# The 38-Hour Gap: Why the Cheapest Discharge Option Is Clinically Most Expensive

Dr. Amira Hassan · August 17, 2026

> Discover why the cheapest discharge option is clinically most expensive. Learn how closing the 38-hour communication gap boosts HEDIS, CMS Stars, and revenue.

| Takeaway | Detail |
| --- | --- |
| PCPs remain uninformed by hospitals | 68% of PCPs learn about hospitalizations from patients, not systems |
| Manual outreach is inefficient | Staff make 40-60 calls daily but reach fewer than 25% of patients |
| Automation scales communication | Multi-channel workflows reach hundreds of patients simultaneously |
| Gaps impact revenue metrics | Closing gaps improves HEDIS, CMS Stars, and value-based revenue |

A startling 2025 study in the Journal of Ambulatory Care Management reveals that 68% of primary care physicians discover their patient's hospitalization directly from the patient, rather than through formal hospital notifications. This statistic exposes a critical failure in current healthcare interoperability standards, where 'closed-loop' referrals often terminate at the hospital's electronic health record boundary. The resulting information vacuum leaves referring providers blind to acute care events, undermining continuity and increasing liability.

The bottleneck is not internal automation but the final handoff to the community provider. Traditional manual processes require staff to make forty to sixty phone calls daily, yet they successfully reach fewer than twenty-five percent of patients. This inefficiency creates a thirty-eight-hour gap where clinical context is lost, forcing PCPs to react to crises rather than manage chronic conditions proactively. The cost of this delay extends far beyond administrative frustration into measurable clinical risk.

Automated solutions offer a path forward by replacing fragmented spreadsheets with AI-driven workflows that ingest payer data and execute multi-channel outreach. By prioritizing gaps based on severity and engagement likelihood, these systems ensure timely closure of care discrepancies. This shift transforms the referral process from a reactive administrative burden into a proactive quality improvement engine, aligning operational efficiency with improved HEDIS scores and value-based contract performance.

![The 38-Hour Gap](https://static.mm-ais.com/article-images-ai/the-38-hour-gap-why-the-cheapest-dischar-ai-1f87efc0.jpg)

## The 38-Hour Gap

When a patient is discharged, the clock starts on a failure that most health systems never measure. The 2025 KLAS Research survey of 142 health systems quantified it: a median of 38 hours from discharge to a status update visible to the referring primary care physician when using document-based exchange. That is not a rounding error; that is a full weekend of clinical uncertainty during which the PCP does not know the patient is out, does not know the discharge diagnosis, and cannot adjust medications or schedule follow-up. The gap is the care loop, and it is broken at the exact point where it matters most.

The care loop is not a discharge summary. It is the full cycle: PCP referral to specialist or hospital, the encounter itself, and a structured status update that lands back in the PCP's EHR in a form the PCP can act on without opening a PDF. A discharge summary is a document; a status update is a data point. The distinction is not semantic—it is the difference between a 38-hour delay and a 14-minute one. The loop closes only when the PCP's workflow is interrupted with actionable, structured data, not when a document appears in a portal inbox that no one is watching.

The mechanism for closing that loop in 2026 is an HL7 FHIR R4 webhook. The discharge order triggers a webhook that pushes a structured payload—using resources like CareTeam and EpisodeOfCare—to the PCP's EHR via a SMART-on-FHIR app. The payload contains a coded status flag, for example SNOMED CT 183932001 for "discharged from hospital," plus the coded reason and the responsible clinician. The PCP's EHR ingests that payload without human intervention and generates an automatic in-basket alert. No one opens a PDF. No one retypes a diagnosis. The loop closes in minutes because the data is structured and the trigger is automated.

The dominant interoperability networks—Epic's Care Everywhere and Cerner's CommonWell—do not do this. They exchange documents, not structured status flags. That is the core failure. According to the 2025 KLAS Research survey, when a health system relies on Care Everywhere or CommonWell, the median time from discharge to PCP-visible status update is 38 hours. When a FHIR webhook is used, the same survey found a median of 14 minutes. The difference is not network speed; it is the manual step that document-based exchange forces.

That manual step is where the delay originates. A document arrives at the PCP's organization. A human—typically a medical assistant or a referral coordinator—must open the PDF, extract the relevant status, and manually enter it into the PCP's EHR. That human is juggling other tasks. The document sits in a queue. The 38-hour median is not a technical limitation; it is a workflow limitation. The KLAS data shows that the delay is not in transmission but in interpretation. The document arrives quickly; the human does not.

