Objectives

The Harvard T. H. Chan School of Public Health, the Massachusetts General Hospital (MGH) Global Health Innovation Lab, and the Kisumu Medical and Education Trust (KMET), a non-profit organization in Kisumu, Kenya partnered to assess contextual barriers to PPH management in Kenya. The purpose of this qualitative assessment was to explore the main challenges to management of PPH among providers at the primary health care facility level in Kenya, and to understand how decisions about referrals are made to get women to the care they need. The study was motivated by the following overarching research questions:

1. How did PHC providers manage patients with PPH that were referred to higher-level facilities?
2. What is the typical experience of a PHC provider when managing PPH and the referral process for these cases?
3. What are the main challenges providers at PHCs face in managing and referring PPH cases?
4. What recommendations do PHC providers have for improving PPH management and referral processes?

Study setting

The Kenyan health system is decentralized to the county level, where county governments are responsible for the coordination of health services, including referral [6]. The health system has six levels of care. Level 1 includes community health units and community health workers who provide basic preventive and promotive health services at the community level. These units play a role in disease surveillance and health promotion activities, contributing to the early detection and prevention of outbreaks [5]. Levels 2 and 3 are primary health care facilities (PHCs) that provide basic obstetric care and refer emergency cases to Level 4 or 5 county-level referral health facilities when surgical interventions or blood transfusions are required. Level 2 facilities, often referred to as dispensaries, provide outpatient services, including antenatal care, immunizations, and treatment of common illnesses. Certain Level 2 healthcare facilities perform delivery services; however, they are required to discharge mothers by the end of the day due to the lack of inpatient care services. Level 3 facilities, known as health centers, offer more comprehensive services, including delivery services, basic emergency obstetric care, and inpatient care. These facilities are typically staffed by clinical officers and nurses. Level 4 facility staff include medical officers and occasionally specialists. They manage common medical conditions and minor surgical procedures [6]. Level 5 hospitals, also known as county referral hospitals, offer a wider range of services compared to PHCs, including specialized outpatient services, emergency surgery, and inpatient care for more severe cases. They often have better infrastructure and more specialized staff, such as surgeons and anesthetists, compared to lower-level facilities [7, 8]. The major difference between level 4 and 5 facilities is that Level 5 are teaching hospitals and are better equipped. Level 6 includes national-level referral hospitals which are under the national government and semi-autonomous in their operations, of which there are five facilities across the country [7, 8]. These national referral hospitals provide specialized care, advanced diagnostic services, and are equipped for major surgeries and complex medical conditions. They serve as training centers for medical professionals and are often involved in medical research and the development of clinical guidelines [7].

This study was conducted in three counties–Nairobi, Kisumu, and Kakamega—in western Kenya. There is wide variation between counties in the proportion of births that take place in a facility, ranging from 47% to 89% of deliveries in the three study counties [9]. Each county has at least one Level 5 referral hospital, which serves as the coordinating and referral hospital for lower-level facilities within the county. These referral hospitals have the capacity to provide basic as well as comprehensive emergency obstetric care (CEmOC), including blood transfusion and surgical intervention, while the PHCs (Levels 2 and 3) provide basic emergency obstetric care (BEmOC) [5, 10]. BEmOC includes administration of IV antibiotics, magnesium sulphate, and parental oxytocics as well as performing manual removal of the placenta, removal of retained products of conception, assisting in vaginal delivery (e.g., by vacuum extraction), and performing newborn resuscitation. CEmOC includes all of the BEmOC functions as well as performing caesarean sections, providing emergency obstetric anesthesia, and administering blood transfusions. Providers are trained on and expected to adhere to the Kenyan National Guidelines for Quality Obstetrics and Perinatal Care [5].

Study design

We employed a qualitative case study and phenomenological approach to understand the experiences of PHC providers managing and referring patients who developed PPH. The case study approach allowed us to review the details of specific patient cases to better understand exactly which steps were taken at the PHC to identify, manage, and refer for PPH. Building off these cases, the phenomenological approach allowed us to understand the personal and lived experience of providers working at PHCs in Kenya managing PPH referral cases.

