Table 3

Information for data extraction and subsequent summary and appraisal

DomainKey items
Source of dataSource of data (eg, cohort, case–control, randomised trial participants, registry data, etc)
ParticipantsParticipant eligibility and recruitment method (eg, location, number of centres, setting, inclusion and exclusion criteria)
Participant description (age, sex, primary VL or relapse case, comorbidities including HIV coinfection)
Details of treatments received
How VL diagnosis is defined (whether consistent for all participants, using serology and/or microscopy, molecular testing, clinical history and physical signs, etc)
Study dates
Outcome(s) to be predictedType of outcome (eg, single or combined endpoints)
Definition and method for measurement of outcome (for example, is mortality disease-specific or all-cause, is cure/initial failure/relapse diagnosed based on clinical symptoms and/or diagnostic testing)
Was the same outcome definition (and method for measurement) used in all patients?
Time of outcome occurrence or summary of duration of follow-up
Was the outcome assessed without knowledge of the candidate predictors (ie, blinded)?
Candidate predictorsNumber and type of predictors (eg, demographics, patient history, physical examination, laboratory parameters, HIV status, disease characteristics, etc)
Definition and method for measurement of candidate predictors (including whether defined and measured in a similar way for all participants)
Timing of predictor measurement (eg, at patient presentation, at diagnosis, at treatment initiation or otherwise)
Handling of predictors in the modelling (eg, continuous, linear, non-linear transformations or categorised)
Sample sizeNumber of participants and number of outcomes/events
Events per candidate predictor
Whether the authors describe a sample size calculation
Missing dataNumber of participants with any missing value (including predictors and outcomes)
Number of participants with missing data for each predictor
Handling of missing data (eg, complete-case analysis, imputation or other methods)
Model developmentModelling method (eg, logistic, survival or other)
Modelling assumptions satisfied
Description of participants that were excluded from the analysis with justification
Method for selection of predictors for inclusion in multivariable modelling (eg, all candidate predictors, preselection based on unadjusted association with the outcome)
Method for selection of predictors during multivariable modelling (eg, full model approach, backward or forward selection) and criteria used (eg, p value, Akaike information criterion)
Shrinkage of predictor weights or regression coefficients (eg, no shrinkage, uniform shrinkage, penalised estimation)
Model performanceCalibration (calibration plot, calibration slope, Hosmer-Lemeshow test), discrimination (C-statistic, D-statistic, log-rank) and overall performance measures with confidence intervals
Classification measures (eg, sensitivity, specificity, predictive values, net reclassification improvement) and whether a priori cut points were used
Model evaluationMethod used for testing model performance: development dataset only (apparent performance, random split of data, resampling methods, eg, bootstrap or cross-validation, none) or separate external validation
For external validations; data source and participants to be described as per ‘source of data’ and ‘participants’ domains. Definitions and distributions (including missing data) of outcome and candidate predictors
In case of poor external validation, whether model was updated or extended (eg, intercept recalibrated, predictor effects adjusted, or new predictors added)
ResultsFinal and other multivariable models presented, including predictor weights or regression coefficients, intercept, baseline survival, model performance measures (with SEs or CIs)
Any alternative presentation of the final prediction models, for example, sum score, nomogram, score chart, predictions for specific risk subgroups with performance
Comparison of the definition and distribution of predictors (including missing data) for development and validation datasets
Interpretation and discussionStudy authors’ interpretation of presented models (intended use, clinical utility, etc)
Study authors’ reported strengths and limitations
MiscellaneousSource of funding/sponsor
Any declared conflicts of interest
Methodological guidelines used
  • Adapted from CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies (CHARMS) and Prediction model Risk Of Bias ASsessment Tool (PROBAST).

  • CI, confidence interval; HIV, human immunodeficiency virus; SE, standard error; VL, visceral leishmaniasis.