Table 1 Summary of study characteristics

Hemilä et al, 1999, Finland, 198–1993

26,975 male smokers participating in RCT on the effect of nutritional support with a-tocopherol, P-carotene, or a-tocopherol + P-carotene for cancer prevention

Clinical diagnoses of TB ascertained from the discharge register of hospitals. 167 incident cases of TB registered from 1985 to 1993.

Self reported at baseline. Alcohol use categorized as 30 gram alcohol per day or more.

Age, BMI, martial status, education, residential neighbourhood, smoking, nutritional intervention

Adjusted relative risk: 1.03 (95% CI: 0.70–1.53)

Eight years follow up and change in drinking pattern not ascertained.

Prevalence of exposure among controls: 20%

Moran-Mendoza, 2004, British Columbia, Canada, 1990–2000

33,146 contacts of active TB cases recorded in division of disease control 1990–2000, who had a TST performed, excluding those with TB history and those with HIV, followed until 2001

Any type of TB, registered in the division of TB control database. 228 active cases identified.

Alcoholism as noted in medical record

Age, sex, Canadian-born, aboriginal, DM, malnutrition, malignancy, immunosuppressant treatment, BCG, no of contacts, type of contact, TST size, SES (geographical location), latent TB treatment, intravenous drug use, recent arrival from high TB incidence country

Adjusted relative risk: 2.9 (1.3–6.5)

Entire study population are TB infected. RR reflect risk of progress to active disease.

Prevalence of alcoholism among whole cohort: 0.8%

Thomas et al 2005, Tiruvallur district, Tamil Nadu, India, 2000–2001

503 cured new smear positive pulmonary patients as per TB district register, followed prospectively

TB recurrence within 18 months (62 recurrencess recorded)

Self reported during initial treatment. Exposure was "Habitual drinking", which was not defined in terms of amounts or frequency

Adjusted OR from multivariate analysis not reported.

Factors accounted for were sex, age, occupation, education, smoking, adherence, drug sensitivity, smear conversion, initial weight

Crude relative risk: 2.3 (1.3–4.1)

Level of exposure not provided, but since the prevalence of exposure of "habitual drinking" in the cohort was 33% in this rural Indian district, it not likely to correspond to high level consumption.

Case control studies

Author (Year), Setting

Cases and controls

Exposure Measure

Confounders Controlled for

Effect size (95% confidence interval)

Comments

Brown and Campbell. 1961, Hospital for ex-servicemen, Victoria, Australia, 1950s

Cases (100): All consecutive new admissions

Controls (100): Randomly selected from surgical ward in same hospital (excluding orthopedic cases)

Self reported daily consumption

Moderate to heavy drinking defined as 26 ml alcohol per day or more. Crude numbers for different level of exposure were reported, allowing calculation of association also at the > 50 ml (40 g) and other cut-off points.

Stratified by smoking status. All subjects were men. All ex-army staff in the age bracket 20–70. Age distribution very similar between cases and control. Pre- HIV era

Crude OR of moderate to heavy alcohol vs. none/low: 4.88 (95% CI: 2.59–9.24)

For > 50 ml vs =< 50 ml: OR 8.18 (4.05–16.53)

For 1–50 ml vs. none: 1.98 (0.89–4.43)

Significant (p < 0.0001) dose response relationship:

OR

0 (reference) 1.00

10–25 ml/day: 1.66

26–50 ml/day: 2.38

51–75 ml/day: 9.27

76–100 ml/day: 8.50

101–125 ml/day: 27.82

126- ml/day: 43.27

OR not analysed in original study. The ORs reported here are calculated based on crude data reported in the paper

Smoking possibly effect modifier. Stratified for none smokers and smokers respectively (any alcohol vs. no alcohol):

Non smokers: 2.25 (0.54–9.86)

Smokers: 5.22 (1.83–15.61)

Prevalence of "moderate to heavy alcohol intake" in controls: 39%

Lewis and Chamberlain, 1963, Hospital, London, 1962

Cases (100): Male, active cases of pulmonary TB

Controls (200): Matched for age and social class: A (100): From medical and surgical wards at the same hospital. B (100): From emergency department at another, general, hospital

Self-reported average daily consumption 6 months before symptoms started

"Regular drinking" defined as the equivalent of 2 or more pints per day.

Only men, stratified by age, social class, marital status and smoking. Pre-HIV era.

Crude OR for regular drinkers vs. not regular drinker 2.64 (95% CI: 1.50-4-66)

Did not change when stratified for smoking status: OR 2.68 and 2.61 in respective stratum

OR not analysed in original study.

Social class effect modifier? Stratified for SES:

Class I-II: OR = 1.16 (0.42–3.22)

Class III-V: OR = 4.07 (1.98–8.41)

Prevalence of "regular drinkers" among controls: 19.5%

UK pint = 568 ml. 2 pints of 5% beer contains about 45 g alcohol

Mori et al, 1992, Indian Health Service hospital, Pine Ridge Reservation, South Dakota, USA

Cases (46): All new, active, adult (18 years and above), cases registered between 1983–1989.

