Treatment outcomes and risk factors of death in childhood tuberculous meningitis in Kandahar, Afghanistan: a prospective observational cohort study

Abstract Background Tuberculous meningitis (TBM) is the most severe form of TB. We prospectively documented the treatment outcomes and the risk factors for death in children with TBM from Kandahar, Afghanistan. Methods This prospective observational cohort study was conducted from February 2017 to January 2020 in hospitalised TBM children. All the patients were prospectively followed up for 12 mo. Data were analysed by using descriptive statistics, χ2 and multivariate logistic regression. Results A total of 818 TBM hospitalised patients with median age 4.8 (0.8–14.5) y were recruited. Females accounted for 60.9% (498/818). Upon admission 53.9% (n=441) and 15.2% (n=124) had TBM stages II and III, respectively, and 23.2% (n=190) had focal neurological signs. The case fatality rate was 20.2% (160/794) and 30.6% (243/794) survived with neurological sequelae. Independent risk factors for death were being unvaccinated for BCG (adjusted OR [AOR] 3.8, 95% CI 1.8 to 8.1), not receiving dexamethasone (AOR 2.5, 95% CI 1.5 to 4.2), being male (AOR 2.3, 95% CI 1.5 to 3.6), history of recent weight loss (AOR 2.2, 95% CI 1.3 to 3.9) and having stage III TBM (AOR 2.0, 95% CI 1.2 to 3.3). Conclusions TBM continues to cause high morbidity and mortality in Afghan children. Strategies to reduce mortality should emphasise early diagnosis and treatment, routine use of dexamethasone and increased BCG vaccination.


Introduction
TB, an infectious disease caused by the bacillus Mycobacterium tuberculosis , ranks globally in the top 10 causes of death and is the leading cause of death from a single infectious disease. 1 In 2019, approximately 10 million people had TB and 1.4 million died. 1 Tuberculous meningitis (TBM) is the most severe form of TB and is associated with high mortality rates of approximately 15-30%, despite anti-TB chemotherapy. 2 -5 TBM is also responsible for neurological sequelae and severe disability in 43-79% of survivors 6 -8 and for socioeconomic hardship. 9 In 2007, a study in India observed neurological sequelae in 78.5% of patients, including cognitive impairment (55%), motor deficit (40%), optic atrophy (37%) and other cranial nerve palsies (23%). 10 These sequelae were common in patients with baseline focal motor deficits and impaired consciousness. 10 Delays in the diagnosis and treatment of TBM increases mortality, 11 with one Peruvian study reporting a 70% increase in mortality with a treatment delay of > 3 d after admission. 12 Diagnosing TBM by the acid fast staining of a cerebrospinal fluid (CSF) has a very low sensitivity of < 10%. 13 -15 CSF culture of M. tuberculosis has a sensitivity of approximately 50-60% but is slow, requiring up to 8 wk on solid media and, ideally, requires a biosafety level 3 laboratory. 16 GeneXpert, a WHO-recommended molecular diagnostic method, has a sensitivity similar to culture (50-60%), but with a run time of 2 h. [17][18][19] Chest radiography and neuroimaging studies are useful to support a diagnosis of clinically suspected TBM. 20 Afghanistan is a resource-poor country. Therefore, diagnostic tools for the rapid diagnosis of TB are limited to Ziehl Neelson (ZN) staining of slides and ancillary diagnostic tools such as chest radiograph. 21 Molecular methods (e.g. GeneXpert) are not readily available, but this limitation is common in many settings. As a result, algorithms that combine clinical features (e.g. duration of illness) with CSF findings and radiology are used to diagnose TBM. 22 , 23 Thwaites et al. used five features in adults to predict TBM over bacterial meningitis (i.e. age, length of history, white blood cell count, total CSF white-cell count and CSF neutrophil proportion). This algorithm was 97% sensitive and 91% specific and was validated in a later prospective study. 22 , 24 TB remains a major public health challenge in Afghanistan. TB services (diagnosis and treatment) are free of charge and the public health surveillance system is based on confirmed and clinically suspected cases. 25 25 Overall, 56% were females, approximately 20% (9732) were children and about 26% had extra pulmonary TB. 25 A small, retrospective, hospital-based study of extrapulmonary TB showed that 23/118 (19.5%) had TBM; the most common TBM symptoms were headache (69.6%), fever (60.9%) and vomiting (60.9%). At the time of presentation, 5/23 (21.7%) patients were unconscious, while 6/23 (26.1%) had neck stiffness. Treatment outcomes were not reported. 26 Afghanistan follows the WHO-recommended treatment for drug-sensitive TBM for adults and children, which consists of a four-drug regimen (isoniazid, rifampicin, pyrazinamide and ethambutol) for 2 mo, followed by 7-10 mo of rifampicin and isoniazid 27 , 28 with high dose adjunctive dexamethasone, tapered over 6-8 wk. 27 Evidence supporting the use of dexamethasone in adults is small but of high quality, 29 while, for children, the evidence is weaker and is based on smaller randomised trials with supportive findings from observational studies. 30 -33 To our knowledge, there has never been a prospective study of TBM in Afghanistan. We, therefore, conducted a study to describe the features of paediatric TBM, document treatment outcomes and determine risk factors for death.

