The Epidemiology and Outcome of Prehospital Respiratory Distress

Patients with respiratory distress often seek emergency medical care and are transported by emergency medical services (EMS). EMS encounters with patients in respiratory distress have not been well described. The study objective was to characterize the epidemiology of prehospital respiratory distress and subsequent patient outcomes. This was a population-based cohort study of noninjured adults transported by EMS to any of 16 hospitals between 2002 and 2006. EMS records were linked to hospital administrative data for encounters categorized by EMS personnel as primarily “respiratory distress” in nature. The authors described prehospital patient and encounter characteristics, interventions, hospital discharge diagnoses (using ICD-9-CM codes), and patient outcomes. The association between prehospital variables, defined a priori, and hospital admission were described using multivariable logistic regression. There were 166,908 EMS encounters, of which 19,858 were for respiratory distress (11.9%, 95% confidence interval [CI] = 11.7% to 12.1%). Half of the patients were admitted to the hospital (n = 9,964), one-third of those required intensive care (n = 3,094), and 10% of hospitalized patients died prior to discharge (n = 948). Fifteen percent of hospitalized patients received invasive mechanical ventilation (n = 1,501), over half of whom were intubated during prehospital care (n = 896). The most common primary discharge diagnoses among prehospital respiratory distress patients admitted to the hospital were congestive heart failure (CHF; 16%), pneumonia (15%), chronic obstructive pulmonary disease (COPD; 13%), and acute respiratory failure (13%). Few EMS patients with respiratory distress were coded with a primary diagnosis of acute myocardial infarction (3.5%, n = 350) or underwent percutaneous coronary intervention (0.7%, n = 71). In a multivariable regression model, prehospital factors that were independently associated with hospital admission included initial respiratory rate (odds ratio [OR] = 1.29 for an increase in respiratory rate of five breaths/min, 95% CI = 1.24 to 1.35) and an encounter that originated at a nursing home (OR = 2.80, 95% CI = 2.28 to 3.43). In a population-based cohort, EMS personnel commonly encounter prehospital respiratory distress among medical patients, many of whom require hospital admission to the intensive care unit. These data may help to inform targeted therapy or more efficient triage and transport decisions. Los pacientes con dificultad respiratoria a menudo demandan atención médica urgente, y son transportados por los Servicios de Emergencias Médicas (SEM). Las atenciones del SEM con los pacientes con dificualtad respiratoria no han sido bien descritas. El objetivo del estudio fue caracterizar la epidemiología de la dificultad respiratoria extrahospitalaria y los resultados subsiguientes de dichos pacientes. Estudio de cohorte de base poblacional de adultos sin lesiones trasladados por SEM a uno de los 16 hospitales entre 2002 y 2006. Las historias clínicas del SEM se vincularon con los datos administrativos hospitalarios para las atenciones clasificadas por el personal de SEM con el diagnóstico principal de “dificultad respiratoria”. Los autores describieron las caracteristicas del paciente y de la atención extrahospitalaria, las intervenciones, los diagnosticos al alta hospitalarios (mediante códigos CIE-9), y los resultados del paciente. La asociación entre las variables extrahospitalarias, definidas a priori, y el ingreso hospitalario se describió mediante un modelo de regresión logística multivariable. Hubo 19.858 atenciones del SEM, de las cuales 166.908 fueron por dificultad respiratoria (11,9%, intervalo de confianza [IC] = 11,7% a 12,1%). La mitad de los pacientes fueron ingresados en el hospital (n = 9.964), de los cuales un tercio requirió cuidados intensivos (n = 3.094), y un 10% fallecieron previamente al alta (n = 948). Un 15% de los pacientes hospitalizados recibieron ventilación mecánica invasiva (n = 1.501), de los cuales más de la mitad fueron intubados durante la atención extrahospitalaria (n = 896). Los diagnósticos principales al alta más frecuentes entre los pacientes con dificultad respiratoria extrahospitalaria ingresados en el hospital fueron la insuficiencia cardiaca congestiva (16%), la neumonía (15%), la enfermedad pulmonar obstructiva cronica (13%) y la insuficiencia respiratoria aguda (13%). Pocos pacientes del SEM con dificultad respiratoria fueron catalogados con un diagnóstico principal de infarto agudo de miocardio (3,5%, n = 350), o sometidos a una intervención coronaria percutánea (0,7%, n = 71). En un modelo de regresión multivariable, las factores extrahospitalarios que se asociaron independientemente con el ingreso hospitalario incluyeron la frecuencia respiratoria inicial (odds ratio [OR] 1,29 por cada incremento de 5 respiraciones por minuto en la frecuencia respiratoria, IC 95% = 1,24 a 1,35), y la atención en una residencia (OR 2,80, IC 95% = 2,28 a 3.43). En una cohorte de base poblacional, el personal de SEM halla frecuentemente dificultad respiratoria en los pacientes médicos, muchos de los cuales requieren ingreso hospitalario en la unidad de cuidados intensivos. Estos datos pueden ayudar a asesorar la terapia dirigida o las decisiones de triaje y traslado más