Performance of active drug safety monitoring and management for pharmacovigilance strengthening in Ethiopia, experience from a multidrug-resistant tuberculosis management program: a cross-sectional study

Abstract

Objective

The study aimed to assess the performance of active drug safety monitoring and management (aDSM) in the context of multidrug-resistant and rifampicin-resistant tuberculosis (MDR/RR-TB) treatment programmes to generate insights that could support potential scale-up efforts to strengthen the pharmacovigilance (PV) system in Ethiopia and beyond.

Design

A cross-sectional study design was employed. The records of patients with MDR/RR-TB enrolled and treated at selected hospitals during April 2018 through February 2023 were reviewed. Data were collected using a structured data extraction form, and Research electronic Data Capture (REDCap) was used for data entry and management. The data were analysed using SPSS version 25, employing descriptive statistics.

Setting

The study was carried out at ten (10) public hospitals in Ethiopia.

Participants

A total of 1377 patients with MDR/RR-TB were included from 10 hospitals.

Results

Out of the 1377 patients, the majority (59.3%) were males, and 39.9% were new to anti-TB treatment. Deviations from the schedule or missed safety monitoring and other investigations were observed. No records were available for monitoring adverse events (AEs) in 94 (6.8%) of the patients during their treatment follow-up. Overall, 455 (33.0%) patients experienced at least one AE during the treatment period. A total of 774 AE records were identified, of which 134 (17.3%) and 225 (29.1%) were serious adverse events (SAEs) and adverse events of special interest (AESI), respectively. Only 135 (17.4%) of all identified AEs were reported to the PV Centre, of which 70 (51.9%) and 48 (35.6%) were AESI and SAE, respectively. However, the trend of AE detection and reporting showed improvement over the surveyed aDSM implementation period.

Conclusion

Despite the implementation of aDSM, AE detection and reporting remained suboptimal in the MDR/RR-TB treatment cohort. Therefore, routine aDSM needs to be strengthened and consistently implemented to improve the national pharmacovigilance system and enhance patient safety monitoring and management.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Uses of real-world programmatic data reflect the actual implementation of active drug safety monitoring and management under routine conditions, providing valuable insights for programme improvements.

• Multiple sites study involving diverse patients from various healthcare settings enhances the credibility of the findings to inform decision-making.

• This study used a substantial sample size, which enhances the robustness and generalisability of the results.

• Limitations include retrospective design, incomplete or missing data, and the exclusive use of descriptive analysis.

Introduction

Multidrug-resistant and rifampicin-resistant tuberculosis (MDR/RR-TB) remains a significant challenge in tuberculosis (TB) treatment. Globally, the burden of MDR/RR-TB remains stable. In 2023, an estimated 3.2% of new TB cases and 16% of previously treated cases were affected by MDR/RR-TB. An estimated 410 000 people worldwide developed MDR/RR-TB annually.¹ Ethiopia is one of the 30 countries with high TB and TB/HIV burdens. It has recently transitioned out of 30 high MDR-TB/RR-TB countries.² In Ethiopia, the prevalence of RR-TB is 1.1% among new and 7.5% among previously treated TB cases, respectively, while the MDR-TB prevalence is 1.0% among new and 6.5% among previously treated TB patients.³

The management of MDR/RR-TB requires the concurrent use of many and new drugs,⁴ and medication-related safety issues are a major concern.⁵ Most patients with MDR/RR-TB experienced at least one adverse event (AE), and a recent study has shown that two-thirds of such patients have stopped at least one medicine temporarily or permanently as a result of AEs.⁶ The most common AEs reported in various studies include gastrointestinal disturbance, peripheral neuropathy, hypothyroidism, psychiatric symptoms and hypokalaemia.⁷

Following the introduction of new and repurposed anti-TB drugs, the WHO launched the implementation of active drug safety monitoring and management (aDSM) in the MDR/RR-TB management programme.⁸ The aDSM entails active and systematic clinical and laboratory assessment of patients while on treatment and applies to patients on therapy with new anti-TB drugs, novel MDR/RR-TB regimens, or extensively drug-resistant tuberculosis (XDR-TB) regimens.³ Continuous AE reporting is instrumental in ensuring patient safety. The success or failure of any AE monitoring depends on detecting suspected cases.⁹ Early detection and management of AEs help to reduce hospital admissions and save a substantial amount of financial and human resources that could be spent on patient treatment.¹⁰

Cognisant of this, the Ethiopian National TB Program integrated the aDSM implementation guide into the national drug-resistant tuberculosis (DR-TB) treatment guidelines, and the AE report has been sent to national pharmacovigilance centres in accordance with aDSM requirements since its initiation in 2018.^{11 12} However, the extent and quality of detecting and reporting MDR/RR-TB-related AEs were not well studied after aDSM implementation.

