Screening and testing for tuberculosis among the HIV-infected: outcomes from a large HIV programme in western Kenya

In 2016 there were an estimated 10.4 million incident tuberculosis (TB) cases worldwide, 10% of whom were people living with HIV (PLHIV) [1]. About 1.7 million deaths were attributed to TB, with more than 374,000 of these occurring among PLHIV. About one-third of deaths among PLHIV are attributed to TB [2].

Global TB incidence rates have been falling since the year 2000. However, this decline has been slow, estimated at 1–2% per annum. If the decline in incidence continues at this rate it will not be possible to reach the 80% reduction in TB incidence that is necessary for meeting the End TB Strategy target and the Sustainable Development Goal of Ending the TB epidemic by 2030 [3].

Kenya is among the top 14 countries worldwide burdened by TB, multidrug-resistant TB (MDR-TB) and TB/HIV [1]. With 1.6 million PLHIV, Kenya is also among the top countries globally with the highest numbers of HIV-infected individuals [4]. A recent nationwide prevalence survey showed that there were an estimated 169,000 incident TB cases in the country in 2016 [1], 17% of which were among PLHIV. However, the country notified only 77,376 cases [1], representing a gap of about 54% who were either undiagnosed or unreported to the national TB programme. This has necessitated a strategic shift in focus to ‘finding the missing cases’.

HIV infection increases the risk of TB disease at least 20-fold [2]. The risk remains substantially elevated even after immune reconstitution with antiretroviral therapy [5]. There is thus need for a collaborative approach to identify and manage TB in this population. One of the pillars of such an approach is intensified case finding (ICF) [6]. ICF can be defined as “the systematic identification of people with presumptive active TB, in a predetermined target group, using tests, examinations or other procedures that can be applied rapidly” [7]. The key to effective ICF is the systematic application of the active screening for TB disease in the predetermined high-risk group by the health workers rather than only screening in response to a specific request or complaint by an individual seeking care. The latter (termed passive case finding) has consistently resulted in over three million infectious TB cases per year being missed by the health system [8]. In a variety of settings ICF strategies, coupled with other routine TB control activities, have been used to reduce TB prevalence, incidence and mortality [9].

With respect to HIV, ICF is the systematic screening for tuberculosis at every encounter in those infected with HIV, are at high risk of HIV, or live in congregate settings [10]. Although different screening strategies yield varying results in different groups of people [8, 10–17], in PLHIV, ICF can result in a high yield with the number needed to screen to identify one case of TB being 10 in high TB incidence countries [12].

PLHIV should be routinely and systematically screened for TB at every clinical encounter using a World Health Organization-approved four-symptom questionnaire inquiring of the presence of cough of any duration, fever, weight loss, and night sweats [18]. A positive response – defined as the presence of one or more of these symptoms – should lead to further confirmatory diagnostic evaluation for TB by Xpert MTB/RIF (Cepheid, CA, USA) [19–21]. Other possible investigations include sputum smear microscopy, culture for Mycobacterium tuberculosis and chest radiography [22–24]. Those found by testing to have TB are initiated on appropriate TB therapy while those without TB may be considered for isoniazid preventive therapy [25].

In 2014, only seven million PLHIV were reported to have been screened for TB globally [26], a dismal figure compared to the burden of HIV. Whenever screening for TB is done, quality may not always be guaranteed. Given that mortality among PLHIV during TB treatment is high – and mostly occurs within the first two months – missed and delayed diagnosis play contributory roles [27–29]. While different screening and diagnostic strategies vary in their sensitivities and specificities, [18, 28, 29] the applications of those which are especially human-dependent (like symptom screening questioning) are more subjective. Responses to such questions may depend on language, how they are framed, patient current psychological status and existing rapport between staff and the patient. Further, about 15–25% of PLHIV with bacteriologically-confirmed TB disease may be asymptomatic [30]. Other system- and patient-related aspects may also play a significant role.

Although it is part of routine programme performance indicators, ICF among PLHIV has not been systematically evaluated in the programme setting in Kenya. The quality of the screening process has also not been evaluated in large programmes in Kenya. Since being screened for TB does not necessarily guarantee completion of the TB diagnostic pathway, evaluation of the downstream processes are also important.

This study therefore aimed to evaluate, over a period of two years from 2015 to 2016 in a large HIV programme in Kenya; i) the screening for TB among PLHIV by encounters and patients, ii) diagnostic evaluation of identified presumptive TB cases, and iii) factors associated with PLHIV not being screened for TB during their clinical encounters.

In this study, slightly more than half of the PLHIV had documented screening for TB in all their clinical encounters. When screening was documented symptoms were reported in less than 1% of encounters and 4% of PLHIV during the study period. Among those with recorded symptoms less than one-in-ten had sputum microscopy or CXR results documented. Being on ART, having more clinical encounters, high patient volumes in a clinic and attendance at paediatric clinics increased the risk of not being screened.

