Medication Adherence Assessment

Instructions

Step 1: Identify Patient and Active Chronic Medications

Tool: fhir_search
resourceType: "MedicationRequest"
queryParams: "patient=[patient-id]&status=active&intent=order&_count=100"

Filter for chronic medications (those with repeating dispense requests or ongoing conditions). Exclude PRN-only and short-course medications (antibiotics, acute pain).

Step 2: Pull Dispense History for Each Medication

For each chronic medication identified:

Tool: fhir_search
resourceType: "MedicationDispense"
queryParams: "patient=[patient-id]&medication=[medication-code]&_sort=-whenhandedover&_count=50"

If medication-specific search fails, pull all dispenses:

Tool: fhir_search
resourceType: "MedicationDispense"
queryParams: "patient=[patient-id]&status=completed&_sort=-whenhandedover&_count=200"

Extract for each fill:

• whenHandedOver (dispense date)
• quantity.value (tablets/units dispensed)
• daysSupply.value (days supply)
• medicationCodeableConcept (drug identity)

Step 3: Calculate Adherence Metrics

Medication Possession Ratio (MPR):

MPR = (Total days supply dispensed in period) / (Number of days in period) x 100

Evaluation period: typically 12 months or since first fill

Proportion of Days Covered (PDC) (preferred measure):

PDC = (Number of days with medication available) / (Number of days in period) x 100

Key difference from MPR: PDC does not double-count overlapping fills.
When fills overlap, shift the later fill start to after the previous supply runs out.

See references/adherence-metrics.md for detailed calculation formulas and interpretation.

Step 4: Assess Fill Pattern Anomalies

Refill Gaps > 7 days: Calculate gap between each fill's expected end date (whenHandedOver + daysSupply) and the next fill date. Flag any gap exceeding 7 days.

Gap = next_fill_date - (current_fill_date + current_days_supply)
If gap > 7 days: FLAG as potential non-adherence

Early Refills: If a fill occurs more than 7 days before the previous supply should have run out:

Early_days = (current_fill_date + current_days_supply) - next_fill_date
If early_days > 7 and pattern is recurring: FLAG as potential hoarding/diversion concern

Escalating Gaps: If gaps between fills are progressively increasing, flag as declining adherence trend.

Step 5: Assess Polypharmacy Burden

Count total active medications from Step 1.

Identify opportunities to reduce pill burden:

• Combination products (e.g., lisinopril/HCTZ instead of separate)
• Once-daily formulations instead of BID/TID
• Therapeutic duplication removal

Step 6: Pull Clinical Outcome Correlations

For key conditions, pull relevant lab values to correlate with adherence:

Diabetes (HbA1c):

Tool: fhir_search
resourceType: "Observation"
queryParams: "patient=[patient-id]&code=4548-4&_sort=-date&_count=4"

Hypertension (BP readings):

Tool: fhir_search
resourceType: "Observation"
queryParams: "patient=[patient-id]&code=85354-9&_sort=-date&_count=6"

Hyperlipidemia (LDL):

Tool: fhir_search
resourceType: "Observation"
queryParams: "patient=[patient-id]&code=18262-6&_sort=-date&_count=4"

Correlate: Poor adherence (PDC <80%) + uncontrolled outcomes (HbA1c >7%, BP >140/90, LDL >100) = strong evidence adherence is the barrier.

Step 7: Apply WHO 5-Dimension Framework

Assess each dimension (see references/adherence-interventions.md):

1. Socioeconomic: Insurance coverage, copay burden, health literacy
2. Health system: Pharmacy access, provider relationship, appointment frequency
3. Condition-related: Symptom burden, depression, cognitive impairment
4. Therapy-related: Side effects, complexity, pill burden
5. Patient-related: Beliefs, motivation, knowledge, forgetfulness

Step 8: Generate Adherence Report

Structure output:

MEDICATION ADHERENCE ASSESSMENT - [Patient Name]
Assessment Date: [date]
Evaluation Period: [start] to [end]

SUMMARY:
- Total chronic medications: [N]
- Polypharmacy status: [Low/Moderate/High/Extreme]
- Medications with PDC >= 80%: [N] of [total]
- Medications with PDC < 80%: [N] of [total]

MEDICATION-LEVEL DETAIL:
| Medication | Fills (12mo) | MPR | PDC | Max Gap | Status |
|------------|-------------|-----|-----|---------|--------|
| Metformin 1000mg | 10 | 85% | 82% | 12 days | Borderline |
| Lisinopril 20mg | 7 | 58% | 55% | 45 days | Non-adherent |
| Atorvastatin 40mg | 11 | 92% | 91% | 5 days | Adherent |

FLAGS:
- Lisinopril: PDC 55% with BP readings averaging 158/94. Adherence likely primary barrier.
- [Early refill pattern on X medication - Y occurrences in 12 months]

CLINICAL CORRELATION:
- HbA1c: 7.8% (metformin PDC 82% - borderline adherence may contribute)
- BP: 158/94 average (lisinopril PDC 55% - strong adherence-outcome gap)

RECOMMENDED INTERVENTIONS:
1. Lisinopril: Explore barriers. Consider once-daily combination product.
2. Metformin: Assess GI side effects (common cause of gaps).
3. General: [motivational interviewing prompt based on identified barriers]

Examples

Example 1: Chronic Disease Adherence Review

User says: "Assess medication adherence for patient 45678 who has uncontrolled diabetes and hypertension."

