An observational study of adults seeking emergency care in Cambodia

Abstract Objective To describe the characteristics and chief complaints of adults seeking emergency care at two Cambodian provincial referral hospitals. Methods Adults aged 18 years or older who presented without an appointment at two public referral hospitals were enrolled in an observational study. Clinical and demographic data were collected and factors associated with hospital admission were identified. Patients were followed up 48 hours and 14 days after presentation. Findings In total, 1295 hospital presentations were documented. We were able to follow up 85% (1098) of patients at 48 hours and 77% (993) at 14 days. The patients’ mean age was 42 years and 64% (823) were females. Most arrived by motorbike (722) or taxi or tuk-tuk (312). Most common chief complaints were abdominal pain (36%; 468), respiratory problems (15%; 196) and headache (13%; 174). Of the 1050 patients with recorded vital signs, 280 had abnormal values, excluding temperature, on arrival. Performed diagnostic tests were recorded for 539 patients: 1.2% (15) of patients had electrocardiography and 14% (175) had diagnostic imaging. Subsequently, 783 (60%) patients were admitted and 166 of these underwent surgery. Significant predictors of admission included symptom onset within 3 days before presentation, abnormal vital signs and fever. By 14-day follow-up, 3.9% (39/993) of patients had died and 19% (192/993) remained functionally impaired. Conclusion In emergency admissions in two public hospitals in Cambodia, there is high admission-to-death ratio and limited application of diagnostic techniques. We identified ways to improve procedures, including better documentation of vital signs and increased use of diagnostic techniques.


Introduction
Emergency medicine is a neglected component of health-care systems in most low-and middle-income countries. Nearly half of deaths and one third of disability adjusted life-years lost in these countries could be avoided by applying the basic principles of emergency care. 1 However, despite the renewed interest shown by World Health Organization calls to expand emergency care infrastructure 2 -these systems remain underdeveloped in most low-and middle-income countries and access to high-quality care is limited. 1,3,4 The current health-care system in Cambodia exemplifies this gap in the provision of essential emergency care. Since the Cambodian Ministry of Health was established in 1993, the country has relied on a combination of public and private providers and international nongovernmental organizations to fulfil the medical needs of its people. In 2012, in the public sector, 1080 health centres provided primary care, first aid and maternal and child care (designated the minimum package of activity) in rural areas, while 90 referral hospitals provided secondary and tertiary care -designated the complementary package of activity. Hospitals offering care at complementary package of activity level 2 and above are required to provide emergency care at all times and each has an onsite physician and nurse, with a midwife available for obstetric emergencies. 5 Nevertheless, many of these hospitals lack designated emergency departments, a formal triage process and staff trained in emergency medicine. Furthermore, emergency medicine is not recognized as a specialty by the Ministry of Health and there is no formal programme for training residents in the discipline.
As functional emergency care systems vary between countries and regions, 3 accurate characterization of patients seeking emergency care is essential for developing locally appropriate systems. The 2013 academic emergency medicine consensus conference identified the lack of data on chief complaints as a critical gap in global emergency care research. 6 The chief complaint is the primary consideration used by health-care providers to structure the evaluation and management of patients who present with an acute condition or with an acute presentation of a chronic disease. Moreover, the information obtained from a good understanding of chief complaints differs from that gleaned from discharge diagnoses. 7 In countries with mature emergency care systems, chief complaints are standardized and mapped. 8 However, few developing nations have the means to report or collate information on chief complaints, 6 even though these data are essential for resource allocation, training, research and syndromic surveillance.
The objectives of this study were to identify the characteristics and chief complaints of adults presenting for emergency care in Cambodia. Understanding the pattern of adult emergencies in Cambodia could help guide the development of policy on health-care interventions, such as triage systems and staff training for common conditions. This information could also provide a basis for future studies of the effectiveness and quality of emergency medical services and of patient outcomes.