Study design, site and participants
This was a prospective observational cohort study that was conducted from February 2017 to January 2020 (3 y) at the Mirwais regional hospital in Kandahar city, the largest tertiary referral hospital in southwest Afghanistan ( Figure 1 ).
Included children had to satisfy the following criteria: (i) a signed consent form by a legal guardian, (ii) age < 18 y, (iii) probable or definite TBM as per the scoring system described by Marais et al., 34 (iv) admitted to the paediatric ward of Mirwais regional hospital and (v) were residents of Kandahar province. Children were excluded if they had confirmed or CSF findings favouring non-TBM, acute encephalitis or were on anti-TB treatment before admission.

Clinical care of admitted children
All children underwent a history and physical examination, which included a detailed neurological examination and a lumbar puncture (21-or 23-gauge needles) for standard CSF analysis, including a ZN-stained slide. Blood was taken for routine haematology and biochemistry and repeated as clinically indicated.

Data collection and analysis
A standardised case record form (CRF) was used to record sociodemographic data, clinical features, laboratory examinations, treatment and outcomes. The CRF was completed by expert paediatricians of the paediatric ward. Histories were taken from the patients' parents or caretakers. Anonymised data were double entered into Microsoft Excel and cleaned before analysis in SPSS version 22 (Chicago, IL, USA).
Frequencies/percentages and means/medians were used to summarise categorical and continuous variables, respectively. χ 2 test was used to assess associations of categorical variables and the unpaired t test or non-parametric Mann-Whitney U tests were used to compare continuous data, as appropriate. Statistically significant (p < 0.05) variables were assessed for independence by multivariate logistic regression to determine the adjusted OR (AOR) for factors associated with mortality. The cumulative probability of death was determined by Kaplan-Meier survival analysis. p < 0.05 was considered statistically significant.
Transactions of the Royal Society of Tropical Medicine and Hygiene

Treatment and treatment outcomes
All children were treated with anti-TB drugs for 12 mo. For the first 2 mo, intensive therapy comprised isoniazid (INH, 10 mg/kg), rifampicin (15 mg/kg), pyrazinamide (35 mg/kg) and ethambutol (20 mg/kg), followed by 10 mo of maintenance therapy with INH (10 mg/kg) and rifampicin (15 mg/kg). When prescribed, dexamethasone was given for 4 wk: 0.6 mg/kg/day intravenous for 7 d, reducing to 0.3 mg/kg/day for 21 d. Treatment outcomes were assessed at 1 wk, 1 mo, 6 mo and 12 mo using the Paediatric Cerebral Performance Category (PCPC) scale. 35

