Parts of this text have been reproduced from Pufulete et al.4 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.
Aim C: to identify an outcome measure representing a definitive change in clinical management, conditional on having had cardiovascular magnetic resonance, that would be credible to cardiologists and other stakeholders as a measure of the ‘value-added’ by doing cardiovascular magnetic resonance
- Objective C1: to define, using formal consensus methods, a treatment/process outcome that will constitute a definitive change in clinical management arising from having CMR.
- Objective C2: to identify patient subgroups in whom CMR use is indicated, using formal consensus methods and illustrative data from the registry.
- Objective C3: to determine whether or not hospital episode data adequately capture the main changes in clinical management identified using formal consensus.
- Objective C4: to pilot the implementation of a primary outcome based on the changes in management for which consensus was achieved using hospital episode data.
We used a formal consensus method based on the modified nominal group technique66 to identify important changes in management that can be used to define the composite outcome. Formal consensus is a method that combines both research evidence and expert opinion. We used this approach because we knew from a preliminary scoping of the literature that there were few studies that reported the impact of CMR on patient management.4
We used clinical expertise to define ‘clinical events’ that captured the important changes in management and a clinical coder to identify collaborated with codes (ICD-1031 and the OPCS Classification of Intervention and Procedures32) corresponding to these clinical events. We then searched for these codes in hospital episode data for participants in the cohort study during the 12 months following the index admission (PPCI pathway activation).
Literature review
We initially examined a small number of key reports, reviews and primary studies to determine the extent of the literature reporting how CMR influences patient management. There were few studies that directly reported the impact of CMR on patient management, and none of these related to our population of interest. Figure 13 shows the flow diagram of the systematic review and the different types of studies identified. We identified a total of 171 studies reporting the use of CMR in patients with acute MI who activated the PPCI pathway. There were no studies that directly compared groups of patients having CMR or not with respect to patient management or clinical outcomes in this population.
FIGURE 13 Flow diagram of the study selection process and types of studies identified. (Figure not reproduced here; see the original article.)
There were 38 prognostic studies assessing the value of CMR in predicting MACEs; 12 did not distinguish the contributing CMR markers, and 26 assessed individual CMR markers of myocardial damage [e.g. microvascular obstruction (MVO), infarct size (IS), LV ejection fraction (EF), myocardial salvage, intramyocardial haemorrhage] as predictors of MACE. These studies did not all evaluate the prognostic value of CMR indices in addition to indices available from alternative imaging modalities; most were small (20–1217 patients, median 257), single centre and heterogeneous in terms of the CMR parameters reported and how these were assessed. Nevertheless, they showed that CMR markers (particularly MVO, IS and myocardial salvage) were consistently associated with MACEs in STEMI patients.
There were 38 studies of the role of CMR in ‘diagnosing’ a specific feature of the index MI or complications arising from the index MI or procedure. CMR was used to investigate the type of MI (i.e. acute, chronic, aborted) in 13 studies, to detect complications of MI [e.g. LV thrombus, ventricular septal defect (VSD) ventricular aneurysms and pericarditis] in 19 studies and to detect injury from PPCI in five studies. Many of these were case reports, some were diagnostic test accuracy studies comparing CMR with ECHO as the reference standard (e.g. diagnosis of thrombus) and some also followed up patients and assessed the prognostic implications of the CMR findings (e.g. aborted MI).
There were 18 studies in which CMR was used to ‘diagnose’ conditions other than MI (e.g. myocarditis and Takotsubo cardiomyopathy in patients with unobstructed arteries on angiography) and incidental cardiac findings. All but two of these were case reports. There were 71 studies in which CMR was compared with one or more other imaging modalities [i.e. ECHO, SPECT, computed tomography (CT), angiography or with standard coronary angiography with respect to indices of heart function (e.g. myocardial viability, necrosis, perfusion, myocardium at risk, ischaemia)]. Twenty-four of these were diagnostic test accuracy studies (those that included a reference standard and provided statistics of test performance, i.e. sensitivity, specificity, positive or negative predictive values, receiver operating characteristic curves) and 47 were validation or correlation studies. Most focused on infarct characteristics and myocardial viability, but the definitions and techniques assessing viability were highly variable for both CMR and the comparator, making it difficult to compare studies.
Formulation of consensus statements
Thirty-seven draft statements were generated by the three cardiologists and one methodologist (see Appendix 7, Table 31). Examples of draft statements, categorised by patient subgroup, are shown in Appendix 7, Table 32. Statements relating to the same patient subgroup/condition (e.g. patients at risk of complications after PPCI who develop LV thrombus) were condensed into one statement. This reduced the number of statements from 37 to 16. All statements were then reworded to standardise their format. Examples of reworded statements are shown in Appendix 7, Table 32. The 16 statements were discussed in turn at the internal group meeting and the following changes were made:
- The total number of statements was reduced from 16 to 12. Two statements relating to device implantation were condensed into one, three statements were removed because it was felt that their content was covered adequately by other statements and one statement was removed because it was felt that it did not relate to a change in patient management.
