This is a cluster randomized controlled trial with two arms comparing standard passive case detection with household contact investigation in eight provinces in Vietnam. In the active intervention clusters, household contacts of patients with TB will be assessed for the disease four times over a two-year period. The assessment comprises clinical evaluation, chest radiography and, in TB suspects, microbiological testing of sputum.
Inclusion and exclusion criteria
All patients aged 15 years or over with pulmonary TB and at least one sputum sample positive for acid-fast bacilli on direct smear microscopy will be eligible to enter the study. Patients are excluded if they have no eligible household contacts.
Household contacts of any age are eligible for inclusion if they lived in the same dwelling as the index patient during the two months prior to the diagnosis of the index patient. Pregnant women are eligible for inclusion, but women known to be pregnant are not offered a chest radiograph.
Study endpoints
The following endpoints are being measured in all contacts (in both groups). The primary endpoint is the proportion of contacts with clinically diagnosed TB. This will be ascertained based on the results of screening (in the active screening districts) and routine notifications to the NTP (in the active screening and control districts). Secondary endpoints, as indirect measures of study feasibility, are: (a) the weighted proportion of contacts with microbiologically proven TB (defined as either smear or culture positive TB), smear-positive TB and extrapulmonary TB; (b) cost effectiveness, determined by calculating the direct and indirect costs for each additional case diagnosed; and (c) the proportion of eligible contacts enrolling in the study, and the proportion of contacts participating in scheduled follow-up.
Description of intervention
Recruitment of index cases
Index cases with smear-positive pulmonary TB and their contacts are identified in both active and control study districts. The study intervention is only implemented for contacts in the active intervention districts.
Consecutive patients with smear-positive pulmonary TB, who live in the study districts, are identified by TB staff working within district hospitals or district tuberculosis units. Those who agree to participate are given a participant information statement and asked to sign a written consent form.
Recruitment of contacts
Contacts are identified using a screening questionnaire completed in consultation with the index case. Household contacts include all members of the household occupied by the index case, including children and infants, who were present during the period of recent infectivity, that is, two months up to the time of diagnosis. Household contacts are approached by the index patient in their family and given information explaining the study. Contacts are invited by the patient to enroll in the study at a district clinic.
Active intervention
All contacts in the active intervention districts will be evaluated at the time they are first identified, and then 6, 12, and 24 months post-exposure. The procedures will be free of charge to the participants. On each occasion, the assessment will include a symptom questionnaire, physical examination, and chest X-ray. Contacts will have a chest X-ray at a local clinic or nearby private radiology service. X-ray films will be read by local clinical staff for routine clinical action and then be transferred to a central location for standardized re-reading by provincial level staff.
Contacts will be defined as having possible TB if they have: (a) cough lasting for two weeks, (b) sputum for two weeks, (c) any hemoptysis, or (d) an abnormal chest X-ray meeting the criteria for possible TB. Contacts meeting one or more of these criteria will be asked to submit two spontaneous sputum specimens collected on consecutive days for acid-fast bacilli (AFB) examination. Two direct smear examinations and one culture test will be undertaken in the Provincial Hospital laboratory. For children younger than five years, the diagnosis of TB will be made on clinical and radiological grounds, in accordance with local policy at the time of commencement of the study. Treatment of active disease will be according to the policies of the NTP. An experienced clinician in each province will be available to advise district staff regarding the diagnosis of childhood TB and the investigation of suspected extrapulmonary disease.
Control districts
Contacts enrolled in control districts, will be given general information about the typical symptoms of TB and advised to report to the district hospital or TB unit if they develop these symptoms. During the study period, self-referring contacts with TB, who are participating in the study, will have two sputum specimens transferred to the reference laboratory for culture, in addition to any routine diagnostic procedures. This will allow comparison of rates of culture-confirmed TB between active and control districts. Contacts diagnosed with TB in both the control and active intervention groups will receive standard treatment for TB through the NTP.
