Persistence of delays in presentation and treatment for patients with acute myocardial infarction: The GUSTO-I and GUSTO-III experience☆☆☆

Introduction 

A delay in receiving thrombolytic therapy for acute myocardial infarction (AMI) is associated with diminished treatment effect and increased mortality.1 Although treatment administered up to 12 hours after symptom onset has proved beneficial, maximum benefit is achieved when therapy is provided within the first few hours.2 Over the past decade, substantial public health efforts have successfully educated health care providers about the importance of timely treatment with thrombolytic therapy.3 As a result, the time required to identify and treat patients with myocardial infarction in the emergency setting has been nearly halved.4

Despite such success in decreasing time to treatment in the emergency department, no similar reduction in time from symptom onset to patient arrival at the hospital has been observed. In a study of all 41,021 patients enrolled in the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO-I) trial, certain patient groups arrived to the hospital late in the course of their AMI.5 Female, elderly, diabetic, and hypertensive patients were found to be susceptible to delays in hospital arrival and treatment in the ED.

We hypothesized that improvement in the inhospital time to treatment for US patients with AMI would be associated with a similar decrease in the delay from symptom onset to patient arrival to the hospital. Given the publication of national guidelines on time to thrombolytic treatment and rigorous public education initiatives since 1990, the Global Use of Strategies to Open Occluded Coronary Arteries (GUSTO-III) trial, completed in February 1997, afforded us a unique opportunity to examine this hypothesis. The parallel designs of these 2 studies also allowed us to study 2 large and very similar patient groups separated by several years in which awareness and treatment of myocardial infarction were rapidly evolving. We sought to isolate the factors associated with both out-of-hospital and inhospital delays. Furthermore, through socioeconomic information obtained in the GUSTO-III trial, we examined the impact of educational level, occupation, health care insurance status, and living alone on time delays.

Materials and methods 

Our analysis population consisted of the patients enrolled in the United States in the GUSTO-I and GUSTO-III trials. This study was approved by the institutional review boards of all participating clinical sites. GUSTO-I enrolled 41,021 patients in a multicenter, international trial evaluating 4 fibrinolytic regimens of streptokinase and tissue plasminogen activator. Complete methods for the GUSTO-I trial have been published.6 In this trial, patients were randomized to 1 of 4 fibrinolytic regimens if they presented within 6 hours after symptom onset and had chest pain lasting 20 minutes or more. In addition, patients must have had no contraindications to thrombolysis and must have met electrocardiographic criteria for infarction (ST-segment elevation ≥0.1 mV in ≥2 limb leads or ≥0.2 mV in at least 2 contiguous precordial leads). The GUSTO-III trial enrolled 15,060 patients; complete methods have been published.7 Inclusion criteria for patients in GUSTO-III were identical to GUSTO-I. This randomized, controlled clinical trial compared 2 different fibrinolytic regimens: alteplase versus reteplase. Electrocardiographic criteria for determining infarction were similar to those in GUSTO-I. Because practice patterns vary substantially between countries, our evaluation was limited to the 23,105 GUSTO-I patients and 4,744 GUSTO-III patients enrolled in the United States.

In GUSTO-I and GUSTO-III, 3 time variables were defined prospectively: (1) time from symptom onset to hospital arrival (out-of-hospital delay), (2) time from hospital arrival to administration of thrombolytic therapy (inhospital delay), and (3) total time from symptom onset to treatment. The temporal frames of 0 to 2 hours, 2 to 4 hours, and more than 4 hours were based on a prespecified analysis in GUSTO-I for time to hospital arrival from symptom onset and total time to treatment. For GUSTO-III, time from arrival to treatment was prospectively divided into 2 groups reflecting early treatment (<1 hour) and late treatment (>1 hour).

For these analyses, the time of symptom onset was provided by the patient, or a family member if the patient was unable to communicate. The time of patient arrival to hospital was derived from hospital records, as was time of administration of fibrinolytic therapy.

For the GUSTO-I and GUSTO-III trials, information on baseline characteristics including age, sex, race, height, and weight were prospectively collected. We prospectively defined the following baseline characteristics for our analysis of their association with time to hospital arrival: history of hypertension, diabetes, smoking, hypercholesterolemia, family history, history of angina, myocardial infarction, coronary artery bypass grafting, and angioplasty. Physical examination data, including pulse rate, systolic and diastolic blood pressure, presence of congestive heart failure, and anatomic location of the infarction were also gathered.

