Life-threatening infections are not as commonly acquired from hospitals as they used to be, but they still cause much pain and suffering in both poor and wealthy countries. Depending on the year and how causes of death are categorized, hospital-acquired infections in the United States generally rank somewhere around the tenth leading cause of death. Certainly, many of these deaths occur because patients' defenses are severely compromised by surgical procedures, drugs that suppress the immune system, and other problems, such as severe burns. But just as certainly, attendants could be more conscientious in their hygienic practices. How far have we come? In 1991 a study of a nursery ward for newborns in Chicago showed that nurses followed appropriate hand-washing guidelines about half the time; doctors in the ward followed the guidelines half as frequently as that. Apparently the effectiveness of antibiotics in U.S. hospitals has caused generations of hospital staff to rely on antibiotics rather than on hygiene.

The reliance on antibiotics has generated an arms race with microbes. Most antibiotics are more poisonous to the microbes than they are to us, but clearly, it is in the patient's interest to use the lowest dosage that will eliminate the disease. Within microbe populations, however, there typically exists variation in the vulnerability to antibiotics. Some microbes may have biochemical machinery that binds to the antibiotic, or machinery that destroys the antibiotic, or biochemical pumps that can pump out the antibiotic. If the dosages of antibiotics are not sufficiently high to knock out all of the microbes, the resistant microbes will be left to repopulate the next generation, which will, as a result, be more resistant to the antibiotic than was the previous generation. This increment in antibiotic resistance then forces an increment in dosage. If the dosage is already at a threshold of acceptability, a new antibiotic must be found. This process is an evolutionary arms race: microbes evolve increased resistance, forcing humans to use higher dosages and new antibiotics, which in turn drive increased resistance, and so on.

The problem of antibiotic resistance in hospitals has been widely acknowledged, but the connection between antibiotic resistance and the virulence of hospital-acquired pathogens is often overlooked. Many researchers have had a sense that the antibiotic-resistant pathogens

circulating in hospitals were particularly nasty, but when they compared the pathogens isolated within hospitals, the expected difference between the harmfulness of the antibiotic-resistant and antibiotic-sensitive organisms was often not detectable. Antibiotic-resistant Staphylococcus aureus, for example, is particularly lethal in hospitals. Much of the damage is attributable to the compromised defenses of patients—burn patients are particularly likely to succumb. This damage is especially evident when antibiotics have little effect. But staph outbreaks also appear to cause more overt disease in nurses than one would expect from studies of the outside community, where although about one third of all people carry S. aureus, only a tiny percentage show any sign of serious disease. An evolutionary perspective suggests that comparison of one hospital form with another may be like looking for a lost key in the lamplight—though the hospital environment is the most convenient place to look for representative antibiotic-sensitive strains, it may not be the right place. It may be more informative to compare hospital strains with strains isolated in the community. Such direct comparisons have been sorely lacking.

Although the evolution of virulence in hospitals has been badly understudied, the available information supports the idea that cycling in hospitals makes pathogens more harmful; for example, a review of all hospital outbreaks of the sometimes benign, sometimes deadly diarrheal pathogen Escherichia coli showed that the longer the strains had been circulating in the hospital environment, the more lethal the outbreak—strains that had been circulating for just a week rarely caused death, but strains that had circulated for many months killed about one in ten infants.

Though the most frightening hospital strains of pathogens are the highly virulent

antibiotic-resistant strains, evolutionary considerations do not suggest that genetic instructions for antibiotic resistance make germs more harmful. Evolutionary considerations suggest instead that transmission in the hospital environment makes germs both more harmful and more resistant to antibiotics. Hospital strains—whether antibiotic-resistant or antibiotic-sensitive—appear to be inherently more harmful than the normal strains that circulate outside hospitals. To remedy this problem, we will need to heed the ghost of Semmelweis: hygienic practices must be improved.