Post-Discharge Follow-Up Calls: Evidence, Workflow and ROI

Why do the first days at home matter?

Discharge is a handoff from a setting where every dose is administered and every symptom is watched to one where the patient, and often a family caregiver, is in charge. The first weeks at home are when recoveries go off track. The landmark Medicare analysis found 19.6% of beneficiaries rehospitalized within 30 days of discharge. Roughly one in five patients experiences an adverse event in the first weeks home, and about two-thirds of those events are medication-related.

Those medication problems are the kind a phone call can catch: the prescription never filled, the dosing confusion, the side effect nobody explained. So are the other early signals of trouble, such as a red-flag symptom, a follow-up visit that was never scheduled, or a missing caregiver at home. Hospitals have no near-real-time signal for which patients left without understanding their medications or their follow-up plan. A structured call is the earliest available detector.

That is why it helps to think of the post-discharge follow-up call as a sensor, not just a service. A sensor has to run on every patient to be useful. A call program that reaches the patients who happen to pick up on the first try, and misses the rest, leaves the hospital blind to exactly the group that is hardest to reach.

The timing is not arbitrary either. The nurse-led program behind the most-cited recent result called within three days. One quasi-randomized trial placed its call at 48 hours. Medicare's Transitional Care Management benefit is built around interactive contact within two business days of discharge. When CMS pays for a behavior, it is telling you where it believes the value lies.

What does the evidence say?

Here is the result that has drawn the most attention. At one hospital, nurses called every discharged patient within three days and walked through a short, structured conversation covering medications, warning signs and the follow-up visit. Among patients who did not get a call, 17% were back in the hospital within 30 days. Among those who did, 3.5%. The study (Wright et al.) was published in 2026 in Worldviews on Evidence-Based Nursing, a peer-reviewed nursing journal, and the nurses' script came from a discharge toolkit developed by the federal Agency for Healthcare Research and Quality.

That is one study, at one site, with a quality-improvement design rather than a randomized trial, so it should not be read as the effect any program will get. It matters because it does not stand alone. It sits on top of two decades of transitional-care research pointing the same way, summarized in the table below.

Two caveats belong next to those numbers. First, the strongest trials tested full nurse-led transitional-care programs in which the call was one component, so a call program alone should be expected to capture part of the effect, not all of it. The Du study makes that concrete: ED visits fell, but readmissions on their own did not move. One cluster-randomized trial likewise found that a single call did not move readmissions, and its authors concluded that higher-intensity support was needed. Second, results depend on how reliably patients are actually reached. A 2024 VA systematic review of post-discharge contact programs found that virtually none of the studies it examined checked whether patients had received the contact at all.

Read together, the evidence supports a specific conclusion. A structured call is a sound intervention, and one call is not a program. What works is persistence, structure and follow-through, and those are operational problems before they are clinical ones. The full caveats are in A Phone Call Cut 30-Day Readmissions From 17% to 3.5%.

Why do most post-discharge calls never happen?

The evidence was settled years ago, and nearly every health system has a post-discharge calling protocol on paper. The question that decides whether any of the value is realized is much narrower: what fraction of the eligible cohort is actually reached? That number is usually lower than leadership assumes, and it is also a number the evidence base barely tracks.

Where reach has been measured, the picture is sobering. Manual dialing yields roughly one real conversation per four attempts. One 100-nurse team reached only a quarter of its patients. Clinical leaders at GuideWell described a dedicated nurse team whose only job is following up with high-risk members, and 75% of those members were never reached. Not for lack of care or effort, but because nurses spend their days calling, not connecting. Even under study conditions, with a research team making at least five attempts per patient, one trial reached 69% of its intervention group. That is close to the ceiling. A discharge unit working a daily list does not get five attempts per patient.