eficientes. Respiratory distress is a common reason to engage the emergency health care system in the United States, accounting for more than 3.7 million emergency department (ED) visits annually.1 Respiratory distress may result from a variety of problems including chronic obstructive pulmonary disease (COPD), asthma, pneumonia, or congestive heart failure (CHF) and can be generally defined as respiratory demand being greater than capacity.2 Among the elderly, respiratory distress is the second most common reason to visit an ED. When hospitalized, patients with respiratory distress severe enough to result in respiratory failure are almost four times less likely to survive to hospital discharge than patients with cancer or heart disease.3 Emergency medical services (EMS) personnel play a prominent role in triage, transport decisions, and initial management of patients with respiratory distress. Evidence suggests that interventions in the prehospital setting reduce mortality among patients with respiratory distress, particularly when care is provided by advanced rather than basic life support (ALS vs. BLS).4 Yet, paramedics are hampered in their delivery of early, targeted therapy by the challenges in accurately differentiating between the various etiologies of respiratory distress.5, 6 Certain therapies, if misapplied, may be harmful to patients with undifferentiated respiratory distress.7-9 A notable example is prehospital unopposed oxygen compared to titrated oxygen in patients with presumed COPD exacerbation.10 Little is known about the frequency with which EMS personnel encounter prehospital respiratory distress, its etiologies, or outcomes. We sought to characterize the epidemiology and outcomes of prehospital respiratory distress, to explore the associations between patient and incident characteristics and hospital admission, and to describe the concordance of EMS clinical impressions with hospital diagnoses. These data provide a framework on which to consider how early diagnosis and therapy may be offered to high-risk patients with prehospital respiratory distress. This was a community-based, retrospective cohort study that included all EMS patient encounters between 2002 and 2006 in greater King County, Washington (excluding the city of Seattle). This data set is part of a larger cohort of King County EMS records used for previous prehospital studies.11-13 The institutional review boards of the Washington State Department of Health and the University of Washington approved the study with waiver of informed consent. We analyzed patients transported by a countywide EMS system between January 2002 and December 2006. The EMS catchment area includes urban, suburban, and rural areas over approximately 2,000 square miles containing a population of 1.2 million people. This EMS system is structured to respond in a tiered fashion to medical emergencies. The first-tier response is emergency medical technician (EMT)-firefighters who evaluate all patients and provide BLS. BLS providers operate 29 fire departments in the county, receive common initial and ongoing training, and follow standardized patient care protocols. The second tier of the EMS system includes paramedics trained in ALS who work in pairs to evaluate and manage higher-acuity patients. ALS providers are trained in interventions such as intravenous (IV) access, medication administration, and advanced airway management. King County has 25 ALS ambulances on duty at any one time. All paramedics receive common training and provide care under a common set of care guidelines. Dispatchers for EMS activate the ALS and BLS response concurrently if standardized criteria are fulfilled, based on the patient's presenting complaint.14 For patients with breathing difficulty, dispatch criteria for an ALS response include at least one of the following: 1) unconscious and unresponsive; 2) is in respiratory distress, defined as either the inability to speak normally (increased work of breathing) or sitting, standing, or leaning forward to breathe (tripod position); 3) breathing difficulty accompanied by chest discomfort if the patient is older than 25 years old; or 4) has audible wheezing or stridor. Patients are transported to 16 acute care hospitals within the county. BLS transport is the default; ALS providers will transport a patient with isolated breathing difficulty if any of the following criteria are met or if in their judgment the patient may imminently require ALS (in consultation with medical control): 1) respiratory rate > 30 breaths/min, 2) abnormal respiratory pattern, 3) patient in tripod position, 4) failure to respond to inhaled bronchodilators, or 5) acute asthma with history of previous intubation. We used probabilistic matching to remove duplicates among EMS records and used hierarchical deterministic matching to link EMS records to 1) the Comprehensive Hospital Abstract Reporting System (CHARS), a statewide database of detailed diagnostic, procedural, and discharge data on all hospitalizations and 2) the Washington State death record. Additional details on cohort development and the success of data linkage were reported previously.11 We excluded patients with traumatic injury or cardiac arrest and those younger than 18 years of age. Among the remaining EMS encounters, we included all patients transported to a hospital who had physical examinations, including vital