In Ethiopia, MDR/RR-TB regimen-related AEs are known to occur frequently.¹³ A recent study reported that 92% of patients developed at least one AE during MDR/RR-TB treatment, and 19.2% of these patients developed a severe AE, leading to the suspension of the suspected medication.¹⁴ However, the detection and reporting of AEs is believed to lag behind the true incidence. According to the Ethiopian Food & Drug Authority (EFDA) report, only 392 adverse events were received for the 2017–2020 cohort of 2665 patients with MDR/RR-TB.¹⁵ Ethiopia, focusing on the intermediate package, has been implementing the aDSM since 2018; however, how well the programme is functioning is not yet well understood. Hence, this study was intended to assess the aDSM implementation status and contribution during the treatment of DR-TB in Ethiopia for possible scale-up in strengthening the national pharmacovigilance (PV) system.

Methods

Study setting

In Ethiopia, all newly diagnosed patients with MDR/RR-TB start treatment at one of the 60 treatment initiation centres (TICs). We selected the 10 TICs that provide treatment and services to the highest number of TB patients. Together, these TICs provided treatment to 1396 (39.9%) of all patients with MDR/RR-TB (3500) in the country. Therefore, the study was conducted at All Africa Leprosy, Tuberculosis and Rehabilitation Training Centre (ALERT), Bishoftu, Boru Meda, Dil-Chora, Adama, Gondar, Shenen Gibe, St Peter, Tulu Bollo and Yirgalem Hospitals.

Study design

A descriptive, hospital-based, cross-sectional study was employed using the medical records of patients with MDR/RR-TB who were enrolled for treatment from April 2018 to February 2023.

Study participants

All patients with MDR/RR/XDR-TB who were treated with bedaquiline/delamanid-containing or injectable shorter regimens for at least 1 month from April 2018 to February 2023 at the study sites were included in the study (table 2). Patients who were on follow-up for treatment with a conventional or previous longer regimen during the period were excluded. Under programmatic TB management, a person with signs and symptoms corresponding to TB is examined clinically and radiologically along with at least one of the laboratory tests, including smear microscopy, culture, rapid molecular tests for the detection of Mycobacterium tuberculosis and anti-TB drug resistance via the Xpert Mycobacterium tuberculosis/rifampicin (MTB/RIF) test, first-line Line Probe Assay (LPA) or second-line LPA, sputum or phenotypic drug susceptibility tests (DSTs). The patients will also be screened for underlying comorbidities that will be used as a baseline during the treatment follow-up.¹⁶

Patient monitoring during treatment

The National Tuberculosis Control Program has been deploying a clinic-based ambulatory care model since 2013 for programmatic management of DR-TB (PMDT) services. However, patients with either serious medical or social issues may be admitted to referral centres. DR-TB patient monitoring throughout treatment is implemented in line with the national PMDT treatment monitoring plan.¹²

Per standard of care, hospitalised patients with MDR/RR-TB are evaluated for clinical signs and symptoms of any adverse events daily, and patients who start treatment as outpatients will have a weekly clinical and adherence assessment until they become clinically stable for at least the first 2 to 4 weeks and then monthly until the end of treatment. Laboratory examinations, including sputum smear and culture, liver and renal function tests, electrocardiogram (ECG), are scheduled every month during the intensive phase of treatment, irrespective of inpatient or outpatient treatment. The ECG monitoring continues monthly throughout the continuous phase of treatment, and other investigations are performed when indicated.^{12 16}

Treatment regimens

The treatment regimen options that have been used in the management of MDR/RR-TB during 2018‒2023 in Ethiopia were as follows:

All-oral shorter regimen lasts 9‒11 months and comprises bedaquiline (Bdq) (for 6 months) in combination with levofloxacin (Lfx), clofazimine (Cfz), pyrazinamide (Z), ethambutol (E), high-dose isoniazid (Hh) and ethionamide (Eto) in the intensive phase for 4‒6 months, followed by treatment with Lfx, Cfz, Z and E for 5 months. It is recommended for patients with MDR/RR-TB who have not been exposed to second-line drugs for more than 1 month and have no history of fluoroquinolone resistance.^{16 17}

Full oral longer regimen: This regimen lasts 18‒20 months. It comprises Bdq (for 6 months), Lfx/moxifloxacin (Mfx), linezolid (Lzd), Cfz and cycloserine (Cs). This regimen is recommended for patients with MDR/RR-TB who are not eligible for a shorter duration.^{16 17}

