Introduction
In sub-Saharan Africa, one in six people live more than two hours away from a public hospital, and one-eighth of the population is farther than one hour away from any health facility. Reference Falchetta, Hammad and Shayegh1 For critically ill and medically complex patients who make it to an initial hospital for evaluation, the process is further compounded by the need for transfer to a higher level of care. Reference Iwashyna2,Reference Mueller, Shannon, Dalal, Schnipper and Dykes3 Previous studies suggest that transfer to a higher level of care is correlated with increased patient morbidity and mortality rates, Reference Durairaj, Will and Torner4,Reference Mohr, Harland and Shane5 higher health care costs, longer durations of stay, Reference Mohr, Harland and Shane6 and that these transfers tend to relocate patients farther from their homes and support systems. Delayed or unnecessary transfers can further negatively impact patients. In developing countries, nine out of ten patients lack access to timely care. Reference Meara, Leather and Hagander7 For acutely ill and injured patients, timely transfer to definitive care can significantly predict patient outcomes. If there are complications, transfer delays can lead to increases in resource consumption and patient morbidity and mortality. The unnecessary transfer of stable patients places undue burdens on receiving tertiary hospitals, such as overcrowding and an increased frequency of immediate counter-referral. Reference Nkurunziza, Toma and Odhiambo8 When transfer systems are inappropriately utilized, it can lead to poor allocation of available resources Reference Southard, Hedges, Hunter and Ungerleider9,Reference Amedee, Maronge and Pinsky10 and overall poor health outcomes.
The coordination and efficiency of the transfer process relies heavily on communication. Communication between the specialized receiving physician and transferring physician determines which patients are appropriate for transfer, helps to stabilize critically ill patients, and ensures that a proper transfer mode is used. Reference Kulshrestha and Singh11 This is a crucial component of the health care system, Reference Rosemann, Wensing, Rueter and Szecsenyi12 and specifically, the Rwandan system. The Rwandan health care system maintains a pyramidal model, whereby each unit offers a specific package of services. There is a need for referral systems to ensure the proper application of services. 13 The current referral system demonstrates a multi-directional pattern with district hospitals referring to each other, as well as sending referrals to tertiary hospitals. Though feasible, transfers from a tertiary hospital to district hospital are uncommon. 13
In Rwanda, the health structure is in pyramid starting from health centers staffed by nurses, district hospital staffed mostly by general practitioners (GPs), and followed by referrals hospitals with specialists. The process of interhospital transfer helps to escalate from lower to higher level of care using hospitals’ ambulances, and this lacks standardization and is plagued by challenges in communication across facilities. Interhospital transfers have historically been associated with the problems in continuity of care due to communication errors and information gaps. Reference Mueller, Zheng, Orav and Schnipper14 Since 1990, on-call medical care and triage systems have been in place in certain parts of the world. These systems facilitate conversations between transferring and receiving physicians, Reference Bunik, Glazner, Chandramouli, Emsermann and Hegarty15 thus strengthening health care accessibility in different countries, including the United Kingdom, the United States, Denmark, and Switzerland. Reference Downes, Mervin, Byrnes and Scuffham16 Developing a strategy to facilitate communication between district hospitals and the receiving transfer center can be an effective way to improve the outcomes for transfer patients. Therefore, transfer delays, unnecessary transfers, and negative patient outcomes might be reduced through the implementation of an on-call consultation service.
This pilot study aimed to provide introductory data on the demographic characteristics and care of patients transferred from district hospital to tertiary hospital in Rwanda. Furthermore, the study aimed to examine the effects of an “[Emergency Medicine] EM Doc on Call” intervention on these variables.
