|
| Criterium | Issue of Previous Trials | Improvement |
|
| QoL instrument | lack of sensitivity and specificity of older methods | use of validated thyroid-specific methods |
|
| Detectable effect on QoL | small effect size | moderate effect size |
|
| Statistical power | very low | low |
|
| Sample size requirement | very large | large |
|
| Patient selection | selection bias due to inclusion of heterogenous patient groups by etiology and prognosis | inclusion of homogeneous diagnostic categories, use of stratified randomisation |
|
| Proportion of symptomatic patients | dilution of the true effect | randomized controlled designs for subgroups with persistent symptoms |
|
| Treatment-related improvement | healthy control group lacking | inclusion of a healthy control group |
|
| Dose adequacy | TSH targets may be misguided. | Treatment-related altered equilibria have to be considered. |
|
| Response heterogeneity | wide variation in the treatment response | physiologically based categorisation |
|
| Specific confounders | T4 to T3 conversion efficiency | identify conversion issues and apply strata |
|
| Statistical analysis | presence of unknown hierarchies and latent groups | latent class analysis |
|
| Statistical method | amalgamation bias (Simpson’s paradox), disaggregation of within-group and between group effects over time | multilevel models,
cross-over design |
|
| Patient expectancies | expectancy bias from treatment uncertainty in RCTs vs treatment certainty under actual conditions of intended drug use | randomization to randomization probabilities (R2R) adjusting for differences in patient expectancies |
|
| Tissue effects | not addressed by RCTs due to lack of differential markers for organ-specific effects | limited usefulness of surrogate markers, requirement for novel markers |
|
| Actions of non-classical thyroid hormones | not addressed | improvement of assay technology, evaluation as possible additional treatment targets |
|
| Safety profile | not addressed by RCTs | prospective acquisition and analysis of big data, especially from T3 users |
|
| Drug-related issues of LT4 | generally reduced and variable T4 to T3 conversion rates | measuring conversion efficiency and targeted T3 addition |
|
| Drug-related issues of LT3 | pharmacological properties, among others short half-life, high peak levels | slow-release preparations |
|
| Drug-related issues of natural desiccated thyroid extracts | popular choice among patients, but few studies | effective large-scale trials |
|
