Saturday 22 April 1978
thought that the role of home care for patients suspected myocardial infarction merited further scrutiny. To ensure that the results of any new trial were applicable to the majority of patients with suspected myocardial infarction, we wished to secure as high a randomisation-rate as possible and to use criteria for entry and exclusion which were clearly defined and rigorously followed. Mather et al.2 had used a large number of general practitioners in a wide area to enter patients into their study. We sent a hospital-based team to patients’ homes in response to calls from general practitioners. The team made an initial working diagnosis, provided emergency treatment, excluded unsuitable patients, and randomly allocated the remainder to home or hospital care. We
A RANDOMISED TRIAL OF
with
HOME-VERSUS-HOSPITAL MANAGEMENT FOR PATIENTS WITH SUSPECTED
MYOCARDIAL INFARCTION J. D. HILL*
J. R. HAMPTON J. R. A. MITCHELL Department of Medicine, General Hospital, Nottingham NG1 6HA
Summary
Home and tients with
hospital management of pasuspected myocardial infarc-
tion were compared in a randomised trial in which a hospital-based team responded to calls from general practitioners. 500 calls were received, and 349 patients (70%) were suspected of having myocardial infarction. Of these, 24% were excluded from the trial on predeter-
mined medical and social
Methods
grounds;
for the remainder (76%) there was no significant difference in the 6-week mortality between the home group (13%) and the hospital group (11%). For the majority of patients to whom a general practitioner is called because of suspected infarction, hospital admission confers no clear advantage.
Introduction
continuing claims for the life-saving value of and hospital prehospital coronary care, and despite very considerable investment of financial and human resources in their provision, community mortality from myocardial infarction has not changed in the past few years.’ Mather and his colleagues2 reappraised the role of hospital admission in a randomised home-versus-hospital study in- Bristol and the South-West, but only 24% of their patients with suspected myocardial infarction could be allocated randomly to the two treatment groups, the remainder being electively admitted to hospital or retained at home. A joint working-party of the Royal College of Physicians and the British Cardiac Society3 dismissed the results of this study because "there were defects in its design, many patients were seen late after the onset of their symptoms and there was a small and ill-defined minority of the patients who were randomised". It was "the unanimous view of the Working Party that some form of specialised accommodation for the care of patients after cardiac infarction is essential" and that "the D.H.S.S. should actively encourage the development of mobile coronary care, wherever possible with doctor-manned mobile coronary-care units". DESPITE
*Present address: Bradford
Royal Infirmary.
’
60 general practitioners working in and around the City of Nottingham agreed to participate in the study. 9 areas within a radius of 15 miles of Nottingham General Hospital were chosen, all served by fast arterial roads. The largest group practice taking part (population 24 000) was 4 miles from the hospital. The hospital-based team, which consisted of a senior houseofficer and a coronary-care-unit (c.c.u.) nurse, drove an ambulance-type estate-car to the patient’s home, taking with them a battery-operated electrocardiograph (E.C.G.), defibrillator, and monitor, plus resuscitation equipment and appropriate drugs. The occupants of the car were insured under the hospital casualty team policy. No age-limit was applied to the study, and when a patient in one of the designated areas reported symptoms which caused his general practitioner to suspect a myocardial infarction, the doctor contacted the team with the hospital emergency bleep system. He could do so either before or after he had seen the patient, depending on his commitments
and
on
the information received. The team went to the as quickly as traffic and weather conditions
patient’s home allowed. Once in the
patient’s home, the team gave any necessary emergency treatment and recorded the history and the results of physical examination and an initial E.C.G. On the basis of this information, a provisional diagnosis was made. If myocardial infarction was not suspected, the patient was managed according to the presumed diagnosis. If myocardial infarction was suspected, the patient’s suitability for entry into the trial was assessed. Patients were considered unsuitable if (1) they had been successfully resuscitated by the team; (2) they had complications of myocardial infarction which were thought to need hospital care-e.g., complete heart-block requiring pacemaker insertion; (3) they had coincidental disease requiring hospital admission; or (4) they were socially unsuitable because of their housing or family circumstances. The remaining patients were observed in the home for 2 hours. Sealed envelopes containing a card indicating "home" or "hospital" management had been prepared from a table of random 8069
838
numbers, and
at
the end of the 2-hour
period,
the
next
envel-
ope was opened.
