Abstract
The aim of this study was to analyse temporal changes in treatments for and outcomes of multidrug-resistant (MDR)/rifampin-resistant (RR)-tuberculosis (TB) in the context of national economic status.
We analysed data collected by the Collaborative Group for the Meta-Analysis of Individual Patient Data in MDR-TB Treatment on MDR/RR-TB patients from 37 countries. The data were stratified by three national income levels (low-/lower-middle, upper-middle and high) and grouped by time of treatment initiation (2001–2003, 2004–2006, 2007–2009, 2010–2012 and 2013–2015). Temporal trends over the study period were analysed. The probability of treatment success in different income groups over time was calculated using generalised linear mixed models with random effects.
In total, 9036 patients were included in the analysis. Over the study period, use of group A drugs (levofloxacin/moxifloxacin, bedaquiline and linezolid) recommended by the World Health Organization increased and treatment outcomes improved in all income groups. Between 2001–2003 and 2013–2015, treatment success rates increased from 60% to 78% in low-/lower-middle-income countries, from 40% to 67% in upper-middle-income countries, and from 73% to 81% in high-income countries. In earlier years, the probability of treatment success in upper-middle-income countries was lower than that in low-/lower-middle-income countries, but no difference was observed after 2010. However, high-income countries had persistently higher probability of treatment success compared to upper-middle income countries.
Improved treatment outcomes and greater uptake of group A drugs were observed over time for patients with MDR/RR-TB at all income levels. However, treatment outcomes are still unsatisfactory, especially in upper-middle-income countries.
Abstract
Introduction of group A drugs is probably responsible for improved outcomes, even in resource-poor countries. However, a gap in treatment outcomes, which could not be fully explained by group A drugs, persists between high- and upper-middle-income countries https://bit.ly/2XJpO8U
Introduction
Drug-resistant tuberculosis (TB) is a major threat to global health. In 2018, approximately half a million cases of rifampin-resistant (RR)-TB occurred globally [1]. Among patients with RR-TB, 78% had multidrug-resistant (MDR)-TB, defined as resistance to both isoniazid and rifampin [1]. The number of reported cases of MDR/RR-TB is increasing, with ∼214 000 MDR/RR-TB-associated deaths occurring in 2018 [1].
Globally, the treatment success rate for MDR/RR-TB has been a disappointing 56% [1]. Antimicrobial therapy for MDR/RR-TB requires long durations of treatment. Under some circumstances, parenteral administration of drugs such as amikacin or streptomycin for 6–7 months may be required [2]. Treatment-associated adverse events are reported frequently [3–5].
Over the past two decades, several improvements have been made in the treatment of patients affected by MDR/RR-TB [6]. Later-generation fluoroquinolones, such as levofloxacin and moxifloxacin, have replaced the earlier-generation fluoroquinolones. Novel anti-TB drugs (e.g. bedaquiline and delamanid) have been introduced [7–9] and existing drugs (e.g. linezolid, carbapenem and clofazimine) have been repurposed to treat patients with MDR/RR-TB [10–12]. These drugs enabled the adoption of all-oral regimens in the current guideline [2]. Finally, universal adoption of the Xpert MTB/RIF assay has shortened the turnaround time for MDR/RR-TB diagnosis, enabling earlier treatment initiation [13].
Together, 20 countries, including upper-middle-income, lower-middle-income and low-income countries, account for 86% of global MDR/RR-TB incidence [1]. In some of these countries, the availability of later-generation fluoroquinolones, second-line injectable drugs and linezolid have been sparse [14–16]. Since the early 2000s, the Global Fund to Fight AIDS, Tuberculosis and Malaria has reduced the prices of second-line drugs in resource-poor settings [17]. In addition, bedaquiline is distributed at reduced prices or free of charge in low-income countries [18].
With the introduction of new drugs and improved treatment strategies for MDR-TB, treatment outcomes might improve. However, there have been few reports of changes in treatment modalities and outcomes among MDR/RR-TB patients over time, and there have been no analyses of these factors in the context of national income levels. This study aimed to analyse temporal changes in patient characteristics, drug susceptibility patterns, treatment modalities and treatment outcomes in the context of national economic status using a large MDR/RR-TB patient dataset.
