User Tools
Sidebar
**This is an old revision of the document!**
Table of Contents
- Primary Outcomes
- Secondary Outcomes
- Reported Result
- “When compared with AC alone, CDT had lower mortality but high major bleeding and numerically higher ICH”
- “The risk of morality and ICH was high with ST when compared with CDT.
- Findings were similar when analysis was restricted to intermediate risk PE.
Problems
The Definition of Risk Groups is not Stated
- Uses “intermediate risk,” “high risk”, and “intermediate-high risk,” thus mixing terminologies
- 2019 ESC: low, intermediate-low, intermediate-high, high
- 2011 AHA: massive, sub-massive, low risk
- 2016 CHEST: low high, PE without hypotension, PE with hypotension
Very few RCT patients got CDT
| Total Papers (n=45) | ||
|---|---|---|
| patient_type | number | percent |
| AC | 19976 | 24.4% |
| CDT | 9610 | 11.8% |
| ST | 52119 | 63.8% |
| total | 81705 | NA |
| Intermediate-Risk Papers (n=20) | ||
|---|---|---|
| patienttype^number^percent^ |AC|8873|75.9%| |CDT|1929|16.5%| |ST|883|7.5%| |total|11685|14.3% (of $n{total}$) | ||
| RCT Trials Only (n=17) | ||
|---|---|---|
| patienttype^number^percent^ |AC|1101|49.8%| |CDT|78|3.5%| |ST|1031|46.7%| |total|2210|2.7% (of $n{total}$) | ||
This means that the number of CDT patients from RCTs is only $\frac{n{CDT}}{n{total}}=\frac{78}{81611}=0.096\%$ of the study total!!
ULTIMA trial (2013) was only CDT RCT looked at, and $N = 59 (n = [30,29])$
TATED (2021 in India), CDT vs ST ($N = 50$).
CANARY (2022 in Iran), CDT vs AC ($N = 94$)
The Primary Outcome is not reported correctly, given likely intransitivity
The paper utilized a network meta-analysis (1,2,3).
They list that ”[t]he primary analysis compared CDT and systemic fibrinolysis with AC alone.“
However, they combine RCTs, prospective, and retrospective studies, raising serious questions of intransitivity.
Statistical Issues
No attempts to control family-wise error rate
They had to change their statistical analysis strategy
Interestingly, they do NOT report p values for their efficacy outcome – just 95% CI.
Publication inconsistency for their efficacy outcome was significant ($p = 0.036$), but there was no inconsistency at the loop level using a loop inconsistency plot.
Thus, they had to perform a direct meta-analysis. For this analysis, they reported p values (?!). Why would they only report p-values for a “backup” analysis method.
