A bug in calibrateGroup

[ggamit]

When running CohortMethod with multiple analyses, if there are no negative control outcomes, but a large number (>4) of outcomes of interest, summarizeResults fails with the following error:

Error in optim(c(0, 100), logLikelihoodNullMcmc, logRr = logRr, seLogRr = seLogRr) :
function cannot be evaluated at initial parameters

The reason is that in the function calibrateGroup, called by runCmAnalyses->summarizeResults->calibrateEstimates:
the line:
ncs <- group[group$trueEffectSize == 1 & !is.na(group$seLogRr), ]
is returning a non-empty empty tibble even if there are no negative control outcomes (because NA==1 returns NA and not FALSE).
This causes the condition if (nrow(ncs) >= 5) to be TRUE even though there is no need to compute the systematic error.

it should be corrected, similarly to the next line, to:
ncs <- group[!is.na(group$trueEffectSize) & group$trueEffectSize == 1 & !is.na(group$seLogRr), ]