Controls (46). Randomly selected, matched for age and residence, from health care register in Reservation, where all residents are included

Chart review: Alcohol abuse/alcoholism listed in medical record, or alcohol related admission within 10 years or outpatient visit within 5 years

Matched by age and residence.

OR adjusted for sex, isoniazid profylaxis, and diabetes

All study subjects from same Indian community.

Adjusted OR (AOR) for alcohol abuse vs. no alcohol abuse: 3.8 (1.15–12.3)

Prevalence in control group: 32%

Buskin, et al, 1994, Seattle, King County Tuberculosis Clinic, Washington State, 1988–1990

Cases (151): Active TB cases, aged > 17 registered at TB clinic 1988–1990

Controls (545): Individuals seeking care at the clinic, but no TB diagnosed

Self reported frequency of drinking and amount consumed.

Heavy drinkers defined as 3 or more drinks/day or more than 5 drinks on average on each drinking occasion.

OR adjusted for age and smoking.

Sex, SES, BMI, and race were analysed, but did not influence result

Adjusted OR heavy drinking vs. non-drinkers 2.0 (95% CI: 1.1; 3.7)

1 US standard drink is 14 gram, thus 3 standard drinks is 42 gram.

Prevalence of heavy drinking in control group: 12.5%

Rosenman et al,1996, New Jersey, USA, 1985–1987

Cases (148): All active male, HIV-negative, cases over age of 35, born in USA, notified 1985–87

Controls (290): From Medicade finance administration files, matched for age and race

Self reported. "Heavy drinking" defined as > 22 alcohol equivalents/week

Only HIV- men in study, controls matched for age and race. Alcohol association not controlled for other variables in study, since alcohol was treated purely as confounder

Crude OR: 3.33 (1.99.5.59)

Prevalence "heavy drinkers" among controls: 14% 1 US standard drink is 14 gram, thus > 22 drinks per week = > 44 grams per day

Schluger et al, 1999, Social services agencies and chest clinic, NY, USA. 1994–1997

Cases (20): Persons screened positive for active TB among 3,828 individuals seeking social services

Controls (3,245): Those not screened positive for active TB

Self reported "moderate to heavy alcohol use". This was not defined further

None, but all subjects are social service clients

Crude OR 2.38 (0.88–6.58)

The authors did analyse, the study as a case control study.

Considering that the subjects were all social service clients and alcohol problem was common in this group, it can be assumed that "moderate to heavy" correspond to at least 40 g per day and/or alcohol abuse

Prevalence among controls: 43%

Spletter, 2000, TB Control Clinic, Phoenix, Ariziona, USA, 1993–1999

Cases (43): active pulmonary TB, 25–64 years old, excluding refugees, HIV positive, and comorbidity such as gastrectomy, jejunuilial bypass, DM, silicosis, renal failure, immunosuppressive treatment, malignancies.

Controls (258): Patients infected with M.tuberculosis, but active disease ruled out.

Medical record review: Heavy drinking defined as those with chart entries indicating alcohol abuse or alcohol history recorded as "heavy drinking"

See list of exclusion criteria. Controlled for age, sex, smoking, race, US born, high risk residence, illicit drug use.

Adjusted OR for heavy alcohol use vs. no heavy alcohol use: 6.1 (1.4; 26.2):

Entire study population are TB infected. OR reflect risk of progress to active disease.

Prevalence of heavy alcohol consumption in controls: 2.3%

Dong et al, 2001, 12 communes in Chengdu, China, 1996–97

Cases (174): All active TB cases recorded between March 1996 and March 1997

Controls (174): Random sample from community (population registry), matched for age, sex, and place of residence

Self reported use.

Definition of alcohol use or amounts not reported.

Matched for age and sex and district. Smoking, crowding, darkness in dwelling, air-pollution and BMI are reported variables, but not reported what was actually controlled for in the logistic regression

Adjusted OR (alcohol vs no alcohol): 1.76 (0.90–3.42)

Tocque et al 2001, Liverpool, UK, 1989–1996

Cases (112): All notified in the city

Controls (198): From Liverpool general practitioner database, matched for sex, age and residential area

Self reported, high consumption defined as > 30 units per week (> 4.3/day), both at time of interview and 2 years prior to diagnosis

Matched for age, sex, and residence area

Alcohol not included in multivariate analysis

Crude OR for drinkers vs. non-drinkers: 1.01 (0.67–1.70, at 2 years before diagnosis

One UK alcohol unit is 8 gram, thus 4.3 units/day = 34 gram

Tekkel et al, 2002, Hospital, Tallinn, Estonia, 1999–2000

Cases (248): consecutive, incident pulmonary TB cases admitted to one hospital in Tallinn

Controls (248): From population registry, matched for age, sex, and country of residence.

