Associated characteristics of multidrug-resistant tuberculosis (MDR-TB) cases in Sub-Saharan Africa: a systematic review and meta-analysis

Abstract

Background

The persistent spread of tuberculosis (TB), driven by multidrug-resistant TB (MDR-TB), poses a major public health threat in Sub-Saharan Africa (SSA), worsening the region’s already high burden of infectious disease mortality. Despite fragmented evidence, a comprehensive quantitative assessment of MDR-TB patient characteristics that fuel resistance across the continent has been limited. This study aimed to synthesize and quantify the demographic, clinical, socio-economic, and behavioral patient profiles associated with MDR-TB in SSA to inform targeted control strategies.

Methods

We conducted a systematic review and meta-analysis of observational studies published between January 2000 and November 2025 across five databases (PubMed/MEDLINE, Scopus, Web of Science, AJOL, and Google Scholar). Pooled proportions of demographic, clinical, socio-economic, behavioral, and treatment-related characteristics reported among MDR-TB patients were calculated using a random-effects model, with heterogeneity quantified by the I² statistic.

Results

Thirty-six studies from 15 African countries, predominantly Ethiopia (36%), involving 10,792 participants and 3,864 MDR-TB cases, were included. The pooled proportion of MDR-TB cases across these heterogenous study populations was 39.0% (95% CI: 29–49; I² = 99.9%). Demographically, males (70%) were disproportionately affected. Key socioeconomic characteristics included being a farmer or day labourer (77%; 95% CI: 53–100) and large family size (>5 members) (55.0%; 95% CI: 48–62). HIV co-infection was reported in 34% of cases. Major behavioral and clinical characteristics included alcohol use disorder (48%) and record of previous TB treatment (63%). A history of direct contact with known TB patients was observed in 49% of the cases. While a lack of BCG vaccination (69%) and consultation with traditional healers (86%) were noted, though these parameters were evaluated in only two studies and cannot be generalized across the wider subcontinent.

Conclusion

The findings suggest that MDR-TB across the included study population in Sub-Saharan Africa is commonly reported alongside systemic treatment failures, behavioral factors, and socio-economic vulnerabilities. Addressing these challenges requires strengthening directly observed therapy short-course (DOTS) through community engagements, integrating traditional healers into referral systems, and enhancing social support to improve adherence.

Clinical trial number

Not applicable.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12879-026-14008-9.

Introduction

The global burden of tuberculosis (TB) remains pronounced in Sub-Saharan Africa (SSA), which continues to be unduly affected by one of the world’s leading infectious causes of death. Although the region accounts for only about 14% of the world’s population, it contributes approximately 24% of global TB cases [1]. Efforts to control TB across Sub-Saharan Africa are being compromised by the rising incidence of multidrug-resistant tuberculosis (MDR-TB). Defined as resistance to at least rifampicin and isoniazid, MDR-TB constitutes a severe threat to treatment efficacy and destabilizes regional health safety frameworks. In 2008, MDR-TB was estimated to have caused about 150,000 deaths globally, with 390,000–510,000 new MDR-TB cases emerging worldwide (best estimate: 440,000). At that time, 3.6% (95% CI: 3.0–4.4) of all incident TB cases were estimated to be MDR-TB, based on expanded datasets and revised analytical methodologies [2]. Furthermore, recent advancements in predictive modeling and advanced biomathematical analyses underscore that without targeted, data-driven interventions, the epidemiological trajectory of drug-resistant TB will continue to outpace traditional public health responses, compounding the region’s overall mortality burden [3, 4].

Systematic reviews indicate that MDR-TB prevalence among newly diagnosed TB patients in SSA remains relatively low but not negligible. Pooled prevalence estimates range from approximately 1.5 - 2.1%, with a modest decline observed over the past two decades [5, 6]. However, due to the high underlying TB incidence in the region, these proportions translate into a substantial absolute disease burden. In 2015, this was estimated at about 54 MDR-TB cases per 100,000 population among new TB patients [5]. A recent continent-wide geospatial analysis estimated approximately 148,000 drug-resistant TB cases annually in Africa, with particularly high prevalence observed in Eswatini (formerly Swaziland), South Africa, Morocco, and Tunisia, alongside pronounced subnational heterogeneity [7]. Global burden studies similarly identify Africa as one of the regions with the highest MDR-TB incidence and associated disability burden worldwide [8].

Data collected across Sub-Saharan Africa shows that a history of previous anti-TB chemotherapy is a characteristic frequently shared by patients diagnosed with MDR-TB, highlighting the role of inadequate or incomplete treatment in the development of resistance. Other major contributors include HIV co-infection, overcrowding, and systemic weaknesses within health-care systems that facilitate ongoing transmission and poor treatment outcomes [9–11]. Molecular epidemiological evidence indicates that a substantial proportion of MDR-TB in Africa results from primary transmission of resistant strains often linked to lineage 4, rather than resistance acquired during therapy. Community-based and nosocomial outbreaks, as well as widespread MDR/XDR-TB transmission clusters, have also been documented across several settings [12, 13].

