Macrolide resistance: Cause for concern?

In the PURL, “Consider adding this drug to fight COPD that’s severe”(J Fam Pract. 2012;61:414-416), Drs. Hobbs and Brown state that “there was an increase in the prevalence of macrolide-resistant respiratory pathogens in patients on daily azithromycin.” This statement is technically correct but terribly misleading. It implies that azithromycin caused increased resistance, which it did not.

Prevalence is a proportion, and in this case refers to the proportion of all isolates that were macrolide resistant. An increased proportion may be due to either an increased numerator (resistance) or a decreased denominator (isolates).

In the study in question¹ there were actually fewer macrolide-resistant pathogens isolated during treatment with azithromycin compared with placebo. All else being equal, this would have resulted in a decreased prevalence. However, there were also far fewer total isolates in the azithromycin group. This relatively larger decrease in the denominator prevailed, resulting in “increased” prevalence, due to fewer pathogens, not more resistance. This finding (of fewer pathogens isolated) has a clinical correlate. The 2 largest trials comparing azithromycin with placebo both found decreased acute respiratory illnesses in the azithromycin groups compared with the placebo groups.^2,3

The correct way to assess resistance would have been to calculate the incidence of newly detected resistant pathogens over a defined period of time in both the azithromycin and placebo groups. In fact, the incidence of macrolide resistance was 24% lower in the azithromycin group (11.1 per 100 patients per year vs 14.9 per 100 per year in the placebo group).⁴ Thus, the increased “prevalence” referred to by Hobbs and Brown does not indicate increased resistance, but rather decreased pathogens.

David L. Hahn, MD, MS
Madison, Wis

Drs. Hobbs and Mounsey respond

We thank Dr. Hahn for his comments and agree that further clarification of the impact of azithromycin on macrolide resistance is appropriate. As Dr. Hahn notes, the number of colonized patients in the azithromycin group (66/479) was lower than in the placebo group (172/476), as would be expected because they had been on azithromycin for one year.¹ Dr. Hahn calculates the incidence of macrolide resistance using as the denominator all the patients in both the azithromycin and placebo groups and shows that the rate is higher in the azithromycin group.

We chose to determine macrolide resistance by comparing resistance rates only in the colonized patients (66 on azithromycin and 172 on placebo), not the whole group—the majority of whom were not colonized at all. Albert et al used similar methodology, reporting that “the incidence of resistance to macrolides was 81% [in the azithromycin group] and 41% [in the placebo group].”¹

So as Dr. Hahn states, patients on azithromycin were less likely to become colonized with bacteria, but when they did, the organisms were more likely to be macrolide resistant.

Whichever way the data are presented, the finding of macrolide-resistant organisms in 81% of the isolates after only a year must raise concern about the long-term use of prophylactic azithromycin. In a recent commentary on the use of prophylactic azithromycin, Wenzel et al called this a “major concern” and stated that the Albert trial was not long enough to elucidate the extent or clinical implications of the problem.²

Keia Hobbs, MD
Anne Mounsey, MD
Chapel Hill, NC

In the PURL, “Consider adding this drug to fight COPD that’s severe”(J Fam Pract. 2012;61:414-416), Drs. Hobbs and Brown state that “there was an increase in the prevalence of macrolide-resistant respiratory pathogens in patients on daily azithromycin.” This statement is technically correct but terribly misleading. It implies that azithromycin caused increased resistance, which it did not.

Prevalence is a proportion, and in this case refers to the proportion of all isolates that were macrolide resistant. An increased proportion may be due to either an increased numerator (resistance) or a decreased denominator (isolates).

In the study in question¹ there were actually fewer macrolide-resistant pathogens isolated during treatment with azithromycin compared with placebo. All else being equal, this would have resulted in a decreased prevalence. However, there were also far fewer total isolates in the azithromycin group. This relatively larger decrease in the denominator prevailed, resulting in “increased” prevalence, due to fewer pathogens, not more resistance. This finding (of fewer pathogens isolated) has a clinical correlate. The 2 largest trials comparing azithromycin with placebo both found decreased acute respiratory illnesses in the azithromycin groups compared with the placebo groups.^2,3

The correct way to assess resistance would have been to calculate the incidence of newly detected resistant pathogens over a defined period of time in both the azithromycin and placebo groups. In fact, the incidence of macrolide resistance was 24% lower in the azithromycin group (11.1 per 100 patients per year vs 14.9 per 100 per year in the placebo group).⁴ Thus, the increased “prevalence” referred to by Hobbs and Brown does not indicate increased resistance, but rather decreased pathogens.

