**4. Discussion**

This is the first study in Nairobi, Kenya, to compare TB case detection, TB treatment outcomes, HIV testing and referral to ART on a month-by-month basis between the COVID-19 and pre-COVID-19 periods. There were three important findings, which require explanation and interpretation.

First, with respect to TB programme activities, there was considerable variation in TB case detection numbers during the pre-COVID-19 period. Numbers with presumptive PTB increased between April and July 2019 (the summer months), and this is in line with what usually happens each year at this particular time, both at national level and in the city county of Nairobi (source: NTLD-P). From July to October 2019, however, there was widespread strike action among the healthcare workers which severely affected health services, including TB case finding which declined dramatically. When the strike action was over, numbers with presumptive TB did not increase. There were two possible reasons for this that included (i) TB case numbers between 2018 and 2019 already being in decline, and this may have been part of the process, and (ii) community confidence in the health sector being adversely affected and taking time to pick up. In contrast, numbers with registered TB did increase, showing that the health services had returned to some degree of normality.

In spite of these pre-COVID-19 variations, there was still an overall significant decline in numbers presenting to the health facilities with presumptive PTB and in numbers being diagnosed and registered with TB in the COVID-19 period. The expected upward trend in TB case finding from April to July 2020 (the summer months) did not happen. Children and women were particularly affected, maybe because there was more difficulty for mothers and children to move around during lockdown and a heightened sense of fear for the family about accessing health facilities. The national lockdown and the ongoing challenges after lockdown, which included health facility closures and strike action in the health sector, posed ongoing problems. While numbers with presumptive TB picked up in the second six months of the COVID-19 period and particularly in the last month of the study, numbers with registered TB remained below pre-COVID-19 levels throughout this time.

These findings of declines in TB case detection are similar to reports from other health facilities and clinics in Nigeria [24], Brazil [25], China [26] and India [27]. Our study and these clinic reports align with a recent report from WHO showing an overall 21% shortfall in TB case notifications in 84 countries in 2020 compared with 2019 [28]. A modelling analysis at the start of the pandemic suggested that a three month suspension of TB services due to COVID-19 lockdown followed by ten months restoration back to normal would cause over five years an additional 25,000 TB cases and 12,500 TB deaths in Kenya, mainly as a result of the accumulation of undetected TB during lockdown [29]. A further modelling study in high-burden, low- and middle-income countries predicted a 20% increase in TB mortality, the greatest impact coming from reductions in timely diagnosis and treatment of new TB cases [30]. However, these modelling studies were done early on in the pandemic when there was hope that service disruption would be temporary. The reality is that service disruption has continued throughout the year, and has been further exacerbated by industrial action in the health sector, health facility closures and continued community fear and stigma about COVID-19.

The city of Nairobi, therefore, has done well to introduce an array of innovative measures such as integrated screening for TB and COVID-19, TB self-testing, active case finding and contact tracing to mitigate the challenges. These seem to have paid off with improvements in case detection and diagnosis in the later months of the COVID-19 period. The measures undertaken in Nairobi are in line with those recently recommended by the Stop TB Partnership [31], which include screening patients with respiratory symptoms for both TB and COVID-19, creating, developing and supporting networks of TB survivors and TB communities and implementing real-time surveillance data.

Second, it was encouraging to see that while TB treatment success initially decreased, it then picked up to between 70% and 80% in the last four months of the study. At the start of the study, we were concerned that COVID-19 restrictions would prevent patients from collecting anti-TB medications, compromise drug adherence and reduce the ability of TB programme staff from obtaining information about final treatment outcomes. TB patients with associated COVID-19 coinfection have an increased risk of mortality [32,33]. We were worried that undetected COVID-19 in our TB patients might increase TB deaths during treatment. In the event, however, there was a minimal increase in the risk of death. The TB programme implemented patient-centered measures to help patients comply with treatment and adhere to medication, and attention was paid to reducing the "not evaluated" treatment outcome. All these measures were associated with improvements in treatment success.

