Setting up a new register for biocide sales in the Netherlands

3. Practical aspects of monitoring biocide sales

This chapter discusses the practical aspects that are relevant for setting up a monitoring system for biocide sales. Conversations with (and specifications of) the Ministry of IenW have been included in this advisory report. The topics that are covered are based on research on important aspects of (meta)data quality and quality indicators for administrative data sources (Daas & Ossen, Metadata quality evaluation of secondary data sources., 2011) (Daas P., et al., 2011). These aspects are in line with most of the elements of question four in the research proposal:
d) What are the starting points for a data registration system (e.g. confidentiality)? In general terms, what software and hardware requirements (one or more intended uses, minimisation of administrative costs, etc.) and what institutional setting would be preferable (independence, private or public, etc.)?

3.1 Purpose and use

3.1.1 Purpose of the register

The purpose of the register is to gain insight into the use of biocides in the Netherlands and the resulting (toxic) burden on humans and the environment. Information about biocide sales is an important link in the learning process. This is because the use of biocides is difficult to measure, but biocide sales (via authorisation holders) are easier to track. Since there is a relationship between sales and actual use, one of the strategies is to develop a sales register. This will enable us to base our priorities and the direction of biocide policies on evaluations of the effects of previously implemented policy measures.

3.1.2 Intended use of the register

The register will be used to determine the quantities of biocides and/or active substances in biocides that are sold on an annual basis in the Netherlands. Data for multiple years will provide policy makers with insight into increases and decreases over time. In the long term, this will also help us make connections with environmental emissions.

3.1.3 Public access to the register

Microdata (data collected at the level of the authorisation holders) are usually considered confidential and are not available to the public. The totals of these figures, which cannot be used to identify individual authorisation holders, may be published, however. These totals represent relevant environmental information, which is in the interest of the general public. A legal basis, taking into account the Århus Convention, will be needed to determine exactly which totals (i.e. which aggregation level) can be published. If the information is competition-sensitive, we could opt for delayed publication, for example after a year or two. Section 3.5 discusses the confidentiality of (micro)data in greater depth.

3.2 Time frame for intended use and outcomes

3.2.1 Frequency of outcome calculations

We recommend adapting the frequency of outcome calculations to the frequency needs of policy makers. The maximum frequency depends on how often new data are submitted to the register. If there is a continual inflow of new information into the register, outcomes can be calculated more frequently. This process should always take into account statistical security (see section 3.5). For the time being, we are assuming a frequency of once a year, with reports covering the preceding concluded year. This will give the authorisation holders enough time to submit data and it gives the register holder enough time to process the data and make corrections if necessary. This frequency is sufficient for monitoring and for policy adjustments.

3.2.2 Publication date

Because the information may be competition-sensitive for authorisation holders, we recommend publishing the calculated totals no sooner than one (or two) years after the end of the period to which the data refer. Biocide sales for 2021 could be published in January 2023 (see also section 3.1.2).

3.2.3 Developments with regard to authorisation holders

As a group, the authorisation holders from the Ctgb and ECHA registers are a good source of information for the register. Tracking product types (PTs) and the start and end date of each authorisation gives an idea of how the market for biocides develops over time for each PT. This also tells us, once it’s time to request data, which (new) authorisation holders need to be approached. Some biocides are authorised for multiple PTs, which can in turn be used in other products as well. To prevent double entries, this will be taken into account in the structure of the different tables in the register and later calculations (see section 3.10.3). In some cases, an expired authorised product may be supplied (and used up) for a certain period after the end date. So it would be useful to approach authorisation holders for (recently) expired authorisations as well when requesting information.

3.3 Communication

3.3.1 Communication between register holder and respondents

The register holder must clearly explain the purpose of the biocide register and specify what it will not be used for. No competition-sensitive information will be shared with other authorisation holders, but the authorisation holders will receive general information, for instance via Platform Biociden, about statistically secure aggregated data. The most suitable mode of communication (e.g. via regular mail, email or telephone) will have to be investigated further and will depend on the budget.

3.3.2 Communication about data quality

The register holder will contact the authorisation holders if the submitted data are of insufficient statistical quality, for example because data are missing or because there are clear indications that the submitted data may be incorrect. In that case, the party in question will either be asked to supply additional data or to check whether the submitted data are correct. There may be additional communication about the situation if necessary. This communication may be orally and in writing, or in writing only for evidence purposes. Whether the register holder is legally authorised to initiate contact by telephone (or in writing) will have to be determined. This only works if there is a legal requirement stipulating who is authorised to initiate contact.