Material and methods
Data collection
We analysed data collected by the Collaborative Group for the Meta-Analysis of Individual Patient Data in MDR-TB Treatment, which included datasets from 50 studies conducted in over 37 countries (supplementary table S1) [11]. The analysis was restricted to patients enrolled in observational studies who commenced treatment between January 1, 2001 and December 31, 2015. To reflect real-world practice, patients enrolled in randomised controlled trials were excluded. Additionally, to prevent distortion of results by a single study with many participants, we randomly selected a 10% sample from the 3626 patients included in a South African cohort study conducted in 2015.
Patients were classified into three groups based on World Bank categories of per capita gross national income in 2015 (in USD) [19] as follows: low/lower-middle-income (≤4035), upper-middle-income (4036–12 475) and high-income (≥12 476). In addition, the patients were grouped by 3-year intervals according to the year that treatment commenced: 2001–2003, 2004–2006, 2007–2009, 2010–2012 and 2013–2015.
Treatment outcomes were defined based on the recommendations of the World Health Organization (WHO) [20] and Laserson et al. [21] (supplementary table S1). Outcomes were classified as “success”, “treatment failure or relapse”, “death” and “loss to follow-up or transfer”. In our analysis, treatment success was defined as the sum of cures and treatment completions without evidence of relapse. Follow-up data after treatment completion were not available to define “cure” [22].
Data analysis
Baseline demographic characteristics (age, sex, body mass index, smoking status, HIV infection and diabetes mellitus), previous TB treatment history with first-line or second-line anti-TB drugs, acid-fast bacilli (AFB) smear status, radiographic features (presence of cavity or bilateral involvement), drug susceptibility patterns, treatment modalities (including drug use, hospitalisation and surgical resection) and treatment outcomes were compared across World Bank categories of per capita gross national income and 3-year intervals of treatment initiation.
Dichotomous variables were presented as frequencies (%) and continuous variables were described using medians and interquartile ranges (IQRs). Descriptive analyses were performed using complete case data without missing data imputation. Comparisons of continuous and dichotomous variables by income groups were made using the Kruskal–Wallis and Chi-squared tests, respectively. Tests for trend were used to analyse temporal changes over the study period.
Regression analyses used data where missing values were multiply imputed. Data on patient characteristics and drug susceptibility testing (DST) were imputed separately. Patient characteristics were imputed using demographic factors, previous treatment history, radiographic features, treatment outcomes and income level. Data on DST were imputed based on demographic factors, previous treatment history, DST, radiographic features, treatment outcomes and income level.
The primary analysis was treatment success versus all other outcomes (treatment failure or relapse, death and loss to follow-up or transfer), using imputed data. We also assessed survival (death versus all other outcomes). The odds ratio was calculated with 95% confidence intervals to identify treatment-related factors affecting treatment outcomes among different income groups over time, using generalised linear mixed-models with each study acting as the clustering variable. The resulting models included the unadjusted model (model 1) and stepwise models adjusted for 1) demographic characteristics (model 2); 2) all variables included in model 2 plus radiographic features, AFB smear status and previous TB treatment history (model 3); 3) all variables included in model 3 plus DST (model 4); and 4) all variables included in model 4 plus group A drugs (levofloxacin/moxifloxacin, bedaquiline and linezolid) according to a recent WHO classification [2] (model 5).
Data were collated and analysed using R (version 3.5.1) and the R lme4 package (version 1.1.21). Multiple imputations were performed using the R Multiple Imputation with Chained Equations (MICE) package (version 3.6.0). Values of p<0.05 were considered statistically significant. The study was approved by the institutional review board of McGill University Health Centre (Montreal, QC, Canada) and ethics approval was obtained from each participating institution.
Results
Patient classification, demographics and clinical characteristics
9036 patients with MDR/RR-TB started treatment between January 1, 2001 and December 31, 2015 and were included in the analysis. The dataset included 2612 (29%) patients from low-/lower-middle-income countries, 3926 (43%) from upper-middle-income countries and 2498 (28%) from high-income countries (table 1). The list of countries included in the analysis along with the number of patients from each country is shown in supplementary table S2.