In GUSTO-III only, socioeconomic data were prospectively obtained to identify the impact of education level, occupation, payer status, and living alone on time to hospital arrival and time to treatment.

Descriptive statistics, including percentages for discrete variables and medians with 25th and 75th percentiles for continuous variables, were used for baseline characteristic determination and time to treatment data for GUSTO-I and GUSTO-III patients. Statistical testing was performed using the χ2 test for categoric variables and Cox modeling for continuous variables. A P value less than .05 was considered significant.

Because enrollment criteria and baseline data in GUSTO-I and GUSTO-III were identical, these databases were combined to maximize power for the analysis. No distributional assumptions were made for the dependent variables, time to hospital arrival, and time from hospital arrival to treatment.

On the basis of the GUSTO-I analysis that identified the principal features associated with delay in presentation, we performed a univariate analysis that identified age, sex, race, and diabetes as independent predictors of delay in hospital arrival for patients. Similarly, the patient's age, sex, race, pulse rate, previous angina, previous infarction, previous bypass surgery, and infarct location were found to be predictive of delay in time to treatment after hospital arrival. We then used stepwise regression modeling to evaluate the influence of socioeconomic factors on delay obtained in GUSTO-III. The socioeconomic factors that were evaluated included years of education, occupation, type of insurance, and living alone.

Results 

The baseline characteristics of patients enrolled in GUSTO-III were similar to those of patients enrolled in GUSTO-I. Thirty-day mortality was also similar for the 2 groups (Table 1).

Table 1. GUSTO-I and GUSTO-III baseline characteristics and mortality.*

	Characteristic	GUSTO-I (n=23,105)	GUSTO-III (n=4,744)
	Age, y	61.1 (51.5, 69.8)	61.5 (51.5, 71.2)
	Female sex	6,278 (27.2)	1,346 (29.7)
	Weight, kg	80.2 (70.1, 91)	81.8 (71, 93)
	Height, cm	172.7 (165.1, 180)	172 (165, 179)
	Race
	Black	1,113 (4.8)	222 (4.9)
	Other	962 (4.2)	199 (4.4)
	White	20,962 (91.0)	4,108 (90.7)
	Hypertension	9,821 (42.7)	1,994 (44.0)
	Diabetes mellitus	3,820 (16.6)	819 (18.1)
	Hypercholesterolemia	8,473 (37.9)	1,873 (41.5)
	Previous myocardial infarction	3,943 (17.1)	809 (17.9)
	Systolic blood pressure, mm Hg	126 (110, 142)	133 (116, 150)
	Pulse rate, beats/min	74 (63, 86)	74 (62, 87)
	Killip class
	I	19,950 (87.1)	4,092 (91.4)
	II	2,507 (11.0)	334 (7.5)
	III	250 (1.1)	32 (0.71)
	IV	187 (0.82)	21 (0.47)
	Anterior infarction	8,591 (37.4)	1,994 (44.0)
	Inferior infarction	13,694 (59.6)	2,385 (52.6)
	Inhospital mortality	1,495 (6.5)	246 (5.4)
	Mortality by 30 d	1,562 (6.8)	286 (6.3)
	*Results are expressed as numbers (percentages) or median (25th, 75th percentiles).

The time from symptom onset to hospital arrival did not change during the 7-year period from the start of GUSTO-I in 1990 to the end of GUSTO-III in 1997 (Table 2).

Table 2. Comparison of the GUSTO-I trial to the GUSTO-III trial: Time from symptom onset to hospital arrival, hospital arrival to treatment, and time from symptom onset to treatment.*

	Interval Times	GUSTO-I (n=23,105)	GUSTO-III (n=4,744)	P Value
	Enrollment years	1990–1993	1995–1997	—
	Symptom onset to hospital arrival, h	1.4 (0.9, 2.3)	1.4 (0.8, 2.3)	.09
	Hospital arrival to treatment, h	1.1 (0.8, 1.5)	0.8 (0.6, 1.2)	<.0001
	Total time to treatment, h	2.7 (1.9, 3.8)	2.3 (1.6, 3.3)	<.0001
	*Data are median (25th, 75th percentiles).

View Large Image Download to PowerPoint

Fig. 1. Cumulative distribution plots of time from symptom onset to hospital arrival for the GUSTO-I and GUSTO-III trials.

