Demographic, Belief, and Situational Factors Influencing the Decision to Utilize Emergency Medical Services Among Chest Pain Patients
Background—Empirical evidence suggests that people value emergency medical services (EMS) but that they may not use the service when experiencing chest pain. This study evaluates this phenomenon and the factors associated with the failure to use EMS during a potential cardiac event.
Methods and Results—Baseline data were gathered from a randomized, controlled community trial (REACT) that was conducted in 20 US communities. A random-digit-dial survey documented bystander intentions to use EMS for cardiac symptoms in each community. An emergency department surveillance system documented the mode of transport among chest pain patients in each community and collected ancillary data, including situational factors surrounding the chest pain event. Logistic regression identified factors associated with failure to use EMS. A total of 962 community members responded to the phone survey, and data were collected on 875 chest pain emergency department arrivals. The mean proportion of community members intending to use EMS during a witnessed cardiac event was 89%; the mean proportion of patients observed using the service was 23%, with significant geographic differences (range, 10% to 48% use). After controlling for covariates, non-EMS users were more likely to try antacids/aspirin and call a doctor and were less likely to subscribe to (or participate in) an EMS prepayment plan.
Conclusions—The results of this study indicate that indecision, self-treatment, physician contact, and financial concerns may undermine a chest pain patient’s intention to use EMS.
Every year, ≈1 250 000 persons in the United States experience an acute myocardial infarction (AMI).1 Of these, >50% die before reaching a medical facility. A majority of these deaths occur within 1 hour of the onset of acute symptoms.1 2 Thrombolytic therapy and other coronary reperfusion strategies are critical in altering the course of an AMI; they can reduce mortality by 25% if initiated within 1 hour of the onset of acute symptoms.3 Unfortunately, only a fraction of patients who are eligible for thrombolytic therapy receive treatment; this is due, in large part, to the time delay between the onset of acute symptoms and arrival at the hospital.4 5 6 7 8 9 10
Little is known about a patient’s decision to use the emergency medical service (EMS) system during a chest pain event. EMS system use can be crucial to receiving prompt therapy for a possible AMI. Benefits include early diagnosis and treatment, emergency department (ED) forewarning of patient arrival, and the ability to address life-threatening complications, such as dysrhythmias, during transport.11 12 However, studies indicate that only 50% to 60% of patients with chest pain use the EMS system.13 14
Factors associated with EMS use among chest pain patients presenting to EDs were previously investigated in 2 concurrent studies in King County, Washington.9 15 The first study focused on the association between EMS use and demographic, situational, and clinical factors; the authors of this study reported that greater education and being physically active at the time of symptom onset were related to decreased EMS system use.9 The second study evaluated knowledge and belief issues surrounding EMS use and found that chest pain patients fail to use EMS because they do not perceive their symptoms as being life-threatening, they did not think of calling 911, or they thought self-transport would be faster.15 An important limitation in the current literature is that all published studies evaluating EMS use among chest pain patients originate from one state with a tax-based, prepaid EMS system.9 13 15 16 17 18 Thus, geographic differences and the impact of cost concerns on EMS use remain uninvestigated.
The objective of the current study was to determine if community members recognize the benefit of the EMS system in a cardiac emergency and to compare these findings to actual EMS usage. This study documented geographic variations in bystander intention to use EMS services among 20 diverse communities in the United States and compared these findings to actual EMS utilization rates among chest pain patients in each community. In addition, survey data provided by chest pain patients presenting to participating EDs were used to determine how demographic factors, situational attributes, and patient perceptions influence the decision to access the EMS system.
The data for this study were drawn from a subgroup of all patients included in the REACT trial.19 REACT was a multicenter, randomized, controlled community trial designed to evaluate the effects of a community intervention on the time interval between onset of AMI symptoms to contact with hospital-based emergency medical care.19 20 In brief, 20 communities were pair-matched by demographic characteristics in 5 regions throughout the United States. One community of each pair was randomly assigned as the intervention site and the other served as a control site. Four months of baseline data were collected in all communities; this was followed by an 18-month, multifaceted education program in the intervention communities. Data used in this study were collected from all 20 communities during the baseline period (December 1995 through March 1996) before the intervention was initiated. In the REACT trial, patient consent requirements were reviewed and approved by all participating hospitals.
For this study, data were provided by 2 sample sources: a random-digit-dial (RDD) community telephone survey and a telephone follow-up survey of chest pain patients presenting to participating EDs and either released or admitted to the hospital with a possible or confirmed coronary event. A review of the medical records for patients participating in the telephone follow-up survey was also conducted.
