This case, originally published in May of 2009, sparked a lively debate online. Here are some highlights.
Continue reading for the final analysis.
When readers responded to the scenario (which appeared originally in the May issue) I was glad to see that most people thought the patient and physician acted reasonably. I changed the facts to make the scenario apply to an emergency department setting, but the patient was me and the physician was my primary care physician. My symptoms were typical of my previous bouts with diverticulitis and I really didn’t want the radiation from another CT scan. My physician was uncomfortable treating me without a CT scan, but trusted me enough to know that I would follow up as she requested. Fortunately, my symptoms resolved in a week or so and there were no bad outcomes.
It is entirely appropriate to begin outpatient management in a case of mild diverticulitis. Patients with diverticulitis who present with systemic signs of infection, peritonitis, who are immunocompromised, or who have comorbidities may benefit from inpatient management. Of those patients presenting with diverticulitis, 15-25% develop complications requiring surgery including abscess, intestinal rupture, perotinitis, and fistula formation.Everyone was right on the mark in their answers (some highlights are printed below). Given the presumptive diagnosis, a trial of antibiotics and close follow up were reasonable alternatives to further testing provided that the patient has a good understanding of the potential for a bad outcome and that the patient agrees to seek further evaluation if the symptoms do not improve. In short: standard of care met.The idea of discharging a patient without doing a full work-up pits physicians who are “risk tolerant” against those who are “risk adverse.” It is encouraging to see many physicians agreeing that exhaustive testing is not always necessary on patients who have persistent or recurrent symptoms. However, as several commenters noted, the decision on whether to perform repeat testing depends largely upon the history and the patient. For example, many elderly patients with abdominal pain can be considered “high risk.” Up to 40% of elderly patients with abdominal pain require surgery and the etiology for abdominal pain in elderly patients is misdiagnosed in more than a third of cases. Young relatively healthy patients with abdominal pain are less likely to suffer from problems such as cancers, aneurysms, or mesenteric ischemia. History may lead the practitioner to consider alternate diagnoses as well – patients who have chronic abdominal pain associated with food intake may be more likely to be suffering from biliary colic, gastric issues, food allergies, or celiac disease.
Some patients don’t follow the rules, though. One physician recalled a patient with a presentation similar to this one in which the patient had acute cholecystitis and situs inversus. Another physician described a patient who was discharged after receiving five days of intravenous antibiotics for diverticulitis with a small perforation. The patient returned with left upper quadrant tenderness, a WBC count of 11,000, and minimally elevated pancreatic enzymes. He was discharged home and died two days later from a large left upper retroperitoneal abscess that abutted the pancreas. These “outlier” cases and the ensuing lawsuits often give physicians cause for concern.
Many readers noted the presence of a normal WBC in reviewing the physician’s thought processes. Several studies have questioned the predictive value of WBC counts in diagnosing causes of abdominal pain. A study by Gerhardt et al. in the American Journal of Emergency Medicine (AJEM) showed that WBC counts, in addition to most other diagnostic studies, did not have sufficient sensitivity to exclude serious causes of nonspecific abdominal pain. Only one diagnostic test had sufficient sensitivity to diagnose urgent causes of abdominal pain: the noncontrast CT scan. A study by Silver et al in AJEM showed that WBC counts “rarely affect clinical decision-making” in young women with abdominal pain. In that study, no clinically significant diagnoses would have been missed had a CBC not been performed. In fact, according to eMedicine.com, 20-40% of patients with acute diverticulitis have normal white blood cell counts. The utility of a WBC count in diagnosing appendicitis is somewhat more controversial. For example, one 2005 study in AEM concluded that there was no threshold WBC count that was of clinical utility in diagnosing or excluding appendicitis while a 2007 study in Pediatric Emergency Care showed a high correlation between elevated WBC counts and appendicitis in children, including a high negative predictive value for appendicitis when WBC counts were normal or low. Excepting appendicitis, there is little data demonstrating the utility of WBC counts in determining whether abdominal pain is or is not benign.
Several commenters wondered whether an abdominal aortic aneurysm may have been present. No aneurysm was seen on previous CT scans, but their questions raise another issue in dealing with patients who have recurrent symptoms: how often should normal testing be repeated? Can a clinician safely exclude the presence of an aneurysm based on the absence of an aneurysm on a CT scan two years ago? Unfortunately, even though medical studies are very good at determining the positive and negative predictive value of diagnostic testing, few if any studies definitively state the degree or frequency of testing necessary to “rule out” a diagnosis. How many of us have seen patients who recently had a “normal” cardiac stress test come to the ED with an ST elevation myocardial infarction? The lack of a clear standard for the frequency of repetitive testing leaves physicians to use their clinical judgment in determining what tests to order and how often to order them. Varying clinical experiences result in varying amounts of diagnostic testing, which brings me to my final point.
Dr. Spontak noted that the outcome of the case may have swayed the decision about the standard of care in one way or another. Unfortunately, retrospective bias is a significant problem when evaluating a physician’s clinical decision-making. How many physicians would still believe that this care was acceptable if the patient died two days later from a retroperitoneal abscess? It would be interesting to structure a study comparing evaluation of clinical decision-making using the same clinical data while altering patient outcomes. But we’ll leave that for another column.