Sorting medication-related problems out of free text into a standardized taxonomy — Indication, Effectiveness, Safety, and Adherence categories
A medication-related problem (MRP) is any undesirable event a patient experiences that involves, or is suspected to involve, drug therapy and that interferes with achieving desired health outcomes. Described in unstructured clinical narrative, every problem reads differently even when the underlying issue recurs constantly across a population. A standardized classification system replaces free text with a shared, finite set of categories — turning anecdote into data.
When a pharmacist or clinician writes "patient still short of breath despite inhaler, may need dose increase," that observation is clinically useful but computationally inert — it cannot be aggregated, searched, or compared across the next thousand patients. Classification systems (such as PCNE, Cipolle's Pharmacotherapy Workup, or HTA problem lists) translate that sentence into a coded category: here, an effectiveness-related problem, insufficient dosage.
Once coded, the same underlying issue can be counted whether it appears in patient 1 or patient 10,000, in this clinic or another, this month or last year. That counting is what makes quality improvement, safety signal detection, and outcomes research possible at all.
A classification is only as useful as it is standardized: the same clinical scenario must always map to the same category, regardless of who is doing the coding. This consistency is the entire value proposition of the exercise.
A useful medication-problem taxonomy separates issues along their root cause, not merely their symptom. Two patients might both be "not doing well on their blood pressure medication" — but one has a drug that is simply too weak for their disease severity (effectiveness), while the other is skipping doses because of cost (adherence). Lumping both under "poor control" would erase the very information a clinician needs to act.
The four broad domains used throughout this simulator — Indication, Effectiveness, Safety, and Adherence — are chosen because each implies a different, largely non-overlapping category of intervention. Knowing which domain a problem falls into tells the care team roughly what kind of fix is needed before any further detail is reviewed.
The payoff of standardized coding shows up at the population level. A clinic that codes every MRP can ask: are safety-related problems clustering around a specific drug class after a formulary change? Is the rate of adherence-related problems higher in patients discharged without teach-back counseling? Did a new dosing protocol reduce effectiveness-related problems for a given indication over the following year?
None of these questions are answerable from a folder of free-text notes. They become straightforward queries once every problem instance carries a standardized category label, a severity rating, and a timestamp.
This category captures two mirror-image situations: a medication being taken with no valid medical indication to justify it, and a diagnosed condition left without appropriate drug therapy at all. Both represent the same underlying failure — a mismatch between what the patient actually needs and what they are actually prescribed — just approached from opposite directions.
This sub-type appears constantly in patients with long medication lists and multiple prescribers. A drug started for an acute, now-resolved problem (e.g., a short-course proton pump inhibitor never stopped), a medication continued after the diagnosis it treated was ruled out, or a duplicate therapy from two different specialists prescribing overlapping drugs for the same target — all fall here.
The review question is simple to ask but often hard to answer without a careful history: "what is this drug actually for, right now, in this patient?" If no current, valid indication can be identified, the medication itself is the problem, independent of whether it is causing any harm.
The inverse case is a diagnosed condition with no corresponding drug therapy, or with therapy that does not match evidence-based standards for that indication. Examples include a patient with documented osteoporosis on no bone-protective therapy, a post-myocardial-infarction patient without a beta-blocker with no documented contraindication, or newly diagnosed hypertension with no antihypertensive initiated at all.
This is often the least visible category of MRP, because it is defined by absence rather than by an event — nothing "goes wrong" that triggers a chart flag. Systematic medication reviews and indication-based checklists are usually required to surface it.
Both sub-types are corrected the same conceptual way: align the medication list with the actual, current clinical picture — remove what lacks a reason, add what the diagnosis requires.
This category captures situations where the chosen drug, or the dose at which it is given, is inadequate to reach the therapeutic goal. The medication is not causing harm — it is simply not doing enough good. This distinction from a safety problem matters: the intervention needed is to intensify or change therapy, not to withdraw it.
Even a correctly indicated drug can be the wrong pharmacologic choice for a given patient or target. A first-generation agent selected when guidelines favor a more effective newer class, a drug with known reduced efficacy in a specific population (e.g., pharmacogenomic non-response), or an agent that does not adequately address the severity of disease all belong here.
