Why do we get strep screens in the ED? Does it promote the practice of value medicine?
Value medicine stops asking the question, “What is the best?” and instead asks, “What is the value?” Value medicine has four components: the magnitude of benefit, the magnitude of harm, the magnitude of patients unaffected, and the charge to the patient (and cost to society). These are negotiated by shared decision-making.
The party line given in the ED for using antibiotics for strep throat is that they prevent acute rheumatic fever and peritonsillar abscess and reduce the spread of disease and symptoms by about one day. None of these assumptions is true, but if they were, they are clinically irrelevant for most patients.
Antibiotics do not prevent acute rheumatic fever in the United States. The ability for group A beta-hemolytic strep (GABHS) to cause rheumatic fever depends on a strain infrequently found in the mainland United States; it has not caused an epidemic since the 1980s. The current incidence is less than one per 100,000 population. (World Health Organization, 2001; Clin Infect Dis 2006;42:448.) It would be impossible to affect such an infinitesimally low incidence with antibiotics.
Antibiotics do not prevent peritonsillar abscess either. The majority of patients with peritonsillar abscesses have negative rapid strep tests, and a majority of patients with abscesses are on antibiotic therapy that covers strep. (J Laryngol Otol 1991;105:553; Fam Pract 1996;13:317; J Fam Pract 2000;49:34.) Fairly good evidence shows that rapid testing and antibiotics do not prevent peritonsillar abscess.
Antibiotics may decrease symptoms only if given early, but who cares? Articles published in the 1980s and a few recent reviews show an advantage of antibiotics if given early after symptoms appear (J Pediatr 1985;106:870; Pediatr Infect Dis 1984;3:10; JAMA 1985;253:1271; BMJ Med 2000;320:150), but there are several problems with this apparent benefit. Most patients do not present to their doctor on the first day of symptoms, and it is not clear that antibiotics provide any symptom benefit if they present on day two or three. And the primary symptom is pain, and no data have compared analgesia with an antibiotic and analgesia. It is common sense that an antibiotic may be irrelevant if you have good analgesia. Symptoms for 16 to 24 fewer hours does not necessarily translate into returning sooner to school, work, or activity.
Antibiotics probably don't decrease spread of disease. Twenty-four hours of antibiotics do reduce the recovery of GABHS from cultures of the pharynx, but this is a surrogate marker, not the same thing as 24 hours of an antibiotic decreasing spread (which is the way these data are often translated). These are different for several reasons. Most patients have symptoms at school or work before they receive antibiotics, and they have likely already spread GABHS. Many patients also are treated empirically with antibiotics for presumed GABHS. The majority of these patients have a virus; even 50 percent of adult patients with four of the four Centor criteria have a viral illness. (J Gen Intern Med 2007;22:127.) Returning patients prematurely just because they have had an antibiotic for 24 hours will actually increase the spread of viral disease. Many patients are tested with rapid streps to determine whether they are candidates for antibiotic treatment; unfortunately, the rapid streps performed in a community setting falsely reassure patients and parents more than they accurately treat.
Thirty percent of culture-proven GABHS patients tested with “rapid strep tests” have an initial negative screen in clinical community studies. (Pediatrics 2009;123:437; Clin Microbiol Rev 2004;17:571.) Sending these “negative streps” back to school or work prematurely before the culture, which can take two to four days, could actually increase GABHS exposure if the patient is not on antibiotics. Fifteen percent of pediatric patients have false-positive strep screens (carriers). (2012 Red Book. 28th ed., Elk Grove Village, IL: American Academy of Pediatrics, 2009.) They are sent back into the community after 24 hours of an antibiotic to potentially spread viral disease.
We forget that “strep throat” includes non-group A beta-hemolytic strep (group C and G), which is not uncommon. Non-group A BHS is not tested for on the rapid test or culture. (Arch Intern Med 1990;150:825; J Clin Microbiol 1997;35:1.) So all cases of non-group A BHS that routinely go without testing or antibiotics are sent back into the community, with no apparent sequelae from untreated “strep throat” for the individual, but they do spread strep throat in the community.
The premise that a person can return to work or school after antibiotics for 24 hours is an unproven assumption. Given the poor sensitivity of rapid streps in clinical community practice, the false-positives from carriers, and the neglect to test or treat non-group A BHS, it is, in fact, a more reasonable assumption that we increase the spread of viral disease and strep throat with this current policy.
The party line against the current indiscriminate use of antibiotics for “strep throat” in the ED is increased antibiotic resistance in the community and increased risk of anaphylaxis. I am against antibiotic usage for viral illness, but it does not appear that we have created any penicillin-resistant strains of strep throat nor have we been able to measure any increase in anaphylaxis from the overuse of antibiotics in sore throat.