The reasons are operational. Persistent dialing competes for the scarcest resource in healthcare: nursing time. Most of that time goes to work that consumes nursing capacity without using nursing skill:

• Dialing and redialing numbers that do not pick up.
• Leaving voicemails, and tracking which patients were left one.
• Retrying at different times of day, inside the window that matters.
• Documenting each attempt and its outcome, a record-keeping gap that EHRs have never handled well.

Why do the calls that happen go unanswered?

The second failure point is the phone network. A hospital can place a call that is consented, clinically approved and fully compliant, and the patient's phone can still label it "Spam Likely" and send it to voicemail. As far as the hospital knows, the outreach program is running. As far as the patient knows, no one ever called.

Carriers and their analytics partners score calls in real time on traffic patterns, complaint data and number reputation, and calls that score badly are labeled or blocked before they ring. Industry data suggests roughly half of outbound calls are never answered at all, with only about a quarter of dials reaching a live person, and Hiya's State of the Call 2026 found that 86% of unidentified calls now go unanswered. Organizations are not reliably informed which calls were blocked, or why.

Worse, the ordinary behavior of a follow-up program looks like spam to an analytics engine: high volume from one number, short call durations because most calls reach voicemail, and repeated attempts to numbers that never answer. A program can degrade its own number reputation simply by running as designed. A blocked post-discharge call means an unfilled prescription or an unreported symptom that nobody hears about. See The Call Was Compliant. It Still Never Rang. and the Getting Through research summary.

This is the first step of Consig's post-discharge workflow. Consig handles dialing patterns, number reputation and branded caller ID through carrier relationships, so the call shows up as the patient's own hospital rather than an unknown number. Once the patient answers, identity is confirmed with passkeys, device biometrics or one-time passcodes before any health information is discussed.

What should the call cover?

The research gives a consistent answer: a short, structured conversation. In the Wright et al. program, nurses walked through medications, warning signs and the follow-up visit, using a script from an AHRQ discharge toolkit. Structure is what makes the call a reliable sensor. The same questions, asked the same way, on every call, produce answers that can be compared, escalated and audited.

A protocol your clinical team approves would typically cover the following, with each step designed so that an answer either closes the item or triggers a defined escalation:

• Identity and consent. Confirm the right person answered before any health information is discussed, and capture consent where the workflow requires it.
• Medications. Whether each prescription was filled, whether the patient understands the dosing, and whether there are side effects nobody explained. This is the category behind about two-thirds of early adverse events.
• Warning signs. A structured red-flag screen, with the answers that require a clinician defined in advance.
• The follow-up visit. Whether it is scheduled, and whether the patient can get there.
• Support at home. Whether a caregiver is present and whether anything needed at home is missing.
• Instruction read-back. Confirming the patient understood the clinical instructions they were given.
• Care gaps. Calls surface more than the problem they were designed for; in Du et al., 40% of completed calls surfaced at least one care gap.

What should automation do, and what must a clinician do?

Large language models have changed what an automated call can be. A system can now ask an open-ended question, interpret a free-form answer and decide what to do next. That capability arrived just as the staffing arithmetic made purely nurse-led calling hard to sustain. Which raises the question to ask first: where exactly is the line?

The right split is specific. The automation handles the mechanics of connection and structured intake: it dials, verifies identity, walks the structured protocol, collects the patient's answers and documents the call. It does not assess the patient, interpret symptoms or give medical advice. Clinical judgment stays with clinicians. What the automation removes is the dialing, the voicemail, the retries and the documentation.

Within the automated part, match the conversational technology to the step. Identity verification, consent capture, red-flag screening and clinical-instruction read-back have to happen the same way on every call, so they run as deterministic, scripted flows. The parts that benefit from sounding natural, such as acknowledging a frustrated patient or finding a convenient callback time, can use generative AI. A well-built system chooses deliberately, step by step, rather than letting a probabilistic model improvise a legally or clinically binding moment. More in What AI Should (and Should Never) Do on a Patient Call.

