| icpower {icensmis} | R Documentation |
Study design in the presence of error-prone diagnostic tests and self-reported outcomes
Description
This function calculates the power and sample in the presence of error-prone diagnostic tests and self-reported outcomes. Two missing mechanisms can be assumed, namely MCAR and NTFP. The MCAR setting assumes that each test is subject to a constant, independent probability of missingness. The NTFP mechanism includes two types of missingness - (1) incorporates a constant, independent, probability of missing for each test prior to the first positive test result; and (2) all test results after first positive are missing.
Usage
icpower(
HR,
sensitivity,
specificity,
survivals,
N = NULL,
power = NULL,
rho = 0.5,
alpha = 0.05,
pmiss = 0,
pcensor = 0,
design = "MCAR",
negpred = 1
)
Arguments
HR |
hazard ratio under the alternative hypothesis. |
sensitivity |
the sensitivity of test. |
specificity |
the specificity of test |
survivals |
a vector of survival function at each test time for baseline(reference) group. Its length determines the number of tests. |
N |
a vector of sample sizes to calculate corresponding powers. If one needs to calculate sample size, then set to NULL. |
power |
a vector of powers to calculate corresponding sample sizes. If one needs to calculate power, then set to NULL. |
rho |
proportion of subjects in baseline(reference) group. |
alpha |
type I error. |
pmiss |
a value or a vector (must have same length as survivals) of the probabilities of each test being randomly missing at each test time. If pmiss is a single value, then each test is assumed to have an identical probability of missingness. |
pcensor |
a value or a vector (must have same length as survivals) of the interval probabilities of censoring time at each interval, assuming censoring process is independent on other missing mechanisms. If it is the single value, then we assume same interval probabilities as the value. The sum of pcensor (or pcensor * length(survivals) if it is single value) must be <= 1. For example, if pcensor = c(0.1, 0.2), then it means the the probabilities of censoring time in first and second intervals are 0.1, 0.2, and the probability of not being censored is 0.7. |
design |
missing mechanism: "MCAR" or "NTFP". |
negpred |
baseline negative predictive value, i.e. the probability of being truely disease free for those who were tested (reported) as disease free at baseline. If baseline screening test is perfect, then negpred = 1. |
Details
To calculate sample sizes for a vector of powers, set N = NULL. To calculate powers for a vector of sample sizes, set power = NULL. One and only one of power and N should be specified, and the other set to NULL. This function uses an enumeration algorithm to calculate the expected Fisher information matrix. The expected Fisher information matrix is used to obtain the variance of the coefficient corresponding to the treatment group indicator.
Value
result: a data frame with calculated sample size and power
I1 and I2: calculated unit Fisher information matrices for each group, which can be used to calculate more values of sample size and power for the same design without the need to enumerate again
Note
When diagnostic test is perfect, i.e. sensitivity=1 and
specificity=1, use icpowerpf instead to obtain significantly
improved computational efficiency.
See Also
Examples
## First specificy survivals. Assume test times are 1:8, with survival function
## at the end time 0.9
surv <- exp(log(0.9)*(1:8)/8)
## Obtain power vs. N
pow1 <- icpower(HR = 2, sensitivity = 0.55, specificity = 0.99, survivals = surv,
N = seq(500, 1500, 50), power = NULL, rho = 0.5, alpha = 0.05,
pmiss = 0, design = "MCAR", negpred = 1)
plot(pow1$result$N, pow1$result$power, type="l", xlab="N", ylab="power")
## Calculate sample size, assuming desired power is 0.9
pow2 <- icpower(HR = 2, sensitivity = 0.55, specificity = 0.99, survivals = surv,
N = NULL, power = 0.9, rho = 0.5, alpha = 0.05, pmiss = 0, design = "MCAR",
negpred = 1)
## When missing test is present with MCAR
pow3 <- icpower(HR = 2, sensitivity = 0.55, specificity = 0.99, survivals = surv,
N = NULL, power = 0.9, rho = 0.5, alpha = 0.05, pmiss = 0.4, design = "MCAR",
negpred = 1)
## When missing test is present with NTFP
pow4 <- icpower(HR = 2, sensitivity = 0.55, specificity = 0.99, survivals = surv,
N = NULL, power = 0.9, rho = 0.5, alpha = 0.05, pmiss = 0.4, design = "NTFP",
negpred = 1)
## When baseline misclassification is present
pow5 <- icpower(HR = 2, sensitivity = 0.55, specificity = 0.99, survivals = surv,
N = NULL, power = 0.9, rho = 0.5, alpha = 0.05, pmiss = 0, design = "MCAR",
negpred = 0.98)
## When test is perfect and no missing test
pow6 <- icpower(HR = 2, sensitivity = 1, specificity = 1, survivals = surv,
N = NULL, power = 0.9, rho = 0.5, alpha = 0.05, pmiss = 0, design = "MCAR",
negpred = 1)
## Different missing probabilities at each test time
pow7 <- icpower(HR = 2, sensitivity = 0.55, specificity = 0.99, survivals = surv,
N = NULL, power = 0.9, rho = 0.5, alpha = 0.05, pmiss = seq(0.1, 0.8, 0.1),
design = "MCAR")