Risk scores for predicting early antiretroviral therapy mortality in sub-Saharan Africa to inform who needs intensification of care: a derivation and external validation cohort study

BACKGROUND

Clinical scores to determine early (6-month) antiretroviral therapy (ART) mortality risk have not been developed for sub-Saharan Africa (SSA), home to 70% of people living with HIV. In the absence of validated scores, WHO eligibility criteria (EC) for ART care intensification are CD4 < 200/μL or WHO stage III/IV.

METHODS

We used Botswana XPRES trial data for adult ART enrollees to develop CD4-independent and CD4-dependent multivariable prognostic models for 6-month mortality. Scores were derived by rescaling coefficients. Scores were developed using the first 50% of XPRES ART enrollees, and their accuracy validated internally and externally using South African TB Fast Track (TBFT) trial data. Predictive accuracy was compared between scores and WHO EC.

RESULTS

Among 5553 XPRES enrollees, 2838 were included in the derivation dataset; 68% were female and 83 (3%) died by 6 months. Among 1077 TBFT ART enrollees, 55% were female and 6% died by 6 months. Factors predictive of 6-month mortality in the derivation dataset at p < 0.01 and selected for the CD4-independent score included male gender (2 points), ≥ 1 WHO tuberculosis symptom (2 points), WHO stage III/IV (2 points), severe anemia (hemoglobin < 8 g/dL) (3 points), and temperature > 37.5 °C (2 points). The same variables plus CD4 < 200/μL (1 point) were included in the CD4-dependent score. Among XPRES enrollees, a CD4-independent score of ≥ 4 would provide 86% sensitivity and 66% specificity, whereas WHO EC would provide 83% sensitivity and 58% specificity. If WHO stage alone was used, sensitivity was 48% and specificity 89%. Among TBFT enrollees, the CD4-independent score of ≥ 4 would provide 95% sensitivity and 27% specificity, whereas WHO EC would provide 100% sensitivity but 0% specificity. Accuracy was similar between CD4-independent and CD4-dependent scores. Categorizing CD4-independent scores into low (< 4), moderate (4-6), and high risk (≥ 7) gave 6-month mortality of 1%, 4%, and 17% for XPRES and 1%, 5%, and 30% for TBFT enrollees.

CONCLUSIONS

Sensitivity of the CD4-independent score was nearly twice that of WHO stage in predicting 6-month mortality and could be used in settings lacking CD4 testing to inform ART care intensification. The CD4-dependent score improved specificity versus WHO EC. Both scores should be considered for scale-up in SSA.

Tuberculosis preventive treatment opportunities at antiretroviral therapy initiation and follow-up visits

Setting

Twenty-two clinics providing HIV care and treatment in Botswana where tuberculosis (TB) and HIV comorbidity is as high as 49%.

Objectives

To assess eligibility of TB preventive treatment (TPT) at antiretroviral therapy (ART) initiation and at four follow-up visits (FUVs), and to describe the TB prevalence and associated factors at baseline and yield of TB diagnoses at each FUV.

Design

A prospective study of routinely collected data on people living with HIV (PLHIV) enrolled into care for the Xpert^® MTB/RIF Package Rollout Evaluation Study between 2012 and 2015.

Results

Of 6041 PLHIV initiating ART, eligibility for TPT was 69% (4177/6041) at baseline and 93% (5408/5815); 95% (5234/5514); 96% (4869/5079); and 97% (3925/4055) at FUV1, FUV2, FUV3, and FUV4, respectively. TB prevalence at baseline was 11% and 2%, 3%, 3% and 6% at each subsequent FUV. At baseline, independent risk factors for prevalent TB were CD4 <200 cells/mm³ (aOR = 1.4, P = 0.030); anemia (aOR = 2.39, P < 0.001); cough (aOR = 11.21, P < 0.001); fever (aOR = 2.15, P = 0.001); and weight loss (aOR = 2.60, P = 0.002).

