5.2 Digestive and Hepatobiliary Disorders
Key Takeaways
- GI hemorrhage needs acuity, anatomic source, and etiology; unspecified gastrointestinal bleeding is not the destination when endoscopy or the provider already names a bleeding lesion.
- Ulcer, obstruction, peritonitis, and GI malignancy require the features that change treatment: hemorrhage or perforation, complete versus partial obstruction, localized versus generalized peritonitis, and whether cancer occasioned the admission.
- Cirrhosis, hepatitis, pancreatitis, and biliary obstruction need etiology and the decompensating or obstructive events actually treated this stay—not problem-list history alone.
- Hepatic encephalopathy is reportable when the provider links altered mentation to liver disease; confusion next to cirrhosis is not a linkage.
- Compliant queries for these charts cite sourced indicators (EGD, CT, lipase, lactulose, ERCP) and must not mention reimbursement or quality scores.
5.2 Digestive and Hepatobiliary Disorders
Quick Answer: For gastrointestinal (GI) hemorrhage, capture acuity, source, and etiology (ulcer, varices, diverticular, angiodysplasia, malignancy)—not unspecified bleeding when the record can support more. Ulcer, obstruction, peritonitis, and GI malignancy need the features that change treatment and grouping: hemorrhage, perforation, complete versus partial obstruction, localized versus generalized peritonitis, and whether cancer occasioned the admission. Cirrhosis, hepatitis, pancreatitis, and biliary obstruction need etiology and complications. Hepatic encephalopathy is reportable when the provider links altered mentation to liver disease—not when “confusion” sits next to cirrhosis without a stated relationship.
GI and hepatobiliary stays are high-volume Domain II application items because the first documented phrase is often a symptom (melena, hematemesis, abdominal pain, jaundice) while the principal diagnosis (PD) after study is a more specific disease. Clinical documentation integrity (CDI) closes that gap with clinical indicators and a nonleading query, then leaves sequencing to the Uniform Hospital Discharge Data Set (UHDDS) and the ICD-10-CM Official Guidelines. This independent OpenExamPrep teaching is not an Association of Clinical Documentation Integrity Specialists (ACDIS) reprint.
Do not invent unofficial complication or comorbidity (CC) / major CC (MCC) lists for every K-code. Centers for Medicare & Medicaid Services (CMS) publishes those designations in the fiscal-year Inpatient Prospective Payment System (IPPS) tables. Specified bleeding ulcers, peritonitis, and some hepatic-failure constructions are familiar examples of conditions that often change severity capture compared with unspecified abdominal pain or unspecified GI bleeding; verify the current tables and the CC exclusion list before treating any code as a guaranteed tier change.
GI hemorrhage: source, etiology, acuity
Unspecified gastrointestinal hemorrhage is a residual when the source is truly unknown after study. It is not the destination when endoscopy, angiography, or the provider’s impression already names a bleeding lesion. Query along three axes:
- Acuity — acute, chronic, or acute-on-chronic blood loss; acute blood-loss anemia when clinically supported and documented by a provider.
- Anatomic source — esophagus, stomach, duodenum, small bowel, colon, rectum; upper versus lower when that is all that is known.
- Etiology — gastric or duodenal ulcer with hemorrhage, esophageal varices with bleeding, diverticular bleeding, angiodysplasia, Dieulafoy lesion, malignancy, ischemic colitis, hemorrhoids, or another stated cause.
Hematemesis, melena, and hematochezia are signs. They do not replace a bleeding diagnosis when a more specific condition is documented. Do not code both unspecified GI hemorrhage and a specific bleeding ulcer for the same event. If two potential sources exist (varices and ulcer) and the endoscopist treats one, query which lesion caused the hemorrhage that occasioned admission.
A proton-pump inhibitor infusion, octreotide, massive-transfusion protocol, and intensive hemoglobin checks are pharmacologic and monitoring indicators. They support a query; they are not a hemorrhage etiology.
Scenario: “GI bleed” as the copied PD
A 64-year-old is admitted for melena and orthostasis. Hemoglobin falls from 11.2 to 7.4 g/dL. Esophagogastroduodenoscopy (EGD) shows a duodenal ulcer with an actively bleeding visible vessel, treated with clips. The hospitalist continues to write “GI bleed” as the assessment. After study, acute duodenal ulcer with hemorrhage is the PD candidate, not unspecified gastrointestinal bleeding. Query if the impression never connects the ulcer to the hemorrhage. Acute blood-loss anemia, if documented and treated (transfusion, monitoring), may meet UHDDS secondary-diagnosis criteria; a falling hemoglobin number alone is not a diagnosis.
Ulcer, obstruction, peritonitis, malignancy
Peptic ulcer disease needs site (gastric, duodenal, peptic unspecified), acuity (acute versus chronic when known), and the complications that drive resources: hemorrhage, perforation, or both. “Gastritis” on an emergency-department differential is not a duodenal ulcer. Nothing-by-mouth status and serial abdominal exams are indicators, not codes.
Intestinal obstruction needs location (small bowel versus large bowel), completeness (complete versus partial), and complications (ischemia, perforation, peritonitis). Postoperative ileus versus adhesive mechanical small-bowel obstruction is a frequent query. Do not upgrade “constipation” or “ileus” to mechanical obstruction without provider documentation. Nasogastric decompression and a transition point on computed tomography (CT) support a mechanical-obstruction query; the radiologist’s impression is an indicator, and the attending (or another treating provider) still diagnoses the obstruction.