Definitions
Confirmed TBM: clinical criteria (symptoms and signs of meningitis, including ≥1 of the following: headache, irritability, vomiting, fever, neck stiffness, convulsions, focal neurological deficits, altered consciousness or lethargy) plus ≥1 of the following: acidfast bacilli (AFB) seen in the CSF, M. tuberculosis cultured from the CSF or a CSF-positive commercial nucleic acid amplification test. 34 Probable TBM: for this, we also used the criteria from Marais et al.: clinical criteria plus a total diagnostic score of either ≥10 points (cerebral imaging not available) or ≥12 points (cerebral imaging available) and exclusion of alternative diagnoses. At least two points should come from either CSF findings or cerebral imaging criteria. 34 Staging/grading of TBM severity: we used the Medical Research Council (MRC) TBM severity grades: (i) Grade I, alert and orientated without focal neurological deficit; (ii) Grade II, Glasgow coma score (GCS) 11-14 with or without focal neurological deficit or GCS 15 with a focal neurological deficit; and (iii) Grade III, GCS ≤10 with or without a focal neurological deficit. 36 Tuberculin skin test: using WHO guidelines, the tuberculin skin test (TST) was regarded as positive when there was > 5 mm of induration in high-risk children (including HIV-infected and severely malnourished children) or > 10 mm of induration in all other children, irrespective of previous BCG vaccination. 37 Socioeconomic status: low socioeconomic status, monthly income < 2500 Afghanis ( < US$30); middle socioeconomic status, monthly income 2500-20 000 Afghanis (US$30-250); and high socioeconomic status, monthly income > 20 000 Afghanis ( > US$250).

Results
In total, 818 children met the inclusion criteria and were selected for the final analysis ( Figure 2 ). As each patient was followed up prospectively for 12 mo during anti-TB treatment, this study lasted 3 y. The median admission age was 4.8 y and the majority of the patients (60.9%; 498/818) were female ( Transactions of the Royal Society of Tropical Medicine and Hygiene  Table 3 ).
Of the 160 deaths, 151 (94.4%) occurred in hospital. The times to death as a function of receiving dexamethasone were rapid for both groups (Figure 3 ), but the dexamethasone recipients had a threefold increased survival rate, hazard ratio = 3.3 (95% CI 2.1 to 5.2, p < 0.001). Upon admission, 156 and 459 children had PCPC scores of 3 and 4, respectively, which improved significantly at 12 mo (Table 4 ). Moreover, there was a trend of increasing numbers of children with mild disability. However, this increase in mild disability was due to a significant decrease in the number of patients with coma or vegetative state, severe disability and moderate disability.

