Pulmonary tuberculosis in outpatients in Sabah, Malaysia: advanced disease but low incidence of HIV co-infection

Tuberculosis (TB) is a disease of major public health significance in Malaysia, with a current estimated incidence of 80 new cases annually per 100,000 population [1]. While important successes have been achieved in TB control nationwide [1,2], TB control has proven to be more challenging in Malaysian Borneo. The state of Sabah in eastern Malaysia on the island of Borneo contains approximately 10% of the country’s total population, but a disproportionally high burden of the country’s TB cases, estimated at 20-30% [3,4]. The case notification rate for Sabah has been sustained at higher levels of 144 to 217 per 100,000 population during recent decades while rates have fallen in other parts of Malaysia [5].

Reasons for the greater difficulty in achieving TB control in Sabah are unclear. The extent to which the local epidemic might be driven by HIV, migration from regional high-burden settings, drug resistance, delayed diagnosis, high smoking rates, diabetes or other factors, has not been clearly elucidated. Our own data indicate that drug resistance in the outpatient setting is uncommon (publication pending); this is supported by data from the local laboratory [4]. Nations which are close neighbours of Sabah have higher reported TB incidence than Malaysia (Philippines: 265/100,000; Indonesia 185/100,000 [1]), but the extent to which immigrants contribute to local TB case notifications is unclear. Migrants are thought to comprise approximately 27.8% of the Sabah population [6]. Regarding HIV as a contributor to TB rates, the rate of HIV co-infection in people with TB appears relatively low by international standards for Malaysia as a whole at 6.1% [1], with HIV status established in a high proportion of TB patients (97%) [1]. Rates of HIV among TB patients have not been published from Eastern Malaysia, but were found to be 7.7% in a 2012 study of 530 people in Western Malaysian hospitals [7]. Our unpublished data from the largest hospital in Sabah show an average of 63 new diagnoses of TB-HIV co-infection and 15 deaths annually between 2010 and 2013. Therefore although we recognise TB-HIV co-infection as a serious problem in the hospital setting, accurate data on TB-HIV co-infection in Sabah are lacking.

HIV point-of-care (POC) blood tests have been an essential and cost-effective component of HIV care in outpatient clinics, especially in resource-limited settings [8] but increasingly also in higher income settings [9]. A large number of manufacturers provide low-cost HIV POC tests. Such tests usually have reported sensitivity of >95% and specificity >99% under laboratory conditions [10]. The World Health Organisation (WHO) recommends a minimum HIV test sensitivity of 99% and specificity of 98% [11]. Problems with POC tests can exist however, including inadequate sensitivity [12], inadequate specificity [13,14] and faulty batches [15]. The simplicity of the tests also can provide a false sense of assurance about requirements for quality control measures and staff training. Lowest-cost tests can be especially appealing to budget-conscious decision makers but may underperform.

In order to better understand the factors contributing to high ongoing TB rates in Sabah, our aims were to determine epidemiological characteristics and HIV status of patients with tuberculosis at a large outpatient clinic in Kota Kinabalu Sabah. Additionally, we sought to test the sensitivity and specificity of the HIV POC test kit stocked by the TB clinic against another used-widely POC test. In the absence of readily-available onsite mycobacterial culture, we restricted the study to smear-positive pulmonary TB patients is order to include only those most likely to have true TB.

During the 2-year study period from July 4^th 2012 to July 3^rd 2014, 176 consenting participants were enrolled out of 395 people referred. 79 were ineligible, 12 were considered unsuitable by the research staff (e.g. too unwell or pregnant) and 128 did not consent. Chest x-rays were available for review in 145 people.

We have found that TB patients in Kota Kinabalu have advanced disease at diagnosis, long diagnostic delays, low HIV co-infection rates, high smoking rates among males, and that migrants may be over-represented. These findings provide important insights to guide local TB control efforts. Although Malaysia is an upper middle-income country which has achieved impressive TB control at a national level, our study participants had severity measures similar to or worse than we have previously found in comparable patients in a low-resource Asian setting [20]. Forty six percent of participants in the current analysis had haemoptysis, nearly a third had a microscopy grade of 3+, and two thirds had cavitary disease. Although recall bias may influence people’s reported symptom duration, the median reported diagnostic delay of 8 weeks is 4 weeks longer than has been reported on average from elsewhere [21,22], and was associated with worse disease.

The presence of advanced disease suggests that earlier detection of pulmonary TB must become a priority for local healthcare services. This could be achieved through active case detection efforts, improved contact tracing, and a lower clinical threshold for considering TB, especially targeted at apparent geographical hotspots or at-risk communities. Mathematical modelling has indicated that control efforts targeting TB hotspots can be an efficient strategy to reduce wider TB rates [23]. Improved TB control among the migrant population group would have major individual and public health benefits, and impact favourably on Malaysian national TB statistics. Other Malaysian studies have estimated the proportion of cases occurring in migrants as being substantially fewer, at 10% in a 2004 study [2] and 17% in a study from 2000–2007 [24]. It makes sense that as TB control in Malaysia continues to improve, an increasing proportion of total TB cases will occur in people from neighbouring high-burden settings. In Australia for example, >85% of all TB cases now occur in migrants or visitors from high-burden countries [25].

