Table 1 Summary of studies included in the review

 Morano, 2014 [21]

United States

To evaluate the efficacy of a mobile medical clinic screening program for detecting latent and active TB

Analysis of data routinely collected by the clinic, January 2003 to June 2011

Ever used cannabis. No further details provided

Latent TB was identified through TST. Those with a positive TST (≥10 mm for HIV-negative individuals and ≥ 5 mm for HIV-positive individuals) were assessed for active disease by screening for symptoms and chest X-ray. Those with symptoms or abnormal chest X-ray results underwent sputum culture

N/A

4650 people were screened of which 779 were identified as prevalent TB cases (newly and previously diagnosed)

Of all those with TB (newly or previously diagnosed), 62% were male, 48% foreign-born

Cannabis use was associated with incident (but not prevalent) TB infection in adjusted analyses - aOR: 1.57; 95% CI:1.05–2.37. Adjusted for foreign-born, unstable housing, Hispanic ethnicity, Black ethnicity, employed, age, hazardous drinking, crack cocaine use

 Munckhof, 2003 [11]

Australia

To report a cluster of TB cases and investigate whether shared use of a cannabis water pipe was associated with TB transmission

Retrospective cohort study / outbreak investigation

Smoking a cannabis water pipe (‘bong’)

Active TB cases were identified via contact screening and self-referral and diagnosis confirmed with clinical examination and positive sputum culture (all isolates were identical on typing)

All contacts were screened with an initial TST and chest radiograph. If the initial TST was not significant, a second TST was performed 3 months after the last contact with the relevant index case

5 pulmonary TB cases (all were culture positive, 4 of whom were sputum AFB smear positive), 149 identified contacts

All cases were young (median age 20 years), Australian-born, HIV-negative, Caucasian males

Of 45 contacts who had shared a cannabis bong with a case, 29 (64%) had a significant TST reaction. There was some (weak) evidence that sharing a cannabis water pipe with a pulmonary TB case was associated with an increased risk of acquiring TB infection (univariable association p = 0.05 [odds ratio not presented], adjusted OR: 2.22, 95% CI 0.96–5.17). The model adjusted for gender of contact, whether shares a house with case, and previous BCG vaccination. In this cluster, cannabis smoking was frequently performed in closed rooms or cars

 Thu, 2013 [22]

Australia

To report on 3 cases of cavitating disease, which seemed to be associated with smoking cannabis using a makeshift pipe or bong

Case series plus retrospective cohort of contacts - reports on 3 cases of active TB and results of the screening of their contacts

Smoking a cannabis water pipe ‘bong’

The 3 index cases all had cavitating pulmonary TB and were AFB sputum smear positive

Contacts were screened using TST

3 TB cases who had a total of 111 identified contacts (34 were positive for latent TB, one of whom developed active disease)

All index cases were young Australian-born, non-indigenous adults

All 3 index cases regularly smoked cannabis through a bong. The unadjusted odds of having TB infection (positive TST) in contacts who shared a bong with an active TB case vs. those who did not report doing so was 6.5, 95% CI: 1.4–30.4, p = 0.016

 Oeltmann, 2006 [9]

United States

To describe a TB outbreak fuelled by illicit drug use

Retrospective cohort study / outbreak investigation. Isolates from all culture positive TB patients in Seattle and King County, Washington, 2003–2004 were genotyped. Patients who had an isolate that matched the outbreak strain or who had a social link to an already included patient were included in the study

‘Hotboxing’ (all cases reported frequent hotboxing)

TB cases were those that were culture-confirmed

Contacts were defined as friends or others. Friends were contacts of cases who spent time within a closely connected network of young men who exhibited similar cannabis-using behaviour. Other contacts were family or relatives and others not closely associated with the network. Contacts were evaluated for TB including a TST

11 TB cases (8 were AFB sputum smear positive), 121 contacts (54 friends; 67 other)

8/11 TB patients were born in East Africa.

