Commercialization of obstetric and neonatal care in the Democratic Republic of the Congo: A study of the variability in user fees in Lubumbashi, 2014

Objective In the Democratic Republic of the Congo, insufficient state financing of the health system produced weak progress toward targets of Millennium Development Goals 4 and 5. In Lubumbashi, almost all women pay out-of-pocket for obstetric and neonatal care. As no standard pricing system has been implemented, there is great variation in payments related to childbirth between health facilities and even within the same facility. This work investigates the determinants of this variation. Methods We conducted a cross-sectional study including women from admission through discharge at 92 maternity wards in Lubumbashi in March 2014. The women’s payments were collected and validated by triangulating interviews of new mothers and nurses with document review. We studied payments related to delivery from the perspective of women delivering. The total was the sum of the payments linked to seeking and accessing care and transport of the woman and companion. The determinants were assessed by multilevel regression. Results Median payments for delivery varied by type: for an uncomplicated vaginal delivery, US$45 (range, US$17-260); for a complicated vaginal delivery US$60 (US$16-304); and for a Cesarean section, US$338 (US$163-782). Vaginal delivery was more expensive at health centers than in general referral hospitals or polyclinics. Cesarean sections done in corporate polyclinics and hospitals were more expensive than those done in the general referral hospitals. Referral of delivering women, use of more highly trained personnel, and a longer stay in the maternity unit contributed to higher expenses. A vaginal delivery in the private sector was more cost-effective than in the public sector. Conclusion To guarantee universal coverage of high-quality care, we suggest that the government and funders in DRC support health insurance and risk pool initiatives, and introduce and institutionalize free mother and infant care. The city of Lubumbashi has an area of 747 km2 and an estimated population, in 2014, of more than 2 million, for a density of 2,543 inhabitants per km2. Nearly 70% of its population lives on less than US$1 per day [38]. It is subdivided into eleven HZs, each of which has a general referral hospital (GRH), on average, 15 health centers (HC). One HZ, Kowe, does not have a GRH. In the two-level health care system in DRC, the health centers (HC) are the first level of contact with the population; the GRHs are the first level to which the HCs refer complicated cases. GRHs can in turn refer cases to the highest level, a provincial hospital like the Jason Sendwe Hospital or the University Clinics of Lubumbashi. In practice, this hierarchy is not always respected by users. There are more than 350 HCFs in Lubumbashi (hospitals, polyclinics, and HCs), and 70% of them are urban. The private sector represents more than 60% [41]. Nearly 180 of the HCF in Lubumbashi provide maternity services [49]. In 2012, the percentage of deliveries with skilled attendance at Lubumbashi was 94%, but the rate of Cesarean sections was low, at 4.5% [40]. Until 2005, the United Nations Population Fund (UNFPA) supported the maternity ward at the general provincial referral hospital, Sendwe, by supplying drugs, supplies, and equipment for emergency obstetric and neonatal care (EmONC), though they did not manage payments to personnel. Since 2005, no other organization has stepped in to subsidize maternity care. The price of obstetric care is covered by private or parastatal companies for women whose partners work for those companies, but these women account for no more than 1% of those who deliver in Lubumbashi. Occasionally, a charitable organization or politician may pay the fees for obstetric care for insolvent women. The provincial hospital, Sendwe, which formerly had close to 9,000 childbirths a year, now has only 1,600 (Fig 1; S1 Table; S1 Fig). The University Clinics of Lubumbashi (UCL) is a university hospital center for the training of doctors and nurses under the management of the University of Lubumbashi. It is a tertiary-level referral hospital. However, because of the precarity of its funding, it does not function to the full potential of its activities. Due to the lack of an appropriate public GRH in its HZ, it also serves as the GRH for the Lubumbashi HZ. Since 2010, the average annual number of deliveries there is 1200. The two parastatal hospitals, GCM-Sud and the one belonging to the Société Nationale des Chemins de Fer du Congo (SNCC), are also two large health care facilities in Lubumbashi. In addition to care provided to employees of these parastatal companies, they offer care to the general population. The number of deliveries in 2014 at these facilities was, respectively, 1800 and 900. GCM-Sud is the GRH for the Mumbunda HZ and SNCC is the GRH for the Tshamilemba HZ. For this study, we included all the HZ, and in each of them, all HCFs that performed at least 25 deliveries in the month before the survey (March 2014). These facilities were officially recognized as providing maternal health care according to the provincial division of health’s data. Facilities not included in this study were not yet officially recognized as providing maternity care, but had occasional deliveries without having set up the appropriate services ahead of time. In each of the 92 HCF that fit the inclusion criteria, we targeted all the women admitted to the maternity ward during the study period, but only those capable of communicating with the researchers (by speaking French or at