The RDD community survey was administered among ≈60 adults who were ≥18 years of age in each of the 20 communities. Telephone exchanges and a count of households with listed phone numbers were obtained for specific zip code areas designating the geographic boundaries of each community. Counts of listed households were supplemented with estimates of unlisted households. Disproportionate stratified sampling was used to increase the overall household rate. To adjust for the complex sample design, survey responses were weighted by the reciprocal of the probability of selection. For purposes of this study, only community respondents ≥30 years of age were included in the analysis to facilitate comparison with the follow-up survey.
The telephone follow-up survey included both an ED telephone survey and a hospital inpatient telephone survey. The ED follow-up telephone survey was conducted 7 to 13 weeks after the ED visit for patients presenting to EDs with chest pain but who were subsequently released without a hospital admission. The inpatient follow-up telephone survey, which was conducted 7 to 13 weeks after hospital discharge, was administered to admitted patients with a confirmed International Classification of Diseases21 discharge code of AMI (410) or acute cardiac ischemia (411). Disproportionate stratified random sampling was applied with sampling fractions adjusted for community size and patient response for both the ED survey and inpatient survey. Because patient sampling and survey response rates differed by community, responses were weighted by the number of eligible persons (released from the ED or admitted to the hospital) divided by the number of completed interviews.
The 2 follow-up telephone surveys were appended and merged with hospital medical chart data. This combined database, referred to as the patient follow-up survey, was limited to patients who were ≥30 years of age who presented to the hospital with non-traumatic chest pain.19 Patients were excluded if they were institutionalized or transferred from another hospital.
Additionally, each EMS and fire service agency in each REACT community was queried regarding the availability of a prepayment system. EMS prepayment systems indemnify citizens against the cost of EMS treatment and transport.13 Systems may be tax-based (publicly funded EMS) programs, which do not bill patients for services, or hybrid EMS programs that offer an optional prepayment service that, on the basis of an annual membership fee, indemnifies the patient against any charges not covered by health insurance.
Data contained in the RDD community telephone survey were used to identify community perceptions regarding the value of EMS services during a cardiac event. Specifically, the following question addressed bystander intentions during a coronary emergency: “If you thought someone was having a heart attack, what would you do?” Two optional responses, among many, were the following: (1) call 911 or an ambulance and (2) drive the person to the hospital. By comparing the community telephone survey findings with the EMS utilization data contained in the patient follow-up survey, we could compare community perceptions regarding intended bystander EMS usage with actions taken by community members experiencing a suspected coronary event.
The patient follow-up survey also contained questions assessing demographic, situational, and belief factors associated with the chest pain event that led patients to seek medical attention. Thus, we could also associate EMS use with patient demographics, patient appraisals of their medical condition, actions taken before seeking medical attention, and various beliefs and perceptions that facilitated or hindered quick action when seeking medical care.
Descriptive statistics were used to assess the similarity among the independent samples used in this study. In addition, an exploratory analysis was conducted with patient follow-up survey data to identify demographic, belief, and situational factors associated with the decision to activate (or not activate) the EMS system. Demographic factors and other variables associated with EMS activation in the exploratory analysis were included in a mixed-effects logistic regression model predicting the primary mode of transport (EMS versus other). Design effects associated with the REACT trial were incorporated into the model, in which “study pair” was nested within “geographic region,” and “community” was nested within “pair” and “region” using the glimmix macro for the SAS system.22 Contributions to the model are reported as adjusted odds ratios. All analyses were conducted using SAS, version 6.12.
Survey Response Rates
In the RDD community telephone survey, 36.9% of the randomly generated telephone numbers were for zip code–eligible households (n=2067). In addition, 55 calls to households resulted in no contact after 15 attempts. Among those contacted, 520 resulted in refusals, 62 were ineligible due to a language barrier (non-Spanish or English) or illness, and 136 provided incomplete interviews. The overall interview rate (completed interviews divided by potentially eligible households) was 62.5%. The total sample (≥30 years of age) included 962 respondents.
Response rates for the ED telephone survey and hospital inpatient telephone survey that were appended into the patient follow-up survey are reported separately. For the ED telephone survey, 426 people provided complete interviews out of the 1338 we attempted to contact. Because of a slow study start-up, 18.1% (n=243) of cases were excluded because the 13-week interview window had expired before consent could be obtained. An additional 300 people could not be contacted (eg, non-working phone number). Among those contacted (n=795), 46.4% of people refused the interview or were found to be ineligible during the interview process (ie, too ill, died, deaf, or currently in a nursing home). The overall response rate (number interviewed/[number selected−number ineligible]) was 34.4%.