Median age: 22 years

After those with a past positive TST were removed, 14 (64%) of 22 screened friends and 6 (23%) of 26 other contacts had TB infection (positive TST). The (unadjusted) risk of a positive TST result was 2.8 times greater among friends than among other contacts (95% CI:1.3–6.0). 54% friend contacts self-reported or were observed hotboxing; among those who had a TST 79% (11/14) had a positive result

 Han, 2010 [19]

United States

To investigate associations between duration of use of specific illicit drugs (including cannabis) and lifetime specified health conditions (including TB)

Cross-sectional analysis of National Surveys on Drug Use and Health data for 2005–2007; a nationally representative sample of non-institutionalised civilians. Analyses were restricted to adults aged 35–49 years

Duration of cannabis use (age at initiation minus age at last use). No further details provided

Self-reported whether a doctor or other medical professional had ever diagnosed them with TB

N/A

N/A

Adults aged 35–49 years

Cannabis use (of any duration) was not associated with the risk of TB disease (reference group never used cannabis, adjusted odds ratios [aORs]: 0.79 (95% CI:0.33–1.87) for those with ≤1 year use, 0.72 (95% CI:0.25–2.06) for those with 2–10 years use, and 0.73 (95% CI:0.33–1.58) for those with ≥11 years use). Adjusted for non-medical use of prescription drugs, duration of alcohol use, duration of tobacco use, daily cigarette smoking history, age, gender, race/ethnicity, years of education, health insurance, and family income

 Davis, 2017 [20]

Kazakhstan

To examine associations between various risk factors (including drug consumption) and TB

Matched case-control study. The sample comprised 562 individuals with TB (identified by the national TB programme) plus one household and one community control were identified for each index case. Study conducted June 2012 to May 2014)

Ever used cannabis; used cannabis in the past 90 days. No further details provided

Cases had to have been diagnosed with new pulmonary TB by positive culture or on clinical and radiographic grounds within 3 months before study enrolment and respond to anti-TB treatment. Household and community controls self-reported at baseline that they did not have TB

N/A

Study population consisted of 562 TB cases and 1308 controls

Participants were ≥ 18 years

In unadjusted analyses having ever used cannabis was associated with an increased odds of being a TB case (OR: 1.86, 95% CI:1.21–2.85, p = 0.005), as was having used cannabis in the last 90 days (OR:2.18, 95% CI:1.00–4.73, p = 0.049). However, in the multivariable model there was no evidence of an association with ever having used cannabis and TB (aOR:1.64, 95% CI: 0.76–3.54, p = 0.210) [cannabis use in last 90 days not analysed in multivariable model]. Model adjusted for: age, country of origin, marital status, education, criminal history, chronic disease, and comorbidities (HIV and diabetes mellitus)

 Livengood, 1985 [23]

United States

To report on a community outbreak of isoniazid-resistant TB

Retrospective cohort study / outbreak investigation. Following the identification of a patient with culture and AFB sputum smear positive isoniazid-resistant TB extensive contact tracing (family, work and community contacts) was undertaken

Smoking cannabis with the index case at home, work and/or via ‘hotboxing’ in a car

Clinical and radiological diagnosis plus positive sputum smear in case

Initial close contacts plus extensive investigation of other community contacts after resistant strain realised. TST was used to identify TB in contacts (induration > 10 mm = positive TST)

One index case was identified, and a total of 104 contacts screened. One secondary case with active disease was identified

The index case was a 35-year-old male. Most of the TST-positive contacts (36/42) were associated with a local tavern frequented by the index patient

42 contacts had a positive TST result. Smoking cannabis with the index case was identified as the most important social risk factor for TB transmission - 100% (14/14) of contacts who smoked cannabis with the index case had a positive TST result

 Merritt, 2007 [24]

Australia

To characterise a pulmonary TB cluster in the Hunter Area of New South Wales using a combination of traditional epidemiological methods and molecular typing

Outbreak investigation. All notifications for TB in the Hunter Area January 1994 to June 2005 were reviewed, and genotyping conducted for isolates from people who were born in Australia or New Zealand

No information

Cluster cases were confirmed if the isolate genotype was indistinguishable by MIRU, spoligotyping and IS6110 RFLP analysis. Cases missing only the RFLP analysis, or with an RFLP pattern that differed by only one band, were included if an epidemiological link was confirmed

For each case contact tracing records were reviewed and the person re-interviewed where traceable. Additional contacts were sought, and previous uncontactable contacts were followed up

9 cases were identified to be part of the cluster, and two probable cases. Number of contacts not reported

Cluster cases were people born in Australia or New Zealand. The median age at diagnosis was 35 years.8 of the 11 cases were women