Despite evidence of a gradual decline in MDR/RR-TB incidence from historically high levels, diagnostic capacity and routine surveillance across the region remain inadequate, resulting in significant under-detection of cases and delayed response to transmission hotspots [7, 14]. Considerable regional variation persists in different African regions and countries. For instance, East African MDR-TB prevalence is estimated at approximately 4% among newly diagnosed TB patients and up to 21% among previously treated cases, with risk factors such as poor living conditions, substance use, diabetes, HIV co-infection, and prior TB treatment [9]. Works from Ethiopia and several West African countries have report particularly high burdens among previously treated patients, reinforcing the dominant role of treatment history [10, 11].

Though an expanding body of primary research exist on MDR-TB, a comprehensive synthesis and quantitative assessment of these risk drivers across the African continent remain limited. This systematic review and meta-analysis therefore aimed to identify, synthesize, and quantify key modifiable and contextual factors associated with increased MDR-TB cases across Africa, with the goal of informing targeted interventions, strengthening TB control strategies, and supporting progress toward achieving the WHO End TB and Global Stop TB targets by 2035.

Methods

Study design and search strategy

We performed this study in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework [15]. Pertinent literature was sourced by querying PubMed/MEDLINE, Scopus, Web of Science, Google Scholar, and African Journals Online (AJOL). Data collection covered the period between January 2000 and November 2025 and did not restrict results based on language. The retrieval strategy relied on a combination of Medical Subject Headings (MeSH) and free-text terms related to multidrug-resistant tuberculosis, associated features, risk determinants, and Sub-Saharan Africa. A detailed breakdown of the search syntax used for each database is provided in Table Supplementary document (S2). To ensure a comprehensive review, we also hand-searched the bibliographies of eligible articles and related reviews to uncover any studies that may have been missed by electronic indexing.

Eligibility criteria

We used observational research studies (cross-sectional, case-control, and cohort studies) that reported MDR-TB incidence and associated risk/characteristic factors among TB patients. Inclusion criteria were: (1) studies involving human participants with confirmed TB diagnosis; (2) studies reporting MDR-TB prevalence data (number of MDR-TB cases and total sample size); (3) studies reporting at least one associated factor with MDR-TB; and (4) studies with original data.

The primary outcome was the pooled prevalence of MDR-TB among TB patients. Secondary outcomes included pooled estimates of associated features connected with MDR-TB development. We excluded reviews, case reports, editorials, case series with fewer than 10 participants, conference abstracts without full data, and studies with incomplete methodological information.

Study selection and data extraction

Two authors (M.Z.P and S.T) independently screened all search results, first reviewing titles and abstracts and then examining full texts of eligible papers. A third reviewer (A.O) acted as an arbiter for any unresolved disagreements. Data were captured using a structured form piloted on a subset of five studies. Extracted data points comprised: study metadata (author, year, country, period), methodology, setting, participant demographics, and diagnostic techniques. We also extracted MDR-TB prevalence rates and patient profile estimates. We grouped these characteristic factors into five specific categories: demographic, clinical, behavioral, socio-economic, and treatment-related.

Quality assessment

We assessed the methodological quality of the included literature using the Newcastle-Ottawa Scale (NOS) adapted for observational research [16]. The appraisal was performed independently by authors A.O and M.Z.P, who resolved any discrepancies through consensus. The NOS framework assigns up to 9 points across three categories: participant selection (maximum 4), group comparability (maximum 2), and outcome assessment (maximum 3). Based on the aggregate score, we stratified study quality as high (7–9 points), moderate (5–6 points), or low (0–4 points), as detailed in Table S3.

Statistical analysis

All statistical assessments were executed using R statistical software version 4.5.1 (2025–06–13 ucrt) with the `meta`, and `metafor`, packages. The pooled proportion of MDR-TB prevalence was estimated using a generalized linear mixed model (GLMM) with logit transformation, which is recommended for meta-analysis of proportions. Between-study heterogeneity was quantified by means of the I² statistic, with values of 25, 50, and 75% signifying low, moderate, and high heterogeneity, accordingly. The τ² statistic was used to estimate between-study variance. Pooled estimates are presented with 95% confidence intervals (CIs). For each characteristic profile, we calculated pooled proportions of patients with the factor among MDR-TB cases. When studies reported odds ratios (ORs) with 95% CIs for risk/associated factors, these were pooled using the DerSimonian and Laird random-effects model. Forest plots were generated to visually display study-specific and pooled estimates. We performed subgroup analyses (country, study design and study setting) to explore potential sources of heterogeneity. To evaluate the stability of our results, we performed a leave-one-out sensitivity analysis, determining if any single study disproportionately skewed the pooled estimates. We utilized random-effects exploratory meta-regression to investigate potential associations between MDR-TB prevalence and specific covariates. Rather than an over-saturated multivariable structure, covariates and patient profiles including region, setting, design, year, and diagnostic method, were strictly screened sequentially in isolated regression iterations. Finally, we screened for publication bias using a dual approach: visual inspection of funnel plots and statistical verification via Egger’s regression test.

Ethical considerations

The study did not use any primary data but drew information from publicly available domain with the protocol registered in PROSPERO systematic review and meta-analysis database (ID. No: CRD420251185459).

