Can Sierra Leone maintain the equitable delivery of their Free Health Care Initiative? The case for more contextualised interventions: results of a cross-sectional survey

Sampling

World Vision’s PPA/AIM-Health programme services two locations and one municipality in the riverine (Sittia, Dema, and Bonthe Municipality) and four chiefdoms on the mainland (Imperi, Jong, Sogbeni, and Kpanda Kemoh). A two-stage probability sampling method was applied to obtain a sample of the population across in the riverine and on the mainland. In the first stage of sampling, a list of all the 199 villages from selected chiefdoms of the mainland and 121 villages of the selected riverine chiefdoms was compiled. The probability of a village being selected was therefore set as proportional to the number of households within that village. In this first stage, all 12,037 households on the mainland and 4712 households in the riverine therefore had an equal chance of being selected regardless of whether they contained the target population or not. The total number of households to include was then calculated assuming a confidence level of 95 % (α = 0.05), with an additional 5 % added to account for non-responses. This resulted in a minimum target of 373 households in the riverine, and 391 households on the mainland.

In the second stage of sampling, village leaders led field teams to the village centre where a pen was spun to determine the field team’s walking direction. A random number generation table was subsequently used to decide which household was to be visited first. A household was defined in terms of persons who were co-resident and shared common cooking arrangements, and were able to recognise one person as the head of household [18]. In the event that residents were absent from their home or that the target group was not present, field teams were instructed to proceed to the “next” household, which was defined as the one whose front door is closest to the one just visited. Enumerators proceeded to the next household, until the total number of households to be sampled from that village was completed.

Survey tool

The survey tool [see Additional file 1] was developed in consultation with the Bonthe DHMT and with assistance from maternal and child health experts within the World Vision Partnership. A total of 30 community health workers (CHWs) were selected as enumerators to participate in the household survey training, hosted by staff from neighbouring health centres and the DHMT. As part of the survey training enumerators were taught how to administer the questionnaire, record responses from participants, verify patient health cards, interpret the mid-upper arm circumference (MUAC) tapes, and weigh and measure children.

The survey included questions about household demographics; food intake; child health, including symptoms of acute respiratory infection (ARI), diarrhoea, and fevers, as well as treatment at a health centre within 24 h of the aforementioned symptoms’ onset; child vaccination (calculated for appropriate ages including measles at 9 months, OPV at birth, and 9-months for full immunisation, etc.); maternal care services, including Intermittent Preventive Treatment (IPT), use of insecticide treated nets (ITNs), access to antenatal (ANC), postnatal (PNC) and HIV services; delivery in health centres and in the presence of a skilled birth attendant (SBA); use of any family planning method (calculated by household use rather than individual); hand washing and latrine use. To minimise the risk of response bias, answers were verified through child and maternal health cards (i.e., child vaccinations, ANC visits), where available. The presence of a mosquito net and latrine was verified by the enumerator, as was the cleanliness of the latrine. Piloting of the questionnaire took place in villages not included as part of the final sampling frame. Where appropriate, questions were phrased in binomial (i.e., yes or no) format to facilitate the collection of data. Though the questionnaire was printed in English, training was conducted in a mixture of Krio and Mende and CHWs were instructed to conduct the interview in whichever language they felt best suited the household.

Exclusion and inclusion criteria

To be considered for secondary analysis a household had to contain at least one pregnant woman and/or a child under the age of 5. Interviews were conducted with the child’s primary caregiver, defined as the person who was, “primarily responsible for the health, safety and comfort of that child”. A sample of 393 households across all 121 villages in the riverine, and 397 households across 130 of the 199 villages on the mainland were ultimately represented in the study sample, for a total of 790 households.

Data analysis

Quantitative analysis was conducted using SPSS Statistics 17 & 22 (Release Versions 17.0.0 and 22.0.0). Using point prevalence data for various binomial variables (and variables which could be transformed into binomial variables), prevalence data were compared across households in the riverine and the mainland using Pearson Chi-Squared test with Yates’ Continuity Correction. To maximise all data available, missing data was handled using pairwise deletion. Rank variables were compared across the two areas using the Mann-Whitney test. Maternal health, child health and sanitation indicators in Bonthe District were compared across two geographically different areas; the riverine and the mainland. All tests were conducted for 95 % confidence with α = 0.05.

Response rates

According to the enumerators’ logs, 504 households were visited on the mainland, with 397 surveys collected, indicating a 78.8 % response rate. In the riverine, 491 households were visited and 393 surveys were collected, amounting to a response rate of 80.0 %. Non-responders include those that did not meet the criteria for inclusion as well as those that refused to participate.

