Implementation of systematic screening for tuberculosis disease and tuberculosis preventive treatment among people living with HIV attending antiretroviral treatment clinics in Ghana: a national pilot study

Strengths and limitations of this study

  • Key strengths include the diverse health facilities included, relatively large sample size, comprehensive assessment of participants and detailed cost assessment which included downstream costs associated with treatment.

  • Only 20% of people living with HIV attending participating health facilities were included in this study, which could result in selection bias and affect generalisability.

  • The majority of tuberculosis (TB) disease diagnoses were clinical rather than microbiological, so associations with diagnosis may be confounded by clinical judgement.

  • Cost estimates were derived assuming all participants completed TB disease treatment and TB preventive treatment, which may result in an overestimation of costs.

Introduction

Tuberculosis (TB) and HIV remain significant public health threats globally.1 Among people living with HIV (PLHIV), TB is the most frequent life-threatening opportunistic disease and the leading cause of death, even among those receiving antiretroviral treatment.2 Therefore, a priority is integrating TB care into HIV programmes to reduce morbidity and mortality stemming from TB.

PLHIV are at substantially higher risk of developing and subsequently dying of TB disease than people without HIV.3–7 Therefore, providing TB preventive treatment (TPT) and prompt identification and initiation of treatment for TB disease is an urgent priority for both HIV/AIDS and TB programmes. To achieve this, the WHO recommends a multifaceted approach to TB prevention and care among PLHIV, which includes effective infection control measures, enhanced and intensified case finding activities, and provision of TPT for PLHIV without TB disease.8–10

In Ghana, routine programmatic data suggest only 72% of PLHIV were screened for TB disease in 2017; TPT initiation was unknown.11 To address this gap, Ghana revised its TB screening questionnaire and screening algorithm for PLHIV engaged in care to include Xpert MTB/RIF and digital X-ray. These updates resulted in additional diagnostic tools—48 digital X-ray machines and 127 Xpert MTB/RIF machines—being deployed to high-burden health facilities to support enhanced case finding in key populations.12 13

Following the updates to screening guidelines and deployment of enhanced diagnostic tools, we conducted a national pilot study to increase the systematic evaluation of PLHIV for TB disease and initiation of eligible PLHIV on TPT, and assessed the yield and costs of the programme.

Methods

This study was reported according to Strengthening the Reporting of Observational Studies in Epidemiology guidelines. Patients and the public were not involved in the design of this study.

Study setting

The pilot study aimed at assessing the yield and cost of systematic TB screening and prevention in PLHIV. It was conducted from August 2019 to December 2020. During the study period, Ghana was considered one of the thirty high TB/HIV burden countries in the world. The TB incidence in 2019 and 2020 was 144 and 143 per 100 000 population, respectively; among those with TB, HIV coinfection occurs in approximately 15%.14 The uptake of TPT among PLHIV is unknown.14 At the onset of the study, Ghana had ten administrative regions (online supplemental appendix figure 1), and one high-volume facility in each region—which provided antiretroviral treatment (ART) services and had both a digital X-ray and Xpert MTB/RIF machine—was purposively selected.

Supplemental material

Study population, inclusion and exclusion criteria

This was a prospective pilot study conducted in 10 participating health facilities. The eligible population included all PLHIV attending one of the participating health facilities who were receiving ART, irrespective of their immune status (eg, CD4 count or viral load). PLHIV who were not screened for TB, who lived a distance from the health facility that did not permit routine monthly or quarterly clinic visits, or did not consent to take part in the study were excluded.

Procedures

Prior to the start of the study, training on the TB symptom tool (see online supplemental appendix 1; the tool was developed by the National TB Programme and used for participants of all ages), the algorithms for screening for TB disease among PLHIV, and the use of digital data collection tools, was given to all health workers (eg, pharmacists, medical officers, physicians, nurses) at participating health facilities. Staff were also trained on the administration of the TPT regimen (6 months of daily isoniazid)—specifically, the eligibility criteria (no evidence of TB disease, no previous TPT, did not use alcohol, did not have elevated liver transaminases, no contraindications to isoniazid, no known contact with drug-resistant TB patient), common side effects and the appropriate dosage for different age groups.

