Table 2.

Selected methodological issues to consider when interpreting CML studies including PROs

Selected key methodological PRO issues	Brief explanatory text
A priori definition of the objective	It is important to specify, a priori, the specific objective of the PRO analysis, unless explicitly stated the exploratory nature of the study. As PRO measures might often include several scales (each of which measures different health aspects), it is important to state a priori which is the primary outcome of the analysis and which are considered as secondary or exploratory outcomes.
Adequacy of the PRO measurement	As more than one PRO measure can be used, the decision for using one, in place of others, should be guided by the specific purpose of the study. For example, in a RCT setting comparing 2 drugs or treatment modalities, it is important to select highly sensitive PRO measures (that are at least leukemia specific), which are more likely to capture subtle differences between arms. In any case, the availability of a full-length manuscript documenting a number of details of the development process, such as evidence of input from patients in item generation, is crucial.
Mode of PRO administration	As mode of administration can influence patients’ outcome reporting, information regarding the setting of administration (eg, via telephone, Web-based, paper copy in the clinic) should be provided. Details of the setting of PRO administration are to be provided.
Documentation of baseline PRO compliance	In prospective studies where PRO is a secondary end point, the number (or percentage) of patients providing PRO data at baseline (ie, at study start) should be provided. This is necessary to show how the sample included in the PRO analysis is representative of the overall population of the study.
Adequacy of sample size for the specific PRO purpose	Sample size considerations are necessary to interpret statistically significant differences in PROs. The study should have enough power to possibly detect differences either between groups or within a group over time (depending on the specific design).
Documentation of PRO missing data	Lack of information regarding missing PRO data over time (ie, at different time points during the study period) might introduce bias with regard to generalizability of study findings or, if too high, can also reduce the power of statistical analysis.
Statistical handling of PRO missing data over time	Several statistical approaches (eg, imputation techniques) are available to account for missing PRO data. Using available PRO data only (complete case analysis) is a limited statistical approach.
Clinical significance of PRO findings	Where available for the specific PRO measures being used, results should be documented in terms of clinical significance and not only in terms of statistical significance.