Sampling procedure

From 31^st October 2018 to 28^th February 2019, we identified patients who had delivered in a PHC and were subsequently referred for PPH management in one of three referral hospitals in Nairobi, Kakamega, and Kisumu counties where the study team was collecting data on postpartum hemorrhage quality of care at referral hospitals [10]. The three referral hospitals were part of a larger study and were purposively selected because they manage high volumes of deliveries (between 17 and 50 per day in 2018) and therefore see large numbers of PPH cases. Patients were recruited into the study at the referral facilities, where they were asked a series of questions about their PHC experience prior to referral. Patients were eligible if they were over 15 years old, had delivered at a PHC and later were referred to the study facility for management of PPH or suspected PPH, and were sufficiently stable to provide consent. Enumerators approached eligible patients once they were stable (and before discharge) to ask for consent to participate. The enumerator requested information on where the patient received care, including the name of the facility, and the date and time at which she arrived at the PHC for delivery, and asked for permission to contact the PHC(s).

Data collection

Using the information provided by patients, an enumerator visited the facility in person to invite the relevant providers to participate in the study. Interviews were scheduled at times convenient to the provider and followed a semi-structured guide to capture details about the specific PPH case and how it was managed; demographic information such as background and training on identifying and managing PPH, job satisfaction, and motivation; and perspectives on the typical PPH experience at PHC facilities (see the full guide in S1 Text. A snowball sampling technique was used, where interviewed providers identified other members of the care team involved in each case. If another member of the care team was identified, enumerators used that information to follow up with the additional provider and complete the same process outlined above. The overall aim was to interview every provider involved in a patient’s care.

Data analysis

A semi-structured interview guide was developed by study team members at Harvard and MGH. The guide was reviewed and revised by the KMET team in Kenya to ensure the appropriateness of each question for the study context. After the first few interviews, a few additional probes were added to the study guide, including around the coordination of transportation as well as availability and use of data, as these topics emerged from early interviews as potentially important themes.

The interviewers were all trained Kenyan researchers with backgrounds in qualitative methods. Two of the three interviewers also had clinical backgrounds in maternity care. The interviewers participated in a refresher course on research ethics and methods before conducting interviews. Interviews were conducted in English in a private area within the health facility and were audio recorded. Interviews lasted approximately one hour. While we did not pre-define saturation, our goal was to follow a case series approach to interview all PHC providers who managed and referred a PPH case to one of the referral facilities during the study period. We therefore stopped data collection once the study period was over.

The recordings were transcribed verbatim by a Harvard research assistant and were checked by a research assistant on the KMET team. All identifying information within the final transcripts was redacted, including facility and provider names.

We used an inductive thematic approach to code and organize the data. The first author read the transcripts multiple times to understand the data and developed an initial codebook without pre-defined assumptions. Codes were then categorized into overarching themes and subthemes before applying the codebook to the entire dataset using Atlas.ti (version 8.4.4). Two of the study authors applied the codebook to 20% of the transcripts independently and discussed discrepancies in coding decisions to ensure consistency before coding the remainder of the transcripts. As new codes emerged, they were discussed by the data analysts, added to the codebook, and applied to all transcripts such that both the codebook and the emerging themes and sub-themes were revised in an iterative process. Once completed, codes were reviewed to identify the most salient themes regarding the challenges that PHC providers experience in managing and referring PPH cases. The dictionary of codes is included in the “S2 Text”. We also reviewed deviant findings across all transcripts to assess whether the themes were valid. The themes were shared with subject matter experts on the study team, including medical doctors in the United States and trained Kenyan nurses. Thematic saturation was reached within approximately 60% of analysis, such that the themes identified in the first 8 interviews were consistent with the themes in the remaining interviews.

Ethics statement

This study was approved by the Institutional Review Boards of Jaramogi Oginga Odinga Teaching & Referral Hospital and the Harvard T.H. Chan School of Public Health (#IRB00047360). Permission to conduct the study was obtained in each study facility. Verbal consents were obtained from each participant after reading a consent form, including patients who consented to their case being reviewed in this study. Interviews with providers were conducted in settings where confidentiality could be maintained. All identifying information was removed from transcripts before analysis, including references to various referral hospitals and other facilities in the geographic area. The audio files were deleted as soon as transcription was complete, per IRB guidelines.

Inclusivity in global research

Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research is included in the “S1 Checklist”.