least one of the local languages: Swahili, Tshiluba, Lingala or Kikongo) and who agreed to participate in the study were retained. In all, 1404 were admitted to maternity units during this period; all were eligible for our study. The women who read and understood French signed a written consent after having read it; for those who could not read French, the consent form was read and explained to them by the researcher in a local language, in the presence of a witness who was not a member of the study team. They then signed the consent for after confirming that they had satisfactorily understood its contents. Data were kept confidential. This study was approved by the medical ethics committee of the University of Lubumbashi (CEM-UNILU: UNILU/CEM/010/2011). We conducted a cross-sectional study, including women admitted to the selected HCF from March 1–8, 2014. We followed the women from admission to discharge. During this period, we tracked their payments [50]. The last women followed up were discharged from the study in May of the same year. Thirty trained researchers collected data via interviews and document review. The structured questionnaire (S1 Text) collected data on payments made and the reason for each payment, as well as sociodemographic and economic information not included in maternity records. We interviewed the head of the maternity unit to determine for each parturient which items were covered by the payments. We reviewed each woman’s medical record to collect data about her obstetric situation (reasons for admission, complications, type of delivery and care received), and reviewed invoices for fees related to delivery. Payments made were followed daily during the whole length of the stay, by tallying receipts held by the women or their relatives, or from the maternity records when the receipt was not available [50]. Payments were validated after triangulating information i) from the parturient or family, ii) in the maternity record, and iii) by confirmation of the payment by the head of the maternity unit. For medications purchased outside of the HCF, we tallied the purchase receipts; if the receipt was missing, we took the woman’s word, provided that the medicine had been identified or that the information had been validated by the birth attendants. This was also the case for laboratory examinations or blood bought at the blood bank. If the total of the payments declared by the woman was higher than that stated by the maternity unit team and proofs of payment were not available, the difference was included in the category “other payments”, which included any gratuity given to the health staff [50]. For drugs, we took into account those prescribed and purchased during the facility stay whether the woman was to continue taking them at home or not. For example, during data collection, we noted that women received prescriptions before discharge, and staff ensured that the women had obtained these drugs before discharge. We did not take into account any drugs bought outside the HCFs after the woman’s discharge. For the women whose care was subsidized by an organization, standard pricing was used to determine the price of care and the charges covered by the payments. Payments associated with round-trip transport, when utilized, were determined based on the woman’s or a family member’s statement. If these were paid by a third party, they were estimated by taking into account the going rate for the trip. For public transportation, this total was multiplied by the number of companions accompanying the woman upon her arrival at the maternity ward. For the trip home, since we were following the women, we asked them what means of transport they would take. For those who had a personal means of transport, we estimated the price based on private taxi fare to the woman’s home. For those who took public transport, we likewise estimated the price as a function of distance. We used the same estimate whether the women planned to take a shared taxi or a minibus. Data were double-entered in Excel. We studied payments for care from the perspective of women admitted to maternity units over the course of the study period. Household expenses include direct and indirect costs. In this study, direct costs comprise direct medical costs (obstetric and neonatal care: medicine, equipment, delivery, Cesarean section, episiotomy, bandages, laboratory tests, newborn care, stay, and medical record) and direct non-medical costs (transport of the woman). Indirect costs were additional payments by companions of the women for trips home to get food and clothing, and time lost by companion for supervision and stay). The total payments for delivery were the sum of payments related to: i) obstetric and neonatal treatments (direct medical costs) as well as transport of the woman (direct non-medical costs), and ii) transport of the companion (indirect costs) [50]. Time lost was estimated in terms of daily lost revenue, calculated according to the current occupation of the woman and companion as declared the day of the survey. Given the difficulty of obtaining data about household income, we utilized data from household surveys in Lubumbashi. Daily income was the quotient of the monthly income divided by 28 days. The lost income was the product of the daily income and the number of days in the maternity ward. For the women for whom housework was the only occupation, the average monthly income used to calculate the time lost was US$45 [50]. We did not include in the calculation the expenses related to the purchase of the layette (clothing for the mother and newborn). Likewise, intangible effects such as stress and emotions related