The True Costs of Cesarean Sections for Patients in Rural Rwanda: Accounting for Post-Discharge Expenses in Estimated Health Expenditures

Anne Niyigena (  niyianne@gmail.com ) Partners In Health https://orcid.org/0000-0001-6140-8083 Barnabas Alayande Harvard Medical School Laban Bikorimana Partners In Health Elizabeth Miranda Harvard Medical School Niclas Rudolfson Lund University: Lunds Universitet Deogratias Ndagijimana Partners In Health Fredrick Kateera Partners In Health Robert Riviello Brigham and Women's Hospital Bethany Hedt-Gauthier Harvard Medical School Department of Global Health and Social Medicine


Introduction
Access to emergency obstetric care, including cesarean sections (c-sections), is an essential part of a functional health system [1]. However, inequalities in c-section access have been reported across and within countries [2,3]. Financial barriers are the most cited driver of c-section inaccessibility [4], with poor access most affecting the economically deprived [5][6][7]. Limited access to c-sections is associated with higher risks of poor outcomes for mothers and their babies [4,8].
C-sections are considered cost-effective interventions, costing US$251 to US$3,462 per disability adjusted life year saved [9]. However, women who deliver via c-section are at risk of nancial hardship. Studies in sub-Saharan Africa have estimated the direct costs of c-section to be $144-$426 [10][11][12], a considerable amount compared to the average regional gross domestic product (GDP) per capita of $4,195 in 2019 [13] .Even when direct costs are heavily subsidized, indirect costs put a woman and her family in nancial risks [14]. Studies have shown that surgery in general [15,16], and c-sections speci cally [17,18], can be nancially catastrophic for a patient's family. However, these studies fail to include the extended costs for surgery, and potentially underestimating the true risk of catastrophic health expenditures (CHEs) due to surgery.
In Rwanda, the location of this study, 28% of patients undergoing peritonitis surgery suffered CHEs as a result of the surgery [19]. While these studies include direct and indirect medical costs up to the time of discharge, we are unaware of any study of surgery in Africa that considers costs associated with postoperative follow-up after discharge. In this paper, we describe the nancial costs of c-section care for Rwandan women delivering via c-section at a rural district hospital, including direct and indirect costs of all care received up to postoperative day (POD 30), and estimate the full risk of CHE for these women.

Study setting
This study was nested in a prospective cohort study conducted at Kirehe District Hospital, which aimed to evaluate the feasibility and acceptability of a telemedicine intervention for the diagnosis of post-hospital discharge surgical site infections by community health workers. Kirehe District Hospital is located in the Eastern Province of Rwanda and is managed by Rwanda's Ministry of Health with technical support from Partners In Health/Inshuti Mu Buzima (PIH/IMB), a Boston-based non-governmental organization that provides technical support to the Ministry of Health.
In Rwanda, c-sections are typically performed at district hospitals by general practitioners (GPs) [20], and at Kirehe District Hospital, c-section is the most commonly performed surgery. After delivery, women are monitored in a post-c-section ward and usually discharged on POD 3. In Rwanda, there are no standardized guidelines for c-section follow-up; however, at Kirehe District Hospital, c-section patients are asked to visit the local peripheral health centers three days post-discharge for wound inspection and dressing change and to continue follow-up until deemed unnecessary by the health center nurse.
Approximately 83% of the Rwandan population has health insurance and 96.1% of insured rural residents are enrolled in the community-based health insurance (CBHI) program [21]. Rwanda's CBHI is based on a 4-tier wealth system called Ubudehe, with the bottom tier including the poorest and the upper tier including the wealthiest Rwandans. For those in Ubudehe 1, CBHI premium is fully subsidized by the government; individuals in Ubudehe 2 and 3 pay CHBI premiums of approximately US$3 per person each year and those in Ubudehe 4 pay a premium of US$7 per person per year [22]. Individuals in Ubudehe 1 pay no copayment at point of care while those in Ubudehe 2-4 incur a 10% copayment for direct medical services.

