# Discharge summary: what it contains, examples and how to review > What goes in a hospital discharge summary, the CMS rules, what studies show it leaves out, a worked example with a pending result, a template and a review method. Canonical page: https://medrecords.ai/guides/discharge-summary/ --- New — **Missing Records Detection:** flags every visit, provider, and date missing from the file. [See how →](https://medrecords.ai/product/missing-records-identification/) Record anatomy, September 2026. 10 chapters. ## Discharge summary: what it contains and how to check it against the stay For attorneys, legal nurse consultants, claims reviewers, underwriters and clinicians. You get the parts of a discharge summary, the federal hospital rules behind it, what research shows about what it leaves out, a worked example with a result that came back after discharge, a discharge summary template and a review log. By [Ahmed Jemaa](https://medrecords.ai/authors/ahmed-jemaa/), Co-Founder and CEO, Medrecords AI. Published 26 September 2026. A discharge summary is the physician's closing account of a hospital stay: why the patient was admitted, what was found and done, the diagnoses at discharge, the medications going home, the patient's condition and where they went, and the follow-up plan. Federal hospital rules require one in every record. It is often the first document a reviewer reads. The discharge summary is the page everyone reads first. It is short, it is written by the physician in charge, and it claims to tell the whole stay in 2 or 3 pages. That is exactly why it cannot be taken on trust. It is a summary written at the end, often from memory and copied text, and the research on what it leaves out is consistent. This guide is built from the Medicare hospital Conditions of Participation for medical records and discharge planning (42 CFR 482.24 and 482.43), the CMS State Operations Manual interpretive guidance for those rules, a systematic review of communication at hospital discharge (JAMA, 2007), 2 studies of test results pending at discharge (Annals of Internal Medicine, 2005; Journal of General Internal Medicine, 2009) and a study of adverse events after discharge (Annals of Internal Medicine, 2003). The worked example is hypothetical. Nothing here is legal or medical advice. 8 numbers ### The discharge summary in 8 numbers 30 days after discharge for the hospital record, including the final diagnosis, to be complete 42 CFR 482.24(c)(4)(viii) 12% to 34% of first follow-up visits had the discharge summary available, across the studies reviewed Kripalani et al., JAMA, 2007 65% of discharge summaries lacked test results still pending at discharge Kripalani et al., JAMA, 2007 16% of tests with pending results were mentioned in discharge summaries Were et al., J Gen Intern Med, 2009 41% of patients discharged from hospitalist services had test results return after discharge Roy et al., Ann Intern Med, 2005 61.6% of potentially actionable post-discharge results surveyed were unknown to the physicians responsible Roy et al., Ann Intern Med, 2005 19% of patients discharged home had an adverse event after discharge Forster et al., Ann Intern Med, 2003 66% of those adverse events were adverse drug events Forster et al., Ann Intern Med, 2003 Chapter 1 Everyone ### What a discharge summary is A discharge summary is written when a patient leaves the hospital. It tells the next clinician, usually a primary care doctor, a nursing home or a home health agency, what happened during the stay and what needs to happen next. Every hospital record must have one, and outpatient surgery and emergency visits have a shorter equivalent. It is not the same as the discharge instructions. Instructions are written for the patient in plain language: medications, activity, warning signs, appointments. The summary is written for clinicians. A file usually contains both, and they should agree. For a reviewer, the summary is a map, not the territory. It points to the admission, the key events and the diagnoses, and it saves time on a first pass. But it is written at the end of the stay by 1 physician, often one who did not see the patient every day, and it compresses days or weeks into a few paragraphs. Every important statement in it should be traced back to the record it summarizes. #### Terms you will meet Hospital courseThe narrative of the stay, usually by problem or by day. The longest and most compressed section. Principal diagnosisThe condition established after study to be chiefly responsible for the admission. It drives coding and billing. Secondary diagnosesOther conditions present or arising during the stay. Complications appear here, or do not. DispositionWhere the patient went: home, home with services, skilled nursing, rehabilitation, hospice, another hospital, against medical advice, or death. Discharge medicationsThe