Actions:

1. Pull active MedicationRequests for patient 45678: finds metformin 1000mg BID, glipizide 10mg BID, lisinopril 40mg daily, amlodipine 10mg daily, atorvastatin 40mg daily.
2. Pull MedicationDispense for each medication over past 12 months.
3. Calculate PDC: metformin 72%, glipizide 68%, lisinopril 45%, amlodipine 81%, atorvastatin 88%.
4. Pull HbA1c (8.9%) and BP (162/98).
5. Identify: lisinopril has 3 gaps >30 days. Glipizide and metformin have intermittent 10-14 day gaps.

Result:

ADHERENCE ASSESSMENT - Patient 45678

PDC BELOW THRESHOLD (<80%):
- Lisinopril 40mg: PDC 45% [CRITICAL - 3 gaps >30 days]
- Metformin 1000mg: PDC 72% [MODERATE - intermittent gaps 10-14 days]
- Glipizide 10mg: PDC 68% [MODERATE - pattern matches metformin gaps]

CLINICAL IMPACT:
- HbA1c 8.9% (goal <7%) - metformin+glipizide non-adherence contributing
- BP 162/98 (goal <130/80) - lisinopril PDC 45% is primary driver

INTERVENTION SUGGESTIONS:
- Lisinopril: Major gap pattern suggests intentional discontinuation. Explore: side effects (cough, dizziness), cost, perceived need.
  MI prompt: "I notice your blood pressure medication has some gaps in fills. What has your experience been with this medication?"
- Metformin/Glipizide: Parallel gap pattern suggests shared barrier. Explore: pharmacy access, insurance renewal timing, forgetfulness.
  MI prompt: "Sometimes it can be hard to keep up with multiple medications. What makes it difficult for you?"
- Consider: Combine lisinopril+amlodipine into single pill. Switch metformin to extended-release (once daily, less GI upset).

Example 2: Polypharmacy and Simplification

User says: "Review medication adherence for elderly patient Mary Johnson, MRN 99887."

Actions:

1. Search Patient by MRN. Find age 82.
2. Pull 16 active MedicationRequests (extreme polypharmacy).
3. Pull dispense records. Calculate PDC for each.
4. Find 7 medications with PDC <80%. 3 medications with no fills in past 90 days.

Result:

ADHERENCE ASSESSMENT - Mary Johnson (Age 82)

POLYPHARMACY: EXTREME (16 active medications)

MEDICATIONS WITH NO RECENT FILLS (>90 days):
1. Omeprazole 20mg - Last fill: 142 days ago
2. Ferrous sulfate 325mg - Last fill: 98 days ago
3. Docusate 100mg - Last fill: 112 days ago

PDC BELOW THRESHOLD:
[... 7 medications listed ...]

SIMPLIFICATION OPPORTUNITIES:
1. Combine lisinopril 20mg + HCTZ 12.5mg into lisinopril/HCTZ combo (eliminate 1 pill)
2. Switch metoprolol tartrate 50mg BID to metoprolol succinate 100mg daily (eliminate 1 daily dose)
3. Omeprazole: no active indication found. Candidate for deprescribing.
4. Docusate: limited evidence for efficacy. Consider discontinuation.
5. Net reduction potential: 16 -> 12 medications

PILL BURDEN ANALYSIS:
Current: 22 pills/day across 16 medications
After simplification: 15 pills/day across 12 medications

Troubleshooting

MedicationDispense not available for this FHIR server

• Not all EMR FHIR endpoints expose MedicationDispense. This is common with EPIC patient-facing APIs.
• Fallback: Use MedicationRequest authoredOn dates as proxy for fill dates, combined with dispenseRequest.expectedSupplyDuration.
• Alternative: Check MedicationStatement for adherence extension if the system populates it.
• Note in the report that dispense-based PDC was unavailable and estimates are based on prescription dates.

Days supply not populated in MedicationDispense

• Calculate from quantity and the dosing instructions in the corresponding MedicationRequest.
• Example: quantity 60 tablets, dose "1 tablet BID" = 30 days supply.
• If dosing frequency is unclear, use common defaults: tablets usually 30-day supply, inhalers 30-day supply, insulin vials 28-day supply.

Patient has medications from multiple pharmacies

• MedicationDispense records may only reflect one pharmacy's data.
• Check if MedicationStatement has additional reported medications.
• Note in the report which medications have verified fill history and which are self-reported only.

Related Skills

• medication-reconciliation - Ensure medication list is accurate before adherence assessment
• prescription-appropriateness-review - Assess if non-adherence is to an appropriate medication
• drug-interaction-checker - Non-adherence to one drug in an interacting pair changes risk profile
• clinical-summary-generator - Include adherence metrics in clinical summaries