topeeng.gif (8383 bytes)

[ Scientific Activity - Actividad Científica ] [ Brief Communications - Temas Libres ]

Validation of a telephone questionnaire for ischemic cardiopathy in Emergency Health Services

Martin-Castro Carmen; Unidad de Investigación EPES.

Empresa Pública de Emergencias Sanitarias 06
Granada, Spain

Abstract
Introduction
Objectives
Material and Methods
Results
Discussion
Conclusions

Abstract
Study Objective: To validate a telephone questionnaire to identify patients with ischemic cardiopathy that request emergency care.
Design: Transversal study in an extrahospital setting.
Setting: The study was performed in four Andalusian provinces (Almeria, Cadiz, Granada and Jaen) of the "Empresa Pública de Emergencias Sanitarias (061)".
Patients: The study sample comprised 763 patients with chest pain recruited from a larger sample of 13689 individuals who called the 061 number for health care
Instruments: A questionnaire was designed by a committee of 12 experts and based on a search of the literature and records of telephone calls made to the Emergency Co-ordination Centers for ischemic cardiopathy. The questionnaire comprises 9 questions adapted to the emergenc setting. Hospital records and/or telephone calls enabled the identification of patients diagnosed with ischemic cardiopathy (acute myocardial infarction and/or unstable angina) The questions were weighted and the questionnaire was validated, with an evaluation of its calibration and discrimination.
Results: Logistic regression analysis yielded a telephone questionnaire with six questions: (History of ischemic cardiopathy? Age over 40 years? Chest pain irradiating to the left arm? Vegetative symptoms within a multi-symptom picture? Does the chest pain abate after intake of nitrites? Which sex?) The calibration was assessed with the chi-squared test: c 2= 4.3613, p = 0.7374, and the discrimination with the area under the ROC curve: 0.6904, p < 0.05.
Conclusions: The proposed protocol calibrates and discriminates well and adequately differentiates between patients with and without ischemic cardiopathy. Our results suggest that it should be systematically employed at all health emergency coordination centers in order to obtain an improved distribution of resources. Similar studies would be of value to address other types of care frequently requested from emergency coordination centers.

Top

Introduction:

The activity of the different coordination centers of the Emergency services is largely based on prehospital telephone triage. The aim is to prioritize the calls received so as to assign the most appropriate resource and avoid the saturation of either the Emergency Service itself or the referral hospital center. Emergency teams are decisive for administering the earliest possible treatment and for reducing the time to hospital admission for this type of patient. The telephone diagnosis of these critically ill patients in an Emergency Service co-ordination suite is based on a standardized set of questions. The co-ordination centers in our region do not currently use scientifically validated protocols, although they are reported in the literature. Failure to accurately distinguish patients with ischemic heart disease implies either a waste of resources or the increased risk of an individual’s death.
The defensive medicine approach is a major pitfall for emergency services, and can lead to an indiscriminate employment of their resources. Diagnostic decisions in this setting require the support of scientifically validated telephone protocols to achieve optimal resource utilization.

Objectives:

The aim of the present study is to validate a telephone protocol to identify patients with ischemic heart disease from among those calling the 061 telephone for emergency health care.

Top

Material and Methods:

A cross-sectional study was performed in an extrahospital setting on patients requesting emergency care via the 061 telephone number in four provinces of Southern Spain (Almeria, Cadiz, Granada and Jaen) during May, June, July and August 1997. The four emergency services in these provinces serve a population of 1 245 030 people. We gathered a sample of 736 calls reporting chest pain suggestive of ischemic heart disease (acute myocardial infarction or unstable angina) from a larger sample of 13 689 calls. Out of the 736 calls with suggestive chest pain, 6% of the total sample, the diagnosis of ischemic heart disease (acute myocardial infarction and/or unstable angina: codified as 410 and 411 respectively in the 9th revision of the International Classification of Diseases) was confirmed in 263 cases. All calls received by the coordination centers for chest pain were collected, in order to avoid interference with their normal functioning.

We examined the protocols published in the literature and established a committee of experts to propose a questionnaire for evaluation. We compiled the questions normally asked when this type of call is received, including standard 061 call data: location of patient, age, date and hour, and specific questions for chest pain [ history of ischemic heart disease, localization of pain (epigastric and/or precordial), pain features (fixed and/or radiating), type of radiation (to left arm and/or mandible), receiving antianginal medication, having taken nitrites to relieve the pain and over or under 30 minutes since onset of symptoms, presence of choking, choking plus symptoms of autonomic dysfunction (sweating, nausea and/or vomiting), symptoms of autonomic dysfunction accompanied by many other symptoms], as well as data (telephone and address) to enable follow-up when the mobile intensive care unit (ICU) was not sent. In the following 48 hours the referral hospital records were examined to register whether the patient was diagnosed with ischemic heart disease (acute myocardial infarction and/or unstable angina).
The presence of ischemic heart disease was related to the different variables in the questionnaire, and contingency tables were constructed between ischemic heart disease and each of the predictors independently. The Chi-squared test was used for categoric variables and the Student’s t test for continuous variables. We considered the relationship between the presence of ischemic heart disease (outcome variable) and the different independent variables to be statistically significant when the expected alpha error presented a probability below 0.05.
Using logistic regression, a model was constructed from the questionnaire and other common questions in order to construct a prognostic index. This prognostic index was used to construct a ROC curve, as well as to give the sensitivity, specificity, positive predictive value and negative predictive value, optimizing the cut-off point according to different criteria.