Individualised longer regimen is constructed for eligible patients by adding one or more medicines from group C (E, delamanid (Dlm), Z, imipenem-cilastatin (Ipm-Cln)/meropenem (Mpm), amikacin (Am)/streptomycin (S), Eto/protonamide (Pto) and para-aminosalicylic acid (PAS)) than the recommended, if one or more drugs in groups A (Lfx/Mfx, Bdq, Lzd) and B (Cfz, Cs) are suspected for resistance or contraindicated on the basis of complications that need individualisation.¹⁶

Isoniazid monotherapy: 6 months of isoniazid preventive therapy (IPT) should be offered to HIV-exposed infants receiving nevirapine-based prophylaxis who are also exposed to a pulmonary TB case.¹⁶

Injectable containing a shorter regimen refers to a course of treatment for MDR/RR-TB lasting 9–12 months and is composed of Mfx, Cfz, Z and E throughout, supplemented by kanamycin (km), Pto and Hh in the intensive phase for 4‒6 months.¹⁸

Adverse event detection and reporting

All healthcare providers involved in the management of DR-TB screen the patients for symptoms of AEs during daily direct observation of treatment and scheduled follow-up visits. Suspected AEs should be reported to the pharmacovigilance centre using the appropriate form as soon as possible after all relevant information is compiled.¹⁶

The reporting of AEs is vital in the assessment of drug safety profiles and for informing regulatory decisions.¹⁹ Ethiopia implements an active pharmacovigilance or intermediate package of aDSM in which serious adverse events (SAEs), adverse events of special interest (AESIs), AEs leading to treatment discontinuation or changes in drug dosage, and AEs that are judged as clinically significant by the clinician are reportable. ⁶Regarding the timeline and tools of reporting, SAEs should be reported within 48 hours to the EFDA via the online reporting system, Med-safety mobile application or yellow form, whereas non-serious AEs are routinely reported using the AE line-listing form. To be considered complete, the AE report should contain at least four variables: patient identifier information, suspected drug information, suspected AE and reporter identifier information.¹⁶

Active drug safety monitoring and management

aDSM is a systematic approach for active and real-time monitoring, detection, management and reporting of AEs associated with the use of new and repurposed anti-TB drugs, particularly in the treatment of DR-TB. The aDSM contains three packages: the core package, which focuses on the active detection and reporting of SAEs; the intermediate package, which adds AESIs and other medically important AEs; and the advanced package, which involves comprehensive detection and documentation of all AEs. The choice of these packages depends on the country’s healthcare system capacity and resource availability.⁸

Sampling and sample size

The total number of patients treated with MDR/RR-TB regimens in Ethiopia during the study period was 3500, as obtained from the Ministry of Health TB database. The sample size was calculated using Open-Epi version 3 (https://www.openepi.com) for a single population proportion with a prespecified margin of error. Assuming a prevalence of AE of 41.8%,²⁰ with a 95% confidence level and a 5% marginal error, 338 participants would need to be included. However, all patients (1377) who fulfilled the eligibility criteria from the 10 selected hospitals were included in the study. This census approach was intended to reduce the impact of incomplete/missing data,²¹ and enhance the precision of rare-event estimation.²²

Data collection

A structured data extraction form, which was adopted from the aDSM implementation guidelines, was used to extract data on clinical signs and symptoms, laboratory results, and AE records for each patient from medical charts, registration logbooks, and patient files. Data were collected by trained nurses working in the MDR-TB clinics. The Research Electronic Data Capture (REDCap) software (https://project-redcap.org) was used for data collection and management. All procedures were closely supervised by the investigators to ensure the accuracy, consistency, completeness and validity of the data.

Data analysis

After cleaning and validation, the data were exported from REDCap to SPSS version 25.0 for analysis. Descriptive statistics were used to characterise the data. Frequencies and percentages were computed to present the findings, and the generated results were summarised and illustrated in tables and graphs. The normality of continuous variables was assessed using the Shapiro–Wilk test, and visual inspection of histograms and line plots to guide the choice of statistical methods. Accordingly, the time to report AEs was not normally distributed; therefore, the median and IQR were used to summarise these variables.

Definition of terms

Adverse event

Any untoward medical occurrence that may be present during treatment with a medicine but does not necessarily have a causal relationship with this treatment.²¹

AEs of special interest

All AEs for which the monitoring programme is specifically sensitised to report, regardless of their seriousness, severity or causal relationship with TB treatment. These include peripheral neuropathy, myelosuppression (anaemia, thrombocytopenia or neutropenia), prolonged QT interval, psychiatric disorders, optic nerve disorders (optic neuritis), hepatitis, impaired hearing, acute kidney injury, hypokalaemia and hypothyroidism.¹⁶

Serious adverse events

Any untoward medical occurrence that at any dose results in death, is life-threatening, requires hospital admission or prolongation of existing hospital stay, results in persistent or significant disability or incapacity or a congenital abnormality.²³

Patient and public involvement

No patient or public involvement occurred in this study; all data were obtained retrospectively from existing medical records.