Results

Search result

The systematic search across electronic databases yielded a total of 4,181 initial records. Specifically, the search identified 1,312 records from PubMed/MEDLINE, 1,668 from Scopus, 1,091 from Web of Science, and 107 from African Journals Online (AJOL). An additional 3 records were identified through manual searches of Google Scholar. Following the elimination of 1,164 duplicate entries, we screened 3,017 unique citations based on their titles and abstracts. This initial review led to the exclusion of 2,972 papers that fell outside the study scope, leaving 45 articles for full-text eligibility review. Nine reports were subsequently excluded for reasons including the lack of reported MDR-TB prevalence data (n = 4), incomplete data or failure to establish specific risk factors (n = 3), and a focus solely on Rifampicin resistance (n = 2). Ultimately, 36 studies [17–52] satisfied all inclusion criteria and were included in the systematic review and meta-analysis (Fig. 1).

Fig. 1.

PRISMA chart for study selection

Study characteristics

The included studies were published between 2004 and 2024, representing data from across 15 African countries (Fig. 2). Most of the studies were conducted in Ethiopia (n = 13), followed by Nigeria (n = 5), and Tanzania (n = 3). Other countries represented that had 2 studies each include Ghana, Sudan, and Uganda, while Cameroon, Burundi, Kenya, Namibia, South Africa, Eswatini, Botswana, Equatorial Guinea, and the Central African Republic all had single studies (n = 1). The study settings varied, with researchers employing both multicenter (n = 21) and single-site (n = 15) locations. The total population across the included studies comprised 10,792 participants confirmed with TB, from which 3,864 MDR-TB cases were identified. The median age of participants in the included studies ranged from approximately 24 to 44.5 years [Table S1].

Fig. 2.

Country distribution of studies across Africa

Quality assessment

We assessed the quality of all included observational studies using the NOS. The detailed quality assessment for each included article is presented in Table S3. Of the studies assessed, approximately half (n = 17) were classified as high quality, achieving scores between 7 and 9. These studies demonstrated robust methodologies, particularly in the selection of participants and the assessment of outcomes. Notable examples of high-quality studies included works by Moore et al. (2015), Mesfin et al. (2018), and Okumu et al. (2024) [27, 30, 42], which scored 8 points each. The remaining studies (n = 18) were classified as moderate quality, with scores ranging from 5 to 6. While these studies generally performed well in the selection and outcome domains, they often lost points in the comparability domain. Many moderate-quality studies did not control for all relevant confounding factors or lacked a non-exposed cohort description, which is common in descriptive cross-sectional designs often used in TB prevalence surveys. No studies were excluded due to low quality (scores 0–4), indicating that all included literature met the minimum methodological standards required for this meta-analysis.

Pooled proportions of MDR-TB and subgroup analyses

Using a random-effects model, we calculated a pooled proportion of MDR-TB prevalence of 39.0% (95% CI: 29–49) across the included literature. The analysis revealed substantial heterogeneity (I² = 99.9%; τ² = 0.0883; p < 0.001), demonstrating that the data are derived from highly diverse patient populations across different institutional settings rather than reflecting a single continental average.

Subgroup analysis by country

Stratifying the data by country highlighted considerable variation from included works revealing 36% (13/36) from Ethiopia, 13.9% (5/36) from Nigeria, 8.3% (3/36) from Tanzania, 2 each from Ghana, Sudan and Uganda (5.6%) and 1 single study each (2.8%) from the remaining 9 African countries. To quantify potential geographic bias driven by the overrepresentation of studies from Ethiopia, we performed a subgroup analysis stratifying Ethiopia against all other Sub-Saharan African (SSA) regions (Figure S27). The pooled proportion of MDR-TB prevalence in Ethiopia was 29.0% (95% CI: 15.0–43.0) across 13 studies. In contrast, the combined pooled proportion of MDR-TB occurrence for the other SSA regions was notably higher at 45.0% (95% CI: 32.0–57.0) across 23 studies. Despite this observed difference in estimates, the test for subgroup differences indicated that the variation between Ethiopia and the other SSA countries was not statistically significant (p = 0.10). There was existence of varying MDR-TB rates throughout the region. The highest proportion was recorded in a single study from Botswana (95%; 95% CI: 87–99), which must be recognized as an isolated, highly selected clinical cohort rather than a nationwide projection. Conversely, the lowest proportions were observed in cohorts from Equatorial Guinea (3%) and Kenya (4%) (Fig. 3).

Fig. 3.

Country-stratified forest plot showing pooled proportion of MDR-TB across Africa

Subgroup analysis by study design

To further dissect the source of between-study heterogeneity, a subgroup analysis was performed based on study design (Figure S28). The pooled proportions differed significantly across methodologies (p = 0.0091). The highest pooled proportion of MDR-TB was documented within cohort studies at 79.0% (95% CI: 52–100). Case-control yielded a pooled proportion of 37.0% (95% CI: 31–42), while cross-sectional designs had a pooled proportion of 34.0% (95% CI: 17–51).

Subgroup analysis by study setting

We also stratified data in this review by the study setting (Figure S29). Multicentre designs demonstrated a higher pooled proportion of 42.0% (95% CI: 27–57) compared to single-site hospital, which showed a pooled proportion of 35.0% (95% CI: 23–46).

Patient profiles and proportions among MDR-TB cases

To ensure sufficient data for pooling, we only analyzed patient characteristics if they appeared in multiple (≥2) publications (Table S4–S27). A total of 24 distinct demographic, clinical, socio-economic, behavioral, and treatment-related features met this criterion (Table 1). Forest plots illustrating the descriptive distribution of these traits are presented in Figures S1–S26. Traits limited to isolated, single reports were excluded from statistical pooling (Tables S28–S35). These include residing in peri-urban area, having tertiary education, self-medication, being unemployed, being self-employed, being a civil servant by occupation, being unmarried (single) and being married.