Conclusion

Eligibility for TPT initiation is higher at visits post-ART initiation, while most cases of active TB were identified at ART initiation. Missed opportunities for TB further compromises TB control effort among PLHIV in Botswana.

Outcomes of HIV-positive patients with non-tuberculous mycobacteria positive culture who received anti-tuberculous treatment in Botswana: Implications of using diagnostic algorithms without non-tuberculous mycobacteria

BACKGROUND

Patients with non-tuberculous mycobacteria (NTM) or Mycobacterium tuberculosis (MTB) pulmonary disease may have similar clinical presentation. The potential for misdiagnosis and inappropriate treatment exists in settings with limited testing capacity for Xpert® MTB/RIF (Xpert), phenotypic culture and NTM speciation. We describe treatment outcomes among people living with HIV (PLHIV) who received anti-tuberculosis treatment and were found to have NTM or MTB positive sputum cultures.

METHODS

PLHIV attending one of the 22 participating HIV clinics, who screened positive for ≥1 tuberculosis (TB) symptoms (cough, fever, night sweats, or weight loss) were asked to submit sputa for culture and speciation from August 2012 to November 2014. The national intensified TB case finding algorithms were followed: initially symptomatic patients were evaluated by testing sputum samples using a smear (smear-based TB diagnostic algorithm) and, after GeneXpert instruments were installed, by testing with Xpert (Xpert-based TB diagnostic algorithm). Within the study period, TB diagnostic algorithms used for MTB did not include screening, diagnosis, and management of NTM. Despite MTB negative culture, some symptomatic patients, including those with NTM positive culture, received empirical anti-TB treatment at the discretion of treating clinicians. Per the World Health Organization treatment outcomes classification: died, treatment failure or loss-to-follow-up were classified as unfavorable (unsuccessful) outcome; cured and treatment completed were classified as favorable (successful) outcome. Empiric treatment was defined as initiating treatment without or before receiving a test result indicating MTB. We compare treatment outcomes and characteristics among patients with NTM or MTB positive culture who received anti-TB treatment.

RESULTS

Among 314 PLHIV, who were found co-infected with TB, 146 cases had microbiological evidence; and for 131/146 MTB positive cultures were reported. One-hundred fifty-two of the 314 were clinically diagnosed with TB and treated empirically. Among those empirically treated for TB, 36/152 had culture results positive for NTM, and another 43/152 had culture results positive for MTB, reported after patients received empirical anti-TB treatment. Overall, MTB positive culture results were reported for 174 (131 plus 43) patients. Treatment outcomes were available for 32/36 NTM and 139/174 MTB; unfavorable outcomes were 12.5% and 8.7% for NTM and MTB, respectively, p = 0.514, respectively. For 34/36 tested NTM patients, all Xpert results indicated 'no MTB'. Among patients who initially received empiric anti-TB treatment and ultimately were found to have MTB positive culture, the unfavorable outcome was 11.8% (4/34), compared to 12.5% (4/32) of patients with NTM positive culture, Fisher's exact test p = 1.00.

CONCLUSIONS

While the higher unfavorable outcome was non statistically significant, the impact of inappropriate treatment among NTM patients should not be overlooked. Our findings suggest that Xpert has the potential to rapidly rule-out NTM and avoid sub-optimal treatment; further research is needed to evaluate such potential.

Improving sputum collection processes to increase tuberculosis case finding among HIV-positive persons in Botswana

SETTING

Twenty-two human immunodeficiency virus (HIV) clinics in Botswana.

OBJECTIVE

To compare sputum collection rates, sputum quality and volume, and tuberculosis (TB) diagnosis rates before and after field efforts to improve sputum collection among individuals newly diagnosed with HIV with TB symptoms.

DESIGN

Newly diagnosed individuals living with HIV attending 22 HIV clinics in Botswana were screened for TB from August 2012 to March 2014. Starting in May 2013, a field intervention composed of the introduction of a tracking log for presumed TB patients, and patient instructions and sputum induction to improve sputum collection rates was implemented.