3.3.3 Communication about overdue data

The register holder has an active policy of reminding parties if data are not submitted on time. When reminders are issued and how (by mail, email or orally) will be specified.

3.4 Legal matters

3.4.1 Legal basis

The design of the register will depend on the legal basis, if any. Mandatory reporting may have a positive effect on the quality and quantity of the data. Voluntary reporting entails the risk of methodological challenges such as selective responses and varied response times.

There is a realistic chance that NGOs or the public will ask for the aggregated information to be made public. This is why we recommend stipulating from the start how the information will be made available. A useful starting point at the international level is the biocides regulation (European Parliament, (EU) Regulation no. 528/2012 concerning the making available on the market and use of biocidal products, 2012) and at the national level the Plant Protection Products and Biocides Act, (Wet gewasbeschermingsmiddelen en biociden, 2024). The requirement (based on Article 67 of EC1107/2009) to provide information to the competent authority about the sale of plant protection products may serve as inspiration (European Parliament, (EC) Regulation no. 1107/2009 concerning the placing of plant protection products on the market, 21 October 2009).

3.4.2 Purpose limitation

The purpose of the register is to collect aggregated information about biocides and emphatically not to impose fines. If the latter were the goal, this would not be conducive to the willingness of authorisation holders to provide accurate data. Enshrining the purpose in law will therefore promote data quality and data quantity.

3.5 Confidentiality of the (micro)data

3.5.1 What measures will be taken to protect the confidentiality of the microdata?

The following measures may be taken:

  • It is crucially important that microdata are never emailed (or otherwise communicated) outside the register holder’s IT environment.
  • Ensure that users of the register can only be identified by their own internal ID number and that no other identifying information is visible. The link between the authorisation holder’s number / Chamber of Commerce number and the user’s own ID number will be stored separately (in a different location). This will limit the damage in the event of a leak.
  • The register holder must be an independent, certified party. Further research is needed on the exact certifications that will be required; ISO 9001 and ISO 27001 are two possible certification options.
  • Ensure that the IT environment of the register cannot be accessed by more people than absolutely necessary: specify exactly who will have access. Specify which (direct) employees of the Ministry of IenW will have (remote) access to the data, and which employees of other government or research organisations will have (partial) access.
  • We also recommend specifying which (supervisory) organisations are explicitly not permitted to access the microdata. This is to reassure authorisation holders that the data will not be shared with supervisory authorities.

3.5.2 What measures will be taken to protect the statistical security of the data?

With regard to the security of public aggregated data, it is important to apply statistical security methods, such as non-traceable classification or rounding (Hundepool & de Wolf, 2010). Which of these methods is suitable for the microdata and which for the intended public output will be determined when more specific descriptions are available for both.

It would also be a good idea to think about the intended use of the microdata. Will the biocide data be combined with other data later on for the purpose of further breakdowns? For example, additional data on toxicity and other factors will be needed to make a connection with the environment. This is why it is important that any expansion of data to be included in the register is technically feasible. If there is a need later on to further categorise the authorisation holders (e.g. based on sector or size) in addition to the breakdown into professional and non-professional markets, the statistical security will have to be re-evaluated for each breakdown.

Every breakdown increases the risk of statistical disclosure and must be weighed carefully against the usefulness of the breakdown. The plant protection products register publishes results at the national level, not broken down by region. It is worth considering whether we should use the same approach for the biocide register. Sales in one region are not a guarantee that these products are used in that same region, so regional numbers may lead to false precision. The content of the data will have to be known in order to determine whether breakdown by region is useful.

3.6 Target population

The term target population refers to all units that would ideally be included in the register if we want to determine the total sales figures for biocides in the Netherlands.

3.6.1 Definition of administrative units

The target population of the register consists of all authorisations that apply to the sale of biocides in the Netherlands. There may be cases where one authorisation holder (a company that has been granted authorisation to sell biocides in the Netherlands) has multiple authorisations. An authorisation grants permission for the sale of one or more specific products of an authorisation holder.
This is why it is important that the register includes descriptions of the authorisations (units of the target population with specific products) as well as the authorisation holders and the relationship between the authorisations and the authorisation holder.

The authorisation holders are the appropriate units to ask for information for the purpose of collecting data on the sale of biocides. This is to prevent authorisation holders being contacted separately for each authorisation.