Table 1.

Pooled proportion of characteristics reported among MDR-TB cases

Category and Characteristics	No. of Studies	Population	MDR-TB cases	Proportion (95% CI)	I2 %	p value
Demographic features
Male	11	1,480	995	70 (64–75)	80.9	<0.001*
Female	2	181	302	60 (54–65)	0.0	0.586
Clinical Characteristics
Having HIV	31	5,826	1,743	34 (25–43)	99.3	0.000**
History of direct contact with TB patient	15	1,985	953	49 (38–60)	97.7	<0.001*
Previous TB history	2	1,401	501	37 (21–53)	97.4	<0.001*
No BCG vaccination	2	44	27	69 (41–97)	71.2	0.062
Being Malnourished (Baseline BMI < 18.5 kg/m2)	5	658	265	36 (18–54)	97.2	<0.001*
Socio-economic Characteristics
History of no formal education	13	1,397	370	28 (17–39)	95.1	<0.001*
Living in rural area	12	1,454	397	36 (24–48)	98.0	<0.001*
Having large family size (>5 members)	3	248	136	55 (48–62)	16.1	0.304
Being a Farmer and Day labourer	2	83	68	77 (53–100)	49.9	0.158
History of migration	2	509	198	32 (10–53)	94.6	<0.001*
Having Primary education	2	83	35	42 (32–53)	0.0	0.969
Residing in Urban area	2	480	253	53 (48–57)	0.0	0.468
Behavioral Characteristics
Alcohol use/Disorder	14	1,244	584	48 (34–62)	99.1	<0.001*
Smoking	12	926	348	35 (23–46)	95.2	<0.001*
History of imprisonment	6	896	128	15 (5–26)	91.4	<0.001*
Being a non-alcoholic	2	300	222	68 (33–100)	97.6	<0.001*
Consulting traditional healers	2	212	181	86 (81–90)	0.0	0.386
Treatment-related Characteristics
History of previous treatment (Acquired resistance)	28	4,059	2,208	63 (53–72)	99.2	0.000**
History of TB treatment failure	14	2,944	1,109	37 (21–52)	99.3	0.000**
Treatment interruption	6	923	363	39 (23–55)		<0.001*
Bad treatment compliance	4	371	253	69 (47–91)	98.3	<0.001*
Poor adherence to drug intake hours	4	350	89	28 (4–52)	95.3	<0.001*

Key: p-value ≤ 0.05 = Significant; ** = Strongly significant, * = Significant

Footnote: Proportions represent distribution within the diagnosed cases rather than relative causal risks

Demographic characteristics

Among demographic parameters, male patients accounted for a pooled proportion of 70.0% (95% CI: 64–75; p < 0.0001) of the evaluated cases.

Clinical features

Several clinical history parameters were commonly reported among MDR-TB patients. A history of direct contact with a known TB patient had a pooled proportion of 49.0% (95% CI: 38–60; p < 0.0001) among cases. A history of previous TB disease accounted for 37.0% (95% CI: 21–53; p < 0.0001). HIV co-infection was present in 34.0% (95% CI: 25–43; p < 0.0001) of the aggregated MDR-TB populations. Also, malnutrition (having baseline BMI < 18.5 kg/m²) was reported in 36.0% (95% CI: 18–54; p < 0.0001) of the cases.

Socio-economic characteristics

Socio-economic determinants played a notable role. Residing in an urban area accounted for a pooled proportion of 53.0% (95% CI: 48–57) among MDR-TB cases, whereas living in a rural area represented 36.0% (95% CI: 24–48; p < 0.0001). A lack of formal education (history of no formal education) was noted in 28.0% (95% CI: 17–39; p < 0.0001) of the patients. Having a large family size (>5 members) accounted for a pooled proportion of 55.0% (95% CI: 48–62).

Behavioral characteristics

Behavioral features reported among MDR-TB patients included the use of alcohol 48.0% (95% CI: 34–62; p < 0.0001), smoking 35.0% (95% CI: 23–46; p < 0.0001), and history of imprisonment 15.0% (95% CI: 5–26; p < 0.0001). Consultation with traditional healers showed a high pooled proportion of 86.0% (95% CI: 81–90) among assessed populations; however, this estimate was based on only two studies and cannot be generalized broadly.

Treatment-related characteristics

Treatment compliance and history were among the most frequently reported findings across the included studies. Bad treatment compliance had a pooled proportion of 69.0% (95% CI: 47–91; p < 0.0001), while treatment interruption was reported in 39.0% (95% CI: 23–55; p < 0.0001) of MDR-TB cases. Furthermore, a history of TB treatment failure accounted for a pooled proportion of 37.0% (95% CI: 21–52; p < 0.0001). However, it should be noted that the proportion among patients with a history of previous treatment (secondary/acquired resistance) was nearly twice that of treatment-naïve patients (primary resistance) (Table 1).

