Bulletin of the World Health Organization

Improving tuberculosis case detection rate with a lay informant questionnaire: an experience from the Lao People’s Democratic Republic

Peter Odermatt a b, Sayphone Nanthaphone a, Hubert Barennes a, Khamsay Chanthavysouk a, Duc-Si Tran a, Bounsou Kosanouvong a, Siamphay Keola a, Phetsamone Mathouchanh a, Khamloune Choumlivong a, Valy Keoluangkhot a, Phoumindr Niranh c, Sixomxay Nanthanavone a, Souraxay Phrommala d, Antoine Degrémont a, Michel Strobel a

Background

Tuberculosis accounts for almost 2 million deaths per year worldwide.1,2 The DOTS strategy is implemented to curb the burden of disease. The target for 2005 included 100% DOTS coverage, 70% annual case detection and 85% cure rates. In WHO’s Western Pacific Region, substantial progress has been achieved in the last 10 years.3

The Lao People’s Democratic Republic is a landlocked country in south-east Asia. Its multiethnic and rural (87%) population (5.6 million) is young (50% below 20 years of age) and scattered (population density 24 persons/km²) in mountainous areas (National Census 2005). The Lao People’s Democratic Republic ranks 133rd among 177 countries in the Human Development 2006 index (available at: http://hdr.undp.org/hdr2006/statistics/). The provision of health services is challenging due to the scattered population, lack of resources and difficult access to remote areas.

The national TB control programme started in 1995.4 The DOTS coverage and cure rates in 2000 reached 98% and 80%, respectively.3 The prevalence of TB estimated by WHO was 318 per 100 000 population in 2004.3 The low case detection rate of 55% is a major challenge for TB control, as is low access to health facilities.

The DOTS strategy relies on self-presentation of patients with chronic cough (> 3 weeks) at the health facility and on sputum smear examination for acid-fast bacilli. Actively identifying patients with chronic cough in the community has the potential to improve case-detection rates. Hence, rapid, low-cost and easily applicable approaches are demanded. Chronic cough is an easily recognizable symptom for laypersons.

The objective of our study was to validate a rapid and low-cost questionnaire to identify patients with chronic cough and assess its relevance in TB case detection.

Population and procedures

Study area

We selected six provinces representative of the Lao People’s Democratic Republic’s geographical and sociocultural diversity (Attapeu, Luang Namtha, Luang Prabang, Saravane, Savanakhet and Vientiane). One district was randomly selected in each province.

Rapid identification of patients

Village headmen were targeted as key informants in a three-item questionnaire. First, they were asked to state three main health problems among the villagers, then they were asked whether chronic cough patients (>3 weeks) were present and to identify them by name. It was explained to them in writing that their information would help to clarify the village’s health priorities. We deliberately did not give more information to avoid introducing bias. Furthermore, the village headmen were asked to consult other village key informants, i.e. health volunteers, elders, monks and schoolteachers, during the study. Finally, they were requested to return the questionnaires to the district health offices.

Questionnaire distribution and collection

Questionnaires were sent to all villages in the districts. A member of the National Institute of Public Health in Vientiane dispatched the letters and questionnaires, in collaboration with the provincial health office (Fig. 1), to the district health offices which forwarded them to the villages using normal information channels. No financial incentive was offered at any level. The village informants received no preliminary information about the study.

Fig. 1. Utilization of rapid and low-cost questionnaire targeted at village key informants to detect patients with chronic cough (schematic representation)
Fig. 1. Utilization of rapid and low-cost questionnaire targeted at village key informants to detect patients with chronic cough (schematic representation)

Validation surveys

Based on the number of reported patients, villages were classified in four quartiles. From each quartile, five villages were randomly selected for the validation survey. A house-to-house investigation was conducted in each village by a two-member study team. Any villager reported as having a chronic cough was examined. At the same time, additional chronic cough patients were identified.

Etiology of chronic cough

In each district, the four villages with the highest number of people with confirmed chronic coughs were re-visited. Three sputum samples per patient were collected over two days and microscopically examined on-site for Paragonimus ova (unstained specimens) and Mycobacterium tuberculosis (Ziehl-Neelsen stain). All patients diagnosed with TB were referred to the national TB programme. Paragonimiasis cases were treated with praziquantel (3 x 25mg/kg/day for 3 days).

Data analysis

All data were entered in EpiInfo version 6.04 and cross-checked with original data sheets. Analyses were performed with STATA version 8. The questionnaire return rate (number returned/number distributed), correlation coefficients (number notified/number confirmed chronic cough cases) and the questionnaires’ sensitivity and positive predictive value were calculated for each district.

Ethical clearance was obtained from the Council of Medical Science in Vientiane. Approval was obtained from provincial and district health authorities and oral informed consent was obtained from the villagers involved. The study was carried out from May 2002 to June 2003.