Data collection
Enrollment data collection: Women who delivered via c-section at Kirehe District Hospital between September 23rd, 2019 and February 22nd, 2020 were enrolled after c-section delivery and prior to discharge. All participants provided informed consent prior to data collection. Data collectors administered sociodemographic and clinical characteristics questionnaires before patients were discharged from the hospital; data were directly entered into REDCap data management software [23].
Patients also responded to a nancial survey, described below. Data on healthcare expenditures was extracted from OpenMRS, an online database tracking details on patients' medical care and expenses.
Follow-up data collection: At enrollment, respondents provided cell phone numbers (their own, a relative's or a neighbor's) on which they could be contacted. On POD 30 (± 1 day), data collectors administered a phone-based follow-up interview to assess post-discharge follow-up activities. The costs of postdischarge c-section follow-up were assessed in terms of expenses for medical care, expenses for transport, and lost wages due to seeking follow up care at the health centers. Study participants that could not be reached by the phone number they provided at discharge were contacted in person by a local community health worker and a telephone survey was administered on the community health worker's telephone. Three attempts on three different days were made in an effort to maximize the response rate; individuals not contacted after three attempts were considered lost-to-follow-up. The POD30 response rate was 84%.
Financial Survey: The nancial survey included questions adapted from the Program in Global Surgery and Social Change National Surgical Obstetric Anesthesia Plan surgical indicator questionnaire [24].We added the following variables to the core questionnaire: estimates of monthly household income, selfreported routine monthly household expenditure, whether the patient had to borrow money or sell possessions to pay for the current hospitalization, and household monthly consumption as a sum of expenditures for food and drink, transportation, livestock, housing, transportation fees, school fees, and healthcare in the past months. Non-monetary income such as agricultural harvest was converted into Rwandan Franc (RWF) using the price of local goods at the time of data collection. Lost wages were estimated using daily wages for the occupation of the patients and caregiver at the time, and reported in RWF.
De nition of key terms: We strati ed expenses into two main categories: in-hospital costs and postdischarge follow-up costs. The in-hospital costs include expenditures from when a woman left her home to seek care for delivery until the time of discharge. In-hospital costs were further grouped into direct medical, direct non-medical, and indirect costs. Direct medical costs included payments for medical supplies, medications, laboratory exams, surgical procedure, imaging, consultation, and hospital bed. Direct non-medical costs included expenses of a caregiver during hospitalization, food and transport from home to hospital. Indirect costs included household lost wages due to hospitalization. Postdischarge follow-up costs included direct medical cost paid at the health center in addition to indirect follow-up costs. The direct non-medical follow-up costs included transport from hospital to home for the patient and caregiver, transportation to the health center for patient and caregiver; while indirect follow up costs included lost wages due to delayed return to work and lost wages of both the patient seeking follow-up care and that of the accompanying caregiver. We chose to include the cost of transport from the hospital to home after the c-section in the follow-up care costs as these expenses are generally not factored in CHE studies and allows for direct comparability of our results.
Poverty was de ned using the World Bank de nitions, de ned as a daily expenditure below $1.90 per person per day [25]. Catastrophic health expenditure (CHE) has been variously de ned as out-of-pocket healthcare expenses that exceeds 10% of total annual household expenditure or income [26,27], or as spending greater than 40% of the annual household income, excluding subsistence needs, on health care [28]. For this study, we de ned CHE as healthcare spending of greater than 10% of annual household consumption to align with the de nition of the United Nations' Sustainable Development Goals 3.8.2 [29].The annual household consumption was de ned as a sum of annual expenditures on food and drink, transportation, livestock, housing, transportation fees, school fees, healthcare and other expenses.

Statistical analysis
We restricted our analysis to patients who responded to the nancial questionnaires at both time points. We also restricted our analyses to individuals who sought follow-up care at the health center at least once during the rst 30 postoperative days so we could estimate the costs associated to follow-up care. All tradeable nancial expenses, such as in-hospital expenses and transport fees, were converted into US dollars (US$) using the nominal exchange rate at study start date (October, 2019), and US$1 equated RWF916.17 [30] .All non-tradeable expenses, including salaries and lost wages were converted to US$ using the 2019 Rwanda purchasing power parity (PPP) conversion factor for personal consumption of 317.18 [31] We describe our sample using frequency and percentages for categorical variables, mean and standard deviation (SD) for normally distributed continuous variables, and median and interquartile range (IQR) for continuous variables with non-normal distributions. We summarize the nancial cost of c-section care strati ed by Rwanda's four-tier wealth classi cation by in-hospital and follow-up care components, using median and interquartile range. We summarize each of the main cost contributors as a percentage of the overall costs. We also calculated incidence of CHE for all expense categories and reported the frequencies and percents.
We determined the daily expenditure per person in the household as a sum of individual expenses of the household divided by the household size, and report the proportion of participants who lived below the poverty line, before c-section delivery. We then calculated the total expenditure remaining after paying csection cost, and estimated the proportion of women whose spending is below poverty line. The percentage of people who were pushed into extreme poverty by c-section delivery re ects people whose annual total expenditure were above poverty line at baseline, but who fell below poverty after paying for the costs of c-section care up through POD30.