Selection of study districts
The study will be conducted in 70 districts within eight provinces. These provinces have been selected to include four provinces in the south (Ho Chi Minh City, Can Tho City, and the rural provinces of An Giang and Tien Giang), where the prevalence of TB is highest, two provinces from the central area (Da Nang Province and the rural province of Binh Dinh), and two provinces from the north (Hanoi City and the semi-rural province of Vinh Phuc). This geographic and urban-rural stratification is important because the incidence of TB increases from the north to the south within Vietnam [7]. There are also substantial differences in lifestyle and infrastructure between rural and urban areas. Hence, the selection of this diverse range of sites will increase the generalizability of the study findings.
Randomization
Among the 97 districts within these eight provinces, 70 have been randomly selected for participation in the trial. These districts were selected with a probability proportional to their population size. The initially selected districts will enter a run-in period, prior to randomization, during which time they will be required to recruit two patients and six contacts and to demonstrate that they have the necessary resources to conduct the study. If districts fail or withdraw during this pre-randomization run-in period, they will be replaced by the next randomly selected district.
We will randomly allocate districts to the active intervention or control group using the minimization method with predefined randomization lists [27]. The minimization criterion will be province, to ensure balance of intervention and control districts within each province. District size will not be taken into account in the randomization process. Randomization will be performed by a person who is not otherwise engaged in the project, and is blinded to the identity of districts.
Sample size calculation
Based on an earlier meta-analysis, we estimated that the prevalence of confirmed active TB in close contacts of smear-positive cases would be 2.3% [28]. In the 2006 to 2007 National Prevalence Survey, approximately 50% of prevalent cases of TB did not report typical symptoms of the disease, suggesting that a substantial proportion of patients with prevalent disease are not diagnosed by routine passive case detection within the NTP. We estimate that screening with the combination of X-ray and symptoms will identify at least 75% of prevalent cases of active pulmonary TB. This represents an increase in case detection rate of 50% above that expected for passive case finding alone. To have 80% power to detect an effect of this size, we would need 3,313 contacts in each group (6,626 in total), without accounting for clustering [29].
The sample size was then adjusted for the clustered design of the study. The first level of clustering is at the level of the index case. We estimated that, on average, each index case would identify four close contacts who would consent to participate. Assuming that the intraclass correlation coefficient within households is 0.1, the design effect at this level will be 1.3. The second level of clustering is at the level of the districts. There were 152 cases of smear-positive pulmonary TB per district in 11 representative provinces in 2007. Assuming that 70% of smear-positive cases consent to participate in the study, there will be 106 index cases. If we assume an intraclass correlation coefficient for index cases within districts of 0.01, the design effect at this level will be 2.05. Hence, we anticipate an overall design effect of 2.67.
On this basis, we estimate that we need to recruit a total of 3,338 index cases and 8,829 contacts (of 2,207 index cases) in each group. Based on our assumptions, this would require us to approach all available patients diagnosed with smear-positive pulmonary TB in 21 districts in each group: 42 districts in total. To ensure that we have a conservative sample size calculation, we will recruit all eligible and consenting index cases and their contacts from 70 districts.
Specific quarterly recruitment targets will be set for each district, based upon the number of notified cases of smear-positive TB during the most recent year for which data were available, as a way of enabling districts to track their progress during the study.
Analysis of effectiveness
The effectiveness of the active case finding in contacts will be assessed by comparing the case detection rate between contacts of index cases in the active intervention and control districts. There are two levels of clustering in the selection of contacts: the district in which the contact lives and the index case with whom he or she was in contact. To account for these two levels of clustering, we will undertake a mixed model analysis in which districts and index cases within districts are random effects and active versus control status is the fixed effect. The measure of effect will be relative risk, estimated using a log link and binomial error distribution. In this primary analysis, there will be no adjustment for baseline characteristics. However, in a subsidiary analysis we will examine the impact of baseline characteristics of the districts, including geographical characteristics and baseline case notification rate, and individuals, including age and sex, on the risk of disease by testing these characteristics as potential effect modifiers.