Third, with respect to HIV services, there was already a marked decline in HIV testing numbers at the health facilities in the pre-COVID-19 period. While the health sector strike action between July and October 2019 played a part, there were additional explanations. HIV testing had been institutionalized in public health facilities over the years and many people visiting these facilities had already been HIV tested. NASCOP had started promoting HIV-self testing, only doing confirmatory tests in health facilities for those who had tested HIV-positive. The country was also moving towards a more targeted HIV testing approach directed at high-risk groups as recommended for Sub-Saharan Africa as a whole and for Kenya [34,35].

Despite the pre-COVID-19 decline, HIV testing numbers in health facilities remained low during the 12 months of COVID-19. The reduction in numbers is similar to the reductions in HIV testing volumes that have been documented in Europe, the USA and other countries in Africa [36–39]. The decrease in HIV testing threatens access to diagnosis and treatment, resulting in excess HIV-related deaths and ongoing transmission of HIV in the community. HIV testing started to improve in the second six months, perhaps due to implementing community outreach services for clients and the promotion of assisted partner testing as recommended nationally [40]. It was encouraging to see that referrals to ART were maintained above 90% over the 12 months, with a slight increase observed during the COVID-19 period.

There were several strengths to this study. First, we embedded monthly surveillance within the routine services of the health facilities. Second, there was cross-checking and validation of monthly data between the country coordinator and the overall study monitoring and evaluation officer. Third, we used two 12-month periods to compare data, which enabled us to account for any seasonal changes that might affect access to health facilities. Finally, the conduct and reporting of the study were in line with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [41].

There were, however, some limitations. Our study was limited to health facilities in Nairobi, and therefore may not be representative of Kenya as a whole. The use of aggregate data meant it was impossible to fully understand the individual cascade of care for TB case detection and HIV testing. We only assessed referral to ART and did not measure initiation and retention on ART. Previous studies have suggested that ART interruption has been a problem during the COVID-19 pandemic [42], and it would have been interesting to assess this in Kenya. We also did not assess treatment outcomes of patients with drug-resistant TB, but these patients would have been referred to other centers where the ability to track outcomes might have been difficult with COVID-19 restrictions in place. Finally, the official monthly reports to WHO of COVID-19 cases and deaths may have underestimated the actual burden, and therefore impact, of COVID-19 in the country. A seroprevalence study of anti-SARS-CoV-2 antibodies in Kenya between

April and June 2020 estimated a crude national seroprevalence of 5.6%, with levels being highest in the counties of Mombasa (8.0%) and Nairobi (7.3%) [43]. A study on deceased people at the University Teaching Hospital morgue in Lusaka, Zambia, found that only 6 (9%) of 70 people, who had SARS-CoV-2 confirmed from postmortem nasopharyngeal swabs within 48 hours of death, had ever been tested before death [44]. These data sugges<sup>t</sup> that there are many unreported cases and deaths due to COVID-19, a situation likely to be present in many other African countries.

Despite these limitations, there are some important programmatic implications from this study. First, the strengthened monthly surveillance system worked well with the disease control programme heads appreciating and closely reviewing the monthly reports. Although cause and effect are difficult to infer with this type of study, the frequent access to data may have helped with the implementation of interventions on the ground to counteract declining trends in TB case detection and treatment outcomes and HIV testing. However, to continue with this system requires funding and resources, which would need to be obtained.

Second, with COVID-19 likely to become endemic, TB and COVID-19 care and treatment programmes need to think about further integration with respect to screening, laboratory infrastructure and diagnosis, contact tracing and sound infection, prevention and control measures, especially in health facility settings [45]. Resources permitting, more consideration should be given to the use of digital platforms to facilitate case finding and treatment [46]. Kenya also has a large private sector, which in the case of TB serves 20% of all patients registered and treated in the country [47], and this must not be forgotten when it comes to integration and innovation. HIV self-testing would undoubtedly fill a significant gap in clinic-based HIV testing at a time of crisis such as this [34,35], but more attention needs to be paid at the health facility level to recording and reporting on numbers self-tested along with the results.

Finally, research into improving the responses to TB and HIV/AIDS amidst the COVID-19 pandemic must continue [48]. This needs to be mainly focused on safely and effectively delivering the key programme activities and ensuring that patients can more easily access diagnostic, treatment and prevention services.