- A supporting paragraph was drafted to put the statement into context, summarise the research evidence and include any key references. These references were chosen to reflect the best available evidence; although, for several statements, the cited studies were small prognostic studies or case series (i.e. the only available evidence).
- The statements were reworded to clarify the care pathway compared with which the additional benefit of CMR was anticipated (e.g. alternative imaging modalities such as ECHO or SPECT).
The final 12 statements are shown in Table 22 and examples of supporting paragraphs are shown in Appendix 7, Table 33. These 12 statements describe changes in management relating to six patient subgroups:
- Patients with a poor prognosis or at risk of complications after PPCI [e.g. those who develop LV thrombus, those with low EF who require an implantable cardioverter defibrillator (ICD), those at risk of impending cardiac rupture] whose treatment could potentially be optimised (statements 1, 4, 6, 7 and 9).
- Patients with a good prognosis after PPCI who could be discharged earlier and followed up less often (statement 2).
- Patients with multivessel disease, in whom revascularisation for the additional diseased arteries could be optimised (statements 8 and 10).
- Patients with unobstructed arteries on angiography [who may have had a heart attack that resolved spontaneously (e.g. MI with spontaneous recanalisation/embolic MI) or another condition, such as myocarditis or cardiomyopathy] for whom treatment planning could be optimised based on a definitive diagnosis (statement 5).
- Patients with an out-of-hospital-cardiac arrest (OHCA) in whom the cause could be definitively identified, leading to appropriate treatment (statement 3).
- Patients with incidental cardiac (e.g. mitral regurgitation, pericardial effusion, valve disease, papillary rupture, congenital cardiac anomalies) and extra-cardiac findings (e.g. pleural effusion, lung mass, mediastinal lymph adenopathy), in whom appropriate treatment could be started (statements 11 and 12).
| Statement number | Statement | Patient subgroup |
|---|---|---|
| 1 | Compared with ECHO, CMR after PPCI allows patients with CMR markers that indicate a poor prognosis (e.g. impaired LV function, large infarct size, MVO) to be followed up more appropriately and undergo more aggressive medical therapy for secondary prevention | Poor prognosis/risk of complications after PPCI |
| 2 | Compared with ECHO, CMR after PPCI allows patients with CMR markers that indicate a good prognosis (e.g. normal LV function, high myocardial salvage, no MVO, no residual ischaemia) to be discharged earlier and followed up less frequently | Good prognosis after PPCI |
| 3 | Compared with ECHO, CMR better identifies the cause of OHCA (e.g. large MI, arrhythmogenic right ventricular cardiomyopathy, aberrant coronary arteries, hypertrophic cardiomyopathy) to optimise further treatment for the patient (e.g. defibrillator for primary arrhythmia or PCI) or family members | Out-of-hospital cardiac arrest |
| 4 | CMR after PPCI identifies patients at high risk of having a ventricular septal defect or impending cardiac rupture, who may require a ventricular patch or other urgent cardiac surgery, and guides the optimal management of these patients | Poor prognosis/risk of complications after PPCI |
| 5 | In patients with a ‘normal’ (unobstructed) coronary angiogram, CMR can differentiate patients who have had a MI with spontaneous reperfusion or distal embolisation from patients with a non-ischaemic diagnosis (e.g. myocarditis, Takotsubo cardiomyopathy, aortic dissection), resulting in a patient treatment plan appropriate for the definitive diagnosis | Unobstructed arteries on angiogram |
| 6 | Compared with ECHO, CMR after PPCI better identifies patients at high risk of sudden cardiac death who would benefit most from an implantable cardiac device (e.g. ICD or CRT) | Poor prognosis/risk of complications after PPCI |
| 7 | Compared with ECHO, CMR after PPCI better identifies patients who would not benefit from cardiac resynchronisation therapy | Poor prognosis/risk of complications after PPCI |
| 8 | Compared with stress ECHO or SPECT, CMR after PPCI in patients with multivessel disease better assesses ischaemia and viability of the myocardium to optimise the revascularisation strategy for the patient and avoid additional diagnostic tests | Multivessel disease |
| 9 | Compared with ECHO, CMR after PPCI better identifies patients with LV thrombus for treatment with anticoagulation therapy | Poor prognosis/risk of complications after PPCI |
| 10 | Compared with other imaging modalities, in patients with multivessel disease, CMR after PPCI better identifies the artery that caused the MI and guides the subsequent treatment plan in relation to additional revascularisation (e.g. PCI or coronary artery bypass graft surgery) | Multivessel disease |
| 11 | Compared with ECHO, CMR after PPCI better identifies incidental cardiac findings that may need further investigation and treatment | Incidental cardiac findings |
| 12 | Compared with ECHO, CMR after PPCI identifies significant incidental non-cardiac findings (e.g. oesophageal and lung tumours, pulmonary embolus, aortic aneurysm) that may need further investigation and treatment | Incidental non-cardiac findings |
The expert panel
Nineteen consultant cardiologists were invited to participate in the consensus process. Seven cardiologists (37%) agreed to participate. Of these, two had CMR, three had interventional, one had ECHO and one had heart failure expertise. All seven cardiologists completed the first survey independently.
