Review MS-DRG documentation to prepare for Recovery Auditors

By Christina Benjamin, MA, RHIA, CCS, CCS-P

Connolly Healthcare, the Recovery Auditor (RA) for Region C, continues to focus on complex inpatient reviews of a long list of MS-DRGs. Regardless of whether this is your hospital’s RA, you need to ensure that your records reflect the quality of care provided.

Although RAs validate the medical necessity of uncomplicated hospital admissions with a short length of stay, they also audit:

• MS-DRG
• Principle diagnosis
• Any secondary diagnoses
• Any procedures that affect—or could affect—the DRG

On its website, Connolly states it will review diagnostic and procedural information as well as the discharge status of the beneficiary to ensure that information is coded and reported by the hospital on its claim. It also reviews documentation to ensure the attending physician’s description matches the information included in the beneficiary’s medical record.

In this article, I’ll address some of the most important documentation pitfalls and coding guideline challenges related MS-DRGs under particular RA scrutiny.

Several of the MS-DRGs under recent scrutiny relate to the digestive system including MS-DRG:  

• 377, gastrointestinal (GI) hemorrhage with MCC),
• 378, GI hemorrhage with CC
• 379, GI hemorrhage without CC/MCC

One of the major problems with these DRGs relates to whether documentation clearly supports a link between the bleeding or hemorrhage and the GI disorder (e.g., ulcers, inflammation, diverticulosis, or angiodysplasia).

Documentation of gastritis with bleeding (as a biopsy finding on a pathology report) is not sufficient to report the code for the GI condition and the hemorrhage. Per official ICD-9-CM coding guidelines, the attending physician must document the significance of such abnormal findings in order for coders to report them as a principle or secondary diagnosis.

However, a bleed or hemorrhage is not a prerequisite to reporting one of the relevant codes. Coders may report one of these codes even in the absence of a finding of current bleeding during a diagnostic exam if a physician clearly documents the relationship between the patient’s typical bleeding and his or her GI disorder. For example, a patient’s bleeding may be intermittent and simply not present at the time of the exam. See Coding Clinic, first quarter 1991, p. 15, for more information.

In the case of bleeding esophageal varices due to alcoholic cirrhosis/hepatitis, the code for the bleeding varices is a secondary diagnosis due to the etiology/manifestation coding convention. If the physician clearly links alcoholism to a GI condition (e.g., cirrhosis or hepatitis) in a diagnostic statement as alcoholic hepatitis, then coders can report 303.90. See Coding Clinic, second quarter 2007, p. 6, for more information.

In many cases, physicians may document a GI bleed (e.g., hemoptysis or melena) when the patient has coagulopathy due to use of Coumadin^®. The principle diagnosis in this case is the adverse effect of the Coumadin. Coders sometimes incorrectly assign other diagnoses, such as a code from category 286 (coagulation defects) or the poisoning code for Coumadin, as the principal diagnosis. See Coding Clinic, second quarter 2006, p. 17, for more information.

Oftentimes, patients with a history of thrombi or emboli are on Coumadin prophylactically to avoid a recurrence. Coumadin increases the time it takes for the blood to clot. Therefore, when the level of Coumadin is too high, the risk for internal bleeding greatly increases. A code from category 286 is inappropriate because the bleeding results are from the use of the medication—not an inherent disorder in the patient’s blood clotting factors.

Patients on anticoagulants require constant monitoring. The international normalized ratio (INR) reflects the time it takes for blood to clot. For patients who don’t have a coagulation disorder and who are not taking any anticoagulants, the blood clots at a normal rate. This means that the blood will clot normally on a daily basis and more readily when is the person suffers an injury.

The terms subtherapeutic and nontherapeutic denote that the time it takes for the blood to clot is below the value needed to provide extra protection against recurring thrombi, emboli, or even strokes. To correct this problem, the dosage of the anticoagulant is increased until the time for clotting is long enough. This proper timing or level is also referred to as therapeutic INR.

An elevated INR is a condition in which the clotting rate or time is longer than the intended therapeutic rate. Over-anticoagulation can be the adverse effect of correctly prescribed anticoagulation. It can also occur as a result of incorrect dosing. When over-anticoagulation occurs, the patient’s blood takes so long to clot that various symptoms, such as serious internal bleeding and death, can result . This is why it’s also incorrect to assign a poisoning code unless the physician specifically documents an error in the prescription or dosage of the Coumadin.

