Tuberculosis diagnosis cascade in Blantyre, Malawi: a prospective cohort study

Tuberculosis (TB) is the leading infectious cause of death worldwide and an estimated 10 million people developed TB disease in 2019 [1, 2]. TB control relies on early diagnosis and treatment, as reflected in the World Health Organization (WHO) End TB 2025 target of ≥90% of people who develop TB being notified and treated [3]. To achieve this the WHO recommends systematic TB screening for priority risk groups in order to reduce poor disease outcomes and TB transmission [4]. These recommendations are reflected in many National TB Programme (NTP) guidelines [5‐8].

The TB care cascade model assesses patient progression and retention through sequential stages of care required to achieve a successful treatment outcome [9], in order to quantify gaps in care delivery and adherence to guidelines. Care cascades have been extensively used to evaluate HIV care delivery [10], but have only recently been applied by TB programmes [9] to expand analysis beyond standardised treatment outcome reporting [11] and ad hoc diagnostic pathway analysis [12].

In Subbaraman et al’s generic model for a care cascade for active TB the first gap is identified as not accessing a TB diagnostic test [9]. This first gap is repeatedly the largest in many settings [13, 14], in keeping with the numerous issues relating to sputum-based tests [15].

Recent studies have emphasised variability in TB diagnosis cascades in high burden countries. In India, only 12–17% of standardised patients were correctly asked to test for TB [16], whereas in Nairobi, Kenya [17] this was 50% and a systematic review found a range for all patients from 4% in Thailand to 84% in South Africa [18]. However, most high-burden TB countries, do not routinely collect data to estimate adherence to systematic TB screening guidelines in health facilities.

WHO recommends people living with HIV are systematically screened for TB each time they visit a health facility and that in high-burden TB settings systematic screening for TB in other selected high risk groups may also be appropriate [4]. These risk groups include older people and those previously treated for TB. However, an estimated 29% of new TB cases are still not identified or officially notified, partly due to failure to diagnose active TB in people accessing healthcare [1, 19]. Examining TB test access for these risk groups and subsequent steps in the TB diagnosis cascade will be critical for efficient TB programme design.

The aims of this study were to: construct a TB diagnosis care cascade; describe the proportion of “clinically-indicated” patients (defined by the Malawi National guidelines [5]) who progressed through each step of the diagnosis cascade in a primary care clinic; and investigate factors associated with being offered a TB test.

This study found that same day sputum submission for TB testing following all steps of the diagnosis cascade was achieved for only 4.7% of participants among whom sputum testing was indicated according to Malawi national guidelines, with patients lost at every stage of the TB diagnosis care cascade. Failure to request sputum by clinicians despite elicited symptoms led to the biggest single gap in the diagnosis care cascade, followed by not asking about symptoms. This suggests that: interventions focusing on health worker behaviour may have the greatest potential for retaining presumptive TB patients within the diagnosis cascade; there appears to be inconsistent application of guidelines and infection control practices; and that we must formalise and strengthen reporting on the early steps in the TB care cascade. Additional important epidemiological groups such as men [1] should be given equal priority to PLHIV within national TB guidelines. However, if guideline adherence is improved, novel high-throughput triage testing approaches will also be needed to reach the required capacity.

Adherence to sputum-request guidelines in 4.7% (21/445) of patients sits at the bottom of the range (4–84%) identified in a recent systematic review [18]. When taken together with a TB treatment initiation rate of 85–94% [25] and TB treatment success rate of 82% in Malawi [19], our data suggests that the overall TB cascade in Malawi is more similar to that for India than that for South Africa. In India gap 1 (did not access a TB diagnostic test) accounted for 50% of all patient losses, whereas in South Africa, low treatment success led to the largest gap in the cascade [9, 13, 16].

To reduce these substantial gaps in accessing TB tests a multi-faceted approach is required to identify logistical barriers and change health worker behaviours. Facility-based screening relies on health worker behaviour (asking about symptoms and requesting sputum) which leads to the biggest gaps and therefore offers the greatest potential for improvement. Suspicion of malaria or bacterial investigations may contribute to not requesting sputum [26] but further investigation is needed to confirm what structural factors drive health worker behaviour.