This is how Consig's post-discharge workflow is built. The call rings with trust through branded, reputation-managed numbers; it authenticates the patient; it runs the check-in on medications, warning signs and the follow-up visit in deterministic, scripted stages; and it escalates answers that match a rule to a nurse, with the call's audio and structured answers attached. Each attempt and its outcome is documented and returned to your platform.

How should escalation work?

Escalation is where the safety of an automated program lives. When an answer matches an escalation rule, the call goes to a person: live within seconds for an urgent red flag, or into the care team's own follow-up queue, with the call's full context attached, for everything else. The second principle is as important as the first: every escalation needs an owner. A flagged finding that reaches no one is worse than no program at all.

That handoff should be a warm handoff, in which the reason for the transfer, the answers collected so far and the audio travel with the patient, so the patient does not have to start over. A cold transfer drops the patient into a queue with no context. Decide the following before the first call is placed:

• Which answers require a person, written as explicit rules your clinical team signs off on.
• Which of those are urgent (a live transfer within seconds) and which go to a follow-up queue.
• Who owns each escalation type, and how quickly they must respond.
• What happens after hours, on weekends and when the owning team is at capacity.
• What travels with the handoff: at minimum the call's audio and structured intake.
• How a closed escalation is recorded, so the loop is visibly closed rather than assumed.

What do TCM and HRRP mean for the economics?

A follow-up call costs a few dollars to place. The value it protects flows through four mechanisms, all concentrated in the narrow window after discharge. The Slam-Dunk Case for Post-Discharge Phone Calls walks through them in detail.

The readmission itself. AHRQ's Healthcare Cost and Utilization Project puts the average 30-day all-cause adult readmission at $16,300, more than the $14,500 the admission before it cost. For a hospital or a risk-bearing plan, one avoided readmission pays for a very large volume of outreach.

The penalty on top. Under the Hospital Readmissions Reduction Program, CMS reduces base Medicare payments by up to 3% for hospitals with excess readmissions across six conditions and procedures: acute myocardial infarction, heart failure, pneumonia, COPD, coronary artery bypass graft surgery, and elective primary total hip or knee arthroplasty. The cap is 3% of base operating DRG payments, and because it applies to base payments for all of a hospital's Medicare discharges, a penalty triggered by a handful of conditions affects revenue broadly. In FY2026 roughly 78% of the hospitals evaluated carry a penalty of some size, and the number penalized 1% or more rose to 240, up from 208 the year before. Critical access hospitals are excluded.

The revenue left on the table. Transitional Care Management reimburses about $201 under CPT 99495 and about $273 under CPT 99496. Both require interactive contact with the patient within two business days of discharge, by phone, electronic communication or in person, by the practitioner or clinical staff under their direction. 99495 covers moderate-complexity decision making with a face-to-face visit within 14 calendar days; 99496 covers high-complexity decision making with a visit within 7 days. Medication reconciliation must be done no later than the visit, and only one practitioner may report TCM in the 30-day service period. CMS guidance allows the service to be reported when two or more separate, timely attempts at contact are unsuccessful and documented and the other requirements are met, which makes recording every attempt part of the workflow. A cohort not reached inside the window is a clinical gap and unbilled revenue.

The quality-measure overlay. Post-discharge contact feeds the measures behind Star Ratings and shared-savings performance. It also feeds HCAHPS, where communication about medicines and discharge information score persistently among the lowest areas of the survey.

How do you estimate ROI? (worked illustration)

This is an illustration, not a claim of results. It shows how the published figures combine, using hypothetical inputs that you should replace with your own. It does not predict how many readmissions any program will avoid. The 17%-to-3.5% result comes from one site and should not be used as an expected effect; the strongest trials tested full programs, and a call alone should be expected to capture part of the effect.

Readmissions. At AHRQ's average of $16,300, each avoided 30-day readmission is worth about $16,300 to whoever bears the cost. Ten avoided readmissions would be about $163,000; twenty, about $326,000. The useful question is the break-even: divide your program's annual cost by $16,300 to find how many avoided readmissions it needs to pay for itself on this mechanism alone.