RESULTS

Prior to the intervention, sputum collection rates were 44.1% (384/870). Subsequently, sputum collection increased to 58.3% (579/993) (P < 0.001). Sputum quality and volume also improved. Although rates of TB diagnosis increased from 9.7% (84/870) to 12.5% (120/993), this difference was not significant (P = 0.143).

CONCLUSION

Sputum collection rates among presumptive TB cases, as well as sputum quality and volume improved after implementation of the field intervention. To improve sputum collection rates, efforts at the program level should be ongoing.

Effect of tuberculosis screening and retention interventions on early antiretroviral therapy mortality in Botswana: a stepped-wedge cluster randomized trial

BACKGROUND

Undiagnosed tuberculosis (TB) remains the most common cause of HIV-related mortality. Xpert MTB/RIF (Xpert) is being rolled out globally to improve TB diagnostic capacity. However, previous Xpert impact trials have reported that health system weaknesses blunted impact of this improved diagnostic tool. During phased Xpert rollout in Botswana, we evaluated the impact of a package of interventions comprising (1) additional support for intensified TB case finding (ICF), (2) active tracing for patients missing clinic appointments to support retention, and (3) Xpert replacing sputum-smear microscopy, on early (6-month) antiretroviral therapy (ART) mortality.

METHODS

At 22 clinics, ART enrollees > 12 years old were eligible for inclusion in three phases: a retrospective standard of care (SOC), prospective enhanced care (EC), and prospective EC plus Xpert (EC+X) phase. EC and EC+X phases were implemented as a stepped-wedge trial. Participants in the EC phase received SOC plus components 1 (strengthened ICF) and 2 (active tracing) of the intervention package, and participants in the EC+X phase received SOC plus all three intervention package components. Primary and secondary objectives were to compare all-cause 6-month ART mortality between SOC and EC+X and between EC and EC+X phases, respectively. We used adjusted analyses, appropriate for study design, to control for baseline differences in individual-level factors and intra-facility correlation.

RESULTS

We enrolled 14,963 eligible patients: 8980 in SOC, 1768 in EC, and 4215 in EC+X phases. Median age of ART enrollees was 35 and 64% were female. Median CD4 cell count was lower in SOC than subsequent phases (184/μL in SOC, 246/μL in EC, and 241/μL in EC+X). By 6 months of ART, 461 (5.3%) of SOC, 54 (3.2%) of EC, and 121 (3.0%) of EC+X enrollees had died. Compared with SOC, 6-month mortality was lower in the EC+X phase (adjusted hazard ratio, 0.77; 95% confidence interval, 0.61-0.97, p = 0.029). Compared with EC enrollees, 6-month mortality was similar among EC+X enrollees.

CONCLUSIONS

Interventions to strengthen ICF and retention were associated with lower early ART mortality. This new evidence highlights the need to strengthen ICF and retention in many similar settings. Similar to other trials, no additional mortality benefit of replacing sputum-smear microscopy with Xpert was observed.

TRIAL REGISTRATION

Retrospectively registered: ClinicalTrials.gov (NCT02538952).

Tuberculosis treatment outcomes among people living with HIV diagnosed using Xpert MTB/RIF versus sputum-smear microscopy in Botswana: a stepped-wedge cluster randomised trial

Background

Xpert® MTB/RIF (Xpert) has high sensitivity for diagnosing tuberculosis (TB) compared to sputum-smear microscopy (smear) and can reduce time-to-diagnosis, time-to-treatment and potentially unfavorable patient-level treatment outcome.

Methods

People living with HIV (PLHIV) initiating antiretroviral therapy at 22 HIV clinics were enrolled and underwent systematic screening for TB (August 2012–November 2014). GeneXpert instruments were deployed following a stepped-wedge design at 13 centers from October 2012–June 2013. Treatment outcomes classified as an unfavorable outcome (died, treatment failure or loss-to-follow-up) or favorable outcome (cured and treatment completed). To determine outcome, smear was performed at month 5 or 6. Empiric treatment was defined as initiating treatment without/before receiving TB-positive results. Adjusting for intra-facility correlation, we compared patient-level treatment outcomes between patients screened using smear (smear arm)- and Xpert-based algorithms (Xpert arm).