list the patient goes home on, ideally marked as new, changed, stopped or continued. Pending resultsTests sent during the stay whose results were not back at discharge, such as cultures, pathology or send-out labs. Transfer summaryThe equivalent document when a patient moves to another facility. Death summaries cover patients who died in hospital. Co-signatureWhen a resident, nurse practitioner or physician assistant writes the summary, the attending physician co-signs to verify it. If you remember one thing Use the discharge summary as a map of the stay. Trace every statement that matters back to the record it summarizes. Chapter 2 Everyone ### The parts of a discharge summary, section by section The mock summary below closes a 4-day admission for pneumonia. The layout is invented; every hospital's template differs. The 9 marked parts appear in some form in almost all discharge summaries. Discharge summary Hypothetical Patient R, admitted 05/02/2026, discharged 05/06/2026 Header — Admission and discharge dates, attending, author, dictated and signed times1 Diagnoses — Principal: community-acquired pneumonia. Secondary: atrial fibrillation, type 2 diabetes2 Procedures — None3 Hospital course — Treated with IV antibiotics, improved, transitioned to oral on day 34 Key results — Chest X-ray: right lower lobe infiltrate. WBC 16.2 falling to 9.85 Pending — None listed6 Medications — Oral antibiotic 5 days; home medications unchanged7 Disposition — Home, stable. Follow up with PCP in 1 to 2 weeks8 Signature — Dictated by Dr. N, resident; co-signed by Dr. W, attending9 HYPOTHETICAL 1. 1 **Dates and times** Admission, discharge, dictation and signature. A summary signed weeks after discharge was written from the chart, not from memory of the patient, and may not have reached the next doctor in time. 2. 2 **Diagnoses** Principal and secondary. Compare with the final coded diagnoses on the UB-04 and with the problem list. A complication that appears in the notes but not here is a finding. 3. 3 **Procedures** Operations and bedside procedures. Each should have its own procedure or operative note. 4. 4 **Hospital course** The narrative. It is the section most likely to be copied from earlier notes and the one most likely to smooth over events such as a fall, a rapid response or a medication error. 5. 5 **Key results** Selected labs and imaging. Selected is the important word; the chart has the rest. 6. 6 **Pending results** Tests not back at discharge. Studies find this section is missing or incomplete far more often than not. 7. 7 **Discharge medications** What the patient goes home on. Check it against the discharge orders, the MAR for the last day and the patient instructions. 8. 8 **Disposition and follow-up** Where the patient went, in what condition, and who is responsible next. "Follow up with PCP" without a name or date is common. 9. 9 **Authentication** CMS expects the author to sign, date and time the summary, and the attending to co-sign a delegated one. #### What each section should be checked against | Summary section | Source record | What a reviewer checks | | --- | --- | --- | | Reason for admission | ED record, history and physical | The presenting complaint and first findings match | | Hospital course | Progress notes, nursing notes, rapid response and consult notes | Events in the notes that the summary leaves out | | Procedures | Procedure and operative reports | Every procedure listed, with complications | | Key results | Lab and imaging reports | Abnormal results not mentioned | | Pending results | Lab orders and result dates | Every test resulted after the discharge time | | Discharge medications | Discharge orders, MAR, patient instructions | New, stopped and changed drugs match across all 3 | | Diagnoses | Problem list, UB-04 codes | Coded conditions missing from the summary, and the reverse | | Disposition | Case management notes, transfer forms | Destination and services arranged | If you remember one thing Every section of the summary has a source record. The hospital course and the pending results are where they most often disagree. Chapter 3 Deciding ### What discharge summaries leave out The best evidence on discharge summaries comes from studies of what reached the next doctor. Kripalani and colleagues reviewed 55 observational studies. Direct communication between hospital and primary care physicians happened in 3% to 20% of cases. The discharge summary was available at the first follow-up visit in 12% to 34% of cases, and still only in 51% to 77% at 4 weeks. When it did arrive, it often lacked diagnostic test results (missing from 33% to 63%), the hospital course (7% to 22%), discharge medications (2% to 40%), test results pending at discharge (65%) and follow-up plans (2% to 43%). Pending results are the most studied gap. Were and colleagues reviewed discharge summaries for 668 patients who all had tests pending at discharge: **What discharge summaries documented for patients with pending tests** — share of summaries — Were et al., 2009 Every patient in the study had at least 1 test pending. 