Non-Communicable Disease and the Need to Strengthen Prehospital Care: Experience from Kigali, Rwanda

Background Non-communicable diseases (NCDs) are increasing in incidence in low-and middle-income countries (LMICs). Many NCDs present as emergencies which require emergency response. We aimed to review NCDs transported by Service d’Aide Medicale Urgente (SAMU), the public ambulance service in Rwanda. Methods Descriptive analysis of a REDCap database was performed for medical cases between December 2012 to May 2016 and compared to non-medical cases. Student’s t-test, ANOVA, and chi square test were run with p < 0.05 considered significant. Results A total of 832 patients were seen. 51% were female, and the average age was 57 years old + 22. Cardiac indications were most common (52%) followed by diabetic complications (22%), cancer (16%) and stroke (10%). Patients with NCDs were more likely to be tachycardic, HR>100 bpm (OR 2.61 [P<0.001]), hypoxic with SPO2 <90% (OR 3.28 [P<0.0001]), hypotensive with blood pressure <100 mmHg (OR 2.65 [P<0.0001]), and tachypneic with respiratory rate >20 breaths per minute (OR 2.06 [P<0.0001]). Conclusions Access to emergency services is essential for NCD management. Compared to other emergencies treated by SAMU in Kigali, Rwanda, the NCD cohort had a higher acuity despite being a smaller proportion. As NCDs increase, investment in robust prehospital care will have substantial value in LMICs.


4
An epidemiological transition from communicable infectious diseases to NCDs is occurring in low-and middle-income countries (LMICs), as the incidence of NCDs in LMICs is rising. 3 Annually, 85% of all premature deaths in the world from NCDs occurred in LMICs. 1 By 2020, NCDs are estimated to contribute to 7 out of every 10 deaths in LMICs. 4 On the African continent, NCDs are estimated to become the most common cause of death by 2030. 5 This is leading to an impending crisis for many African LMICs which have made great strides in addressing communicable diseases but have not addressed NCDs.
Specific to Rwanda, NCDs are an increasingly important cause of morbidity and mortality, accounting for an estimated 36% of all deaths annually. 6 As of 2013, NCDs accounted for at least 52% of district hospital outpatient consultations and 22% of district hospital admissions according to the Rwanda Health Management Information Systems. The probability of premature death from one of the four major NCDs (cardiovascular disease, diabetes, cancer, and chronic respiratory diseases) in Rwanda is 19%. 7 These NCDs result in life-threatening emergencies such as myocardial infarction, sepsis, coma, and seizures as well as significant morbidity, like limb amputation from delayed treatment. However, there is limited literature on management of NCD patients in the prehospital setting in Rwanda or in LMIC settings.
Service d'Aide Medicale Urgente (SAMU) is Rwanda's only public emergency medical service and sees nearly 4,000 patients per year. 8 SAMU has used an electronic prehospital registry since 2013 to track prehospital care in Kigali. In this paper we aimed to describe the incidence of NCD patients managed by SAMU in Kigali, Rwanda.

Study Context and Setting
Rwanda has a population of nearly 13 million people with 80% living in rural settings. 9 SAMU was established by the Rwandan Ministry of Health in 2007. It receives roughly 3,000 calls per month at the country's only 912 emergency call center located in Kigali. 8 SAMU has its own fleet of 13 ambulances in Kigali, but works with 270 total ambulances located at district hospitals throughout the country to provide emergency and transfer care for the entire population.
Data Collection and Patient Sample SAMU staff complete a 'run sheet' for every patient encounter consisting of patient information including age, gender, vitals, location of incident, as well as management on the field and during transport. Data from patient run sheets was manually transferred and stored using an electronic registry in REDCap (Vanderbilt University; Nashville, Tennessee, USA) created in collaboration with Brigham Women's Hospital/Harvard Medical School, Virginia Commonwealth University (VCU), and SAMU. 8 For this study, we evaluated the registry for all available data since inception on all patients with NCDs treated by SAMU. We focused on complaints related to cardiovascular diseases, diabetes, cancer and stroke. This cohort was compared to all-cause emergency patients seen by SAMU.

