Estimation Techniques

CDS (Cross Design Synthesis)


1. Description

CDS (Cross Design Synthesis) combines randomised clinical trials evidence with evidence from clinical databases or observational data. CDS is intended to improve BR evidence by eliminating biases and complementing the weaknesses of one study design with another's strengths.[1][2][3]

2. Evaluation

2.1 Principle
  • The central idea in CDS is the "extrapolation to empty cellá??where available benefits and risks evidence from one population are used to predict the benefits and risks in a slightly different population ("an empty cellá?? by assuming proportionality in the effects measure.
  • This "empty cell" is a population with both characteristics from the two study designs combined.
  • CDS requires users to go through four main tasks or steps.[1]
  • Uncertainties are modelled and are dependent on the data.
  • Investigators' value judgements play important roles in determining the "value" of and which datasets to be combined.
  • The elaborate tasks and over-reliance on investigatorsá??judgments in CDS could potentially overlook inappropriate data pooling and giving false impression of scientific rigour.[4]
  • Uncertainty arising from indirect use of evidence should also be taken into account having CDS providing a "reasonable first approximation"[5]

2.2 Features
  • CDS focusses on synthesising the evidence, rather than on making quantitative comparison of the outcomes.
  • Multiple criteria of benefits and risks can be modelled from the multiple diverse but complementary data á??which is the main purpose of CDS.[4]
  • Sensitivity analysis is addressed in the second task of CDS á??through the estimation of bias.

2.3 Visualisation
  • Visualising benefit-risk trade-offs from CDS analysis depends on the specific model chosen, therefore is not relevant to be appraised here.

2.4 Assessability and accessibility
  • Based on adjustment for bias, the results of benefit-risk assessment from a cross-design synthesis analysis would be acceptable and attractive to many stakeholders.
  • The actual interpretability and acceptability rely on the specific decision model and parameters use.
  • Heavy data resources from clinical trials and clinical databases may restrict the practicality of CDS when used in real-life decision-making.
  • CDS may also be used with other sources of evidence such as case-control studies may also provide relevant information.
  • Cross-design synthesis helps decision-makers to make better decisions by making the sources of and bias in the evidence transparent, alongside maximising strengths and minimising weaknesses of the evidence.

3. References

[1] GAO/PEMD. Cross design synthesis. A new strategy for medical effectiveness research. Washington, D.C.: United States General Accounting Office / Program Evaluation and Methodology Division; 1992 Mar 17. Report No.: B-244808.
[2] Deal L, Gold BD, Gremse DA, Winter HS, Peters SB, Fraga PD, et al. Age-specific questionnaires distinguish GERD symptom frequency and severity in infants and young children: development and initial validation. J Pediatr Gastroenterol Nutr 2005 Aug;41(2):178-85.
[3] Droitcour J, Silberman G, Chelimsky E. A new form of meta-analysis for combining results from randomized clinical trials and medical-practice databases. Int J Technol Assess Health Care 1993;9(3):440-9.
[4] Sacrist??n JA, Soto J, Galende I, Hylan TR. A review of methodologies for assessing drug effectiveness and a new proposal: randomized database studies. Clinical Therapeutics 1997;19(6):1510-7.
[5] Ades AE, Sutton AJ. Multiparameter evidence synthesis in epidemiology and medical decision-making: current approaches. Journal of the Royal Statistical Society: Series A (Statistics in Society) 2006;169(1):5-35