Results

Patients’ baseline and medical information

Out of 1377 patients with MDR/RR-TB, the majority (59.3%) were males. More than half (65.2%) of the patients had a Body Mass Index (BMI) below the normal range. At enrolment, only 550 (39.9%) patients were new to anti-TB treatment, while the rest had used either or both first-line drugs (FLDs) and second-line drugs (SLDs). A large proportion of patients (51.0%) were initiated on a full oral longer regimen, whereas a few, 23 (1.7%), were initiated on other regimens. Of the 1377 patients, 904 (65.6%) had known treatment outcomes, whereas treatment outcome had not been assigned for 473 patients. The majority, 421 (89%) of those for whom the outcome was not yet declared, were on treatment at the time of data collection. Among the patients with known outcomes, 378 (41.8%) were cured, and 360 (39.8%) completed treatment, which implies an 81.6% treatment success rate. Other outcomes were death (10.4%), lost to follow-up (LTFU) (6.7%) and failure (1.2%) (table 1).

Table 1. Characteristics of patients enrolled on new drugs or shorter MDR/RR-TB treatment regimens during 2018‒2023 (n=1377).

Variable category	Variables	Frequency (n)	Percentage (%)
Age groups (years)	Less than 15	79	5.8
	15 to 24	299	21.7
	25 to 34	450	32.6
	35 to 44	250	18.2
	Greater than 44	256	18.6
	Missing	43	3.1
Sex	Female	561	40.7
	Male	816	59.3
Body Mass Index (BMI), (kg/m²)	Less than 18.5	898	65.2
	18.5 to 24.9	402	29.2
	Greater than 24.9	27	2.0
	Missing	50	3.6
Presence of any comorbidity	Yes	288	20.9
	No	1083	78.6
	Missing	6	0.5
History of anti-TB drug exposure	No previous history (new patient)	550	39.9
	Yes: first-line TB drugs	707	51.3
	Yes: second-line TB drugs	39	2.8
	Yes: both first-line and second-line drugs	78	5.7
	Missing	3	0.3
Treatment regimen	All oral shorter regimen	349	25.3
	Individualised regimen	82	6.0
	Full oral longer regimen	702	51.0
	Injectable shorter regimen	213	15.5
	Other treatment regimen	23	1.7
	Missing	8	0.5
Known treatment outcome (904, 65.6%)	Treatment completed	360	39.8
	Cured	378	41.8
	Died	94	10.4
	Failure	11	1.3
	Lost to follow-up	61	6.7
Treatment outcome not assigned (473, 34.4%)	Still on treatment	421	30.6
	Unknown*	52	3.8

^*Unknown treatment outcome=it was not recorded in any of the documents from which data were extracted.

MDR/RR-TB, multidrug-resistant and rifampicin-resistant tuberculosis; TB, tuberculosis.

Clinical and laboratory safety monitoring at baseline and during treatment

We assessed whether clinical monitoring and laboratory tests were conducted according to the schedules recommended in the national guidelines. As shown in table 2, among the assessments required at the initiation of the anti-MDR/RR-TB regimen, weight measurement was performed for most patients (1346; 97.7%), followed by vital sign assessments (1335; 96.9%), liver function tests (1317; 95.6%), complete blood count (CBC) (1315; 95.5%) and renal function tests (urea/creatinine) (1288; 93.5%). Thyroid-stimulating hormone (TSH) testing was conducted for only 455 (33.0%) patients at baseline (table 2).

Table 2. Clinical and laboratory safety monitoring of novel regimens, or bedaquiline, or delamanid-containing anti-MDR/RR-TB medications.

Clinical and laboratory safety monitoring (n=1377)	Follow-up monitoring at baseline	Complete follow-up monitoring every month for 6 months	Complete clinical monitoring until the end of treatment	No follow-up monitoring during the treatment period
Clinical safety monitoring
Vital signs	1335 (96.9%)	736 (53.4%)	469 (34.1%)	97 (7.0%)
Peripheral neuropathy screening	1161 (84.3%)	620 (45.0%)	420 (30.5%)	263 (19.1%)
Audiometry	622 (45.2%)	118 (8.6%)	Not applicable	342 (24.8%)
Visual acuity and colour screening	1166 (84.7%)	468 (34.0%)	321 (23.3%)	260 (18.9%)
Assessment of AEs		781 (56.7%)	541 (39.3%)	94 (6.8%)
Weight	1346 (97.7%)	824 (59.8%)	498 (36.2%)	49 (3.6%)
Laboratory/radiology testing
ECG	1130 (82.1%)	246 (17.9%)	Not applicable	533 (38.7%)
Full blood count	1315 (95.5%)	397 (28.8%)	390 (28.3%)	212 (15.4%)
Urea, creatinine	1288 (93.5%)	349 (25.3%)	Not applicable	270 (19.6%)
Serum electrolytes*	1219 (88.5%)	395 (28.7%)	Not applicable	315 (22.9%)
Liver function tests*	1317 (95.6%)	416 (30.2%)	260 (18.9%)	242 (17.6%)
TSH	455 (33.0%)	139 (10.1%)†	82 (6.0%)	1230 (89.3%)
Chest X-ray	1188 (86.3%)	633 (46.0%)‡	Not applicable	661 (48.0%)