Discussion
In Afghanistan, a confirmed diagnosis of TBM is very difficult due to limited diagnostic tools and resources. Therefore, TBM diagno-sis in our study site was based on the combination of clinical features (e.g. duration of illness) with CSF findings and chest radiography (Marais criteria). 34 These conditions are common where TBM is most prevalent. There was a long median delay in presentation of approximately 4 wk in our patients, consistent with other studies, 33 , 41 and there was favourable response to treatment in many children. Therefore, we believe that most of our children had TBM, cognizant that the differential diagnosis of chronic meningitis is very wide. 41 Our study is noteworthy for its large size, prospective nature and high proportion of children aged < 5 y who are underrepresented in some studies. We did not set out to examine specifically Transactions of the Royal Society of Tropical Medicine and Hygiene  the effect of dexamethasone and dexamethasone recipients had several baseline characteristics that were significantly different vs the non-dexamethasone group. Nevertheless, dexamethasone significantly reduced mortality by approximately 70% in those with MRC grade III. Randomised trials confirming the mortality reduction of dexamethasone in TBM have been reported in Vietnamese adolescents and adults, 42 in 'young' South African children (ages not specified) 32 and Egyptian patients across the age spectrum 31 ; a later retrospective study of 300 Egyptian children aged < 5 y also found reduced mortality in steroid recipients. 33 These results are consistent with several paediatric studies from India 30 , 43 , 44 and Vietnam, 45 but not with one Thai study. 46 A Cochrane review of nine TBM studies found an overall mortality benefit of steroids of 25% in HIV-1-negative adults and children, but their effect on reducing disabling neurological deficits remains open and requires more research. 47 Afghanistan's health system is facing serious challenges, mostly due to insecurity. This has negatively affected the treat-ment outcome of all forms of TB in Afghanistan, especially in rural areas, where the security situation is worse. In our study, the treatment success rate was 79.8%. A retrospective cohort study (2011-2014) on all forms of TB was conducted in a war-affected region of Khyber Pakhtunkhwa in the northwest of Pakistan. This study revealed that among 181 extra-pulmonary TB (including TBM) patients, 80.1% had successful treatment outcomes. 48 This treatment success rate is nearly the same as observed in our study.
Our overall mortality rate was high, approximately 20%, which is higher than the mortality rate reported in childhood TMB from China (4.0%), 2 Thailand (6.4%), 49 Romania (8.0%), 50 South Africa (13.0%) 51 and Vietnam (15.7%), 3 but lower than in India (29.0%) 4 and Ethiopia (48.1%). 52 In our study, worryingly, most deaths occurred soon after admission and the identified, independent risk factors associated with death were male gender, recent weight loss, no BCG vaccination and no dexamethasone given in hospital. TBM mortality risk factors vary across settings but several key factors have been consistently identified, namely, more severe disease at presentation, 53 , 54 which, in turn, is related to delayed presentation and diagnosis, 5 2, 5 5-57 young age, 53 HIV positivity, 58 -60 the lack of BCG vaccination, 61 -63 low socioeconomic status 9 and the presence of raised ICP. 64 In Afghanistan, additional factors for mortality can probably be linked to Afghanistan's fragile healthcare system, national insecurity, the unavailability of advanced life support for seriously ill patients, misdiagnosis and poor referral linkage between primary care and tertiary centres. 65 Receiving BCG vaccination at birth has a reported efficacy of approximately 73% 66 and it has been estimated that had BCG vaccinations been given to 100.5 million children born in 2002, some 30 000 TBM cases would have been prevented (one case/3500 inoculations). 66 However, one Indian study reported a reduced benefit in children aged > 5 y with severe malnutrition and a positive household contact. 67 Another effective strategy is post-exposure prophylaxis in young (aged < 5 y) children, 68 but none of our children received this, illustrating again the significant gap between policy and practice in a developing country. 4 , 69 Our study had several limitations. It was conducted at a tertiary referral hospital and this may have introduced a mortality bias, because the severity spectrum of TBM we observed may not be seen in smaller district hospitals. The nature of the study is biased by having the diagnosis of TBM already assigned to the cohort. Although Mirwais regional hospital is a referral hospital, its diagnostic capacity is limited and all TBM diagnoses were made clinically with supporting evidence from CSF findings and chest radiograph; we do not have facilities for TB culture or neuroimaging. We did not screen for HIV status because the prevalence of HIV is very low in Afghanistan and HIV tests are not performed routinely for inpatients.
Diagnostic tests and treatment in Mirwais regional hospital are free of charge and attract patients from all the districts of Kandahar province, as well as the neighbouring provinces of Helmand, Uruzgan, Zabul, Farah and Nimruz. With such a wide catchment area, challenges with transport may have introduced delays in patients' presentation.

Conclusions
TBM continues to cause high morbidity and mortality in Afghan children and, in our setting, death occurred rapidly. Administering BCG vaccination and giving dexamethasone to all TBM patients are two simple and inexpensive interventions that could have a profound impact on TBM-related mortality. Our data support the use of dexamethasone for 4 wk, but future research should determine the optimal dose of dexamethasone in vulnerable children aged < 5 y.

Transactions of the Royal Society of Tropical Medicine and Hygiene
We are also very thankful to the staff members (ward chiefs, clinicians, nurses and laboratory technicians) of Mirwais regional hospital, research team members of faculty of medicine and all of the study participants.
Funding: This study did not receive any specific funding. W. R. Taylor is part-funded by Wellcome under grant 220211. For the purposes of Open Access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.
Competing interests: All the authors do not have any competing interests.
Ethical approval: Written informed consent was obtained from all children's parents or primary caregivers prior to study enrolment. Children aged < 15 y but old enough to answer were given assent forms for their participation. Information regarding the participants will not be disclosed. The study was approved by the Kandahar University Ethics Committee (KDRU-EC-2017.13). For data collection, only patients' initials were used. All data used for analysis were coded and de-identified.
Data availability: Data is available on request to the email drbilal77@yahoo.com .