- Email: [email protected]
Prevalence and incidence of Q-wave unrecognized myocardial infarction in general population: Diagnostic value of the electrocardiogram. The REGICOR study
International Journal of Cardiology 225 (2016) 300–305
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Prevalence and incidence of Q-wave unrecognized myocardial infarction in general population: Diagnostic value of the electrocardiogram. The REGICOR study Rafel Ramos a,b,h,⁎,1, Xavier Albert b,c,d,1, Joan Sala b,c,1, Maria Garcia-Gil a,b,h,1, Roberto Elosua e,1, Jaume Marrugat e,1, Anna Ponjoan a,h,1, María Grau e,1, Manel Morales c,1, Antoni Rubió f,1, Pedro Ortuño g,1, Lia Alves-Cabratosa a,h,1, Ruth Martí-Lluch a,h,1, on behalf of the REGICOR Investigators 2: a
ISV Research Group, Institut Universitari d'Investigació en Atenció Primària Jordi Gol (IDIAP Jordi Gol), Catalunya, Spain Department of Medical Sciences, School of Medicine, University of Girona, Spain c Coronary Unit and Cardiology, Hospital Josep Trueta, Girona, Biomedical Research Institute, Girona (IdIBGi), ICS, Catalunya, Spain d Doctoral Program in Public Health and Biomedical Research Methods, Autonomous University of Barcelona, Spain e Registre Gironí del COR (REGICOR) Group, Cardiovascular, Epidemiology and Genetics Research Group (EGEC), Municipal Institute for Medical Research (IMIM), Barcelona, Spain f Department of Nuclear Medicine, Hospital Josep Trueta, Girona, Biomedical Research Institute, Girona (IdIBGi), ICS, Catalunya, Spain g Department of Diagnostic Radiology, Hospital Josep Trueta, Girona, Biomedical Research Institute, Girona (IdIBGi), ICS, Catalunya, Spain h Girona Biomedical Research Institute (IDIBGI), Catalan Institute of Health (ICS), Girona, Spain b
a r t i c l e
i n f o
Article history: Received 27 June 2016 Received in revised form 1 October 2016 Accepted 4 October 2016 Available online 05 October 2016 Keywords: Myocardial infarction Asymptomatic disease ECG Cardiac imaging techniques Epidemiology
a b s t r a c t Background: Diagnosis of unrecognized myocardial infarction (UMI) remains an open question in epidemiological and clinical studies, inhibiting effective secondary prevention of myocardial infarction. We aimed to determine the prevalence and incidence of Q-wave UMI in asymptomatic individuals aged 35 to 74 years, and to ascertain the positive predictive value (PPV) of asymptomatic Q-wave to diagnose UMI. Methods: Two population-based cross-sectional studies were conducted, in 2000 (with 10-year follow-up) and in 2005. A baseline electrocardiogram was obtained for each participant. Imaging techniques (echocardiography, cardiac magnetic resonance imaging, and myocardial perfusion single-photon emission computerized tomography) were used to confirm UMI in patients with asymptomatic Q-wave. Results: The prevalence of confirmed Q-wave UMI in the 5580 participants was 0.18% (95% confidence interval [CI]: 0.10–0.33) and the incidence rate was 27.1 Q-wave UMI per 100,000 person-years. The proportion of confirmed Q-wave UMI with respect to all prevalent MI was 8.1% (95% CI: 4.4–14.2). The PPV of asymptomatic Qwave to diagnose Q-wave UMI was 29.2% (95% CI: 18.2–43.2%) overall, but much higher (75%, 95% CI: 40.9– 92.9%) in participants with 10-year CHD risk ≥10%, compared to lower-risk participants. Conclusion: Opportunistic identification of asymptomatic Q-waves by routine electrocardiogram overestimates actual Q-wave UMI, which represents 8% to 13% of all myocardial infarction in the population aged 35 to 74 years. This overestimation is particularly high in the population at low cardiovascular risk. In epidemiological studies and in clinical practice, diagnosis of a pathologic Q-wave in asymptomatic patients requires detailed analysis of imaging tests to confirm or rule out myocardial necrosis. © 2016 Published by Elsevier Ireland Ltd.
1. Introduction The diagnosis of unrecognized myocardial infarction (UMI), understood as a myocardial necrosis accompanied by minimal or no coronary
⁎ Corresponding author at: Research Unit, Family Medicine, Carrer Maluquer Salvador, 11, 17003 Girona, Spain. E-mail address: [email protected] (R. Ramos). 1 This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. 2 Full roster of REGICOR Investigators is available at http://www.regicor.org/regicor_inv.
http://dx.doi.org/10.1016/j.ijcard.2016.10.005 0167-5273/© 2016 Published by Elsevier Ireland Ltd.
symptoms, remains an open question in the estimates of ischemic heart disease incidence in epidemiological and clinical studies [1,2]. Given that UMI prognosis is similar to that of recognized myocardial infarction (MI) [3–8] and a substantial subgroup of patients is affected [9], its identification and secondary prevention is an important challenge. Some studies have assumed that the presence of an asymptomatic Q-wave is equivalent to UMI [10]. Other authors have rejected the possibility of classifying UMI due to the difficulties and uncertainties of its diagnosis [11] or because the value of Q-waves to detect UMI was extrapolated from the characteristics of recognized MI in the electrocardiogram (ECG) [12].