^*Number of patients who had at least one of these tests.

^†Number of patients tested at month 3.

^‡Number of patients tested at month 6.

AEs, adverse events; MDR/RR-TB, multidrug-resistant and rifampicin-resistant tuberculosis; SLI, second-line injectables; TSH, thyroid-stimulating hormone.

During the first 6 months of treatment, AE and vital sign monitoring were completed for 781 (56.7%) and 736 (53.4%) patients, respectively. Other assessments were conducted for fewer than half of the patients: chest X-ray in 633 (46.0%), peripheral neuropathy (PN) screening in 620 (45.0%), vision assessment in 468 (34.0%) and liver function tests in 416 (30.2%). From 6 months to the end of treatment, the frequency of these assessments further declined; thus, AE monitoring was performed in 542 (39.3%) patients, vital signs in 469 (34.1%), PN screening in 420 (30.5%), vision assessment in 320 (23.3%), CBC in 390 (28.3%), liver function tests in 260 (18.9%) and TSH testing in 82 (6.0%). Consequently, a significant proportion of patients missed at least one scheduled monitoring or assessment for each parameter during their first 6 months of treatment course (table 2).

Additionally, a considerable proportion of patients with MDR/RR-TB who were on treatment did not receive essential clinical or laboratory monitoring during their follow-up period. Specifically, TSH, X-ray, ECG, PN, visual/vision and AE monitoring were not performed at any point during the follow-up period for 1230 (89.3%), 661 (48.0%), 533 (38.7%), 263 (19.1%), 260 (18.9%) and 94 (6.8%) patients, respectively (table 2).

Adverse event detection and management

A total of 774 AEs were identified in 455 patients during treatment; 134 (17.3%) and 225 (29.1%) of these were SAE and AESI, respectively. Thus, 455 (33%) patients experienced at least one AE, of whom 85 (18.7%) patients had at least one SAE. Of the 774 AEs, 135 (17.4%) were reported to the National Pharmacovigilance Center (NPC), of which 51.9% and 35.6% were AESIs and SAEs, respectively. Among 134 SAE and 225 AESI cases, only 35.8% and 31.1%, respectively, were reported to NPC. The median times to report SAEs and AESIs were 1.0 (IQR: 0–14) and 3.0 (IQR: 0–22) days, respectively (table 3).

Table 3. Detection, reporting and management of adverse events.

Variables	Denominator	Frequency (n)	Percentage (%)
Prevalence of adverse events (AEs)
Number of AEs registered over the treatment period	Not applicable	774
Number of SAEs registered over the treatment period	774	134	17.3
Number of AESI registered over the treatment period	774	225	29.1
Number of patients who experienced at least one AE	1377	455	33.0
Number of patients who experienced at least one SAE	455	85	18.7
AE reported to National Pharmacovigilance Center (NPC)
Any AE reported to NPC	774	135	17.4
Proportion of SAEs reported out of all reported AEs	135	48	35.6
SAE reported from total SAE cases	134	48	35.8
Number of AESI reported from 135 reported AEs	135	70	51.9
Number of AESI reported from total AESI cases	225	70	31.1
Completeness of reported AEs	135	110	81.5
Management undertaken in response to AEs
The proportion of AE managed (out of 774)	774	549	70.9
Auxiliary treatment is given (out of the managed AEs)	549	346	63.0
Withdrawal of suspected anti-TB drug	549	165	30.1
Dose of suspected anti-TB drug modified (n=549)	549	30	5.5
Regimen changed	549	8	1.5
Time to event and timeline for reporting
Time to report any AEs (in days)	Median (IQR): 9.0 (0–28)
Time to report SAE (in days)	Median (IQR): 1.0 (0–14)
Time to report AESI (in days)	Median (IQR): 3.0 (0–22)

Completeness of reported AEs = included patient information, descriptions of AEs, drugs administered, including concomitant, and reporter information.

^*(%) calculated by dividing the nominator/frequency (column right to the denominator) by the denominator × 100.

AEs, adverse events; AESI, adverse events of special interest; NPC, National Pharmacovigilance Center; SAEs, serious adverse events; TB, tuberculosis.