If "home" was indicated, the team returned to the hospital, leaving two 50 mg pethidine tablets at the home. The patient was managed thereafter by his general practitioner, who could, if he wished, call the team back to see the patient. Patients initially randomised to the home group but who subsequently required hospital admission, with or without a further team visit, will be referred to as "failures of home management" and will remain in their original group for analysis. If "hospital" was indicated, the patient was admitted by ambulance (occasionally by a coronary ambulance) to the c.c.u. at Nottingham General Hospital under the care of the University Department of Medicine. The pattern of care adopted was that followed for all patients with myocardial infarction-i.e., 1-2 days in the c.c.u. followed by 7-8 days of mobilisation in the general ward. Patients were not-given any drugs routinely, but complications—e.g., dysrhythmias, cardiac faure-were treated as they occurred. In both the home and hospital groups evidence of infarction was collected according to a set plan. At their initial visit, the team recorded an E.c.G. and collected a blood-sample for cardiac enzyme estimations; further E.c.G.s and blood-samples were obtained for the next 2 days by the general practitioner or his district nurse if the patient was at home, and by the normal services if the patient was in hospital. All cardiac-enzyme estimations were done by one laboratory. After all the evidence had been collected, it was analysed by one observer (J.D.H.), and the patient was allocated to one of the following fi_nal diagnostic categories: (a) Definite myocatdial infarction.-Patients with a convincing clinical history, together with unequivocal E.C.G. evidence (pathological Q waves and sequential S-T segment changes) and a pronounced rise in cardiac enzymes (serumaspartate-transaminase). (b) Probable myocardial infarction.-Patients with a convincing clinical history together with either unequivocal E.C.G. changes or a pronounced rise in serum-enzyme values. (c) Possible myocardial infarction.-Patients with a corfvincing clinical history together with either an abnormal E.C.G. which was not typical of a recent myocardial infarction or a
unaware of the patients’ management group. Their conclusions were compared with those reached by J.D.H. All survivors were seen in the outpatient clinic 6 weeks after the event requiring the first team visit, and their clinical state and B.C.G. were recorded. As part of a series of studies on the ambulance services in Nottinghamarecord was being kept of all patients with suspected myocardial infarction who were referred to the Nottingham hospitals throughout the period of the home-or-hospital study, from both participating and non-participating general
practices. Results The study lasted 4 years, and this report describes the first 500 calls received by the team: 14 patients were dead when the team arrived, and 125 patients were thought to have diagnoses other than myocardial infarction (table I). There were thus 361 patients in whom myocardial inTABLE II-PATIENTS WITH SUSPECTED MYOCARDIAL INFARCTION WHO WERE NOT ENTERED INTO RANDOMISED TRIAL
TABLE TII-SOCIAL REASONS FOR NON-INCLUSION IN THE TRIAL
marginal rise in serum-enzymes. (d) /schaemic heart-disease.-Patients with a clear history of a previous myocardial infarction or angina together with an abnormal but unchanging E.C.G. and normal serum-enzyme values. (e) Chest paincause,-Patients who were initially suspected of having a myocardial infarction but in whom no adequate final diagnosis could be made. Categories (a) and (b) combined correspond approximately to the W.H.O. definition of "definite myocardial infarction". The clinical records, E.C.G. tracings, and enzyme results of a sample of 100 consecutive cases were classified independently by the two other observers (J.R.A.M. and J.R.H.), who were TABLE I-DIAGNOSIS IN PATIENTS IN WHOM MYOCARDIAL
INFARCTION WAS NOT SUSPECTED
farction was suspected. All but 2 of these had chest pain, with or without other symptoms: 1 patient had pain in the wrists, which he had also had in a previous infarction, and 1 patient had an unexplained fall in bloodpressure. 12 patients were excluded from further consideration-: 2 because they were not on the list of a participating general practitioner; 6 because they were already in the trial from a previous episode; and 4 because of administrative errors (E.C.G. machine not
working, wrong envelope opened, &c.). Of the 349 apparently suitable patients with suspected myocardial infarction, 85 (24%) could not be randomised into the trial (table n). The medical reasons for non-inclusion were complications of infarction in 39 patients and other medical diseases in 9. The social reasons
for non-inclusion
are
shown in table
III.