First survey
Figure 14 shows the distribution of responses for the first survey. There was consensus for three of the 12 statements (25%) regarding the change in management being clinically important: statement 3, relating to the ability of CMR to identify the cause of OHCA and, therefore, optimise treatment for the patient; statement 5, relating to the ability of CMR to provide a definitive diagnosis in patients found to have unobstructed arteries on angiography; and statement 9, relating to the ability of CMR to identify patients with LV thrombus and initiate treatment with anticoagulation therapy. There was no consensus for any statement that the change in management was not clinically important.
FIGURE 14 Median and IQR for the 12 statements in the first survey (n = 7). Boxes represent the median and IQR and the whiskers represent the upper and lower adjacent values (upper quartile + 1.5*IQR, and lower quartile – 1.5*IQR).67 Dots represent extreme values. Statements 3, 5 and 9 were considered to be in consensus (median score ≥ 7 and IQR 6–9). (Figure not reproduced here; see the original article.)
Examples of respondents’ comments to the first survey regarding individual statements are shown in Appendix 7, Table 34. Consensus was sought on the importance of the change in management and most respondents commented on this issue. However, some respondents also considered other factors when rating statements, for example the quality of the supporting evidence, the proportion of patients likely to benefit, the ability of the NHS to provide a service in line with the statement and whether or not the cost of CMR justified the perceived benefit.4
Face-to-face meeting
The face-to-face meeting was attended by six of the seven cardiologists and three non-clinical members of the initial working group. As a result of the face-to-face meeting, the survey was modified as follows:
- The number of statements was reduced from 12 to 10. Statement 10 (relating to patients with multivessel disease) was removed because cardiologists felt that this statement overlapped significantly with statement 8. Statements 11 and 12 (relating to incidental cardiac and non-cardiac findings) were combined into one statement.
- Respondents were asked to rate five aspects of each statement, organised hierarchically – (1) whether or not CMR is better than the comparator (e.g. ECHO), (2) whether or not the information from CMR leads to a change in management, (3) whether or not the change in management is clinically important (i.e. likely to reduce risk of MACEs in the long term), (4) whether or not the change in management is likely to reduce NHS costs in the long term and (5) whether or not the anticipated benefit was sufficiently large to make CMR cost-effective among the patients in whom it would be indicated. Consensus was based on the distribution of responses to questions 1–3 of each statement (i.e. median ≥ 7 and IQR 6–9 for clinically important changes in management, and median ≤ 3 and IQR 1–3 for questions 1, 2 and 3). Responses to questions 1–3 were used because question 1 was constructed to interrogate the respondent’s appraisal of the evidence, question 2 was constructed to determine whether or not the respondent believed the NHS service delivery changed as a result of CMR and question 3 was constructed to determine whether or not the respondent believed that any change recognised in the response to question 2 was clinically important. Discussions in the meeting revealed that consideration of costs and cost-effectiveness for the NHS influenced some panel members’ responses to the initial survey. Questions 4 and 5 were added to isolate these aspects of consideration and separate them from the question of change in management, which was the aim of the consensus process.
- The free-text box in which respondents could justify their score was removed.
The modified survey is shown in Table 23.
| Statement number | Statement |
|---|---|
| 1 | The following statements relate to the ability of CMR to identify patients who have a poor prognosis after PPCI:CMR markers (e.g. impaired LV function, large infarct size, MVO) better identify patients with a poor prognosis after PPCI than markers based on ECHOBetter identification of patients with a poor prognosis after PPCI allows these patients to be followed up more appropriately and treated more aggressivelyMore appropriate follow-up and more aggressive treatment in these patients is expected to lead to a reduced risk of MACEs in the long term |
| 2 | The following statements relate to the ability of CMR to identify patients who have a good prognosis after PPCI:CMR markers (e.g. normal LV function, high myocardial salvage, no MVO, no residual ischaemia) better identify patients with a good prognosis after PPCI than markers based on ECHOBetter identification of patients with a good prognosis after PPCI allows these patients to be followed up less frequentlyLess frequent follow-up in these patients is expected to lead to less NHS resource use in the long term |
| 3 | The following statements relate to the ability of CMR to identify the causes of OHCA in patients who undergo an emergency angiogram:CMR better identifies the cause of OHCA (e.g. large MI, ARVC, aberrant coronary arteries, HCM) than ECHOBetter identification of the cause of OHCA allows treatment to be optimised for these patients (e.g. defibrillator for primary arrhythmia or PCI) or their family members (e.g. genetic screening and counselling, primary prevention)The ability to optimise treatment for these patients or family members is expected to lead to a reduced risk of MACEs in the long term |
