Error when using estimatePeriodPrevalence

[rfherrerac]

Describe the bug
I created my cohorts as mentioned in the package. When using the function estimatePeriodPrevalence
prev <- estimatePeriodPrevalence(cdm = cdm,
denominatorTable = "denominatorfull",
outcomeTable = "outcome",
interval = "years",
fullContribution = FALSE,
completeDatabaseIntervals =TRUE,
outcomeLookbackDays = NULL)

I am getting this error
Error in UseMethod("cohortCount") :
no applicable method for 'cohortCount' applied to an object of class "cdm_reference"

R version 4.2.3 (2023-03-15)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Red Hat Enterprise Linux 8.7 (Ootpa)

other attached packages:
[1] IncidencePrevalence_0.4.1 dbplyr_2.3.2
[3] dplyr_1.1.2 DatabaseConnector_6.1.0
[5] RPostgres_1.4.5 CDMConnector_0.5.1

Additional context
Add any other context about the problem here.

[edward-burn]

Hi @rfherrerac, could you try updating CDMConnector to the latest cran version?

[rfherrerac]

Thanks @edward-burn. I did it. Now the error changed to:
Getting prevalence for analysis 1 of 12
Error in seq.Date(from = s, to = e, by = i) : 'from' must be of length 1

My reference cohort is defined as:
cdm <- generateDenominatorCohortSet(
cdm = cdm,
name = "denominator",
cohortDateRange = c(as.Date("2021-01-01"), as.Date("2023-01-01")),
ageGroup = list(c(0,100),c(0,17), c(18,74),
c(75,100)),
sex = c("Male", "Female", "Both"),
daysPriorHistory = 0,
requirementInteractions=TRUE
)

[edward-burn]

Hmmm sorry about that uninformative error message @rfherrerac. I think it is indicating that it is not finding any full years to estimate period prevalence for (as you have the requirement completeDatabaseIntervals =TRUE). What is your earliest observation period start date and latest observation period end date in your database?

[rfherrerac]

thanks for seeing this. I was checking and in the getPrevalence.R line 74, when doing

start date

start <- studyPop %>%
dplyr::summarise(min(.data$cohort_start_date, na.rm = TRUE)) %>%
dplyr::pull() %>%
as.Date()
Is getting NULL. When removing the na.rm =TRUE and putting na.rm=FALSE works. Any suggestion?

[edward-burn]

Hmm that's odd, what database management system are you using?

[rfherrerac]

redshift

[edward-burn]

Thanks @rfherrerac, looking into this there was a slight change in behaviour since dplyr v1.1 that has caused this problem for redshift. I'm going to do a quick update and release of IncidencePrevalence with a fix for this (I have a fix working locally, but just need to go through a few steps to get the release out).

[rfherrerac]

Super! thanks a lot.

[edward-burn]

@rfherrerac could you please try with the latest version here on github?
remotes::install_github("darwin-eu/IncidencePrevalence")

[rfherrerac]

Perfect! Worked. Thanks a lot!

[rfherrerac]

Perfect! Worked. Thanks a lot!