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305
If UMI is identified as equivalent to asymptomatic Q-wave, it would represent more than one-fourth of overall MI [9], which could have important implications for study design and interpretation of results [13]. However, the positive predictive value (PPV) of ECG Q-waves to identify a Q-wave UMI in the general population has not been well established; therefore, the actual prevalence and incidence of Q-wave UMI remain unknown. The aims of this study were to determine the prevalence and incidence of Q-wave UMI in asymptomatic individuals aged 35–74 years in whom abnormal Q-waves were observed in a routine ECG and to ascertain the PPV of Q-waves in the diagnosis of UMI confirmed by imaging diagnostic techniques. 2. Methods 2.1. Study design and participants To assess our objectives, we used data from two different studies, selecting participants aged 35 to 74 to achieve a uniform age range. Both studies were part of the Registre Gironi del Cor (REGICOR), a clinical and epidemiological study that has monitored ischemic heart disease and the related risk factors at population scale in Spain since 1978. Both studies were approved by the local ethics committee. All the participants were duly informed and signed their consent. First, to determine the prevalence of asymptomatic Q-wave and actual Q-wave UMI, we used data from a population-based cross-sectional study of individuals aged 35 to 79, recruited from 2002 through 2005 in Girona, a province in Catalunya, northeast Spain; the recruitment response rate was 73.8% [14]. After excluding 658 participants older than 74 years and 114 participants with personal history of previous MI, we analyzed data from 5580 individuals. Second, to determine the population incidence of actual Q-wave UMI, we analyzed a population-based cohort of individuals aged 25 to 74, recruited from 1999 through 2000 and followed up from 200 to 2010, also in Girona province (recruitment response rate: 70%) [14]. After excluding 518 participants younger than 35 years and 31 participants with personal history of previous MI, we analyzed baseline and 10-year follow-up data from 2509 individuals included in the cohort study.
301
2.4. Other covariates Personal history of hypercholesterolemia, diabetes mellitus, arterial hypertension, and smoking were recorded. Body mass index (BMI) was calculated as weight divided by squared height (kg/m2). Blood was drawn after fasting for 10–14 h. The methods and quality control used to determine glycemia, total cholesterol, high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), and triglyceride concentrations are detailed elsewhere [14]. Blood pressure was measured with a calibrated oscillometric sphygmomanometer (OMRON 705 IT). Hypertension was defined when systolic blood pressure was ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg, or when participants reported treatment for hypertension. A continuous Doppler device was used in both studies to measure systolic pressure of the posterior tibial and dorsalis pedis arteries in each leg. Right and left ankle-brachial index (ABI) were calculated as the ratio of the highest systolic pressure in each lower limb to the highest (right or left) brachial systolic pressure. The lowest ABI value was used for analysis. Based on a standardized smoking questionnaire [15], participants were classified as smokers (current smokers or quit b1 year), former smokers (quit ≥1 year), or never smokers. Coronary heart disease (CHD) risk was calculated in all participants (35– 74 years old and free of CVD) using a Framingham function adapted to and validated in the Spanish population [18–19]. 2.5. Estimation of statistical power The available sample of 5580 individuals guaranteed a power of 80% to estimate an expected prevalence of asymptomatic Q-wave of 0.7% with an accuracy of 0.22 percentage units. To determine the population incidence of asymptomatic Q-wave, 20,716 personyears allowed us to estimate an expected annual incidence of 0.1%, with a 95% confidence interval (CI) and an accuracy of 0.04 percentage units, assuming that individuals who presented an incident asymptomatic Q-wave followed a Poisson distribution. 2.6. Statistical analysis Continuous variables were presented as mean and standard deviation or median and interquartile range when their distribution departed from the normal (e.g., glycemia, triglycerides). Prevalences were presented as proportions and incidences as rates with their 95% CI. The PPV of asymptomatic Q-wave to diagnose Q-wave UMI was calculated by dividing the number of actual Q-wave UMIs confirmed in imaging tests by the number of asymptomatic Q-waves in the ECG. Chi-square test was used to compare proportions. A p-value b0.05 was considered statistically significant.
2.2. Electrocardiogram assessment
3. Results Both studies followed the same methods for data collection [14] and used the MONICA protocols [15]. Briefly, examinations were performed by trained nurses and interviewers using standard and validated questionnaires and measurement methods. A standard 12-lead ECG was recorded for each participant in the cross-sectional study and ECG results at baseline and at the follow-up visit were recorded for each participant in the cohort study. For all ECGs, we used a digital electrocardiograph (CARDIOLINE® Delta 60 Plus) with a 50-Hz notch filter and a 40-Hz high-frequency cutoff of the band-pass filter. All the ECGs were interpreted blindly (with no clinical information about the participant) by the same senior cardiologist using the Minnesota Code classification system [16]. The presence of asymptomatic Q-waves in individuals with no history of MI was recorded when ECGs were classified as major Q-wave (Q or QS pattern 1–1–X) or minor Qwave (1–2–X or 1–3–X).
2.3. Imaging techniques To confirm myocardial necrosis, all the participants with major or minor Q-waves in the ECG were invited to undergo a series of imaging tests: echocardiography, cardiac magnetic resonance imaging (MRI), and myocardial perfusion single-photon emission computerized tomography (SPECT). Final classification was established by consensus in an expert committee composed of one expert in each of the imaging techniques and a clinical cardiologist. Echocardiograms were done by a trained cardiologist using a Sonos 5500 echocardiograph (Philips®), aiming to detect the presence of akinetic or dyskinetic segments. With the patient in the supine position, 40 ml of gadolinium (GD-DTPA-BMA) was injected using a five-element phased-array cardiac coil. The cardiac MRI was done using a 1.5 T MRI scanner (Gyroscan Intera, Philips Medical Systems, Best, The Netherlands) with a 25 mT/m gradient system. The presence of subendocardial or transmural late gadolinium enhancement images was considered diagnostic of myocardial necrosis [17]. Myocardial perfusion imaging with SPECT was performed on two different days. On the first day, an ergometric bicycle test was completed. It consisted of 99 m Technetium sestamibi injection (925 MBq) at the maximum peak of exercise, with image acquisition 45 min after the injection. On the second day, the same dose of the radionuclide drug was injected, and images corresponding to the resting phase were acquired. Myocardial perfusion images during the stress and rest phases were then compared. The absence of perfusion at stress and rest, with akinesia was considered diagnostic of myocardial necrosis.