The prevalence of HIV infection (39%) was highest in patients from upper-middle-income countries (table 1), and yet the use of antiretroviral therapy was lowest in the upper-middle income group (47%) as compared to the low-/low-middle (90%) and high (78%) income groups (supplementary table S3). Full data on patient demographics including smoking history and diabetes are provided in supplementary table S3. History of treatment with second-line anti-TB drugs was more common in upper-middle-income (31%) countries than in low-/lower-middle-income (13%) and high-income (22%) countries. History of treatment with second-line anti-TB drugs increased among all patients (pooled irrespective of income) over time, peaking in the 2007–2009 period at 26% and maintaining at this level for the remaining two time periods (table 2). Presence of cavities on chest radiographs was also more common in upper-middle-income countries (68%) than in low-/lower-middle-income (55%) and high-income (58%) countries (supplementary table S4).
Resistance to fluoroquinolones was more common in upper-middle-income countries (27%) than in low-/lower-middle-income (21%) and high-income (23%) countries (table 3). Resistance to any second-line injectable drug was also more common among patients from upper-middle-income countries (33%) than among those from low-/lower-middle-income (17%) and high-income (28%) countries (table 3). Among all patients (pooled irrespective of income), resistance to fluoroquinolones and second-line injectable drugs increased over time (table 3). Drug susceptibility patterns for other drugs are shown in supplementary table S5.
Treatment modalities
Overall, use of later-generation fluoroquinolones was more common among patients in high-income (77%) and low-/lower-middle-income (74%) countries than among those in upper-middle-income (32%) countries (table 4). Over time, use of later-generation fluoroquinolones increased in all income groups. By 2013–2015, most patients in low-/lower-middle-income countries (100%), upper-middle-income countries (90%) and high-income countries (87%) were treated with later-generation fluoroquinolones (table 4).
Linezolid was more frequently prescribed in high-income countries (26%) than in low-/lower-middle-income (4%) and upper-middle-income (11%) countries (table 4). As with later-generation fluoroquinolones, use of linezolid has increased over time in all income groups. Although the introduction of linezolid occurred last in upper-middle-income countries, treatment of patients with linezolid in these countries in 2013–2015 was more common (44%) than in low-/lower-middle-income countries (34%), but less common than in high-income countries (54%) (table 4).
Since 2010, the use of bedaquiline has increased rapidly in all countries, especially in upper-middle-income countries. Detailed patterns of other individual drug usage are shown in supplementary table S6.
During the study period, patients in the low-/low-middle and high-income countries were treated with a median of five drugs whereas individuals in the upper-middle-income countries were treated with an average of four drugs, with all three groups treated with an average of four effective anti-TB drugs (IQR 4–5) (table 5). Total treatment duration in patients who achieved treatment success was the longest in upper-middle-income countries (median 23 months, IQR 19.6–24.7 months), although it declined from 24.1 months (IQR 22.3–26.4 months) to 21 months (IQR 18.8–24.7 months) over time (p<0.001). Surgical resection was more common in high-income countries (11%) than in low-/lower-middle-income countries (3%) and upper-middle-income countries (5%) (table 5).
Treatment outcomes
Over the study period, treatment outcomes improved over time in all income groups. Overall, treatment success rates were highest in high-income countries and lowest in upper-middle-income countries. Deaths were more common among patients in upper-middle-income countries (32%) than in low-/lower-middle-income countries (17%) and high-income countries (8%) (p<0.001). While the proportion of patients who died decreased over time in upper-middle- and high-income countries, this proportion remained unchanged in low-/lower-middle-income countries (table 6).
