Menu
Menu
Menu
Menu
Introduction We conducted an implementation science study to increase TB case detection through a combination of interventions at health facility and community levels. We determined the impact of the study in terms of additional cases detected and notification rate and compared the yield of bacteriologically confirmed TB of facility based and community based case finding. Methodology Over a period of 18 months, similar case finding activities were conducted at George health facility in Lusaka Zambia and its catchment community, an informal peri-urban settlement. Activities included awareness and demand creation activities, TB screening with digital chest x-ray or symptom screening, sputum evaluation using geneXpert MTB/RIF, TB diagnosis and linkage to treatment. Results A total of 18,194 individuals were screened of which 9,846 (54.1%) were screened at the facility and 8,348 (45.9%) were screened in the community. The total number of TB cases diagnosed during the intervention period were 1,026, compared to 759 in the pre-intervention period; an additional 267 TB cases were diagnosed. Of the 563 bacteriologically confirmed TB cases diagnosed under the study, 515/563 (91.5%) and 48/563 (8.5%) were identified at the facility and in the community respectively (P<0.0001). The TB notification rate increased from 246 per 100,000 population pre-intervention to 395 per 100,000 population in the last year of the intervention. Conclusions Facility active case finding was more effective in detecting TB cases than community active case finding. Strengthening health systems to appropriately identify and evaluate patients for TB needs to be optimised in high burden settings. At a minimum, provider initiated TB symptom screening with completion of the TB screening and diagnostic cascade should be provided at the health facility in high burden settings. Community screening needs to be systematic and targeted at high risk groups and communities with access barriers.
This study was undertaken between July 2017 and December 2018 in a TB programmatic setting at George primary health care TB diagnostic facility and its catchment population. George community is an informal, poor, high density peri-urban settlement in Lusaka district in Zambia: Fig 1. Lusaka province with a prevalence of 932/100,000 population, has the second highest burden of TB in Zambia after the Copper belt province [22]. The notification rate of TB in Lusaka district in 2016 (pre intervention period) was 640/100,000 (Lusaka District TB data, unpublished), above the country average of 236/100,000 population [23]. In the same year George health facility had a notification rate of 246/100,000 population. George health facility has an outpatient department (OPD), antiretroviral therapy (ART) clinic, Maternal Child Health (MCH) clinic, a voluntary counselling and testing (VCT) point and TB clinic. The catchment community population was 166,975, 173,130 and 179,360 people in 2016, 2017 and 2018 respectively. Before the study, the clinic had onsite LED microscope with no onsite chest x-ray and geneXpert; a mobile digital x-ray and a geneXpert were installed during the study. Similar case finding activities were conducted at the health facility and in the community; they included awareness and demand creation activities, TB screening, diagnosis and linkage to treatment. First, we re-oriented facility health workers and trained community health workers on TB to raise their index of suspicion of the disease. At the health facility, we displayed posters on TB symptoms, community health workers provided daily health talks on TB in all the departments of the clinic and distributed flyers on TB. In the community, we provided door to door sensitization on TB, conducted drama sensitization and displayed posters in places that have/attract large numbers of people and distributed flyers on TB. All these activities had messaging encouraging people to screen for TB. At the health facility, a trained community health worker was stationed at each department to register patient details and refer patients for X-ray screening. In addition, an open access point manned by community health workers was set up to provide fast track TB screening and diagnostic evaluation for clients that were referred by the clinicians and the community health workers and clients presenting directly from the community. In the community, screening and sputum collection points were set up in each mapped zone and identified congregate settings in a rolling fashion with repeated rounds to ensure saturation. History of the four World Health Organisation (WHO) recommended symptoms for TB screening (cough, fever, night sweats and weight loss) [13] and 2 additional symptoms from the Zambia TB guidelines (chest pain and loss of appetite) [24] was documented for all patients presenting for TB screening. One mobile digital chest x-ray (CXR) from Delft Imaging Systems with Computer Aided Diagnosis (CAD4TB) version 5 was used both for community and facility TB screening. Two WHO recommended algorithms [13], both similar to the standard of care algorithms in Zambia except for duration of symptoms when symptom screening is used [24] were used to evaluate for TB: 1) When CXR was available, all patients were screened with CXR-CAD4TB irrespective of symptoms followed by Xpert for those with abnormal CXR; abnormal CXR was defined as CAD score above 60 and 2) When CXR was not available, individuals with any of the above symptoms, irrespective of duration submitted a sputum sample for Xpert. Additionally, clinicians had the discretion to request for GeneXpert for patients who were symptomatic but with a CAD score below 60. Each patient was instructed on how to collect a quality sputum sample by a community health worker. All samples were triple packaged before transportation to the laboratory by community health workers on the same day of collection. Samples were rejected by the laboratory if: i) the specimen was leaking out into biohazard bag, ii) the sputum contained many food particles, iii) the volume was less than <0.5mls and if the sputum contained a lot of blood. HIV status was either self-reported or obtained through opt out HIV testing. All patients diagnosed with TB that did not return to the screening point for results within 2 days had a home visit carried out by a community health worker to facilitate linkage. Contact tracing was done for TB cases identified during the study per routine service requirements. Data was collected from the study TB screening registers and the existing approved National TB laboratory register, TB treatment register and household contact register. The study TB community and facility screening registers were a modification of the nationally approved presumptive TB register whose additional data elements included history of TB treatment, history of contact to a TB case, duration of cough and CAD score. Data from contact tracing was reported under community screening. Data from the facility and community screening registers was entered into a customized web application operating with a Microsoft SQL Server database backend. Transact SQL queries were used to generate weekly/biweekly reports. Error reports were used to flag data inconsistencies that needed corrective actions to be taken ensuring data integrity. Incremental database backups were made on a daily basis. Data comparing community and facility case finding was obtained from the screening registers while data on impact of the interventions in terms of additional cases and notification rate was obtained from the facility TB treatment register. Data was analysed using STATA Statistical Software (Stata Corporation Version 14. College Station, Texas 77845, USA). To show the flow of patients through the diagnostic cascade, 2 flow diagrams were generated for facility and community based case finding and each showed the following steps: individuals screened with presumptive TB, individuals who submitted a sputum sample, individuals with sputum sample evaluated and individuals with sputum evaluated who were diagnosed with bacteriologically confirmed TB(yield). To determine any facility level and community level population characteristic differences among those screened that might account for the differences in case detection, a 2X2 table was constructed and categorical variables were compared using the Chi-squared test and continuous variables using the student t-test. Additional analysis was done to determine the contribution from the community and facility to the total cases detected; contribution from facility was disaggregated further by entry point to determine which entry point had the highest yield. To determine the impact of the case finding on TB notifications, additional cases detected were calculated by comparing TB notifications during the intervention period to a corresponding pre-intervention period. The intervention period included notification data from 3rd quarter 2017 to 4th quarter 2018 and the pre- intervention period included notification data from 3rd quarter 2015 to 4th quarter 2016. Additionality was the difference in TB notification between the intervention period and the pre-intervention period and the percentage change was the additional cases divided by the total notifications in the pre-intervention period multiplied by 100 percent. Lastly, changes in notification rates were also determined taking into consideration the catchment population. Approval to conduct the study was provided by the University of Zambia Biomedical Ethics Research Committee (UNZA BREC) No: 012-05-17 and National Health Research Authority. A waiver of written consent was given by UNZA BREC as the study operations were routine. However, verbal consent was given before participation in the study.