Sample description

A total of 14 in-depth interviews were conducted with health workers at PHCs. All providers were nurse-midwives. Providers had an average age of 41 years and approximately 16 years of experience, with an average of 8.5 years working in the facility in which they were interviewed. The providers had an average of 4.7 years’ experience in their current position and reported attending to approximately 25 deliveries per month. The sample included a range of both early-career health workers and experienced health workers. We provide a breakdown by provider cadre in “S3 Text”). The providers reported that the facilities conducted an average of approximately 75 deliveries per month, ranging from a low of 10 to a high of 200. While we did not collect data on referrals from PHCs, providers estimated that their facilities referred maternity patients between 1–5 times per month for a range of reasons including PPH. All of the facilities were open for both day and night shifts, though the number of staff on duty was typically lower for night shifts versus day shifts. The PHCs cared for a variety of health conditions outside of maternity care, including preventative and other primary care functions. On average, the facilities saw a total of approximately 350 patients per month, ranging from a low of 70 to a high of 800. Table 1 presents a summary of the provider and facility characteristics in this study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Provider and facility characteristics.

https://doi.org/10.1371/journal.pgph.0003842.t001

Main results

Two key themes emerged from the data to describe the challenges providers face in managing and referring PPH cases at primary care facilities. The first set of challenges were related to the identification and management of PPH at PHCs, which ultimately necessitated referral to higher levels of care. Within this theme, important sub-themes included limitations in facility infrastructure, limitations in provider knowledge, training, and experience, and patient-level factors that made the provider’s ability to identify and manage PPH more challenging.

A second key theme included barriers to effective and timely referrals once a decision had been made to transfer the patient. At this stage, providers faced internal delays in decision-making, interpersonal and communication challenges between facilities, structural barriers to efficient referrals, and additional patient-level factors that influenced the timeliness of referrals. These themes and subthemes are described in more detail below.

Key Theme 1: Challenges in PPH management at primary care facilities.

PHCs are readied to provide basic emergency obstetric care, including actions to prevent or treat PPH such as administration of oxytocic drugs, manual removal of the placenta and other products of conception, and monitoring for and treating blood loss [9]. However, the providers we interviewed cited numerous challenges in identifying and managing PPH, including facility-related limitations (Table 2), gaps in provider knowledge and training (Table 3), and other patient-level factors that made PPH identification and management difficult (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. PPH management—limitations in facility infrastructure.

https://doi.org/10.1371/journal.pgph.0003842.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. PPH management–limitations in provider knowledge, training, and experience.

https://doi.org/10.1371/journal.pgph.0003842.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. PPH management—patient-level factors leading to delays in PPH identification and management.

https://doi.org/10.1371/journal.pgph.0003842.t004

Sub-theme 1.1: PPH management—limitations in facility infrastructure. Providers attested to limited staffing, especially during night and weekend shifts. On average, there were 2–3 nurses on staff during the day, and only one nurse on duty at night in each facility. Many providers found it difficult to adequately address patient needs due to limited staff, especially when emergencies arose (Table 2, Quote 1). Some providers noted that staffing in the maternity ward was ad hoc and a provider was only assigned to the ward if a woman was in labor. In severely understaffed facilities, a single provider covered the entire facility on their own, resulting in high workloads and limited ability to prevent emergencies such as PPH. Providers in these situations sometimes called for help from security guards, janitors, or other staff in the facility (Table 2, Quote 2).

In addition to low levels of staffing, facilities lacked higher-cadre staff to advise on and provide support in emergencies. Some providers suggested that they would benefit from having a Medical Officer (MO) on duty at all times in order to comply with Kenyan clinical guidelines that suggest nurses should consult MOs if they are unsure of how to treat specific patients. In some facilities, an MO was always available either in-person or by phone. In others, MOs were difficult to reach and were rarely available for in-person consultations. One nurse commented that they would make fewer referrals if they had more qualified staff on duty, including MOs (Table 2, Quote 3).

Supply shortages were relatively common. Distribution of supplies is managed by the Kenya Medical Supplies Authority (KEMSA), with budgets disbursed to each facility by their county government to purchase the supplies they need. Five of the fourteen interviewed providers cited challenges with limited supplies of gloves, sutures, and IV fluids, as well as some medications such as misoprostol and antibiotics (Table 2, Quote 4). Most of the facilities in this sample were not equipped to provide blood transfusions. However, one facility mentioned that they could transfuse but were out of blood when dealing with this particular emergency and had to borrow from another facility before transferring the patient (Table 2, Quote 5). In Kenya, PHCs are not typically expected to have blood transfusion capacity. Providers stated that if a shortage of medication or supplies occurs, either the patient or provider must pay out-of-pocket for the necessary supplies at a local pharmacy. In some cases, as one nurse described, facilities rely on other facilities to fill commodity gaps (Table 2, Quote 6). This practice can lead to additional delays in care.