Methods
Setting
This is a prospective, observational pilot study conducted in the emergency department (ED) at Kigali University Teaching Hospital (CHUK [Centre Hospitalier Universitaire de Kigali]), a tertiary hospital and the largest referral hospital in the city of Kigali with a 519-bed capacity. It receives referred trauma and medical patients from all over the country for diagnostic and management purposes exceeding 20,000 patient visits per year. The ED has a 24-bed capacity but is often overcrowded with 48 beds. The CHUK ED uses the Triage Early Warning Score (TEWS) to categorize patients based on the severity of illness. Patients are categorized as RED, ORANGE, YELLOW, or GREEN, with RED assigned to patients with the most severe illness. Reference Wallis, Gottschalk and Wood17
Study Design
This study featured a pre-intervention and intervention phase. In the pre-intervention phase, data were collected regarding patient transfers from ten randomly selected district hospitals in the CHUK catchment area (out of 19 possible hospitals) from November 19, 2019 through January 19, 2020.
The pilot intervention consisted of a senior EM Resident Physician (post-graduate [PGY]-3 or PGY-4) or EM Attending Physician, both known as the “EM Doc on Call” at CHUK, calling GPs at randomly selected district hospitals by mobile phone. The “EM Doc On Call” would respond or inquire about patients currently present at the district hospital and ask about any patients that might require transfer. This call would take place every day after morning rounds in the CHUK ED (approximately 10:00am). The “EM Doc On Call” would remain available during daytime hours (7:00am to 7:00pm) to receive calls and questions from the district hospitals. The number of calls and content of calls received by the “EM Doc On Call” was not recorded.
Due to staff and resource constraints at CHUK during the coronavirus disease 2019 (COVID-19) pandemic, the intervention was restricted to five of the ten district hospitals included in the pre-intervention phase (Figure 1). Specifically, staff from CHUK were relocated to other health facilities dedicated to providing care to patients with COVID-19. This significantly limited the abilities of the research team.
Figure 1. Map of Hospitals Included in the Study.
Abbreviation: CHUK, Kigali University Teaching Hospital.
The five hospitals where the intervention was deployed were selected using a random number generator. Data on patient transfers were collected from these hospitals from March 19, 2020 through May 19, 2020. Pre-intervention and intervention data were then compared.
It should be noted that all COVID-19 patients in Rwanda were treated at specialized facilities erected during the pandemic. Few, if any, COVID-19 patients were treated at district or tertiary care facilities within the data collection period. As such, the data acquired during this study does not include patients with COVID-19.
Data Collection and Management
All patients who were transferred from selected hospitals during daytime hours (7:00am to 7:00pm) were enrolled into the study. Following enrollment, prospective data were gathered by ED research staff on patient demographics, the reason for transfer, patient type, training of transferring doctor, vital sign documentation, injury severity, initial disposition, and final disposition. Information was collected on counter-referral, defined as the transfer of a patient back to the district hospital site of initial transfer. Attention was given to immediate counter-referral upon presentation at the CHUK ED and counter-referral at final disposition. Counter-referral is determined by the ED receiving physician. Immediate counter-referral is used here as a proxy marker for inappropriate transfer, as the emergency physician is deciding on arrival that the patient does not meet criteria for treatment at the receiving facility. Data were collected using a pre-designed questionnaire (Appendix; available online only). Data collection began when a transferred patient arrived at the ED and continued until final disposition (ie, admission, discharge, death, or counter-referral).
Ethics
Research activities were reviewed and approved by the CHUK Institutional Review Board (reference: EC/CHUK/181/2019). The ethics committee determined that informed consent was not necessary for this study.
Statistical Analysis
Data analysis was performed using STATA version 15.0 (Stata Corp; College Station, Texas USA). Descriptive statistical analyses were completed for the overall cohort. Summary statistics were calculated using frequencies and percentages for all categorical variables, and continuous variables were summarized using median values and interquartile ranges. Cases were arrogated and stratified by the observational and intervention phase. Differences in characteristics were assessed using χ² or Fisher’s exact tests for categorical variables and independent sample t-tests for normally distributed continuous variables. A significance level of P <.05 was utilized in the analyses.