Meta-regression analysis

Meta-regression was performed to analyze potential sources of the wide between-study variation based on region, diagnostic method, and reported characteristics (Table 2). Crucially, because 36 studies revealed dozens of individual patient characteristics, the regression models are structurally underpowered and subject to severe statistical instability if interpreted causally. Many descriptive categories such as lack of BCG vaccination, consultation with traditional healers, and primary education history, contain data from only two studies, making their regression weights highly unstable. Consequently, the model parameterization was restricted to evaluating isolated exploratory trends. The analysis revealed that the country of study significantly influenced prevalence estimates. Specifically, studies from countries (South Africa: p = <0.0001 and Sudan: p = 0.007); conducted in hospital vs community settings (p = 0.003 vs 0.045) and diagnosed using Xpert (MTB/RIF) p = 0.009, showed statistically significant positive associations with higher MDR-TB prevalence. Regarding patient profiles, “previous TB disease history” and “history of no formal education” were a strong positive predictor (p = 0.001 vs 0.009). Given the high ratio of covariates to the 36 included studies, this meta-regression output should be interpreted cautiously as exploratory trends rather than definitive causal links.

Table 2.

Meta-regression for study features and proportions

Feature	Estimate	SE	z-value	p-value	95% CI (Lower)	95% CI (Upper)
(Intercept)	95.4644	19.9342	4.789	<0.001***	56.3941	134.5347
Country
Botswana	−25.4644	62.1691	−0.4096	0.682	−147.314	96.3849
Burundi	−35.4644	63.0159	−0.5628	0.574	−158.973	88.0445
Cameroon	26.5356	63.9922	0.4147	0.678	−98.8868	151.958
Central African Republic	−25.4644	63.2535	−0.4026	0.687	−149.439	98.5102
Equatorial Guinea	−87.4644	63.4587	−1.3783	0.168	−211.841	36.9123
eSwatini (formerly Swaziland)	78.5356	64.2495	1.2224	0.222	−47.3911	204.4624
Ethiopia	−22.5333	26.1011	−0.8633	0.388	−73.6905	28.6238
Ghana	2.441	46.5933	0.0524	0.958	−88.8801	93.7622
Kenya	−81.4644	64.6607	−1.2599	0.208	−208.197	45.2682
Namibia	21.5356	64.4903	0.3339	0.738	−104.863	147.9343
Nigeria	−6.408	33.3993	−0.1919	0.848	−71.8693	59.0534
South Africa	518.5356	67.2012	7.7162	<0.001***	386.8237	650.2476
Sudan	133.8149	49.2478	2.7172	0.007*	37.291	230.3389
Tanzania	36.7759	40.1149	0.9168	0.359	−41.8478	115.3996
Uganda	−35.9523	47.0992	−0.7633	0.445	−128.265	56.3604
Publication Year
2012	−0.7766	1.9725	−0.3937	0.6938	−4.6426	3.0894
2013	4.023	2.1656	1.8577	0.0632	−0.2215	8.2674
2015	−0.2962	2.0344	−0.1456	0.8843	−4.2835	3.6912
2016	−0.3539	2.1473	−0.1648	0.8691	−4.5626	3.8548
2017	1.5429	1.9743	0.7815	0.4345	−2.3266	5.4125
2018	−1.0235	1.948	−0.5254	0.5993	−4.8416	2.7946
2019	0.3849	1.9215	0.2003	0.8412	−3.3813	4.1511
2020	−0.2907	1.7231	−0.1687	0.866	−3.6679	3.0865
2021	0.1584	1.9694	0.0804	0.9359	−3.7016	4.0184
2022	20.9086	2110.1519	0.0099	0.9921	−4114.9131	4156.7304
2023	−0.7098	1.9162	−0.3704	0.7111	−4.4655	3.0459
2024	−2.5918	2.0286	−1.2777	0.2014	−6.5677	1.3841
Study setting
Hospital	3.8642	1.293	2.9887	0.003**	1.3301	6.3984
Community	3.1495	1.5724	2.003	0.045*	0.0677	6.2313
Diagnostic Method
Xpert (MTB/RIF)	2.5223	0.9604	2.6262	0.009**	0.6399	4.4047
Microscopy	1.9805	1.2753	1.5529	0.120	−0.5191	4.4801
LPA / Molecular	0.3554	0.8348	0.4258	0.670	−1.2808	1.9916
Culture/DST	0.018	1.1471	0.0157	0.988	−2.2303	2.2663
Risks Factor
Bad treatment compliance	−32.2953	35.622	−0.9066	0.365	−102.113	37.5226
Being a Farmer and Day labourer	−61.7657	45.9866	−1.3431	0.179	−151.898	28.3663
Being a Non-alcoholic	14.6749	46.5707	0.3151	0.753	−76.6021	105.9519
Being Female	−5.039	46.5791	−0.1082	0.914	−96.3323	86.2544
Being Male	−0.9635	27.4003	−0.0352	0.972	−54.6672	52.7402
Consultation of traditional healers	−5.017	46.3374	−0.1083	0.914	−95.8365	85.8026
Having Large Family size (>5 members)	−50.1451	39.4477	−1.2712	0.204	−127.461	27.1709
Having Primary Education	−78.11	45.9866	−1.6985	0.089	−168.242	12.022
History of direct contact with known TB patient	−34.0773	25.1427	−1.3554	0.175	−83.356	15.2014
History of drinking alcohol/disorder	−53.8106	25.4186	−2.117	0.034*	−103.63	−3.9911
History of migration	−0.4572	47.0873	−0.0097	0.992	−92.7467	91.8322
History of no formal education	−67.3494	25.8165	−2.6088	0.009**	−117.949	−16.75
History of previous TB treatment	−21.8833	22.984	−0.9521	0.341	−66.931	23.1644
History of prison/imprisonment	−74.5956	31.419	−2.3742	0.018*	−136.176	−13.0154
History of TB treatment failure	−20.1941	25.5751	−0.7896	0.430	−70.3203	29.9322
HIV	−41.245	22.6417	−1.8216	0.069	−85.622	3.1319
Living in rural area	−62.5335	26.2674	−2.3807	0.017*	−114.017	−11.0503
Malnutrition	−43.3298	33.198	−1.3052	0.192	−108.397	21.7371
No BCG vaccination	−81.9973	45.882	−1.7871	0.074.	−171.924	7.9297
Poor adherence to drugs intake hours	−73.2086	35.6036	−2.0562	0.040*	−142.99	−3.4269
Previous TB disease history	159.5534	49.4353	3.2275	0.001**	62.662	256.4447
Residing in Urban area	30.9558	47.0546	0.6579	0.511	−61.2694	123.1811
Smoking	−66.5762	26.2029	−2.5408	0.011*	−117.933	−15.2195
Treatment interruption	−36.6119	31.4306	−1.1648	0.244	−98.2148	24.9909