Sub-theme 1.2: PPH management–limitations in provider knowledge, training, and experience. In general, PPH was perceived to be rare, and providers had limited hands-on experience managing PPH patients. This lack of experience influenced the level of confidence providers felt in being able to identify causes of PPH and describe the steps that should be taken to manage excessive bleeding. For several less-experienced providers, the patient under discussion was among the first PPH cases that they had managed on their own. These providers described patients’ bleeding as being ‘uncontrollable’ or ‘too far along’ and outside of their skill set, leading to the decision to refer.

Monitoring of patients in the postpartum period is a critical action for early identification and management of PPH [5]. However, monitoring was relatively infrequent in this sample and often the responsibility for monitoring blood loss was placed on the mother herself, who was told to alert staff if she was bleeding heavily (Table 3, Quote 1). Providers noted that they were aware of guidelines for the recommended frequency of monitoring (e.g., hourly) but their description of the PPH cases suggested that adherence to these guidelines was irregular. Low levels of monitoring led to catching PPH when it was already well-established and challenging to manage. Even when providers were aware of PPH risk factors for specific patients, they described being surprised by the development of excessive bleeding (Table 3, Quote 2).

Quantifying blood loss was highlighted as a key challenge in diagnosing PPH. While most providers defined PPH as 500 ml or more of blood loss after delivery, they mentioned that they often struggled to tell how much blood had been lost, relying on gauze pads, the amount of blood on the sheets or floor, or the alertness of the patient to assess the severity of the situation. A nurse described this estimation method as being an unreliable source of information on the mother’s condition (Table 3, Quote 3). A more experienced nurse highlighted the need for additional data points such as skin pallor, patient history, and other considerations that might indicate whether the patient needed additional steps for PPH. These data points, however, were learned through experience managing PPH cases, which not all providers had.

Providers suggested that gaps in knowledge are greatest regarding how to manage PPH once it has started (Table 3, Quote 4). Another nurse suggested that trainings were outdated, and they needed information on the latest practices for PPH management (Table 3, Quote 5). Providers stressed that trainings should be hands-on and practical so that providers gain confidence in their decision-making during emergencies.

Sub-theme 1.3: PPH management—patient-level factors leading to delays in PPH identification and management. Providers additionally highlighted factors at the patient level such as delays in seeking antenatal or delivery care that made it challenging to screen for, prevent, and manage PPH. When patients arrive at a facility in later stages of labor (i.e., needing to go straight into delivery) and without any antenatal screening, providers are often unable to check for PPH risk factors or refer them immediately to a more suitable facility (Table 4, Quote 1). In one case, it was only in the postpartum period when a patient told the nurse that she had a history of bleeding in her six prior deliveries (Table 4, Quote 2). Providers felt that better antenatal screening and more information on their patient’s histories and risk factors would improve their ability to prevent and manage PPH.

Key Theme 2: Barriers to effective and timely referrals.

The Kenyan guidelines for obstetric and perinatal care stress the importance of timely referrals in cases where the first facility is unable to provide adequate care [5]. Providers highlighted some of the challenges they face in coordinating these referrals, including internal delays in decision-making, interpersonal and communication challenges with receiving facilities, structural barriers such as lack of transportation and poor road quality, and patient-level factors that influence the referral process. These factors lead to delays in care for patients with PPH, sometimes putting the patient at risk of serious complications or death.

Sub-theme 2.1: Referral challenges—internal delays in decision-making. Slow decision-making by providers was recognized as a source of delays for referrals. In some cases, providers suggested that a referral should have been made immediately rather than accepting patients with significant risk factors (Table 5, Quote 1). In other cases, there was slow decision-making about referral after patients had delivered and were recovering in the postpartum ward (Table 5, Quote 2). Providers described a ‘wait and see’ approach in many cases before ultimately referring, hoping that the patient would improve on their own (Table 5, Quote 3). In other scenarios, providers inherited a case during shift turnover and were dismayed that their colleagues had not already started the referral process (Table 5, Quote 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Referral challenges—internal delays in decision-making.

https://doi.org/10.1371/journal.pgph.0003842.t005

Sub-theme 2.2: Referral challenges—interpersonal and communication challenges between facilities. After deciding to refer patients, providers cited interpersonal challenges in the relationships between PHC providers and higher-level referral providers, where providers in referral facilities were perceived to be rude or dismissive of PHC providers. Providers commented that the referral facilities insinuated poor clinical management of deliveries at PHCs (Table 6, Quote 1) One particularly poor working relationship is demonstrated by the account of a nurse who expressed hesitation around calling the referral facility out of fear of poor treatment by the staff (Table 6, Quote 2). However, poor inter-facility communication challenges were not the case in all settings, and some providers described the relationship with the referring facility as generally positive (Table 6, Quote 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Referral challenges—interpersonal and communication challenges between facilities.