Self reported frequency of drinking during last year. Not defined in amounts of alcohol

Age, sex, and country of residence matched for. OR adjusted for smoking, drug abuse, nutrition, weight loss, contact with TB, place of birth, marital status, and education

Adjusted OR for people who consumed alcohol several times a week/day vs. rarely: 13.63 (4.63–40.10);

Prevalence of alcohol consumptions several times per week: 7.3%

Crampin et al, 2004, Karonga district, Malawi, 1996–2001

Cases (598):All new TB cases, aged > 15, residing in district

Controls (992): Random sample from community register, matched for age, sex and areas of residence.

Self reported as current (1/week or < 1/week), past, or never

Matched for age, sex, area of residence. Adjusted for SES, HIV, TB contacts, BCG

Adjusted OR for current 1/w vs. never: 0.9 (0.5–1.7)

Prevalence of drinking 1/week among controls: 11%

Kim and Crittenden, 2005, County Prison, USA, 1992–1998

Cases (441): All inmates screened positive for active TB 1992–1998

Control (478): Sex matched, random sample from prison pop.

Alcohol abuse as recorded in prison health record

Sex, age, ethnicity, marital status, education, homelessness, IV drug use, HIV, length of stay in prison, type of crime.

Adjusted OR for alcohol abuse vs. no alcohol abuse: 1.59 (p < 0.01, no confidence interval reported)

Prevalence of alcohol abuse among controls: 40.2%

Lienhardt et al 2005, Multicenter, Guinée, Guniea Bissau, and The Gambia, 1999–2001

Cases (687): Newly detected smear positive TB

Controls: For each case: A (687): Age-matched household control, and: B (687): Residence area matched community control

Self reported as never/past/current

A large set of host related and environmental factors

Crude OR for current/past vs. never: 1.84 (1.28–2.66)

When controlling for age, sex, family history of TB, HIV and smoking, this association was no longer significant. However, no adjusted OR is reported in paper.

Prevalence of current/past use among controls: 19%

Selassie et al: 2005, South Carolina, USA, 1970–2002

Cases (437): All recurrent pulmonary TB cases, after at least 12 months from time of treatment completion between 1970 and 2001

Controls (442): Random sample of people who remained free of TB > 12 months after completion, matched for year of initial diagnosis

Medical records reviewed. "Alcoholism" as recorded in medical record

Age, sex, race, treatment duration, adherence, regimen, HIV/AIDS, other chronic condition, country of residence, initial sputum, reported side effects.

Adjusted OR for alcoholism vs. no alcoholism: 3.90 (2.49–6.12)

Entire study population are TB infected and previously successfully treated. OR reflect risk of recurrent TB.

Prevalence of recorded alcoholism among controls: 12.4%

Riekstina, et al 2005, Latvia, 1996

Cases (48): New pulmonary cases who had early (within 4 years) recurrence after successful treatment, adults only, excluding those with any resistance to first line drugs, and prisoners

Control (96):successful treatment, no recurrence, matched for sex and bacteriological status

Alcohol problem according to medical records

Sex and bacteriological status matched for. Age, sex, unemployment, treatment facility, treatment interruption

Adjusted OR for alcohol problems vs. no alcohol problem: 16.63 (3.63–76.10)

Entire study population are TB infected. OR reflect risk of progress to active disease.

Prevalence among controls (all TB patients): 23%

Shetty et al, 2006, Medical college hospital, Bangalore, India, 2001–2003

Cases (189): all consecutive new active pulmonary TB

Controls (189): age and sex matched relatives of non-TB patients in same hospital

Self reported as never, past (> 6 months ago), or current use. Amounts not reported.

Age and sex matched. Education, income, crowding, religion, marital status, BMI, cooking fuel, smoking, chronic illness.

Adjusted OR for current vs.- never use 2.37 (0.95–5.93)

Prevalence of current alcohol use in control group: 11.1%

Coker et al, 2006, TB clinic, Samara town, Russia, 2003

Cases (334): Culture confirmed pulmonary TB

Controls (334): Age and sex matched from population registry

Self reported "heavy drinking" at least once per month during last year, but "heavy drinking" not further defined

Age and sex matched. Adjusted for exposure (family contact and drinking unpasteurized milk)

Adjusted OR for heavy drinking at least once a month vs. no drinking: 2.43 (1.22–4.85)

Not clear if also smoking, illicit drug use, imprisonment, and household assets were controlled for. Alcohol not included in final multivariate analysis, reason not reported, alcohol listed as "not appropriate" in table.

Kolappan et al, 2007, Prevalence survey 2001–2003, Rural district, Tamil Nadu, India

Cases (429): Bacteriologically positive cases, aged > = 15, detected during prevalence survey

Controls (93,516): Those not diagnosed with TB in the prevalence survey, aged > = 15

Self reported, alcohol intake in ml. Alcoholism not defined.

Age, sex, smoking

Adjusted OR for alcoholism vs. no alcoholism: 1.5 (1.2–2.0)

Prevalence among controls: 11%