducted in 2011. The experiment consisted of two waves; one wave with random assignment to one of the modes web, paper, telephone and face-to-face, and one follow-up wave to the full sample with interviewer modes only. The objective of the experiment is to estimate total mode effects and more specifically the corresponding mode effect components arising from undercoverage, nonresponse and measurement. In this paper, mode-specific selection and measurement bias are defined, and estimators for the bias terms based on the experimental design are introduced and discussed. The proposed estimators are applied to a number of key survey variables from the Labour Force Survey and the Crime Victimization Survey.