Primary and secondary endpoints and stratification
The primary endpoint will be the weighted proportion of contacts with clinically diagnosed TB in screening and control districts during the two-year follow-up period of the study. Secondary endpoints will include the weighted proportion of contacts with microbiologically proven TB, smear-positive TB, and extrapulmonary TB. We will also estimate the proportion of contacts with TB who complete treatment. Stratum specific weighted prevalence rates of clinically diagnosed and microbiologically proven TB will be determined for region and rurality.
Analysis of cost effectiveness
Direct costs of diagnosis and treatment will be determined based upon questionnaires for staff and patients at provincial and district level healthcare facilities. Indirect costs will be determined based upon detailed questionnaires for patients, contacts, and staff. Costs will be classified into one of two categories: those related to the identification of new TB suspects among the contacts of TB cases and those related to investigation and management of TB suspects. The former are costs that are unique to this intervention. The latter costs are already met by the NTP but these will be increased by the intervention because there will be more TB suspects.
The incremental cost-effectiveness ratio will be calculated as the difference in total cost for TB control measures between the active and control districts divided by the difference in the number of cases detected.
Data collection and management
Data will be recorded on paper forms at district clinics and provincial hospitals. Each district will submit a monthly report on the recruitment index cases and contacts and the occurrence of study endpoints. These data will be entered within one month of collection into an online database. Written data and chest radiographs from each district will be digitally photographed, and the images transferred to the National Lung Hospital for electronic data entry and quality control. Data will be entered using a custom-designed, internet-based database, and monthly data checking and quality control will be performed at a central level. District staff will be provided regularly with a list of additional information required. A record of the proportion of children diagnosed with TB by chest X-ray will be compiled every three months, to ensure that the proportion of children receiving tuberculosis treatment does not exceed that recognized in the published literature.
Quality control
The technical quality of chest radiographs will be assessed at the National Lung Hospital by examining all radiographs labeled as abnormal and a random sample of all other radiographs. All radiographs read at the district level will be re-read at the provincial level, and the district staff will be informed of discrepancies. Quality control of laboratory testing will be performed according to the routine protocols of the NTP, including structured quarterly monitoring visits by staff from the National Lung Hospital.
Evaluation and monitoring
Members of the project research staff will conduct monitoring visits to each province every quarter, and to each district at least once each year. Provincial TB control staff will also conduct quarterly monitoring visits to each district in conjunction with routine monitoring for the NTP. Monitoring visits will include checking participant consent forms, confirmation that enrolled index patients have been included in the district notification registry, checking research forms for completeness, and compliance with study protocols.
To verify subject consent, randomly selected contacts from each district will be telephoned by study staff to confirm they have consented to participate in the research. Irregularities in recruitment will be followed up by provincial staff and reported back to research staff.
Ethical issues
Ethical approval
The protocol, patient information sheet, and patient consent form have been approved by the Human Research Ethics Committee at the University of Sydney, the Scientific Committee of the Vietnam Ministry of Health and the Scientific Committee of the Vietnam National Lung Hospital.
Informed consent
Subjects may only enter the study if written informed consent is obtained. Written consent by parents will be required for children under 15 years of age. The healthcare worker enrolling the subject will be responsible for obtaining written informed consent and consent forms will be checked during regular monitoring visits by research staff.
Withdrawal from the study
Research participants may voluntarily withdraw from the study for any reason. If they withdraw then there is to be no adverse impact upon the treatment of the TB patient within the NTP.
Confidentiality
A unique study code will be assigned to each research subject, and used to identify radiographs and laboratory specimens. Written records will be stored securely in the district clinics and provincial hospitals or at the National Lung Hospital. The online database and electronic records will be password protected.