After a patient consented to the study, they were verbally screened by a health worker using the TB symptom tool for any of the four signs and symptoms of TB—unintended weight loss, fever, night sweats and cough. Regardless of the result of clinical screening, all patients were then given a chest X-ray, administered by a trained and qualified radiologist. Chest X-rays were evaluated using a computer-aided detection software (CAD4TB, Delft Imaging, Netherlands) as well as interpreted by trained clinicians at each facility for abnormalities suggestive of TB; if there was uncertainty, X-rays were shared with a central panel of three radiologists to arrive at a final interpretation. All participants with chest X-ray abnormalities suggestive of TB or with any symptom of TB were asked to provide a spot sputum sample. Sputum samples were assessed using Xpert MTB/RIF by trained scientists and laboratory technicians. For all participants who were negative on Xpert MTB/RIF, but had chest X-ray abnormalities suggestive of TB, the attending clinician discussed the X-rays and patient profile (eg, clinical history and other findings) with the central panel of radiologists to arrive at a final TB diagnosis. For all patients diagnosed with TB disease, appropriate TB disease treatment was offered. For patients without evidence of TB disease and who were eligible for TPT, TPT was offered.

Data collection and outcomes

At each patient visit, data was collected on paper charts and then transposed to a digital data collection tool (KoboToolbox). ART staff in participating facilities were trained in data collection prior to the study and collected all data during study execution. The data elements collected included sociodemographic and educational data (eg, age, sex, body mass index), clinical data (eg, ART use), TB symptoms and chest X-ray results, Xpert MTB/RIF results (if applicable), final TB diagnosis and initiation of TPT (if applicable). The data collection form is included in online supplemental appendix 2.

We collected information on costs associated with systematic TB disease screening and provision of TPT in 2020 US$ from the health system perspective (online supplemental appendix 3). Where possible, costs were locally collected. We collected costs associated with travel and salaries for training; we distributed time estimation questionnaires to each facility to determine the amount of health worker time associated with symptom screening, X-rays, sputum collection, Xpert MTB/RIF and patient management during treatment, as well as median salaries for health workers responsible for these tasks; we collected relevant information from each participating health facility on Xpert MTB/RIF maintenance and usage; and we collected costs associated with medications and diagnostics from the Global Drug Facility.

The primary study outcome was the proportion of PLHIV with microbiologically or clinically diagnosed TB disease. The secondary outcomes were (i) the proportion of eligible PLHIV initiating TPT and (ii) the cost per TB disease diagnosed and per person initiating TPT.

Data analysis

Sample size

This study did not have a formal sample size calculation. We initially aimed to enrol 500 participants at each participating health facility over the original study period (12 months). However, during the study, the onset of the COVID-19 pandemic impacted recruitment as the number of outpatient attendees including at ART clinics declined. Though the study was extended an additional 5 months (17 months total), these disruptions continued and recruitment was terminated at the end of 2020.

Statistical analysis

All data was analysed in R. We estimated the proportion of TB among PLHIV and exact 95% CIs overall, by microbiological confirmation, by the presence of TB symptoms, by abnormal X-ray and by ART use. We performed logistic regression using generalised linear mixed models—with each health facility as the clustering variable—to estimate the adjusted OR and 95% CI for different risk factors potentially associated with TB disease. We did univariable analysis on the following factors: age (continuous), sex (male, female), employment status (employed, unemployed/retired), education (none, elementary school, high school, university), previous TB screening (yes, no, unknown), newly initiating ART (yes, no, unknown) and previous TB diagnosis (yes, no, unknown). We retained age, sex and all factors that were p<0.2 in univariable analysis for multivariable models. We did secondary analysis following these same methods on the subpopulation with symptoms and/or abnormal chest X-ray to evaluate risk factors associated with TB disease.

Among those eligible for TPT, we estimated the proportion and exact 95% CI of PLHIV who initiated TPT.

Costing analysis

We employed a microcosting approach to estimate the costs associated with systematic TB disease screening and the provision of TPT for those eligible. We converted costs to 2020 US$ using purchasing power parity or direct exchange rates, as appropriate, and when costs were not available in 2020, we used consumer price indices to inflate costs to 2020. We considered the following costs: the costs associated with training health workers at each health facility; the costs associated with symptom screening PLHIV for TB disease; the costs associated with digital chest X-ray; the costs associated with Xpert MTB/RIF analysis; and the costs associated with treatment TB disease and providing TPT. As the study time horizon was short, we did not include discounting.