3 in 4 summaries mentioned none. The rate did not depend on the author's experience or the length of stay. Those results do come back, and some matter. Roy and colleagues followed 2,644 patients discharged from hospitalist services: 41% had test results return after discharge, and 9.4% of those results were potentially actionable. Of the potentially actionable results they surveyed physicians about, the physicians had been unaware of 61.6%. The period after discharge is also when harm shows up. Forster and colleagues followed 400 patients discharged home from a general medical service. 19% had an adverse event within about 3 weeks, and 66% of those were adverse drug events. Some were judged preventable. That is why the medication list in the summary gets the closest reading. If you remember one thing A blank or missing pending-results section is the norm, not the exception. Find every test resulted after the discharge time yourself. Chapter 4 Everyone ### The rules behind the discharge summary Federal hospital rules require a discharge summary in every record and set a deadline for completing the record. A separate rule requires the hospital to send the information the next providers need. | Rule | What it requires | What it means for review | | --- | --- | --- | | 42 CFR 482.24(c)(4)(vii) | A discharge summary with the outcome of hospitalization, disposition of the case and provisions for follow-up care | A record without one is incomplete | | 42 CFR 482.24(c)(4)(viii) | Final diagnosis, with the medical record completed within 30 days after discharge | Summaries signed later than 30 days are late under the rule | | 42 CFR 482.24(c)(1) | Every entry legible, complete, dated, timed and authenticated | Check author, co-signer and times | | 42 CFR 482.43(a) | A discharge planning evaluation for patients at risk, included in the medical record and discussed with the patient | The case management notes show what was planned | | 42 CFR 482.43(b) | At discharge, send necessary medical information to the post-acute and outpatient providers responsible for follow-up | Ask what was sent, to whom and when | [](https://www.law.cornell.edu/cfr/text/42/482.24) [](https://www.law.cornell.edu/cfr/text/42/482.24) [](https://www.law.cornell.edu/cfr/text/42/482.24) [](https://www.law.cornell.edu/cfr/text/42/482.43) [](https://www.law.cornell.edu/cfr/text/42/482.43) CMS guidance for surveyors adds detail. Follow-up care provisions include post-hospital appointments and plans for care by home health, hospice, nursing homes or assisted living. The admitting physician is responsible for the summary; it can be delegated to a nurse practitioner or physician assistant where state law allows, and then: "For delegated discharge summaries we would expect the MD/DO responsible for the patient during his/her hospital stay to co-authenticate and date the discharge summary to verify its content." #### Copied text and late summaries Many hospital course sections are built by copying earlier notes forward. That is not improper by itself, but copied text can carry forward findings that changed, or leave out events that happened between the copied note and discharge. A summary dictated weeks after discharge was written from the chart, and it was written after any complaint or readmission. Our guides to [altered medical records](https://medrecords.ai/guides/altered-medical-records/) and [EHR audit trails](https://medrecords.ai/guides/ehr-audit-trail-medical-malpractice/) cover how to test when and how a summary was written. If you remember one thing The hospital has 30 days to complete the record. Note when the summary was signed and when it was sent, and compare both with what happened next. Chapter 5 Deciding ### Patterns that show up in discharge summaries Some problems recur often enough to check for every time. Most are differences between the summary and the records it summarizes. | Pattern | Example | Why it is a question | | --- | --- | --- | | Event left out | A fall on day 2 is in the nursing notes, not in the hospital course | The summary is the document most readers rely on | | Pending result not listed | Blood culture drawn the day before discharge, "Pending: none" | Who was responsible for the result when it came back? | | Medication mismatch | Summary says "home medications unchanged"; MAR shows a new anticoagulant | Which list did the patient and the next doctor follow? | | Diagnosis mismatch | Coded secondary diagnosis of a pressure injury; summary silent | Present on admission, or acquired? | | Vague follow-up | "Follow up with PCP in 1 to 2 weeks," no name, no date | Was an