Results
Pre-Intervention Phase
During the pre-intervention phase, 285 patients were transferred from district hospitals to CHUK (Table 1). Most patients presented with trauma (56.1%), with the most common indication for transfer being orthopedic evaluation (34.3%). Most common hospitals of origin were Kabgayi (21.0%), Ruhengeri (20.35%), and Kibagabaga (19.3%) Hospitals. The most common length of stay at CHUK was two to three days (26.8% of patients). A GP was the most likely medical provider specialty to initiate a transfer (90.0%). Vital signs were only recorded around one-half of the time (53.2%) before being transferred to CHUK.
Table 1. Characteristics of Transfers During Observational Phase
Abbreviations: TEWS, Triage Early Warning System; CHUK, Kigali University Teaching Hospital.
A total of 216 transferred patients (76.3%) were admitted to CHUK while 67 patients (23.7%) were immediately counter-referred. Of the admitted patients, 166 patients (58.6%) were discharged home, 79 (27.9%) were referred back to their original institution, five (1.77%) were transferred to another hospital, and 33 (11.7%) died (Table 1).
Intervention Phase
A total of 93 patients were transferred from district hospital to CHUK during the intervention phase (Table 2). Similar to what was seen in the pre-intervention phase, patients were most commonly categorized as trauma (53.7%). However, 28% of indications were critical care, as compared to 10.2% in the pre-intervention phase, making critical care the most common indication for transfer (Table 1). This difference in critical care indications was statistically significant (P <.001). Transfer times to CHUK during the intervention phase were quicker (P <.001), with 95.7% of patients being transferred within one day compared to only 34% during the pre-intervention phase. Patients were more likely to be showing emergency signs (55.9%) compared to 31.6% in the pre-intervention phase (P <.001). Vital signs were more likely to be documented before transfer (89.2%) compared to 50.9% during the observational phase (P <.001). During the intervention phase, patients were less likely to be counter-referred and mortality decreased from 11.7% to 9.7%. However, these changes were not statistically significant, likely because the study was not adequately powered for these variables.
Table 2. Comparison of Transfer Characteristics During the Observation and Intervention Phases
Abbreviations: TEWS, Triage Early Warning System; CHUK, Kigali University Teaching Hospital.
Discussion
This study provides introductory data on the transfer of patients from district hospitals to a tertiary hospital in Rwanda. The majority of patient transfers involved traumatic injury with many patients requiring transfer for orthopedic or neurosurgical evaluation.
Pre-intervention data demonstrated several problems with the interhospital transfer process. These data showed significant delays in transfer (43.8% occurring after two days), inconsistent documentation of pre-transfer vitals (46.8% missing vitals), and a high degree of inappropriate transfers as evidenced by immediate counter-transfer (23.7%).
The “EM Doc On Call” intervention indicated improvements in district hospital to tertiary hospital transfers. As a pilot study, it hoped to identify promising improvements of patient care and determine the feasibility of the future implementation of a similar system geared towards improving interhospital transfer in Rwanda. This pilot generated many intriguing data that point towards improvements in this process. Specifically, the pilot intervention led to more timely patient transfers, with 100% of patients transferred within two days in the intervention phase when compared to 66.2% in the observation phase. This may indicate that communication between facilities resulted in a more streamlined transfer process, as patients inevitably transferred are not admitted to the hospital in the Rwandan system but instead stay in the ED until transfer. The “EM Doc On Call” pilot intervention also led to more effective communication between health care facilities, with appropriate pre-transfer vital sign documentation increasing to 89.2% from 53.2%. Such process improvements are likely to lead to better patient outcomes. Prior studies have demonstrated a positive association between clinical outcomes and faster transfer times, Reference Notrica, Evans, Knowlton and Kelly Mcqueen18 as well as with the complete and timely delivery of transfer documentation. Reference Usher, Fanning and Wu19
The intervention was also associated with notable changes in the types of patient transfers. The proportion of patients transferred for a critical care indication greatly increased from 11.9% to 28.0%. There was also a significant decrease in the proportion of patients transferred with the presence of emergency signs, from 66.3% in the observation phase to 44.1% in the intervention phase. This may indicate that patients were more adequately stabilized prior to transfer secondary to discussion with the “EM Doc on Call,” especially when contextualized by the higher proportion of patients considered “critical” by the referring physician during the intervention phase.