Key: SE = Standard Error; CI = Confidence Interval; *** = Very Strongly significant p-value; ** = Strongly significant p-value; * = Significant; LPA = Line probe Assay; DST = Drug Susceptibility Test

Sensitivity analysis, publication bias and trim-and-fill analysis

Our leave-one-out sensitivity analysis confirmed that the pooled proportion estimate of MDR-TB across the included study populations was generally stable (Fig. 4). Most individual exclusions resulted in minor fluctuations ranging from −1.37% to +2.34%. The notable exception was the removal of the study by Moore et al. (2015) [42], which caused a shift of −3.521% in the pooled sensitivity. Regarding publication bias, visual inspection of the funnel plot demonstrated clear asymmetry (Fig. 5). This concern was statistically confirmed by Egger’s regression test, which yielded a significant bias coefficient of 9.2870 (SE = 2.6925; p = 0.0015). While the trim-and-fill procedure did not impute further theoretical missing studies (Fig. 6), the highly significant Egger’s test indicates that publication bias is a valid concern that must be factored into the interpretation of these findings.

Fig. 4.

Sensitivity analysis of pooled prevalence estimates

Fig. 5.

Funnel plot of publication bias

Fig. 6.

Study’s trim-and-fill funnel plot

Discussion

This study comprehensively examined the demographic, clinical, socio-economic, behavioral, and treatment-related characteristics reported among patients with MDR-TB across Sub-Saharan Africa. Drawing on evidence from thirty-six eligible studies conducted in fifteen African countries, the review demonstrates that MDR-TB occurs across heterogeneous study populations with demographic, clinical, socio-economic, behavioral, and treatment-related characteristics commonly reported among affected patients. Given the substantial between-study heterogeneity observed, these findings should be interpreted as descriptive summaries of the included study populations rather than evidence of causal relationships.

Demographically, MDR-TB was commonly reported among young to middle-aged adults, reflecting the disease burden among economically active populations. Several studies reported peak MDR-TB prevalence within similar age ranges across Ethiopia [38, 53], Gabon and Tanzania [54, 55]. Male sex was repeatedly identified as the predominant demographic characteristic across most included studies, with men accounting for most reported cases. This trend aligns with evidence from Nigeria and Central African Republic, among others, and may reflect increased exposure to transmission-prone environments, occupational risks, and higher prevalence of behavioral factors such as smoking and alcohol consumption [39, 50]. Nonetheless, exceptions were observed, particularly in parts of Southwest Ethiopia, where higher MDR-TB incidence was reported among young adult females [56], suggesting that demographic patterns may vary by context, warranting localized investigations.

Clinical history and treatment-related behaviors form a highly interconnected continuum commonly reported among patients with MDR-TB. Rather than operating as isolated variables, previous TB infection, poor treatment compliance, interruption of therapy, and ultimate treatment failure frequently co-occur and likely reflect multiple stages within the treatment pathway where resistance may emerge or be sustained. While our meta-regression models indicated that a history of previous TB disease was associated with higher proportion estimates, this trend should be interpreted with caution due to the potential for overfitting in underpowered regression models. Clinically, these findings reflect long-standing insights showing that irregular or incomplete first-line therapy creates the biological conditions necessary for selecting resistant strains. Nevertheless, these observations are consistent with reports from multiple African settings indicating that inadequate or incomplete first-line therapy contributes to the emergence of drug resistance [17, 50, 53]. Suboptimal drug exposure, whether driven by missed doses, irregular timing, premature cessation, or poorly supervised regimens, creates the precise biological environment required for resistance selection [18, 25, 28]. Consequently, the distinction between clinical failure and behavioral non-adherence is often blurred, highlighting persistent systemic weaknesses in treatment delivery and follow-up across the region. Collectively, these findings underscore the importance of strengthening treatment quality, improving adherence support, ensuring uninterrupted drug supply, and reinforcing patient-centered Directly Observed Therapy (DOTS) programmes by tackling health systems gabs to reduce opportunities for the emergence and transmission of drug-resistant TB.