Peritonitis needs localized versus generalized or acute when documented, and the cause (perforated viscus, spontaneous bacterial peritonitis (SBP) in cirrhosis, anastomotic leak). Peritoneal findings on a CT report are indicators. The provider must diagnose peritonitis. SBP in a patient with ascites is not “abdominal pain in cirrhosis.”
GI malignancy as PD requires that the cancer, after study, occasioned the admission. Bleeding, obstruction, or perforation from a known tumor may be the acute reason for admission; neoplasm-and-complication guidelines govern sequencing. Metastatic disease, obstruction, and malnutrition are separate documentation opportunities—not automatic add-ons from a problem list. A screening colonoscopy that finds an incidental polyp is not this chapter’s inpatient PD problem; a colon mass with obstruction that requires urgent diversion is.
Scenario: partial obstruction that is not “unspecified abdominal pain”
A 71-year-old with prior colectomy presents with distention and vomiting. CT: transition point, fecalization of small-bowel loops, no free air. Nasogastric decompression, serial exams, no immediate laparotomy. The emergency-department PD of “abdominal pain” should not survive the stay. Query for mechanical small-bowel obstruction (adhesive versus other), complete versus partial, and whether ischemia or peritonitis is present or ruled out.
Cirrhosis, hepatitis, pancreatitis, biliary obstruction
Cirrhosis documentation should name etiology when known (alcohol, viral, metabolic, other) and the decompensating events this stay actually treats: ascites requiring large-volume paracentesis, variceal bleeding, SBP, hepatorenal physiology, or encephalopathy. “History of cirrhosis” on a problem list is not decompensated liver disease unless this stay evaluates or treats it. Alcoholic hepatitis, if documented as a distinct acute injury, is not interchangeable with stable cirrhosis.
Hepatitis needs acuity (acute versus chronic) and type (viral letter, alcoholic, drug-induced, autoimmune) when known. Do not assign acute viral hepatitis from an isolated enzyme bump. Chronic viral hepatitis that is only a listed history may still be reportable if it is evaluated (viral loads, hepatology) this stay; it is not PD when bleeding varices occasioned the admission.
Pancreatitis needs acute versus chronic, etiology (gallstone, alcohol, drug, idiopathic, hypertriglyceridemia), and complications (necrosis, infection, organ failure, pleural effusion). Lipase elevation is an indicator. Query necrosis or infected necrosis when imaging and antibiotics or intervention support it; do not infer infected necrosis from a single fever. Intravenous fluids and opioid analgesia do not prove necrosis.
Biliary obstruction (choledocholithiasis, cholangitis, pancreatic-head mass, stricture) needs the presence of obstruction, infection (cholangitis), and whether pancreatitis or jaundice is a linked condition. Charcot’s triad and imaging of a dilated common duct are indicators for a cholangitis or obstruction query, not a license to code from the CDI note. Endoscopic retrograde cholangiopancreatography (ERCP) with sphincterotomy and stone extraction is both a treatment indicator and an ICD-10-PCS documentation opportunity (later Domain VI procedure rules); here, it tells you obstruction was real enough to intervene.
Encephalopathy linkage
Altered mental status in a patient with cirrhosis is not automatically hepatic encephalopathy. Metabolic encephalopathy, alcohol withdrawal, intracranial event, medication effect, and infection can look the same. Hepatic encephalopathy requires a provider link between the mentation change and liver disease. Assign from the current ICD-10-CM index once that diagnosis exists; do not code from ammonia. Ammonia is neither required nor sufficient.
A yes/no query may not introduce a new diagnosis. Use multiple-choice with clinically relevant options, a required open-ended “Other, please specify,” and ruled-out / unable-to-determine choices when they fit the 2026 ACDIS/AHIMA query guidelines. If the neurologist writes “toxic-metabolic encephalopathy” and the hepatologist writes “overt hepatic encephalopathy,” query to reconcile. Do not recode a documented hepatic encephalopathy as a generic metabolic encephalopathy solely because a consultant used broader wording.
Lactulose, rifaximin, and an asterixis exam are indicators. Starting lactulose “in case this is liver-related” is not a diagnosis.
Scenario: ammonia, lactulose, and a missing link
Day 1: “confusion, cirrhosis.” Lactulose is started empirically. Day 2: asterixis documented; lactulose titrated to bowel movements; “encephalopathy” appears without “hepatic.” Query whether the condition is hepatic encephalopathy linked to the documented cirrhosis, another encephalopathy type, an alternative diagnosis, or other. Do not mention SOI, ROM, or MS-DRG.
| Finding in the record | Not enough by itself | What to clarify |
|---|---|---|
| Melena or hematemesis | Unspecified GI hemorrhage | Source, etiology, and acuity |
| Lipase several times the upper limit | “Abdominal pain” | Acute pancreatitis, etiology, necrosis |
| Cirrhosis plus confusion | Hepatic encephalopathy | Explicit linkage and alternative causes |
| Dilated ducts on imaging | Biliary obstruction | Obstruction, cholangitis, cause |
| Free air or peritonitis language on CT | A peritonitis code | Provider diagnosis; localized versus generalized |
EGD shows an actively bleeding duodenal ulcer treated with clips, but daily notes still list only “GI bleed.” The most accurate next CDI step is:
A patient with known cirrhosis is confused. Lactulose is started. Ammonia is mildly high. No provider has linked the mentation change to liver disease. Which statement matches compliant CDI practice?
After study, an acute duodenal ulcer with hemorrhage occasioned the admission. Melena was the presenting symptom. Principal-diagnosis selection should:
A patient admitted for epigastric pain has lipase four times the upper limit, gallstones, a dilated common bile duct, and a CT read of acute pancreatitis without necrosis. The attending writes only “abdominal pain, gallstones.” The highest-yield query cluster is: