What is the best way to predict postoperative pulmonary complications?
Pneumonia and respiratory failure are among the most common serious complications following surgery. Models that predict the risk of these complications may help clinicians identify patients who require more intensive monitoring and surgery preparation and identify potentially modifiable risk factors.
Two prediction models have been developed and validated by researchers from the National Veterans Affairs Surgical Quality Improvement Program.1,2 The models use data from patients at Veterans Affairs medical centers to predict the likelihood of post-operative respiratory failure and pneumonia. The first model (Table 1) was developed using data from 81,719 noncardiac surgeries at 44 medical centers between 1991 and 1993; it was validated using data from 99,390 noncardiac surgeries at 132 medical centers between 1994 and 1995.1 Minor procedures and major transplantations were excluded. Women were also excluded because they were much younger and healthier than the average male patient.1 Patients were followed for 30 days. The model was designed to predict respiratory failure, defined as mechanical ventilation required for at least 48 hours after surgery or as reintubation after having been extubated postoperatively.
|Type of surgery|
|Abdominal aortic aneurysm repair||27|
|Neurosurgery, upper abdominal, or peripheral vascular||14|
|Albumin level less than 3.0 g per dL (30 g per L)||9|
|Blood urea nitrogen level greater than 30 mg per dL (10.7 mmol per L)||8|
|Partially or fully dependent functional status||7|
|History of chronic obstructive pulmonary disease||6|
|60 to 69||4|
The same researchers developed a second, more complex model (Table 2) to predict the likelihood of postoperative pneumonia.2 The model was developed using data from 160,805 noncardiac surgeries at 97 medical centers between 1997 and 1999; it was validated using data from 155,266 noncardiac surgeries at 100 medical centers between 1995 and 1997.2
Both models performed well in the validation studies, and physicians can use them onfidently to predict the likelihood of pulmonary complications in men undergoing noncardiac surgery. Although it seems reasonable to assume that the models also would be accurate in women, further validation is needed.
Applying the Evidence
A 68-year-old man is about to undergo elective surgery to repair an abdominal aortic aneurysm; general anesthesia will be used. He has a history of chronic obstructive pulmonary disease (COPD) and smokes one pack of cigarettes per day. At the time of hospital admission, his albumin level was 3.0 g per dL (30 g per L), and his blood urea nitrogen (BUN) level was 35 mg per dL (12.5 mmol per L). He has not lost weight recently, has no sensorium impairment, has no history of cerebrovascular accident or steroid use, and consumes one or two alcoholic drinks per day. He is functionally independent and lives with his wife. What is the patient's probability of postoperative respiratory failure and pneumonia?
Using Table 1,1 the patient receives 45 points (27 points for type of surgery, 8 for elevated BUN, 6 for COPD history, and 4 for his age), giving him a very high risk of postoperative respiratory failure. Using Table 2,2 he receives 39 points (15 points for type of surgery, 9 for his age, 3 for BUN, 5 for COPD his-tory, 4 for the use of general anesthesia, and 3 for smoking), giving him a 4.6 per-cent probability of postoperative pneumo-nia. Optimizing his pulmonary function, reducing his BUN level before surgery, and closely monitoring his pulmonary status postoperatively may reduce the patient's risk of complications.