appointment made, and was the summary sent? | | Late signature | Signed 3 weeks after discharge, after a readmission | Written with knowledge of what happened next | | Disposition mismatch | "Home, stable," while case management notes a refused rehabilitation placement | Was the patient's condition and the risk recorded? | The medication pattern is the one to check first. The summary's list, the discharge orders, the last day of the [medication administration record](https://medrecords.ai/guides/medication-administration-record/) and the patient instructions are 4 versions of the same list. They often differ, and the difference is where adverse drug events after discharge start. If you remember one thing Compare 4 medication lists: summary, discharge orders, last-day MAR and patient instructions. They should match and often do not. Chapter 6 Building ### Worked example: a result that came back after discharge **Hypothetical.** The patient, physicians, hospital, dates and pages are invented to show the method. It is not a real case or a customer file. Patient R, 71, is admitted with pneumonia on 05/02/2026 and discharged home on 05/06. A blood culture drawn on 05/05 turns positive on 05/07. On 05/12 the patient is readmitted with sepsis. The file holds the first admission's progress notes (pages 40 to 62), labs (pages 70 to 78), the MAR (pages 80 to 92), discharge orders and instructions (pages 94 to 97), the discharge summary (pages 98 to 99) and the readmission record (pages 120 onward). **The discharge summary beside the records it summarizes** — hypothetical — Illustration Discharge review log, 05/02 to 05/21, hypothetical — 9 entries | Date | Source | Record says | Question | Page | | --- | --- | --- | --- | --- | | 05/02 | H&P | Pneumonia; history of atrial fibrillation | Baseline | 40 | | 05/04 | MAR | New anticoagulant started, continued to discharge | Not in the summary's medication list | 86 | | 05/05 18:40 | Labs | Blood culture drawn for fever 38.6 | Fever the day before discharge | 74 | | 05/06 11:00 | Discharge orders | Home; oral antibiotic 5 days; anticoagulant listed | Orders include the new drug | 94 | | 05/06 11:20 | Instructions | "Continue your home medications" | Anticoagulant not explained to the patient | 96 | | 05/07 09:15 | Labs | Blood culture positive, gram-positive cocci | Resulted after discharge; who was notified? | 76 | | 05/12 | Readmission | Sepsis; blood cultures positive | The event the timeline builds toward | 120 | | 05/20 | Discharge summary | "Afebrile, improving." Pending: none. Home medications unchanged | Signed 14 days after discharge, after the readmission | 98 to 99 | | 05/21 | Fax log | Summary sent to primary care | 15 days after discharge | 100 | Gold rows: entries that need a question. Green row: the event the timeline builds toward. 9 entries, 5 questions. The summary describes an uneventful stay. The labs and MAR describe a fever, a pending culture and a new drug. #### What each flag means 1. **A new drug missing from the summary.** The anticoagulant was started on day 3 and ordered at discharge. The summary and the patient instructions say home medications are unchanged. Which list the patient followed is a question for the patient and the pharmacy records. 2. **A fever the day before discharge.** The hospital course says afebrile. The lab log shows a culture drawn for a fever of 38.6. 3. **A pending culture not listed.** "Pending: none," though the culture was in progress at discharge. The result came back the next morning. 4. **No record of who acted on the result.** The file needs the lab's notification log and any call to the patient or the primary care office between 05/07 and 05/12. 5. **A summary written after the readmission.** Signed on 05/20, 8 days after the patient came back, and faxed the next day. The primary care doctor could not have had it before the readmission. #### The same stay on 1 timeline **Every dated entry that matters, first admission to readmission** — hypothetical — Illustration; pages refer to the hypothetical file 1. 05/02 **Admitted with pneumonia** History of atrial fibrillation. H&P p. 40 2. 05/04 **New anticoagulant started** Given daily to discharge. MAR p. 86 3. 05/05 **Fever 38.6, blood culture drawn** Result pending at discharge. Labs p. 74 4. 05/06 **Discharged home** Instructions: "continue your home medications." Orders p. 94; instructions p. 96 5. 05/07 **Culture positive** No notification found in the file. Labs p. 76 6. 05/12 **Readmitted with sepsis** 5 days after the positive result. Readmission p. 120 7. 