Need help with network meta-analysis in Stata - Ranking results inconsistency and ranking results

[Ying lii]

Hello everyone,

I hope this message finds you well. I am currently working on a network meta-analysis in Stata and I am facing a perplexing issue. I am seeking your guidance and expertise to help me unravel this confusion.

My study focuses on investigating the comparative effectiveness of various psychosocial interventions for anxiety. During the initial steps of network setup in Stata, I designated "no treatment or waitinglist" as the reference treatment for Standardized Mean Difference (SMD). However, when I conducted the network rank analysis, I discovered that the "no treatment or waitinglist" intervention was ranked fifth, whereas conventionally, it should be ranked as less effective. What's more, I'm encountering inconsistency between the ranking results obtained from the probabilities (%) of each treatment having each rank and the SUCRA (Surface Under the Cumulative Ranking) mean rank.

I have invested considerable effort in reviewing my methodology, examining the extracted data, and validating the analysis process. Regrettably, the results have continued to perplex me, and I am unable to identify the underlying problem.

To provide you with a more comprehensive understanding, here are some additional details:

• I am utilizing Stata for the network meta-analysis.
• In my analysis, I have employed Model 1, and the ranking results of the network rank analysis have generated inconsistencies with the expected ranking.
• I have carefully scrutinized the data extraction process, ensuring accuracy and consistency in the extracted data.

Despite these efforts, I have been unable to pinpoint the root cause of this issue. It is precisely this conundrum that brings me to seek your guidance and advice.

I kindly request your assistance in shedding light on this matter. If you possess experience in network meta-analysis using Stata or have encountered similar challenges before, your insights would be greatly appreciated.

If you require more specific details or code snippets, I am more than willing to provide them. Your assistance will significantly aid me in comprehending and resolving the ranking results inconsistency and the reference treatment issue.

Thank you sincerely for your time and support. I eagerly await your valuable input.

Best regards, Li Ying

[Ying lii]

Code I have used:
network rank min, zero all reps(1000)gen(prob)
sucra prob* ,lab(ABT CBT BT MBI MPT PP SM ST EMDR EFT NTW yoga TAU AP )rankog

[Nick Cox]

This particular forum is for discussing Mata -- Stata's underlying programming language -- not meta.

If you don't get a good reply, I recommend re-posting in the General forum.