Methods
We conducted a four-week observational study of unscheduled visits to two provincial government referral hospitals Objective To describe the characteristics and chief complaints of adults seeking emergency care at two Cambodian provincial referral hospitals. Methods Adults aged 18 years or older who presented without an appointment at two public referral hospitals were enrolled in an observational study. Clinical and demographic data were collected and factors associated with hospital admission were identified. Patients were followed up 48 hours and 14 days after presentation. Findings In total, 1295 hospital presentations were documented. We were able to follow up 85% (1098) of patients at 48 hours and 77% (993) at 14 days. The patients' mean age was 42 years and 64% (823) were females. Most arrived by motorbike (722) or taxi or tuk-tuk (312). Most common chief complaints were abdominal pain (36%; 468), respiratory problems (15%; 196) and headache (13%; 174). Of the 1050 patients with recorded vital signs, 280 had abnormal values, excluding temperature, on arrival. Performed diagnostic tests were recorded for 539 patients: 1.2% (15) of patients had electrocardiography and 14% (175) had diagnostic imaging. Subsequently, 783 (60%) patients were admitted and 166 of these underwent surgery. Significant predictors of admission included symptom onset within 3 days before presentation, abnormal vital signs and fever. By 14-day follow-up, 3.9% (39/993) of patients had died and 19% (192/993) remained functionally impaired. Conclusion In emergency admissions in two public hospitals in Cambodia, there is high admission-to-death ratio and limited application of diagnostic techniques. We identified ways to improve procedures, including better documentation of vital signs and increased use of diagnostic techniques.

Research
Emergency care in Cambodia Lily D Yan et al. We enrolled all patients aged 18 years or older who presented at the two study hospitals without an appointment between each Sunday and Friday evening during four consecutive weeks in July and August 2012. Because of study logistics, data collection at weekends differed between the two hospitals. At Sampov Meas Provincial Hospital, patients who presented between 17:00 on Friday and 17:00 on Sunday and who were still in hospital on Monday morning were enrolled; at Battambang Provincial Hospital, no weekend patients were enrolled. A repeat unscheduled visit was considered a separate visit. Verbal consent was obtained from patients or guardians and interviews were conducted in Khmer through native-speaking translators.
Patients presented infrequently during weekends when hospital staffing was limited. Sensitivity analyses showed that including the 63 weekend patients we enrolled did not substantially alter our study results: inclusion resulted in odds ratios (ORs) for predictors varying by around 2%. Thus, weekend patients were included in our multivariate analysis of surgery and functional impairment but not of admissions.
Demographic and clinical data were collected by one research team at each hospital using a standardized, secure, web-based application: Research Electronic Data Capture (REDCap, Vanderbilt University, Nashville, USA), which is designed to support data capture for research studies. 10 Up to three chief complaints were recorded for each patient. The full list of chief complaints is available in Appendix A (available from: http://emed.stanford.edu/education/ international/pubs.html). Information was collected from patients, their friends or family and hospital staff and records, and was entered into REDCap via tablet computers connected to mobile telecommunications networks. Two research staff reviewed each patient record for consistency and inconsistencies were resolved by repeating patient interviews and by reviewing hospital records. Follow-up data on patient outcomes were collected 48 hours and 14 days after the patient's initial visit, either directly from patients remaining in hospital or by telephone. Patients were considered lost to follow-up if they could not be contacted on three successive days.

Data analysis
The study's primary outcomes were admission, medical interventions, surgery, functional impairment and death. In Cambodia, patients may remain in the emergency department for treatment and monitoring for a week or more without being transferred to another department. We recorded only interventions completed within 48 hours of the initial presentation, thereby identifying the most urgent. Patients who stayed in hospital overnight were regarded as having been admitted. Functional impairment was defined as significant pain, significant limitation in performing daily activities, confinement to bed or a comatose state.
Multivariate modelling was performed to identify factors associated with admission, surgery and functional impairment. To avoid including patients who underwent minor procedures and were treated and released, we only used data on admitted patients when identifying factors were associated with surgery. Although we included women who presented with presumed labour in the demographic data, we excluded them from our analysis of outcomes in admitted patients and from multivariate analyses since all women in labour were admitted. As vital signs were commonly not observed or documented, we included records with missing data for the respiratory rate, blood pressure or heart rate in the multivariate analyses by assuming that absent values were normal. Records with missing values for any other independent variable were excluded.
Multivariate logistic regression models, which controlled for age and gender, were built for the primary outcomes using predictors identified by univariate analysis. Stepwise methods were not used. Statistical analysis was performed using SAS Enterprise Guide for Windows, version 4.3 (SAS Institute Inc., Cary, USA). ORs and 95% confidence intervals (CI) were calculated for all model variables.