| 4 | The following statements relate to the ability of CMR to identify patients with VSD after MI:CMR identifies the location and characteristics of post-infarct VSD better than ECHOBetter identification of the location and characteristics of post-infarct VSD guides the optimal management of these patientsOptimal management of patients with post-infarct VSD is expected to lead to a reduced risk of MACEs in the long term |
| 5 | The following statements relate to the ability of CMR to differentiate MI from other diagnoses in patients found to have unobstructed coronary arteries on emergency angiography:Unlike ECHO, CMR can provide a definitive ischaemic diagnosis (e.g. MI with spontaneous reperfusion or distal embolisation) or a non-ischaemic diagnosis (e.g. myocarditis, Takotsubo cardiomyopathy, aortic dissection) in patients with unobstructed coronary arteries on angiographyA definitive diagnosis results in a patient treatment plan appropriate for that diagnosisA treatment plan appropriate for the diagnosis is expected to lead to a reduced risk of MACEs in the long term |
| 6 | The following statements relate to the ability of CMR to identify patients at high risk of sudden cardiac death after PPCI who would benefit most from an ICD:CMR identifies PPCI patients who are at high risk of sudden cardiac death better than ECHOBetter identification of PPCI patients at high risk of sudden cardiac death allows optimal patient selection for an implantable cardiac device (ICD or CRT)Optimal patient selection for an ICD is expected to lead to a reduced risk of MACEs in these patients in the long term |
| 7 | The following statements relate to the ability of CMR to identify patients who would not benefit from CRT after PPCI:CMR identifies patients who would not benefit from CRT better than echocardiographyThe ability to identify patients who would not benefit from CRT would reduce CRT use in patients who do not need itReducing CRT use in patients who do not need it is expected to lead to reduced risk of MACEs in these patients in the long term |
| 8 | The following statements relate to the ability of CMR to assess ischaemia and viability in patients with multivessel disease:CMR assesses ischaemia and viability of the myocardium better than ECHOBetter assessment of ischaemia and viability of the myocardium optimises the revascularisation strategy for patients with multivessel disease and avoids additional diagnostic testsThe ability to optimise the revascularisation strategy for patients with multivessel disease is expected to lead to a reduced risk of MACEs in the long term |
| 9 | The following statements relate to the ability of CMR to identify patients with post-infarct LV thrombus:CMR identifies post-infarct LV thrombus better than ECHOBetter detection of post-infarct LV thrombus in PPCI patients allows more affected patients to be treated with anticoagulation therapyTreatment with anticoagulation therapy in patients with post-infarct LV thrombus is expected to lead to a reduced risk of MACEs in the long term |
| 10 | The following statements relate to the ability of CMR to detect incidental cardiac and non-cardiac findings if offered routinely to patients who undergo an emergency angiogram:CMR identifies more incidental cardiac/non-cardiac findings than ECHOImproved detection of potentially significant incidental findings allows affected patients to be investigated further and/or treatedFurther investigation and treatment is expected to reduce the risk of MACE/increase overall survival in affected patients in the long term |
Modified survey
Fifty-four cardiologists (including five of the cardiologists who attended the formal consensus meeting) completed the survey. There was consensus that five of the 10 statements (50%) described clinically important changes in management (Figure 15). These were statements 3, 5 and 9 (which were in consensus in the first survey) and two additional statements: statement 1, relating to the ability of CMR to identify patients who have a poor prognosis after PPCI, and statement 8, relating to the ability of CMR to assess ischaemia and viability in patients with multivessel disease. There was no consensus that any of the other statements described a change in management that was not clinically important.4 The final statements from the modified survey are in Appendix 8.
FIGURE 15 Median and IQR for the 10 statements in the modified survey (n = 54). Boxes represent the median and IQR and the whiskers represent the upper and lower adjacent values (upper quartile + 1.5*IQR, and lower quartile – 1.5*IQR).67 Dots represent extreme values. Reproduced from Pufulete et al.4 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. (Figure not reproduced here; see the original article.)
Implementation of a primary outcome based on the changes in management for which consensus was achieved
Patient subgroups
We identified five patient subgroups (Table 24) for which there was consensus that CMR changed management in a clinically important way, either from HISs or from HES/PEDW.
Patients who underwent PPCI could be identified both from HISs and from HES/PEDW. Patients with unobstructed arteries could not be directly identified from either HIS or HES, but they were indirectly identified from HIS (consented into study and underwent angiography but had no data in BCIS) and from HES/PEDW (had a record for angiography but no record for PCI). Patients with multivessel disease could only be identified reliably from HISs; the presence of disease in multiple vessels is not coded by ICD-10 but might be inferred from multiple procedure codes in HES/PEDW for coronary revascularisation (PCI or CABG). Patients who experienced OHCA, and PPCI patients who went on to develop LV thrombus, could be identified only from HES/PEDW.