The adjusted odds ratio (aOR) for treatment success versus all poor outcomes by income group (with upper-middle-income countries (the income group with the worst treatment outcomes) taken as reference) is provided in table 7. After adjusting for demographic factors, radiographic features, drug susceptibility patterns and use of group A drugs (model 5), the odds of treatment success were higher in low-/low-middle-income countries than in upper-middle-income countries (aOR 1.9, 95% CI 1.26–2.85) between 2001 and 2003, until 2007–2009 and thereafter there was no difference. The odds of treatment success in high-income countries were higher than those in upper-middle-income countries during the study periods (table 7).
The odds of death (versus all other outcomes) did not differ between patients in low-/low-middle-income countries and those in upper-middle-income countries over time except for the 2007–2009 period. Meanwhile, the probability of mortality was generally lower among patients in high-income countries than among patients in upper-middle-income countries during the study periods (supplementary table S7).
Discussion
In this study, we analysed data on 9036 patients with MDR/RR-TB treated between 2001 and 2015 according to national income levels. Over the period studied, MDR/RR-TB treatment outcomes improved over time in all patients despite the growing prevalence of resistance to fluoroquinolones and second-line injectable drugs. Overall, the treatment success rate was best in high-income countries (73%), followed by low-/lower-middle-income (64%) and then upper-middle-income (48%) countries. The probability of treatment success in upper-middle-income countries was lower than that in low-/lower-middle-income countries during the early study period, but was not different starting in 2010. However, treatment success was higher in high-income countries compared to upper-middle income countries throughout the study period.
MDR/RR-TB patients in upper-middle-income countries had lower treatment success and there are plausible reasons for this finding. Indeed, the MDR/RR-TB patients in the upper-middle-income countries had a higher prevalence of HIV infection, lower use of antiretroviral therapy, more cigarette smoking and presented with more advanced disease (presence of cavity and bilateral involvement) compared to patients in other countries. In addition, resistance to pyrazinamide, second-line injectable drugs and/or fluoroquinolones (the latter two reflected by greater numbers of extensively drug-resistant TB patients) were more commonly associated with patients in the upper-middle-income countries. HIV infection [23], presence of cavities [24] and additional drug resistance [25, 26] have all been reported as poor prognostic factors for MDR-TB patients. These may have contributed to the worse treatment outcomes observed in upper-middle-income countries.
Limited use of group A drugs may have also played a part in the worse outcomes observed among patients in upper-middle-income countries. During the earlier years of the study, later-generation fluoroquinolones were used much less commonly in these countries. During the 2001–2003 period, no patients received later-generation fluoroquinolones in upper-middle-income countries, whereas 12% of patients in low-/lower-middle-income countries and 91% of patients in high-income countries received these drugs. Furthermore, linezolid was introduced in upper-middle-income countries in 2008, long after its introduction in high-income (2001) and low-/lower-middle-income (2004) countries. Finally, access to newer drugs has in some cases been better in low-/lower-middle-income countries. For example, collaborations between the private and public sectors have enabled access to later-generation fluoroquinolones in Tanzania, Uganda and Zambia [27, 28]. Similarly, approval of a generic version of linezolid in the early 2000s facilitated access to this drug in India, whereas the patent for linezolid was retained until 2014 in South Africa, an upper-middle-income country [29, 30]. Finally, disbursements by the Global Fund to Fight AIDS, Tuberculosis and Malaria have been primarily concentrated in low-/lower-middle-income countries. Currently, the 10 countries that have benefited most from the Global Fund were all low-/lower-middle-income, except China [31].
Beginning in 2010, MDR/RR-TB treatment outcomes improved in upper-middle-income countries. The treatment success rate was 33% during the 2007–2009 period, but increased to 63% during the 2010–2012 period in these countries. Comparing the same periods, increased use of group A drugs occurred in upper-middle-income countries. Increased use of new and repurposed anti-TB drugs may have resulted in the improved treatment outcomes observed in upper-middle-income countries.