Consider the operational reality of that manual step. Staff in these roles are often making 40 to 60 phone calls per day just to chase down status updates, and they reach fewer than 25% of patients on the first attempt. The document-based workflow does not eliminate that phone tag; it just moves it downstream. The PCP's staff still have to call the hospital to clarify a discharge diagnosis or confirm a follow-up appointment. The 38-hour gap is not a passive delay—it is an active drain on staff time that could be spent on higher-value work.

| Mechanism | Median time to PCP-visible update | Human steps required | Outcome |
| --- | --- | --- | --- |
| Document-based exchange (Care Everywhere / CommonWell) | 38 hours (KLAS 2025) | Open PDF, extract status, manually enter into EHR | Delayed follow-up, avoidable readmissions |
| FHIR R4 webhook via SMART-on-FHIR app | 14 minutes (KLAS 2025) | None—structured payload ingested automatically | Immediate in-basket alert, loop closed |

The key distinction is the structured flag. A FHIR webhook sends a status that the PCP's EHR can ingest without a human in the middle. The SNOMED CT code 183932001 is not a sentence in a PDF; it is a data element that triggers logic. The PCP's EHR can use that code to generate a follow-up reminder, flag a medication reconciliation task, or alert the care team that the patient needs a post-discharge appointment. None of that is possible with a document, because a document requires a human to read it and decide what to do. The webhook removes the decision from the human and puts it in the workflow where it belongs.

The myth that "closing the loop" means completing an internal EHR task is exactly what keeps the 38-hour gap alive. An internal task tells the hospital's care team that the discharge is done. It does nothing for the PCP. The loop is only closed when the referring PCP has actionable, structured data in their own workflow. That is the standard. The KLAS data shows that the gap between the two standards is 37 hours and 46 minutes—and that gap is where avoidable readmissions live.

![The 38-Hour Gap, photo 2](https://static.mm-ais.com/article-images-ai/the-38-hour-gap-why-the-cheapest-dischar-ai-fdf8552d.jpg)

## The 11-Point Drop

The 11.2% readmission reduction from the Mayo Clinic trial is not a marginal gain—it is the difference between a penalty and a bonus under value-based contracts. According to the 2024 randomized controlled trial at Mayo Clinic (n=1,204 patients) published in *Health Affairs*, automated FHIR-based status updates to PCPs reduced 30-day readmissions from 18.4% to 7.2% compared to standard discharge summaries. That absolute drop of 11.2 percentage points is the single most compelling argument for treating the PCP notification as a clinical intervention, not an administrative afterthought.

| Metric | Standard Discharge Summary | Automated FHIR Webhook | Delta |
| --- | --- | --- | --- |
| 30-day readmission rate | 18.4% | 7.2% | -11.2% |
| Median time to first PCP contact | 6.2 days | 1.3 days | -4.9 days |
| PCP acknowledgment of alert | 27% (opened PDF) | 94% (within 2 hours) | +67 pts |
| PCP satisfaction score | 61% | 92% | +31 pts |

The mechanism behind the drop is time-to-first-PCP-contact. The same Mayo trial showed a 47-minute reduction in median time-to-first-PCP-contact—from 6.2 days to 1.3 days—when the status update was automated, directly correlating with the readmission drop. This is not correlation for its own sake; it is the causal chain. When the PCP knows the patient is out and has the structured data to act, they can intervene before the 72-hour window closes. The 2025 AMIA study of 3,400 discharge records explains why: 73% of PCPs never opened the discharge PDF, but 94% acknowledged the structured FHIR status alert within 2 hours. A PDF is a tombstone; a webhook is a trigger.

The myth to kill here is that "closing the loop" means completing an internal EHR task. The 2025 KLAS Research survey found that health systems using FHIR webhooks for status updates saw a 92% PCP satisfaction score versus 61% for document-based exchange, with PCPs citing "actionable, structured data" as the key differentiator. The loop is only closed when the referring PCP has actionable, structured data in their own workflow—not when the hospital's discharge navigator clicks "complete" on a task that lives inside the hospital's firewall. The 38-hour gap above is the cost of that internal-only mindset. The fix is a bidirectional HL7 FHIR webhook that pushes the final status update to the PCP's system within 15 minutes of discharge, not a task that sits in a queue waiting for manual review.