| Subgroup | Data source | Definition |
|---|---|---|
| PPCI | BCIS | BCIS data recorded at site |
| Multivessel disease | BCIS | Two or more vessels with > 50% stenosis (pre PCI) |
| Unobstructed coronary arteries | BCIS | BCIS data not recorded |
| OHCA | HES/PEDW | Index admission includes HES/PEDW diagnosis I46 ‘Cardiac arrest’ |
| LV thrombus | HES/PEDW | Index admission includes HES/PEDW diagnosis I23.6 ‘Thrombosis of atrium, auricular appendage and ventricle as current complications following acute myocardial infarction’ |
The proportions of patients in each subgroup (across all hospitals and by hospital) are shown in Table 25. Across all hospitals, 89% of patients underwent PPCI, 11% were found to have unobstructed coronary arteries and 44% of patients had multivessel disease. The proportions of patients in each subgroup were similar between hospitals, except for patients with multivessel disease (19% in hospital D vs. 48%, 51% and 53% in hospitals A, B and C, respectively).
| Patient subgroup | Hospital, n (%) | Total (N = 1655), n (%) | |||
|---|---|---|---|---|---|
| A (N = 758) | B (N = 272) | C (N = 316) | D (N = 309) | ||
| PPCI | 651 (86) | 246 (90) | 291 (92) | 279 (90) | 1467 (89) |
| Multivessel disease | 361 (48) | 138 (51) | 169 (53) | 58 (19) | 726 (44) |
| LV thrombus after PPCI | 1 (0.1) | 0 (0) | 0 (0) | 0 (0) | 1 (0.1) |
| Unobstructed coronary arteries | 107 (14) | 26 (10) | 25 (8) | 30 (10) | 188 (11) |
| OHCA | 57 (8) | 22 (8) | 16 (5) | 14 (5) | 109 (7) |
Manual review of patient notes
We conducted a manual review of patient notes because of the uncertainty regarding the identification of the unobstructed coronary arteries subgroup from HISs. Only 38 out of 47 (81%) patients were identified as having unobstructed coronary arteries in the manual review of patient notes (see Appendix 9). Of the remaining nine patients, six had a culprit lesion identified (therefore, the likely reason for their emergency angiography was expected to be MI), and the reason for the coronary angiography in the other three patients was unknown.
We compared the diagnoses from the patient notes with the primary diagnoses recorded in HES for the index admission (see Appendix 9). Four of the 38 patients identified as having unobstructed coronary arteries in their medical notes had ‘myocardial infraction’ as a primary diagnosis in HES. Of the six patients who had a culprit lesion identified in their medical notes, only two had ‘myocardial infraction’ recorded as their primary diagnosis in HES; two had a diagnosis of ‘chronic ischaemic heart disease’, one had a diagnosis of ‘pain in throat and chest’ and one had a diagnosis of ‘atrial fibrillation and flutter’.
Identifying changes in management in hospital episode data
The important changes in management identified from the statements in consensus, the patient subgroups these changes in management refer to, the HES/PEDW data sources used to identify them and the events identified from hospital episode data sources in the 12 months of follow-up after the index admission are shown in Appendix 10. Medication data are not available in hospital episode data; therefore, any changes in medication during follow-up could not be identified. This was particularly relevant for the LV thrombus after PPCI subgroup, for which the start of anticoagulation therapy was identified as the important change in management. All relevant non-ischaemic diagnoses were identifiable in hospital episode admissions data, as were procedures, such as implantation of devices, repeat revascularisation and additional diagnostic tests. In contrast, we could not identify these from the outpatient data set because ICD-10 and OPCS codes associated with each outpatient episode were missing for a large proportion of patients (25% of patients had missing diagnoses codes, 74% of patients had missing procedure codes and 74% of patients had ‘unknown and unspecified causes of morbidity’ as diagnosis). This will lead to an underestimation of diagnostic tests in the cohort, because most diagnostic tests are carried out on an outpatient basis. However, we could identify the frequency of outpatient appointments under the cardiology specialty.
Implementation of the consensus statements
The number of events representing changes in management in the relevant patient subgroups, identified in HES/PEDW up to 12 months after the index admission by CMR status are shown in Table 26. The frequency of events in patients who did and in those who did not have CMR was similar across all subgroups. Patients who had CMR had, on average, more outpatient appointments across all subgroups.