signs, documented by EMS personnel. We defined patients as having respiratory distress based on the initial prehospital provider documenting “respiratory” as the primary reason for prehospital care. EMS personnel also documented a diagnostic impression for each patient with respiratory distress: “shortness of breath/hyperventilation,” “asthma,” “chronic obstructive pulmonary disease (COPD)/emphysema,” “respiratory depression/respiratory arrest,” or “other respiratory.” BLS and ALS providers may separately document diagnostic impressions on the same patient if they are both dispatched (i.e., a two-tier EMS response). We linked these data and used the BLS diagnostic impression as the default, except as noted below. For EMS patients admitted to the hospital, we describe inpatient diagnoses and procedures using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), codes. We defined patient outcomes as intensive care unit admission (identified using critical care revenue codes15), length of stay, hospital mortality, and discharge destination (e.g., new nursing home placement). We abstracted prehospital clinical data from EMS records, including triage, demographic, procedural, and transport data. Prehospital vital signs were captured from the first provider on scene and included respiratory rate, pulse rate, blood pressure, oxygen saturation, and Glasgow Coma Scale (GCS) score. EMS personnel on scene provided a severity of illness assessment by classifying each encounter as “nonurgent,” “urgent,” or “life-threatening.” We abstracted prehospital interventions including supplemental oxygen, electrocardiographic monitoring, IV access, bag-valve-mask ventilation, and endotracheal intubation. Mechanical, noninvasive positive pressure ventilation was not used prior to hospital arrival. We calculated the overall proportion of nontrauma, noncardiac arrest EMS activations due to respiratory distress with 95% confidence intervals (CIs) and the proportion with respiratory distress by age groups. We summarized EMS encounter characteristics for those with respiratory distress, and for the subset of patients admitted to a hospital, we identified inpatient procedures and patient outcomes. To explore associations between prehospital characteristics and hospital admission, we constructed a multivariable logistic regression model using robust variance estimation. In the model, we included a priori patient variables (e.g., age, sex, systolic blood pressure [sBP], pulse rate, respiratory rate, and pulse oximetry), incident data (e.g., home, public building, street/highway, adult family home, nursing home, or medical facility), and overall assessment by EMS personnel (severity index and diagnostic impression). We used single imputation with normal values substitution for the GCS score and pulse oximetry because multiple imputation in prior analyses of this data set did not substantially change results.11, 13 We included three additional variables as fixed effects in the final logistic model: year in the cohort, the responding EMS agency, and the receiving hospital. We included the latter two variables to mitigate the nonindependence of observations in these clusters. We decided a priori that provider level of training (i.e., ALS vs. BLS) may be a clinically relevant effect modifier of the association between severity of illness and hospital admission; we included this interaction term in the model and evaluated its significance using a likelihood ratio test. We used STATA 12.0 for data analysis, and two-sided p-values of <0.05 were considered significant. To investigate the accuracy with which prehospital personnel are able to identify “phenotypes” of respiratory distress, we compared primary hospital diagnoses at the time of discharge (using ICD-9-CM codes) to the EMS diagnostic impression. We further classified EMS diagnostic impression into symptom-based (e.g., shortness of breath) or disease-based (e.g., asthma, COPD, respiratory depression). Diagnostic concordance was analyzed separately for patients transported by ALS and BLS. In cases where both ALS and BLS evaluated the patient and documented diagnostic impressions, we used the diagnostic impression of the transporting provider. Approximately one in eight EMS nontraumatic, nonarrest encounters was for a patient classified by EMS as respiratory distress (19,858 of 166,908 [11.9%], 95% CI = 11.7% to 12.1%; Figure 1). This estimate ranged between 10.9 and 12.4% annually over the 5-year study. Approximately one in six patients evaluated by ALS providers had respiratory distress (17.9%, 95% CI = 17.6% to 18.3%), compared to one in 12 for those evaluated by BLS providers only (8.6%, 95% CI = 8.4% to 8.7%). The proportion of EMS encounters for respiratory distress increased across patient age groups (Figure 2). BLS providers responded to all patients with respiratory distress (Figure 3). ALS providers were dispatched to approximately one-half of the patients with respiratory distress (54%, n = 10,666). ALS providers transported more than one-half of these patients (55%, n = 5,815). We observed that most EMS responses for respiratory distress were to the patient's home (n = 13,807, 70%; Table 1). A higher proportion of patients seen jointly by ALS and BLS received supplemental oxygen compared to BLS-only encounters (95% vs. 56%, respectively). ALS providers determined illness severity as