April 2002, Model Operation Review
April 30 - May 1, 2002 FIFRA SAP Meeting
Cumulative and Aggregate Risk Evaluation System (CARES)
Model Operation Review
Documentation and Operation
CARES Documentation in the form of a User Guide, Technical Manual, and Code Manual were provided to the SAP to assist in their review of the CARES software.
Question 1: The User Guide provides installation instructions and 5 case study tutorials to illustrate operational features of the CARES software. Is the Technical Guide complete and understandable? Were Panel members able to load the software on to their computers and complete some or all of the case study tutorials?
Question 2: The Technical Manual provides an overview of the CARES model and detailed descriptions of key model components. Is the Technical Manual complete and understandable, and are the descriptions of specific model components scientifically sound?
Question 3: The Code Manual provides annotated code for select risk assessment algorithms used in the model. Do the algorithms in the annotated code perform the functions defined in the CARES Technical Manual?
Question 4: The key data sources used in CARES include the US Census /PUMS and the CSFII/FCID. Is the methodology used by the CARES Population Generator to create the CARES Reference Population based on appropriate consideration of demographics factors (vector of individual characteristics) and statistical representativeness?
Question 5: The CARES reference population constructs a 365-day dietary profile by matching similar individuals in the CSFII database using specific matching rules. Does the CARES approach provide a reasonable and realistic construct with respect to temporal variability in magnitude and frequency of food consumption?
Question 6: What can the panel say about how these databases were used together in estimating dietary exposure? Are there other publicly available sources of relevant data? What research opportunities would support a refined calendar based, probabilistic exposure and risk analyses using CARES?
Question 7: The CARES residential model uses an Event Allocation Module to create a temporal profile of residential-related product use occurrences, and includes a product co-occurrence matrix that can be derived from available survey instruments. Does the CARES approach provide a reasonable and realistic construct with respect to temporal variability in magnitude and frequency of residential use and the likelihood of exposure co-occurrence events?
Question 8: Case studies described in the Users' Manual and provided on the CARES - CD illustrate the complexity of conducting Monte Carlo simulations in a PC environment ( multiple parameters for multiple chemicals, where dose and risk from multiple vectors of exposure are calculated over 365 days). Does the SAP believe that an iterative testing strategy with CARES could be developed (conduct of multiple simulations with progressive refinement once risk drivers have been identified) that will permit the user to obtain information on source contribution necessary for decision making?
Question 9: What types of contribution/sensitivity analysis
are recommended by the panel to be most useful in making scientifically-sound
regulatory decisions for one or more pesticides and their associated
agricultural, professional and consumer uses? What should be routinely
reported as part of a CARES assessment with respect to inputs and outputs?
Are there certain key graphs and tables that should be reported? What
types of model evaluation steps does the panel recommend to further
refine and advance models such as CARES?