The Centor criteria (fever, exudate, cervical adenitis, no cough) was endorsed nationally as a scoring system to balance the arguments for and against; it uses a clinical scale and tests and treats only when necessary. This approach when followed can decrease testing and antibiotic prescribing, which is good. Sadly, less than one percent of physicians document any scoring. (Arch Intern Med 2006;166:1374.) This is not just a documentation problem because 70 to 75 percent of adults get antibiotics for sore throat when the prevalence is only 10 percent for GABHS. (Ann Intern Med 2003;138:525; JAMA 1997;278:901.) Fifty-three to 69 percent of children get antibiotics when the prevalence is only 15 to 30 percent. (JAMA 2005;294:2315.) Physicians on the whole are just not using it, even after more than 30 years of having the nationally recommended scoring system.
The American Academy of Pediatrics and others have strongly emphasized discretion in testing children under age 3 or testing patients with obviously viral symptoms (hoarseness, runny nose, wheezing, diarrhea, other people in the house with a “cold”). Still, 80 percent of pediatricians and family practitioners test children with obvious viral symptoms. (New Engl J Med 2011;364:648.) Again, physicians continue to ignore recommendations in spite of aggressive educational campaigns.
Even when the science is strongly supportive for a certain course of action, we tend to shift the blame to the patient (or the mother). She will not accept a “no test/no antibiotic approach!” I believe this is also unlikely to be true. One recent trial found that symptom duration was most closely related to patient satisfaction measured by addressing patient concerns rather than the use of antibiotics. (BMJ 1997;314:722; BMJ 1999;319:736.) Trust, not prescriptions, translates into patient satisfaction.
Honestly, testing or not testing, antibiotic or no antibiotic, probably does not make any clinical difference in the ED, but a little common sense significantly influences the emergency physician's satisfaction and certainly the patient's.
- Not testing is faster.
- Not testing is cheaper for the patient. (A strep test in our ED is $127.)
- Not offering antibiotics will save at least one in 10 patients from getting diarrhea, a rash, or a vaginal yeast infection, which can incur more cost for another doctor visit. (JAMA 2010;304:2161.)
- Good information and good analgesia with a note for two to three days away from school or work (regardless of the etiology) will do more for symptom reduction, patient satisfaction, and the communicable spread of disease than any current strategy that involves testing and antibiotics.
If this approach seems too cavalier because it is not the party line of the CDC or the IDSA or AAP, it might be reassuring to consider that almost all European guidelines consider acute sore throat to be a benign self-limiting disease where testing is not routinely performed and antibiotics are not routinely prescribed. (Clin Ther 2011;33:48.) (See table for this question presented in a value medicine format.) Try a “no test-no antibiotics approach, and be sure to provide adequate analgesia for the individual and write a note letting him stay out of school or work for two to three days if you really want to prevent the spread of all diseases in the community.
The way we present this to the patient and parent is critical. Try to touch on several points:
- Tell the patient that you think it is important to know viral sore throat and strep throat go away without antibiotics.
- Explain that pain brought the patient to the ED, and that you will offer good pain relief to help him get through meals.
- Note that you can save the patient 30 to 60 minutes and about $130 if you forgo testing, which can be wrong in about one of three patients.
- Ask the patient if he feels comfortable with this approach.
Number Needed to Treat
Number Needed to Benefit with Antibiotics
- 0 of 100 to prevent rheumatic fever (U.S. prevalence approaches 0, no epidemics since 1980s).
- 0 of 100 to prevent peritonsillar abscesses (majority of peritonsillar abscesses have negative rapid strep tests).
- 0 of 100 feel better at one week compared with no antibiotics (antibiotics do decrease symptoms 16–24 hours quicker).
- No good data exist that antibiotics decrease communicability (would only be true in theory if one received antibiotics immediately at symptom onset and initial testing were reliable).
Number Need to Harm with Antibiotics
- No data available on the prevalence of anaphylaxis from unnecessary antibiotics (but prevalence is very low).
- No data available on the amount of diarrhea, rash, or side effects from unnecessary antibiotics (data from otitis media suggest about 10 of 100).
- No data available on worsening antibiotic resistance from unnecessary antibiotics (but presumed it could increase spread of viral illness by prematurely sending patients back into school or the workplace after “24 hours on an antibiotic,” but likely high).
Number Not Affected by Antibiotics
- 100 of 100 patients will not be significantly helped by antibiotics.
- The most positive spin one could make is that antibiotics may reduce symptoms 16–24 hours faster and in theory could reduce spread of GABHS. (Symptoms are probably helped better by analgesia, and the antibiotics approach is offset because 10 of 100 patients experience side effects.)
- Penicillin and amoxicillin are on the $4 list and bedside strep screens in offices (I'm told) are inexpensive, making antibiotics vs. no antibiotics somewhat of a wash. This is not true in the ED.
- The 2011 charge to the patient for a rapid strep screen in our ED is $127. (This cost is higher for a second ED visit for diarrhea from amoxicillin.)
- Testing 100 patients in the ED and treating half of them costs $13,000.
- Testing and prescribing antibiotics for 100 adult or pediatric patients in the ED with sore throat/strep throat costs $13,000 and provides no benefit for the individual (which analgesia can't equal) but may cause side effects in 10 of 100 patients.
- The benefit to the community is theoretical for immediate antibiotics only for true GABHS. In the real world of sore throats, this is unrealized and outweighed by false reassurance which spreads viruses and non-group A BHS.