Regarding management undertaken, more than two-thirds of the AEs (70.9%) necessitated intervention, and the majority were managed by auxiliary drugs (63.0%). In total, 22.2% of the AEs required temporary or permanent withdrawal of the suspected TB drugs, 4.0% required dose modification and 1.2% led to a change in the regimen (table 3).

The AE detection and reporting status was assessed across study sites. Accordingly, ALERT Hospital reported that of 264 patients, 167 (63.3%) experienced AEs, and followed by Bishoftu, 23 (46.0%), and Tulu Bollo, 26 (45.6%). In addition, the total number of individual AEs recorded and reported to the NPC was evaluated based on study sites. As the analysis showed, St. Peter Hospital detected a greater number of AEs, 267 (34.5%), followed by ALERT Hospital, 263 (34%), compared with others. Regarding reporting, out of the total reported AEs, 41.5% and 26.7% were reported by St Peter and ALERT Hospitals, respectively. In addition, the proportion of AEs reported out of detected cases at each site was evaluated. Accordingly, Dil-chora Hospital reported a higher proportion (55.6%) of recorded AEs. Boru Meda and Tulu Bollo Hospitals did not report any AEs at all (table 4).

Table 4. Adverse event detection and reporting status per study site.

Name of the study sites (hospitals)	Number of patients included per site	Patients developed AE per site n (%)	All AE cases detected per site (n=774) n (%)*	Number of AEs reported to NPC per site n (%)	Proportion of reported all AEs per sites (n=135) (%)
St. Peter Hospital	307	127 (41.4)	267 (34.5)	56 (21.0)	41.5
ALERT Hospital	264	167 (63.3)	263 (34.0)	36 (13.7)	26.7
Bishoftu Hospital	50	23 (46.0)	61 (7.9)	11 (18.0)	8.1
Gondar Hospital	166	39 (23.5)	50 (6.5)	7 (14.0)	5.2
Tulu Bollo Hospital	57	26 (45.6)	43 (5.6)	-	-
Dil-chora Hospital	140	19 (13.6)	27 (3.5)	15 (55.6)	11.1
Shenen Gibe Hospital	55	22 (40.0)	27 (3.5)	1 (3.7)	0.7
Adama Hospital	126	21 (16.7)	23 (3.0)	7 (30.7)	5.2
Boru Meda Hospital	82	7 (8.5)	8 (1.0)	-	-
Yirgalem Hospital	130	4 (3.1)	5 (0.6)	2 (40.0)	1.5
Total	1377	455	774 (100)	135	100

^*All AE cases detected at each site were greater than the number of patients who experienced AEs because each patient can develop more than one AE at a time or during the period.

AEs, adverse events; NPC, National Pharmacovigilance Center.

The occurrence of AEs was analysed by regimen type and treatment outcomes. A higher proportion of AEs was observed among patients receiving a shorter treatment regimen (STR) containing an injectable drug (36.9%) and the new longer regimen (35.0%) compared with other regimen types. With respect to treatment outcomes, AEs were more frequently reported among patients who completed treatment (47.0%) (table 5).

Table 5. Distribution of adverse events by regimen types and treatment outcomes.

Variable	Category	Frequency	Denominator	Percentage
Regimen type	Full oral STR	107	350	30.6
	New longer	247	706	35.0
	Individualised	19	82	23.2
	STR injectable	79	214	36.9
	Others	3	23	13.0
Known treatment outcomes	Completed	170	362	47.0
	Cured	105	379	27.7
	Died	24	94	25.5
	LTFU	16	61	26.2
	Failure	1	11	9.1
Treatment outcome not assigned	Still on treatment	130	421	30.9
	Uknown	9	52	17.3

LTFU, lost to follow-up; STR, shorter treatment regimen.

Among the 774 AEs identified, gastrointestinal tract (GIT) symptoms (37.0%), peripheral neuropathy (12.0%), vision impairment (10.1%), central nervous system (CNS) disorders (8.3%) and arthralgia (7.0%) were common (table 5). More than one-fourth (225 (29.1%)) of the observed AE cases were classified as AESIs and included peripheral neuropathy (n=93), psychiatric issues (n=19), optic nerve disorders (n=38), myelosuppression (n=21), ototoxicity (n=18), prolonged QT interval (n=6), hepatitis (n=21), hypokalaemia (n=4) and acute renal failure (n=5). Among the 134 SAEs, the most commonly recorded were vision impairment (14.9%), GIT (13.4%), peripheral neuropathy (12.7%), hepatotoxicity (11.9%) and CNS-related disorders (11.2%) (table 6).

Table 6. Frequency distribution of adverse events by type and seriousness.