Characteristics of the sample

Maternal health

Of the pregnant women interviewed, 85.3 % of riverine respondents and 84.1 % of mainland respondents had attended at least one ANC visit during their current pregnancy, with no significant differences between the two settings (p = .824). Almost half (44.8 %) of women on the mainland reported using a method of family planning compared to less than one-third, (28.4 %) of women in the riverine, with a statistically significant difference between the two locations ([9.8, 23.2] p < 0.05).

Children under-5

Sanitation

Availability of transport and access to appropriate health services

Among other environmental factors, accessibility and availability of transport are commonly noted throughout the literature to impact on health service [34]. In this study, the difficult terrain characteristic of the riverine may partly explain why women in the riverine are more likely to give birth at home rather than in a health centre. The topography of the riverine requires that one navigate a complex system of rivers, lagoons and estuaries both on foot and by boat. With previous studies linking long travel times to increases in child mortality [35–37], it is important that pregnant women access a health centre at the time of delivery. As a recommendation, the DHMT might consider the use of maternity waiting homes, or a facility within reach of a health centre that provides emergency obstetric care [38]. Maternal waiting homes have been employed in other low-income contexts, including neighbouring Liberia, where they were found to be an effective strategy for increasing the use of skilled birth attendants and improving maternal and neonatal health [39]. As, if not more important, than ensuring a safe and clean environment for delivery however, is the availability of a skilled health worker.

In addition to restricting access to care, difficult geographical conditions can also result in poor living and working conditions for health staff. Studies have highlighted the lowered performance of health workers on remote islands such as those represented in the riverine [40] and other remote peripheral health units in Sierra Leone have also shown significant inefficiencies [41]. Poor working conditions in turn, are associated with poor performance, increases in attrition rates, and difficulties recruiting health workers, as individuals prefer not to be posted to remote and difficult locations with poor infrastructure and far away from their family [42–44]. The terrain and remoteness of Bonthe District make it an unpopular location for health workers, with the riverine being particularly understaffed. The DHMT may want to consider additional incentives as a means of enticing health workers to the riverine. Salary top-ups and other non-financial incentives such as providing free housing and further education opportunities were found to impact on the willingness of nurses to work in remote areas of Tanzania [45]. In addition, the DHMT may want to select health workers who are more intrinsically motivated [46] and who originate from very remote areas, as they have been found to express a greater willingness to take up a remote position [45]. To maintain the equitable delivery of MCH services, it is imperative that less resourced areas have in place strategies to maintain a satisfied health workforce.

Widely recognised as effective preventative practices, child immunisation and appropriate water and sanitation are important determinants of maternal and child health [47, 48]. In contexts with scare human resources for health, enlisting the help of alternative cadres of health worker, such as the community health workers (CHWs), is recommended for the effective delivery of preventative maternal and child health care interventions [49]. CHWs are considerably less expensive to train, are selected by the very communities they serve, and with adequate training, can help address many of the cultural barriers often associated with delayed or inconsistent access to health care services [50, 51]. The use of CHWs may be particularly appropriate for the riverine, given its remoteness and difficult topography.

Affordability is also an important factor when considering the equitable access to transport and health services. Though the FHCI theoretically removed user fees, studies report that women have continued to pay for services since the initiation of the scheme [52]. Unfortunately, these findings are not uncommon in free health care initiatives [53, 54], with patients continuing to pay out-of-pocket for transport, medical supplies, and for informal payments requested by health centre staff. These expenses are likely higher in areas such as the riverine, where there are longer distances to travel to the health facility, resources are scarcer or more difficult to obtain, and where the working conditions for health staff are considerably worse. A more recent study conducted in Sierra Leone on the impact of removal of user fees showed that more educated women were more likely to benefit from their removal, with no impact on service utilisation across wealth quartiles [55]. As McKinnon et al.’s [55] review notes, the removal of user fees alone is not enough to reduce inequalities in the delivery of health care. Barriers that impact on socioeconomic inequalities, and access to quality service need to be addressed and included within national plans if Sierra Leone is to implement their FHCI equitably and efficiently amongst its population.

