Table 3. The effect of prophylactic INH on the 1-year incidence of TB according to the presence of risk factors.

	Non-high-risk subgroup (n = 1526)	High-risk subgroup (n = 92) a
Control group, n	1390	76
INH group, n	136	16
Number of TB cases / Observation period in the control group	12 / 1362.2 person-years	7 / 69.5 person-years
Number of TB cases / Observation period in the INH group	2 / 132.7 person-years	0 / 13.5 person-years
Crude hazards ratio of the INH group (95% CI)	0.76 (0.38–7.62)	0.31 (0.002–2.46)
Adjusted hazards ratio of the INH group (95% CI)	1.01 (0.20–5.08) b	0.37 (0.002–5.10) c

CI, confidence interval; INH, isoniazid; TB, tuberculosis.

^a Firth penalized maximum likelihood was used due to complete separation of outcome.

^b Included concomitant steroid pulse, duration of high-dose steroid treatment from the baseline and cumulative steroid dose as covariates, and was also adjusted for clustering.

^c Included age, male sex, duration of high-dose steroid treatment from the baseline and mean steroid dose used during the prior 6 months as covariates, and was also adjusted for clustering.