For the inpatient survey, 449 of 1787 patients provided complete interviews. Among contacted patients (n=1521), 23.3% refused the interview and 47.1% of respondents were found to be ineligible during the interview. The overall response rate was 42.0%. The final sample sizes for the surveys were 962 and 875 for the RDD community survey and the patient follow-up survey, respectively.
Table 1⇓ lists demographic variables for each of the survey samples. The inpatient survey respondents were older and more frequently reported their ethnicity as non-Hispanic white. A greater proportion of ED survey respondents were male. Participants in the RDD community survey reported higher levels of education.
Intention to Use EMS and Actual EMS Use
Table 2⇓ uses data from the RDD community telephone survey and the patient follow-up survey to compare bystander intent to use EMS with self-reported EMS use in each study community. On average, 89.4% of respondents in each study community indicated that they would call 911 if they witnessed a cardiac event. Very few (8.1%) would consider driving someone with a coronary emergency to the hospital.
The patient follow-up survey provided EMS use information for chest pain patients presenting to participating EDs in each study community. Contrary to the bystander intentions expressed in the community survey, few actual chest pain victims used EMS (23.2%). Most victims were driven to the ED by someone else (60.4%) or drove themselves to the hospital (15.6%).
Factors Associated With Actual EMS Use
Using the patient follow-up survey data, demographic, situational, and belief factors were compared among EMS and non-EMS users. Several demographic variables were significantly associated with EMS use, including increasing age, white ethnicity, living alone, and presence of an ambulance service prepayment plan (Table 3⇓).
When considering actions taken by patients before calling 911 or going to the hospital, patients taking an antacid or aspirin were less likely to use EMS services. However, patients taking nitroglycerin were twice as likely to choose EMS transport. Regarding communications with others, requesting advice from family or friends before seeking medical attention was not associated with EMS use. However, patients communicating with a physician were less likely to use EMS transport to the hospital.
The following question was significantly associated with EMS use (Table 3⇑): “Did any factors or things cause you to go quickly (or wait to go) to the hospital?” Post hoc analyses of answer subcategories indicated that certainty that a patient’s symptoms were caused by a “heart attack” was associated with an increased likelihood of choosing EMS transport, whereas patients who thought their symptoms would go away were significantly less likely to use EMS. Pain severity was not associated with EMS use.
Using a multivariable logistic regression model, we examined the associations of the following factors with EMS use: sex, ethnicity (white versus non-white), living alone, taking nitroglycerin, communicating with a physician, and being prompted to “go quickly” or “waiting” to go to the hospital. The variable identifying the presence of an EMS prepayment system was trichotomized to independently assess the effect of subscription services verses tax-based programs. The variables “took antacid” and “took aspirin” were combined to address the issue of a patient’s self-medicating during a potential cardiac event. Age was excluded from the model because of its strong association with 2 other variables, “living alone” and “taking nitroglycerin.” Separate models were analyzed using weighted and unweighted survey responses. Regression coefficients between the models were similar; thus, we report only the unweighted results.
The overall fit of the logistic model was good; it correctly classified 76% of all cases (Table 4⇓). The variables “living alone,” “taking nitroglycerin,” and being prompted to “go quickly” to the hospital were strong predictors of EMS use. The presence of a tax-based, prepaid EMS system doubled the likelihood of using EMS compared with communities with no such system. Because the presence of an EMS prepayment plan was measured on the community level rather than on an individual level, including random effects associated with community appropriately inflated the confidence band associated with this variable. Thus, the 95% confidence interval associated with the prepayment variable included unity, so that statistical significance could not be attributed to a prepayment effect. This variable should be interpreted with some care. Being prompted to “wait before going,” taking an antacid/aspirin, or consulting with a physician significantly decreased the likelihood that respondents would use EMS services.
Findings indicate that, in general, community members recognize the benefit of EMS transport when acting as a bystander to a “public” cardiac event but individuals personally experiencing symptoms of an AMI often choose not to use EMS services. One should note, however, that bystander intentions may favor an EMS response simply because respondants assumed they were unacquainted with the victim and his/her extenuating circumstances. Bystander decisions can be decisive if personal circumstances do not complicate bystander decision-making. Alternatively, actual patients may not have considered their symptoms to be indicative of a heart attack and were, therefore, less inclined use EMS. It is unclear if similar findings would be present if intentions and actual events were documented for the same subject. Nevertheless, the magnitude of difference between bystander intentions and actions for self and the uniformity of this finding across geographic regions suggest that further investigation may prove useful in determining why the public would choose alternative transportation when faced with a cardiac emergency.