Close or household contact with TB patients was another commonly reported clinical characteristic, reflecting ongoing primary transmission of resistant strains within communities. Close contact with family members of patients with MDR-TB is a well-recognized factor for TB transmission and disease progression, often reflecting underlying household vulnerabilities such as malnutrition and overcrowding. A case–control study in Uganda among household contacts of MDR-TB patients identified malnutrition as an important determinant and recommended that TB programmes routinely screen MDR-TB patients and their household contacts for nutritional deficiencies [57]. Similarly, research conducted at a Cape Coast teaching hospital in Ghana found that sharing the same room with an MDR-TB patient was significantly more common among MDR-TB cases than controls, indicating that household crowding increases the likelihood of transmission [19]. The communal living patterns common in many African settings characterized by overcrowding, large household sizes, and frequent close contact, may facilitate sustained TB/MDR-TB transmission [36, 50]. Thus, these findings highlight the need for MDR-TB control strategies that address not only clinical management but also household-level social determinants such as poor housing and overcrowding. Several interventions have been recommended to reduce transmission among close household contacts and boost MDR-TB control. These include rapid diagnosis and effective treatment of the index case to reduce infectiousness [58, 59], systematic household contact tracing with symptom screening and rapid molecular testing [60, 61], and preventive therapy for eligible high-risk contacts after excluding active disease [62–64]. Additional environmental and supportive interventions, including improved ventilation, avoiding bed sharing, mask use, nutritional and adherence support, have been reported to further strengthen household MDR-TB prevention efforts [52, 57, 61, 65].

HIV co-infection was also commonly reported among MDR-TB patients, probably because immunosuppression increases the risk of progression from latent infection to active TB disease, while frequent health-system contact may also increase opportunities for exposure to resistant strains among susceptible individuals. This observation aligns with earlier studies conducted across Central and Eastern Africa [36, 53]. However, because this review synthesized pooled proportions rather than comparative effect estimates, the present findings should not be interpreted as demonstrating that HIV independently increases the risk of MDR-TB. Furthermore, although less frequently reported, lack of BCG vaccination was observed in a high pooled proportion of MDR-TB patients; however, this finding should be interpreted cautiously because it was derived from only two studies and may not be generalizable across Sub-Saharan Africa. Therefore, additional well-designed studies are needed before drawing firm conclusions regarding the relationship between BCG vaccination and MDR-TB.

Socio-economic characteristics were frequently reported among MDR-TB patients. Poverty, low educational attainment, and poor living conditions were recurrent themes across studies. Lack of formal education was commonly reported, suggesting that limited health literacy may contribute to delayed care-seeking, poor treatment adherence, and limited understanding of infection prevention measures [39]. Residential setting also appeared to influence the distribution of MDR-TB cases, with urban areas experiencing increased transmission due to overcrowding and informal settlements, while rural areas faced challenges related to low healthcare personnel deployment to rural communities, limited amenities/access to care that potentially cause delayed diagnosis and inconsistent drug supply. Although findings from some studies reported higher MDR-TB occurrence in rural settings [38, 66], de Dieu Longo et al. (2023) [50] observed more MDR-TB cases among peri-urban residents, suggesting that these relationships may vary by setting. Household size and living arrangements were also important, as larger households (>5 members) were commonly reported among MDR-TB patients, stressing the role of overcrowding and sustained intra-household transmission [67]. Occupational and mobility-related factors, including migration and farming or day labor, were also reported, likely due to unstable living conditions, interrupted treatment, and poor continuity of care.

Behavioral features represented another group of characteristics commonly reported among MDR-TB patients. Alcohol use and smoking were commonly reported among the included MDR-TB populations, reflecting their potential influence on immune function, treatment adherence, and social interactions in high-risk environments. These acts were in tandem with similar assertions previously reported to increasing MDR-TB incidences [36, 39, 53]. History of imprisonment was also commonly reported, stressing the role of overcrowded and poorly ventilated correctional facilities as reservoirs for MDR-TB transmission [17, 20]. Consultation with traditional healers was reported in a high pooled proportion of MDR-TB patients in some settings. However, this finding should be construed with care because it was derived from only two studies. Although deferred presentation to biomedical healthcare services may contribute to delayed diagnosis and treatment initiation [33], additional evidence is needed before drawing firm conclusions regarding the role of traditional healer consultation in MDR-TB development across Sub-Saharan Africa

Collectively, these findings highlights the urgent need to strengthen DOTS, enhance patient-centered support mechanisms, ensure uninterrupted drug supply chains, and proactively manage treatment side effects. The frequent reporting of previous TB treatment, treatment failure, and poor adherence across the included studies underscores the importance of treatment quality, continuity of care, and broader health-system performance in MDR-TB control. Addressing these interconnected treatment and health-system-related challenges is likely to strengthen MDR-TB prevention and control efforts across Africa.

