enbs {voi}

Expected net benefit of sampling

Description

Calculates the expected net benefit of sampling for a typical study to inform a health economic evaluation, given estimates of the per-person expected value of sample information, decision population size and study setup and per-participant costs. The optimal sample size for each willingness-to-pay, population size and time horizon is also determined.

Usage

enbs(
  evsi,
  costs_setup,
  costs_pp,
  pop,
  time,
  dis = 0.035,
  smooth = FALSE,
  smooth_df = NULL,
  pcut = 0.05
)

Arguments

evsi

Data frame giving estimates of the expected value of sample information, as returned by evsi. This may contain multiple estimates, one for each sample size and willingness to pay.

costs_setup

Setup costs of the study. This can either be a constant, or a vector of two elements giving a 95% credible interval (with mean defined by the midpoint), or a vector of three elements assumed to define the mean and 95% credible interval.

costs_pp

Per-participant costs of the study, supplied in the same format as cost_setup.

pop

Size of the population who would be affected by the decision.

time

Time horizon over which discounting will be applied.

dis

Discount rate used when converting per-person to population EVSI.

smooth

If TRUE, then the maximum ENBS is determined after fitting a nonparametric regression to the data frame x, which estimates and smooths the ENBS for every integer sample size in the range of x$n. The regression is done using the default settings of gam from the mgcv package.

If this is FALSE, then no smoothing or interpolation is done, and the maximum is determined by searching over the values supplied in x.

smooth_df

Basis dimension for the smooth regression. Passed as the k argument to the s() term in gam. Defaults to 6, or the number of unique sample sizes minus 1 if this is lower. Set to a higher number if you think the smoother does not capture the relation of ENBS to sample size accurately enough.

pcut

Cut-off probability which defines a "near-optimal" sample size. The minimum and maximum sample size for which the ENBS is within pcut (by default 5%) of its maximum value will be determined.

Details

pop,time and dis may be supplied as vectors of different lengths. In that case, the ENBS is calculated for all possible combinations of the values in these vectors.

Value

Data frame with components enbs giving the ENBS, and sd giving the corresponding standard deviation. The rows of the data frame correspond to the rows of evsi, and any n and k are inherited from evsi. Additional columns include:

pce: the probability that the study is cost-effective, i.e. that the ENBS is positive, obtained from a normal distribution defined by the estimate and standard deviation.

enbsmax: The maximum ENBS for each willingness-to-pay k.

nmax: The sample size n at which this maximum is achieved.

A second data frame is returned as the "enbsmax" attribute. This has one row per willingness-to-pay (k), giving the optimal ENBS (enbsmax) the optimal sample size (nmax) and an interval estimate for the optimal sample size (nlower to nupper).

If pop, time or dis were supplied as vectors of more than one element, then additional columns will be returned in these data frames to identify the population, time or discount rate for each ENBS calculation. An index ind is also returned to identify the unique combination that each row refers to.

References

Value of Information for Healthcare Decision Making (CRC Press, eds. Heath, Kunst and Jackson: forthcoming)