View Large Image Download to PowerPoint

Fig. 2. Cumulative distribution plots of time from hospital arrival to treatment for the GUSTO-I and GUSTO-III trials.

View Large Image Download to PowerPoint

Fig. 3. Cumulative distribution plots of time from symptom onset to treatment for the GUSTO-I and GUSTO-III trials.

The GUSTO-III data revealed a patient profile similar to GUSTO-I. Increased age, female sex, non-white race, hypertension, and diabetes were all associated with a longer time from onset of symptoms to treatment. Conversely, patients who smoked or had an elevated cholesterol level experienced a shorter total time from onset of symptoms to fibrinolytic therapy. Thirty-day mortality rates in GUSTO-III are comparable with those in GUSTO-I, suggesting that increased time to treatment, for any reason, resulted in increased mortality (Table 3).

Table 3. The GUSTO-III trial: Characteristics associated with delay from symptom onset to hospital arrival with associated inhospital, 30-day mortality (early, 0 to 2 hours; late, >4 hours).*

	Characteristic	0 to 2 Hours (n=3,087)	2 to 4 Hours (n=1,146)	>4 Hours (n=324)	P Value†
	Age, y	60 (50, 70)	64 (53, 73)	64 (55, 74)	.001
	Men	73	66	65	.001
	Race
	White	92	89	90	.02
	Black	4	7	6	.02
	Diabetes	16	20	25	.001
	Current smoker	46	42	37	.005
	Previous angina	19	16	14	.006
	Previous infarction	19	16	14	.001
	Previous angioplasty	11	8	6	.001
	Inhospital mortality	5	7	8	.0001
	Mortality (30 d)	5	9	9	.001
	*Data are median (25th, 75th percentiles) or percentage of patients. †P values for comparison of early versus late presenters to hospital.

In GUSTO-III, patient characteristics associated with greater delay in the time from symptom onset to hospital arrival were similar to those in GUSTO-I (Table 3). In GUSTO-III, increased age (64 years versus 60 years; P =.001), female sex (35% versus 27%; P =.001), black race (6% versus 4%; P =.02), and diabetes (25% versus 16%; P =.001) were associated with delay in hospital presentation. Current smokers (46% versus 37%; P =.005) and patients with known coronary artery disease (previous angina [19% versus 14%; P =.006], infarction [19% versus 14%; P =.001], or angioplasty [11% versus 6%; P =.001]) tended to arrive at the hospital earlier. The finding that patients with preexisting angina arrived earlier to hospital in GUSTO-III did not parallel the GUSTO-I results, in which patients with previous angina tended to come to the hospital later. Patients with previous infarction or angioplasty, however, arrived earlier in both GUSTO-I and GUSTO-III. Inhospital and 30-day mortality rates demonstrated that a delay in arrival to the hospital is associated with increased mortality risk (8% versus 5% inhospital, P =.0001; 9% versus 5% 30-day, P =.001).

When socioeconomic variables were added to this model, black race was no longer retained in the model and was not a significantly indicated risk for out-of-hospital delay. Educational status, however, was now a risk factor for delay in presentation. Patients who had completed college were significantly more likely (30% versus 26%; P =.002) to come to the hospital earlier (median difference 9 minutes) than patients who did not complete high school or who completed high school but not college.

Whereas time to treatment inhospital decreased from 1990 to 1997, significant delays in therapy were evident for many patient groups. In GUSTO-III, increased age (64 years versus 60 years; P =.0001), female sex (34% versus 27%; P =.001), black race (7% versus 4%; P =.001), hypertension (48% versus 42%; P =.001), diabetes (21% versus 16%; P =.001), previous angina (46% versus 36%; P =.001), and previous infarction (21% versus 16%; P =.001) were associated with delays in treatment (Table 4).

Table 4. The GUSTO-III trial: Characteristics associated with delays in treatment after hospital arrival with associated inhospital, 30-day mortality.*

	Characteristic	0 to 1 Hour (n=3,034)	>1 Hour (n=1,514)	P Value†
	Age, y	60 (51, 70)	64 (53, 73)	.0001
	Men	73	66	.001
	Race
	White	92	88	.001
	Black	4	7	.001
	Hypertension	42	48	.001
	Diabetes	16	21	.001
	Hypercholesterolemia	43	39	.006
	Current smoker	47	38	.001
	Previous angina	36	46	.001
	Previous infarction	16	21	.001
	Inhospital mortality	4	8	.001
	Mortality (30 d)	5	9	.001
	*Data are median (25th, 75th percentiles) or percentage of patients. †P values for comparison of patients treated in the first hour after hospital arrival versus later.