05/20 to 05/21 **Summary signed, then faxed** "Pending: none." Summary pp. 98 to 99; fax log p. 100 Read first, as reviewers usually do, the summary frames the stay as routine. Read by date, it is the last document written, after the outcome it failed to anticipate. ##### What the discharge record can show When checked against the chart - What the summary left out that the chart records. - Which results came back after discharge, and when. - Which medication list went home, in which document. - When the summary was signed and sent. ##### What it cannot show alone Where you need another record - Whether anyone was called about a late result (lab and phone logs). - Which list the patient actually followed (pharmacy records, testimony). - Whether discharge was premature (a clinical expert). - What was said at the bedside (testimony). If you remember one thing Put the summary's signature date on the timeline. If it falls after the next event, the summary cannot have guided the care that followed. Chapter 7 Building ### How to review a discharge summary, step by step The method works for a 2-day stay or a 2-month one. Read the summary first for orientation, then set it aside and test it. 1. **Note the dates.** Admission, discharge, dictation, signature, co-signature and date sent. 2. **List the claims.** Every diagnosis, procedure, event, result and medication the summary states. 3. **Trace each claim.** Find the source note, result or order for each. Mark the ones with no source. 4. **Read the chart for omissions.** Rapid responses, falls, transfers, new diagnoses, consults and complications that the summary does not mention. 5. **Find pending results.** Every test ordered during the stay with a result time after the discharge time. 6. **Compare the 4 medication lists.** Summary, discharge orders, last-day MAR and patient instructions. 7. **Check the diagnoses against the codes.** Summary diagnoses vs UB-04 diagnosis codes and present-on-admission flags. 8. **Follow the handoff.** Who was named for follow-up, whether an appointment was made, and when the summary reached them. #### Can this discharge summary be relied on as written? 1 **Was it signed within 30 days and before any later event you are reviewing?** YesGo to 2. NoNote the timing. It was written after the fact and could not have guided later care. 2 **Does every stated event and result trace to a source page?** YesGo to 3. NoLog each unsupported statement. 3 **Are there chart events or pending results it leaves out?** NoGo to 4. YesLog each omission with its source page. 4 **Do the 4 medication lists agree?** YesRely on it for orientation, cited to its page. NoLog each drug that differs, with every list's page. ##### 1. Discharge summary template Covers the CMS elements and the gaps the studies found. Adapt it to your hospital's template. DISCHARGE SUMMARY Patient: [NAME / ID] Admitted: [DATE] Discharged: [DATE, TIME] Attending: [NAME] Author: [NAME, ROLE] Reason for admission: [ ] Principal diagnosis: [ ] Secondary diagnoses (including complications): [ ] Procedures, with dates: [ ] Consultants: [ ] Hospital course, by problem: [Problem]: [what was found, done, and the result] Significant events (falls, rapid responses, transfers): [ ] Key results: [labs, imaging, pathology, with dates] RESULTS PENDING AT DISCHARGE: [test, date sent, who will follow up] Discharge medications: NEW: [ ] CHANGED: [ ] STOPPED: [ ] CONTINUED: [ ] Allergies: [ ] Condition at discharge: [ ] Disposition: [home / home health / SNF / rehab / hospice / other] Follow-up: [NAME, specialty, DATE or timeframe] Patient and family informed of: [ ] Dictated: [DATE, TIME] Signed: [NAME, DATE, TIME] Co-signed (if delegated): [NAME, DATE] Sent to: [PROVIDERS, DATE] ##### 2. Discharge summary cross-check log 1 line per claim in the summary, and 1 line per omission you find in the chart. DISCHARGE SUMMARY CROSS-CHECK Patient: [ID] Admission: [DATES] Summary signed: [DATE] co-signed: [DATE] sent: [DATE, TO] Type (claim / omission / pending / medication) | Summary says (page) | Source record says (page) | Match? | Question Examples: Pending | "None" (p. 98) | Blood culture drawn 05/05 18:40, resulted 05/07 (p. 74, 76) | No | Who was notified of the result? Medication | "Home medications unchanged" (p. 99) | New anticoagulant 05/04 to discharge (MAR p. 86) | No | Which list did the patient follow? #### Before you rely on a discharge summary - **Signature and send dates are known** *Including any co-signature.* - **Progress and nursing notes are in hand** *To find events the summary omits.* - **Result times for every test are in hand** *To find results pending at discharge.* - **All 4 medication lists are in hand** *Summary, orders, last-day MAR, instructions.* - **The UB-04 or coded diagnoses are in hand** *With present-on-admission flags.* - **Case management notes are in hand** *For the discharge plan and destination.