Types of adverse events (AEs)	Frequency of AE n (%)	Frequency of SAEs n (%)
GIT-related symptoms	286 (37.0)	18 (13.4)
Peripheral neuropathy	93 (12.0)	17 (12.7)
Vision impairment/optic neuritis	78 (10.1)	20 (14.9)
CNS-related disorders	64 (8.3)	15 (11.2)
Arthralgia/arthritis/joint pain	54 (7.0)	7 (5.2)
Hearing loss/tinnitus	32 (4.1)	11 (8.2)
Skin-related problems	30 (3.9)	-
Loss of appetite	28 (3.6)	-
Blood disorders	25 (3.2)	11 (8.2)
Drug-induced hepatitis/hepatotoxicity	21 (2.7	16 (11.9)
Fatigue	19 (2.5)	-
Cardiovascular disorder	11 (1.4)	9 (6.7)
Respiratory problems	8 (1.0)	2 (1.5)
Renal toxicity	5 (0.7	3 (2.2)
Hypokalaemia (≤3.4 mEq/L)	4 (0.5)	2 (1.5)
Oral discomfort/difficulty in swallowing	4 (0.5)	-
Weight loss	4 (0.5)	-
Other*	8 (1.0)	3 (2.2)
Total	774 (100)	134

^*Other AEs included: Acute AEs included acute generalised exanthematous pustulosis (AGEP), breast swelling, death, dysuria and burning, pregnancy, increased blood glucose levels, spontaneous abortion, and thyrotoxicosis.

CNS, central nervous system; GIT, gastrointestinal tract; SAEs, serious adverse events.

Trends in adverse event detection and reporting

Trends in the detection and reporting of AEs over the study period (2018‒2022) were evaluated. As figure 1 shows, in 2018, eighty (80) cases of AEs were detected, and 6 (7.5%) of them were reported to be NPC. In the next year, 2019, 157 AEs were recorded, of which 12 (7.6%) were reported. For the remaining years, 176, 182 and 160 AEs were detected in 2020, 2021 and 2022, respectively, and 23 (13.1%), 34 (18.7%) and 41 (25.6%) AEs were reported from respective in these years (figure 1).

Figure 1. Trends of adverse event occurrence and reporting during the treatment of MDR/RR-TB in Ethiopia from 2018‒2022. AEs, adverse events; MDR/RR-TB, multidrug-resistant and rifampicin-resistant tuberculosis.

Discussion

This study highlighted the moderate implementation of aDSM within the programmatic management of MDR/RR-TB. We observed a low number of AEs detected and reported during the MDR/RR-TB treatment courses at selected health facilities. The magnitude of the AE detected and the reporting rate can be key indicators for the best performance of aDSM.²⁴ Furthermore, MDR/RR-TB treatment regimens are known to cause excessive AEs,^{25 26} and previous studies reported an association of MDR-TB medications with a greater proportion of AEs.²⁶ However, in this study, only records of 33% of patients showed that they developed at least one AE during treatment. The observed low incidence of AEs in this study may be attributed to inadequate patient monitoring practices and/or incomplete medical records at treatment facilities. Our findings revealed that only 56.7% of patients were assessed for AEs during the first 6 months, and this figure decreased to 39.3% from 6 months to the end of treatment. Notably, 6.8% of patients did not receive any AE evaluations throughout their follow-up period. Furthermore, we observed notable inter-site variation in AE detection. For instance, at ALERT Hospital, 63.3% of patients experienced AEs, whereas at Yirgalem Hospital, AEs were recorded for only 3.1% of patients. This significant difference in AE documentation may reflect gaps in the implementation of patient monitoring practices at certain sites. This underscores the need to strengthen pharmacovigilance systems to improve patient safety and treatment outcomes.

Since an AE can be any abnormal sign or symptom due to treatment or progression of disease symptoms, laboratory abnormalities, or a combination of such abnormalities,^{27 28} the capacity of AE detection mainly depends on continuous clinical and laboratory safety monitoring during treatment periods. In addition, strict patient monitoring during the treatment of MDR-TB is crucial for the early identification and management of AEs.²⁹ Furthermore, it was noted that aDSM provides a system for identifying AEs using systematic clinical and laboratory assessments of people receiving treatment.²⁸Thus, we assessed whether the clinical safety monitoring and laboratory tests were performed following the current MDR/RR-TB management guidelines. Accordingly, we observed that nearly half and three-fourths of the patients missed each monitoring parameter during the first 6 months and then until the end of the treatment period, respectively. This was an indication of the prevailing non-compliance with the recommended assessment schedules and parameters. For example, for at least 15% of patients, there were no records for one or more laboratory tests at all, indicating that the tests were either not done or not recorded in patient files. On the other hand, the limited test reporting and under-detection of AEs could be linked to irregular access to laboratory test services. Therefore, this suggests that the system for continuous clinical safety and laboratory monitoring should be strengthened to improve aDSM implementation and thereby ensure detection of the actual extent of AEs.