3.1. Prevalence of asymptomatic Q-wave and prevalence of actual Q-wave UMI A diagram showing participant selection for the study of prevalences is presented in Fig. 1 (panel A). Characteristics of the sociodemographic and clinical risk factors of the 5580 participants are shown in Table 1. Participants had a high prevalence of risk factors (hypertension, 42.0%; diabetes, 13.5%; current smoking, 23.7%) but low 10-year CHD risk (mean: 4.0%; 95% CI: 3.9–4.1). Thirty-seven of all the ECGs reviewed had asymptomatic Q-wave (14 major Q-wave, 23 minor Q-wave), a prevalence of 0.67% (95% CI: 0.48–0.91). Of these individuals, 33 (89.2%) agreed to undergo additional imaging diagnostic tests. Q-wave UMI was confirmed in 10 cases (4 major Q-wave, 6 minor Q-wave), a prevalence of 0.18% (95% CI: 0.10 to 0.33). Since 114 participants had clinically recognized MI, the frequency of asymptomatic Q-wave, as a proportion of prevalent MI, was 24.5% (95% CI: 18.3–32.0); but actual Q-wave UMI represented 8.1% (95% CI: 4.4 to 14.2) of all prevalent MI. This percentage was particularly high in participants with 10-year CHD risk ≥ 10% (16.7%; 95% CI: 6.7– 35.8), although the differences did not reach statistical significance. 3.2. Incidence of asymptomatic Q-wave and incidence of actual Q-wave UMI Participant selection for the incidence study is diagrammed in Fig. 1 (panel B). Of 2509 eligible participants, 234 were lost to follow-up (died or moved away). Of the 2275 remaining, 1811 agreed to participate in the follow-up (participation rate: 79.5%). The median (1st quartile– 3rd quartile) follow-up was 7.8 (7.4–8.9) years. Table 1 shows the characteristics of these participants.
302
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305
Fig. 1. Participant selection flowchart.
Individuals who participated in the follow-up were slightly younger, less likely to be current smokers or to have diabetes or hypertension, and had lower cardiovascular risk than those who were lost to followup. These differences were not clinically relevant (data not shown). Seventeen participants showed a new Q-wave in the follow-up ECG (5 major Q-wave, 12 minor Q-wave), representing an accumulated incidence of 0.94% (95% CI: 0.59–1.50) in the 7.8 years median follow-up, or an incidence rate of 115.0 per 100,000 person-years. Of these, 15 individuals (88.2%) agreed to the imaging protocol. Necrosis was confirmed in 4 cases (2 major Q-wave, 2 minor Q-wave), representing an actual Qwave UMI incidence of 0.22% (95% CI: 0.09 to 0.57) in the 7.8 year median follow-up and an incidence rate of 27.1 Q-wave UMI per 100,000 person-years. Since 22 participants had clinically recognized MI, the proportion of Q-wave UMI with respect to all incident MI was 15.4% (95% CI, 6.2 to 33.5).
3.3. Positive predictive value of Q-wave in ECG to detect Q-wave UMI Among the 48 participants with asymptomatic Q-wave who participated in the imaging technique protocol, myocardial necrosis was confirmed in 14 individuals, a 29.2% (95% CI: 18.2–43.2%) positive predictive value (PPV). No other relevant cardiac abnormalities were identified. The characteristics of patients with actual Q-wave UMI compared to those with asymptomatic Q-wave but without myocardial necrosis are shown in Table 2. Patients with Q-wave UMI were more likely to be men and to have a 10-year CHD risk ≥ 10%. Therefore, PPV was higher in men than in women, and especially high in participants with 10-year CHD risk ≥10% (75%; 95% CI: 40.9–92.9%), compared to those with lower risk. A comparison of PPV for all categories of all the other variables did not show any statistically significant differences (Fig. 2).
4. Discussion The prevalence and incidence of actual Q-wave UMI were 3.7 and 4.3 times lower, respectively, than those estimated by the presence of asymptomatic Q-wave in the ECG alone. This finding yielded insufficient PPV to consider asymptomatic Q-wave equivalent to Q-wave UMI, suggesting that ECG alone should not be trusted to detect Q-wave UMI in our population with low cardiovascular risk. In a subpopulation with greater known coronary risk (e.g., men, smokers), the PPV value was higher but still not very reliable. Only in participants with 10-year CHD risk ≥10% did the PPV value of an ECG exceed 50%. Previous studies in which Q-wave UMI was assessed using ECG have reported a proportion of asymptomatic Q-wave (compared to all MI) similar to that observed in our study [9,20–23]; however, the PPV of asymptomatic Q-wave in the ECG to detect Q-wave UMI has not been well reported. In the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) study, 248 randomly chosen 70-year-old participants were recruited and examined with both an ECG and a cardiac MRI. In accordance with our data, the PIVUS study found 10 (4.03%) patients with asymptomatic Q-waves, and Q-wave UMI was confirmed by MRI in only 3 of them (30%, 95% CI: 10.8–60.3). A pathological Q-wave was found in 15 (6.05%) participants, of which 8 presented myocardial scars and only 3 were asymptomatic [24]. In the Reykjavik Study, 47 individuals were found to have an asymptomatic Q-wave in the ECG, and only 19 (40%) were confirmed by the MRI (PPV: 41.3, 95% CI: 28.3–55.6) [3]. These results have both epidemiological and clinical implications. The way we consider Q-wave UMI in epidemiological and clinical studies could have a great impact on epidemiological data, affecting how we identify new MI biomarkers, determine causal associations, perform coronary risk functions, and assess the effectiveness of therapeutic
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305 Table 1 Description of participant characteristics and presence of cardiovascular risk factors in both the prevalence and incidence studies.