life-threatening in 13% of the patients they evaluated (n = 1,353) and intubated two-thirds prior to hospital arrival (n = 896). One-half of EMS patients with respiratory distress were admitted to the hospital (n = 9,964), and one-third of those received intensive care (n = 3,094) for a median of 2 days (interquartile range [IQR] = 1 to 4 days; Table 2). Mechanical ventilation was delivered in one-half of ICU patients (48%; n = 1,486), more than half of whom were intubated in the prehospital setting. Hospital mortality was 10% for those who were admitted, 13% for those requiring intensive care, and 23% for those requiring mechanical ventilation. Among patients who survived hospitalization, one in eight required new placement in a skilled nursing facility (13%; n = 1,163). We observed many characteristics independently associated with hospital admission, including increasing age, male sex, and abnormal prehospital physiology (e.g., greater respiratory and heart rates, lower sBP, and oxygen saturation; Table 3). EMS encounters that originated at the patient's home or a nursing home were associated with more than twofold higher odds of hospital admission than encounters elsewhere. We observed an interaction between prehospital severity of illness and the level of prehospital care (BLS only vs. ALS; p < 0.01), whereby a life-threatening severity of illness assigned by ALS was associated with the greatest adjusted odds of hospital admission compared to a nonurgent patient cared for by BLS only (Table 3). Of the 9,443 prehospital respiratory distress patients admitted after EMS transport, 6,564 (70%) received primary discharge diagnoses related to respiratory disease (i.e., pulmonary edema, COPD, asthma, pneumonia, pulmonary embolism, lung cancer, or acute respiratory failure). A higher proportion of respiratory distress identified during ALS transport (74%) compared to BLS transport (66%) included primary diagnoses related to respiratory disease (Tables 4A and 4B). EMS personnel documented a symptom-based impression in two-thirds of respiratory distress encounters. Fewer patients received disease-based impressions, including 22% with asthma and 20% with COPD. We observed fair concordance between EMS impressions and hospital diagnoses of asthma and COPD (Tables 4A and 4B). Few patients in whom ALS providers suspected asthma or COPD were diagnosed with CHF (5%) or acute myocardial infarction (2%). In a large, community-based cohort over 5 years, we found that respiratory distress was responsible for one in eight nontrauma, nonarrest EMS encounters. We observed that EMS patients with respiratory distress are commonly admitted and have a hospital mortality (10%) comparable to those hospitalized after blunt trauma, chest pain, or suspected stroke.16 The and mortality of prehospital respiratory distress in the study support as a 18 prior work found a mortality rate in prehospital respiratory distress in which was by the of ALS to BLS In study of a two-tier EMS almost half of prehospital respiratory distress encounters received BLS care and were not transported by This suggests a robust to explore how triage within the EMS including those with (e.g., the outcomes of patients with prehospital respiratory distress. may also inform in which EMS management of prehospital respiratory distress be into of critical care. in of the United the triage and management of patients with critical including acute myocardial and from the time of prehospital hospital For patients with respiratory distress, data as the for For patients with prehospital respiratory using this framework identify early of mechanical ventilation, for which the to transport to a facility to provide more ICU care. The accuracy of EMS personnel in of of respiratory distress has we are to consider for prehospital respiratory distress, paramedics identify a respiratory distress and targeted We observed that EMS personnel identified (i.e., COPD and with fair used symptom-based impressions (e.g., shortness of breath) more This on symptom-based impression in the of the respiratory diagnostic in the prehospital clinical record. on EMS diagnostic accuracy consider and diagnoses to EMS personnel that were not in this study (e.g., pneumonia, pulmonary the time or advanced of any assessment is in prehospital may from and under study in the and for prehospital We classified patients as having respiratory distress based on the initial EMS assessment and excluded those patients physical are as patients may have been were classified with primary We not have data on prehospital medication by EMS personnel and are to evaluate the association of interventions and patient outcomes. The concordance of EMS diagnostic impression with final hospital diagnoses may be if EMS personnel were by the by hospital We also a in one by a single EMS The of may be in rural of the United States, and where prehospital procedures are used (e.g., noninvasive positive pressure may have been in the of prehospital or in the of in the population the study which may further Prehospital respiratory distress is often by EMS personnel and is associated with and patient and prehospital characteristics may identify patients with respiratory distress at a greater likelihood of requiring hospital into the of prehospital respiratory distress may inform targeted therapy or more efficient triage and transport decisions.