Why Route and Formulation Need Separate Fields
A dose amount does not fully describe a medication event. The same numeric amount can refer to different products, and a product can be used in different ways only when its formulation and route permit them. For a pharmacometric data set, route and formulation therefore need explicit, interpretable representations rather than being left embedded in free-text product descriptions.
These fields are related but not interchangeable. Route describes how the product is administered to the body, such as oral, intravenous, or subcutaneous. Formulation describes the prepared product, such as a tablet, oral solution, or injection solution. A product identifier supplies the link to the specific product and its labeled strength or concentration. Dose amount and dose units describe the quantity recorded for the event; they do not by themselves identify the route or formulation.
A project might use categories such as ORAL, INTRAVENOUS, and SUBCUTANEOUS for route, and TABLET, ORAL_SOLUTION, and INJECTION_SOLUTION for formulation. These are examples of project-defined values, not a universal coding system. The important requirements are that categories have clear definitions, are applied consistently, and can be traced back to the source information.
Designing the Fields and Their Allowed Values
Choose categories that preserve meaningful distinctions for the analysis and can be supported by the source records. Avoid combining several concepts into one value. For example, “IV infusion solution” contains both a route and a formulation; recording it as one category makes it harder to check either attribute independently. Instead, store route as INTRAVENOUS and formulation as an appropriate solution category, with the product identifier carrying the specific presentation.
A controlled vocabulary does not mean that the source data should be overwritten. Retain the source label, code, or product description alongside the standardized fields, or maintain a documented mapping from source values to standardized categories. This makes it possible to audit a classification, revise a mapping when better information becomes available, and distinguish an actual source value from a derived project category.
| Field | What it represents | Example value |
|---|---|---|
| Source route text | Route wording as received from a source | “by mouth” |
| Standard route | Mapped route category | ORAL |
| Source formulation text | Product-form wording as received | “10 mg/mL solution” |
| Standard formulation | Mapped formulation category | ORAL_SOLUTION |
| Product identifier | Specific product presentation used to look up metadata | Product code P104 |
| Dose amount and unit | Recorded quantity and its unit | 5 mL |
A product identifier is especially useful because broad categories do not establish compatibility by themselves. “Solution” does not say whether a product is oral or injectable, and “5 mg” does not establish whether the dose was recorded on the active-ingredient basis, salt basis, or another labeled basis. Product metadata, source documentation, and the project’s unit definitions must be considered together.
For each product, maintain or obtain a reference record that specifies the supported route or routes, formulation, strength or concentration, and labeled dose basis. A product may have more than one valid route only when the authoritative product information supports those routes. Do not infer route from a formulation label alone, or formulation from route alone.
A Compatibility Check Built From Product Metadata
A compatibility check compares the event’s recorded values with the metadata for its identified product. It is a data-quality rule, not a treatment recommendation. It can flag a mismatch for review, but it cannot determine which source field is wrong when the record is contradictory.
Use the product identifier or documented source mapping to find the relevant product metadata. If the product cannot be resolved, mark compatibility as not assessable rather than guessing from a similar name.
Compare both event fields with the routes and formulations allowed for that product. Flag unsupported combinations for source review.
Compare the event’s unit with the product’s strength or concentration and the project’s definition of the dose amount. Confirm that any conversion uses the correct product presentation and basis.
Distinguish a confirmed compatible record, a confirmed incompatibility, and a record that cannot be assessed because required metadata are missing. Preserve the original values and document any correction.
The logic can be stated formally. Let the product metadata for product \(j\) include the allowed route set \(R_j\), allowed formulation set \(F_j\), and dose-unit basis \(U_j\). For an event with product \(j\), route \(r\), formulation \(f\), and dose unit \(u\), the event passes these checks only if:
The last condition is not merely a string comparison. A mass unit such as mg may be compatible with a product whose labeled strength is expressed in mg per mL, but the recorded dose amount must still be on the intended basis. Conversely, mL describes a volume, not an amount of active ingredient. A unit conversion is defensible only when the product concentration and the relevant basis are known.
Worked Example: Mapping Source Route and Formulation Values
Suppose a hypothetical source file contains three dose events. The source route values are “by mouth,” “PO,” and “oral.” The source formulation values are “tablet,” “solution,” and “oral liquid.” The project reference defines route category ORAL and formulation categories TABLET and ORAL_SOLUTION. Show how to standardize these records while preserving their provenance.
Step 1: Map route values. If the source dictionary or study documentation confirms that all three route labels mean administration by mouth, map each to ORAL. Retain each original route value in a source-route field. The standardized value is consistent, but the source wording remains auditable.
Step 2: Map formulation values. “Tablet” maps to TABLET. “Oral liquid” maps to ORAL_SOLUTION only if the product documentation identifies it as a solution; “liquid” alone may be too broad because it does not necessarily distinguish a solution from a suspension. In this hypothetical mapping, assume documentation confirms that the second and third records are solutions, so both map to ORAL_SOLUTION.
Step 3: Check product-specific compatibility. Suppose the product metadata identify the first product as an oral tablet and the other two as oral solutions. The route and formulation values agree with those metadata. If the product record instead identified the second event as a suspension, the source label “solution” would need review rather than automatic recoding.
The result is a reproducible mapping, not an assertion that every label that looks similar is equivalent. The mapping depends on documented meaning, and ambiguous source values remain unresolved until supporting information is available.
Worked Example: Checking Dose Units Against a Product
Consider a hypothetical oral solution labeled 20 mg/mL. Its product metadata specify route ORAL, formulation ORAL_SOLUTION, and strength 20 mg of active ingredient per mL. A dose event records route ORAL, formulation ORAL_SOLUTION, and dose 3 mL. Determine what can be checked and the corresponding active-ingredient amount.
The route and formulation agree with the product metadata. The recorded unit mL is a volume, so it is compatible with a dose expressed as volume for this solution, provided the project defines the event amount as the volume administered. The implied active-ingredient amount is:
The mL units cancel, leaving 60 mg. This calculation is valid because the hypothetical product concentration and dose basis are specified. It would not be valid to apply this conversion to a different product merely because that product is also called an oral solution.
Now suppose another record for the same product contains dose 60 mg. That value may be compatible if the data dictionary says dose amount is stored as active-ingredient mass. It is not directly interchangeable with 3 mL unless the concentration is used and the conversion is documented. The product’s concentration gives \(60\ \mathrm{mg} \div (20\ \mathrm{mg/mL}) = 3\ \mathrm{mL}\), but that calculation does not establish which representation the source actually recorded.
Worked Example: Flagging a Route–Product Mismatch
A hypothetical product catalog lists product Q as an injection solution with route INTRAVENOUS and strength 5 mg/mL. A dose event identifies product Q, records formulation INJECTION_SOLUTION, route ORAL, and dose 2 mL. Decide what the checks show.
The formulation agrees with the product metadata, but the route does not: ORAL is not in the allowed route set for product Q, which contains INTRAVENOUS. The 2 mL value is dimensionally a volume and can be related to the listed concentration: \(2\ \mathrm{mL} \times 5\ \mathrm{mg/mL} = 10\ \mathrm{mg}\). This arithmetic does not resolve the route conflict. The event should be flagged for review of the product identifier, route, and source record; it should not be silently changed to INTRAVENOUS merely because that is the product’s listed route.
If product Q’s metadata were incomplete, the proper result would be “not assessable” for the route check, not “compatible.” A missing rule is not evidence that a combination is allowed.
Diagnostics, Alternatives, and Limitations
Useful diagnostics summarize the combinations found in the data and the reasons for flags. Review counts of unknown route and formulation values, route–formulation pairs not represented in the product reference, dose units that cannot be interpreted on the stated basis, and product identifiers with missing or conflicting metadata. Examine flagged records in context; a rare combination can be valid, while a common combination can still reflect a systematic mapping error.
A useful distinction is between invalid and unresolved. A record is invalid under the check when reliable product metadata contradict a recorded field. It is unresolved when the product is unknown, the metadata are incomplete, or the source wording cannot be mapped unambiguously. Do not convert unresolved records into valid ones by assigning a default category.
The main alternative to a controlled category system is retaining only source descriptions or using one free-text medication field. This preserves wording but makes consistent filtering and validation difficult. At the other extreme, replacing all source values with a single standardized label can improve uniformity but lose traceability and erase distinctions needed for review. A practical approach keeps source values, standardized fields, product metadata, and documented mapping rules together.
Compatibility checks also have limits. Product metadata can be outdated or incomplete; source records can identify the wrong product; and a single product identifier may not distinguish presentations with different concentrations. Automated checks can identify contradictions but cannot establish what was actually administered. Resolution requires authoritative source evidence and a documented decision. Route-specific process fields may also be needed for some administration types; those fields should be handled according to their own data definitions rather than inferred from route alone.
Common Mistakes
- Combining route and formulation. A label such as “IV solution” hides two attributes. Store route and formulation separately so each can be checked.
- Inferring route from the word “solution.” A solution may be prepared for different routes. Use product-specific metadata.
- Treating a unit as proof of compatibility. A volume unit does not prove the product is a solution suitable for the recorded route, and a mass unit does not identify the product basis.
- Silently correcting a mismatch. A route conflict may indicate an incorrect route, product identifier, or source record. Flag and investigate it; do not guess which value to overwrite.
- Mapping ambiguous text too aggressively. “Liquid” may not be specific enough to assign a formulation category. Preserve it as unresolved until the product presentation is verified.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- A source says “PO,” and the project dictionary defines PO as administration by mouth. Which field should receive the standardized value, and what should happen to the source text?
- A product is documented as an oral solution at 12 mg/mL. A dose record gives 4 mL. What active-ingredient amount does this imply, and what assumption makes the calculation valid?
- A product catalog lists a product as an injectable solution for intravenous use. An event records that product with route SUBCUTANEOUS. Should the event be automatically changed to INTRAVENOUS? Explain.
- A dose is recorded as 25 mg for a product whose strength is 5 mg/mL. What volume corresponds to that amount, and what additional information is needed before treating the records as interchangeable?
- A source formulation says only “liquid,” and the product identifier is missing. How should the formulation compatibility check be classified?
Solutions. (1) Map PO to the controlled route value ORAL, while retaining “PO” as the source value or in a documented source mapping. (2) \(4\ \mathrm{mL} \times 12\ \mathrm{mg/mL} = 48\ \mathrm{mg}\). The calculation assumes the stated concentration applies to that specific product and that the 4 mL is the dose volume on the stated product basis. (3) No. Flag the conflict for review because it could reflect an incorrect route, product identifier, or source record; the catalog entry alone does not establish which event value should be changed. (4) \(25\ \mathrm{mg} \div (5\ \mathrm{mg/mL}) = 5\ \mathrm{mL}\). Interchangeability requires confirmation of the exact product and concentration, the dose basis, and that the records refer to the same intended dose representation. (5) Classify it as unresolved or not assessable: neither “liquid” nor a missing product identifier establishes a specific formulation or allowed route.