Like many other countries in the sub Saharan African region, Kenya is committed to reducing deaths due to tuberculosis in line with the WHO’s end TB strategy [3]. To achieve this goal, TB surveillance needs to be reliable to assist in appropriately directing resources, even in difficult to reach areas like informal settlements. The epidemiology of TB mortality in Nairobi informal settlements has evolved over the years as shown in this paper. This paper describes TB deaths that took place in the NUHDSS in Viwandani and Kochorogo in the years 2002 to 2016. The highest number of TB deaths for this period were in the age group 30 to 39 years. This is the same age group that had the highest HIV prevalence for the period of analysis [14, 15]. An HIV prevalence survey done in the same area in 2007 showed a 12% HIV prevalence in the two informal settlements, the highest prevalence being in the 30 to 39 age group [16].
Though HIV -TB co-infection data was not available for analysis, evidence shows that the risk of developing TB is estimated to be between 15 to 22 times greater in people living with HIV than among those without HIV infection [17].
There was a decline in TB deaths between the periods 2005 to 2016. Interventions targeting informal settlements in Nairobi across the years may have contributed to this decline. These include active defaulter tracing mechanisms from 2005, countrywide scale up and utilization of community health volunteers for home-based care follow-ups and community-based direct observed treatment short course (DOTS) from 2007 and creation of support groups for HIV/TB co-infected patients [18,19,20].
There was a higher proportion of TB deaths in males as compared to females in the NUHDSS within the study period. The Kenya 2016 TB prevalence survey found a similar pattern of higher TB cases in males in comparison to females [5]. Contributing factors may include poorer health-seeking behaviour among men as compared to women [21] and higher TB risk factors among men such as smoking, alcohol and occupational exposure to undetected TB cases [22]. The risk of TB disease has been found to increase by 3.3 and 1.6 times respectively as a result of alcohol use disorder and tobacco smoking [23]. Poorer health-seeking behaviour was also noted in the NUHDSS with a higher proportion of females being reported to have sought health care for their illness as compared to males. Similarly, in a study in Zambia, males were less likely to seek care for their presumptive TB symptoms [21].
Findings in this study demonstrate the crucial role of care-seeking in morbidity and mortality patterns. Nearly 43% of the deaths from TB happened at home. The immediate cause of death in most pulmonary TB patients is usually septic shock or respiratory emergencies [24] and these would require urgent hospital care. Though access to care data was not available for analysis, the high number of deaths at home could be an indicator of access to care challenges. People in informal settlements such as Korogocho and Viwandani often have health care access challenges [25], explaining why a proportion of the patients die at home.
Health-seeking behaviour may also have been affected by the low perception of the seriousness of symptoms, stigma for TB related symptoms because of its correlation with HIV, and delayed care-seeking due to poor awareness of the cardinal signs and symptoms of TB [26]. Community awareness on TB signs and symptoms and subsequent follow-up measures have a crucial role in enhancing appropriate and timely care seeking [27]. Though the informal settlements in Nairobi are supported by a strong community health strategy that consists of a network of community health volunteers (CHVs) organized in community health units (CHUs) [28], referral linkages between the CHVs and health facilities are often weak due to incomplete referrals, inadequate documentation tools as well as poor counter referral mechanisms [27]. This makes it difficult for CHVs to track back the referrals made and whether care was received. Improving the CHU linkages with health facilities may therefore have a major role in enhancing early and effective referral of presumptive TB patients.
Furthermore, the high number of deaths at home (43%) and the few numbers of deaths that were reported to have had death certificates issued (9.1%), illustrate the key role that verbal autopsies can play in establishing the cause of death statistics within the urban informal settlements. In Kenya, deaths at home or in the community do not ordinarily undergo medical certification. They are usually registered by the assigned local registration agent (usually the assistant chief) who is only required to identify the most probable cause of death from a list in the death registration form [29].
Health care-seeking patterns for TB symptoms in the two NUHDSS areas were from multiple sources with the majority having sort for care first in a private health facility. TB service diagnosis and treatment availability in most private health facilities are limited, leading to multiple hospital visits before a diagnosis is made [27]. According to the 2017 patient pathway analysis for differentiated service delivery of TB in Kenya, only 46% of people with possible TB who sought health care had access to diagnosis at initial care seeking [30]). Findings in our study, therefore, re-emphasize the need for strengthening of the capacity of the private health care providers in TB knowledge, diagnosis and treatment especially within the informal settlements as per the Kenya Public-Private Mix (PPM) 2017–2020 action plan and the Kenya national strategic plan for tuberculosis, leprosy and lung health 2019–2023. The multiple type of facilities where care was sought from before death, reinforces the role of the PPM approach in reaching out to a wider scope of facilities including pharmacies, small private clinics, nursing homes and stand-alone laboratories that have ordinarily been left out within the TB scope of service to detect TB cases and in making appropriate referrals for diagnosis and treatment [31]. This would lead to improved detection rates, treatment outcomes, enhance access to services and subsequently minimize late case detection and resultant TB mortalities [26].
Some limitations for this paper are that VA data analysis is usually dependent on the review of symptoms generally associated with various diseases or conditions and VA data may under/overestimate TB related deaths [30]. Other challenges are that mortality estimates obtained by VA are susceptible to bias due to misclassification [7]. Even with these limitations, the dataset used in the analysis covered a large population over a long period and expounds on the community patterns of TB mortality within the informal settlements of Nairobi.