review the reliability and validity of PEDro scale

PEDro scale信、效度

回顧三個信、效度驗證研究，PEDro scale在總分層級具有可接受的評分者間信度，以及良好的建構以及收歛效度。但在個別題目層級，評分者間信度的結果在不同研究之間不太一致。有的研究發現大部份的題目具有良好的評分者間信度，但有的研究則發現評分者間信度大多不佳。

11 raters on 25 RCT randomly selected from English language RCTs in the PEDro database.
- those raters knew they were in a experiment for reliability.
item level: kappa = 0.12 ~ 0.73
total score level: ICC(1,1) = .55 (95% CI =.41, .72)

120篇RTC都會經過兩輪的評分。第一輪的評分中，每一篇會先由2位評分者獨自評分（individual rating）。若有不一致之處，再由第3位評分者（consensus rater）進行評分，以形成最終的共識評分（consensus rating）。並非所有的120篇都由相同的2位評分者進行評分，而是由25位評分者評分，每一個人分到1 ~ 56篇（平均一人分到13.8篇）。第二輪則由另外一組評分者（7位）進行獨自評分，若有不一致之處再由第3位評分者進行評分。
individual rating reliability: 由二輪評分中的4組獨自評分結果計算kappa
consensus rating reliability: 由二輪評分的consensus rating結果計算kappa
- Consensus raters were 4 of the authors (CGM, CS, RDH, and AMM) and 2 research assistants who developed the PEDro scale and maintain the PEDro database.
on 120 RCT randomly selected from English language RCTs in the PEDro database.
item level
total score level:
- individual rating: ICC(1,1) = .56 (95% CI.47, .65)
- consensus rating: ICC(1,1) = .68 (95% CI.57, .76)

conclusion: The PEDro scale was valid and reliable for assessing methodological quality in pharmaceutical trials.
本研究根據先前回顧研究所蒐集到的53篇有關疼痛治療的藥物研究進行PEDro scale 的信效度分析。
評分者：2位獨立評分員
inter-rater reliability
- total score: ICC = 0.80 (95% CI 0.68 ~ 0.88)
- item:
convergent validity: moderate convergence
- Spearman's correlation with Cochrane risk of bias tool, r = 0.61 (95% CI 0.46e0.72)
construct validity

conclusion: Interrater reliability of PEDro score of RCTs published in predatory journals (unacceptable level) is lower than that of trials published in non-predatory journals (acceptable level), likely also due to ambiguous language and incomplete reporting.
sample: 298 RCTs
- 這也是一個次級資料分析。RCT是來自先前的回顧性研究，比較掠奪性期刊文章和非掠奪性期刊文章的方法學品質。
PEDro rater: 2 independent raters
inter-rater reliability
- (不分期刊類型)
- （區分期刊類型）