3.6.2 Subpopulation for the pilot

To gain experience with setting up the register and identifying potential issues, we recommend setting up a pilot with a subpopulation consisting of authorisations relating to a small selection of PTs. This selection could consist of one PT per main biocide group, for instance. A few considerations when selecting the product types are discussed in section 2.4.12.

3.7 Identification of units

Identifying target population units (the authorisations) and reporting units (the authorisation holders), hereinafter referred to as respondents, is important to be able to verify whether all (or the main) units have provided data. If there are changes in the structure of the authorisation holder’s company (split-ups, mergers, etc.), these developments must be clear in order to be able to link the correct units (respondents) to each other.

3.7.1 One identification variable

Ideally, the Chamber of Commerce number of each authorisation holder will be known to the register holder. These numbers can serve as unique identifiers for these units. Even so, companies may have complex structures and the Chamber of Commerce may establish and dissolve units. Names of authorisation holders can be found in the authorisation databases of the Ctgb and ECHA. If you know the name, you can look for the Chamber of Commerce number (via Google or web scraping, for example). Given the application procedure for authorisation, companies are probably motivated to make sure that the “in name of” part of the authorisation is properly established and not affected by structural changes.

Identification of the authorisations is just as important as identification of the authorisation holders (respondents). Authorisations are assigned a unique authorisation number and it is a good idea to include this number in the registration to prevent double entries (e.g. because they are listed in both the Ctgb and ECHA registers).

3.7.2 A combination of identification variables

If there is no single identification variable that is available for authorisation holders, the alternative is to use a combination of variables to refer to a unique unit. It is possible to identify authorisation holders on the basis of company name, address and telephone number, but this is not nearly as precise as a unique identification number because there is a risk of erroneous or missing links with the list of companies of the Chamber of Commerce or the general business register (ABR) of the CBS. So we can never be entirely sure that all the targeted respondents have reported the information.

3.8 Target variables

Target variables are variables (other than the identifying variable(s)) that should be included in the register and that can be used to publish aggregated information at a later point.

Some biocides appear in several product type categories, so it would be useful to specify the relationship between biocide and product type in a separate (key) table: the authorisation table. This would enable us to calculate totals by product type (see section 3.10.3). The authorisation table should be as consistent as possible with the information from the Ctgb and ECHA registers. A few examples of possible tables are shown below.

There are various variants we could use to structure the register. It makes sense to use different tables for each unit. There are three possible variants:

1. Asking for active substances quantities;
2. Asking for product quantities;
3. A hybrid variant where the respondent can choose which one to report.

One advantage of variants 1 and 2 is ease of processing. One advantage of variant 3 is the fact that the respondent can choose the option that is most feasible in terms of their own records, which optimises the quality of the submitted data. Variants 2 and 3 would require a conversion table for products and active substances. This table will be stable over time for each product, because active substance concentrations do not change during an authorisation period. For the plant protection register this is similar to the Fytostat database of manufacturers.

3.8.1 Authorisation holder table
VariabeleDescription
Authorisation-id Identification number of the authorisation
Authorisation-idIdentification number of the authorisation holder
Source of authorisationCtgb (Dutch authorisation board) or ECHA
Brand product quantities supplied to the
non-professional market
During the reporting period
Brand product quantities supplied to the
the professional market
During the reporting period
Brand product quantities supplied to the
both markets
During the reporting period
Brand name (or CAS-number)Brand name and quantity, if available
e.g. Racumin Foam, 500 ml

Registration of product quantities may provide additional information for policy makers (in addition to active substance quantities in kg), but that will also make the register more complex and expensive. The Ministry of IenW will have to decide whether the benefits outweigh the costs.

If the contact person for an authorisation holder is known, some of the requested information can be entered in advance.

3.8.2 Product table
Variabele (kolom in producttabel)Description
Authorisation-idIdentification number or the authorisation
AuthorisationCtgb or ECHA
Start dateDate on which the authorisation starts
End dateDate on which the authorisation expires
Grace market dateDate on which the authorisation may not be marketed anymore
Biocide producttypeBased on the standard list
Biocide main groupBased on the standard list
Biocide brand productBrand name (follows from the product )
Hazard indicationHazard-sentences
Total product quantityNumeric values
Product quantity unitEUnit or the above variable, e.g. grams or liters
Specific gravity or liquidif product is flowable, e.g. kg/l or g/l
Name active substance 1Chemical substancename, e.g. propiconazole
Content active substance 1Numerical value
Unit of content active substance 1For example g/kg or g/l
Amount of active substance 1 in productIn gram if possible, or kg
Name active substance 2Chemical name, e.g. tebuconazole
Content active substance 2Numerical value
Unit or content of active substance 2For example g/kg or g/l
Amount of active substance 2 in productIn gram if possible, or kg
This table would serve as an aid, for example for the conversion of submitted product quantities to active substance quantities.