| Patient subgroup | n (%) | Cardiology outpatient appointments, median (IQR) | ||||
|---|---|---|---|---|---|---|
| New diagnoses (non-ischaemic, Takotsubo, myocarditis, pericarditis, endocarditis, coronary spasm) | Changes in medication | Additional diagnostic tests (PET, ECHO, IVUS, pressure wire, radionuclide angiocardiography, CT, angiography) | Implantation of devices (CRT or ICD) | Revascularisation (PCI or CABG) within 3 months | ||
| PPCI | ||||||
| CMR (N = 152) | 0 (0) | No data available | 16 (11) | 0 (0) | N/A | 2 (1–3) |
| No CMR (N = 1312) | 10 (1) | No data available | 129 (10) | 6 (0.5) | N/A | 1 (1–2) |
| Multivessel disease | ||||||
| CMR (N = 104) | 0 (0) | No data available | 11 (11) | 0 (0) | 11 (11) | 2 (1–3) |
| No CMR (N = 622) | 4 (1) | No data available | 61 (10) | 4 (1) | 64 (10) | 1 (1–2) |
| LV thrombus after PPCI | ||||||
| CMR (N = 1) | 0 (0) | No data available | 0 (0) | 0 (0) | N/A | 3 (3–3) |
| No CMR (N = 0) | 0 (0) | No data available | 0 (0) | 0 (0) | N/A | – |
| Unobstructed coronary arteries | ||||||
| CMR (N = 35) | 1 (3) | No data available | 6 (17) | 1 (3) | N/A | 2 (1–3) |
| No CMR (N = 139) | 1 (1) | No data available | 12 (9) | 3 (2) | N/A | 1 (0–2) |
| OHCA | ||||||
| CMR (N = 14) | 0 (0) | No data available | 2 (14) | 0 (0) | N/A | 1 (1–3) |
| No CMR (N = 95) | 1 (1) | No data available | 10 (11) | 5 (5) | N/A | 1 (1–2) |
Table 27 shows the time to first event for each change in management and for all subgroups, by CMR status. Except for patients with unobstructed coronary arteries, patients in other subgroups who did not have CMR had an admission for an additional diagnostic test earlier than patients who had CMR. In the subgroup of patients with unobstructed coronary arteries, patients who did not have CMR had an admission related to a new non-ischaemic diagnosis earlier, and an admission for an additional diagnostic test and implantation of a device later than patients who did get CMR. In the PPCI/multivessel disease subgroups, time to first cardiology outpatient appointment was earlier for patients who did not have CMR than for patients who had CMR.
| Patient subgroup | Number of days until event after the index admission, median (IQR) | |||||
|---|---|---|---|---|---|---|
| New diagnoses (non-ischaemic, Takotsubo, myocarditis, pericarditis, endocarditis, coronary spasm) | Changes in medication | Additional diagnostic tests (PET, ECHO, IVUS, pressure wire, radionuclide angiocardiography, CT, angiography) | Implantation of devices (CRT or ICD) | Revascularisation (PCI or CABG) within 3 months | Cardiology outpatient appointments | |
| PPCI | ||||||
| CMR (N = 152) | – | No data available | 107 (50–240) | – | N/A | 64 (34–99) |
| No CMR (N = 1312) | 16 (8–20) | No data available | 74 (27–171) | 147 (48–171) | N/A | 55 (33–95) |
| Multivessel disease | ||||||
| CMR (N = 104) | – | No data available | 85 (50–221) | – | 99 (50–180) | 76 (52–116) |
| No CMR (N = 622) | 13 (8–110) | No data available | 62 (24–130) | 147 (84–169) | 66 (39–122) | 57 (33–95) |
| LV thrombus after PPCI | ||||||
| CMR (N = 1) | – | No data available | – | – | N/A | 66 (66–66) |
| No CMR (N = 0) | – | No data available | – | – | N/A | – |
| Unobstructed coronary arteries | ||||||
| CMR (N = 35) | 358 (358–358) | No data available | 93 (58–140) | 78 (78–78) | N/A | 67 (52–121) |
| No CMR (N = 139) | 180 (180–180) | No data available | 123 (78–284) | 245 (71–262) | N/A | 67 (44–106) |
| OHCA | ||||||
| CMR (N = 14) | – | No data available | 110 (79–141) | – | N/A | 65 (61–104) |
| No CMR (N = 95) | 48 (48–48) | No data available | 62 (48–93) | 126 (71–167) | N/A | 64 (41–99) |
The frequency of events in CMR versus non-CMR hospitals is shown in Table 28. On average, patients in CMR centres had more outpatient appointments than patients in non-CMR centres. New diagnoses, additional diagnostic tests, implantation of devices and additional revascularisation in patients with multivessel disease were more frequent in CMR hospitals than in non-CMR hospitals. Time to the first event in CMR versus non-CMR centres for all events and by patient subgroup are shown in Table 29. In CMR centres, patients in most subgroups had an admission for an additional diagnostic test and implantation of a device later than patients in non-CMR centres. Time to first cardiology outpatient appointment was earlier for patients in CMR centres across all subgroups.