https://doi.org/10.1371/journal.pgph.0003842.t006

Poor working relationships caused providers to have reservations about contacting specific facilities, who would instead call a facility that was geographically more distant in order to avoid communicating with their assigned facility. In other cases, the PHCs would have to call on the sub-county coordinator as an advocate and intermediary to broker the referral, adding further delays to the process as the coordinator negotiated a solution on behalf of both facilities. One nurse described a case in which she had decided to call a facility outside of her referral network to get faster care for the patient. A staff member she spoke to at her non-assigned referral facility responded to their emergency call, but did not want to accept the patient without involving the sub-county coordinator for permission. A back-and-forth conversation between the PHC, the referral facility, and the sub-county coordinator resulted in a delay in getting the patient to appropriate care (Table 6, Quote 4). In these cases, the sub-county coordinator was described as a critical resource for the PHCs when trying to reach referral facilities.

Sub-theme 2.3: Referral challenges—structural barriers to effective referrals. Kenyan guidelines suggest that it is the responsibility of the referring facility to be aware of the capacity of the receiving facility to manage the patient before making any referrals [5]. In several cases in this study, providers were refused access for their patient based on claims that the referral facility lacked supplies such as blood, or had limited capacity or staffing. This information was relayed by calling various referral facilities and often meant that the provider made multiple attempts before finding a facility that would accept their patient (Table 7, Quote 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. Referral challenges—structural barriers to effective referrals.

https://doi.org/10.1371/journal.pgph.0003842.t007

Transportation was a persistent challenge in nearly all facilities, including the cost of ambulance and taxi services, lack of vehicle fuel even when an ambulance was present, unavailability of ambulance drivers, and poor road quality that further delayed patients from receiving adequate care (Table 7, Quote 2). While some facilities had their own ambulance on site, many more relied on external services for transportation, such as faith-based organizations, local NGOs, or the Kenya Red Cross. Facilities rarely had enough staff to accompany patients to referral facilities, and would sometimes send the patient alone in a taxi. One provider reported that these challenges are frequent and can cause significant delays (Table 7, Quote 3). In some cases, delays may add several hours to the referral process or may require extensive networking and negotiation on the part of the provider on duty to find a mode of transport (Table 7, Quote 4).

While guidelines were available for how referrals should be coordinated, providers described deviating from these procedures when a case required immediate attention (Table 7, Quote 5). Overall, providers referenced these ad hoc decisions as being necessary to manage the emergency that they were addressing. For example, the protocol for emergencies in one facility is to call the referral facility, ask them to accept the patient, and then wait for the referral facility ambulance to arrive. However, the severity of a specific PPH patient meant that they had to skip these steps and send the patient by taxi to the referral facility without advance notice (Table 7, Quote 6).

Sub-theme 2.4: Referral challenges—patient-level factors influencing referral delays. At the patient level, providers mentioned that financial constraints and concerns about the distance of certain referral facilities contribute to delays transferring patients to appropriate care. While maternity services are free in Kenya as of 2013, patients often incur hidden costs such as transport charges to and from referral facilities. These costs are not covered by the national maternity policy [6]. Providers recounted experiences in which patients refused referral because they were unable to afford the bus fare to come home afterward (Table 8, Quote 1). In other cases, providers had to bargain with taxi or ambulance drivers to decrease the out-of-pocket patient charges for quality transportation (Table 8, Quote 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Referral challenges—patient-level factors influencing referral delays.

https://doi.org/10.1371/journal.pgph.0003842.t008

Main findings

This study revealed two main sets of challenges in management and referral of PPH cases from primary care facilities. First, providers at primary care facilities faced a variety of challenges in identifying and managing PPH. PPH was a relatively rare event, and some of the less experienced providers had limited experience managing PPH cases. Results from the larger study estimated a 9% (CI:7%-11%) rate of suspected PPH among observed vaginal deliveries within the three referral hospitals [11]. However, given that these are referral hospitals, this rate is likely more than what might be seen in the broader population or at the PHC sites in which these interviews took place. In addition, facilities were understaffed, lacked higher-level cadres, and had shortages of key medications and supplies for delivery. There was a general limitation in knowledge and training around PPH, though providers suggested that the main challenges were in knowing how to manage bleeding once it had begun (which extend into interpersonal communication challenges), rather than gaps in knowing how to prevent PPH or to identify excessive blood loss. Finally, patient-level delays in seeking care, either through low antenatal care attendance or delays in arriving at a facility to give birth added challenges for PHC providers.