to delivery were not included in the calculation of expenses. Also, given that the HCFs did not provide food for the women during their stay and the women ate their usual household food, we did not include payments for this food in our calculations. On the other hand, payments for the companion’s trip to the get the food were included. All of the expenses were expressed in US dollars (920 Congolese francs = US$1). We used the usual descriptive statistics (percentage, mean and standard deviation, median, minimum and maximum) to describe the profile of the facilities and the women included in the study. The Mann-Whitney and Kruskal-Wallis tests were used to compare median expenses according to the characteristics of the women. The Bonferroni correction was used for pairwise comparisons of expenses by type of delivery: uncomplicated vaginal delivery (UnVD) versus complicated vaginal delivery (CVD); UnVD versus Cesarean; CVD versus Cesarean) [51]. Complicated delivery was defined according to WHO criteria as “medical problems associated with obstetric labor, such as hemorrhage (antepartum and postpartum), obstructed labor, postpartum sepsis, placenta previa, placental abruption and premature rupture of membranes, complications of abortion, severe pre-eclampsia and eclampsia, ectopic pregnancy and ruptured uterus or others. These complications can affect the well-being of the mother, the fetus, or both [52]. The determinants of the variability of payments related to childbirth were investigated by forward stepwise multilevel nonlinear modeling [53] at a significance threshold of 5%. The four levels considered were: 1) the women delivering; 2) the staff managing the delivery; 3) the types of HCFs; and 4) the sector to which the HCF belonged. In this study, we considered any HCF created as a HC as a HC, regardless of the evolution of its technical platform. HCs in which, for example, cesarean sections were carried out, were included in the HC category. All sociodemographic variables, the type of delivery, the length of stay, whether a patient was referred, the person paying for care, were variables linked to the women delivering. For each type of delivery, we presented four models—empty, with the variables linked to the women delivering, with the systemic variables, and with all variables—to study the change in measures of variation of systemic variables with regard to individual variables. All analyses were performed in Stata v13.1. After identifying the determinants of the variability in payments, we explored the most cost-effective alternative in terms of access to EmONC for vaginal deliveries [54]. We defined EmONC as “services for the treatment of direct obstetric complications that arise during pregnancy and childbirth. They include nine signal functions: i) administer parenteral antibiotics; ii) administer uterotonic drugs; iii) administer parenteral anticonvulsants for preeclampsia and eclampsia; iv) manually remove the placenta; v) remove retained products; vi) perform assisted vaginal delivery; vii) perform basic neonatal resuscitation; viii) perform surgery and ix) perform blood transfusion” [55]. In this study, we evaluated the efficacy of each option in terms of the woman’s access to signal functions i-vi of EmONC. We considered any woman who had an obstetric need and who received an appropriate intervention for that need according to the indications for each EmONC signal function as having received EmONC. In the case of VD, every woman should receive at least one EmONC signal function (injection of an uterotonic to prevent post-partum hemorrhage). We did not conduct this cost-effectiveness analysis for cesarean deliveries, since they did not respond to the same efficacy criteria as vaginal deliveries [52]. We carried out the choice of alternatives based on one factor of entry into the health care system: the ownership sector. We asked the question, for women, which of the choices, between delivering vaginally in public or parastatal HCFs (option_1) or in private or religious HCFs (option_2) was more cost-effective. In option_1, women had the choice to deliver in a public or parastatal HCF, which was a HC, GRH, Sendwe, UCL, or a parastatal facility (GCM-Sud, SNCC). In this facility, was the delivery managed by a midwife or by a physician, a generalist or a specialist, and did she receive EmONC or not? In option_2, women were exclusively in a private or religious HCF, where the options for their management were the same as those in option_1. To answer this question, we calculated the incremental cost-effectiveness ratio (ICER) as the relationship of the difference in payments linked to vaginal birth in option_1 and option_2 to the difference in access to EmONC between the two options. To measure the uncertainty in our cost-effectiveness model, we considered a threshold of acceptable payments—willingness to pay—of US$460 (2015 per capita gross domestic product of the DRC) [56]. We categorized the options studied according to whether they were very cost-effective (ICER < 1 x per capita GDP); cost-effective (ICER ≥ 1 x and ≤ 3 x per capita GDP); and not cost-effective (ICER > 3 x per capita GDP [57]. To test the robustness of our conclusions, we carried out a probabilistic sensitivity analysis. We created probability distributions for all the parameters of the model. For the costs, we used the payment data observed in this study. We created 100 samples using Monte Carlo simulations and calculated their expected values. We then calculated the proportion of samples that had a good cost-effectiveness for each alternative. These analyses were performed in TreeAge Pro 2018 R1.1 (Watertown, MA, USA).