David L. Hahn, MD, MS
Madison, Wis

Drs. Hobbs and Mounsey respond

We thank Dr. Hahn for his comments and agree that further clarification of the impact of azithromycin on macrolide resistance is appropriate. As Dr. Hahn notes, the number of colonized patients in the azithromycin group (66/479) was lower than in the placebo group (172/476), as would be expected because they had been on azithromycin for one year.¹ Dr. Hahn calculates the incidence of macrolide resistance using as the denominator all the patients in both the azithromycin and placebo groups and shows that the rate is higher in the azithromycin group.

We chose to determine macrolide resistance by comparing resistance rates only in the colonized patients (66 on azithromycin and 172 on placebo), not the whole group—the majority of whom were not colonized at all. Albert et al used similar methodology, reporting that “the incidence of resistance to macrolides was 81% [in the azithromycin group] and 41% [in the placebo group].”¹

So as Dr. Hahn states, patients on azithromycin were less likely to become colonized with bacteria, but when they did, the organisms were more likely to be macrolide resistant.

Whichever way the data are presented, the finding of macrolide-resistant organisms in 81% of the isolates after only a year must raise concern about the long-term use of prophylactic azithromycin. In a recent commentary on the use of prophylactic azithromycin, Wenzel et al called this a “major concern” and stated that the Albert trial was not long enough to elucidate the extent or clinical implications of the problem.²

Keia Hobbs, MD
Anne Mounsey, MD
Chapel Hill, NC

In the PURL, “Consider adding this drug to fight COPD that’s severe”(J Fam Pract. 2012;61:414-416), Drs. Hobbs and Brown state that “there was an increase in the prevalence of macrolide-resistant respiratory pathogens in patients on daily azithromycin.” This statement is technically correct but terribly misleading. It implies that azithromycin caused increased resistance, which it did not.

Prevalence is a proportion, and in this case refers to the proportion of all isolates that were macrolide resistant. An increased proportion may be due to either an increased numerator (resistance) or a decreased denominator (isolates).

In the study in question¹ there were actually fewer macrolide-resistant pathogens isolated during treatment with azithromycin compared with placebo. All else being equal, this would have resulted in a decreased prevalence. However, there were also far fewer total isolates in the azithromycin group. This relatively larger decrease in the denominator prevailed, resulting in “increased” prevalence, due to fewer pathogens, not more resistance. This finding (of fewer pathogens isolated) has a clinical correlate. The 2 largest trials comparing azithromycin with placebo both found decreased acute respiratory illnesses in the azithromycin groups compared with the placebo groups.^2,3

The correct way to assess resistance would have been to calculate the incidence of newly detected resistant pathogens over a defined period of time in both the azithromycin and placebo groups. In fact, the incidence of macrolide resistance was 24% lower in the azithromycin group (11.1 per 100 patients per year vs 14.9 per 100 per year in the placebo group).⁴ Thus, the increased “prevalence” referred to by Hobbs and Brown does not indicate increased resistance, but rather decreased pathogens.

David L. Hahn, MD, MS
Madison, Wis

Drs. Hobbs and Mounsey respond

We thank Dr. Hahn for his comments and agree that further clarification of the impact of azithromycin on macrolide resistance is appropriate. As Dr. Hahn notes, the number of colonized patients in the azithromycin group (66/479) was lower than in the placebo group (172/476), as would be expected because they had been on azithromycin for one year.¹ Dr. Hahn calculates the incidence of macrolide resistance using as the denominator all the patients in both the azithromycin and placebo groups and shows that the rate is higher in the azithromycin group.

We chose to determine macrolide resistance by comparing resistance rates only in the colonized patients (66 on azithromycin and 172 on placebo), not the whole group—the majority of whom were not colonized at all. Albert et al used similar methodology, reporting that “the incidence of resistance to macrolides was 81% [in the azithromycin group] and 41% [in the placebo group].”¹

So as Dr. Hahn states, patients on azithromycin were less likely to become colonized with bacteria, but when they did, the organisms were more likely to be macrolide resistant.

Whichever way the data are presented, the finding of macrolide-resistant organisms in 81% of the isolates after only a year must raise concern about the long-term use of prophylactic azithromycin. In a recent commentary on the use of prophylactic azithromycin, Wenzel et al called this a “major concern” and stated that the Albert trial was not long enough to elucidate the extent or clinical implications of the problem.²

Keia Hobbs, MD
Anne Mounsey, MD
Chapel Hill, NC