It’s essential for coders to understand the differences between adverse effects and poisonings. Four relevant MS-DRGs that RAs are targeting include the following:

• 915, allergic reactions with MCC
• 916, allergic reactions without MCC
• 922, other injury, poisoning, and toxic effect diagnosis with MCC
• 923, other injury, poisoning and toxic effect diagnosis without MCC

The official ICD-9-CM coding guidelines state that an adverse effect occurs when a drug or substance is properly prescribed and administered. Terms such as hypersensitivity, idiosyncrasy, allergic reaction, or synergistic reaction are used to refer to an adverse effect.

A poisoning occurs when there is an error on the part of the provider in administering or dosing the drug. A poisoning also occurs when in any of the following scenarios:

• Overdose of a drug intentionally taken
• A non-prescribed drug or alcohol taken with correctly prescribed and properly administered medication
• Prescribed medication taken by the patient incorrectly
• Prescription medication taken by someone for whom it was not prescribed

Poisoning excludes alcohol intoxication and illegal drug use, both of which are conditions that are classified to the mental health chapter of ICD-9-CM.

If a patient stops taking medication, this is not classified as a poisoning or adverse effect—even if problems occur as a result of the cessation. Instead, report the codes that reflect the condition for which the patient should have been taking medication. It’s interesting to note that ICD-10-CM classifies this scenario as an underdosing.

The phrase drug toxicity can also cause confusion Physicians may use this phrase in their diagnostic statements. Coders can’t make any assumptions with regard to whether the patient suffered an adverse reaction or a poisoning. Instead, coders must query the physician regarding the effects of the drug, particularly when no specific reaction is identified. See Coding Clinic, November-December 1984, p. 14, for more information.

If the adverse reaction meets the definition of principle diagnosis, report the specific effect of the drug or substance as the principle diagnosis along with a code from the Table of Drugs and Chemicals that reflects the adverse effect.

However, if the physician clearly documents a poisoning, coders must select a poisoning code from the Table of Drugs and Chemicals as the principle diagnosis followed by a code for the effects of the drug or substance as well as an E code to reflect the poisoning. If more than one substance is involved, assign an E code and/or poisoning code separately for each drug or substance.

Other MS-DRGs under scrutiny include the following:

• 637, diabetes with MCC
• 638, diabetes with CC
• 639, Diabetes without MCC/CC

Coders frequently make errors when coding complications of diabetes, such as:

• Ulcers
• Renal failure
• Gangrene
• Hypoglycemia

The combination codes for these conditions cannot be assigned unless the physician specifically documents the relationship between the diabetes and the condition. Per Coding Clinic, third quarter 2008, p. 5, the phrase diabetes with [a certain condition] satisfies this requirement.

If multiple complications of diabetes are present, assign all the relevant diabetic combination codes along with any associated additional codes to further specify the complications.

If the patient is admitted a diabetic complication, code diabetes—not the complication— as the principal diagnosis.

For example, when a diabetic patient is admitted for an ulcer, the coder cannot assume a relationship between the ulcer and the diabetes. However, if the physician describes the ulcer as a diabetic ulcer, then coder should assign a code from the 249 or 250 code category as principle diagnosis.

If the ulcer is described as decubitus in a diabetic patient—and the admission is for treatment for complications resulting from the ulcer (e.g., gangrene or osteomyelitis)—coders should assign the ulcer as the principle diagnosis. See Coding Clinic, first quarter 2004, pp. 14-15, for more information.

One common error is to omit an additional code specifying the nephropathy in a patient with both chronic kidney disease and diabetic nephropathy.  See Coding Clinic, September-October 1984, p. 1.

Diagnostic statements such as poorly controlled or very poor control are not sufficient to assign a fifth digit to reflect uncontrolled diabetes. The diagnostic statement must state diabetes uncontrolled or diabetes not controlled. The only exception is for diabetic ketoacidosis in which case coder should automatically assigner the fifth digit of 3 due to the nature of the condition. That is, unless the physician specifically documents type II diabetes. See Coding Clinic, second quarter 2006, p. 19, for more information.

When hypoglycemia is diagnosed in a diabetic patient, no additional code is required. Coders must only assign codes 250.8x or 249.8x. In the case of hypoglycemic shock due to insulin overdose, report the poisoning code for insulin as the principle diagnossi with the associated E poisoning code as a secondary diagnosis. See Coding Clinic, fourth quarter 1993, pp. 19-21, for more information.

Editor’s Note: This article was originally published on JustCoding.com. You can reach Christina Benjamin, RHIA, CCS, CCS-P, an independent coding and education consultant in Jesup, Ga., via e-mail at handmaiden555@gmail.com.