Skip to main content

Table 1 Three component framework for constructing conceptual definitions and operational criteria for counting number of drugs

From: Application of a framework for determining number of drugs

Component Description and explanation/examples
1. Scope What types of products will be considered drugs?
Each drug product used by a patient can be classified according to these five subcomponents: (a) prescription drug status, (b) drug type, (c) route of administration, (d) dosage form, and (e) common use. This is not an exhaustive list of subcomponents. The investigator can create additional subcomponents as needed for the specific study objectives, such as therapeutic category or pregnancy category. Possible values for each of the five subcomponents are given below for illustrative purposes and are not intended to be exhaustive
Prescription drug status: prescription drug, over-the-counter drug
Drug type: standard drug, vitamin, herbal, dietary supplement, other complementary and alternative medicine
Route of administration: oral, inhalation, topical, transdermal, ophthalmic, subcutaneous, sublingual, rectal
Dosage form: tablet, capsule, cream, solution, suspension
Common use: regular, as needed
For each subcomponent, decisions must be made about which values will be counted as drugs, even if the decision is to include all possible values for a subcomponent. For example, will prescription and over-the-counter products be counted as drugs? Will vitamins, herbals, dietary supplements, or other various complementary and alternative medicine products be counted in addition to standard drugs? Will certain routes of administration be excluded? Will all dosage forms be counted? Will drug products commonly used on an “as needed” basis be included along with those commonly used on a regular basis?
2. Uniqueness Will an ingredient-based or product-based approach be used?
In an ingredient-based approach, drug products are dissected into their ingredients, and unique ingredients are counted. Single ingredient products contribute one to the count while combination products contribute at least two to the count of ingredients. When multiple drug products include the same ingredient, that ingredient is counted only once. When drug products are dissected into their ingredients, those ingredients can be counted based solely on ingredient (e.g. albuterol), or descriptive information can be appended to an ingredient to count at a more detailed level of ingredient-route (e.g. albuterol-oral), ingredient-route-form (e.g. albuterol-oral-tablet), or ingredient-route-form-strength (e.g. albuterol-oral-tablet-2 mg). For the sake of illustration, consider a regimen including: (1) a combination inhaler containing albuterol and ipratropium, (2) albuterol nebulizer solution, and (3) albuterol oral tablets. The ingredient-based count at the ingredient level is 2 drugs, 1 for albuterol and 1 for ipratropium. The ingredient-based count at the ingredient-route level is 3 drugs, 1 for albuterol-inhalation, 1 for ipratropium-inhalation, and 1 for albuterol-oral
In a product-based approach, unique drug products are counted. Combination products contribute one to the count regardless of the number of ingredients in the product. Drug products can be counted at the ingredient(s)-route-form-strength level (e.g. albuterol/ipratropium-inhalation-aerosol-100 mcg/20 mcg) or at a more general level of ingredient(s)-route-form (e.g. albuterol/ipratropium-inhalation-aerosol), ingredient(s)-route (e.g. albuterol/ipratropium-inhalation), or ingredient(s) (e.g. albuterol/ipratropium). Considering the illustrative regimen described above, the product-based count at the ingredient(s)-route-form level is 3 drugs, 1 for albuterol/ipratropium-inhalation-aerosol, 1 for albuterol-inhalation-solution, and 1 for albuterol-oral-tablet
3. Timeframe Will a cross-sectional or longitudinal orientation be used?
For cross-sectional analyses, what criteria will be used to determine whether a drug was in the regimen at the specified point in time?
Cross-sectional vs. longitudinal orientation. A cross-sectional orientation seeks to count the number of drugs in a regimen at a specific point in time. In contrast, the longitudinal orientation asks how many unique drugs the patient has used over some period of time (e.g. 1 year). Thus, the longitudinal orientation includes all drugs present in the regimen at the end of the time interval plus all drugs that had been used and discontinued during the interval
Timeframe criteria for cross-sectional analyses. The choice of criteria for the cross-sectional orientation is driven by the data source. Drug use data collected from patient interview or questionnaires will often ask the patient about what drugs they took yesterday or within some relatively short time frame (e.g. last 7 days). Studies using inpatient drug administration records would typically count unique drugs administered on a given day. Pharmacy refill databases are another common data source and require a more thorough examination (Fig. 1)
Other considerations Additional considerations can be implemented using variations of these 3 basic components to accomplish study specific objectives. Some examples using a standard specification for most drugs, but applying different criteria for selected types of drugs based on study specific needs are:
A study focused on asthma may apply specialized criteria for inhaled dosage forms
A study focused on pain management may apply a wider allowable index gap (defined as the length of time before the index date in which a drug must be filled to be counted, Fig. 1) for drug products commonly used on an “as needed” basis