Results
During the study period, 2162 visits without a prior appointment were documented at the two study hospitals. Of these, 1295 (59.9%) were visits by adult patients aged 18 years or older; these patients comprised the study cohort. We were able to follow up 84.8% (1098) of patients at 48 hours and 76.7% (993) at 14 days. The patients' mean age was 42 years and most patients were female and employed ( Table 1). The predominant mode of transport to hospital was by private vehicle: 55.8% (722) of patients arrived by motorbike and 24.1% (312), by taxi or tuk-tuk, whereas only 9.3% (120) arrived by ambulance. Travel times to the study hospitals were typically short (i.e. less than 2 hours).
Overall, 31.4% (407) of patients presented with an acute complaint (i.e. onset within the past 24 hours), 20.4% (264) with a recent complaint (i.e. symptom onset within 1 to 3 days) and 24.4% (316) with a subacute complaint (i.e. within 4 to 14 days). More than one third were transferred directly to the study hospital from another healthcare facility or referred by a medical provider; only 10.1% (131) had received prior care for their chief complaint and had not been transferred from another facility or referred by a medical provider (Table 1).
Abdominal pain was the chief complaint of 36.1% (468) of patients. Fifteen percent (196) of patients reported respiratory problems and 13.4% (174) reported headache. Only 4.0% (52) of patients presented with vomiting and 1.8% (23) with diarrhoea. Fever was a chief complaint in only 7.5% (97) of cases. Injuries accounted for 11.3% (146) of visits and one death. The proportion of patients admitted was higher for those complaining of abdominal pain, injury, obstetrical problems or vomiting (Fig. 1).
The overall admission rate was 60.5% (783) and the median hospital stay was 3.9 days (Table 2). Of those admitted, 166 patients underwent surgery. Obstetric surgery accounted for 39.1% (65/166) of all surgical procedures ( Table 2). Abdominal pain accounted for 44.8% (351) of all admissions. Even after women with abdominal pain associated with presumed labour were excluded from the analysis, abdominal pain still accounted for 32.3% (253) of all admissions.
Vital signs were either not observed or not recorded for around 20% (245) of patients (Table 1) and 58.4% (756) did not receive any diagnostic tests. The investigations most frequently ordered within 48 hours were laboratory tests and diagnostic imaging (Table 3). Other investigations, such as electrocardiography, were rarely done (1.2% [15]). Computed tomography and magnetic resonance imaging were unavailable at the two hospitals. In the first 48 hours, the intervention most commonly received was medication; (84.1% [1089] of all patients), 40.5% (524) received intravenous fluids (Table 3).
Overall, 14-day mortality in our cohort was 3.9% (39/993). Nearly half of the deaths occurred in the first 48 hours ( Table 2). No patient who was treated and discharged or who underwent surgery died during follow-up. However, 34 patients with conditions that hospital staff deemed untreatable were discharged. By 14 days, six of these patients had died, five were still alive and 23 were lost to follow up. There were 232 pregnancy-related visits and of them 205 pregnant patients were past 20 weeks' gestation. Of the pregnant women past 20 weeks' gestation, 81.9% (168/205) gave birth during the study period, most frequently within 48 hours of presentation (see Appendix B, available from: http://emed.stanford.edu/ education/international/pubs.html). In addition, 28% (65/232) of women with pregnancy-related complaints underwent obstetric surgery. No maternal deaths were noted during follow-up.
Multivariate logistic regression analysis identified several factors as-