Several critical methodological limitations must be considered when evaluating this meta-analysis. First, extreme between-study variation (I² = 99%) was present across almost all categories, driven by combining data from diverse clinical networks, referral centers, and regional survey methods. The pooled estimates must therefore be interpreted as descriptive summaries of the included literature rather than a baseline prevalence for Sub-Saharan Africa. Second, the meta-regression models are underpowered and subject to statistical instability due to the high number of covariates evaluated relative to the 36 included studies. Third, specific parameters such as a lack of BCG vaccination history and consultation with traditional healers, were evaluated in only two studies, preventing broad generalizability across the subcontinent. Fourth, data collection was geographically concentrated in a few countries, such as Ethiopia, Nigeria, and Tanzania, leaving North, Central and Francophone West Africa underrepresented. Finally, the highly significant Egger’s test (p = 0.0015) indicates the presence of potential publication bias, meaning that smaller studies with null findings may be missing from the published literature.

Conclusion

This systematic review and meta-analysis synthesized the demographic, clinical, socio-economic, behavioral, and treatment-related characteristics reported among MDR-TB patients across Sub-Saharan Africa. The overall pooled proportion of MDR-TB cases across the literature was 39.0%. This figure reflects the highly selected, heterogeneous study populations included in the meta-analysis rather than a true continental prevalence; however, when combined with the frequent reporting of poor treatment adherence (69%) and previous treatment failure (37%), it underscores the critical need for health system reform. Similarly, consultation with traditional healers (86%) should be interpreted cautiously because this finding was derived from only two studies, while alcohol use (48%) was commonly reported among MDR-TB patients across the included literature. Ultimately, achieving the WHO End TB targets by 2035 will require integrated, patient-centered care, strengthened surveillance systems, and interventions that address both health-system weaknesses and the broader social determinants of tuberculosis.

Recommendations

1. Broaden TB Health Policy: Policies should prioritize sustaining free Directly Observed Treatment, Short-course (DOTS) while actively reducing out-of-pocket patient expenditures. Evidence from successful SSA interventions demonstrates that formally integrating traditional healers into TB control frameworks may improve early case detection and referral pathways. Programs that train traditional health practitioners to identify TB symptoms and refer patients to biomedical clinics have successfully reduced diagnostic delays and improved early case detection [68, 69]. Although consultation with traditional healers was reported in a high pooled proportion of MDR-TB patients in this review, this finding was based on only two studies and should therefore be interpreted carefully. Nevertheless, strengthening collaboration between traditional and biomedical healthcare providers may enhance timely diagnosis and referral.

2. Strengthen Treatment Adherence via Cost-Effective Models: National TB programs must pivot from passive, facility-based care to Community-based DOTS (CB-DOTS) and patient-centered support mechanisms. Economic evaluations across SSA consistently show that CB-DOTS is highly cost-effective compared to standard care, as it drastically lowers the catastrophic travel and time costs borne by patients. Furthermore, implementing socio-economic safety nets, such as decentralized drug delivery, nutritional support, or conditional cash transfers, has been shown to improve treatment adherence and reduce financial barriers to care. Given the frequent reporting of treatment interruption (39%) and poor treatment adherence (69%) among MDR-TB patients in this review, strengthening these interventions may improve treatment continuity and overall program performance.

3. Enhance Contact Tracing: The frequent reporting of household contact with TB patients (49%) necessitates rigorous contact tracing and screening protocols for families of known TB patients to interrupt primary transmission chains.

4. Address Co-morbidities and Behavior: Integrated care models that simultaneously address HIV co-infection, alcohol use disorders, and smoking cessation should be strengthened, as these factors or characteristics were commonly reported among MDR-TB patients and addressing them is likely to improve overall TB treatment outcomes.

5. Targeted Surveillance in High-Risk Groups: Surveillance and diagnostic resources should be prioritized for population groups frequently represented in the included studies, specifically males and those with a history of imprisonment or previous TB treatment, while recognizing that these findings are descriptive and should not be interpreted as evidence of independent causal danger.

6. Expand Research: Future research should prioritize under-studied regions in Central and West Africa to correct geographic imbalances and provide a more representative continental picture of the MDR-TB landscape.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

TB

Tuberculosis

MDR-TB

Multidrug-resistant tuberculosis

AJOL

African Journals Online

RR-TB

Rifampicin resistant tuberculosis

XDR-TB

Extensively drug-resistant tuberculosis

DOTS

Directly Observed Therapy, Short-course

HIV

Human Immunodeficiency Virus

NOS

Newcastle-Ottawa Scale

GLMM

Generalized linear mixed model

Author contribution

Conceptualization: M.Z.P., and A.O.; Methodology: M.Z.P., and A.O.; Validation: M.Z.P., A.O., S.T., S.M.B., and W.K.A.; Formal analysis: A.O. and M.Z.P.; Investigation: M.Z.P., A.O., and S.T.; Resources: M.Z.P., A.O., S.T., and S.M.B.; Data curation: M.Z.P., A.O., S.T., S.M.B., and W.K.A; Writing–original draft preparation: M.Z.P., and A.O.; Writing review and editing: A.O., M.Z.P., S.T., S.M.B., and W.K.A; Visualization: A.O., and M.Z.P.; Supervision: M.Z.P., and A.O.; Project administration: M.Z.P., and A.O.; Funding acquisition: M.Z.P., and A.O. All authors have read and agreed to the published version of the manuscript.

Funding

This work was done without any external funding aside from that of the authors.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files. The study drew information from publicly available domains.

Declarations

Ethics approval and consent to participate

This study did not require ethical approval or participant consent as it utilized secondary data drawn from the public domain. The study protocol was registered with PROSPERO systematic review and meta-analysis database (ID No: CRD420251185459).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.