* 0 of 6 checked. For diagnosis codes and present-on-admission flags, see the [UB-04 and CMS-1500 guide](https://medrecords.ai/guides/ub-04-and-cms-1500/). For the notes the hospital course summarizes, see the [SOAP note](https://medrecords.ai/guides/soap-note/) and [nursing notes](https://medrecords.ai/guides/nursing-notes/) guides, and for surgical stays, the [operative report guide](https://medrecords.ai/guides/operative-report/). If you remember one thing Read the summary first, then test it. Every claim needs a source page, and every event in the chart needs a place in the summary or a note that it is missing. Chapter 8 Deciding ### AI and discharge summary review A discharge summary is 2 pages. The stay it summarizes can be 2,000. Chapter 7's method is tracing each claim in the summary back to the chart, and scanning the chart for what the summary left out. Doing that by hand is slow; it is the kind of matching software does well. AI medical record review can build the timeline from the source records rather than from the summary, put every lab result, medication and note on it with a page-level citation, and show the summary's claims beside what the chart says. That turns an admission into an AI medical chronology where a pending result, a medication that differs between lists, or an event the summary omits is visible at a glance and can be checked against the page in seconds. There is a specific trap here. A large language model (LLM) asked to summarize a hospital stay will lean on the discharge summary, because it is already a summary, and repeat its omissions. Hallucination and omission are both risks, which is why every line needs its page and why the timeline should be built from the underlying records. Whether a discharge was premature or a result should have been acted on is a clinical judgment. The human in the loop reads the flagged pages. Medrecords AI works this way on the files you upload. It builds a [cited medical chronology](https://medrecords.ai/product/chronology/) from the notes, results, MAR and orders, each entry linked to its page. It answers questions like "every test resulted after the discharge time, and whether the summary mentions it" through [cited Q&A](https://medrecords.ai/product/qa/) and [record search](https://medrecords.ai/product/medical-record-search/), writes [summaries](https://medrecords.ai/product/summaries/) with a citation on every line, and flags [records that should exist but were not produced](https://medrecords.ai/product/missing-records-identification/). Flags are signals, not verdicts. It does not decide whether care met the standard, and it does not retrieve records from providers. It runs under SOC 2 and HIPAA with a signed business associate agreement (see [security](https://medrecords.ai/security/) and [HIPAA](https://medrecords.ai/hipaa/) ). Self-Service bills 10 cents a deduplicated page, down to 5 cents at volume, duplicates free; Enterprise On-Prem is an annual license. If you are comparing tools, our guide to [medical record review](https://medrecords.ai/guides/medical-record-review/) covers what to ask any medical chronology software vendor. See how discharge review fits [medical malpractice](https://medrecords.ai/solutions/medical-malpractice/), [wrongful death](https://medrecords.ai/solutions/wrongful-death-record-review/), [hospital risk management](https://medrecords.ai/solutions/hospital-risk-management/) and [legal nurse consultant](https://medrecords.ai/solutions/legal-nurse-consultants/) work. The offer #### The whole stay on 1 timeline, built from the chart. Book a demo on a file where the discharge summary and the chart disagree, then run your first case free on us. Every line comes back cited to its source page. You review, you revise, you sign. [Book a demo](https://medrecords.ai/demo/) [See the cited chronology](https://medrecords.ai/product/chronology/) Scheduling only. No records move from a public page. If you remember one thing Build the timeline from the chart, not from the summary. AI can do that with a page on every line; a clinician still decides what the gaps mean. Chapter 9 Everyone ### Frequently asked questions What should a discharge summary include?The reason for admission, diagnoses, procedures, the hospital course, key results, results pending at discharge, discharge medications marked as new, changed or stopped, condition and disposition, follow-up plans with names, and the author's signature with date and time. Is a discharge summary required?Yes. Federal hospital rules require every record to have a discharge summary covering the outcome of the stay, the disposition and follow-up care, and the record must be completed within 30 days after discharge. Who writes the discharge summary?The physician responsible for the patient during the stay. It can be delegated to a resident, nurse practitioner or physician assistant under hospital policy and state law, and the responsible physician then co-signs it. What is the difference between a discharge summary and discharge instructions?The summary is written for clinicians and covers the whole stay. The