The hallmark of this sub-type is that the drug is doing something — it has a plausible mechanism relevant to the indication — but is simply not the most effective available option for reaching the stated clinical goal.
The far more common sub-type: the right drug at the wrong (too low) dose. Sources include overly cautious initial titration never advanced, renal or hepatic dose adjustments that undershoot rather than merely account for organ function, and prescribers stopping titration once side effects appear even though the therapeutic target has not been reached.
Laboratory or clinical monitoring — blood pressure still elevated, HbA1c still above goal, INR still subtherapeutic — is usually what surfaces this problem. It is corrected by systematic up-titration guided by the same monitoring parameter that flagged it.
Effectiveness problems are defined relative to a therapeutic goal, not relative to harm — always ask "is this treatment achieving what it is meant to achieve?" before asking whether it is safe.
This category captures adverse drug reactions, dose-related toxicity, and other safety concerns where the medication itself is the source of harm to the patient. It is the mirror image of the effectiveness domain: here the drug is doing something, but that something is damaging rather than merely insufficient.
An adverse drug reaction (ADR) is a harmful, unintended response occurring at doses normally used for prophylaxis, diagnosis, or therapy. This spans predictable, dose-independent reactions (a known class side effect such as ACE-inhibitor cough) to rare idiosyncratic reactions (a hypersensitivity rash to an antibiotic) that could not have been anticipated from the pharmacologic profile alone.
Because ADRs can mimic new medical conditions, they are notoriously under-recognized: a "new" symptom is often the medication, not a new diagnosis requiring its own separate workup and treatment — a distinction that avoids the classic prescribing cascade.
Distinct from an idiosyncratic ADR, dose-related toxicity is a predictable, exposure-dependent harm: too much of the right drug. Renal impairment causing accumulation of a renally cleared drug, a drug interaction raising serum levels of a narrow-therapeutic-index agent, or simple prescribing error resulting in supratherapeutic dosing all fall here.
Because the relationship between exposure and harm is usually well characterized pharmacokinetically, dose-related toxicity is often the most preventable safety category — caught by therapeutic drug monitoring, renal/hepatic dose adjustment protocols, and interaction-checking software before it ever reaches the patient.
Roughly half of clinically significant adverse drug events are considered preventable — dose-related toxicity in particular is the safety sub-type most amenable to systematic prevention through monitoring and dose-adjustment protocols.
This category captures situations where the patient is not taking the medication as prescribed — for a wide variety of underlying reasons. Adherence problems are classified separately from the other three domains because their root cause lies outside the drug and dose selection entirely: even the perfect prescription fails if it is never actually taken.
It is tempting to treat non-adherence as just another cause of "the drug isn't working" and fold it into the effectiveness domain. But doing so would misdirect the intervention: up-titrating the dose of a medication the patient is not taking at all achieves nothing and can even increase risk once adherence improves and exposure suddenly rises.
Classifying adherence separately forces the reviewer to ask a prior question before touching the prescription: is the patient actually taking this medication as intended? Only once that is confirmed does it make sense to evaluate whether the drug and dose themselves are adequate.
Adherence barriers are heterogeneous and each calls for a different fix:
• Financial: the patient cannot afford the copay or the medication is not covered — resolved by formulary alternatives, patient assistance programs, or switching to a lower-cost equivalent • Regimen complexity: too many pills, too many times per day — resolved by simplification, combination products, or pill organizers • Understanding: the patient does not know why or how to take the drug — resolved by counseling and teach-back education • Side effects: the patient stops the drug themselves because of a tolerability issue never reported back — resolved by proactive side-effect screening and dose or agent adjustment • Practical/physical: difficulty swallowing pills, opening packaging, or accessing a pharmacy — resolved by formulation changes or delivery logistics support
Because the fix for non-adherence is barrier-specific rather than pharmacologic, correctly classifying a problem as adherence-related — and identifying which underlying reason applies — is often more important to resolving it than any change to the drug or dose itself.