An epidemiological review of tuberculosis in Sabah, Malaysia, 2012-2018

Background: Tuberculosis (TB) is of high public health importance in Malaysia. Sabah State, located on the island of Borneo, has previously reported a particularly high burden of disease and faces unique contextual challenges compared with peninsular Malaysia. The aim of this study is to describe the epidemiology of TB in Sabah to identify risk groups and hot spots of TB transmission. Methods: We conducted a retrospective review of TB cases notied in Sabah, Malaysia, between 2012 and 2018. Using data from the state’s ‘myTB’ notication database, we calculated the case notication rate and described trends in the epidemiology, diagnostic practices and treatment outcomes of TB in Sabah within this period. Results: Between 2012 and 2018 there were 33,193 cases of TB reported in Sabah (128 cases per 100,000 population). We identied several geographic hotspots, including districts with >200 cases per 100,000 population per year. TB rates increased with age and were highest in older males. Children <15 years accounted for only 4.6% of cases. Moderate or advanced disease on chest x-ray and sputum smear positivity was high (58% and 81% of cases respectively), suggesting frequent late diagnosis. Multi-drug Resistant (MDR) TB prevalence was low (0.3% of TB cases), however, rapid diagnostic test coverage was low (1.2%) and only 18% of all cases had a positive culture result. Treatment success was 83% in those with drug-sensitive TB and 30% in cases of MDR-TB. Conclusion: Between 2012 and 2018, Sabah State accounted for 20% of Malaysia’s TB notications, despite representing only 10% of Malaysia’s population. We found hotspots of TB in urbanised population hubs and points of migration, as well as evidence of late presentation and diagnosis. Ensuring universal health coverage and expansion of GeneXpert® coverage is recommended to reduce barriers to care and early diagnosis and treatment for TB. the reality of deep poverty affecting many residents of the eastern states of Malaysian Borneo. In the present era of the SARS-CoV-2 pandemic, there has been strong recognition of the need for investment in early detection and prevention of transmissible respiratory disease, rather than reactive disease treatment.(3) These lessons, especially the proven effectiveness of case detection, case isolation and contract tracing, provide much-needed inspiration to re-orient healthcare services in high TB burden settings towards prevention. Examining regional TB epidemiology provides the best mechanism for assessing the effectiveness of existing prevention and management strategies. died, lost to follow-up or not evaluated. ‘Treatment success’ included cases conrmed cured by negative culture at completion of treatment, or who successfully completed the full course of treatment. evaluated’ included patients who were transferred out or where the treatment outcome was not recorded. Treatment success was calculated as a proportion of all registered cases. For cases of multidrug resistant TB (MDR-TB), treatment outcome was assessed for years 2012-2017 as data were incomplete for 2018 at the time of analysis. Previous studies have shown that the of TB cases occur within two years of we found supporting that

Borneo. In the present era of the SARS-CoV-2 pandemic, there has been strong recognition of the need for investment in early detection and prevention of transmissible respiratory disease, rather than reactive disease treatment.(3) These lessons, especially the proven effectiveness of case detection, case isolation and contract tracing, provide much-needed inspiration to re-orient healthcare services in high TB burden settings towards prevention. Examining regional TB epidemiology provides the best mechanism for assessing the effectiveness of existing prevention and management strategies.
While Malaysia is successfully reducing TB related mortality,(4, 5) TB incidence is not decreasing in line with global End TB Strategy milestones.(6) Sabah State, located in Malaysian Borneo, has historically had a higher burden of TB. While Sabah State accounts for only 10% of the country's total population, it has been reported to account for 20-30% of all Malaysia's TB cases. (7,8) While TB rates in other parts of the country have fallen over the past decade, sustained high TB case noti cation rates of 144 to 217 cases per 100,000 population have been observed. (9) Previous studies have indicated that the TB epidemic in Sabah is chie y related to delayed health seeking and limited access to TB care, rather than factors such as HIV or drug resistance which may drive TB epidemics elsewhere. (10) Fortunately rates of HIV co-infection and drug-resistance in Sabah have been reported to be relatively low. (10,11) However, presentations with pulmonary TB are often late, with extensive cavitary disease and profound weight-loss frequently evident at diagnosis.(10) Delayed diagnosis and advanced disease are likely to result in poorer patient outcomes and increased risk of transmission prior to starting treatment.
