When a newly diagnosed Type 2 diabetic misses their first metformin refill, the reason is almost never "they forgot." Month two or three post-diagnosis is when the side effects peak — GI distress, nausea, the fatigue that doesn't match what the doctor described. Patients stop filling the prescription because the medication feels worse than the diagnosis did.
When a patient who has been on metformin for 18 months stops filling, the calculus is entirely different. They've probably tolerated the drug fine for over a year. Now something else is happening: a coverage change, a life disruption, the beginning of a quiet decision to see if they "really need it."
Both patients show up the same way in your EHR — overdue on a refill. The intervention that works for one will fail for the other. Getting this distinction right is the operational challenge that generic patient outreach programs miss entirely.
The Adherence Cliff at Months 3–6
Type 2 diabetes medication adherence follows a predictable degradation pattern. The period from month three to month six post-prescription is a high-risk window that shows up repeatedly in pharmacy refill data. The initial prescription gets filled, often a second fill as well, and then the refills become irregular or stop.
For metformin specifically, this correlates with the medication's well-documented GI tolerance period. Most patients adapt within 4–6 weeks on standard release formulations, but some take longer — and patients who don't receive explicit counseling that the GI symptoms are temporary often conclude the drug isn't for them.
The intervention that works here isn't a reminder SMS. It's a clinical touchpoint — a message that specifically addresses the GI side effect question, ideally from a care coordinator or nurse who can actually answer "should I be pushing through this or talk to my doctor?" A simple automated refill reminder will be ignored by a patient who is experiencing daily nausea and has already decided the medication is intolerable.
The channel choice follows directly from the intervention content. If the message needs to carry clinical substance, it needs to either be a phone call or an SMS that invites a phone call. "Your metformin refill is due — click here to refill" is the wrong message for this patient at this time.
The Drift Pattern at Months 12–24
Long-tenured patients on a stable regimen like metformin 500mg twice daily present a completely different non-adherence signature. They've been filling on time for a year or more. Then refill intervals start lengthening — 35 days, then 40, then they miss a fill entirely. This is drift, not intolerance.
Drift has different root causes: benefit year rollovers that reset their deductible and make the refill suddenly feel expensive, a life event that disrupted their routine, or the classic "feeling fine" reasoning where controlled HbA1c leads the patient to question whether they still need the medication.
The intervention here can be lower-touch because you're not managing a clinical barrier — you're managing inertia. A well-timed SMS that acknowledges their history ("You've been managing your diabetes well — your refill is coming up, here's the quickest way to get it") is often sufficient. You don't necessarily need a phone call. You need to catch the drift before it becomes a two-month gap, because a two-month gap in metformin shows up as HbA1c deterioration that requires a follow-up lab, a provider visit, and often an escalation of the medication regimen.
The practical implication: the patient who should get a phone call today is not the patient with the longest gap — it's the patient whose gap pattern is inconsistent with their history. The 18-month adherent patient who has now missed two refills in a row is the high-priority call, not the newly-diagnosed patient who missed their very first refill.
Where the HEDIS Measure Creates the Wrong Incentive
The HEDIS Comprehensive Diabetes Care (CDC) measure set focuses on process and outcome metrics: HbA1c testing, eye exam completion, BP control, statin use. Medication adherence as a continuous measure isn't captured directly in CDC — it shows up indirectly through the Medication Management for People with Asthma (MMA) equivalent, and through PDC (proportion of days covered) calculations that most health plans track for diabetes medications.
Here's the tension: a health system managing to HEDIS CDC will prioritize getting the annual HbA1c done and the eye exam scheduled. Those are the closure events that count in the measure. The refill gap that happens between the annual encounters is invisible in the measure set — it only surfaces when the HbA1c comes back elevated, at which point the care gap has already cost the patient and the plan.
We're not saying HEDIS is the wrong target. The CDC measure set captures genuinely important care quality signals. But optimizing for measure closure events — annual visits, annual labs — without monitoring the between-encounter adherence pattern means you're watching the annual checkpoints while the medication coverage gap quietly accumulates.
