Study Area

We conducted a TB prevalence survey in the communities of Ravensmead and Uitsig, which have a total surface area of 3.5 km². In 2001, the study area had a population of 36,334 (5). Two primary healthcare clinics and an adjacent tertiary care hospital serve the area. The World Health Organization directly observed treatment short-course strategy was introduced in these 2 communities in 1997.

Study Design

The TB prevalence survey was part of a larger community survey, the Lung Health Survey. In addition to prevalence of TB, the Lung Health Survey aimed to determine the prevalence of lung diseases, including asthma and chronic obstructive pulmonary disease. The current report deals only with TB. Information on other lung diseases will be reported elsewhere.

The study included 2 neighborhoods with similar socioeconomic status. We selected a random sample of all addresses in the study area and defined an address as the residential geographic location (either a physical street address or the name and number of a flat). A randomized 15% sample was taken from all residential addresses. A total of 5,592 addresses were situated in the study area, of which 839 households were selected for participation in the survey. The study protocol was reviewed and approved by the ethics committees of Stellenbosch University and the University of Cape Town.

Survey Procedures

The Lung Health Survey was conducted from July 1 through December 15, 2002. Trained community workers counted the number of persons at each selected address. All persons at selected addresses (including the main house and all backyard shacks) were eligible for investigation. If the head of the household did not give consent for the household to be enrolled in the survey, the household was replaced with a household at an adjacent address, e.g., first to the right and then to the left of the household that declined participation. All persons living at the address were informed of the purpose of the survey, and written informed consent was obtained from each participant before enrollment. The Lung Health Survey included both adults and children, but here we report data in our survey only for adults >15 years of age.

All participants, supervised by a trained field worker, completed a questionnaire containing questions on demographic characteristics and earlier TB treatment. Patient category and treatment outcome of previous disease episode(s) were obtained from the local healthcare center.

Participants were then transported to a nearby facility where chest radiographs (35 × 43 cm postero-anterior view) were performed by using a 200-mA chest radiography machine. Within 1 week, a pulmonologist screened the radiographs for abnormalities. Persons with abnormalities that required immediate investigation or treatment were referred to a local hospital. An accredited reader assessed all radiographs when the survey was finished by using a standardized classification system developed for epidemiologic surveys (6). Abnormalities were classified as being either consistent with TB or not related to TB. Parenchymal, pleural, and central structure abnormalities were considered consistent with TB. A second experienced reader then reread a stratified sample of 31% of the radiographs. Kappa agreement between the 2 readers was 0.69 (95% confidence interval [CI] 0.64–0.74) for abnormalities consistent with TB and 0.47 (95% CI 0.42–0.53) for whether the radiographic result was normal (6). On the basis of these results, a second reader was deemed unnecessary.

Upon request, each participant provided a sputum specimen at the health center where the chest radiograph was obtained by using the active cycle of breathing technique (7). Specimens were collected in a wide-mouthed plastic container with a secure screw-top lid. This container was then transported to the laboratory where it was processed within 3 days. A second specimen was obtained from persons who had a 1 positive smear, a 1 positive culture, or both, before they were referred for treatment to a nearby clinic.

Laboratory Procedures

One smear and 1 culture were prepared from each sputum specimen. Smears were stained by using the Ziehl-Nielsen technique with carbol fuchsin and methylene blue. The smears were scored by 1 reader according to the guidelines of the International Union Against Tuberculosis and Lung Disease (8). A smear result was considered positive if we observed >1 acid-fast bacillus per 100 oil-immersion fields. This provision included scanty smears because these are considered indicative of true positivity (9). Sputum samples were liquefied and decontaminated with 4% NaOH by using standard procedures (10). Samples were then centrifuged at 3,000 × g for 15 minutes. Concentrated decontaminated sputum sediment was resuspended in 2 mL phosphate buffer (68 mmol/L, pH 6.8) and placed on Löwenstein-Jensen slants. These slants were then incubated at 37°C for 6 weeks.

Definitions, Data Processing, and Statistical Analysis

A bacteriologically confirmed case of TB was defined as a person with either 2 positive smears or 2 positive cultures, or 1 positive smear and 1 positive culture. Participants with 1 positive smear or 1 positive culture whose chest radiographs showed TB-related abnormalities, or those with a positive sputum smear or culture results from specimens collected at the health center within 2 months after sputum collection in the prevalence survey, were also considered to have bacteriologically confirmed TB.

Data were entered into a Microsoft (Redmond, WA, USA) Access database. Inconsistencies between the 2 entries were checked against original data and errors were corrected. Statistical analysis was performed by using SPSS version 12.0.1 for Windows (SPSS Inc., Chicago, IL, USA) and STATA version 8.0 for Windows (11). Prevalence of TB was calculated by dividing the number of TB cases by the number of participants who attempted to provide a sputum specimen. Those who were unable to provide a sputum specimen were considered smear-negative and culture-negative. Results include crude prevalence rates and prevalence rates adjusted for the cluster sampling design. Households served as the primary sampling units and variation at this level was taken into account.