Discussion
This study of adults presenting without an appointment at two large public referral hospitals in Cambodia helps to describe the patient population seeking acute care in Cambodia and South-East Asia. Demographic data showed that the patients' mean age was 18 years older than the median age of the Cambodian population. In addition, 12.6% of the study cohort was aged 65 years or more, whereas only 7.7% of the country's population is older than 60 years. 11 The most common chief complaint was abdominal pain, which accounted for over one third of visits and 45% of admissions. Other gastrointestinal complaints, such as vomiting and diarrhoea, were infrequent. The next most common chief complaints were respiratory problems, headache, injury and chest pain, respectively. Few comparative studies of chief complaints at emergency departments in health-care systems in low-and middleincome countries have been reported in the literature. A recent study at a single health centre in Kenya found that road traffic injuries, vaginal bleeding and altered mental status were the most common chief complaints; the most common emergency department diagnosis was injury (20%). 12 In our study, injuries accounted for 11.3% of visits and one death. Fever

Characteristic
No ranked seventh as the primary reason for a hospital visit in our study and was the ninth most common reason in the Kenyan study. 12 Although infectious disease still accounts for a large proportion of the disease burden in low-and middle-income countries, our results are consistent with previous research which suggests that noncommunicable disease is an increasingly common reason for seeking acute care. 5,13,14 Abdominal pain is reportedly the chief complaint of 9% of adults seeking care at emergency departments in the United States of America. 15 According to these nationally-representative statistics, 7% of adults presented with chest pain while headache, shortness of breath and back pain each accounted for 3% of emergency department visits. Our observations in Cambodia show that diagnostic tests are seldom used during the first 48 hours following an emergency admission. Electrocardiography, radiography and ultrasound imaging were available at the two hospitals. This meant that standard evaluations were not done for most chief complaints. The lack of diagnostic evaluations was probably due to both limited resources and inadequate training. We found that over 40% (572) of patients held government-subsidized health equity fund insurance cards. Since the health equity fund awards a fixed amount per patient-visit and many employed people live in poverty, patients may not have been able to pay for even low-cost diagnostic tests.
Poor diagnostic and treatment practices may be due in part to gaps in continuing medical education. Currently, continuing education is not mandatory for health-care providers in Cambodia. Furthermore, most educational programmes approved by the Ministry of Health concentrate on identifying and managing infectious diseases, such as tuberculosis, and on maternal care. Although it is important to address common infectious diseases, our data suggest that the majority of patients present with undifferentiated complaints caused by noncommunicable diseases. Focusing diagnostic resources and provider education on the most common conditions would improve the evaluation and management of these patients.
This study identified several simple, yet important, changes that could potentially improve health-care delivery at Cambodian referral hospitals. First, despite a low number of deaths in the short term, a high percentage of patients were admitted (60.5%). Comparable studies in other low-and middle-income countries, report about 40% admissions. 16 Identifying patients who can be safely discharged would substantially reduce unnecessary admissions and, therefore, health-care costs. Most referral hospitals in Cambodia do not have a central point of entry for patients seeking treatment with undifferentiated complaints. Moreover, neither hospital in our study had a triage system. In the absence of an emergency department and a triage system, inefficiencies, delays in care and inconsistencies in treatment are likely to arise. Consequently, admission criteria at the two study hospitals may have varied according to where in the hospital the patient presented. Relatively simple changes could address these deficits, such as deploying a triage system, designating an area for evaluating new patients and improving staff capacity for managing patients with undifferentiated complaints.
Second, nearly half of the study patients had previously consulted an outside provider for their presenting complaint. There is no coordinated system in Cambodia for referring patients, for sharing medical information between facilities or for transferring patients. 17 Strengthening health management information systems and introducing phone-based referral centres could improve communications between government facilities, save time and help avoid repeating procedures during the evaluation and treatment of patients.
Finally, vital signs were not reliably measured or reported at either study A similar proportion of patients had no records of respiratory rate, blood pressure and heart rate. Again, the absence of a central point of entry and of a triage system for new patients may have contributed to inconsistencies in measurements or documentation. As vital signs are important markers of the severity of illness and have been shown to be reliable predictors of death following trauma and acute poisoning in low-and middle-income countries, these should be measured and documented in every patient. 18,19 The study had several limitations: patients were not enrolled 24 hours a day, seven days a week; reliable discharge diagnoses could not always be obtained; and it was difficult to track discharged patients. Although our findings may not be generalizable to other locations, the two study sites did serve different communities and provinces.