The World Health Organization's End TB Strategy sets ambitious targets to which countries must strive in order to end the global TB epidemic.(6) Pillar one of the strategy -'integrated patient centred care and prevention', requires countries to improve the diagnosis and management of TB in high-risk groups. (6) It is imperative that TB control strategies are adapted to t the local epidemiology to ensure effective and e cient use of limited resources and ensure vulnerable and high-risk groups within the population are reached and that their health needs are adequately addressed. While it is clear that the burden of TB in Sabah is disproportionately high, there is little published data characterising the local TB epidemic. The most recent report presents data from two decades ago. (9) The aim of this study is to describe the epidemiology of TB in Sabah to better identify risk groups and hot spots of TB transmission.

Methods
In Sabah, all microbiologically con rmed and clinical cases of TB must be noti ed to the State Health Department where they are recorded in a secure electronic database (myTB). Using this data, we conducted a retrospective review and descriptive analysis of all noti ed cases of TB disease in Sabah State Malaysia, with a date of diagnosis between 1 January 2012 and 31 December 2018.
The study period was based on the availability of complete calendar years of data at the time of the study. Population data used to calculate incidence rates was obtained from the Malaysian Department of Statistics and data on national TB noti cations in Malaysia was obtained from the World Health Organization (WHO). (12) Descriptive analysis of demographic, epidemiological and clinical data was undertaken using the R software package.(13) Maps were produced using QGIS v3. 8.2.(14) We determined numbers of cases of drug-sensitive and multi-drug resistant TB and noti cation rate of drug-sensitive disease per 100,000 population, by year and district of residence. Using national noti cation data, we calculated the proportion of all cases in Malaysia occurring in Sabah State. Cases were characterised by age, sex, country of birth and Malaysian citizenship status. The myTB database records information on three speci c TB risk factors -HIV co-infection, diabetes and smoking. We assessed the prevalence of these risk factors among reported TB cases. Clinical and diagnostic information was used to determine the proportion of cases diagnosed clinically or with bacteriological con rmation, the proportion with pulmonary versus extrapulmonary disease, and various sites of infection. Site of infection data was available for years 2012 to 2017 only. Radiological severity was assessed using established grading criteria and classi ed as no lesion, minimal lesion, moderately advanced or far advanced. (15,16) Bacteriologically con rmed TB was de ned as a case in whom a biological specimen is positive by smear microscopy, culture or WHO recommended rapid diagnostic test such as GeneXpert®. Clinically-diagnosed TB includes cases who do not meet the criteria for bacteriologically con rmed cases, but in whom a clinician has diagnosed TB and decided to provide a full course of anti-TB treatment. (17) Treatment outcome was classi ed based on WHO reporting guidelines, (18) as treatment success, failure, died, lost to follow-up or not evaluated. 'Treatment success' included cases con rmed cured by negative culture at completion of treatment, or who We assessed the proportion of new versus previously treated cases, the method of case detection and type of diagnosing heath facility, the proportion of pulmonary sputum smear positive cases, and the proportion with moderate to advanced disease on chest x-ray. We assessed treatment outcomes, including sputum smear conversion during treatment and nal treatment outcome.
Treatment success was de ned as cured (negative sputum culture at treatment completion) or had completed the full treatment course. The Chi-squared test was used for determining the difference between two proportions.

Patient and public involvement
This research was done without patient involvement. Patients were not invited to comment on the study design and were not consulted to develop patient relevant outcomes or interpret the results. Patients were not invited to contribute to the writing or editing of this document for readability or accuracy.

Demographic characteristics
The median age of TB cases was 38 years. Case noti cation rates increased steadily with age from a low of 23 cases per 100,000 among children aged 5-14 years, to a high of 402 cases per 100,000 among adults aged 65 years and older. Only 1.6% of cases were among children aged less than ve years, and 4.6% of cases were children less than 15. Further, the proportion of total cases that were children decreased from 2012 to 2018: by 2018, only 1.2% of cases were <5 years and 3.6% were <15 years.
There was substantial gender disparity among reported TB cases; 60% of cases were male. This was most notable in older age groups -in particular, those aged ≥55 years had a male to female noti cation ratio of 2.1:1. In both males and females, noti cation rates increased with age, but from adults aged 25 years and older, there was increasing disparity between sexes. The highest noti cation rates were seen in males aged 65 years and older at 540 cases per 100,000 population (compared to 254 cases per 100,000 in women in the same age group). There was little difference in the number of noti cations between males and females among children less than 15 years (Figure 3).