A product may contain multiple active substances, so multiple columns are provided for active substance variables. Depending on the maximum number of active substances in one biocide, a column may appear more than twice. An alternative is to give each substance-product combination a row of its own, so we only need one column for the active substance variable. In that case, the number of rows in which a product appears for each authorisation holder corresponds to the number of active substances.

Many concentrations are expressed as a weight/weight ratio (e.g. grams/kilograms). Liquid products are expressed as grams/litre. If we want to be completely accurate, we will need specific gravity as well. These choices may affect the end totals. Getting information on this variable (amount of biocide in product) may prove challenging. This information is not publicly available in a way that is easy to find in the ECHA authorisation register. However, the information is available in section 2.1 of the Summary of Product Characteristics (SPC) in the Biocidal Fact Sheets, which are publicly available for every biocide. One option would be to use an automatic text analysis method to search these PDF files for this information. An alternative for “name of active substance” is the CAS number, which is also listed in the active substance table. The main purpose of this variable is to provide a link with the active substance table.

There is a standard list for biocide main group number, biocide main group name, biocide product number and biocide product name.

3.8.3 Active substance table
VariableDescription
Name active substanceChemical name, e.g. propiconazole
CAS-numberUnique identification number for chemical
ToxicityTo be defined

There are several ways to measure the toxicity of an active substance, for example based on human toxicity or environmental toxicity. Hazard Statements are associated with the product, so these can be found in the product table and not in the active substance table. Which method is most relevant depends on the requested variable: active substance quantity or product quantity. A pilot would help answer this question (see 4.1).

3.9 Comparability

We recommend asking for a minimum of two years’ worth of information when the information is first requested. At the next request, year t can then be compared with the available years t-1 or t-2, and over time a the quality of the data for this whole time period can be assessed.

3.10 Quality control

3.10.1 Quality control for (groups of) units

(Groups of) units, both respondents and authorisations, can be subjected to (automated) quality controls, for issues such as:

  1. Accuracy of the identifiable characteristics. Possible options are syntactic control, verification of correct authorisation, and inventory of missing values.
  2. Coverage ratio of the target population (over- or undercoverage, selectivity, duplicates, time delays and the addition of new units or removal of old units).

3.10.2 Quality control of target variables

The following (automated) controls can be performed by the register holder on target variables supplied by the respondent:<

  1. Do all the values fall within an accepted range (and is it zero for non-authorised products)?
  2. Are there any extreme values and if so, are they realistic?
  3. Are there inconsistencies between variables from one respondent?
  4. Are there any missing values?

Are there values that are expected to be the same over time (e.g. identification key, product type or certain active substances)?

3.10.3 Product type double entries

When working with statistics that are calculated at several levels, such as “active substance quantities in kg per PT” and “total active substance quantities in kg for all PTs”, we have to factor in overlap between PTs. There are products that appear in multiple product type categories, which results in double entries when the totals for each PT are added to calculate the total for all PTs. Adding up all biocides (individually, not by PT), allows us to calculate the total without any duplicates.

Double entries result in a lower total for active substance quantities in kg for all PTs than the sum of active substance quantities in kg in each PT.

Breakdown by product type is not possible if values are missing in the “biocide product type” variable, but the biocide can still be included in the calculation of the total. If there is uncertainty about the PT of a biocide (for example “PT02” or “PT04”), it might be possible to use weighted categories (for example 0.5 for PT02 and 0.5 for PT04) so it is partially counted for the different PTs.

3.11 Quality improvement

Upon completion of the (automated) quality control, the following steps can be taken to improve quality:

  1. Values with quality issues can be kept or deleted.
  2. The register holder can ask the respondent about the variables with quality issues.
  3. Values that contain obvious errors can be overwritten with correct data.
  4. Missing values can be imputed, for example by using historical data. There are several methods for imputing data. Which method is the most suitable will have to be determined for each variable individually.

3.12 Reproducibility

  1. The software used by the register holder must meet the following requirements:
  2. The software must have low error sensitivity with regard to human actions, for example by performing plausibility controls on entered data.
  3. The software must be well-documented. All components and their interrelation must be described in detail.
  4. The software must be easy to transfer between different colleagues, to prevent dependence on any one individual.