TCM revenue. Assume, hypothetically, that a program reaches 100 additional eligible patients a month inside the two-business-day window, and that those patients go on to meet every other TCM requirement. At about $201 each under CPT 99495, that is about $20,100 a month, or $241,200 a year. If the same 100 qualified under CPT 99496 at about $273, it would be $27,300 a month, or $327,600 a year. Whether any contact satisfies TCM billing requirements is a determination for your billing and compliance team.

HRRP exposure. Assume a hypothetical hospital with $100 million in base operating DRG payments. The HRRP cap of 3% puts its maximum exposure at $3 million a year; a 1% reduction would be $1 million. Readmission performance on the six tracked conditions moves where the hospital lands, which is why reach on those cohorts deserves particular attention.

Reach is the multiplier. Every figure above is realized only for patients who are actually reached. At roughly one real conversation per four manual attempts, reaching 1,000 patients takes on the order of 4,000 dials before any clinical work begins. That dialing load is the cost automation is meant to take off nurses.

How should health plans think about it?

For a risk-bearing plan, an avoided readmission is an avoided cost, and post-discharge contact is one of several outreach programs that move both cost and quality. Preventive care and closed care gaps lift the measures that drive reimbursement, including HEDIS and Star Ratings, and many HEDIS gaps are at root outreach gaps: a member who never schedules a screening or a follow-up visit after hospitalization. Medication adherence, a common contributor to avoidable complications and readmissions, feeds three heavily weighted Part D Star Ratings measures.

The reach problem is the same one hospitals face. The GuideWell example, in which 75% of the high-risk members a dedicated nurse team was meant to follow up with were never reached, came from a plan setting. Well-intentioned outreach also carries calling-law risk: OptumRx settled for about $1.86 million over medication-adherence calls placed to wrong numbers, a consent and data-hygiene gap rather than bad intent. A single member call can implicate TCPA, HIPAA, Medicare contact rules, Do Not Call and state law at once.

The FCC's healthcare exemption lists post-discharge follow-up intended to prevent readmission among its covered call types, but its conditions are narrow: calls only to the wireless number the patient provided, generally one minute or less, no more than one call a day and three a week from a specific provider, and an easy opt-out honored immediately. The FCC treats an AI-generated voice as an artificial voice, so an AI call relying on the exemption must meet the same conditions. Retry cadence is therefore a compliance decision as well as an operational one.

One of Consig's early customers needed to reach more than a million patients proactively every year. Consig automated the painful 80% of that outreach (voicemails, follow-ups, call screening, incomplete interactions) so staff could focus only on the 20% of calls that led to meaningful clinical conversations. The customer reallocated 10 FTEs to the calls that need a clinical conversation and saves $700,000 a year. See customers and Consig for health plans.

How do you measure the program?

The most common gap in post-discharge programs is measurement of reach itself: the 2024 VA review found virtually none of the studies it examined checked whether patients received the contact. A program that reports only calls dialed cannot tell whether the protocol is running or merely scheduled. Measure the funnel from discharge to closed escalation, and segment it by unit, condition and payer so that gaps on HRRP-tracked conditions and TCM-eligible cohorts are visible.

• Reach rate inside the window: the share of the eligible cohort with an interactive contact within two business days, not the share dialed.
• Attempts per patient and attempt documentation: how many separate, timely attempts were made and recorded for patients not reached.
• Answer rate by number and carrier: an early warning that calls are being labeled or blocked before they ring.
• Protocol completion: the share of connected calls that complete every structured step.
• Escalations raised and closed: volume by type, time to response against the owner's target, and after-hours handling.
• Findings surfaced: unfilled prescriptions, red-flag symptoms, unscheduled visits and other care gaps.
• Downstream outcomes: ED visits and 30-day readmissions, read with the caveat that a call alone captures part of a full program's effect.
• Financial capture: TCM claims submitted against the TCM-eligible cohort, and staff hours moved from dialing to clinical conversations.