As a result of these issues, providers referred PPH patients whose needs exceeded their abilities. The second set of challenges arose in the process of coordinating these referrals in an effective and timely manner. Notably, delays in making the decision to refer meant that cases were sometimes kept at the PHC for many hours before action was taken. Once a decision was made to refer, providers met with interpersonal and communication challenges with referral facilities, which often required arbitration by a sub-county coordinator. This tension between PHCs and referral facilities meant that PHC providers sometimes contacted a less convenient referral facility to avoid poor treatment by staff at their assigned referral facility. When referral facilities were willing to take patients, there often were additional delays due to structural barriers such as lack of transportation, supply shortages at referral sites such as blood for transfusion, and shortage of staff to accompany the patient. Finally, patients themselves sometimes resisted referrals based on the high cost of transportation to return home.

Other studies on referrals in low- and middle-income countries have identified similar challenges. For instance, we found that providers delayed making decisions about referrals and chose instead a ‘wait and see’ approach. Studies of obstetric emergencies in Uganda and Ethiopia have similarly found that providers rely on their personal judgment and experience over clinical guidelines [12, 13]. Many studies have similarly found that low provider knowledge and training on PPH management—and emergency obstetric care more broadly—is a major contributing factor to delays in PPH management and referrals [13, 14]. A 2017 review of maternal deaths in Kenya found that 75% of those deaths were driven by provider-related factors such as lack of training or delays in transferring the patient [4].

Clear communication between health centers and referral facilities is highlighted in global and national guidelines as being critical to smooth referrals [7, 10]. Other studies in the region have found that poor pre-referral communication may lead to multiple additional referrals before patients are ultimately placed in an appropriate hospital [14]. Our study found that communication was sometimes hampered due to interpersonal relationship challenges between facilities, where PHC providers avoided calling their assigned referral hospitals out of fear of being treated poorly. This was not true in all cases, but did add to the many reasons for delayed referrals.

Facility-related challenges such as supply shortages, inadequate staff, lack of ambulance services or funds to cover transportation costs were common, both in terms of properly managing PPH as well as supporting timely referrals. These issues have been documented elsewhere as being major contributors to delays [15, 16]. In Kenya specifically, an analysis of the most recent Service Provision Assessment found that 42% of facility births occur in a facility not able to offer the full package of basic emergency care functions [17], suggesting that if an emergency should occur, these births might not be properly managed.

At the patient-level, it has been widely shown that delays in seeking care contribute to complications when women reach the facility [18, 19]. However, we found that there are additional delays at the patient-level after she has reaches the first facility and before being referred onward. For example, patients may object to referral because they cannot afford transport home from the referral site. In other studies, researchers have found that patients may also prefer not to visit a referral site due to the perception that higher-level facilities are overcrowded, have long wait times, or offer poor quality care [13, 18]. In our study, no such objections about higher-level facilities were found.

Strengths and limitations

The strength of our study is in the rich data on challenges not only in PPH management at the primary care level but also in–for the same cases–transfer to higher levels of care. There are several limitations worth noting. First, the patient cases that were examined in this paper may not be a representative sample of the average PPH referral case, as there were some instances in which a patient was not recruited into the study (e.g., if the woman’s condition never stabilized at the referral facility). We also did not predetermine saturation in this study to guide our sampling strategy. Instead, we sought to interview all providers responsible for managing and referring a PPH case during the study period. We did, however, find saturation in themes after analyzing approximately 60% of our interviews where the identified themes repeated in the remaining data. Adding additional interviews, in particular from more varied facility types (e.g., more rural) may have added new perspectives on the identified themes outside of our findings for the more urban sample in these counties. The sample size for this study was relatively small (n = 14 interviews). Therefore, the findings may not be generalizable, nor will they be generalizable to geographies outside of the study sites that may face greater challenges in referrals, for example, for very remote locations. Finally, this study focused on the perspective of PHC providers, but improvements in referrals for emergency obstetric cases will require multilevel, comprehensive interventions. As such, further research should be conducted to understand the experiences of both lower- and higher-level providers, patients, health system coordinators and supervisors, as well as policymakers and funders. Lastly, future research should explore the underlying reasons for the delayed transfers and the challenges around transfers in greater depth.