Implementation of public policy to reduce smoking rates is greatly needed. Smoking increases the risk of TB infection and disease [26,27]. In this study, 51.1% of participants were current or ex-smokers, with many of the ex-smokers having quit only since the onset of respiratory symptoms. By comparison, in a recent study from Mexico, 24.4% of pulmonary TB patients were current smokers or smokers within the last 12 months [27]. Policy changes such as increased cigarette prices [28] are needed in Malaysia and other Asian nations in which smoking consumption rates remain very high among men, and are increasing among women [29]. We found that some patients had quit smoking due to pulmonary irritability (i.e. cough), sometimes at a date slightly before the time that they recalled the onset of TB symptoms. This raises the question of whether smoking cessation due to respiratory symptoms could be an early sign of tuberculosis in a high-burden setting, and hence whether smoking cessation due to increasing respiratory symptoms should prompt investigation for pulmonary TB.

Our data indicate that caution is required in the use of low-cost point-of-care HIV tests, which should be carefully evaluated before they become adopted as national standard of care. The low number of HIV positive results meant that test characteristics could not be ascertained with confidence. However, one instance of HIV positivity according to Determine™/EIA was negative on the clinic-provided test, raising major concerns. We had intended to test the sensitivity and specificity of one POC against Determine™ but the test stocked by the clinic changed throughout the study period, depending on clinic purchasing patterns which we could not influence. Our experience with low-cost rapid tests in clinical practice in Kota Kinabalu has included other instances of inadequate sensitivity, including in antenatal screening. We therefore recommend discontinuation of use of low-performing tests. A limitation of this study was that we did not use a gold standard comparator HIV test, although positive results were cross-checked with methods considered as gold standards. A study in Uganda has identified problems with low specificity (false positive results) with the Determine™ rapid tests [14], which we did not observe. One reason for lower performance of POC tests in field settings compared with manufacturer-provided laboratory data may be the temperature at which tests are often stored and used. Routine practice at our study site was for the clinic’s test kits belonging to be kept in the general clinical area which is not air conditioned, where the ambient temperature is often ≥25°C. A transition away from the traditional testing location of the laboratory to point-of-care testing in the clinic must be accompanied by training of clinical staff in the principles of laboratory quality control procedures.

It is welcome news that high ascertainment of HIV status was achieved, and that the prevalence of HIV co-infection among smear-positive TB patients in the outpatient setting was low at 1.7%. Comparison can be drawn with nearby eastern Indonesia (Papua Province), where the HIV rate among TB smear-positive outpatients is 13% [30]. However, one TB-HIV co-infected patient died, and barriers to institution of antiretroviral treatment were faced in another. Barriers are well described regionally [30] and globally [31]. Despite the reassuringly low community HIV rate we found, our clinical experience in the hospital setting in Kota Kinabalu is, contrastingly, of a major and rising problem with HIV and TB-HIV co-infection. Unpublished 2013 data from our Queen Elizabeth Hospital clinic, the main referral centre for HIV management in Kota Kinabalu, show 253 new HIV diagnoses. Seventy seven had TB co-infection, and 12 died in 2013. Thus there is no scope for complacency.

Limitations of this study are that participants may not be representative of Malaysian Borneo TB patients as a whole. Males were over-represented, but male over-representation among PTB patients is not uncommon in Asia [32], for a variety of reasons reviewed elsewhere [33]. Our selection criteria (smear-positive outpatients) would have selected for a group less likely to have HIV. Reluctance to have an HIV test was not cited as a reason for non-consent – people not enrolled in the study also had HIV testing performed as part of routine clinical care. Non-HIV co-morbidity data were based on patient report only; hence co-morbidities were likely to be under-estimated. Determination of true ethnicity can be hampered by a reluctance to admit to non-local status, partly related to fears about health care costs for non- residents, or deportation; we reassured people that TB treatment is free regardless of ethnicity, and ethnicity would remain confidential; also, we did not rely on stated ethnicity alone, as described in the Methods.

A somewhat unexpected finding was that people with co-morbidities had less-severe disease. This might be because people knowledgeable of their co-morbidities may be those more likely to seek health care earlier, and to be engaged in health care services. Just as TB is an important opportunity to detect HIV, so too is it an important opportunity to detect diabetes. Screening for diabetes could be an important additional strategy to add to TB care in Kota Kinabalu, particularly in older individuals with higher BMI. The finding of more advanced disease in males might also have been related to their health-seeking behaviour; alternatively their high smoking rates may have contributed, although we could not find a statistically significant association between smoking and severity per se.

In summary, the low rates of TB-HIV co-infection we identified provide optimism for local TB control. We recommend that careful quality control procedures be applied to HIV point-of-care testing, and that purchasing of test kits be guided by evidence of adequate test characteristics in keeping with WHO recommendations. Our findings have prompted changed local practices regarding HIV test kit use, and improved communication between the TB and HIV clinics. We suggest that a cost effective strategy for controlling local TB rates would be a greater focus on geographical TB foci comprising socioeconomically disadvantaged migrant populations in Sabah. Since TB rates in eastern Malaysia have a disproportionate influence on national rates, successes in Sabah could translate into a fall in national TB rates. Local operational research in strengthening TB control is required.