Multidrug resistance has complicated tuberculosis therapy. We studied antibiotic susceptibilities of Mycobacterium tuberculosis and predictors of multidrug resistance to assist in determining initial drug regimens.
We conducted a case-control study based on chart review of patients with and without multidrug-resistant tuberculosis, including outpatients and inpatients with culture-proved tuberculosis seen at a large New York, NY, hospital during 1991 and 1992. Patient characteristics studied included serologic findings for human immunodeficiency virus and the presence of the acquired immunodeficiency syndrome. Descriptive analysis considered potential initial drug regimens. A theoretically effective regimen was assumed to contain at least two drugs to which an isolate was susceptible.
For 172 patients, 28.5% of isolates were resistant to isoniazid, at least 20.9% to rifampin, 15.7% to ethambutol, 8.1% to pyrazinamide, 18.6% to streptomycin, 9.9% to ethionamide, 8.1% to kanamycin, and none to capreomycin, cycloserine, and ciprofloxacin; 18.6% were resistant to both isoniazid and rifampin. Chart review of 159 patients showed that acquired immunodeficiency syndrome, human immunodeficiency virus seropositivity, female gender, residence in the Bronx, and race were associated with multidrug resistance. The four-drug regimen of isoniazid, rifampin, ethambutol, and pyrazinamide was theoretically effective for 81% to 85% of patients. No subset of patients would have a markedly better theoretical benefit from that regimen. Only five- or six-drug regimens that used the combinations of capreomycin plus ciprofloxacin, capreomycin plus cycloserine, ciprofloxacin plus cycloserine, or all three drugs together theoretically offered significantly higher effectiveness.
Tuberculosis isolates at our hospital have a high frequency of multidrug resistance. Only five- or six-drug regimens are theoretically adequate as initial therapy for our patients.(Arch Intern Med. 1994;154:2161-2167)