For training, we included attendee salaries for the 1 day of training and the costs of travel for staff administering the training. For TB symptom screening, we used the time estimation questionnaire results from each facility to estimate the personnel costs associated with administering the TB symptom screening questionnaire and referring participants for a digital chest X-ray. For the digital chest X-ray, we used reported costs from each health facility to calculate the cost of performing one X-ray. For participants requiring Xpert MTB/RIF testing, we used multiple data sources. First, we used the time estimation questionnaire results to estimate the personnel costs associated with collecting sputum samples. Second, from each site we collected costs of sputum transport to the laboratory from the health facility and based on annual usage metrics of Xpert MTB/RIF at each laboratory, estimated costs of transport on a per sample basis. Third, from annual usage metrics at each site and annuitising capital costs of the Xpert MTB/RIF machine at 3% over a 5-year useful life (all laboratories used a four-cartridge machine, Xpert MTB/RIF G4), we calculated costs associated with the machine on a per sample basis. Fourth, we used costs from the Global Drug Facility for one Xpert MTB/RIF cartridge. Fifth, we used previously collected data from Benin to estimate the laboratory personnel costs, laboratory overhead and annual maintenance costs for Xpert MTB/RIF on a per sample basis. For the costs associated with TB disease and TPT, we used the time estimation questionnaire results from each facility to estimate the personnel costs associated with initial and follow-up visits during treatment and the costs of 6 months of daily isoniazid and rifampin coupled with pyrazinamide and ethambutol during the first 2 months (for TB disease) and 6 months of daily isoniazid (for TPT) from the Global Drug Facility, assuming all participants completed treatment.

Patient and public involvement

In this study, patients and the public were not involved in the study design or its dissemination.

Results

Description of participants

During the study period, 12 916 people attended participating ART clinics and 2639 (20%) agreed to participate in the study (figure 1). Characteristics of the included participants are in table 1. Of those consented to participate, 1824 (69.1%) were already receiving ART and 815 (30.9%) were newly initiating ART. Overall, 2129 (80.7%) participants were female and 121 (4.6%) were children under 15 years of age; the majority of participants were employed (82.7%). Approximately one-third of participants (805; 30.5%) reported they had been previously screened for TB and 97 (3.7%) reported they had been previously diagnosed with TB disease. Most participants were recruited from the Eastern Region (673; 25.5%), while the fewest participants were recruited from the Upper West Region (103; 3.9%).

Figure 1
Figure 1

Flow diagram of included participants. ART, antiretroviral treatment; TB, tuberculosis; TPT, tuberculosis preventive treatment.

Table 1

Characteristics of included participants

TB screening

Among the 2639 participants, 272 (10.3%) reported at least one TB symptom. The most common symptom was weight loss (136; 5.2%), followed by cough of at least 2 weeks (113; 4.3%). All participants underwent chest X-ray examination, of which 90 (3.4%) had an abnormal chest X-ray result (69 (2.6%) had an abnormal X-ray without symptoms). Overall, 341 (12.9%; 95% CI 11.7% to 14.3%) participants had at least one TB symptom or an abnormal chest X-ray and were further evaluated for TB disease using Xpert MTB/RIF.

Prevalence of TB among PLHIV and TPT initiation

Among the study participants, 50 (1.9%; 95% CI 1.4% to 2.5%) were diagnosed with TB disease, of which 36/50 (72%) were diagnosed clinically and 14/50 (28%) were bacteriologically confirmed (table 2). Among the 341 participants with at least one TB symptom or an abnormal chest X-ray, the prevalence was 14.7% (95% CI 11.1% to 18.9%). When stratifying on indication for microbiological testing for TB disease, prevalence was lowest among those with at least one TB symptom and a normal chest X-ray (10.0%; 95% CI 6.5% to 14.4%) and highest among those with at least one TB symptom and an abnormal chest X-ray (71.4%; 95% CI 47.8% to 88.7%). Among the 69 asymptomatic participants with an abnormal chest X-ray, 10 (14.5%; 95% CI 7.2% to 25.0%) had TB disease; none were microbiologically confirmed. When evaluated by ART use, the prevalence of TB was 1.6% (30/1824) among those already on ART and 2.5% (20/815) among those newly initiating ART. Among the 2589 participants without TB disease and thus were eligible for TPT, 2581/2589 (99.7%) were initiated on TPT.

