In Figure 1, we show the chance of the woman remaining free of genital chlamydial infection after treatment with either azithromycin or doxycycline, assuming that a rectal swab was not taken. The ranges reflect the autoinoculation probability varying from zero to one. When the probability of autoinoculation is one, the chance of the patient remaining free of genital infection after treatment with doxycycline is 96.8%, and 81.9% for azithromycin. That is, a 3.2% and 18.1% chance, respectively, of not clearing the infection. This corresponds to a 5.7-fold greater chance of not clearing an infection with azithromycin compared with doxycycline.
If the woman was known to have a rectal infection before treatment (i.e., if a positive rectal swab had been taken) then the ranges become wider: the chance of remaining free of genital infection after treatment with azithromycin is 78.2-94.3%, and after treatment with doxycycline 96.7-97.1%. This corresponds to a 2.0-6.6-fold greater chance of not clearing a chlamydial infection with azithromycin compared with doxycycline.
Using estimates from the literature of the efficacy of azithromycin and doxycycline in treating genital and rectal chlamydial infections, along with some simple assumptions, we have obtained estimates of the percentage chance of remaining free of chlamydia after treatment for genital chlamydia when the possibility of autoinoculation from rectum to genitals is taken into account. If autoinoculation does not occur, the efficacies of azithromycin and doxycycline are as reported in the systematic review (94.3% for azithromycin and 97.1% for doxycycline) and there is approximately a 2-fold greater chance of not clearing the infection with azithromycin. If autoinoculation has a high probability of occurring, then the efficacy of azithromycin may be as low as 81.9% when a patient’s rectal infection status is unknown, and as low as 78.2% if the patient is known before treatment to have a rectal infection. This means that the chance of not clearing an infection could be 6 times greater with azithromycin compared with doxycycline. It has usually been assumed that new infections detected after treatment for genital infection are due to re-infection by a partner, but it may be that some are due to autoinoculation. The disparity in efficacies provides further support for the careful re-evaluation of azithromycin as the preferred treatment for chlamydia.
Measuring the probability that autoinoculation from rectum to genitals occurs would be difficult. Assuming that there is some daily chance that autoinoculation occurs, whether autoinoculation has taken place would be a function of time since treatment. However, this is also the case for re-infection from a partner, and it would be difficult to separate the effects of these two mechanisms of re-infection. We recommend post-treatment tests for re-infection at both genital and rectal sites, as this will of course capture re-infection regardless of its source.
The ‘use-effectiveness’ of doxycycline does seem to be high, with a study that monitored adherence using microchipped medication bottles finding that chlamydial infection resolved in 76 of 81 (93.8%) of patients [17]. However, adherence did have an impact, with all 4 of the patients who failed therapy and were evaluable (i.e., returned their medication bottles) having at least two 24-hour intervals during which they did not take medication. Additionally, none of the evaluable 58 patients who took at least 10 doses failed therapy, while 4 of the evaluable 20 patients who took less than 10 doses failed therapy. While it seems that good use-effectiveness can be had from doxycycline, excessively low adherence is clearly to be avoided. Providers proscribing doxycycline for treatment of urogenital chlamydia should continue to encourage patients to take their full courses of medication.