Over the course of the study period, 50% of cases were born in Malaysia. Of cases born outside Malaysia, the majority were born in the Philippines (17%) and Indonesia (7%). The remaining 26% represented a range of other nationalities. There was some variation in the distribution year to year, though no obvious trend ( Figure 4). Most cases (71%) were Malaysian citizens, and no trend in the proportion of cases with or without Malaysian citizenship was observed over the study period. However, there was a substantial difference by district of residence in the proportion of cases who were non-citizens -ranging from <5% of cases in Nabawan, Kota Belud and Pitas, to >50% of cases in the eastern districts of Lahad Datu and Kunak and 62% of cases in Kinabatangan ( Figure   5). Among overseas-born cases, only 14% had been in Malaysia <5 years at their time of diagnosis; 52% of overseas-born cases had a length of stay in Malaysia of 10-29 years at their time of diagnosis.
Over the study period, monthly median household income of TB cases was 1,000 Malaysian Ringgit (MYR) (equivalent to ~USD 250).

Case detection and diagnosis
Overall, 57% of TB cases were diagnosed in Government hospitals, with 42% diagnosed in public primary health or village clinics and 2% diagnosed in private facilities. Over time, there was a decrease in the proportion of cases diagnosed in government hospitals, from 61% of cases in 2012, down to 52% in 2018. A corresponding increase in diagnoses occurred in primary care (primary health and village clinics) and a small increase at private facilities (clinics and hospitals). ( Figure 6).
Between 2012 and 2018, most cases (89%) were detected passively. Of those that were diagnosed through active case nding, 74% (N= 2,244 cases) were detected during contact investigation -equivalent to 6.8% of all TB cases during this period. The remaining 26% of cases were detected through various screening mechanisms, such as community or occupational screening -equivalent to 4.2% of all TB cases during this period. The overall proportion of TB cases detected through community screening did increase from 2014 onwards, though accounted for only a very small proportion of overall case detection (reaching just 1.7% in 2018).
Seventy-three percent of TB cases between 2012 and 2018 were bacteriologically con rmed (range 70-74%). However, among children less than ve years, only 19% were bacteriologically con rmed. Both culture and smear microscopy are mandatory components of the diagnostic algorithm in Malaysia. (19) However, among all cases during this period, only 74% were tested by culture, varying year to year from 51% to 87%. Culture positivity was low at 24% overall, ranging from 11% to 29% between years.
Culture positivity among pulmonary cases only was 26%. A total of 403 cases were tested by GeneXpert® during the study period, equating to test coverage of only 1.2% (range 0.3-3.0%). GeneXpert® positivity was higher than culture at 66% overall, but also varied substantially from 0% (of 12 samples tested) in 2012 to 89% (of 149 samples tested) in 2016. Most cases (97%) had smear microscopy results available, and this was consistent across the period. (Table 1)  available, however we would anticipate that there is a substantial gap between the incidence and noti cation rates within Sabah, at least equivalent to that seen at the national level. As such, the case noti cation rates presented are likely to underestimate the true burden of disease.
Sabah State is a setting which poses unique challenges to Malaysia's national TB control program. Sabah is less economically developed than peninsular Malaysia, has an interior which can be topographically challenging and di cult to access for healthcare provision, has substantial communities of marginalised persons living in overcrowded informal settlements, and a porous land/sea border with the Philippines.(21) Understanding the local epidemiology in this unique context is critical to implementing effective measures for TB control and prevention. TB noti cation data is the best available source of data in Sabah for describing the local epidemiological situation. However, the ndings are potentially limited by the quality and completeness of data recording and entry.
Malaysia has a complex system of both paper based and electronic mechanisms for collection of patient data and noti cation of TB, and enumeration of cases may be imperfect. There was no scope to cross-check data from the myTB database against primary records. However, the State Department of Health undertakes regular auditing and cleaning of data to improve its quality and completeness. Thus, our ndings are likely to also underestimate the true burden of disease, particularly in some groups of the population which face additional barriers to care, such as those living in remote areas and migrants, particularly those without legal status.