Of the 85
patients not randomised, 57 (67%) were subsequently diagnosed as having definite or probable myocardial infarction. The complications of myocardial infarction which led to elective admission to hospital and therefore to
of
non-randomisation are shown in table iv. Of the 349 suitable patients whom the team suspected
having
a
myocardial infarction,
264
(76%)
were
839 TABLE IV-COMPLICATIONS OF MYOCARDIAL INFARCTION OR ITS
TREATMENT WHICH LED TO ELECTIVE HOSPITAL ADMISSION
therefore randomised into the trial. Table v shows the composition of the random groups. Table vi shows the treatment given in the home by the team. 79 (60%) of the 132 patients in the home group and 71 (54%) of the 132
patients in the hospital group were subsequently diagnosed as having definite or probable myocardial inwas no significant difference between the groups for any of the parameters measured, including the numbers of patients in each group with given levels of cardiac enzymes, although the patients randomised to home care had a higher definite/probable infarct rate. The accompanying figure shows the time from the onset of symptoms to the first request for help from a doctor, and from the onset of symptoms to the team’s arrival at the house. The times elapsed do not differ significantly between the two randomised groups. Our trial was intended to answer the question facing general practitioners who are called to the home of patients whom they suspect of having a myocardial infarction-namely, "If I exclude patients whose medical and social condition makes hospital admission appear desirable, can I safely leave the remainder at home?" The
farction. There
TABLE V--COMPOSITION OF THE RANDOM GROUPS
Cumulative percentage of randomised patients who had called their general practitioner and had been visited by the team at specified intervals after the onset of symptoms.
trial
must be analysed to treat and on the basis when the decision must be
the initial intention of the information available taken. In all the analyses the therefore patients are compared by their initial randomisation categories, so that a death occurring after hospital admission because home management had failed is still attributed to the home-care group. 26 patients randomised to the home group subsequently required hospital admission. Failure of home management constituted an important trial end-point, and table vn shows the reasons for failure of home care. Half of those who were admitted came to hospital within 48 hours of the team’s initial visit. Table vni shows the total 6-week mortality in all the patients in whom the initial working diagnosis was suspected myocardial infarction. The death-rate in the two randomised groups does not differ significantly, so for these patients, hospital admission did not influence 6-week mortality. There is, however, a significantly
according to
TABLE VII-REASONS FOR FAILURE OF HOME MANAGEMENT
TABLE
TABLE VI-DRUGS GIVEN BY TEAM TO THE RANDOM GROUPS
VIII-6-WEEK
MORTALITY
(%) IN PATIENTS WITH
SUSPECTED INFARCTION
840 TABLE IX——CUMULATIVE MORTALITY IN THE GROUPS
groups is due to the ability of hospital intervention defer death but not necessarily to prevent it. The mortality figures in the two groups therefore move closer together as time elapses from the initial infarction epitwo
to
*
3 died in hospital.
higher mortality in the patients who were excluded from for all the trial than in the randomised groups (P
thought that the role of home care for patients suspected myocardial infarction merited further scrutiny. To ensure that the results of any new trial were applicable to the majority of patients with suspected myocardial infarction, we wished to secure as high a randomisation-rate as possible and to use criteria for entry and exclusion which were clearly defined and rigorously followed. Mather et al.2 had used a large number of general practitioners in a wide area to enter patients into their study. We sent a hospital-based team to patients’ homes in response to calls from general practitioners. The team made an initial working diagnosis, provided emergency treatment, excluded unsuitable patients, and randomly allocated the remainder to home or hospital care. We
A RANDOMISED TRIAL OF
with
HOME-VERSUS-HOSPITAL MANAGEMENT FOR PATIENTS WITH SUSPECTED
MYOCARDIAL INFARCTION J. D. HILL*
J. R. HAMPTON J. R. A. MITCHELL Department of Medicine, General Hospital, Nottingham NG1 6HA
Summary
Home and tients with
hospital management of pasuspected myocardial infarc-
tion were compared in a randomised trial in which a hospital-based team responded to calls from general practitioners. 500 calls were received, and 349 patients (70%) were suspected of having myocardial infarction. Of these, 24% were excluded from the trial on predeter-
mined medical and social
Methods
grounds;
for the remainder (76%) there was no significant difference in the 6-week mortality between the home group (13%) and the hospital group (11%). For the majority of patients to whom a general practitioner is called because of suspected infarction, hospital admission confers no clear advantage.
Introduction
continuing claims for the life-saving value of and hospital prehospital coronary care, and despite very considerable investment of financial and human resources in their provision, community mortality from myocardial infarction has not changed in the past few years.’ Mather and his colleagues2 reappraised the role of hospital admission in a randomised home-versus-hospital study in- Bristol and the South-West, but only 24% of their patients with suspected myocardial infarction could be allocated randomly to the two treatment groups, the remainder being electively admitted to hospital or retained at home. A joint working-party of the Royal College of Physicians and the British Cardiac Society3 dismissed the results of this study because "there were defects in its design, many patients were seen late after the onset of their symptoms and there was a small and ill-defined minority of the patients who were randomised". It was "the unanimous view of the Working Party that some form of specialised accommodation for the care of patients after cardiac infarction is essential" and that "the D.H.S.S. should actively encourage the development of mobile coronary care, wherever possible with doctor-manned mobile coronary-care units". DESPITE
*Present address: Bradford
Royal Infirmary.
