| Autor: |
Kevin Veen, A Joseph, F Sossi, P Blancarte Jaber, E Lansac, Z Das-Gupta, S Aktaa, JJM Takkenberg |
| Rok vydání: |
2023 |
| DOI: |
10.21203/rs.3.rs-2252328/v1 |
| Popis: |
Aims Standard outcome sets enable the value-based evaluation of health care delivery. Whereas the attainment of expert opinion has been structured using methods such as the modified-Delphi process, standardized guidelines for extraction of candidate outcome measures from literature are lacking. As such, we aimed to describe a novel methodology to obtain a comprehensive list of candidate outcome measures for potential inclusion in standard outcome datasets. Methods We designed a three-key steps to develop a list of candidate outcome measures to evaluate healthcare and applied these steps for the development of the international consortium of health outcome measures Heart valve disease dataset to illustrate the method. Our methodological approach involves: 1) Benchmark review of relevant registries and Clinical Practice Guidelines; 2) Applying machine learning to screen (using frequent words in abstracts) the studies that have been extracted from a systematic search of the literature using only the disease term; and 3) Extracting the candidate outcome measures from randomly selected batches of the retrieved studies iteratively until saturation is reached. Batch cutoff choices were investigated using data of 1000 simulated cases. Results Simulation showed that on average 98% (range 92% to 100%) saturation is reached using a 100-article batch initially, with 25 articles in the subsequent batches. On average 1.7 repeating rounds (range 1-5) of 25 new articles were necessary to achieve saturation. Conclusion In this paper a standardized three-pillar approach is proposed to identify relevant outcome measures for a standard dataset. This approach creates a balance between comprehensiveness and feasibility in conducting literature reviews for the identification of candidate outcome measures. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