Although the probability of treatment success in upper-middle-income countries was lower than that in low-/lower-middle-income countries between 2001 and 2009, there was no difference from 2010 to 2015. However, the gap between upper-middle- and high-income countries persists even after adjusting for use of group A drugs. In low-income and middle-income countries, access to healthcare does not guarantee adequate treatment for diseases such as TB [32]. Only 13–45% of TB patients were correctly managed in these countries [32]. Untrained primary care providers or inadequate supervisory support could result in low quality of care in resource-poor settings [33]. In high-income countries, comprehensive and patient-centred approaches for TB patients, electronic device-based adherence monitoring [34], psychosocial interventions, social worker assistance and financial support [35, 36] may have contributed to improved treatment outcomes. In addition, the better overall healthcare systems of high-income countries may have played a role in improved treatment outcomes. Although improvements in personal healthcare access and quality have been observed globally over the past 25 years, these attainments have been slower to take hold in southern sub-Saharan Africa and south Asia, where TB is most prevalent [37].
To best interpret these results, it is important to consider the limitations of our study. First, the data analysed in this study were obtained from published or “to be published” articles rather than from nationwide reporting systems. Consequently, the results could have been affected by selection bias. Different proportions of patients from different countries were included in each period. This could cause fluctuations in several variables over the periods. For example, the fluctuating frequency of HIV co-infection in upper-middle-income countries was driven by different number of patients for each period from South Africa, the country with the highest burden of HIV infection in the world [38]. Second, the role of wealth inequality, which also affects TB incidence and prevalence [39, 40] could not be investigated in this study. Even in high-income countries, impoverished populations are at higher risk of TB [41] and such inequality may affect TB mortality within a country [42]. Third, the impact of healthcare systems and policies in each country could not be considered in this study. Different health policies could result in different TB treatment modalities and outcomes, even in countries with similar income levels [38, 43]. Despite these limitations, to our knowledge ours is the first study to use extensive patient data (representative >9000 patients with MDR/RR-TB) to assess changes in demographic factors, disease severity, drug susceptibility patterns, treatment modalities and treatment outcomes according to national income levels and over time.
In summary, treatment outcomes for patients with MDR/RR-TB have improved in all income groups, which in part reflects the effectiveness of group A drugs that have become widely available in recent years. Despite these improvements, treatment outcomes, especially in upper-middle-income countries, remain unsatisfactory. Greater investment in diagnosis, treatment, and support for MDR/RR-TB patients is urgently needed.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material ERJ-01394-2020.SUPPLEMENT
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-01394-2020.Shareable
Footnotes
This article has supplementary material available from erj.ersjournals.com
Author contributions: N. Kwak and J-J. Yim designed the study and protocol. N. Winters and J.R. Campbell did the data analysis. N. Kwak and J-J. Yim wrote the initial draft of the manuscript and all authors were involved at all stages of critical revision of manuscript. All the authors read and approved the final manuscript.
Conflict of interest: N. Kwak has nothing to disclose.
Conflict of interest: N. Winters has nothing to disclose.
Conflict of interest: J.R. Campbell has nothing to disclose.
Conflict of interest: E.D. Chan has nothing to disclose.
Conflict of interest: M. Gegia has nothing to disclose.
Conflict of interest: C. Lange reports personal fees for lectures from Chiesi, Gilead, Janssen, Lucane, Novartis, Oxoid, Berlin Chemie and Thermofisher, personal fees for advisory board work from Oxford Immunotec, outside the submitted work.
Conflict of interest: M. Lee has nothing to disclose.
Conflict of interest: V. Milanov has nothing to disclose.
Conflict of interest: D. Menzies has nothing to disclose.
Conflict of interest: J-J. Yim received donations of linezolid (Zyvox) from Pfizer Inc. and Delamanid (Deltyba) from Otsuka Pharmaceutical Co. and served as principal investigator on clinical trials.
Support statement: Initial assembly of the IPD was supported by grants from the European Respiratory Society, Centers for Disease Control and Prevention, Infectious Diseases Society of America, American Thoracic Society and the Canadian Institutes of Health Research (CIHR). In this analysis, N. Kwak was supported by CIHR (FRD331745), and J.R. Campbell by Fonds de Recherche Santé (award #258907). C. Lange was supported by the German Center for Infection Research (DZIF). Funding information for this article has been deposited with the Crossref Funder Registry.
- Received April 25, 2020.
- Accepted June 2, 2020.
- Copyright ©ERS 2020