N Age a b
Women (%) Hypertension (%) Diabetes (%) Smoking Current or former smoker ≤1 year (%) Former smoker N1 year (%) Never smoker (%) Systolic blood pressurea Diastolic blood pressurea Glycemiab Total cholesterola HDL cholesterola LDL cholesterola Triglycerides mg/dlb Total cholesterol N250 mg/dl HDL cholesterol b46 mg/dl (W)/b40 mg/dl (M) LDL cholesterol N130 mg/dl Triglycerides N200 mg/dl Hypertension treatment Diabetes treatment Lipid-lowering treatment ABI b0.9 10-year CHD risk 10-year CHD risk ≥10% (%)
Prevalence study
Incidence study
5580
1811
54.0 (10.9) 54 (45.0–63.0) 53.6% 42.2% 13.5%
53.1 (10.6) 53 (44.0–63.0) 53.3% 46.1% 10.1%
23.9% 25.2% 51.0% 124.6 (22.7) 77.9 (13.0) 93.0 (85.0–102.0) 209.0 (46.8) 52.4 (13.9) 136.1 (36.4) 93.0 (68.0–131.0) 15.6% 31.9%
23.2% 19.1% 57.7% 132.0 (20.1) 82.2 (10.1) 99.0 (92.0–107.0) 223.9 (41.6) 52.7 (14.7) 149.5 (37,0) 91.0 (69.0–126.0) 22.9% 25.1%
54.8% 7.9% 18.4% 5.9% 10.6% 3.3% 4.0 (3.8) 7.0%
61.5% 5.1% 13.8% 4.7% 6.8% NA 4.2 (3.7) 8.5%
UMI, unrecognized myocardial infarction; HDL, high-density lipoprotein; LDL, low-density lipoprotein; ABI, ankle-brachial index; CHD, coronary heart disease. a Mean (standard deviation). b Median (1st quartile–3rd quartile).
interventions. Our findings indicate that asymptomatic Q-waves found in the ECG cannot be directly considered as Q-wave UMI in clinical trials or epidemiologic studies, as has been suggested [13]; however, we cannot ignore them because of the noteworthy proportion of total MI classified as Q-wave UMI [9] and the similarity between the clinical Table 2 Characteristics of patients with confirmed Q-wave unrecognized myocardial infarction, compared to those with asymptomatic Q-wave but without myocardial necrosis.
N Agea Women (%) Hypertension (%) Diabetes (%) Smoking Current or former smoker ≤1 year (%) Former smoker N1 year (%) Never smoker (%) Systolic blood pressurea Diastolic blood pressurea Glycemiab Total cholesterola HDL cholesterola LDL cholesterola Triglycerides mg/dlb 10-year CHD risk ≥10% (%) Major Q-wave Q-wave location on the ECG Anterior Lateral Inferior
UMI
No UMI
p value
14 61.3 (9.4) 7.1% 71.4% 35.7%
34 60.9 (10.5) 38.2% 67.6% 29.4%
0.91 0.03 0.79 0.67
28.6%
8.8%
28.6% 42.8% 142.4 (24.2) 86.7 (10.8) 113.5 (96.7–135.0) 216.1 (42.6) 43.2 (12.9) 143.4 (36.7) 110.0 (91.3–225.0) 42.9% 42.9%
29.4% 61.8% 140.2 (25.9) 81.6 (12.9) 96.0 (91.0–116.0) 206.6 (40.6) 50.9 (15.0) 133.5 (34.3) 100.5 (76.8–130.3) 5.9% 29.4%
14.3% 7.1% 78.6%
8.8% 29.4% 61.8%
303
implications and prognosis of Q-wave UMI and those of symptomatic MI [3–8]. Furthermore, non-Q-wave UMI, known to be a large proportion of all UMIs, is not detected by ECG. In the PIVUS study, cardiac MRI identified UMI in 19.8% of the elderly participants, and 80% of them had no Q-waves on the ECG [24]. In another study, the prevalence of non-Q-wave UMI in patients with suspected coronary artery disease was 27.0% (95% CI: 21.1–33.8), compared to the 8% (95% CI: 5.0–12.9) prevalence of Q-wave UMI [25]. In the Multi-Ethnic Study of Atherosclerosis (MESA), cardiac MRI identified UMI in 7.9% of the participants, 78% of them were not detectable by ECG or clinical evaluation [26]. Our data clearly reinforce the need for an imaging study to discard a false positive when an asymptomatic Q-wave is detected by ECG in clinical practice. This need is all the more urgent because UMI has such important therapeutic and prognostic implications for the individual: it increases the risk of death [3], atrial fibrillation [4], stroke [5], heart failure [6], and new episodes of MI [7]. In addition, cardiac necrosis cannot be confirmed or discarded on the basis of individual concomitant risk factors when an asymptomatic Qwave is found in an apparently healthy individual. But as previously described [27], our results show that 10-year CHD risk is markedly associated with the presence of UMI. Then, only male sex or 10-year CHD risk ≥10% justifies increased urgency in the diagnostic process; once again, an imaging study could contribute critical information needed to start preventive treatment immediately in these patients. One surprising result was the lack of association between the presence of major Q-wave and higher probability of UMI. A possible explanation is the higher proportion of major Q-wave ECG in the group with UMI, which had limited statistical power due to the sample size. It is likely that sample size also explains the failure to confirm our hypothesis that pathological ABI would be associated with higher PPV, compared to normal ABI. The incorporation of imaging techniques into clinical studies appears to be the best option to assess UMI. Along this line, delayed enhancement cardiac MRI showed 99% sensitivity to detect Q-wave MI, and 91% to detect non-Q-wave MI [28]. The use of imaging techniques could contribute several benefits: a) reduction of sample size, duration, and ethical concerns in clinical trials, avoiding unnecessary prolonged patient exposure to potential harms from experimental therapies; b) the opportunity to exclude patients with asymptomatic MI from primary prevention studies; c) the potential to assess a therapy's effectiveness in reducing UMI in intervention studies; and d) potential improvement in the efficient use of research funding resources [13]. 4.1. Study limitations
0.10 0.78 0.17 0.47 0.08 0.37 b0.01 0.36
0.24
HDL, high-density lipoprotein; LDL, low-density lipoprotein; CHD, coronary heart disease; ECG, electrocardiogram. a Mean (standard deviation). b Median (1st quartile–3rd quartile).