| Patient subgroup | n (%) | Cardiology outpatient appointments, median (IQR) | ||||
|---|---|---|---|---|---|---|
| New diagnoses (non-ischaemic, Takotsubo, myocarditis, pericarditis, endocarditis, coronary spasm) | Changes in medication | Additional diagnostic tests (PET, ECHO, IVUS, pressure wire, radionuclide angiocardiography, CT, angiography) | Implantation of devices (CRT or ICD) | Revascularisation (PCI or CABG) within 3 months | ||
| PPCI | ||||||
| CMR centres (N = 897) | 6 (1) | No data available | 112 (12) | 5 (1) | N/A | 2 (1–3) |
| Non-CMR centres (N = 567) | 4 (1) | No data available | 33 (6) | 1 (0.2) | N/A | 1 (0–1) |
| Multivessel disease | ||||||
| CMR centres (N = 499) | 4 (1) | No data available | 61 (12) | 4 (1) | 53 (11) | 2 (1–3) |
| Non-CMR centres (N = 227) | 0 | No data available | 11 (5) | 0 | 22 (10) | 1 (0–1) |
| LV thrombus after PPCI | ||||||
| CMR centres (N = 1) | 0 | No data available | 0 | 0 | N/A | 3 (3–3) |
| Non-CMR centres (N = 0) | – | No data available | – | – | N/A | – |
| Unobstructed coronary arteries | ||||||
| CMR centres (N = 133) | 2 (2) | No data available | 16 (12) | 4 (3) | N/A | 1 (0–2) |
| Non-CMR centres (N = 41) | 0 | No data available | 2 (5) | 0 | N/A | 0 (0–1) |
| OHCA | ||||||
| CMR centres (N = 79) | 0 | No data available | 9 (11) | 4 (5) | N/A | 2 (1–3) |
| Non-CMR centres (N = 30) | 1 (3) | No data available | 3 (10) | 1 (3) | N/A | 1 (0–1) |
| Patient subgroup | Number of days until event after the index admission, median (IQR) | |||||
|---|---|---|---|---|---|---|
| New diagnoses (non-ischaemic, Takotsubo, myocarditis, pericarditis, endocarditis, coronary spasm) | Changes in medication | Additional diagnostic tests (PET, ECHO, IVUS, pressure wire, radionuclide angiocardiography, CT, angiography) | Implantation of devices (CRT or ICD) | Revascularisation (PCI or CABG) within 3 months | Cardiology outpatient appointments | |
| PPCI | ||||||
| CMR centres (N = 897) | 14 (8–17) | No data available | 79 (33–205) | 167 (126–171) | N/A | 46 (30–81) |
| Non-CMR centres (N = 567) | 20 (14–34) | No data available | 45 (24–116) | 48 (48–48) | N/A | 90 (57–126) |
| Multivessel disease | ||||||
| CMR centres (N = 499) | 13 (8–110) | No data available | 69 (32–159) | 147 (84–169) | 71 (41–151) | 52 (32–87) |
| Non-CMR centres (N = 227) | – | No data available | 59 (15–122) | – | 63 (43–88) | 95 (68–136) |
| LV thrombus after PPCI | ||||||
| CMR centres (N = 1) | – | No data available | 0 | 0 | N/A | 66 (66–66) |
| Non-CMR centres (N = 0) | – | No data available | – | – | N/A | – |
| Unobstructed coronary arteries | ||||||
| CMR centres (N = 133) | 269 (180–358) | No data available | 103 (68–205) | 162 (75–254) | N/A | 65 (46–103) |
| Non-CMR centres (N = 41) | – | No data available | 184 (69–299) | – | N/A | 89 (54–133) |
| OHCA | ||||||
| CMR centres (N = 79) | – | No data available | 79 (62–141) | 147 (99–169) | N/A | 63 (42–95) |
| Non-CMR centres (N = 30) | 48 (48–48) | No data available | 48 (5–74) | 48 (48–48) | N/A | 86 (52–105) |
Cardiovascular magnetic resonance imaging questionnaire
The changes in management identified by the CMR questionnaire and from HES/PEDW data are shown in Table 30. Changes in management identified by the CMR questionnaire could also be identified in hospital episode data, although there were clearly some gaps in the data we had available to us, namely: (1) poor coding of outpatient hospital episode data (diagnoses), (2) failure to obtain medication data and (3) poor coding of diagnostic tests and investigations (both acute care and outpatient data sets).
