TABLE 5

Comparison of shorter regimens and individualised longer regimens for outcome of failure or relapse versus success, using propensity score matched individual patient data meta-analysis

	Studies shorter:longer	Shorter events/total	Longer events/total	Propensity score matched multivariable meta-regression
	Studies shorter:longer	Shorter events/total	Longer events/total	Pairs	aOR (95% CI)	aRD (95% CI)
Overall	9:38	118/2282	112/1926	1926	2.0 (0.96–4.0)⁺	0.02 (0.00–0.05)⁺
HIV status strata
People living with HIV	5:10	24/295	55/750	295	2.1 (0.6–7.7)	0.03 (−0.02–0.07)
HIV negative	9:38	94/1978	56/1162	1162	1.9 (0.9–4.0)	0.02 (−0.00–0.04)
Extensiveness
Extensive	9:36	91/1969	83/1320	1320	1.2 (0.6–2.6)	0.01 (−0.02–0.03)
Not extensive	8:26	20/259	28/602	259	2.9 (0.8–10)	0.05 (0.00–0.09)
Pyrazinamide DST^#
Resistant	5:26	36/270	11/349	270	10.7 (1.8–64.5)^¶,+	0.12 (0.07–0.16)⁺
Susceptible	5:23	12/248	13/428	248	1.3 (0.3–6.7)	0.01 (−0.04–0.05)
Ethionamide/prothionamide DST^#
Resistant	5:26	23/249	4/149	149	3.9 (1.0–15.1)^¶,+	0.07 (−0.01–0.16)⁺
Susceptible	4:30	7/660	26/613	613	0.1 (0.0–1.5)	−0.03 (−0.05– −0.01)
Ethambutol DST^#
Resistant	8:37	39/692	27/554	554	3.1 (1.8–5.3)^¶,+	0.09 (0.04–0.13)⁺
Susceptible	3:23	1/297	9/334	297	0.2 (0.0–1.9)	−0.02 (−0.04–0.01)

Data are presented as n, unless otherwise stated. aOR: adjusted odds ratio; aRD: adjusted risk difference; DST: drug susceptibility testing. All models adjusted for age, sex, HIV status, previous treatment with first-line tuberculosis medications and extensiveness of disease. Results were adjusted as described in the Methods. ^#: analyses restricted to patients with DST-confirmed fluoroquinolone susceptibility (i.e. excluding those with no DST data for fluoroquinolones); ^¶: aOR calculated from fixed effect model; ⁺: confidence intervals suggestive of increased odds or risk of failure or relapse with the shorter regimen.