In observational studies, the response rate in recruitment and willingness to participate in follow-up could seriously affect the study population and introduce selection bias. In the present study, the participation rate was high in both contexts (73.8% in the cross-sectional study and 70% in the follow-up cohort study), which guaranteed external validity and a population representative of the study area. A limitation is the small number of asymptomatic Q-waves observed in our population, which limited the power to detect statistically significant differences between subgroups. Another limitation could be the 40-Hz high-frequency cutoff used in the ECG recording, as this might eliminate high-frequency signals that could be important from a clinical point of view [29]. However, a recent study [30] concludes that a 40-Hz high-frequency cutoff of the band-pass filtering yields a higher percentage of optimal quality ECGs and fewer non-interpretable traces, and allows detection of the same percentage of significant Q-waves (suggestive of myocardial necrosis), compared to the 150-Hz filtering. No inter-observer analysis was done for REGICOR ECG performance, but all ECGs were performed by a small team of highly experienced and trained nurses, thereby minimizing any variability. Subjectivity in ECG interpretation by only one cardiologist was another potential limitation. However, interpretation by experienced cardiologists has been demonstrated to be the gold standard [31]. In our
304
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305
Fig. 2. Positive predictive value of asymptomatic Q-wave in the electrocardiogram to diagnose necrosis-confirmed unrecognized myocardial infarction, by participant characteristics. PPV: positive predictive value; ABI: ankle-brachial index; CHD: coronary heart disease; LDL: low density lipoproteins; HDL: high density lipoproteins.
study, all ECGs were interpreted by the same senior cardiologist, who has done all ECG interpretation for the REGICOR study for 35 years. The final classification by expert consensus also has the potential to introduce some subjectivity, but could also be considered a strength of the study. The participation in this process by experts in each of the imaging techniques and a clinical cardiologist allowed us to take advantage of all
available information and to mirror the care process used in clinical practice. Furthermore, we did not include participants with a normal ECG in this study and therefore could not assess the sensitivity, specificity, or negative predictive value of ECG to detect Q-wave UMI. Finally, a significant proportion of UMIs could present as non-Q-wave and were not identified in this study [3,24,25].
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305
5. Conclusions In the population aged 35–74 years, Q-waves identified opportunistically in a routine ECG overestimated the actual Q-wave UMI prevalence, which was shown to be 8% to 13% of all MI. This overestimation questions the PPV of this finding in the general population. Only in participants with a high 10-year CHD risk did the PPV exceed 50%. Therefore, identifying a pathologic Q-wave in asymptomatic patients requires confirmation by imaging tests, both in clinical practice and in epidemiological studies, to confirm or rule out myocardial necrosis. Funding sources This project was supported by clinical research grants from the Ministerio de Salud (PI20471); Spain's Ministry of Science and Innovation through the Carlos III Health Institute, co-financed with European Union ERDF funds (Network for Prevention and Health Promotion in primary Care) [RedIAPP RD12/0005]; the Red de Investigación Cardiovascular [RD12/0042/0061, RD12/0042/0013], and Miguel Servet Contract (CP12/03287); and by the Departament de Salut, Generalitat de Catalunya, Agency for Health Technology Assessment [AATRM 034/33/ 02; AATRM 078/28/06] and Agency for Management of University and Research Grants [2014 SGR 240; 2014 SGR 902]. Conflict of interest No author has reported any potential conflicts of interest involving the work under consideration for publication. Acknowledgments The authors are grateful to participants in the REGICOR and HERMES studies and to Susana Tello, Martina Sidera and Isabel Ramió for field work. The authors also appreciate the revision of the English text by Elaine Lilly, Ph.D., of Writer's First Aid. References [1] M.J. Leening, M. Kavousi, J. Heeringa, F.J. van Rooij, J. Verkroost-van Heemst, J.W. Deckers, F.U. Mattace-Raso, G. Ziere, A. Hofman, B.H. Stricker, J.C. Witteman, Methods of data collection and definitions of cardiac outcomes in the Rotterdam study, Eur. J. Epidemiol. 27 (2012) 173–185. [2] A. Linnersjö, N. Hammar, A. Gustavsson, C. Reuterwall, Recent time trends in acute myocardial infarction in Stockholm, Sweden, Int. J. Cardiol. 76 (2000) 17–21. [3] E.B. Schelbert, J.J. Cao, S. Sigurdsson, T. Aspelund, P. Kellman, A.H. Aletras, C.K. Dyke, G. Thorgeirsson, G. Eiriksdottir, L.J. Launer, V. Gudnason, T.B. Harris, A.E. Arai, Prevalence and prognosis of unrecognized myocardial infarction determined by cardiac magnetic resonance in older adults, JAMA 308 (2012) 890–896. [4] B.P. Krijthe, M.J. Leening, J. Heeringa, J.A. Kors, A. Hofman, O.H. Franco, J.C. Witteman, B.H. Stricker, Unrecognized myocardial infarction and risk of atrial fibrillation: the Rotterdam study, Int. J. Cardiol. 168 (2013) 1453–1457. [5] M.A. Ikram, M. Hollander, M.J. Bos, et al., Unrecognized myocardial infarction and the risk of stroke: the Rotterdam study, Neurology 67 (2006) 1635–1639. [6] M.J. Leening, S.E. Elias-Smale, J.F. Felix, J.A. Kors, J.W. Deckers, A. Hofman, B.H. Stricker, J.C. Witteman, Unrecognised myocardial infarction and long-term risk of heart failure in the elderly: the Rotterdam study, Heart 96 (2010) 1458e1462. [7] R.Y. Kwong, H. Sattar, H. Wu, G. Vorobiof, V. Gandla, K. Steel, S. Siu, K.A. Brown, Incidence and prognostic implication of unrecognized myocardial scar characterized by cardiac magnetic resonance in diabetic patients without clinical evidence of myocardial infarction, Circulation 118 (2008) 1011e1020. [8] A. Dehghan, M.J. Leening, A.M. Solouki, E. Boersma, J.W. Deckers, G. van Her pen, J. Heeringa, A. Hofman, K. JA, O.H. Franco, M.A. Ikram, J.C. Witteman, Comparison of prognosis in unrecognized versus recognized myocardial infarction in men versus women N55 years of age (from the Rotterdam study), Am. J. Cardiol. 113 (2014) 1–6. [9] S.E. Sheifer, T.A. Manolio, B.J. Gersh, Unrecognized myocardial infarction, Ann. Intern. Med. 135 (2001) 801–811.
