We performed a retrospective cohort study of TB patients in Attapeu Province South of Laos from February to June 2005.
Study area
The study was performed in the rural province of Attapeu, located on the southern border with Vietnam and Cambodia. Attapeu province has 108,000 inhabitants (10 inhabitants per km2) scattered in 5 districts and 280 villages. The population is predominantly poor, illiterate and belongs to ethnic minority groups. Mean monthly household income is estimated to be 29 USD [1], and access to health care is generally limited. In the Attapeu provincial hospital, TB is the third most frequent cause for outpatient attendance (247 of 8281 in 2004) and the fifth most frequent cause of hospitalization (54 of 1632 in 2004). HIV prevalence in incident TB cases is 2% in Laos [3].
Hospital records, patients definition and follow-up
We reviewed the records of all patients hospitalized from January 2002 to December 2004 and identified all TB patients. Tuberculosis was defined as either positive AFB (3 sputum sample examination at Attapeu Hospital and grading according to NTP criteria) or sputum negative with a suggestive chest x-ray result and a positive result of the clinical algorithm provided by the NTP [7].
From the records, we collected information on hospitalizations including symptoms, weight, treatments, length of stay, bacteriology, smear positivity and grading according to the NTP [7].
All patients were traced in their residential villages. On site, we interviewed the patient with a questionnaire which was tested on comprehensiveness and subsequently improved questions. We obtained the following information: current symptoms, chemotherapy, quality of the relationship with the health facility's medical staff, attitudes regarding treatment and compliance, and knowledge about TB. Subsequently, we performed a general physical examination.
Patients were weighed lightly clothed using an electronic SECA scale for adults (precision ± 100 g). Patients' height was measured with a 0.1 cm precision using wooden measuring boards and a measuring tape. The patient's body mass index was calculated (BMI: weight/height2). Adults were considered underweight if BMI was ≤ 18.5 and severely underweight if the BMI was < 16 [8]. We recorded body weight before treatment, during treatment and during home visits for patients with initially available data.
Definition
We used definitions of the NTP for compliance and treatment achievement (completeness). A patient was defined as compliant if, at the time of the survey, the patient had taken at least 90% of the prescribed number of tablets days [7]. On the other hand, non-compliant patients (as determined by the pill counts of the NTP, cross-checked with the district hospital data) were considered defaulters. A TB treatment was regarded as completed if the patient finished at least 95% of his treatment days according to his treatment category [7].
Clinical outcomes
Survival rate was the primary outcome. Compliance with treatment, and clinical outcome including nutritional status and symptoms were the secondary endpoints.
Laboratory and field procedures
All patients were initially checked for AFB in their sputum by the hospital. Assessing the sputum during follow-up was done by referring the patients to the provincial hospital for the scheduled 2-5 monthly sputum checks by the NTP. NTP, with the help of Service Fraternel d'Entraide, a non-governmental organisation, was responsible for the training and regular quality assessments of the laboratory.
A parallel survey was performed at the same time to detect active TB cases and to conduct contact tracing in the patients' villages. Of the 84 villages with TB patients, 10 villages were randomly selected, and all inhabitants with a chronic cough of more than 3 weeks (and/or haemoptysis) were screened using a questionnaire approach [9]. We collected two sputum samples from patients with chronic cough, one immediately and one the following morning, according to standard procedures of the NTP [7]. Samples were collected each day and sent to the district hospital where a Ziehl-Neelsen stained slide was examined for the presence of acid-fast bacilli. In addition, a sputum smear was directly examined for the presence of Paragonimus eggs using a light microscope [10]. Positive patients were referred to the provincial hospital for treatment. We felt that the detection of paragonimiasis (often clinically misdiagnosed as TB) was justified as there has been some evidence of it in Laos [10].
Data management and analysis
Data was entered in Epidata freeware (http://www.epidata.dk, Odense, Denmark) and cross-checked against original data sheets. Data analysis was carried out with Stata, version 8 (Stata Corporation, College Station, TX, USA). We used Fisher's exact test to assess associations between categorical variables, student's t-test for two normally-distributed continuous variables, log rank test to compare the durations of treatment and hospitalisation between compliant and non compliant, and between survivors and non survivors and t-test for two-sample mean-comparison of paired data for the patients weight evolution. We assessed the compliance factors. First, we analyzed the factors affecting the compliance in an univariate analysis. Second, all factors with p values ≤ 0.2 were then fitted into a multivariable logistic regression model. We considered p < 0.05 as significant.
All participants gave informed witnessed verbal consent. The study was approved by the Ministry of Health's Ethical Council of Medical Sciences for Health Research and the provincial health authorities. The study was performed in accordance with the Declaration of Helsinki.