Table 1. Summary of studies.

First author, year	Country	Years of study	Study description	Study conclusion	Number in cohortg	Ethnicity	CD4 count category	Number of TB cases	Patient years	TB incidence per 1,000 person-years	Bacteriologicaly confirmed TB (%)	TST positive (%)	Age range (median)	Male (%)	Epidemic type	Qualitya (selection/ comparability/outcome)
Assebe, 2015	Ethiopia	2008–2012	Retrospective cohort study of patients in pre-ART care at a tertiary hospital focusing on the effect of IPT. Cohort comprised IPT (n = 294) and non-IPT (n = 294) patients sampled from 3,476 in pre-ART care. Median follow-up 24.1 months. Adjusted hazard ratio from Cox PH model reported for CD4 categories (adjusting for IPT use) based on n = 585 patients.	IPT found to halve TB incidence.	588	Black African	<350	49	1297.5	aHR= 3.16 (1.04–3.92)	–	–	15–64	38	Heterosexual	B/A/B
							350–499			aHR = 2.87 (1.37–6.03)
							≥500			aHR = 1g
Collins, 2015	Haiti	2005–2012	Open-label RCT, using the CIPRA-HT001 cohort to receive early ART (n = 408) or delayed ART (i.e., when CD4 < 200; n = 408). Median follow-up of 4.4 in the early treatment arm and a median 3.9 years in the delayed treatment arm. 43 participants with prevalent TB at initiation excluded from incidence analysis, which reported an adjusted HR of TB incidence per 50 cells/mm3 CD4 decrement from Cox PH models (this included person-time on ART). Model including adjustments for treatment group and PPD-status/IPT used.	Delayed ART initiation results in persistent increased TB risk.	816	–	–	96	aHR = 1.24 (1.11–1.38) per 50 CD4 cells/mm3 decline		61	27e	≥18 (40)	42	–	B/A/A
Grant, 2009	UK	1996–2005	An observational cohort study of patients (UK Collaborative HIV Cohort, restricted to those who joined prior to 2006); median follow-up 4.2 years. TB within 3 months of enrolment excluded, as were individuals without CD4 count or who were lost to follow-up or died within 3 months of first visit. TB incidence reported separated for those not taking ART.	Early HIV diagnosis and IPT key to reducing the particularly high TB incidence in non-white & low-CD4 PLHIV in the UK.	22,833	Black African	<50	9	176	51.1	–	–	≥16 (34)	36	Heterosexual	A/B/B
							50–199	13	612	21.2
							200–349	14	1,453	9.6
							350–500	7	1,443	4.9
							>500	7	1,553	4.5
						White	<50	9	850	10.6			≥16 (35)	92	MSM
							50–199	25	2,631	9.5
							200–349	20	6,636	3
							350–500	9	7,548	1.2
							>500	3	9,560	0.3
						Other	<50	3	252	11.9			≥16 (34)	77	MSM
							50–199	4	535	7.5
							200–349	5	1,422	3.5
							350–500	11	1,587	6.9
							>500	1	2,180	0.5
Markowitz, 1997	USA	1988–1990	Prospective, multi-center cohort study of HIV seropositive patients (the Pulmonary Complications of HIV Infection Study) representative of US PLHIV; median follow-up of 53 months. Around 4% of patients were taking IPT. TB incidence increased with PPD induration.	TB incidence highest in patients with CD4 <200 and PPD patients.	1,130	White/Black /Hispanic	<200	17	1,417	12	71	6d	18–67 (37b)	87	MSM/IDU /heterosexual	C/B/A
							≥200	14	2,800	5
Monge, 2014	Spain	2004–2010	Open, multi-center, prospective cohort of ART naïve patients (CoRIS = Spanish AIDS Research Network Cohort). Around half of TB cases were in people on ART. Adjusted rate ratios by CD4 category from Poisson regression (adjusting for ART-status) were therefore used. TB episodes during follow-up and up to 3 months prior to enrolment were included.	Early HIV diagnosis and ART should be used to help address high TB incidence in PLHIV.	6,811	–	<200	124	aHR = 5.20(3.25–8.33)		65	–	>13	80	Heterosexual /IDU	B/A/C
							200–350	32	aHR = 1.54(1.01–2.34)
							>350	37	aHR = 1g
Nicholas, 2011	Guinea, Kenya, Malawi, Mozambique, Nigeria & Uganda	2006–2008	Multi-center, retrospective cohort study based on the FUCHIA database from all sub-Saharan Médicins sans Frontières HIV programmes. TB at, or within 15 days of, enrolment excluded; patients with <15 days of follow-up excluded. Median follow-up 9 months. TB recorded pre-ART and during ART. Median pre-ART follow-up of 9 months.	Importance of early HIV diagnosis and treatment and implementation of the 3Is.	8,998	Black African	<50	59	98	602	–	–	27–40c (33)	27	Heterosexual	A/A/B
							50–99	68	170	401
							100–199	162	732	221
							≥200	398	6,180	64
Wolday, 2003	Ethiopia	1997–2001	A prospective cohort study (Ethio-Netherlands AIDS Research Project = ENARP) to study biomarkers associated with TB progression. HIV-positive and negative factory workers with an overall median follow-up of 3.8 years and 6-monthly assessments.	Low CD4 and high viral load associated with TB incidence; successful TB treatment does not reduce viral load.	95	Black African	<200	5	46	107.6	50	38d	(34)	52	Heterosexual	C/A/B
							200–499	5	121	41.2
							≥500	0	43	0

Notes.

^aQuality assessed with a modified Newcastle-Ottawa scale for cohorts with A/B/C according to all/some/few criteria met in each domain.

^bMean.

^cInter-quartile range.

^d5 mm cut-off used in tuberculin skin test.

^eCut-off used in tuberculin skin test not stated.

^fReference value.

^gCohort number restricted to those included in TB incidence calculation.

aHR

adjusted hazard ratio

ART

antiretroviral therapy

IPT

isoniazid preventive therapy

TB

tuberculosis

RCT

randomized controlled trial

PPD

purified protein derivative

TST

tuberculin skin test

PLHIV

people living with HIV

IDU

injecting drug users

MSM

men who have sex with men