| Change in management | n/N (%) | |
|---|---|---|
| CMR questionnaire (n = 187) | HIS or HES (n = 185) | |
| Risk stratification after PPCI | ||
| Patients with markers of a good prognosis | 102/160 (64) | Not possible to identify |
| Patients with markers of a poor prognosis | 58/162 (36) | |
| Change in management? | 55/123 (45) | Yes, potentially identifiable in HES by considering the frequency and timing of events representing changes in management |
| Patients with multivessel disease | 121/165 (73) | 104/185 (56) |
| Change in management (additional diagnostic tests)? | 18/111 (16) | 11/104 (11) |
| Difficult to identify all diagnostic tests because procedure codes were not well recorded in HES/PEDW outpatient data | ||
| Change in management (did CMR optimise revascularisation)? | 95/120 (79) | Yes, potentially identifiable in HES by considering timing of revascularisation and frequency of additional diagnostic tests |
| Patients with LV thrombus | 9/187 (5) | 2/185 (1) |
| Change in management (anticoagulation therapy)? | 5/9 (60) | Not possible to identify in HES as medications not available |
| Potentially possible to identify in HIS if medications on discharge data are provided | ||
| Patients with unobstructed arteries | 19/183 (10) | 35/185 (19) |
| New diagnosis? | 13/16 (81) | 1/35 (3) |
| What were the diagnoses? | Myocarditis (n = 4) Takotsubo cardiomyopathy (n = 5) Heart attack with spontaneous LCX/DS reperfusion and myocarditis (n = 1) Spontaneous LAD reperfusion (n = 1) Borderline low normal EF, possibly consistent with early DCM (n = 1) Prior infarcts (n = 1) | Acute MI (n = 88) Acute MI AND cardiac arrest (n = 1) Acute MI AND cardiomyopathy (n = 1) Acute MI AND chronic ischaemic heart disease (n = 2) Acute MI AND other diseases of the digestive system (n = 1) Acute MI AND pain in throat and chest (n = 1) Angina pectoris (n = 1) Chronic ischaemic heart disease (n = 4) Chronic ischaemic heart disease AND acute MI (n = 2) Heart failure AND chronic ischaemic heart disease AND acute MI (n = 1) Other acute ischaemic heart diseases (n = 1) Other diseases of pericardium AND acute MI (n = 1) |
| Change in management? | 15/19 (79) | Yes, potentially identifiable |
| Patients with OHCA | 24/180 (13) | 14/185 (8) |
| Cause identified by CMR? | 19/24 (79) | 0/14 (new diagnosis recorded) |
| Change in management? | 11/24 (46) | Yes, potentially identifiable |
Summary of main findings
We identified five subgroups of ACS patients who activate the PPCI pathway and for whom there was consensus that CMR changes patient management in a clinically important way: (1) patients who have an OHCA (≈7% of those who activate the PPCI pathway),69 (2) patients who have a ‘normal’ (unobstructed) coronary angiogram (≈10% of those who activate the PPCI pathway),15,16 (3) patients who develop LV thrombus (3% overall and 9% in patients with anterior STEMI),70 (4) patients who have multivessel disease (between 40% and 65% of those who undergo PPCI)7,10,71 and (5) patients in whom CMR markers indicate a poor prognosis (up to 60% of patients after STEMI).2,72 There was consensus about the first three subgroups in both rounds of the survey, and consensus was reached about the remaining two subgroups in the modified survey. These results suggest that CMR benefits a large proportion of patients who activate the PPCI pathway.4 The population investigated was generally representative of the population of patients who activate the PPCI pathway; 11% of patients did not undergo PPCI and were categorised as having unobstructed coronary arteries, 44% of patients had multivessel disease and 7% of patients had an OHCA. Fewer patients than reported in the literature were identified as having LV thrombus (0.3% in our data vs. 3% reported in PPCI populations).
Two subgroups, patients with OHCA and patients who developed LV thrombus after PPCI, could not be identified from HIS data submitted by hospitals but could be identified in the HES/PEDW admissions data relating to the index admission. We defined patients as having unobstructed coronary arteries if they had no BCIS data, because we did not have other information in our HIS data sets for positive identification of this subgroup. We conducted a manual review of these patients’ medical notes because we were uncertain whether or not all patients identified as such truly had unobstructed coronary arteries. The majority of patients (81%) were defined as having unobstructed arteries from their medical notes, although 13% did have a culprit lesion identified, suggesting that the most likely reason that these patients activated the PPCI pathway was MI. We also compared the diagnosis from medical notes with the primary diagnosis in HES for all patients in the unobstructed coronary artery subgroup. Again, there were discrepancies, with some patients who were identified as having unobstructed coronary arteries in their medical notes having a primary diagnosis of MI. We conclude that some patients were incorrectly classified as having unobstructed coronary arteries from HIS, and that classification of this patient subgroup is likely to vary depending on the data source used (medical notes, HIS or hospital episode data).
Some important changes in management identified through the formal consensus (e.g. additional revascularisation in the multivessel disease group, implantation of devices) were readily identifiable in HES/PEDW admission data sets. Some changes in management (e.g. changes in medication) were unavailable because medications data are not available in HES/PEDW. Other changes in management (e.g. additional diagnostic tests) were likely to be underestimated because a large proportion of diagnostic tests are performed on an outpatient basis but procedure codes are not well recorded in outpatient data sets.
When implementing the consensus statements, we considered both the frequency of the events representing the important changes in management and time to the first event. This is because, for some changes in management (e.g. additional revascularisation for multivessel disease patients or a new diagnosis in patients with unobstructed coronary arteries), it is the timing to the event that represents the main change in management resulting from CMR. Although there were differences by CMR status in both the frequency and timing of events for all patient subgroups, we chose not to interpret these because of the small numbers involved and because our main objective was to test the feasibility of implementing the consensus statements using hospital episode data.
The CMR questionnaire highlighted several gaps in hospital episode data (i.e. lack of medications data and poor coding of diagnostic tests and investigations). Without the former data, it would be difficult to characterise patients at baseline (index admission), which is important for identifying and controlling for, confounding, and without the latter data we would be unlikely to have reasonable sensitivity for important changes in management during follow-up in a registry.