The study had several strengths: it followed up a large number of patients over a considerable duration of time and the assessment was based on data obtained from the routine programme setting giving a realistic picture of the situation on the ground. These make the results easily generalizable at the programmatic level.

To ‘find the missing cases’ screening for TB should be systematically conducted in all the high-risk populations, including PLHIV. Those screening positive for any of the symptoms should be tested using appropriate tools while those diagnosed with TB be promptly initiated on appropriate TB therapy. Missed opportunities in this care cascade have been reported in other settings as well. A recent study in South Africa reported that 39% of patients with TB were not screened at the primary care setting, and among those screened, 62% were not tested for TB [40]. These screening rates were comparable to those of our study which also reported far less testing – < 10% of the symptomatic. In Mozambique, 61% of a nationally-representative sample of patients newly initiating ART were screened for TB [41]. However, this varied from 2 to 98% according to the setting. In the same study, similar to the current study, screening rates increased over the years.

The World Health Organization advises that if screening is carried out properly, between 10 and 20% of PLHIV report at least one symptom suggestive of TB [21]. The < 1% and overall 4% positive symptom screen in clinical encounters and PLHIV reported in our study, respectively is thus far lower than that expected in such a setting. Conversely, a study in a hospital HIV clinic in Ethiopia reported higher positive symptom screen at 39% [42].

The identified factors associated with higher risk of undesirable screening like being on ART, more clinical encounters/visits per patient, higher patient volume in a clinic, and paediatric clinic setup are likely more than just patient-related; they are also system- and health worker-related. A clinician is likely to spend more effort and time managing adherence and medication related issues for a patient on ART than focus on screening for TB. While some clinicians are aware that ART reduces the risk of TB disease, this is unlikely to play a major role in the lack of screening in this setting.

More visits by a patient to the clinic slightly reduced the chances of being screened for TB. While it is expected that more clinical encounters/ visits by a patient may lead to familiarity with the clinicians and the system, it has the unintended risk of skipping some processes (for example, TB screening or recording) especially if not regularly emphasized. Increased patient volume ultimately results in less time spent per patient thus affecting quality of care especially if the number of clinicians does not increase in tandem. The challenges associated with screening and diagnosis of TB in the paediatric age-group may explain the reduced screening in this subgroup.

The findings in this study have several implications: First, the study reports poor screening and testing for TB despite having structured clinical encounter forms with specific checks for TB screening, TB-related diagnostics and other aspects of care. Similar findings were reported by a clinical audit in a large UK urban clinic which found that only 6–12% of routine tuberculosis screening occurred among patients enrolled in the HIV programme despite implementing screening prompts [43]. Thus there is need for a systematic audit of the vast amount of data being generated by the AMPATH HIV programme. Such an audit should assess, but not limit itself to, i) the quality of screening for TB in the programme, ii) access to TB diagnostics including Xpert, iii) the recording processes, and iv) the overall quality of TB data in the database.

Second, there is need to strengthen the weaker links in TB/HIV integration which enhance screening and testing for TB, including documentation. These include fully adopting the ‘two diseases, one patient, one clinic, one appointment, one health worker’ approach [37]. Such an approach helps avoid patient juggling, increases knowledge and expertise of health workers in managing both diseases and improves documentation.

Third, there is need to strengthen sputum transportation system and access to Xpert testing platform including timely result feedback. Specimen transportation should be customized to the individual site characteristics and can include the use of motorcycle transporters and/ or health workers. Result relay can be further strengthened by optimizing the use of technology to transmit the results, for example, automated SMS platforms. Commodity supply and logistics must also be streamlined for full functionality.

Finally, HIV programmes should shift their focus to also include quality of the screening and attendant processes among the PLHIV. This can include regular audits as aforementioned and qualitative reviews.

The study had some limitations. We report on few co-variates likely affecting adequate adjustments for confounders. We do not, in particular, report nor adjust for direct provider-related factors like attitude and knowledge. Another limitation is the non-independence and clustering of data analyzed at encounter/ observation level. We, however, also describe the data per-patient as necessary. The binary cut-off of < 90% for undesirable screening though based on programmatic targets, is arbitrary and outcome may depend on duration of evaluation/ follow up. Finally, as a retrospective study, there were significant challenges to the data that included missing and/or inconsistent data on Xpert uptake and TB treatment – which are key to evaluating any TB/HIV programme. It is thus possible that some of our findings could be attributed to poor documentation and reporting bias.

This study found slightly more than half of PLHIV having documented screening for TB and low symptom and TB-testing recording. ART, more encounters, high patient volumes, and paediatric clinic setup reduced screening rates. Programmes should strengthen the ICF process for PLHIV and focus on the quality of the screening and recording of TB in the HIV clinics.