Inferensys

Glossary

Best Possible Medication History (BPMH)

A comprehensive and verified list of all medications a patient was taking prior to a care transition, obtained through a systematic interview and review of at least two different sources of information.
Developer building agentic RAG system, retrieval pipeline diagram on laptop, technical workspace with notes.
DEFINITION

What is Best Possible Medication History (BPMH)?

A Best Possible Medication History (BPMH) is a comprehensive and verified list of all medications a patient was taking prior to a care transition, obtained through a systematic patient interview and corroboration with at least two different sources of information.

The Best Possible Medication History (BPMH) is the foundational artifact of the Medication Reconciliation (MedRec) process. It is not a simple transcription of a patient's pharmacy list; rather, it is an actively curated and verified inventory generated by a trained clinician—typically a pharmacist or pharmacy technician—who systematically interviews the patient or caregiver and cross-references the reported history against at least one other objective data source, such as a community pharmacy dispensing record, a previous discharge summary, or a blister pack inspection.

The primary clinical utility of the BPMH lies in its role as the gold-standard comparator against admission, transfer, and discharge orders to identify Unintentional Discrepancies. By resolving conflicts between what the patient was actually taking and what is newly prescribed, the BPMH directly prevents Adverse Drug Events (ADEs) caused by Omission Errors and Duplicate Therapy. In automated systems, AI models are tasked with synthesizing a draft BPMH from heterogeneous sources, applying Temporal Reasoning and Dose Normalization to construct a chronologically accurate Medication History Longitudinal Record before human verification.

DEFINING THE GOLD STANDARD

Core Characteristics of a BPMH

A Best Possible Medication History is defined by its rigorous methodology, not just its output. These core characteristics distinguish a BPMH from a simple medication list.

01

Multi-Source Verification

A BPMH is fundamentally defined by the systematic triangulation of information from at least two independent sources. This process moves beyond a simple patient interview to create a high-reliability record.

  • Primary Sources: Patient/caregiver interview, inspection of medication vials or blister packs, and review of the patient's own medication list.
  • Secondary Sources: Community pharmacy dispensing records, insurance claims data, previous hospital discharge summaries, and nursing home transfer records.
  • Discrepancy Resolution: When sources conflict, the pharmacist must use clinical judgment to determine the most accurate regimen, documenting the rationale for the final decision.
2+
Minimum Sources Required
02

Structured Patient Interview

The BPMH interview is a standardized, systematic conversation, not an informal chat. It employs a brown-bag review and open-ended questioning to uncover non-prescription items and actual usage patterns.

  • Show-and-Tell: Patients are asked to physically present all medications, including inhalers, patches, drops, and over-the-counter drugs.
  • Probing Questions: The interviewer explicitly asks about adherence ('How many doses did you miss last week?'), recent changes, and medications prescribed by specialists.
  • Non-Verbal Cues: The process accounts for health literacy barriers by using visual aids and pill identification guides to confirm what the patient is actually taking.
03

Comprehensive Scope

A true BPMH captures the complete pharmacotherapeutic picture, extending far beyond prescription pills to include all substances that could cause an adverse event or interaction.

  • Prescription Medications: All regularly scheduled and PRN (as needed) drugs.
  • Non-Prescription Items: Over-the-counter drugs, vitamins, herbal supplements, and traditional medicines.
  • Dose & Frequency: The exact strength, route of administration, and the actual dosing schedule the patient follows, which may differ from the label.
  • Temporal Context: The date the medication was last taken and the duration of therapy are critical for preventing withdrawal syndromes or duplicate loading doses.
04

Formal Documentation Standard

The output of a BPMH is a structured, coded, and attributable record, not a free-text narrative. This standardization is essential for interoperability and downstream clinical decision support.

  • Coded Data: Medications are mapped to standardized terminologies like RxNorm to enable automated drug-drug interaction and allergy checking.
  • Source Attribution: Each medication entry is explicitly linked to its source of information (e.g., 'per patient vial' or 'per CVS dispensing record') to establish data provenance.
  • Discrepancy Flagging: The BPMH serves as the baseline against which admission orders are compared, with any unintentional discrepancies clearly documented and communicated to the prescriber.
05

Pharmacist-Led Execution

The gold standard for BPMH collection is execution by a trained clinical pharmacist or a certified pharmacy technician under pharmacist supervision. This leverages specialized expertise in medication management.

  • Clinical Inference: Pharmacists can resolve incomplete histories by inferring a missing dose based on a patient's reported frequency and the standard available strengths.
  • Therapeutic Duplication: A pharmacist is uniquely trained to identify a patient taking both a brand-name and generic version of the same drug, a common error missed by non-clinicians.
  • Cognitive Dissonance: The process requires navigating the patient's belief about what they take versus the objective evidence, a reconciliation skill central to pharmacy practice.
06

Baseline for Reconciliation

The BPMH is not an end in itself; it is the definitive pre-admission baseline against which all subsequent medication orders are compared during the formal Medication Reconciliation (MedRec) process.

  • Comparator Role: Every admission, transfer, and discharge order is compared line-by-line to the BPMH to identify unintentional discrepancies.
  • Error Prevention: This comparison catches high-risk omission errors (failing to restart a critical beta-blocker) and commission errors (prescribing a drug the patient is allergic to).
  • Continuity of Care: The BPMH travels with the patient across transitions, ensuring that the verified home regimen is communicated to the next provider of care.
BPMH ESSENTIALS

Frequently Asked Questions

Clear, authoritative answers to the most common questions about the Best Possible Medication History process, its clinical significance, and the technologies that support it.

A Best Possible Medication History (BPMH) is a comprehensive, verified list of all medications a patient was taking prior to a care transition, obtained through a systematic patient or caregiver interview and cross-referenced with at least two different sources of information. The process works by first conducting a structured interview using open-ended questions and medication-specific probes, then validating that self-reported history against objective sources such as community pharmacy dispensing records, previous hospital discharge summaries, medication vials brought from home, and provincial or state prescription drug program databases. Any discrepancies between sources are explicitly resolved and documented, resulting in a single, reconciled list that serves as the gold-standard reference for admission medication orders, preventing unintentional discrepancies from propagating into the inpatient record.

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.