Discussion
The COVID-19 pandemic affected healthcare access and utilisation, especially among vulnerable populations, including people with chronic diseases like diabetes. Currently, there is little evidence of the cost burden imposed on patients with T2D by the COVID-19 pandemic in terms of healthcare and broader societal costs in LMICs. This study provides evidence on the cost of managing T2D before and during COVID-19 in Kenya and Tanzania, answering whether COVID-19 had a significant impact on healthcare and broader societal costs. The COVID-19 pandemic had an adverse impact on the socioeconomic situation of patients in the two countries, despite the total direct costs not being statistically different during the pandemic.
The overall costs of managing T2D in most categories increased in both countries during COVID-19, but some of the increase was not significant. Transport, testing and outpatient costs increased significantly in Tanzania but not in Kenya. T2D medication costs did not change in Tanzania but decreased significantly in Kenya. However, productivity losses increased significantly in both countries. We acknowledge the contextual differences between the two countries. We therefore attempted to compare the effects on costs but with close consideration of contextual differences. The differential effects on costs can be explained by three contextual reasons. First, the presence of lockdown measures in Kenya affected access to services, hence limited increases in costs, which is different from Tanzania.33–35 Second, insurance coverage was relatively higher in Kenya than Tanzania, as observed in this study, possibly explaining the limited cost escalation in Kenya compared to Tanzania. Third, the increase in testing costs in Tanzania is likely due to increased testing in private pharmacies and laboratories following a reduced utilisation of services and severe disruption of care in hospitals, which is evident in our data.36
However, when considering wider societal costs, productivity loss was significantly impacted during COVID-19 in both countries. This is possibly due to the fact that most patients reduced their mobility, avoiding crowded spaces, including workplaces, and allocated more time to improve and sustain their health. Because of the very low number of responses, it was not possible to explore whether COVID-19 was associated with any change in formal and informal care. The divergence of the findings in the two countries after adjusting for the clinical and demographic characteristics of patients with T2D in the two countries indicates the heterogeneity of the costs incurred in different settings of SSA.
Our findings are similar to those of other studies conducted among patients with T2D before and during COVID-19. A study conducted in Bangladesh showed increased glucose levels, complications and costs of care and disruption of access to care among patients with T2D during the COVID-19 pandemic.37 The decreased costs of diabetic medication during the COVID-19 period in Kenya are in contrast with another Kenyan study showing increased costs of insulin medication and reduced healthcare utilisation.38 Healthcare services and access were disrupted in Kenya during the pandemic, specifically the significant decline in the prescription of oral hypoglycaemics in Kenya, hence the overfall reduced costs.36 The reduction in utilisation and availability of T2D medication makes a strong case in our study, where the medication costs incurred were reduced significantly in Kenya and non-significantly in Tanzania during the COVID-19 pandemic. A study conducted in India on the implication of COVID-19 on the management of chronic conditions explained the loss of income as a driver for the inaccessibility of essential medication and job loss as significant predictors of depreciated diabetes symptoms.39 In other LMICs, the healthcare systems experienced severe disruptions in the provision of essential medical infrastructure that affected the service delivery and outcomes for patients with NCDs and other comorbidities.40 Similarly, in high-income countries, the burden of T2D during COVID was also experienced and there was great concern about the increased risks of mortality, morbidity and decreased economic productivity.41
The healthcare system was severely impacted during the COVID-19 pandemic, hence affecting service delivery to patients with T2D.18 This had a significant impact on costs based on the availability of healthcare services to patients in the countries. It was reported in other countries in Africa that lockdowns had an impact on glucose control, where 90.5% of patients had uncontrolled blood sugar during the lockdown, and this was higher than before the lockdown periods (82.9%).42 There was also reduced access to healthcare facilities and clinics, with reduced use frequency during lockdowns.42 In Tanzania, there were no lockdowns, but there was an increased surge of patients with respiratory conditions in the facilities.23 35 This evidence of the slight increases in costs in Kenya and Tanzania may be due to the inaccessibility of services and the lack of essential medication due to lockdowns, among other healthcare system constraints associated with the COVID-19 pandemic. In Kenya, like other LMICs, there was also a confirmed inability to access healthcare during lockdowns, hence reduced healthcare-seeking practices in urban slum areas.43 The differences in the costs can also be attributable to the different insurance mechanisms in Kenya and Tanzania, where the insured patients in Kenya experienced severe disruption, a reverse of what was observed in Tanzania.36 There was a significant decrease in overall outpatient visits in Kenya for patients with diabetes, showing a drop in healthcare system utilisation, which was similar to observations noted in other SSA countries.44 45 Healthcare system utilisation has an impact on the costs incurred by the patients in managing chronic conditions, implying that this might have economically affected the spending on the patients and the government. The overall disruption of the healthcare system in SSA, specifically in Kenya and Tanzania, indicates the implications and dynamics of a constrained environment for management and the economic utilisation and stability of patients with T2D.
Strength and limitations
This study captured the different dynamics of costing in Kenya and Tanzania, which are different. This warranted the use of different approaches: a two-part model in Tanzania to account for the zero costs and the use of the generalised gamma regression model for Kenya. Furthermore, this is a unique study, as no patient-perspective economic analysis has been conducted in these countries. The results of this research are likely to form a blueprint for the formulation of country-specific recommendations that will contribute to the formulation of policies for diabetes management during pandemics.
However, this study has several limitations. The data were collected using a cross-sectional study design while recalling the experience before and during the COVID-19 period. This could have caused a potential recall bias because the costs were incurred at different periods for each person, which means the timescale of costs is less meaningful. However, this does not significantly affect the individual cost breakdowns, but the magnitude of predicted costs. In addition, we did not include whether someone had COVID-19 infection, which may have had a substantial impact on healthcare costs for the patient because less than 10% of the patients reported having tested positive for COVID-19 infection in both countries. The questionnaire did not include questions on food costs and accommodation costs; this could have been important considering that the diet for a patient with diabetes would cost more, and accommodation for those who travelled for specialised diabetes care had cost implications. We also did not capture T2D complications; instead, we used comorbidities. Furthermore, we expected an overestimation of productivity loss costs because patients with chronic illnesses and their caregivers were highly affected by the disruptions during the COVID-19 pandemic.
Furthermore, the sample was not drawn at random. Therefore, it may not be representative of the larger population, and the results may not apply to other groups or populations. Because of this, caution should be exercised when generalising the findings of this study to other settings. Finally, some assumptions on unit costs had to be made, as NHIF costs did not come at the same level of disaggregation of the resource use we collected. Although the sample was not randomly selected, the distribution of key variables like diversified geographical locations and gender in our sample resembles that of the general population. The costs shown in this study are an underestimation of the actual costs incurred by the patients and the public healthcare system because what is represented is an average of what is incurred in the three most disrupted months in 2021. The estimation of these costs per patient for the whole year, assuming that the patients incurred some costs in the other quarters of the year, would make the costs substantially higher. Therefore, despite the slight increase in economic costs of managing T2D in Kenya and Tanzania, there was a larger estimated cost.
