Benchmark Dose Software: BMDS vs. PROAST: Are there important advantages/disadvantages to using one over the other?

Are there important advantages/disadvantages to using one over the other?  Are there other software packages (e.g., R-based) that could be used for calculating benchmark doses?
Biological modeling and pharmacokinetics Biostatistics Dose-response modeling
Felix Kluxen
Next to preference, there are some differences in the functionality between BMDS and PROAST. While both are not ideally suited for count data, BMDS can handle it by setting up classes. PROAST can use bootstrapping to derive the BMD confidence interval. A comparable R package is bmd, which uses the very convenient drc package for dose-response modelling, it is comparable to BMDS and PROAST but can additionally handle count data, can derive Wald-based confidence intervals and be used in hierarchal models (see ToxicR (see is the R equivalent of the Bayesian module of BMDS (ie BMDS can also be used for Bayesian statistics), naturally it is quite complex for non-statisticians to set appropriate (and defensible) priors. 
Mohammad Asaduzzaman Chowdhury
The point of departure (POD) from animal toxicology data for use in human health risk assessments has traditionally been determined using the No-Observed-Adverse-Effect-Level (NOAEL) method. This strategy, however, has well-defined constraints, such as stringent reliance on dosage selection, dose spacing, and the sample size of the study from which the crucial effect was found. In addition, the NOAEL method ignores the shape of the dose-response curve as well as other relevant data. The benchmark dose (BMD) technique, which was first offered in the 1980s as an alternative to the NOAEL methodology, solves several of the NOAEL method's shortcomings. It takes into account the form of the dose-response curve and is less reliant on dose selection and spacing. Toxicity tests are used to discover potential key endpoints that are important to human health. Only a small number of animals and a small number of dose levels are evaluated in dose-response data from experimental animal research (see Appendix E – Basic concepts in the BMD approach).
The first constraint (the small number of animals employed) refers to the fact that in a dose-response experiment, only a small number of animals are examined per dose group, resulting in some uncertainty in the mean responses. As a result, rather than considering mean responses, confidence intervals are a preferable means of understanding experimental dose-response data. The goal of the NOAEL method is to discover the largest experimental dose for which no adverse health effects can be (statistically) observed using preset (i.e. tested) doses.
Instead than focusing on predetermined dosages, the BMD strategy seeks to find a dose that corresponds to a predefined response (the benchmark reaction – BMR; i.e. the occurrence or size of an unfavorable health consequence). As a result, by fitting mathematical models to the data, it considers dose-response information. The BMD is the dosage level associated with a specific change in the reaction, as determined by the estimated dose-response curve (the BMR). The BMD's confidence interval takes into account the statistical uncertainty in the estimation.

The BMDL and BMDU are the lower and upper confidence limits, respectively.
K. Kannan
BMDS is normal distribution based approach and PROAS is a lognormal distribution based approach is developing benchmark doses. BMD approach is more relevant although several factors applied in risk assessment are not included in that approach.

Post an Answer

Sign In to Answer