A Wide Angle Lens on Adverse Events

Blue Health Intelligence attempts to paint a fuller picture of ‘what went wrong’ by limiting exclusions.

Peter Wehrwein
Editor

The Institute of Medicine’s To Err is Human: Building a Safer Health System report came out 19 years ago. Rarely do reports live up to appellations like “landmark” and “watershed.” To Err is Human is the exception. Health Affairs paid homage to the report by making patient safety the theme of its November issue. There’s a documentary film with the same name. To Err is Human has led to countless efforts to reduce errors and improve patient safety. And now there is real money is at stake. One of the main tools in the value-based payment toolkit is linking payment to performance on safety measures.

But which events get tallied as errors or adverse events remains a fraught issue. “Terms like complications, medical errors, adverse events, never events—they are emotionally laden terms,” notes Alan Spiro, senior vice president and CMO at Blue Health Intelligence, a health analytics company in Chicago that is an independent licensee of the Blue Cross Blue Shield Association.

In many contexts, adverse events are identified, with good reason, for the purpose of identifying errors or where processes can be improved. But as a result, some adverse events may be excluded from those counts because they are “clinically expected” given patient comorbidities and other issues. Post-discharge events that were related to the hospitalization may also fall outside the count.

Spiro and his BHI colleagues have conducted an analysis of a commercial health plan claims database and adverse events that is designed to paint a fuller, real-world picture of adverse events. The database included patients, ages 18–64, admitted to the hospital in 2016 and 2017 whose complications happened during their admission.

“We wanted to begin a dialogue with payers, providers, and patient/consumers about how the industry might show, and more transparently communicate, risks associated with inpatient care,” Spiro said in a prepared statement.

The BHI analysis compares adverse event counts and rates using BHI methods with counts and rates using Agency for Healthcare Research and Quality (AHRQ) methods.

One way that BHI cast a wider net was by tapping into professional as well as facility claims. But the more salient difference was that the BHI didn’t include many of the exclusions that the methodologies of AHRQ and other patient safety and quality organizations typically use. For example, certain infections in cancer or transplant patients may be excluded because those patients have compromised immune systems. Again, that is reasonable from an error-prevention point of view; infections can occur in those patients without anyone having made any kind of mistake. But if you are trying to present a complete picture of what happens to cancer and transplant patients—a picture that includes additional costs and, for patients, inconvenience and suffering—that exclusion could be misleading. BHI only excluded cases from its adverse event calculations when the person was admitted for that event or condition.

The results for inpatient falls with hip fracture are the clearest example of the difference between the BHI and AHRQ adverse events. Using just facility claims, BHI’s analysis identified 1,417 inpatients falls with hip fracture. Because of exclusions, the AHRQ methods identified zero such falls.

Comparing complication rates

Four complications: BHI vs. AHRQ rates

Central line–associated bloodstream infections (CLABSI)
BHI facility/ professional claims BHI facility claims AHRQ facility claims
Included cases All inpatient admissions Admits with eligible medical DRG or surgical DRG Admits with eligible medical DRG or surgical DRG
Excluded cases None Principal diagnoses of CLABSI; stays of < 2 days Principal or secondary diagnosis of CLABSI present on admission; stays of less than 2 days; immune-compromised and/or cancer patients
# of measured cases 4,712,596 3,655,053 3,104,040
# of excluded Cases None 1,057,543 1,608,556
# of adverse events 3,393 1,194 507
Complication rate 0.072% 0.033% 0.016%
Inpatient falls with hip fracture
BHI facility/ professional claims BHI facility claims AHRQ facility claims
Included cases All inpatient admissions Admits with eligible medical DRG, or surgical DRG and operating room (OR) procedure Admits with eligible medical DRG, or surgical DRG and OR procedure
Excluded cases Diagnosis-related group (DRG) with hip or femur fractures, procedures Admits with eligible medical DRG or surgical DRG and OR procedure Principal diagnosis of hip fracture; major diagnosis code of musculoskeletal or connective tissue, pregnancy, or childbirth. Excluded diagnoses of seizure disorder, syncope, stroke, occlusion of arteries, coma, cardiac arrest, poisoning, trauma, delirium or other psychoses, anoxic brain injury, or conditions associated with fragile bone (metastatic cancer, lymphoid malignancy, or bone malignancy.
# of measured cases 4,695,571 4,369,780 3,059,798
# of excluded cases 17,025 342,816 1,652,798
# of adverse events 4,116 1,417 0
Complication rate 0.90% 0.03% 0.00%
Iatrogenic pneumothorax
BHI facility/ professional claims BHI facility claims AHRQ facility claims
Included cases All inpatient admissions Admits with eligible medical DRG or surgical DRG Admits with eligible medical DRG or surgical DRG
Excluded cases Principal diagnosis of iatrogenic pneumothorax Principal diagnosis of iatrogenic pneumothorax Principal diagnosis of iatrogenic pneumothorax; major diagnosis code of pregnancy, or childbirth. Excluded diagnoses of chest trauma, pleural effusion. Excluded procedures include thoracic surgery, lung or pleural biopsy, diaphragmatic repair, and cardiac procedures.
# of measured cases 4,712,596 4,500,983 3,376,939
# of excluded cases 0 211,613 1,335,657
# of adverse events 5,568 3,279 617
Complication rate 0.12% 0.07% 0.02%
Sepsis (post-operative)
BHI facility/ professional claims BHI facility claims AHRQ facility claims
Included cases All inpatient admissions Admits with eligible medical DRG or surgical DRG Admits with eligible surgical DRG and OR procedure
Excluded cases Principal diagnosis of sepsis, DRGs for sepsis or infectious disease Principal diagnosis of sepsis Principal diagnosis of sepsis; major diagnosis code of pregnancy, or childbirth. Excluded diagnoses of infections.
# of measured cases 4,712,596 4,559,123 235,469
# of excluded Cases 154,755 153,473 4,477,127
# of adverse events 88,605 50,179 3,371
Complication Rate 1.94% 1.10% 1.43%