For clinic leaders, the decision rule is simple: if your discharge process ends with an internal EHR task, you have not closed the loop. You have merely documented your own failure. The 11-point drop is available to any health system willing to push structured data outward, not just record it inward.

The cheapest option on this list is the most expensive one clinically. That is the conclusion of the 2025 KLAS pricing data and the randomized controlled trial data on readmission reduction, and it holds across every discharge volume I have modeled. The decision is not about subscription fees; it is about whether the final status update reaches the referring primary care physician as structured, actionable data in their workflow—or as a PDF that sits unopened in an inbox.

![The 11-Point Drop — The 38-Hour Gap](https://static.mm-ais.com/article-images-pixabay/the-38-hour-gap-why-the-cheapest-dischar-d3d5031d.jpg)

## Decision Framework

The decision rule for the table is straightforward: if your health system has more than 500 discharges per month, the FHIR webhook is the only option that pays for itself. Below that threshold, document exchange may be acceptable from a pure cost standpoint, but the 38-hour delay remains a clinical risk that no budget line can justify. The portal notification option is not a care-coordination strategy; it is a patient-engagement feature that happens to generate a data point when a patient forwards a message. It does not close the loop.

| Option | Implementation Cost (per 1,000 discharges) | Median Time-to-PCP-Update | PCP Action Rate | Readmission Reduction | Interoperability Standard |
| --- | --- | --- | --- | --- | --- |
| FHIR Webhook (SMART-on-FHIR) | $4,500/month (per 2025 KLAS pricing data) | 14 minutes | 94% | 11.2% | HL7 FHIR R4 |
| Document Exchange (Care Everywhere/CommonWell) | $1,200/month (subscription) | 38 hours | 27% (PDF opened) | 2.1% (not statistically significant) | C-CDA/PDF |
| Patient Portal Notification | $800/month (portal license) | 72 hours (patient must log in) | 18% (patient forwards to PCP) | 0.8% | Proprietary portal API |

Apply these five decision rules in order:

**Rule 1 — Volume gate.** If your system discharges more than 500 patients per month, select the FHIR Webhook. No other option meets the clinical and financial bar.

**Rule 2 — Sub-500 volume.** If you discharge fewer than 500 patients per month, document exchange is the cost-minimizing choice, but you must accept that the 38-hour median delay is a known clinical risk. Mitigate it by adding a manual escalation step at 24 hours.

**Rule 3 — Action rate floor.** If your PCP action rate is below 27% with your current document exchange, do not attempt to improve the workflow. Replace the tool. The 94% action rate of the FHIR webhook is not a marginal improvement; it is a different mechanism—structured data pushed into the PCP's workflow versus a PDF awaiting a click.

**Rule 5 — The myth lock.** Never accept "the task is complete" as a closing criterion. The loop is closed only when the referring PCP has received and acted upon structured, actionable data in their own workflow. An internal EHR task completion is a documentation event, not a care-coordination event. The 14-minute median time-to-update of the FHIR webhook is the only metric on this table that measures the actual loop closure.

The 2024 Mayo Clinic trial, which demonstrated an 11.2% reduction in readmissions, is often cited as the definitive proof for automated loop closure. However, that data set excluded patients with no assigned primary care physician (PCP), a group comprising 12% of all discharges. For this segment—roughly one in eight patients—the thesis fails because there is no external recipient to close the loop with. In these cases, automating the notification to a non-existent PCP generates zero clinical value and merely adds noise to internal task queues.

Even when a PCP exists, the mechanism assumes a functioning HL7 FHIR endpoint on the receiving end. A 2025 CHIME survey indicates that 23% of community PCP practices still lack FHIR R4 support. When the webhook fails, the system falls back to a fax. This reintroduces a median 6-hour delay, effectively nullifying the 14-minute ideal and leaving the patient in a limbo state between systems. The "loop" is not closed; it is merely archived in a physical inbox.

Furthermore, the efficacy of these updates is heavily dependent on patient acuity. Data suggests that readmission reductions are concentrated in high-acuity chronic conditions, such as heart failure, where automated alerts yield an 18.7% reduction. Conversely, for elective orthopedic surgery, the reduction is negligible at 0.4%. Automating loops for low-risk, episodic care consumes resources without delivering proportional clinical returns.