A stepwise regression model combining the US cohorts of GUSTO-I and GUSTO-III identified age, systolic blood pressure, pulse rate, previous infarction, previous bypass surgery, previous angina, sex, race, and infarct location as predictive of time to treatment after hospital arrival. Men were treated 7 minutes faster than women. Patients with previous angina, previous infarction, and previous bypass surgery had delayed therapy (median differences were 5.5, 4.8, and 6.0 minutes, respectively). Older patients were treated later, as were black patients (median delay of 12 minutes). Patients with inferior and anterior infarcts were treated 15 minutes earlier than patients with other infarct sites.

Using GUSTO-III's socioeconomic data, we developed a new model containing the predictive factors of pulse rate, sex, previous angina, black race, infarct location, and whether the patient received public insurance. Patients with public insurance were treated, on average, 6 minutes later than those with private insurance.

Patients with higher educational levels had decreased time from symptom onset to hospital arrival (30% versus 26%; P =.001; Table 5), as did individuals with management or professional occupations (28% versus 21%; P =.001).

Table 5. Socioeconomic factors associated with patient delays from symptom onset to hospital arrival.*

	Socioeconomic Factor	0 to 2 Hours (n=3,087)	2 to 4 Hours (n=1,146)	>4 Hours (n=324)	P Value†
	High school completed	74	68	70	.001
	College/university completed	30	24	26	.002
	Years of school completed	12 (12, 14)	12 (11, 13)	12 (11, 14)	NS
	Current/principal past occupation
	Clerical/sales	19	21	20	NS
	Craftsman	9	9	9	NS
	Homemaker/housewife	10	11	12	NS
	Laborer	35	38	38	NS
	Management/professional	28	21	21	.001
	Type of insurance
	Private	64	60	56	.003
	Public	36	44	44	.001
	Other	2	2	3	NS
	None	7	6	8	NS
	Lives alone	18	21	23	.006
	*Data are median (25th, 75th percentiles) or percentage of patients. †P values for comparison of early versus late presenters to hospital.

NS, Not significant.

Patients with at least a high school education were treated sooner once in the emergency setting than those without this degree (Table 6).

Table 6. Socioeconomic factors associated with delays in treatment after hospital arrival.*

	Socioeconomic Factor	0 to 1 Hour (n=3,034)	>1 Hour (n=1,514)	P Value†
	High school completed	73	69	.017
	College/university completed	29	27	NS
	Years of school completed	12 (12, 14)	12 (11, 14)	NS
	Current/principal past occupation
	Clerical/sales	19	21	NS
	Craftsman	9	8	NS
	Homemaker/housewife	9	12	.004
	Laborer	36	36	NS
	Management/professional	24	23	.026
	Type of insurance
	Private	65	58	.001
	Public	36	45	.001
	Other	2	2	NS
	None	8	6	.013
	Lives alone	19	21	NS
	*Data are median (25th, 75th percentiles) or percentage of patients. †P values for comparison of patients treated in the first hour after hospital arrival versus later.

NS, Not significant.

Private insurance was associated with more rapid delivery of therapy inhospital (65% versus 58%; P =.001), whereas patients with public insurance tended to have delays in therapy (36% versus 45%; P =.001). Individuals without insurance were treated earlier as well (8% versus 6%; P =.013).

Discussion 

Early treatment of patients with AMI is critical to maximizing benefit.8, 9, 10 Our study sought to identify factors associated with delays to hospital arrival as well as inhospital administration of thrombolytic therapy. In the GUSTO-I and GUSTO-III trials, the median time of delay from symptom onset to hospital arrival was 1.4 hours. During the 7 years between these 2 studies, despite widespread knowledge of the benefits of thrombolytic therapy among health care providers and extensive public education programs about AMI, early arrival to the hospital and treatment once in the hospital have not been realized. In particular, the same groups identified in a previous study of GUSTO-I (conducted from 1990 through 1993) as susceptible to delay (elderly, women, patients with diabetes, and minorities) continue to arrive late to the hospital.5 Other studies have confirmed that these patients are at risk for delay.11, 12, 13, 14, 15, 16, 17