’
60 general practitioners working in and around the City of Nottingham agreed to participate in the study. 9 areas within a radius of 15 miles of Nottingham General Hospital were chosen, all served by fast arterial roads. The largest group practice taking part (population 24 000) was 4 miles from the hospital. The hospital-based team, which consisted of a senior houseofficer and a coronary-care-unit (c.c.u.) nurse, drove an ambulance-type estate-car to the patient’s home, taking with them a battery-operated electrocardiograph (E.C.G.), defibrillator, and monitor, plus resuscitation equipment and appropriate drugs. The occupants of the car were insured under the hospital casualty team policy. No age-limit was applied to the study, and when a patient in one of the designated areas reported symptoms which caused his general practitioner to suspect a myocardial infarction, the doctor contacted the team with the hospital emergency bleep system. He could do so either before or after he had seen the patient, depending on his commitments
and
on
the information received. The team went to the as quickly as traffic and weather conditions
patient’s home allowed. Once in the
patient’s home, the team gave any necessary emergency treatment and recorded the history and the results of physical examination and an initial E.C.G. On the basis of this information, a provisional diagnosis was made. If myocardial infarction was not suspected, the patient was managed according to the presumed diagnosis. If myocardial infarction was suspected, the patient’s suitability for entry into the trial was assessed. Patients were considered unsuitable if (1) they had been successfully resuscitated by the team; (2) they had complications of myocardial infarction which were thought to need hospital care-e.g., complete heart-block requiring pacemaker insertion; (3) they had coincidental disease requiring hospital admission; or (4) they were socially unsuitable because of their housing or family circumstances. The remaining patients were observed in the home for 2 hours. Sealed envelopes containing a card indicating "home" or "hospital" management had been prepared from a table of random 8069
838
numbers, and
at
the end of the 2-hour
period,
the
next
envel-
ope was opened.
If "home" was indicated, the team returned to the hospital, leaving two 50 mg pethidine tablets at the home. The patient was managed thereafter by his general practitioner, who could, if he wished, call the team back to see the patient. Patients initially randomised to the home group but who subsequently required hospital admission, with or without a further team visit, will be referred to as "failures of home management" and will remain in their original group for analysis. If "hospital" was indicated, the patient was admitted by ambulance (occasionally by a coronary ambulance) to the c.c.u. at Nottingham General Hospital under the care of the University Department of Medicine. The pattern of care adopted was that followed for all patients with myocardial infarction-i.e., 1-2 days in the c.c.u. followed by 7-8 days of mobilisation in the general ward. Patients were not-given any drugs routinely, but complications—e.g., dysrhythmias, cardiac faure-were treated as they occurred. In both the home and hospital groups evidence of infarction was collected according to a set plan. At their initial visit, the team recorded an E.c.G. and collected a blood-sample for cardiac enzyme estimations; further E.c.G.s and blood-samples were obtained for the next 2 days by the general practitioner or his district nurse if the patient was at home, and by the normal services if the patient was in hospital. All cardiac-enzyme estimations were done by one laboratory. After all the evidence had been collected, it was analysed by one observer (J.D.H.), and the patient was allocated to one of the following fi_nal diagnostic categories: (a) Definite myocatdial infarction.-Patients with a convincing clinical history, together with unequivocal E.C.G. evidence (pathological Q waves and sequential S-T segment changes) and a pronounced rise in cardiac enzymes (serumaspartate-transaminase). (b) Probable myocardial infarction.-Patients with a convincing clinical history together with either unequivocal E.C.G. changes or a pronounced rise in serum-enzyme values. (c) Possible myocardial infarction.-Patients with a corfvincing clinical history together with either an abnormal E.C.G. which was not typical of a recent myocardial infarction or a
unaware of the patients’ management group. Their conclusions were compared with those reached by J.D.H. All survivors were seen in the outpatient clinic 6 weeks after the event requiring the first team visit, and their clinical state and B.C.G. were recorded. As part of a series of studies on the ambulance services in Nottinghamarecord was being kept of all patients with suspected myocardial infarction who were referred to the Nottingham hospitals throughout the period of the home-or-hospital study, from both participating and non-participating general
practices. Results The study lasted 4 years, and this report describes the first 500 calls received by the team: 14 patients were dead when the team arrived, and 125 patients were thought to have diagnoses other than myocardial infarction (table I). There were thus 361 patients in whom myocardial inTABLE II-PATIENTS WITH SUSPECTED MYOCARDIAL INFARCTION WHO WERE NOT ENTERED INTO RANDOMISED TRIAL