Perceived quality of care

It is worth noting that along with the introduction of the FHCI, Sierra Leone also made illegal the practice of deliveries being assisted by a traditional birth attendant (TBA) [56]. The decision to give birth in the presence of a TBA is largely influenced by cultural and societal norms, with individuals reporting a more personal relationship, better quality care, and greater trust in TBAs. Similarly, a lack of available or inaccessible health facilities and/or a poor opinion of hospital services and staff also contribute to a preferred use of TBA [31, 57]. TBAs are also considered to be more affordable, as they are amenable to being paid in-kind or by instalments [58]. These factors may all contribute to our observation of greater TBA use and a greater prevalence of women giving birth at home in the riverine.

While the Bonthe DHMT is well aware of the legal standing of TBAs, they must acknowledge the value that some women place on TBAs. In the absence of a greater number of trained health providers posted to the riverine, TBAs remain an important resource for the transmission of health messaging and to increase service utilisation. For example, TBAs are an important resource to encourage pregnant women to come to the health centre to attend antenatal services and when engaged appropriately, and like community health workers [49], can act as an important bridge between communities and more formal health systems [39]. Consideration should be given for how to best engage TBAs in Bonthe District, as an important strategy to achieve a more equitable maternal health care service.

Equity, not equality

While the Government of Sierra Leone (GoSL) currently provides health resources to the Bonthe DHMT through both the Bonthe Municipal Council (located in the riverine), and Bonthe District Council (located on the mainland), resources are allocated on the basis of population with little consideration for typography, difficulty of access, and the demographic profiles of potential health care users. The results of this study indicate that the riverine may be disproportionately disadvantages by a lack of resources. Contingencies for areas of greater need should be developed and implemented, which includes increased availability of skilled health workers, more focused immunisation campaigns, and greater resources to address the lack of access to water and sanitation. In other words, initiating a free MCH programme alone is not enough to reduce disparities within populations and resources in Bonthe District should be allocated to specific health centres on the basis of need, rather than on the basis population. In addition, routine monitoring of national and district data should be conducted to capture “health systems input, quantities and prices and health services outputs to facilitate regular efficiency analyses” [41]. Early involvement of stakeholders is particularly important in the early stages of MCH programmes as they can provide important suggestions for programme re-alignment, and how to meet end-user expectations [59]. Helping programmes better plan for service delivery is important to ensure access to care for more vulnerable populations and to deliver equitable health services. Additional research into the reasons for the observed discrepancies will be important to further inform the implementation of health activities and the distribution of resources going forward.

Sierra Leone post-Ebola

Data from this study was collected prior to the Ebola outbreak that afflicted Sierra Leone and its neighbouring countries of Guinea and Liberia. The recent outbreak resulted in over 3950 deaths in Sierra Leone alone [60] and has left an already under-resourced and overstretched health system in a severe crisis. With 6.85 % of Sierra Leone’s health workforce being lost to the Ebola virus, some predict an increase in maternal and infant mortality of 74 and 13 %, respectively [61]. The tragic loss of health staff has the potential to result in an estimated 4022 additional women dying during pregnancy and childbirth per year across Sierra Leone, Guinea and Liberia, reminiscent of levels during the civil conflict [62].

Findings during the outbreak have indicated a decrease in RMCH service utilization [63], with evidence that Ebola has disproportionately affected women and children [64]. In the aftermath of Ebola, it is predicted that its devastating consequences will again most be felt by women and children [61]. It is therefore not unreasonable to assume that the discrepancies observed in this study will be exacerbated as a result of the recent Ebola outbreak, causing a further decay of the health care system and an even greater shortage of health workers.

The challenge for Sierra Leone will be to rebuild and strengthen its health system in an efficient and equitable manner. As these findings show, considering existing discrepancies and contextual conditions, over and above populations numbers, are essential to ensure equitable – not just equal - distribution of maternal and child health resources.

Limitations

This study is not without its share of limitations. Notably, sampling errors may have occurred during the development of the sampling frame, impacted by the non-availability of community maps. Similarly, only households where someone was present during the day were sampled, excluding houses where people were away working or caring for their farms. In addition, a pocketing effect may have been introduced due to the random number generators for first house selection. This pocketing effect could influence certain indicators, such as immunisation rates. Bias may also have been introduced during the data collection process. This includes social-desirability bias, which has been noted in other studies regarding reporting of breastfeeding and child health indicators and recall bias, for those indicators (fever, diarrhoea, ARIs) that were collected based on the primary caregiver’s perception of illness without validation by a medical professional. In addition, this survey was conducted during the dry season, which may impact on the reported incidence of certain diseases, notably diarrhoea, which is more prevalent during the rainy season.