Table 2

TB prevalence by diagnosis method, presence of symptoms or abnormal chest X-ray, and ART use

Factors associated with TB among PLHIV

Demographic variables of age, sex, employment and education, as well as clinical characteristics of previous TB disease screening and previous TB disease diagnosis, were not associated with prevalent TB disease among participants (table 3). In multivariable analysis, only ART initiation status was associated with TB disease diagnosis; participants newly initiating ART had 4.1 times (95% CI 2.0 to 8.2) higher odds of being diagnosed with TB disease than participants who were already receiving ART. Trends were similar, but non-significant when only evaluating those with microbiologically confirmed TB disease (online supplemental appendix table 1).

Table 3

Analysis of factors associated with tuberculosis diagnosis among the included population

When limiting the population only to those with at least one TB symptom or an abnormal chest X-ray (n=341), findings were similar (online supplemental appendix table 2). Most demographic and clinical characteristics were not associated with a diagnosis of TB disease. However, ART initiation status and symptom and radiological findings were associated with TB disease. Compared with participants already receiving ART, participants newly initiating ART had 3.2 times (95% CI 1.3 to 8.0) higher odds of being diagnosed with TB disease. Compared with participants with TB symptoms but a normal X-ray, participants with an abnormal X-ray but no TB symptoms were not at significantly higher odds of TB disease diagnosis (adjusted OR (aOR) 2.3; 95% CI 0.9 to 6.3), however, participants with both TB symptoms and an abnormal X-ray were at significantly higher odds of TB disease (aOR 21.6; 95% CI 6.3 to 74.3).

Cost of systematic TB disease screening and TPT

Across the 10 participating health facilities, we estimated the cost (in 2020 US$) of training staff at each facility to be $2687, of symptom screening and chest X-ray per participant to be $21.49, of Xpert MTB/RIF per sample was $43.90, of providing treatment for TB disease per participant to be $107.63 and of providing TPT per person to be $20.35 (table 4). Overall, the total cost of of systematic TB disease screening and TPT in this study was $151 277, equivalent to a cost per person screened of $57.32 and cost per TB disease diagnosed of $3026. If the intervention was only targeted to PLHIV newly initiating ART, the cost per TB disease diagnosed would fall to $2365. The proportional share of costs associated with staff training was 17.8%, associated with screening and TB disease was 47.5%, and associated with provision of TPT was 34.7%.

Table 4

Costs associated with TB screening and TB preventive treatment among people living with HIV

Discussion

In this pilot study assessing the yield and cost of systematic TB screening and prevention at 10 health facilities providing care to PLHIV in Ghana, we found 1.9% (95% CI 1.4% to 2.5%) of participating PLHIV had TB disease and nearly all eligible participants initiated TPT. Odds of TB disease were more than four times higher among PLHIV newly initiating ART as compared with those already established on ART. Among those ultimately diagnosed with TB disease, the majority of diagnoses were based on clinical criteria. The costs of the overall programme were modest per person screened.

Though there was a high prevalence of TB in this population, few patient-level factors were associated with a diagnosis of TB disease. We found only participants who were newly initiating ART were at significantly increased odds of TB disease diagnosis, and the prevalence of TB disease was 2.5%. However, even among PLHIV engaged in care and receiving ART, the prevalence of TB disease was also high at 1.5%. Both metrics are well above the WHO threshold for case-finding of 0.5% and these findings emphasise the importance of following WHO recommendations for systematic screening for TB disease among PLHIV at each visit to a health facility.15

