Reducing delays to multidrug-resistant tuberculosis case detection through a revised routine surveillance system

Tuberculosis (TB) is the highest single cause of death worldwide from an infectious disease and continues to be a major public health problem [1]. The World Health Organization (WHO) estimates, 10.4 million people contracted TB in 2016 of which 4 million of these cases were undiagnosed. 480,000 were estimated to be cases with TB that is resistant to at least Isoniazid (INH) and Rifampicin (RIF), with or without resistance to other first-line drugs (MDR-TB) cases of which 6.2% were Extensively Drug-Resistant TB (XDR-TB). 1.3 million people were estimated to have died of the disease [1]. In a resource-constrained setting, diagnosis of MDR-TB can take many weeks caused by long delays in starting and completing diagnosis with limited diagnostic facilities that can detect drug-resistance [2]. Delay in diagnosis and treatment has serious consequences for disease control at both the individual and the community levels [3]. The strength of the TB laboratory network is often a direct reflection of the success of the TB Control Programmes [4]. Poor TB case detection and rising TB drug resistant are in part the result of historically neglected laboratory services, slow technology transfer, and a lack of new more accurate TB diagnostic tools [5]. The WHO recommends rapid and sensitive diagnostic methods that provide information on drug resistance (i.e. methods such as GeneXpert (Xpert MTB/RIF) and Line-Probe Assays (LPA) [6]). The LPA Genotypic (molecular) tests for identification of isoniazid and rifampicin resistance to first-line anti-TB drugs [7], and Genotype® MTBDRplus assay have shown promise for the diagnosis of drug resistant TB for the second line (SL-LPA) anti TB drugs. These tests can be performed in a single working day and detect the presence of mutations associated with drug resistant TB [8, 9]. Xpert can be used as a point of care technology in some contexts and can reduce the time between specimen collection and diagnostic result [10].

A successful continuous Routine Surveillance System (RSS) is an important element in the process of detecting and monitoring drug resistant TB. The existing RSS of the National TB and Leprosy Program (NTLP) in Tanzania shown in Fig. 1, specifies sputum specimens should be submitted to the Central TB Reference Laboratory (CTRL) for culture and drug susceptibility testing (DST) for 25% of all new TB cases and 100% of previously treated TB cases [11]. However, these percentages for DST at the CTRL have historical not been achieved. Indeed for previously treated TB only 61% of cases were sent for testing even in the best year of the three-year period investigated, suggesting cases of drug resistance are likely to have been missed [12]. This study focused on the previously treated TB cases because that is where most MDR TB cases are found [13, 14]. Failure to appropriately track these cases could lead to the emergence of XDR-TB. Qualitative findings showed that the system of TB specimen transportation in Tanzania was a major problem in remote health facilities as there were no reliable and frequent means of transport [12]. The existing RSS was underperforming in relation to the feedback and communication of the drug susceptibility testing results from the CTRL to the peripheral health facilities [12]. Based on these findings, in this study a revised routine surveillance system was designed and piloted in Mwanza region, Tanzania. The design took account of the weaknesses identified from the earlier qualitative study [12] and changes in the availability of molecular diagnostic techniques. A ‘before and after’ comparison was conducted of the performance of the RSS. Measurements were taken to assess the effect of the revised RSS on the number of specimens received and tested for drug susceptibility, the time for specimen transportation and results feedback, and the number of MDR-TB cases detected.

Consequently, a study designed to understand the potential effects of a revised routine surveillance system for previously treated TB cases on delays and level of MDR-TB diagnosis by piloting the revised approach in Mwanza region, Tanzania.

The study involved the design and piloting of a revised RSS in Mwanza region. The revised RSS aimed to address many of the issues raised in the earlier qualitative study [12], as well as taking into account the wider availability of molecular testing. The primary outcomes of the study were the impact of the revised RSS on specimen transit time and the time for drug susceptibility results to be sent back to the requester (turnaround time). The pilot of the revised RSS showed significant improvements compared to the current RSS. In particular, reduced transit times and turnaround times for testing which would be expected to lead to drug resistant TB patients starting treatment earlier as well as an increase in the number identified (in this study by around 67%). We observed the volume of previously treated specimens received at the CTRL more than doubled (185 in 2017/18 compared to 75 in 2016/17). The use of local Xpert sites reduced times for specimen transportation in Mwanza region for both sputum and isolates to less than 3 days in most cases.

The transit time recorded for both sputum and isolate specimens in 2017/2018 showed a median of 10 days (IQR 6, 15) with 21% over 21 days. In comparison for sputum specimens in 2016/2017, the median was 12 days (IQR 3, 54) with 44% taking more than 21 days. Unfortunately, due to poor data recording in 2016/2017, the turnaround time is only based on 8 samples where the median turnaround time was 62 days when solid culture was used. By comparison, the use of molecular techniques in 2017/18 saw the turnaround time median drop to 7 days.

In addition, the revised RSS led to 93% of the specimens received at the CTRL being examined in comparison to just 64% in 2016/17 before the revised RSS was implemented. This was supported by a reduction in missing information on laboratory request forms.

Social media played an important role in this study. Various groups created by the NTLP Manager such as Vitendanishi for lab supplies, and the Mwanza family created by the RTLC in Mwanza to enhance timely communication with laboratory personnel, and the DTLC in Mwanza. These groups allowed the ground staff responsible for sending out specimens to resolve concerns timely and eased results dissemination consequently shortening both transit and turnaround times. The platform was replicated in other regions among DTLCs, RTLCs and laboratory teams. Use of WhatsApp groups may also be a useful approach to address other communication issues in the TB programme and beyond.

The study had some limitations due to poor record keeping in the historical data which made some comparisons of performance difficult. However, to a certain degree, this poor record keeping was also symptomatic of the poor performing RSS. The revised RSS record keeping has improved and this is an additional benefit assuming it can be sustained. The evidence from this study shows why this is important as it can have a direct effect on patients getting appropriate treatment and in a timely manner. There was great variability by month in the number of specimens transported for testing (Fig. 4) this was mainly due to operational reasons, for example towards the end of May 2017, there was a nationwide employees’ education certificates inspection for irregularities which led to the termination of several laboratory staff and clinicians. This exercise had a negative impact on the study. In addition, in December 2017 and February 2018, fewer specimens were received which might have been due to staff shortages and some being on holidays leading to some specimen batching.

It is clear from this study there is more to do. For instance, even with the focus of the revised RSS, 13% of laboratory request forms had missing laboratory reference numbers and 3% missing addresses, although much lower than in 2016/17. Similarly, there were significant numbers of specimens in 2017/18 that were not processed for a variety of reasons (i.e. in total 75 out of 472, 15.9%) – see Fig. 3, however, again this is much lower than in 2016/17. Moreover, in the National Strategic Plan (NSP) it was estimated that the MDR TB burdens were 699 and 725 for 2016 and 2017 respectively. The targets set for the region were 349 and 435, but the actual drug resistant TB cases notified were 197 and 200, which implies there remains a large detection gap [17‐19].

The study focussed on previously treated cases only, all the interventions looked at could have equal value to new cases, though the level of drug resistance would be expected to be much lower.

In conclusion, the revised RSS led to an increased number of specimens received and tested at the CTRL. The use of social media within the NTLP network led to close follow-ups and timely response to concerns during the piloting. The revised system was shown to reduce delays in diagnosis and increase in drug resistant case detection. The shorter transit times and turnaround times are important in the diagnosis of MDR-TB and TB control. These positive results suggest a larger scale study involving more regions should be considered to determine whether these benefits are robust and sustainable across similar settings.