This study demonstrates a low level of adherence to National TB Programme guidelines. This is the case even with groups identified as high risk within both the Malawi and WHO guidelines, such as those who have previously had TB and PLHIV. Health workers operate in challenging conditions with average patient consultation times < 3 min [27], a high turnover of staff and regular supply stock outs [28]. As such, measures undertaken to improve adherence to guidelines and increase the proportion of clinically-indicated patients who access TB tests need to be pragmatic. Strategies such as FAST - Finding TB cases Actively, Separating safely and Treating effectively – [29] are effective in increasing testing and infection control not only for TB but also other respiratory infections. In Malawi, some elements of FAST, such as cough monitors, have been inconsistently implemented, due to limited availability of resources. However, our analysis shows the large gap in cough and symptom enquiry that could be met by universal cough monitors. Implementing strategies such as FAST consistently is critical for all low and middle income countries (LMICs), especially in the midst of the COVID-19 pandemic.

In addition, enhanced monitoring and central collation of data are essential to tracking individual clinic performance. Malawi, as is typical for LMICs, collects and reports comprehensive data on TB case notification and treatment success at clinic level, but only reports the number of TB tests per facility per quarter, without further diagnostic steps. A WHO recommendation to report numbers of screened presumptive TB cases, disaggregated by age, gender and HIV-status globally, would allow greater focus on the earlier steps of the TB care cascade.

Despite TB prevalence in men being over twice as high as among women in LMICs [30] and in Malawi a ratio of male to female cases of 1.5 [1, 19], in our study sex was not associated with being requested to submit sputum. In Malawi, the ratio of prevalent-to-notified cases of TB – an indication of how long patients take to be diagnosed - is 1.5 times higher among men than women [30]. Men should, therefore, be considered as much of a priority group within TB guidelines as PLHIV in countries with a high male-to-female case ratio. Notably, men are less likely than women to seek health care early on in their illness [31], making it critical to manage them efficiently when they do present to a facility.

Finally, if all patients attending the outpatient clinic were screened for TB as per the guidelines, the current Xpert facilities would only be able to process up to a half of the required samples. It is unknown to what extent this lack of diagnostic capacity may influence test decisions among the health workers. If guideline adherence and increased identification of presumptive TB patients is subsequently improved a novel high-throughput approach to triage testing using new diagnostics (e.g. computer aided diagnostics for X-rays) will also be required for LMICs to increase capacity [32, 33].

Study limitations include the single site nature of this study, limiting generalisability, although the study site is typical of urban primary care clinics in Malawi so likely representative of primary care in the areas with highest TB burden in the country [21]. Due to limited research staff capacity we interviewed only 44% of clinic attendees with men and those with TB symptoms more likely to complete an exit interview (Suppl Table 1), potentially resulting in selection bias and overestimation of the proportion who are clinically-indicated. However, this is mitigated by high participation in those approached. Symptoms, HIV status and testing practices were self-reported, potentially resulting in social desirability bias in measurement of these variables, with HIV-positive status under-reported by up to 40% in Malawi [34] and extensive dual HIV and TB stigma [35]. Although healthcare workers would be dependent on the same self-report of TB symptoms to assess eligibility for sputum testing and the proportions progressing through each step of the diagnosis cascade were similar for all HIV status groups. In addition, 69.5% of patients asked to submit sputum were not clinically-indicated to do so as per our definition and 30.8% of those had no reported TB symptoms at all – it is unknown why sputum was requested from these patients.

Same-day sputum submission for TB testing following all steps of the diagnosis cascade was achieved in only 4.7% of those clinically indicated. Requesting sputum after eliciting symptoms is the key point of the cascade to intervene. Interventions are needed to optimise TB screening guidelines, formalise reporting, increase guideline adherence and improve diagnostic capacity, in order to reduce the most significant gaps early in the TB care cascade and to reach the required testing capacity to meet the WHO End TB goals.