Variables
The registry captures demographics, disease information, field vital signs, presentation, treatment, and transportation. Additionally, the SAMU dispatch team uses a triage system categorizing urgency for each case as "absolute," "relative," or "no urgency" based on mechanism, clinical presentation and vital signs. We further classified the data using vital signs to determine acuity of illness. Systolic blood pressure (BP) less than 100 mmHg was considered hypotensive and greater than 160 mmHg hypertensive. Blood glucose was less than 60 mg/dL or 3.3 mmol/L was considered hypoglycemic and greater than 300 mg/dL or 11.1 mmol/L as hyperglycemic. Heart rate (HR) greater than or equal to 100 beats per minute was considered tachycardia and oxygen saturation (SpO2) less than 90% as hypoxic. Tachypnea was a respiratory rate (RR) greater than or equal to 20. Shock index was calculated for all patients and defined as HR divided by systolic BP. Transport destinations used included referral hospitals (highest level of care), district hospitals (mid-level care), and health centers (primary care). Patients undergoing primary transportation were brought to their initial point of contact with the healthcare system while patients undergoing secondary transport were transferred between health facilities.

Data Analysis
The data was analyzed using descriptive statistics performed on SPSS version 25.
Student's t-test, ANOVA, and chi square test were run with p < 0.05 considered significant.
Significant findings were further quantified by calculating odds ratios. The Proportional Mortality Ratio was determined and compared mortality due to NCDs with all-cause mortality of all patients in the SAMU registry. Only patients with recorded values for a given variable were included in analysis.

Ethical Consideration
The Ministry of Health of Rwanda and VCU approved the study plan. The approval included access to patients' medical records. No informed consent was required. This project falls under a Memorandum of Understanding between the Ministry of Health of Rwanda and VCU to build trauma and emergency capacity in Rwanda.

Results
Demographics SAMU responded to a total of 11,161 patients between December 2012 and May 2016, of which 832 patients (7.5%) had at least one diagnosed NCD. The average age was mean 7 57 years old (standard deviation ± 22) years and the median was 60 years old (IQR: 60-75). The largest age group was > 70 years, accounting for 32% of the NCD population.

Discussion
This is one of the first studies to look at management of NCDs in an LMIC setting. In Kigali, Rwanda, the SAMU prehospital ambulance service routinely manages NCD-related cases, although at a smaller proportion than obstetric and traumatic cases. However, these NCD patients tended to be older and more critically ill and have higher odds of dying than the rest of SAMU patients based on Proportional Mortality Ratio.
SAMU patients with NCDs had more vital sign derangements than other emergency patients. This may be because of delays in recognition of illness, delays in accessing care, and/or delays in availability of quality services. In obstetrics and gynecology, the three delay model describes delays: deciding to seek appropriate medical help for an obstetric emergency, reaching an appropriate obstetric facility, and receiving adequate care when a facility is reached. 10 This model can be translated to delays for NCDs. Patients in many LMICs experience delays in care-seeking generally due to sociocultural factors. 11,12 Lack of awareness of NCDs by the public or by medical staff, cultural practices contributing to delays in accessing care, cost of accessing care, shortages of staff or complexities of the current healthcare system may contribute to challenges for patients with NCDs. The general public may not be aware of the signs and symptoms of NCDs to seek help. The symptoms may be vague; for example, nausea and heartburn are known to be atypical symptoms of myocardial infarction in HIC settings. 13 The medical staff may not be familiar with signs and symptoms of acute NCDs since these are still fairly underrecognized. Lack of training and lack of nurses were previously described as barriers to adequate ICU care in LMICs. 14 Cultural practices may lead to delays in seeking care. Patients may seek care from traditional healers, for example. 15,16,17 These barriers to care have been noted across a variety of settings, in both HICs and LMICs. 18,19 The infrastructure of the health care system may also be a barrier. The referral system acts as a gatekeeping mechanism and may limit getting care for emergent conditions since patients need to progress to higher levels through the system. Furthermore, the Our study has several limitations. This study was limited to the prehospital setting.
Patients were not followed through the emergency department or hospital stays because no infrastructure exists yet to connect data across these settings. This study, therefore, did not aim to study patient outcomes or whether patients received the correct treatment for their complaint. The data collection and entry process may have limitations from entry errors as well as omission and transcription errors. SAMU holds daily morning meetings to discuss patients seen during the last 24 hours which provides an internal check on the information logged into the run sheets, but no formal audit has been conducted of this data set. Finally, the patients with NCDs represented a small subset of the overall population therefore subgroup analyses are limited due to insufficient sample size to make valid statistical comparisons. Nevertheless, it is valuable to understand the care provided by SAMU for NCDs in Kigali using the best available data. Future directions may include creating data infrastructure to study patient outcomes and intervention studies to determine if prehospital interventions may reduce the mortality and morbidity from NCDs in LMIC settings.

Conclusion
This study shows that LMICs such as Rwanda must be more prepared for the rising tide of NCDs. LMIC health systems will need resources to manage these complicated and critically