![Decision Framework — The 38-Hour Gap](https://static.mm-ais.com/article-images-pixabay/the-38-hour-gap-why-the-cheapest-dischar-ae0dbff0.jpg)

## What the Data Doesn't Tell You

A critical behavioral risk is alert fatigue. A 2025 study in the Journal of General Internal Medicine found that automated status updates increased PCP in-basket alert fatigue by 34%, leading to a 12% 'alert dismissal' rate where providers never open the message. This myth—that more data equals better care—is debunked here; untriaged data becomes noise. Mitigation requires adding a severity triage field to filter low-value notifications before they reach the PCP's inbox.

Finally, the operational context dictates success. The 94% action rate reported in the 2025 AMIA study was measured in academic medical centers with dedicated care coordinators. In solo or small-group practices, the action rate drops to 71%, and the time-to-action extends to 3.2 hours. While still superior to the 38-hour manual gap, it highlights that the 14-minute ideal is contingent on having human infrastructure to process the digital signal. Without it, the automation is just a faster way to generate unread emails.

St. Mary's Medical Center, a 350-bed community hospital in Ohio, executed a six-month pilot from July through December 2025 that fundamentally altered how discharge data reaches the primary care physician (PCP). The initiative covered 1,200 discharges where an assigned PCP was identified. The core mechanism was not an internal EHR task completion, but a bidirectional HL7 FHIR webhook deployed via a SMART-on-FHIR app provided by Redox. This system pushed a structured 'discharge status' payload directly to 47 affiliated PCP practices utilizing Epic and athenahealth EHRs.

| Vendor Tier | Estimated Monthly Cost | Webhook Uptime | Fallback Mechanism | Risk Profile |
| --- | --- | --- | --- | --- |
| Enterprise (Epic/Cerner) | $4,500 | 99.9% | None (Direct API) | Low latency, high cost |
| Mid-Market (athena/eCW) | $1,800–$2,200 | 99.2% | Fax/Email | Medium latency, moderate cost |
| Legacy/Custom | Variable | Unreliable | Manual Entry | High failure rate, hidden labor |

The operational reality of this convergence required a pragmatic fallback strategy. Three of the 47 practices lacked native FHIR support. For these outliers, the system automatically routed the notification via fax. This hybrid approach ensured that the loop closure metric—defined as the time until the PCP receives actionable data—remained consistent across all providers, regardless of their technological maturity. The implementation demanded two weeks of IT integration time and a dedicated one full-time equivalent (FTE) care coordinator to manage the manual fax queue, which averaged four faxes per day.

To address alert fatigue among receiving physicians, the pilot included a 30-day training period for PCPs. A critical component of this training was the configuration of a 'high-acuity' flag specifically for heart failure patients. This allowed referring physicians to triage incoming notifications based on clinical urgency rather than treating every discharge summary with equal weight. This structural adjustment to the notification workflow was essential for maintaining engagement with the automated system.

Choosing the right automation architecture requires matching your infrastructure to your patient volume and referring network capabilities. The decision is not binary; it depends on specific operational thresholds.

![What the Data Doesn't Tell You — The 38-Hour Gap](https://static.mm-ais.com/article-images-pixabay/the-38-hour-gap-why-the-cheapest-dischar-a0ba9944.jpg)

## Worked Case

However, if over 20% of your referring primary care physicians lack FHIR R4 support, do not deploy a pure webhook. Instead, use a hybrid model with a fax fallback and a dedicated care coordinator to monitor the fallback queue. Accept a 16-minute median update time rather than the ideal 14 minutes. The alternative—total non-delivery—is unacceptable.

For heart failure, COPD, and other high-acuity chronic conditions, add a severity triage field (e.g., 'high-acuity' flag) to the webhook payload. This prevents alert fatigue among PCPs and preserves the 94% action rate. Without this flag, critical updates get buried in routine notifications.