A multivariate model was constructed in which the dependent variable was the presence of ischemic heart disease, and logistic regression was performed to identify the set of factors that can predict ischemic heart disease. Measurement of the deviation was used to observe the fit of the model first without and then with the inclusion of each variable; the Chi-squared test was used to detect whether the change after each step was significant (p<0.05).

The method to estimate the parameters was that of maximum likelihood. For the selection of the variables the non-automatic back stepwise method was used. To assess the linearity of the continuous predictors existing diagnostic methods for this purpose were employed.
The calibration and discrimination of the protocol model proposed was tested. The calibration used the Hosmer-Lemeshow test, comparing the expected and observed number of patients with and without ischemic heart disease by risk group.

Top

Results:

The first stage of screening identified 13,689 calls for assistance in the four provinces during the study period, of which 736 reported chest pain. The geographical distribution of the chest pain calls is shown in Table 1. Fifty-three per cent were males; the mean age was 63 years, standard deviation 15 years. The frequency and percentages of the different variables are listed in Table 2.
There was a statistically significant association between all independent variables studied and the presence of ischemic heart disease, except for the following: presence of choking plus symptoms of autonomic dysfunction, epigastric localization and pain radiating to the mandible (Table 3).
We initially included in the logistic regression analysis all the variables with statistical significance in the univariate analysis and others we believed should be included for their clinical relevance. The variables that entered the model were: history of ischemic heart disease; age over 40 years: pain radiating to left arm: symptoms of autonomic dysfunction with multiple concomitant symptoms; having taken nitrites to relieve the pain; and, although the gender variable was not significant in the model, gender was retained because we judged it to be clinically relevant. (Table 4).
Table 5 lists the values obtained with the Hosmer-Lemeshow test, showing a Chi-squared of 4.3615 with 7 degrees of freedom and p=0.7374: There were no statistically significant differences between the observed and the expected number of ischemic heart disease cases.
Assessment of the discrimination of the proposed protocol (the degree to which it differentiates between patients with and without ischemic heart disease) was done with a study of the sensitivity, specificity, positive predictive value and negative predictive value of each of the variables (Table 6). The discrimination of ischemic heart disease by the model was assessed with the area under the Receiver Operating Characteristics (ROC) curve, known as the Diagnostic Performance Curve in the clinical setting, which gave an area for our model of 0.6904, p<0.05 (Table 7).

Table 1
table1.gif (5470 bytes)

Top

Table 2
table2.gif (10348 bytes)

Top

Table 3

table3.gif (9666 bytes)

Top


Table 4
table4.gif (8679 bytes)

Top

Table 5
table5.gif (9117 bytes)

Top

Table 6
table6.gif (10821 bytes)

Top


Table 7

table7.gif (7968 bytes)

Top

Discussion:

The results of our study on the calibration and discrimination of the protocol are similar to those of other studies, which emphasize such symptoms as radiating pain, the patient’s history of heart problems, and pain that is refractory to nitrites. When the different questions comprising our questionnaire were applied to patients calling with chest pain, the following isolated symptoms were significant: history of ischemic heart disease, radiating pain, pain radiating to the left arm, precordial localization, fixed pain, that the patient was taking antianginal medication, that the pain was refractory to nitrites, symptoms of autonomic dysfunction accompanied by other multiple symptoms, and being a male over 40 years. There were no statistically significant results for other symptoms proposed in the literature, such as choking, choking plus symptoms of autonomic dysfunction, pain radiating to the mandible (classically regarded as important) or duration of pain over 30 minutes.
Our study shows slightly superior predictive abilities compared with those described by Pozen in 1980 and 1984, Goldman in 1982 and 1988, Tierney in 1983 and Grijseels in 1995. However, these studies evaluated the predictive ability of symptoms suggestive of ischemic heart disease alongside electrocardiogram findings. We concur with reports in the literature that while the symptoms studied catalogue well patients with ischemic heart disease, some of them have inadequate prognostic ability and must be combined with another diagnostic method such as electrocardiography.
Our selection of patients may represent a study limitation, since to avoid interference in the normal activity of the service we included all consecutive calls for chest pain suggestive of ischemic heart disease, regardless of whether the emergency vehicle was sent. It is also possible that a recall bias was produced by the commonly stressful situation of the caller, which could alter the information sent.

Conclusions:

Our model calibrates and discriminates well, and accurately predicts patients with ischemic heart disease from those without it. It is therefore an effective tool to rationally distribute available health emergency resources and to reduce both false positives and false negatives, thus contributing to an improved quality of care. All patients presenting the variables in this questionnaire must be attended rapidly by an emergency team.

 

Questions, contributions and commentaries to the Authors: send an e-mail message (up to 15 lines, without attachments) to coronary-pcvc@pcvc.sminter.com.ar , written either in English, Spanish, or Portuguese.

Top


© CETIFAC
Bioengineering
UNER

Update
Nov/12/1999