The PDC threshold for "adherent" on diabetes medications is typically set at 80% by most pharmacy benefit managers and CMS quality programs. A patient at 78% PDC is functionally non-adherent by that standard, but they might have gone 90 days without a refill and 60 days without a care touchpoint before anyone noticed.
A Concrete Scenario: The Insulin Transition
Consider a patient population scenario we work through with care teams regularly: Type 2 diabetic patients who have recently been transitioned from oral-only regimens to basal insulin — typically glargine or detemir — added on top of their existing metformin and possibly a GLP-1 agonist like semaglutide.
Insulin initiation is a high-anxiety event for patients who have been managing with oral medications for years. Injection technique, hypoglycemia fear, refrigeration requirements, and the psychological weight of "needing insulin" all create adherence barriers that have nothing to do with forgetting. The refill data alone won't tell you which barrier is operating for which patient.
But the signal pattern helps narrow it. A patient who fills the insulin glargine on time but whose metformin refill interval has lengthened since the transition is likely trying to simplify their regimen — they may be under the impression they no longer need the metformin now that they're "on insulin." That's a specific clinical education gap that warrants a targeted call.
A patient who hasn't filled either medication in 45 days is in a different situation entirely — possible coverage disruption, possible financial barrier, possible complete disengagement from the care plan. The channel and message are completely different.
Patientrig's signal matching does exactly this — it looks at the combination of which medications are overdue, how long, and what the prior adherence pattern looked like to triage which patients need a clinical call versus an automated nudge versus a pharmacy outreach.
Channel Selection Is a Clinical Decision
In care management, we sometimes treat channel selection as an IT or preference question. "What does the patient prefer — SMS or phone?" That framing misses the point for medication adherence interventions.
Channel selection is a clinical decision because the appropriate message content determines the appropriate channel. If the message requires two-way dialogue — because you're trying to assess whether the patient is experiencing side effects, or whether they understand the dosing change — SMS is the wrong channel regardless of patient preference. You can open the conversation with an SMS, but you need a follow-up pathway to a clinical voice.
If the message is a well-timed logistical nudge to a previously adherent patient, SMS is often exactly right. It's low friction, it arrives at the right moment, and it doesn't require a 15-minute phone call from a coordinator who has 40 other patients to reach that day.
The rule we've found most useful in practice: if the expected response is "yes, I'll refill it" — SMS. If the expected response requires the patient to explain something, ask a question, or make a clinical decision — phone call, or SMS with an explicit invitation to call back. The patients who need to explain something won't do it via text. They'll either ignore the text or reply "OK" and not refill.
What This Requires Operationally
Getting medication adherence intervention right for a diabetic panel requires a few things that most health system outreach programs haven't built yet:
- Pharmacy claims or PBM feed integration — not just EHR prescription records. EHR records tell you what was prescribed; PBM claims or retail pharmacy data tells you what was actually dispensed. A patient can have 12 active prescriptions in the EHR and not have filled 4 of them in 90 days. Without the fill data, you're flying blind on adherence.
- Adherence history as a feature, not just a flag — the difference between "overdue" and "adherence drift from a 20-month streak" is operationally significant. Both patients show up as having a fill gap. Only one of them has a pattern that suggests the gap is anomalous.
- Channel routing logic tied to intervention type — automated, low-touch outreach for low-complexity nudges; clinical voice outreach for barriers that require dialogue. Not based on patient demographics alone, but on what the specific adherence signal suggests the barrier is.
- Closed-loop tracking back to refill completion — knowing whether the outreach actually resulted in a refill, not just whether the patient responded to the message. Response rate is a vanity metric. Refill completion is the outcome.
Diabetes medication adherence is one of the highest-leverage places in chronic disease management to apply structured outreach logic, because the patient panel is large, the medication regimens are relatively predictable, and the signal-to-noise ratio in pharmacy data is good. The problem isn't data availability — it's building outreach workflows that are actually calibrated to what the data shows.