305
[10] J. Herlitz, M. Dellborg, T. Karlsson, M.H. Evander, A. Berger, R. Luepker, Epidemiology of acute myocardial infarction with the emphasis on patients who did not reach the coronary care unit and non-AMI admissions, Int. J. Cardiol. 128 (2008) 342–349. [11] D. Tsiachris, C. Tsioufis, C. Thomopoulos, D. Syrseloudis, V. Antonakis, L. Lioni, I. Kallikazaros, T. Makris, V. Papademetriou, C.I. Stefanadis, New-onset diabetes and cardiovascular events in essential hypertensives: a 6-year follow-up study, Int. J. Cardiol. 153 (2011) 154–158. [12] K.A. Ammar, J.A. Kors, B.P. Yawn, R.J. Rodeheffer, Defining unrecognized myocardial infarction: a call for standardized electrocardiographic diagnostic criteria, Am. Heart J. 148 (2004) 277e284. [13] Y.B. Pride, B.J. Piccirillo, C.M. Gibson, Prevalence, consequences, and implications for clinical trials of unrecognized myocardial infarction, Am. J. Cardiol. 111 (2013) 914–918. [14] M. Grau, I. Subirana, R. Elosua, P. Solanas, R. Ramos, R. Masiá, F. Cordón, J. Sala, D. Juvinyà, C. Cerezo, M. Fitó, J. Vila, M.I. Covas, J. Marrugat, Trends in cardiovascular risk factor prevalence (1995–2000–2005) in northeastern Spain, Eur. J. Cardiovasc. Prev. Rehabil. 14 (2007) 653–659. [15] Manual of the MONICA Project, World Health Organisation, Geneva, 2000 (http:// www.thl.fi/publications/monica/manual/index.htm, (accessed 27.06.16)). [16] P.W. Macfarlane, Minnesota coding and the prevalence of ECG abnormalities, Heart 84 (2000) 582–584. [17] H. Thiele, M.J.E. Kappl, S. Conradi, J. Niebauer, R. Hambrecht, G. Schuler, Reproducibility of chronic and acute infarct size measurement by delayed enhancementmagnetic resonance imaging, J. Am. Coll. Cardiol. 47 (2006) 1641–1645. [18] J. Marrugat, I. Subirana, E. Comín, C. Cabezas, J. Vila, R. Elosua, et al., for the VERIFICA Investigators, Validity of an adaptation of the Framingham cardiovascular risk function: the VERIFICA study, J. Epidemiol. Community Health 61 (2007) 40–47. [19] J. Marrugat, R. D'Agostino, L. Sullivan, R. Elosua, P. Wilson, J. Ordovas, et al., An adaptation of the Framingham coronary heart disease risk function to European Mediterranean areas, J. Epidemiol. Community Health 57 (2003) 634–638. [20] A. de Torbal, E. Boersma, J.A. Kors, G. van Herpen, J.W. Deckers, D.A. van der Kuip, B.H. Stricker, A. Hofman, J.C. Witteman, Incidence of recognized and unrecognized myocardial infarction in men and women aged 55 and older: the Rotterdam study, Eur. Heart J. 27 (2006) 729–736. [21] M.R. MacDonald, M.C. Petrie, P.D. Home, M. Komajda, N.P. Jones, H. Beck-Nielsen, R. Gomis, M. Hanefeld, S.J. Pocock, P.S. Curtis, J.J. McMurray, Incidence and prevalence of unrecognized myocardial infarction in people with diabetes: a substudy of the Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycemia in Diabetes (RECORD) study, Diabetes Care 34 (2011) 1394–1396. [22] S.E. Sheifer, B.J. Gersh, N.D. Yanez III, P.A. Ades, G.L. Burke, T.A. Manolio, Prevalence, predisposing factors, and prognosis of clinically unrecognized myocardial infarction in the elderly, J. Am. Coll. Cardiol. 35 (2000) 119–126. [23] L.L. Boland, A.R. Folsom, P.D. Sorlie, H.A. Taylor, W.D. Rosamond, L.E. Chambless, L.S. Cooper, Occurrence of unrecognized myocardial infarction in subjects aged 45 to 65 years (the ARIC study), Am. J. Cardiol. 90 (2002) 927–931. [24] C.E. Barbier, T. Bjerner, L. Johansson, L. Lind, H. Ahlström, Myocardial scars more frequent than expected: magnetic resonance imaging detects potential risk group, J. Am. Coll. Cardiol. 48 (2006) 765–771. [25] H.W. Kim, I. Klem, D.J. Shah, E. Wu, S.N. Meyers, M.A. Parker, A.L. Crowley, R.O. Bonow, R.M. Judd, R.J. Kim, Unrecognized nonQ-wave myocardial infarction: prevalence and prognostic significance in patients with suspected coronary disease, PLoS Med. 6 (2009) e1000057. [26] E.B. Turkbey, M.S. Nacif, M. Guo, R.L. McClelland, P.B. Teixeira, D.E. Bild, R.G. Barr, S. Shea, W. Post, G. Burke, M.J. Budoff, A.R. Folsom, C.Y. Liu, J.A. Lima, D.A. Bluemke, Prevalence and correlates of myocardial scar in a US cohort, JAMA 314 (2015) 1945–1954. [27] D. McAreavey, J.S. Vidal, T. Aspelund, G. Eiriksdottir, E.B. Schelbert, O. Kjartansson, J.J. Cao, G. Thorgeirsson, S. Sigurdsson, M. Garcia, T.B. Harris, L.J. Launer, V. Gudnason, A.E. Arai, Midlife cardiovascular risk factors and late-life unrecognized and recognized myocardial infarction detect by cardiac magnetic resonance: ICELAND-MI, the AGES–Reykjavik study, J. Am. Heart Assoc. 