Situational factors that decreased EMS use during a cardiac event included taking an antacid/aspirin or communicating with a doctor before going to the hospital. However, patients taking nitroglycerin and patients believing their condition was heart-related were more likely to use EMS. These findings suggest that patients with familiar symptoms or experience with a heart condition are more likely to rely on EMS care as a valued form of medical care and transport. Additional published work has associated symptom familiarity with increased EMS use.15
The fact that communication with a doctor decreased EMS use is problematic. It is unclear if doctors were acting as managed care “gatekeepers” to EMS care or if they reduced patient anxiety in a way that made EMS transport seem optional. There may be a variety of valid reasons why physicians who are familiar with individual patient histories may not dictate EMS use during phone contact with a concerned patient. However, our data indicate that 83% of patients who spoke with a physician and did not use EMS transport were subsequently admitted to the hospital.
Regarding belief factors, no correlation existed between seeking advice from peers or pain severity and EMS transport, which is contrary to other studies demonstrating a positive correlation between these factors and EMS use.6 9 15 The perception among patients that their symptoms would go away decreased EMS use; this result is similar to findings reported elsewhere.15
Several demographic variables were associated with EMS use. Living alone and increasing age (although unadjusted) enhanced EMS use. These results may reflect the fact that the elderly and those in single-person households have fewer transportation options. Other demographic variables, including ethnicity, sex, and education, were not related to EMS use, which contrasts with the results of previous studies.6 8 9 However, one should note that previous research addressing this question originated in one state with a relatively high EMS use rate.9 13 15 16 17 18 Thus, contradictions between previous findings and current results may represent geographic differences in patient population, EMS structure, etc.
Of interest is the fact that the presence of an EMS prepayment system increased EMS use. One other study documented a similar increase among residents of lower income census blocks.13
There are several important limitations to this study. A potential source of bias relates to the fact that ED and inpatient survey data were obtained retrospectively, 7 to 13 weeks after the cardiac event. The event or the extended period of time between the event and our interviews may have affected patient responses. At least one other study, however, has shown that acute health conditions requiring medical attention often represent “sentinel events” and may be accurately recalled for up to 6 months.23 A second limitation involved the low response rate to the ED and inpatient surveys (<42%). Missing interviews may systematically favor an income group, degree of chronic illness, or some other unmeasured variable that limits the generalizability of our findings. The fact that our study sample included communities with diverse mean incomes and ethnic distributions may temper some potential bias due to sample selection.19
In summary, people seem to understand the prudent actions to take when faced with a public cardiac event, but they may be unwilling to take the appropriate steps when facing a personal cardiac emergency, perhaps due to symptom uncertainty or other behavioral factors. Variables representing demographic, situational, and self-efficacy (or belief) factors can inhibit or promote EMS use during a cardiac event. Subscription services and taxed-based systems that offset the cost of EMS services need to be analyzed further to determine if these programs represent a major factor among patients evaluating options for emergency transportation.
Mr Brown was a summer research student in the Department of Emergency Medicine at Oregon Health Sciences University during the time this research was conducted. The majority of Dr Mann’s efforts on this project occurred during his time as part of the faculty of the Department of Emergency Medicine at Oregon Health Sciences University. The REACT trial was supported by cooperative agreements U01-HL-53141, U01-HL-53412, U01-HL-53149, U01-HL-53155, U01-HL-53211, and U01-HL-53135 from the National Heart, Lung, and Blood Institute, Bethesda, Md. In addition, an American Heart Association Summer Student Award was made to Mr Brown. The authors are solely responsible for the content of the article, and their opinions do not necessarily represent the views of any listed funding source.
- Received November 22, 1999.
- Revision received January 28, 2000.
- Accepted February 8, 2000.
- Copyright © 2000 by American Heart Association
National Heart, Lung, and Blood Institute. Morbidity and Mortality: Chart Book on Cardiovascular, Lung, and Blood Diseases. Bethesda, Md: US Department of Health and Human Services, Public Health Service; 1992.
American Heart Association. Heart Facts. Dallas, Tex: American Heart Association; 1992.
Dracup K, Moser DK, Eisenberg M, et al. Causes of delay in seeking treatment for heart attack symptoms. Soc Sci Med. 1995;40:379–392.
Dean NC, Haug PJ, Hawker PJ. Effect of mobile paramedic units on outcome in patients with myocardial infarction. Ann Emerg Med. 1988;7:1034–1041.
Siepmann DB, Mann CN, Hedges JR, et al. Association between pre-payment systems and EMS use among patients with acute chest discomfort syndrome. Ann Emerg Med. In press.
International Classification of Diseases, 9th revision. Washington, DC: Public Health Service, US Dept of Health & Human Services; 1980.
Littell RC, Milliken GA, Stroup WW, et al. SAS System for Mixed Models. Cary, NC: SAS Institute Inc; 1996.