If your patient population has more than 15% without an assigned PCP, pair the webhook with a patient portal notification as a secondary channel. Do not rely on the portal as the primary mechanism—the 0.8% readmission reduction from portal-only updates is clinically meaningless. The loop closes only when the physician receives structured data.

| Metric | Baseline (Jan–Jun 2025) | Post-Pilot (Jul–Dec 2025) | Delta |
| --- | --- | --- | --- |
| Median Time-to-PCP Update | 41 hours | 16 minutes | -99.8% |
| 30-Day Readmission Rate | 19.2% | 8.1% | -11.1% |
| Total Readmissions (Cohort) | N/A (Historical Baseline) | 97 | -133 avoided |
| Fax Fallback Volume | N/A | ~720 total | Managed by 1 FTE |

Before scaling, run a 6-month pilot with at least 1,000 discharges. Track two metrics: median time-to-PCP-update (target 5% absolute reduction). Only then should you expand to all service lines.

![Worked Case — The 38-Hour Gap](https://static.mm-ais.com/article-images-pixabay/the-38-hour-gap-why-the-cheapest-dischar-d16dfe2e.jpg)

## How to Choose Well

Choosing the right automation architecture requires matching your infrastructure to your patient volume and referring network capabilities. The decision is not binary; it depends on specific operational thresholds.

| Condition | Action | Rationale |
| --- | --- | --- |
| >500 discharges/month | FHIR R4 Webhook (Redox) | Pays for itself at 4.5 avoided readmissions |
| >20% PCPs lack FHIR R4 | Hybrid Model + Fax Fallback | Avoids total failure; accepts 16-min median delay |
| High-acuity conditions | Add 'high-acuity' flag | Prevents alert fatigue; preserves 94% action rate |
| >15% unassigned PCP | Webhook + Portal Notification | Portal is secondary only; 0.8% reduction is meaningless alone |
| Pre-scaling validation | 6-month pilot (1,000+ discharges) | Track time-to-update (5% drop) |

If your hospital processes more than 500 discharges per month, you must implement a FHIR R4 webhook via an intermediary like Redox within 90 days. The economics are straightforward: the 11.2% readmission reduction pays for the $4,500 monthly cost at just 4.5 avoided readmissions. This is not a technology upgrade; it is a financial imperative.

However, if over 20% of your referring primary care physicians lack FHIR R4 support, do not deploy a pure webhook. Instead, use a hybrid model with a fax fallback and a dedicated care coordinator to monitor the fallback queue. Accept a 16-minute median update time rather than the ideal 14 minutes. The alternative—total non-delivery—is unacceptable.

For heart failure, COPD, and other high-acuity chronic conditions, add a severity triage field (e.g., 'high-acuity' flag) to the webhook payload. This prevents alert fatigue among PCPs and preserves the 94% action rate. Without this flag, critical updates get buried in routine notifications.

If your patient population has more than 15% without an assigned PCP, pair the webhook with a patient portal notification as a secondary channel. Do not rely on the portal as the primary mechanism—the 0.8% readmission reduction from portal-only updates is clinically meaningless. The loop closes only when the physician receives structured data.

Before scaling, run a 6-month pilot with at least 1,000 discharges. Track two metrics: median time-to-PCP-update (target 5% absolute reduction). Only then should you expand to all service lines.

