ASA Connect

Hello,

I have several studies in which the research was suspended due to the closure of university buildings and the stay-at-home mandate. All are protocol-driven. Some have not required participant visits or intervention during this time period. Other have enrolled participants at various stages of weekly visits and intervention. My questions are two.  Let's assume a RCT with participants randomized to 12 weeks of weekly treatments with primary outcome assessed at 12 weeks and 3- and 6-month follow-up for secondary endpoints. The between-groups difference in the longitudinal trajectory of outcomes using linear mixed models for repeated measures is the original statistical plan.

1. How would you recommend analyzing the data of participants for which Covid-19 research suspension halted a) early participation, b) intermediate participation and c) late participation?

2. What language would you recommend for describing this analytic approach in the Methods section of the research report? Bonus question: does anyone think there will be a universally adopted journal editorial policy for describing the statistical analysis of research interrupted by the pandemic?

Dear Donald,

The German regulators have provided a helpful document how to deal with discontinued recruitment, see
https://www.bfarm.de/EN/Drugs/licensing/clinicalTrials/news/CT_COVID19.html;jsessionid=65F152E90375C1559879B3991E891F54.2_cid343
In one of our trials we have taken this document as basis and act as follows:
1. We have implemented a telephone interview to collect information about the primary endpoints.
2. At the time when patients come for visits we ask for the same information once again.
3. We will report the agreement.
4. We will provide a description in the flow chart about protocol violations. Corona will most likely be the main reason for a violation of the observation time.
5. The analyses will not be altered because we will be having telephone information about the primary endpoints plus measured endpoints at a later time point. The number of patients with delayed measurements will most likely just be 20 out of approximately 1000, thus a small proportion. And it will affect both treatment groups in the same way.

Ad 1. I am unable to provide a clear answer because you have not described whether the data will be missing or whether the observational time frame will be shifted. If it is shifted you could modify your regression model according to the needs. The advantage is the missing data mechanism because of the pandemic. It is by design!
Ad 2. In the specific study I have mentioned above, we have decided to inform the IRB about our procedure. If the Methods strategy is changed because of Corona, in my opinion it should be described as such. Many of us will experience protocol violations because of the various types of lockdowns implemented.

Andreas

Hello,

I have several studies in which the research was suspended due to the closure of university buildings and the stay-at-home mandate. All are protocol-driven. Some have not required participant visits or intervention during this time period. Other have enrolled participants at various stages of weekly visits and intervention. My questions are two.  Let's assume a RCT with participants randomized to 12 weeks of weekly treatments with primary outcome assessed at 12 weeks and 3- and 6-month follow-up for secondary endpoints. The between-groups difference in the longitudinal trajectory of outcomes using linear mixed models for repeated measures is the original statistical plan.

1. How would you recommend analyzing the data of participants for which Covid-19 research suspension halted a) early participation, b) intermediate participation and c) late participation?

2. What language would you recommend for describing this analytic approach in the Methods section of the research report? Bonus question: does anyone think there will be a universally adopted journal editorial policy for describing the statistical analysis of research interrupted by the pandemic?

Dear Donald,

The German regulators have provided a helpful document how to deal with discontinued recruitment, see
https://www.bfarm.de/EN/Drugs/licensing/clinicalTrials/news/CT_COVID19.html;jsessionid=65F152E90375C1559879B3991E891F54.2_cid343
In one of our trials we have taken this document as basis and act as follows:
1. We have implemented a telephone interview to collect information about the primary endpoints.
2. At the time when patients come for visits we ask for the same information once again.
3. We will report the agreement.
4. We will provide a description in the flow chart about protocol violations. Corona will most likely be the main reason for a violation of the observation time.
5. The analyses will not be altered because we will be having telephone information about the primary endpoints plus measured endpoints at a later time point. The number of patients with delayed measurements will most likely just be 20 out of approximately 1000, thus a small proportion. And it will affect both treatment groups in the same way.

Ad 1. I am unable to provide a clear answer because you have not described whether the data will be missing or whether the observational time frame will be shifted. If it is shifted you could modify your regression model according to the needs. The advantage is the missing data mechanism because of the pandemic. It is by design!
Ad 2. In the specific study I have mentioned above, we have decided to inform the IRB about our procedure. If the Methods strategy is changed because of Corona, in my opinion it should be described as such. Many of us will experience protocol violations because of the various types of lockdowns implemented.

Andreas

Hello,

I have several studies in which the research was suspended due to the closure of university buildings and the stay-at-home mandate. All are protocol-driven. Some have not required participant visits or intervention during this time period. Other have enrolled participants at various stages of weekly visits and intervention. My questions are two.  Let's assume a RCT with participants randomized to 12 weeks of weekly treatments with primary outcome assessed at 12 weeks and 3- and 6-month follow-up for secondary endpoints. The between-groups difference in the longitudinal trajectory of outcomes using linear mixed models for repeated measures is the original statistical plan.

1. How would you recommend analyzing the data of participants for which Covid-19 research suspension halted a) early participation, b) intermediate participation and c) late participation?

2. What language would you recommend for describing this analytic approach in the Methods section of the research report? Bonus question: does anyone think there will be a universally adopted journal editorial policy for describing the statistical analysis of research interrupted by the pandemic?

Hello,

I have several studies in which the research was suspended due to the closure of university buildings and the stay-at-home mandate. All are protocol-driven. Some have not required participant visits or intervention during this time period. Other have enrolled participants at various stages of weekly visits and intervention. My questions are two.  Let's assume a RCT with participants randomized to 12 weeks of weekly treatments with primary outcome assessed at 12 weeks and 3- and 6-month follow-up for secondary endpoints. The between-groups difference in the longitudinal trajectory of outcomes using linear mixed models for repeated measures is the original statistical plan.

1. How would you recommend analyzing the data of participants for which Covid-19 research suspension halted a) early participation, b) intermediate participation and c) late participation?

2. What language would you recommend for describing this analytic approach in the Methods section of the research report? Bonus question: does anyone think there will be a universally adopted journal editorial policy for describing the statistical analysis of research interrupted by the pandemic?