Results
In total, 479 patients were included in this study, of whom 68.7% were aged less than 30 years, and the majority (94.8%) were insured by CBHI (Table 1). Approximately 10% of patients belonged to the lowest Ubudehe category, 84.7% were farmers and the median annual household income was US$532.8 (IQR: $232.8, $859.1). The median travel time from home to health center was 30 minutes (IQR: 15, 60 minutes) and from health center to hospital was 40 minutes (IQR: 5, 60 minutes). For the 433 (81%) patients who rst sought care at the health center prior to going to Kirehe District Hospital, 61.6% were transported to the hospital in ambulance and 26.9% walked to the hospital.
The median household size was 4 people (IQR: 3, 6) ( Table 1)   substantially lower out-of-pocket payments for c-sections in Rwanda most likely re ects the cost cushion provided by Rwanda's robust health insurance system. In our study, nearly all c-section patients were insured through CBHI. Although CBHI may offset the direct medial costs of c-sections, previously estimated in Rwanda as US$339.0 from the health facility perspective [11] c-section patients paid an additional US$31.4 in non-medical costs while still in the hospital. Moreover, the direct and indirect cost of c-section up to discharge was catastrophic for 27.0% of women, which is comparable to previous report of CHE among peritonitis patients in Rwanda [19]; but lower compared to the 60.0% incidence of CHE from c-section reported in India [33].
Surprisingly, the full cost of cesarean section rose to US$122.2 and was catastrophic for 77% of women when follow-up costs were considered. The tripling of out of pocket expenditures by follow-up costs implies that the full nancial picture of c-section care can only be truly understood when post-discharge costs are examined. Most costs covered by health insurance globally are expenditures linked to direct medical services [34][35][36], but this misses the substantial follow-up costs. For example, despite full coverage of direct medical cost for people in Ubudehe 1, the incidence of CHE in this group within 30 days post-cesarean was 74.0%. Interestingly, women in Ubudehe 2 had the highest risk of CHE during their hospital stay, re ecting their increased vulnerability of their low incomes combined with lower coverage of expenses by CBHI. This corroborates the argument that health insurance that covers direct medical costs, though essential, is not su cient to nancially protect poor patients [35,36].
While the majority of women in this study were already poor, c-section delivery exacerbated nancial hardship of poor women and threatens their living standards, as re ected by the fact that more than twothirds of women sold assets or borrowed money to afford c-section surgery and hospitalization. The sales of property in order to afford obstetric surgery care in rural Rwanda was found to be higher than in Ethiopia (4.4%) and most other LMICs [37].
Major contributors to overall c-section costs included post discharge lost wages (55.3%), costs of post discharge follow up at health center (15.8%), costs of caregiving (13.3%), and transportation costs (11%). Similarly, major cost drivers of post-discharge expenditure included lost wages (84.7%) and transportation (11.5%). Follow up interventions and models that reduce lost wages and eliminate transportation costs may contribute to a reduction in CHE [16,38,39]. Examples of innovative holistic interventions including transportation interventions, like the Uganda Reproductive Health Voucher Project, can be adapted to the local context [40,41]. We are also exploring innovative mHealth strategies and contextualized community-based follow-up strategies that allow for home-based care to reduce the cost of follow-up as well as the physical burden of traveling [42].
Addressing lost wages will be challenging. While paid maternity leave for women employed in formal work sectors is a national policy in Rwanda [43], the majority of patients in rural Rwanda are farmers and do not have access to these job-protected maternity leave packages. C-sections have been found to impose further health costs if mothers return to work prior to recovery [44], and further studies are needed to explore the appropriate time to resume work after c-cesarean section, with consideration to mitigating lost wages in both the formal work sectors and for farmers.
Our ndings should be interpreted in light of some limitations. Firstly, the calculations of household expenditure depended on patient memory and based on predetermined expense categories. Patients may have failed to accurately report expenditures or missed expenses that did not align with a category. Secondly, estimates of the cost of post-discharge follow up were based on reports from patients who decided to seek care on their own, because currently, there is no protocol for post-c-section follow-up.
Thus, our ndings do not fully re ect the true cost of c-section follow, if such follow-up protocols existed.

Conclusion
When full costs are considered, c-section care confers signi cant risk of nancial catastrophe on already impoverished households in rural Rwanda, despite the presence of a robust and widespread CBHI policy.
Indirect non-medical cost and the holistic cost of follow up for c-section from a patient perspective exceeds that of receiving initial medical care, and must be considered in development of policy and relevant intervention. Modelling of the nancial implication of various follow up strategies should be encouraged to determine the most e cient, safe, and nancially protective models of cost subsidy.

Ethics statement
This study had ethical approval from the Rwandan National Ethics Committee (Kigali, Rwanda, No.326/RNEC/2019) and Harvard Medical School (IRB18-1033). Adult patients were read information about the study, and voluntarily signed the consent prior to enrollment. We obtained voluntary assent from individuals less than 18 years, with signed consent from their parents or guardians.

Declarations Ethics statement
This study had ethical approval from the Rwandan National Ethics Committee (Kigali, Rwanda, No.326/RNEC/2019) and Harvard Medical School (IRB18-1033). Adult patients were read information about the study, and voluntarily signed the consent prior to enrollment. We obtained voluntary assent from individuals less than 18 years, with signed consent from their parents or guardians.

Consent for publication
We do not disclose any personal information of our participants. Thus, consent for publication is not applicable.
Data availability.
Data is available upon a reasonable request, by emailing niyianne@gmail.com