We calculated prevalence after applying a correction factor for 100% collection coverage. Results were weighted to adjust for unequal coverage in the different sex and age groups (15–34 and >35 years of age). Separate weighting was applied to participants in the original and replacement samples. We calculated the sampling weight as the product of (1/selection probability of a household) and (1/participation rate). The selection probability of a household was 0.15 because we randomly sampled 15% of the households. The participation rate varied by group sampled; e.g., 68% of women 15–34 years of age in the original sample had a chest radiograph performed. Sampling weight was (1/0.15) × (1/0.68) = 9.8. CIs were calculated by using the normal approximation to the binomial distribution.

Study Population and Coverage of Measurements

Figure

Figure. Sample selection of 3,483 adults, Cape Town, South Africa, for the study.

Initially, heads of 625 (74%) of 839 households consented to participate in the study. We replaced 212 of 214 households that refused participation; 2 households could not be replaced. Of 214 nonparticipating households, 81 provided demographic details of persons living in the household. Occupants of these 81 households did not differ from the 212 households that were resampled with respect to sex (odds ratio [OR] 1.00, 95% CI 0.82–1.23) or age (t = 0.33, p = 0.74, by Student t test). In the final sample, residents within 837 households included 3,971 adults, of whom 3,483 (88%) consented and completed a questionnaire (Figure). Of these, 2,608 (75%) had a chest radiograph and attempted to provide a sputum specimen. Persons >35 years of age (78%) had a chest radiograph more often than persons <35 years of age (72%) (OR 1.41, 95% CI 1.21–1.65). More women (78%) than men (71%) had a chest radiograph (OR 1.48, 95% CI 1.26–1.72). Of those who had a chest radiograph, 1,170 (45%) were able to provide a sputum specimen. Thirteen specimens were contaminated or inadequate for culture because of insufficient volume.

Prevalence of TB

Of 2,608 participants with chest radiographs, 702 (26.9%) had an abnormality on the radiograph (95% CI 25.2%–28.6%). In 337 (12.9%) of these radiographs, abnormalities were consistent with TB (95% CI 11.6%–14.2%) (Table 1). Prevalence of TB-related abnormalities on chest radiographs increased with age (χ² for linear trend 59.1, p<0.001) and were more often seen in men (16%) than in women (11%) (OR 1.48, 95% CI 1.16–1.89).

Twenty-nine participants had a positive smear or positive culture, of whom 3 did not fulfill the definition of a bacteriologically confirmed TB case; 2 had a scanty smear and 1 had a positive culture. The remaining 26 participants fulfilled the definition of bacteriologically confirmed TB, giving a prevalence of 10.0/1,000 (95% CI 6.2–13.8/1,000) (Table 1). Of these 26 patients with bacteriologically confirmed TB, 16 (62%) were new patients (no previous treatment), giving a prevalence of new bacteriologically confirmed TB cases of 6.1/1,000 (95% CI 3.1–9.1/1,000). Ten (56%) of the 18 smear-positive patients had been previously treated for TB. The prevalence of previously treated smear-positive TB cases was 3.8/1,000 (95% CI 1.5–6.2/1,000). Only 8 (44%) smear-positive cases were new cases, and the prevalence of new smear-positive TB cases was 3.1/1,000 (95% CI 0.9–5.1/1,000).

Correction for the cluster sampling design and adjustment for noncoverage had little effect on the prevalence estimates (Table 1). Three of the 26 patients with bacteriologically confirmed TB were receiving anti-TB therapy at the time of the survey. Nineteen of the remaining 23 patients with detected cases began treatment <6 months after these cases were detected in the survey. Of those patients, 12 were cured, 1 completed treatment without confirmed smear conversion, 2 interrupted treatment, 1 died, and 3 had an unknown outcome.

History of TB

All 10 patients with bacteriologically confirmed TB who had previously received treatment for TB were smear-positive. Four of the 10 previously treated patients had been considered cured during their previous disease episodes, and 3 had interrupted treatment before completion. For 3 previously treated smear-positive patients, no information was available on the outcome of their previous disease episode (Table 2). Four patients had been treated for TB on more than 1 occasion before their cases were detected in the survey. Also, 3 previously treated patients lived in the same household. Bacteria isolated from 1 of 7 previously treated TB patients who underwent drug sensitivity testing were resistant to isoniazid (Table 2).

A history of TB was reported by 338 (9.7%) of 3,483 adults (95% CI 8.7%–10.7%) (Table 1). The prevalence of bacteriologically confirmed TB was 29.9/1,000 (95% CI 11.4–47.7/1,000) in those who were previously treated for TB compared with 5.1/1,000 (95% CI 2.6–7.6/1,000) for persons who had never had TB (OR 5.96, 95% CI 2.68–13.25). Participants reporting a history of TB were significantly older (mean 41 years) than participants who had never had TB (mean 38 years) (t = 2.59, p = 0.01, by Student t test). A history of TB was more common in men (12%) than in women (8%) (OR 1.60, 95% CI 1.27–2.01).