Education of patients identified as having increased risk for development of atherosclerosis and subsequent myocardial infarction would appear prudent.18 The National Heart, Lung, and Blood Institute, through the National Heart Attack Alert Program, has supported educational initiatives through a mass media project: Rapid Early Action for Coronary Treatment (REACT).19 Targeting such patients individually, through their family physician or pharmacist, may be more effective than mass media projects, which may not provide focus for patients at high risk. It would appear that some benefit of education was obtained in the years between the 2 trials for patients with previous angina. Although patients with pre-existing angina experienced a delay in hospital presentation in GUSTO I, this was not observed in GUSTO III.

Ten years ago, initial studies evaluating time to treatment once the patient arrived at the hospital identified a typical 90-minute delay before infusion of fibrinolytic therapy.20, 21 Although substantial progress has been realized since then, national registries still identify delays of nearly 40 minutes.4 In GUSTO-I, initiated in 1990, the median time from hospital arrival to treatment was 66 minutes. This period had decreased to 48 minutes when measured in GUSTO-III, which was completed 7 years later.

Patients with previous angina, infarction, and angioplasty tended to present to the hospital earlier with AMI. Paradoxically, these patients with known coronary artery disease tended to have delays in delivery of fibrinolytic therapy. There are multiple possible reasons for this inhospital delay in therapy. Patients presenting with pre-existing coronary artery disease often have baseline electrocardiographic abnormalities that can confound the diagnosis of an acute injury. Typically, the old ECG or the patient's medical records, which describe a previous procedure or hospitalization, must then be obtained from another area of the institution or from another hospital system entirely; this can serve as a significant source of delay in diagnosis. In a similar fashion, contacting the patient's primary care physician or cardiologist for consultation while the patient is in the ED is often necessary for these patients and can also retard treatment.

One finding of this study was that certain groups of patients are especially prone to experience inhospital treatment delays. Among these are patient subgroups at risk for out-of-hospital delay as well, including the elderly, women, minorities, and patients with diabetes.5, 13, 22 Although multiple studies have identified female patients as prone to delays in therapy,13, 18 the other groups have received less attention. Elderly patients with diabetes are known to have atypical symptoms associated with AMI.23, 24 These less typical symptoms serve to confound the clinician in the emergency setting caring for these patients, thus increasing the likelihood of a delay in diagnosis and treatment once the patient arrives at the hospital. Further education of health care providers is essential to prospectively identify these high-risk groups and subsequently to pursue approaches to decrease time to treatment for these patients.

The GUSTO-III study prospectively identified the influence of socioeconomic factors on time to hospital arrival and time to treatment. A patient's educational level, occupation, health insurance, and the presence of family or friends in the home appear to affect health care-seeking behavior. Health care provider behavior appears to be influenced as well. A higher educational level and occupations defined as management or professional are associated with earlier times to hospital arrival and treatment. The more highly educated patient may be aware of the symptoms and risk factors associated with AMI, prompting earlier hospital arrival. Once in hospital, the patient with a higher educational level may better articulate this symptom complex, allowing the treating physician to make a more rapid decision.

Similarly, in patients with symptoms of myocardial infarction, having private health insurance is linked with earlier arrival to the hospital and more rapid administration of fibrinolytic therapy compared with public insurance.25, 26, 27 Living alone also led to a delay in arrival, perhaps as a result of the lack of family or friends necessary to convince a patient of the severity or importance of their symptoms. Other factors, such as education level, availability of a telephone, or income level could also influence the behavior of the patient living alone.

This study evaluating nearly 28,000 patients with myocardial infarction treated in the United States during the GUSTO-I and GUSTO-III trials identifies a number of patients at risk for not receiving thrombolytic therapy early in the course of their infarction. Significant delays in hospital arrival and treatment still exist nearly a decade after the original GUSTO-I trial, suggesting that further effort is necessary to educate patients and health care providers. Although special educational efforts should be targeted to those individuals known to be at high risk for delays in hospital arrival and treatment, such as women, the elderly, and those with less formal education, patients who are not members of these particular groups also suffer delays and require education about AMI.28 The time-dependent nature of thrombolytic treatment for myocardial infarction represents a critical public health education challenge.