Findings

Out of 456 questionnaires sent to village headmen, 295 were returned (65%). The rates varied widely between districts (range 48–87%).

In the validation surveys, 1071 and 1002 chronic cough patients were found by village headmen and the team, respectively. In identifying chronic cough patients, a large variability was found between districts in the sensitivity (range 56–98%) and positive predictive values (range 34–88%). For example, in Sanamxay district almost all cases of chronic cough (98%) were reported, but over-reporting resulted in low positive predictive value (34%). By contrast, in Namback and Hinheub districts chronic cough was under-reported (60% and 56%, respectively; positive predictive value 85% and 88%, respectively). The correlation between the number of reported and confirmed people with cough was also highly variable between districts, from weak to moderate correlation (0.26< r <0.78).

Between 35 and 119 patients were screened for TB and Paragonimus per district (Table 1). Both infections were diagnosed in all districts except Hinheub (no TB) and Muang Sing (no paragonimiasis). In 374 patients with chronic cough notified in a population of 13 541 persons, 19 new TB cases and 26 new paragonimiasis cases were identified. An overall case detection rate of 5.1% (range 0–26%) and 7.0% (range 0–10.1%) was found for TB and paragonimiasis, respectively. In four out of six districts, the rate of infection paragonimiasis was higher than TB.

Discussion

The study showed that village headmen and their informants participated well in the questionnaire approach. The return rate was higher than expected, given that these groups received no specific preliminary information or incentives to collect the questionnaire at district health offices, to search for symptomatic villagers or to return the filled data sheets to the health offices. In extremely remote districts like Toumlane (Saravane province), the return rate reached 87% and the information provided was relevant; i.e., a strong correlation was found between the numbers of reported and confirmed chronic cough patients. However, the opposite was also found (e.g. Attapeu province) where the return rate was low (58%) with weak correlation, indicating that laypersons may have more difficulty assessing chronic cough than originally thought. In the context of the Lao People’s Democratic Republic, coughing may be very common given poor housing and daily indoor exposure to tobacco and fuel smoke.

Rapid and low-cost questionnaires to screen large areas for urinary schistosomiasis have shown to be a highly cost-effective community diagnosis tool.5 The questionnaire approach is based on an easily recognizable symptom, haematuria (blood in urine), reported by laypersons. The questionnaire is sent to primary schools through administrative channels from provincial to village level and back. Teachers question their pupils to identify patients with haematuria, which is subsequently treated. Areas with higher numbers of patients represent priority areas for interventions.

Chronic cough is a much less specific symptom or marker. Haematuria is highly specific for urinary schistosomiasis, while chronic cough can be found in various diseases involving upper and lower respiratory tract infections.

However, the moderate performance of the quick questionnaire has to be considered in the light of its extremely low cost and high feasibility. Indeed, it allowed the detection of several hitherto undiagnosed smear-positive TB cases. In Savannakhet province nine TB patients were newly diagnosed, yielding an incidence rate of 342/100 000, which is three times the national incidence rate. Furthermore, the questionnaire fosters contact between the community and health services, which in itself is extremely valuable in addressing access-related issues.

The DOTS strategy promotes a passive detection option. Mathematical models suggest that increasing TB detection and cure rates by a more proactive strategy would be the most effective way to reduce TB incidence and burden,6 and would avert millions of cases and deaths worldwide.7 A recent review listed these proactive strategies as including tuberculin surveys, mass radiography, house-to-house surveys and systematic outpatient screening.8 Our rapid questionnaire approach could be added to the list.

We detected paragonimiasis cases in all but one investigated district. This allowed the discovery of an endemic focus of this ill-known and underdiagnosed parasitic lung disease (Hinheub district).9,10,11

To our knowledge, this is the first attempt to examine the usefulness of administering low-cost, rapid questionnaires targeting village lay informants through the health system. In the 1970s in Kenya, direct questioning of community leaders yielded satisfactory results in detecting suspected TB in individuals with chronic cough, which consequently led to smear-positive rates of 1.5% to 2.3%.12

Although this rapid questionnaire approach is operator-dependent for the identification of chronic cough cases, it results in a consistent increase in TB case detection at an extremely low cost. Repeated application with preliminary additional information to village key informants is likely to increase its performance. The communities’ motivation to participate will increase with the subsequent successful management of chronic cough patients and TB treatment. ■


References

Affiliations

  • Institut de la Francophonie pour la médecine tropicale, Vientiane, the Lao People’s Democratic Republic.
  • Department of Public Health and Epidemiology, Swiss Tropical Institute, PO Box 4002, Basel, Switzerland.
  • Faculty of Medical Sciences, National University of Laos, Vientiane, the Lao People’s Democratic Republic.
  • National Institute of Public Health, Ministry of Health, Vientiane, the Lao People’s Democratic Republic.
Share