How can you tell if a cold is caused by a virus or bacteria?

Many doctors believe that they can accurately distinguish bacterial from viral colds. They are deluded.

Well-established signs algorithms, like the Centor and McIsaac criteria, are not very accurate. Of the patients that show all four elements of strep infection (fever, exudates, tender lymph nodes, absence of cough), only about 60% actually have strep throat[1] . That’s the best-case scenario: patients that show all four signs (most don’t), and in clinical trials, which are highly structured and recruit motivated doctors. Actual performance by the average doctor on an average day is likely to be much worse.

To be fair, the Centor criteria do add value. Only about 10% of colds are bacterial[2] , so increasing the likelihood from 10% (with no additional information) to 60% (with the Centor criteria) is significant. But it’s still not much better than a coin flip.

The prospects for better diagnostics are good. An Israeli group published a beautiful paper comparing the levels of hundreds of serum proteins in patients that had infections[3] . This approach takes advantage of the fact that your immune system reacts differently to viral vs bacterial infections, making a different set of proteins in response. The Israelis formed a company (MeMed) to develop a point-of-care test, and have published some initial clinical data which suggests their test is about 90% accurate[4] . Another company, RPS Diagnostics, is developing a similar test. Results from early clinical trials suggest accuracy of about 85%.

So right now there is no good way – meaning fast, inexpensive and reliable – to distinguish bacterial from viral colds. But that will change in the next few years.

Featured on Forbes

[1] Large-scale validation of the Centor and McIsaac scores to predict group A streptococcal pharyngitis.
[2] Viruses and Bacteria in the Etiology of the Common Cold
[3] A novel host-proteome signature for distinguishing between acute bacterial and viral infections.
[4] Diagnostic accuracy of a TRAIL, IP-10 and CRP combination for discriminating bacterial and viral etiologies at the Emergency Department