Results

Among 6041 patients enrolled (smear arm, 1816; Xpert arm, 4225), 256 (199 per 2985 and 57 per 1582 person-years of follow-up in Xpert and smear arms, respectively; adjusted incidence rate ratio, 9.07; 95% confidence interval [CI]: 4.70–17.48; p < 0.001) received TB diagnosis and were treated. TB treatment outcomes were available for 203 patients (79.3%; Xpert, 157; smear, 46). Unfavorable outcomes were reported for 21.7% (10/46) in the smear and 13.4% (21/157) in Xpert arm (adjusted hazard ratio, 1.40; 95% CI: 0.75–2.26; p = 0.268). Compared to smear, in Xpert arm median days from sputum collection to TB treatment was 6 days (interquartile range [IQR] 2–17 versus 22 days [IQR] 3–51), p = 0.005; patients with available sputum test result had microbiologically confirmed TB in 59.0% (102/173) versus 41.9% (18/43), adjusted Odds Ratio [aOR], 2.00, 95% CI: 1.01–3.96, p = 0.048). In smear arm empiric treatment was 68.4% (39/57) versus 48.7% (97/199), aOR, 2.28, 95% CI: 1.24–4.20, p = 0.011), compared to Xpert arm.

Conclusions

TB treatment outcomes were similar between the smear and Xpert arms. However, compared to the smear arm, more patients in the Xpert arm received a TB diagnosis, had a microbiologically confirmed TB, and had a shorter time-to-treatment, and had a lower empiric treatment. Further research is recommended to identify potential gaps in the Botswana health system and similar settings.

Trial registration

ClinicalTrials.gov Identifier: NCT02538952. Retrospectively registered on 2 September 2015.

The effect of sputum quality and volume on the yield of bacteriologically-confirmed TB by Xpert MTB/RIF and smear

Introduction

The World Health Organization endorsed (2010) the use of Xpert MTB/RIF and countries are shifting from smear microscopy (smear)-based to Xpert MTB/RIF-based tuberculosis (TB) diagnostic algorithms. As with smear, sputum quality may predict the likelihood of obtaining a bacteriologically-confirmed TB when using Xpert MTB/RIF.

Methods

From 08/12-11/2014, all people living with HIV were recruited at 22 clinics. For patients screened positive using the four TB symptoms their sputa were tested by Xpert MTB/RIF and smear. Laboratorians assessed and recorded sputum appearance and volume. The yield of bacteriologically-positive sputum evaluated using Xpert MTB/RIF and smear, likelihood-ratios were calculated.

Results

Among 6,041 patients enrolled 2,296 were presumptive TB, 1,305 (56.8%) had > 1 sputa collected and 644/1,305 (49.3%) had both Xpert MTB/RIF and smear results. Since >1 sputa collected from 644 patients 954 sputa were tested by Xpert MTB/RIF and smear. Bacteriologically-positive sputum was two-fold higher with Xpert MTB/RIF 11.4% versus smear 5.3%, p < 0.001. Sputum appearance and quantity were not predictive of bacteriologically-positive results, except volume of 2ml to < 3ml, tested by Xpert MTB/RIF LR+= 1.26 (95% CI, 1.05–1.50).

Conclusion

Xpert MTB/RIF test yield to bacteriologically-positive sputum was superior to smear. Sputum quality and quantity, however, were not consistently predictive of bacteriologically-positive results by Xpert MTB/RIF or smear.

Higher-than-expected prevalence of non-tuberculous mycobacteria in HIV setting in Botswana: Implications for diagnostic algorithms using Xpert MTB/RIF assay

Background

Non-tuberculous mycobacteria (NTM) can cause pulmonary infection and disease especially among people living with HIV (PLHIV). PLHIV with NTM disease may clinically present with one of the four symptoms consistent with tuberculosis (TB). We describe the prevalence of NTM and Mycobacterium tuberculosis complex (MTBC) isolated among PLHIV who presented for HIV care and treatment.