Of all participants diagnosed with TB disease in this study, 20% (10/50) had subclinical TB disease (ie, no symptoms). This has important implications for the design and implementation of screening algorithms. Prevalence surveys in high TB burden countries consistently demonstrate about half of all TB disease detected is subclinical16 and the contribution of subclinical TB disease to transmission may be substantial.15 17 Among the 20 participants diagnosed with TB disease newly initiating ART, 6 (30%) had subclinical disease (asymptomatic), while only 4 (13%) of the 30 participants already on ART diagnosed with TB disease had subclinical disease. Though the sample size is small and all subclinical diagnoses were based on clinical criteria (ie, not microbiologically confirmed), these data can inform when and how frequently the tests that can detect subclinical disease should be implemented among PLHIV, given the relatively high cost of chest X-ray ($20.49) compared with symptom screening ($1.00) in this study. It could imply that if systematic chest X-ray is not possible for all PLHIV because of resource constraints, at least PLHIV newly initiating ART should have a chest X-ray regardless of the presence of clinical symptoms to avoid missing 30% of the prevalent TB among them.

The overall cost of systematic TB disease screening and provision of TPT for those without disease was $57.32 per person in this study. These costs may seem high, but they account for some of the downstream costs associated with implementing screening programmes—that is, those associated with treatment—which are extremely useful for budget impact assessments of TB programmes. Costs associated with this screening and treatment programme could be reduced through economies of scale, which would reduce the variable costs of overhead and equipment associated with Xpert MTB/RIF. The use of shorter TPT regimens, such as 1 month of daily isoniazid and rifapentine, however, are unlikely to reduce overall costs without substantial reductions in the cost of rifapentine.18 Though the long-term benefits of the programme, such as averted costs due to prevented TB and transmission, were not assessable in this study, modelling evidence has consistently demonstrated interventions to systematically screen for TB and provide TPT are highly cost-effective or even cost-saving in the long-term, including among populations of PLHIV with high ART coverage.19

A large portion of this study took place during the COVID-19 pandemic, which presented challenges and impacted participant recruitment. Despite this, we were able to complete the study after extending it for an additional 5 months due to commitment from staff and willingness of patients to participate as well as supportive supervision from investigators and the National TB Control Programme.

Key strengths of this study include the diverse health facilities included, relatively large sample size, comprehensive assessment of participants and the detailed cost assessment which included downstream costs associated with treatment. This study also has limitations. Only 20% of PLHIV attending participating health facilities were included in this study. We speculate this is due to the more intensive follow-up frequency required for TPT (monthly), as opposed to the bi-annual follow-up frequency for ART offered during the acute phase of the COVID-19 pandemic. This low participation likely resulted in selection of participants willing to accept and complete TPT and may affect generalisability. In data reported by Ghana to the WHO, only 50% of all PLHIV newly initiating ART received TPT in 2022; while not directly translatable as it is unknown if all were offered TPT, it does suggest the high uptake of TPT seen in this study would not be maintained during national implementation. The majority of TB disease diagnoses were clinical rather than microbiological. Factors associated with TB disease diagnosis therefore may be confounded by clinical judgement. Though most (81%) participants were female, this is only slightly higher than UNAIDS estimates20 for the proportion of the HIV burden among females (70%) in Ghana and it is unlikely a result of selection bias. Though we performed a comprehensive cost assessment, we could not include all costs, such as those associated with treatment-related adverse events, monitoring and evaluation, and any additional infrastructure that may be required during further scale-up. We also assumed all participants completed TB disease treatment and TPT, which may result in an overestimation of costs.

In summary, we found about 1 in 50 PLHIV attending participating health facilities had TB disease, and 20% was subclinical. The overall cost of this systematic screening and treatment programme was modest, with costs associated with screening and treatment of TB disease accounting for half of all costs. PLHIV newly initiating ART were at highest risk for TB disease, however, TB disease risk was high among all participants, supporting WHO recommendations that systematic screening of TB disease should be performed routinely among all PLHIV. These findings warrant wider implementation and further evaluation to ensure the yield and costs of the intervention remain acceptable.

Data availability statement

Data are available upon reasonable request. Data underlying this analysis can be made available by contacting the corresponding author.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and ethical approval for the study was obtained from the Ethics Review Committee of the Ghana Health Service with approval number GHS-ERC003/08/18. Approval was sought from the Regional Directors of Health Services and the medical superintendents of the facilities before entry into those facilities. Written informed consent was obtained from participants; parents/guardians provided consent for participants <18 years of age.

Acknowledgments

We would like to thank the study participants, and all staff involved in data collection and study execution, for without them this study would not be possible.

This post was originally published on https://bmjopen.bmj.com