5 (2016) e002420. [28] R.J. Kim, T.S. Albert, J.H. Wible, M.D. Elliott, J.C. Allen, J.C. Lee, M. Parker, A. Napoli, R.M. Judd, Gadoversetamide Myocardial Infarction Imaging Investigators, Performance of delayed-enhancement magnetic resonance imaging with gadoversetamide contrast for the detection and assessment of myocardial infarction: an international, multicenter, double-blinded, randomized trial, Circulation 117 (2008) 629–637. [29] J. García-Niebla, P. Llontop-García, J.I. Valle-Racero, G. Serra-Autonell, V.N. Batchvarov, A. Bayés de Luna, Technical mistakes during the acquisition of the electrocardiogram, Ann. Noninvasive Electrocardiol. 14 (2009) 389–403. [30] D. Ricciardi, I. Cavallari, A. Creta, G. Di Giovanni, V. Calabrese, N. Di Belardino, et al., Impact of the high-frequency cutoff of bandpass filtering on ECG quality and clinical interpretation: a comparison between 40 Hz and 150 Hz cutoff in a surgical preoperative adult outpatient population, J. Electrocardiol. 49 (2016) 691–695. [31] M.C. de Bruyne, J.A. Kors, A.W. Hoes, D.A. Kruijssen, J.W. Deckers, M. Grosfeld, G. van Herpen, D.E. Grobbee, J.H. van Bemmel, Diagnostic interpretation of electrocardiograms in population-based research: computer program research physicians, or cardiologists? J. Clin. Epidemiol. 50 (1997) 947–952.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Prevalence and incidence of Q-wave unrecognized myocardial infarction in general population: Diagnostic value of the electrocardiogram. The REGICOR study Rafel Ramos a,b,h,⁎,1, Xavier Albert b,c,d,1, Joan Sala b,c,1, Maria Garcia-Gil a,b,h,1, Roberto Elosua e,1, Jaume Marrugat e,1, Anna Ponjoan a,h,1, María Grau e,1, Manel Morales c,1, Antoni Rubió f,1, Pedro Ortuño g,1, Lia Alves-Cabratosa a,h,1, Ruth Martí-Lluch a,h,1, on behalf of the REGICOR Investigators 2: a
ISV Research Group, Institut Universitari d'Investigació en Atenció Primària Jordi Gol (IDIAP Jordi Gol), Catalunya, Spain Department of Medical Sciences, School of Medicine, University of Girona, Spain c Coronary Unit and Cardiology, Hospital Josep Trueta, Girona, Biomedical Research Institute, Girona (IdIBGi), ICS, Catalunya, Spain d Doctoral Program in Public Health and Biomedical Research Methods, Autonomous University of Barcelona, Spain e Registre Gironí del COR (REGICOR) Group, Cardiovascular, Epidemiology and Genetics Research Group (EGEC), Municipal Institute for Medical Research (IMIM), Barcelona, Spain f Department of Nuclear Medicine, Hospital Josep Trueta, Girona, Biomedical Research Institute, Girona (IdIBGi), ICS, Catalunya, Spain g Department of Diagnostic Radiology, Hospital Josep Trueta, Girona, Biomedical Research Institute, Girona (IdIBGi), ICS, Catalunya, Spain h Girona Biomedical Research Institute (IDIBGI), Catalan Institute of Health (ICS), Girona, Spain b
a r t i c l e
i n f o
Article history: Received 27 June 2016 Received in revised form 1 October 2016 Accepted 4 October 2016 Available online 05 October 2016 Keywords: Myocardial infarction Asymptomatic disease ECG Cardiac imaging techniques Epidemiology
a b s t r a c t Background: Diagnosis of unrecognized myocardial infarction (UMI) remains an open question in epidemiological and clinical studies, inhibiting effective secondary prevention of myocardial infarction. We aimed to determine the prevalence and incidence of Q-wave UMI in asymptomatic individuals aged 35 to 74 years, and to ascertain the positive predictive value (PPV) of asymptomatic Q-wave to diagnose UMI. Methods: Two population-based cross-sectional studies were conducted, in 2000 (with 10-year follow-up) and in 2005. A baseline electrocardiogram was obtained for each participant. Imaging techniques (echocardiography, cardiac magnetic resonance imaging, and myocardial perfusion single-photon emission computerized tomography) were used to confirm UMI in patients with asymptomatic Q-wave. Results: The prevalence of confirmed Q-wave UMI in the 5580 participants was 0.18% (95% confidence interval [CI]: 0.10–0.33) and the incidence rate was 27.1 Q-wave UMI per 100,000 person-years. The proportion of confirmed Q-wave UMI with respect to all prevalent MI was 8.1% (95% CI: 4.4–14.2). The PPV of asymptomatic Qwave to diagnose Q-wave UMI was 29.2% (95% CI: 18.2–43.2%) overall, but much higher (75%, 95% CI: 40.9– 92.9%) in participants with 10-year CHD risk ≥10%, compared to lower-risk participants. Conclusion: Opportunistic identification of asymptomatic Q-waves by routine electrocardiogram overestimates actual Q-wave UMI, which represents 8% to 13% of all myocardial infarction in the population aged 35 to 74 years. This overestimation is particularly high in the population at low cardiovascular risk. In epidemiological studies and in clinical practice, diagnosis of a pathologic Q-wave in asymptomatic patients requires detailed analysis of imaging tests to confirm or rule out myocardial necrosis. © 2016 Published by Elsevier Ireland Ltd.