Methods

All PLHIV patients presenting for HIV care and treatment services at 22 clinical sites in Botswana were offered screening for TB and were recruited. Patients who had ≥1 TB symptom were asked to submit sputa for Xpert MTB/RIF and culture. Culture growth was identified as NTM and MTBC using the SD-Bioline TB Ag MPT64 Kit and Ziehl Neelsen microscopy. NTM and MTBC isolates underwent species identification by the Hain GenoType CM and AS line probe assays.

Results

Among 16, 259 PLHIV enrolled 3068 screened positive for at least one TB symptom. Of these, 1940 submitted ≥1 sputum specimen, 427 (22%) patients had ≥1 positive-culture result identified phenotypically for mycobacterial growth. Of these 247 and 180 patients were identified as having isolates were NTM and MTBC, respectively. Of the 247 patients identified with isolates containing NTM; 19 were later excluded as not having NTM based on additional genotypic testing. Among the remaining 408 patients 228 (56%, 95% confidence interval, 46–66%) with NTM. M. intracellulare was the most common isolated (47.8%). Other NTMs commonly associated with pulmonary disease included M. malmoense (3.9%), M. avium (2.2%), M. abscessus (0.9%) and M. kansasii (0.4%). After excluding NTM isolates that were non-speciated and M. gordonae 154 (67.5%) of the NTM isolates were potential pathogens.

Conclusions

In the setting of HIV care and treatment, over-half (56%) of a positive sputum culture among PLHIV with TB symptoms was NTM. Though we were not able to distinguish in our study NTM disease and colonization, the study suggests culture and species identification for PLHIV presenting with TB symptoms remains important to facilitate NTM diagnosis and hasten time to appropriate treatment.

Peripheral clinic versus centralized laboratory-based Xpert MTB/RIF performance: Experience gained from a pragmatic, stepped-wedge trial in Botswana

Background

In 2011, the Botswana National Tuberculosis Program adopted World Health Organization guidelines and introduced Xpert MTB/RIF (Xpert) assay to support intensified case finding among people living with HIV enrolling in care. An evaluation was designed to assess performance under operational conditions to inform the national Xpert scale-up.

Methods

Xpert was implemented from August 2012 through November 2014 with 13 GeneXpert instruments (GeneXpert) deployed in a phased approach over nine months: nine centralized laboratory and four point-of-care (POC) peripheral clinics. Clinicians and laboratorians were trained on the four-symptom tuberculosis screening algorithm and Xpert testing. We documented our experience with staff training and GeneXpert performance. Test results were extracted from GeneXpert software; unsuccessful tests were analysed in relation to testing sites and trends over time.

Results

During 276 instrument-months of operation a total of 3,630 tests were performed, of which 3,102 (85%) were successful with interpretable results. Mycobacterium tuberculosis complex was detected for 447 (14%); of these, 36 (8%) were rifampicin resistant. Of all 3,630 Xpert tests, 528 (15%) were unsuccessful; of these 361 (68%) were classified as “error”, 119 (23%) as “invalid” and 48 (9%) as “no result”. The total number of recorded error codes was 385 and the most common reasons were related to sample processing (211; 55%) followed by power supply (77; 20%) and cartridge/module related (54; 14%). Cumulative incidence of unsuccessful test was similar between POC (17%, 95% CI: 11–25%) and centralized laboratory-based GeneXpert instruments (14%, 95% CI: 11–17%; p = 0.140).

Conclusions

Xpert introduction was successful in the Botswana setting. The incidence of unsuccessful test was similar by GeneXpert location (POC vs. centralized laboratory). However, unsuccessful test incidence (15%) in our settings was higher than previously reported and was mostly related to improper sample processing. Ensuring adequate training among Xpert testing staff is essential to minimize errors.