instructions are written for the patient: medications, activity, warning signs and appointments. Both belong in the file and should agree. What are pending results at discharge?Tests sent during the stay whose results were not back when the patient left, such as cultures or pathology. Studies found discharge summaries mention only a small share of them. Can I rely on a discharge summary in a legal case?As a guide to the stay, yes. As proof of what happened, check each statement against the notes, results and orders it summarizes. Summaries often omit events and results. Why does the date a discharge summary was signed matter?A summary signed after a readmission or complaint was written with knowledge of the outcome, and it could not have guided the care that followed discharge. Can AI write or review a discharge summary accurately?It can build a timeline from the underlying records and show where the summary differs from them, with a page cite on every line. An AI summary that relies on the existing discharge summary will repeat its gaps. Is it HIPAA compliant to upload hospital records to an AI tool?It can be, with a vendor that signs a business associate agreement, holds a SOC 2 report and commits in writing not to train on your data. A consumer chatbot without a BAA is a different matter. Chapter 10 Everyone ### Sources and method Regulations and the CMS State Operations Manual (Appendix A, revision 248) were read in full text from primary sources in September 2026, and quoted text is verbatim. Study figures come from the published abstracts. The mock summary, patient, physicians, dates and pages in chapters 2, 6 and 7 are hypothetical. Product facts come from this site's product pages. Nothing here is legal or medical advice. - [42 CFR 482.24](https://www.law.cornell.edu/cfr/text/42/482.24): medical records, the discharge summary and the 30-day completion rule. - [42 CFR 482.43](https://www.law.cornell.edu/cfr/text/42/482.43): discharge planning and transmission of necessary medical information. - [CMS State Operations Manual, Appendix A](https://www.cms.gov/regulations-and-guidance/guidance/manuals/downloads/som107ap_a_hospitals.pdf): interpretive guidance at tags A-0468 (discharge summary) and A-0469 (final diagnosis and record completion). - [Kripalani S, LeFevre F, Phillips CO, et al., "Deficits in communication and information transfer between hospital-based and primary care physicians: implications for patient safety and continuity of care"](https://doi.org/10.1001/jama.297.8.831), JAMA 297(8):831-841, 2007. - [Roy CL, Poon EG, Karson AS, et al., "Patient safety concerns arising from test results that return after hospital discharge"](https://pubmed.ncbi.nlm.nih.gov/16027454/), Annals of Internal Medicine 143(2):121-128, 2005. - [Were MC, Li X, Kesterson J, et al., "Adequacy of hospital discharge summaries in documenting tests with pending results and outpatient follow-up providers"](https://link.springer.com/article/10.1007/s11606-009-1057-y), Journal of General Internal Medicine 24(9):1002-1006, 2009. - [Forster AJ, Murff HJ, Peterson JF, et al., "The incidence and severity of adverse events affecting patients after discharge from the hospital"](https://pubmed.ncbi.nlm.nih.gov/12558354/), Annals of Internal Medicine 138(3):161-167, 2003. ### More guides - [**EHR audit trails in medical malpractice: the field guide** — How an EHR audit trail works in a medical malpractice case: the rules behind it, how to preserve and request…](https://medrecords.ai/guides/ehr-audit-trail-medical-malpractice/) - [**Elder abuse and financial exploitation: the medical records that prove or disprove it** — How medical records prove or rebut elder financial exploitation and nursing home abuse: capacity timelines, 42…](https://medrecords.ai/guides/elder-abuse-financial-exploitation-records/) - [**EMS patient care report: what each part means and how to read it against the ED chart** — What each part of an EMS patient care report means, NEMSIS times, Medicare ambulance rules, what handoff…](https://medrecords.ai/guides/ems-patient-care-report/) - [**EMTALA violations: the emergency department records that prove or defeat the claim** — EMTALA claims under 42 USC 1395dd: the 3 duties, disparate screening, the 2-year limit, and the ED records and…](https://medrecords.ai/guides/emtala-violation-medical-records/) - [**HIPAA and AI medical record review: what to verify** — A buyer checklist for HIPAA compliant AI medical record review: the agreements, encryption, data residency, and…](https://medrecords.ai/guides/hipaa-compliant-ai-medical-record-review/) - [**History and physical: what each part means and how to test it against the chart** — What goes in a history and physical, CMS 30-day and 24-hour rules, what studies show about copied and templated…](https://medrecords.ai/guides/history-and-physical/)