Our results show that while Sabah has a moderate overall burden of TB, there were geographic hotspots of disease -particularly the districts of: Semporna, Pitas, Kota Marudu, Tuaran, and Kota Kinabalu. These districts correspond with areas of frequent crossborder movement between Malaysia and the Philippines (Semporna and Pitas) and the more populous urbanised districts of northwestern Sabah (Tuaran, Kota Kinabalu and Kota Marudu). Given the high burden of TB in Sabah's nearest two neighbours, the Philippines and Indonesia (incidences of 554 and 316 cases per 100,000 population respectively), (12) and the frequent movement of people across these international borders, there is an evident need for the Malaysian healthcare system to actively engage international arrivals in no-cost TB prevention and early detection strategies. It is di cult however, to ascertain the relative burdens of imported versus locally acquired TB disease. Our nding of geographic areas with a high proportion of cases among non-citizens, and high-burden urban areas with a relatively low proportion of cases among non-citizens, may support a hypothesis of seeding of overseas acquired disease in Sabah, with ampli cation of transmission occurring in densely populated urban centres. Investment in healthcare for non-citizens could therefore have a major bene cial impact on overall TB rates. Previous studies have shown that the majority of TB cases occur within two years of infection. (22,23) However, we found that among overseas born TB cases, the majority developed TB ve or more years after arrival in Sabah, further supporting the likelihood that transmission of TB, even among overseas born cases, is occurring locally.
Sabah's migrant population, especially those who are refugees, from stateless minority groups and illegal or undocumented migrants face substantial challenges to accessing healthcare and social protection -including nancial, legal, language and physical access barriers. (24)(25)(26) The nding that non-citizens were signi cantly more likely to have advanced disease on chest x-ray and be sputum smear positive at presentation, supports the hypothesis that these groups face additional barriers to accessing health services. We found that monthly household income of TB cases in Sabah (MYR 1,000 or USD 250) was around one-quarter the median monthly household income across Sabah State (MYR 4,110 or USD 1,031 in 2016). (27) This highlights the risk of nancial barriers limiting access to diagnosis and care.
The United Nations Sustainable Development Goals encourage countries to achieve universal health coverage and WHO's End TB Strategy reinforces the role of universal health coverage in ensuring equal and unhindered access to early diagnosis and treatment of TB. (6,28) Understanding the barriers that both non-citizens and citizens of Malaysia may face in accessing TB services is essential, followed by engagement of communities and reorientation of services to ensure patient-centred care. (6) We note that over the study period there has been a steady shift in diagnosis of TB from government hospitals to primary health and village clinics and the private sector. This shift may indicate a decentralisation of services which could improve access to TB diagnosis and care. It is however important that the TB program engages with and equips primary health, village clinics and private facilities with the support they need to increase their role in TB care.
Our results highlight key gaps in the detection of TB in Sabah. We found that case detection among children was very low, with Uptake of WHO-recommended rapid diagnostics (GeneXpert®) for diagnosis of TB was limited. However, at the time of the study, national guidelines did not recommend GeneXpert® as a rst-line diagnostic test for all suspected drug-sensitive TB cases. (30) Rapidly increasing GeneXpert® coverage should be a priority, especially given that a quarter of cases were not tested by culture, and of those cases that were, culture positivity was extremely low (26% among pulmonary cases). While culture negative pulmonary TB can be an indicator of early disease,(31) the concurrent ndings of high sputum smear positivity and high proportion of cases with moderate-advanced disease on chest x-ray, suggest the low culture positivity rate may be related to sub-optimal test sensitivity in this setting, noted in Kota Kinabalu previously. (32) The low coverage of GeneXpert® and high proportion of culture negative or not tested cases, raises concern of drug-resistant TB being missed.  (12) are considerably higher than what we observed in Sabah. Fast tracking the scale up GeneXpert® coverage would not only aid in improving con rmation of drug-sensitive TB, but help ensure that rifampicin resistance is detected. While current cases of MDR-TB are low, the lack of testing limits our con dence in whether all cases are being detected. Adding to this concern are the poor treatment outcomes recorded among MDR-TB cases. Given one-third of all MDR-TB cases stopped or failed treatment, there is a risk that these cases may continue to be infectious and contribute to primary transmission of drug resistance in Sabah. Efforts to improve treatment outcomes among MDR-TB cases, such as through improved treatment supervision and monitoring, consideration of new all oral or standardised shorter regimens, and patient support measures should be prioritised. (33) Conclusions This study highlights the value of detailed sub-national review of TB noti cation data to evaluate programmatic outcomes. The TB burden remains disproportionately high in Sabah, compared with peninsular Malaysia. Within Sabah, geographic hotspots of TB correspond with migration hubs and dense urbanised population centres. Further micro-level analysis of the epidemiology and service utilisation within these centres may assist in formulating strategies for targeted active case nding activity. However, given the evidence of late diagnosis of TB, particularly among non-citizens, priority should be given to ensuring universal health coverage and removal of any social, nancial, legal, linguistic or other barriers to care. There is an urgent need to scale up use of GeneXpert® in Sabah. This would help improve the diagnostic gap in children and ensure that rifampicin-resistance is detected.