Repeated assessments of informed consent comprehension among HIV-infected participants of a three-year clinical trial in Botswana

BACKGROUND

Informed consent (IC) has been an international standard for decades for the ethical conduct of clinical trials. Yet frequently study participants have incomplete understanding of key issues, a problem exacerbated by language barriers or lack of familiarity with research concepts. Few investigators measure participant comprehension of IC, while even fewer conduct interim assessments once a trial is underway.

METHODS AND FINDINGS

We assessed comprehension of IC using a 20-question true/false quiz administered in 6-month intervals in the context of a placebo-controlled, randomized trial for the prevention of tuberculosis among HIV-infected adults in Botswana (2004-2009). Quizzes were offered in both Setswana and English. To enroll in the TB trial, participants were required to have ≥ 16/20 correct responses. We examined concepts understood and the degree to which understanding changed over three-years. We analyzed 5,555 quizzes from 1,835 participants. The participants' highest education levels were: 28% primary, 59% secondary, 9% tertiary and 7% no formal education. Eighty percent of participants passed the enrollment quiz (Quiz1) on their first attempt and the remainder passed on their second attempt. Those having higher than primary education and those who took the quiz in English were more likely to receive a passing score on their first attempt (adjusted odds ratios and 95% confidence intervals, 3.1 (2.4-4.0) and 1.5 (1.2, 1.9), respectively). The trial's purpose or procedures were understood by 90-100% of participants, while 44-77% understood randomization, placebos, or risks. Participants who failed Quiz1 on their initial attempt were more likely to fail quizzes later in the trial. Pass rates improved with quiz re-administration in subsequent years.

CONCLUSIONS

Administration of a comprehension quiz at enrollment and during follow-up was feasible in a large, international collaboration and efficiently determined IC comprehension by trial participants. Strategies to improve understanding of concepts like placebos and randomization are needed. Comprehension assessments throughout a study may reinforce key concepts.

Risk factors for non-adherence and loss to follow-up in a three-year clinical trial in Botswana

Background

Participant non-adherence and loss to follow-up can compromise the validity of clinical trial results. An assessment of these issues was made in a 3-year tuberculosis prevention trial among HIV-infected adults in Botswana.

Methods and Findings

Between 11/2004–07/2006, 1995 participants were enrolled at eight public health clinics. They returned monthly to receive bottles of medication and were expected to take daily tablets of isoniazid or placebo for three years. Non-adherence was defined as refusing tablet ingestion but agreeing to quarterly physical examinations. Loss to follow-up was defined as not having returned for appointments in ≥60 days. Between 10/2008–04/2009, survey interviews were conducted with 83 participants identified as lost to follow-up and 127 identified as non-adherent. As a comparison, 252 randomly selected adherent participants were also surveyed. Multivariate logistic regression analysis was used to identify associations with selected risk factors. Men had higher odds of being non-adherent (adjusted odds ratio (AOR), 2.24; 95% confidence interval [95%CI] 1.24–4.04) and lost to follow-up (AOR 3.08; 95%CI 1.50–6.33). Non-adherent participants had higher odds of reporting difficulties taking the regimen or not knowing if they had difficulties (AOR 3.40; 95%CI 1.75–6.60) and lower odds associated with each year of age (AOR 0.95; 95%CI 0.91–0.98), but other variables such as employment, distance from clinic, alcohol use, and understanding study requirements were not significantly different than controls. Among participants who were non-adherent or lost to follow-up, 40/210 (19.0%) reported that they stopped the medication because of work commitments and 33/210 (15.7%) said they thought they had completed the study.

Conclusions

Men had higher odds of non-adherence and loss to follow-up than women. Potential interventions that might improve adherence in trial participants may include:targeting health education for men, reducing barriers, clarifying study expectations, educating employers about HIV/AIDS to help reduce stigma in the workplace, and encouraging employers to support employee health.

Trial Registration

ClinicalTrials.gov NCT00164281