Bayesian Power Trial

R
simulation
clinical trial design
bayesian
power calculations
Author

Vitaly Druker

Overview of SPYRAL HTN-ON

Try to recreate power analysis in Böhm et al. (2020) - renal denervation trial that uses this method

  • Specifically the ON-MED group

  • Note that in the table below I transformed the standard errors found in table 3 of the publication for the pilot study.

    parameter_assumptions <- tibble::tribble(
    ~trial,    ~arm,        ~basline_adjusted_mean, ~baseline_adjusted_sd, ~n,
    "pilot",   "treatment", -8.8,                   1.8*sqrt(36),          36,
    "pilot",   "control",   -1.8,                   1.8*sqrt(36),          36,
    "pivotal", "treatment", -6.8,                   12,                    NA,
    "pivotal", "control",   -1.8,                   12,                    NA

)

Weibull discount functyion parameters that were used: Shape \(k = 3\), scale \(\lambda =0.25\)

Interim analyses will happen at 175 and 220 subjects

Treatment effect defined by:

\(\mu = \mu_t - \mu_c\)

Trial success criteria:

\[ P(\mu <0 ) \gt .975 \]

Trial futility is made by imputation of remaining subjects and if

\[ P(\mu <0 ) \lt .05 \]

Trial Performance Characteristics

  • Overall trial pwer to detect treatment difference of -5 was 96%
  • Type I error 3%
  • Power at first and second interim looks was 89% and 94%

Other publications

library(bayesDP) # package that was used in the clinical trial
Loading required package: ggplot2
Loading required package: survival

Haddad et al. (2017) for perspective from device community

specifics about dynamic borrowing Viele et al. (2014)

test

References

Böhm, Michael, Raymond R. Townsend, Kazuomi Kario, David Kandzari, Felix Mahfoud, Michael A. Weber, Roland E. Schmieder, et al. 2020. “Rationale and design of two randomized sham-controlled trials of catheter-based renal denervation in subjects with uncontrolled hypertension in the absence (SPYRAL HTN-OFF MED Pivotal) and presence (SPYRAL HTN-ON MED Expansion) of antihypertensive medications: a novel approach using Bayesian design.” Clinical Research in Cardiology: Official Journal of the German Cardiac Society 109 (3): 289–302. https://doi.org/10.1007/s00392-020-01595-z.
Haddad, Tarek, Adam Himes, Laura Thompson, Telba Irony, and Rajesh Nair. 2017. “Incorporation of Stochastic Engineering Models as Prior Information in Bayesian Medical Device Trials.” Journal of Biopharmaceutical Statistics 27 (6): 1089–1103. https://doi.org/10.1080/10543406.2017.1300907.
Viele, Kert, Scott Berry, Beat Neuenschwander, Billy Amzal, Fang Chen, Nathan Enas, Brian Hobbs, et al. 2014. “Use of historical control data for assessing treatment effects in clinical trials.” Pharmaceutical Statistics 13 (1): 41–54. https://doi.org/10.1002/pst.1589.