## What to do next

| Step | Action | Why it matters |  |
| --- | --- | --- | --- |
| 1 | Configure a bidirectional HL7 FHIR webhook in your EHR to push the discharge status update directly to the referring PCP's system, not just an internal EHR task. | This closes the loop at the exact point where document-based exchange fails, so the PCP receives the update within minutes of discharge. |  |
| 2 | Replace manual call lists with an AI-driven multi-channel outreach workflow that ingests payer data. | Manual calling reaches fewer than 25% of patients; automation scales to reach many patients simultaneously. |  |
| 3 | Trigger the webhook at discharge order entry, not at summary completion. | The KLAS Research survey quantified the median gap from discharge to PCP-visible update; every minute of delay compounds clinical risk. |  |
| 4 | Prioritize gap closure by severity and engagement likelihood, per the Journal of Ambulatory Care Management study. | This aligns Frequently Asked Questions What percentage of primary care physicians learn about hospitalizations from patients rather than formal hospital notifications? A 2025 study in the Journal of Ambulatory Care Management reveals that 68% of primary care physicians discover their patient's hospitalization directly from the patient. How does the median time for a PCP-visible status update compare between document-based exchange and FHIR webhooks according to KLAS Research? The 2025 KLAS Research survey found a median time of 38 hours for document-based exchange versus 14 minutes when using a FHIR webhook. What specific SNOMED CT code is used to trigger an automatic in-basket alert upon hospital discharge? The payload contains the coded status flag SNOMED CT 183932001 for 'discharged from hospital' to trigger logic within the PCP's EHR. By how many percentage points did automated FHIR-based status updates reduce 30-day readmissions in the Mayo Clinic trial? The 2024 randomized controlled trial at Mayo Clinic showed an absolute drop of 11.2 percentage points in 30-day readmissions, reducing the rate from 18.4% to 7.2%. What was the difference in PCP acknowledgment rates between structured FHIR alerts and standard discharge PDFs in the AMIA study? The 2025 AMIA study found that 94% of PCPs acknowledged the structured FHIR status alert within 2 hours compared to only 27% who opened the PDF. How many phone calls do staff typically make daily to chase status updates, and what is their success rate? Staff make 40 to 60 phone calls daily but successfully reach fewer than 25% of patients on the first attempt. Quick answers How do 68% of primary care physicians typically learn about their patient's hospitalization? | They discover it directly from the patient rather than through formal hospital notifications. |
| What is the median time from discharge to a PCP-visible status update when using document-based exchange networks like Care Everywhere or CommonWell? | The median time is 38 hours. |  |  |
| What is the median time to a PCP-visible update when using an HL7 FHIR R4 webhook via a SMART-on-FHIR app? | The median time is 14 minutes. |  |  |
| Why does the 38-hour gap occur in document-based exchange workflows? | It occurs because a human must manually open the PDF, extract the status, and enter it into the EHR, creating a workflow limitation rather than a technical one. |  |  |
| What specific SNOMED CT code is used for the 'discharged from hospital' status flag in the FHIR webhook payload? | The code is SNOMED CT 183932001. |  |  |

### Related reading

- [Lower Clinic No Shows: Michigan 18% to 9% 48-Hour Recall vs Hire 2026](https://getpulse.care/blog/lower-clinic-no-shows-michigan-18-to-9-48-hour-recall-vs-hire-2026.php)
- [Remote Patient Monitoring Billing 2026: Device Code (CPT 99454) 16-Day Bill or Miss](https://getpulse.care/blog/remote-patient-monitoring-billing-2026-device-code-cpt-99454-16-day-bill-or-miss.php)
- [Veterans Day Clinic Closures: 2026 18-Point Surge for Nov 12-13 Recall](https://getpulse.care/blog/veterans-day-clinic-closures-2026-18-point-surge-for-nov-12-13-recall.php)
- [75+ RSV Uptake at 22.4%: 2-Click BPA 11-Point Lift](https://getpulse.care/blog/75-rsv-uptake-at-224-2-click-bpa-11-point-lift.php)
- [0-7-21-60 Pulse Recall vs Manual for Diabetes and Colon](https://getpulse.care/blog/0-7-21-60-pulse-recall-vs-manual-for-diabetes-and-colon.php)
- [Wyoming CME Credits: Live Meetings vs. Telehealth Cost per Hour](https://getpulse.care/blog/wyoming-cme-credits-live-meetings-vs-telehealth-cost-per-hour.php)

### Latest

- [Lower Clinic No Shows: Michigan 18% to 9% 48-Hour Recall vs Hire 2026](https://getpulse.care/blog/lower-clinic-no-shows-michigan-18-to-9-48-hour-recall-vs-hire-2026.php)
- [Remote Patient Monitoring Billing 2026: Device Code (CPT 99454) 16-Day Bill or...](https://getpulse.care/blog/remote-patient-monitoring-billing-2026-device-code-cpt-99454-16-day-bill-or-miss.php)
- [Veterans Day Clinic Closures: 2026 18-Point Surge for Nov 12-13 Recall](https://getpulse.care/blog/veterans-day-clinic-closures-2026-18-point-surge-for-nov-12-13-recall.php)
- [75+ RSV Uptake at 22.4%: 2-Click BPA 11-Point Lift](https://getpulse.care/blog/75-rsv-uptake-at-224-2-click-bpa-11-point-lift.php)

Canonical: https://getpulse.care/blog/the-38-hour-gap-why-the-cheapest-discharge-option-is-clinically-most-expensive.php
Markdown: https://getpulse.care/blog/the-38-hour-gap-why-the-cheapest-discharge-option-is-clinically-most-expensive.php/index.md