Furthermore, our investigation revealed under-reporting of AEs during MDR/RR-TB management, as only less than one-fifth (17.4%) of all types of AEs were reported to the regulatory body, an indication for strengthening consistent and comprehensive reporting practices. Although all AEs are reportable in ideal circumstances, depending on resource availability, the TB programme can focus on SAE and AESI,³⁰ and all SAE and AESI are expected to be reported.³¹ Notwithstanding this, in this study, only 35.8% and 31.1% of SAE and AESI were reported to NPC, respectively. Reporting AEs to the NPC is vital and the primary mechanism for detecting, assessing and responding to drug safety issues in public health programmes that receive medicines directly from donors, bypassing the conventional regulatory frameworks that involve market authorisation holders.⁸ Under-reporting of AEs is a globally recognised challenge, particularly in routine and programmatic care settings, and undermines safety assessment and patient management.^{32 33} This under-reporting results from multiple interrelated factors, including limited resources, weak healthcare infrastructure, inadequate training,³⁴ low awareness³⁵ and the complexity of AEs associated with prolonged multidrug regimens.^{35 36} Failure to promptly detect and report AEs may allow them to progress in severity, thereby compromising treatment adherence and leading to unfavourable outcomes.³⁵ Consequently, gaps in AE reporting hinder the effectiveness of pharmacovigilance in the management of AEs.³⁴ Thus, multifaceted strategies, including ongoing educational and sensitisation programmes for healthcare workers (HCWs)³⁷ and engagement of patients in pharmacovigilance,³⁸ are required to improve AEs reporting. Importantly, HCWs should execute their obligation to report any suspected SAE to the NPC, as stated in the EFDA pharmacovigilance directive.³¹

In addition to under-reporting, a delay in reporting SAEs and AESI was also identified. We observed that the median time to report SAE was 1.0 days (IQR: 0–14), and for AESIs, it was 3.0 days (IQR: 0–22), indicating that 25% of SAEs and AESIs were reported after 14 and 22 days of occurrence, respectively. This may be due to either non-compliance with the reporting timeline or incomplete documentation of reported AEs. Such delayed reporting is not supported by the country’s pharmacovigilance directive, which stated that all suspected SAEs shall be reported within 48 hours of their occurrence.³¹ As the delay in AE reporting will make reports unreliable and put patients’ safety at risk,²³ HCWs should vigilantly detect AEs and must report to the NPC as soon as possible, at least within the established timeline.

The introduction of aDSM has been recognised by the WHO as complementary to national pharmacovigilance systems, helping to address gaps in safeguarding patient safety.⁸ In line with this, our findings revealed a consistent increasing trend in the detection and reporting of AEs over consecutive years, from 2018‒2022. Although the overall number of identified and reported AEs was lower than anticipated, the observed trend over the years indicates that implementation of aDSM has contributed to the improvement of the AE detection and reporting process. Similarly, it was reported that aDSM implementation in TB management programmes was crucial for strengthening AE detection in treatment programmes.³⁹ Thus, our findings clearly indicate that, if strictly implemented, aDSM can significantly contribute to the improvement of PV systems. Therefore, ongoing efforts are necessary to overcome existing challenges and ensure effective and sustainable aDSM implementation.

Limitations

Finally, owing to the retrospective nature of its design, this study may have limitations, specifically: certain information was incomplete, which could lead to the underestimation of some results. Despite these constraints, the study’s multisite approach and substantial sample size enhance the robustness and generalisability of its findings. By encompassing diverse patients across various healthcare settings, the findings of this study provide valuable insights that can inform decision-making at the national and various healthcare system levels.

Conclusion

This study highlighted the moderate performance of aDSM during programmatic management of MDR/RR-TB. Despite the implementation of aDSM, the detection and reporting of AEs have remained suboptimal, likely because of inadequate patient monitoring practices, limited laboratory tests, and/or incomplete medical record-keeping at treatment facilities. Nevertheless, an upwards trend in AE detection and reporting was observed over time during the aDSM execution period, suggesting that comprehensive and strict aDSM implementation can enhance national pharmacovigilance efforts and improve patient safety in MDR/RR-TB and other treatment programmes.

Future research

Future research employing mixed-methods approaches is needed to identify determinants of aDSM implementation, including drivers of under-detection and under-reporting of AEs, as well as facility-level, staff capacity and supply-chain–related constraints, to inform targeted strengthening of aDSM implementation.

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study. All data relevant to the study are included in the article or uploaded as supplementary information.