TABLE TII-SOCIAL REASONS FOR NON-INCLUSION IN THE TRIAL
marginal rise in serum-enzymes. (d) /schaemic heart-disease.-Patients with a clear history of a previous myocardial infarction or angina together with an abnormal but unchanging E.C.G. and normal serum-enzyme values. (e) Chest paincause,-Patients who were initially suspected of having a myocardial infarction but in whom no adequate final diagnosis could be made. Categories (a) and (b) combined correspond approximately to the W.H.O. definition of "definite myocardial infarction". The clinical records, E.C.G. tracings, and enzyme results of a sample of 100 consecutive cases were classified independently by the two other observers (J.R.A.M. and J.R.H.), who were TABLE I-DIAGNOSIS IN PATIENTS IN WHOM MYOCARDIAL
INFARCTION WAS NOT SUSPECTED
farction was suspected. All but 2 of these had chest pain, with or without other symptoms: 1 patient had pain in the wrists, which he had also had in a previous infarction, and 1 patient had an unexplained fall in bloodpressure. 12 patients were excluded from further consideration-: 2 because they were not on the list of a participating general practitioner; 6 because they were already in the trial from a previous episode; and 4 because of administrative errors (E.C.G. machine not
working, wrong envelope opened, &c.). Of the 349 apparently suitable patients with suspected myocardial infarction, 85 (24%) could not be randomised into the trial (table n). The medical reasons for non-inclusion were complications of infarction in 39 patients and other medical diseases in 9. The social reasons
for non-inclusion
are
shown in table
III.
Of the 85
patients not randomised, 57 (67%) were subsequently diagnosed as having definite or probable myocardial infarction. The complications of myocardial infarction which led to elective admission to hospital and therefore to
of
non-randomisation are shown in table iv. Of the 349 suitable patients whom the team suspected
having
a
myocardial infarction,
264
(76%)
were
839 TABLE IV-COMPLICATIONS OF MYOCARDIAL INFARCTION OR ITS
TREATMENT WHICH LED TO ELECTIVE HOSPITAL ADMISSION
therefore randomised into the trial. Table v shows the composition of the random groups. Table vi shows the treatment given in the home by the team. 79 (60%) of the 132 patients in the home group and 71 (54%) of the 132
patients in the hospital group were subsequently diagnosed as having definite or probable myocardial inwas no significant difference between the groups for any of the parameters measured, including the numbers of patients in each group with given levels of cardiac enzymes, although the patients randomised to home care had a higher definite/probable infarct rate. The accompanying figure shows the time from the onset of symptoms to the first request for help from a doctor, and from the onset of symptoms to the team’s arrival at the house. The times elapsed do not differ significantly between the two randomised groups. Our trial was intended to answer the question facing general practitioners who are called to the home of patients whom they suspect of having a myocardial infarction-namely, "If I exclude patients whose medical and social condition makes hospital admission appear desirable, can I safely leave the remainder at home?" The
farction. There
TABLE V--COMPOSITION OF THE RANDOM GROUPS
Cumulative percentage of randomised patients who had called their general practitioner and had been visited by the team at specified intervals after the onset of symptoms.
trial
must be analysed to treat and on the basis when the decision must be
the initial intention of the information available taken. In all the analyses the therefore patients are compared by their initial randomisation categories, so that a death occurring after hospital admission because home management had failed is still attributed to the home-care group. 26 patients randomised to the home group subsequently required hospital admission. Failure of home management constituted an important trial end-point, and table vn shows the reasons for failure of home care. Half of those who were admitted came to hospital within 48 hours of the team’s initial visit. Table vni shows the total 6-week mortality in all the patients in whom the initial working diagnosis was suspected myocardial infarction. The death-rate in the two randomised groups does not differ significantly, so for these patients, hospital admission did not influence 6-week mortality. There is, however, a significantly
according to
TABLE VII-REASONS FOR FAILURE OF HOME MANAGEMENT
TABLE
TABLE VI-DRUGS GIVEN BY TEAM TO THE RANDOM GROUPS
VIII-6-WEEK
MORTALITY
(%) IN PATIENTS WITH
SUSPECTED INFARCTION
840 TABLE IX——CUMULATIVE MORTALITY IN THE GROUPS
groups is due to the ability of hospital intervention defer death but not necessarily to prevent it. The mortality figures in the two groups therefore move closer together as time elapses from the initial infarction epitwo
to
*
3 died in hospital.
higher mortality in the patients who were excluded from for all the trial than in the randomised groups (P