Cannabis use was identified among six of the nine confirmed cases and both of the probable cluster cases

 Sterling, 2000 [10]

United States

To describe a TB outbreak among a highly mobile population

Outbreak investigation. Active TB cases were identified following referral (to the Baltimore Chest Clinic), June 1998 to January 2000, and through a DNA fingerprint database for isolates matching the outbreak isolate. Contacts were identified through traditional means and additional investigations

No information

Cases were culture confirmed TB with an identical DNA fingerprint, or clinical (culture-negative) TB cases with epidemiological links to outbreak cases

Close contacts named by cases, and those identified during TB treatment visits to the home of the source case, were investigated for TB infection or disease. Work-site screenings of close contacts was conducted, and location-based screening in a nightclub frequented by source cases. TST was performed to identify TB infection

20 outbreak cases (18 culture-confirmed); and 114 contacts (77 of whom were screened for TB) were identified

Cases were predominantly young (median age 24 years), black African-American (95%), and male, 11/20 were HIV-positive

35% (7/20) of TB cases reported cannabis use

 Evans, 2011 [26]

United Kingdom

To describe the identification of, and risk factors for, the single most prevalent TB strain in one area of the UK

All TB isolates in the West Midlands region (2004–2008) were genotyped to identify the most prevalent strain. Two epidemiological investigations were then undertaken. The second of these (reported on here) was an analysis of case notes of a cohort of patients in one city of the region with an indistinguishable MIRU-VNTR profile

No information

Cases were culture-confirmed and clinically diagnosed patients with the MIRU-VNTR profile of the “Mercian” strain, 2003–2006. Comparison group was culture-confirmed cases with other strains of TB diagnosed in 2004

N/A

Analyses were conducted on 35 patients with the Mercian strain, compared with 47 patients with non-Mercian strains of TB

Those with the Mercian strain of TB were more likely to be of white ethnicity and UK-born compared with those with a non-Mercian strain

In unadjusted analyses cannabis use was associated with an increased odds of having the Mercian strain: 11/35 (31%) of those with the Mercian strain reported cannabis use compared with 2/47 (4%) of those with a non-Mercian strain (OR: 10.02, 95% CI:1.96–100.33, p < 0.01). In adjusted analyses excess alcohol use and cannabis use were assessed as a combined variable. The two factors combined were associated with an increased odds of having the Mercian strain (aOR 6.26, 95% CI:1.45–27.02, p = 0.01), after adjusting for whether UK-born, previous contact with TB case, and presence of cavitary disease

 McElroy, 2003 [25]

United States

To evaluate whether social network analysis can provide insights into transmission settings that might otherwise go unrecognized by routine practice

Retrospective cohort study / outbreak investigation. A cluster of 22 TB patients diagnosed between 1994 and 2001 was investigated. All adult outbreak-associated cases (n = 19) and selected named and unnamed contacts (n = 26) were re-interviewed using a network analysis questionnaire

Cannabis use in the routine investigation. Drug use practices (including ‘shotgunning’) in network analysis sample

Cases were the adult outbreak-associated TB patients included in the routine contact investigation

Primary contacts - those identified by TB cases and interviewed at time of routine investigation (1994). Secondary contacts named by either primary contacts or cases at time of network analysis interviews (2001) and those identified during original contact investigation. TST performed on contacts (induration ≥5 mm = positive TST)

19 adult TB cases were part of the cluster. 7/9 female cases had sputum acid-fast bacilli smear-positive cavitary pulmonary TB. 90 primary contacts were identified, of whom 26 were selected (convenience sample) for interview

19 adult cases included 10 men and nine women. The 7 female patients with sputum AFB smear-positive cavitary pulmonary TB worked as exotic dancers in a specific geographic area. Epidemiological links to female patients had previously been established for the majority of male patients. Median age of cases was 35 years

The convenience sample of primary contacts included 15 TST-positive and 11 TST-negative contacts. 79% of the 19 active TB cases reported cannabis use, compared with 35% of all primary contacts (33% of those TST positive and 36% of those TST negative) [authors state that “As this is not a standard case control study, the comparison of cases and primary contacts in this instance is purely descriptive”]. 44% of the 9 female cases reported ‘shotgunning’ (all used crack but it is not clear how many also used cannabis)