While not statistically significant, there was a decrease in patient mortality and counter-transfers following implementation of the “EM Doc On Call” intervention. The study set out to test a system that could improve the interfacility transfers, and the data appear to indicate that this method is extremely promising.
Limitations
Of the several limitations that should be considered within the context of this study, perhaps the most apparent is that the intervention was employed during the COVID-19 pandemic. This certainly affected the capabilities for data collection and resulted in a smaller sample size than initially intended, secondary to resource reallocation at CHUK. Any pandemic-related disruptions to Rwanda’s medical landscape at the time of this study could have also altered the results to a degree that might not fully represent what would have been observed prior to 2020.
Other important limitations of this study relate to the collection of the data. First, information was not collected on how patients were transported to their original district hospital. Previous research in other sub-Saharan countries has established that a dysfunctional ambulatory network contributes to delays in interhospital referrals. Reference Tayler-Smith, Zachariah and Manzi20 Determining whether patients were transported by the Emergency Medical Service (SAMU [Service d’Aide Médicale d’Urgence]) or by private vehicle could provide a more complete picture of the many factors impacting patient referrals and outcomes in Rwanda. This information could also help determine if future training for SAMU could direct more emergent patients to CHUK faster.
Secondly, patient outcome data were not obtained for those recommended to stay at district hospitals as part of the referral system. To more accurately assess outcomes of the intervention, further research should thoroughly track outcomes for patients who remain at district hospitals in addition to those transferred to a tertiary hospital. Doing so will better control for the effects of changes in the types of patients transferred as a result of the “EM Doc On Call” pilot intervention, such as differences in illness severity.
Lastly, this intervention was only applied during daytime hours. While it is more difficult to staff referral phone calls overnight, the design of a 24-hour research model which includes nighttime workers will ensure that overnight referrals and transfers are not missed. Doing so will increase the scope of the study’s impact and provide a more complete dataset.
Conclusion
The “EM Doc On Call” intervention set out to begin to classify transfers in Rwanda so that data could be collected on the pilot intervention aimed at improving the timeliness and documentation associated with interhospital transfers. While there are valid limitations in the study, the data collected are extremely promising and warrant further study in the future. Future studies can further explore the effects of this intervention on patient-oriented outcomes while controlling better for confounders, such as lack of nighttime data collection or reduction of study size secondary to unforeseen resource reductions. The epidemiological data from this study, while perhaps not complete or representative of all Rwandan interfacility transfers, are the first attempt at documentation of this critical aspect of medical care and can help to inform further research and interventions regarding interhospital transfer.
Conflicts of interest/funding/disclosures
The authors have no conflicts of interest. The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the views of or any academic organizations. The funders played no role in the collection, analysis, or reporting of the data. Funding for this research was provided through Kigali University Teaching Hospital (CHUK) small research grants. The funders had no role in the study design, data collection, or reporting processes. Preliminary results from this work were presented at the Society for Academic Emergency Medicine and shared to the CHUK-Directorate of research and education.
Author Contributions
Authors contributed as follows to the conception or design of the work; the acquisition, analysis, or interpretation of data for the work; and drafting the work or revising it critically for important intellectual content: VN, AB, AG, and KDM contributed 20.5%; CU, DU, CR, SL, SDN, NJP, GGD, ED, and MVA contributed 2.0%. All authors approved the version to be published and agreed to be accountable for all aspects of the work.
Acknowledgments
The authors would like to thank the emergency department staff at the Kigali University Teaching Hospital (CHUK) for all their help and support of the study, as well as the district hospitals who participated in this research.
Supplementary Materials
To view supplementary material for this article, please visit https://doi.org/10.1017/S1049023X23005927
Open access
