Substance conversion

When concentration data at the level of measured substances have to be converted to concentration data at the level of active substances (or perhaps also inactive substances), then substance conversion rules can be specified to provide the rules. This section first describes the basic substance conversion, and then the refinements using available substance authorisations.

For each measured substance in the concentration data, there may be zero or more conversion rules (records in the substance conversion rules data source), each linking to an active or inactive substance. Substance conversion rules may specify a link to an exclusive substance or not. For an exclusive conversion it is assumed that only one substance is present in the sample, therefore the measured substance is considered to be just one of the linked substances. It can also be that measured substances link to one or more exclusive substances plus one (non-exclusive) substance that is considered a metabolite of the other exclusive substances. The metabolite can occur together with any of the exclusive substances. It is assumed that either all conversion rules linked to a measured substance are marked as exclusive (case 1), or precisely one rule is marked as exclusive and the other rules are marked as not exclusive (case 2). If this is not the case for any set of rules linked to a measured substance, then this is regarded as erroneous data.

Four methods are implemented for substance conversion:

1. Allocate most potent (EC 2018 Tier 1): For each measured substance, the linked substances are restricted to the active substances of interest. The concentration of the measured substance is assigned to the most potent active substance in this set. Potency is specified by the relative potency factors. All other candidate active substances are assigned a zero concentration. I.e., the measured substance concentration is allocated for 100% to the most potent substance specified by the conversion rules and for this allocation, the concentration or LOR is multiplied by the molecular weight correction factor. See EC2018 Tier 1.

2. Random allocation (EC 2018 Tier 2): One of the conversion rules is drawn randomly (with equal probability), including the rules of both active and other substances. This drawn rule is used as follows to generate active substance concentrations:

If the drawn conversion rule is marked as exclusive, the concentration or LOR is allocated to the linked substance.

If the drawn conversion rule is marked as not exclusive, a proportion p, specified by the drawn conversion rule, of the concentration or LOR is allocated to the linked substance. The remaining proportion (1-p) is allocated to one other substance, which is the substance that is linked to the measured substance in a conversion rule marked as exclusive (in this case it is assumed that precisely one record per measured substance is marked as exclusive).

All assigned concentrations are multiplied by the molecular weight correction factor. All unselected candidate substances are assigned a zero concentration. See EC2018 Tier 2.

3. Nominal estimate: The substances specified through the conversion rules are allocated with a nominal value based on all possible conversion rules. This may be regarded as the nominal or average allocation value of the random sampling method.

All conversion rules are marked as exclusive: The measured substance concentration is divided over all n active substances specified with equal proportions 1/n, accounting for the molecular weight correction factor for all substances.

Precisely one conversion rule is marked exclusive and n conversion rules are marked as not exclusive: The measured substance concentration is divided over all active substances specified, with a proportion 1/2 + 1/n for the substance belonging to the exclusive conversion rule, and equal proportions 1/n for the other substances, accounting for the molecular weight correction factor for all substances.

4. Allocate all: The concentration of a measured substance is allocated to each active substance associated with the measured substance as if it were the most potent substance. I.e., the same measured substance is allocated to all associated active substances simultaneously. This method is not sensible when using it in a cumulative assessment, but it is of use in substance screening assessments, where in a combined analysis of multiple substances all active substances are considered independently.

Use of substance authorisations in substance conversion

When substance authorisations are available, then these can be used to exclude conversions of measured substances to unauthorised substances on a given food. The information is used as follows in the substance conversion procedures:

1. Allocate most potent: The set of candidate active substances from which the most potent active substance is to be drawn is reduced to only the substances with authorised uses. However, if none of the candidate active substances is authorised, then the most potent of the unauthorised substances is selected for active substance allocation.

2. Random allocation: The set of conversion rules from which to draw is reduced to the rules linking to authorised substances or the non-exclusive substance (thus allowing the selection of a possibly unauthorised metabolite of an authorised substance). If none of the conversion rules links to an authorised substance, then one rule is drawn from the full set of all (unauthorised) conversion rules.

3. Nominal estimate: The set of conversion rules is reduced in the same way as in Tier 2. Nominal calculation is performed on the resulting set of conversion rules.

4. Allocate all: For this method, the same rules apply as for allocate most potent. The set of candidate active substances that are to be allocated is reduced to only the substances with authorised uses. Hence, a substance is not allocated when it is not authorised and there is at least one other candidate active substance that is authorised. However, if none of the candidate active substances is authorised, then the most potent of the unauthorised substances is selected for active substance allocation.

Multiple allocations of the same active substance in one sample

In some datasets substance conversion can cause the same active substance to be allocated multiple times in one sample. For example, when an active substance is measured directly, but also a measurement is recorded for a (measured) substance that converts to the active substance. By default, MCRA does not accept such cases, because often it can be associated with errors in the data. Therefore, an error will be reported with a message

“Unexpected substance translation in sample xxx: substance X is translated from multiple measured substances.”

However, if such cases are known to exist and accepted in the data, then a method is available to fix duplicate substance allocation inconsistencies. If active substance allocation leads to multiple allocated measurements then the following procedure is implemented for resolving these inconsistencies:

If there is any measurement that is positive or zero then: clone one of these records to create the “aggregate measurement record”; to make this selection deterministic, prefer records that have measured-substance equal to active-substance over other records, and take the record with the highest residue. Of this clone, update the measured value with the mean of all positive/zero measurements.
If all allocated active substance measurements are non-detect, then clone one of the non-detects. Here also, records that have measured-substance equal to active-substance are preferred over other records, and then the measurement with the smallest LOR is used.

Note that these rules are quite generic and would work quite well also in case there are many measurements for the same active substance. In practice, one would expect only a few (two).