1. Introduction The diagnosis of unrecognized myocardial infarction (UMI), understood as a myocardial necrosis accompanied by minimal or no coronary
⁎ Corresponding author at: Research Unit, Family Medicine, Carrer Maluquer Salvador, 11, 17003 Girona, Spain. E-mail address: [email protected] (R. Ramos). 1 This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. 2 Full roster of REGICOR Investigators is available at http://www.regicor.org/regicor_inv.
http://dx.doi.org/10.1016/j.ijcard.2016.10.005 0167-5273/© 2016 Published by Elsevier Ireland Ltd.
symptoms, remains an open question in the estimates of ischemic heart disease incidence in epidemiological and clinical studies [1,2]. Given that UMI prognosis is similar to that of recognized myocardial infarction (MI) [3–8] and a substantial subgroup of patients is affected [9], its identification and secondary prevention is an important challenge. Some studies have assumed that the presence of an asymptomatic Q-wave is equivalent to UMI [10]. Other authors have rejected the possibility of classifying UMI due to the difficulties and uncertainties of its diagnosis [11] or because the value of Q-waves to detect UMI was extrapolated from the characteristics of recognized MI in the electrocardiogram (ECG) [12].
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305
If UMI is identified as equivalent to asymptomatic Q-wave, it would represent more than one-fourth of overall MI [9], which could have important implications for study design and interpretation of results [13]. However, the positive predictive value (PPV) of ECG Q-waves to identify a Q-wave UMI in the general population has not been well established; therefore, the actual prevalence and incidence of Q-wave UMI remain unknown. The aims of this study were to determine the prevalence and incidence of Q-wave UMI in asymptomatic individuals aged 35–74 years in whom abnormal Q-waves were observed in a routine ECG and to ascertain the PPV of Q-waves in the diagnosis of UMI confirmed by imaging diagnostic techniques. 2. Methods 2.1. Study design and participants To assess our objectives, we used data from two different studies, selecting participants aged 35 to 74 to achieve a uniform age range. Both studies were part of the Registre Gironi del Cor (REGICOR), a clinical and epidemiological study that has monitored ischemic heart disease and the related risk factors at population scale in Spain since 1978. Both studies were approved by the local ethics committee. All the participants were duly informed and signed their consent. First, to determine the prevalence of asymptomatic Q-wave and actual Q-wave UMI, we used data from a population-based cross-sectional study of individuals aged 35 to 79, recruited from 2002 through 2005 in Girona, a province in Catalunya, northeast Spain; the recruitment response rate was 73.8% [14]. After excluding 658 participants older than 74 years and 114 participants with personal history of previous MI, we analyzed data from 5580 individuals. Second, to determine the population incidence of actual Q-wave UMI, we analyzed a population-based cohort of individuals aged 25 to 74, recruited from 1999 through 2000 and followed up from 200 to 2010, also in Girona province (recruitment response rate: 70%) [14]. After excluding 518 participants younger than 35 years and 31 participants with personal history of previous MI, we analyzed baseline and 10-year follow-up data from 2509 individuals included in the cohort study.
301
2.4. Other covariates Personal history of hypercholesterolemia, diabetes mellitus, arterial hypertension, and smoking were recorded. Body mass index (BMI) was calculated as weight divided by squared height (kg/m2). Blood was drawn after fasting for 10–14 h. The methods and quality control used to determine glycemia, total cholesterol, high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), and triglyceride concentrations are detailed elsewhere [14]. Blood pressure was measured with a calibrated oscillometric sphygmomanometer (OMRON 705 IT). Hypertension was defined when systolic blood pressure was ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg, or when participants reported treatment for hypertension. A continuous Doppler device was used in both studies to measure systolic pressure of the posterior tibial and dorsalis pedis arteries in each leg. Right and left ankle-brachial index (ABI) were calculated as the ratio of the highest systolic pressure in each lower limb to the highest (right or left) brachial systolic pressure. The lowest ABI value was used for analysis. Based on a standardized smoking questionnaire [15], participants were classified as smokers (current smokers or quit b1 year), former smokers (quit ≥1 year), or never smokers. Coronary heart disease (CHD) risk was calculated in all participants (35– 74 years old and free of CVD) using a Framingham function adapted to and validated in the Spanish population [18–19]. 2.5. Estimation of statistical power The available sample of 5580 individuals guaranteed a power of 80% to estimate an expected prevalence of asymptomatic Q-wave of 0.7% with an accuracy of 0.22 percentage units. To determine the population incidence of asymptomatic Q-wave, 20,716 personyears allowed us to estimate an expected annual incidence of 0.1%, with a 95% confidence interval (CI) and an accuracy of 0.04 percentage units, assuming that individuals who presented an incident asymptomatic Q-wave followed a Poisson distribution. 2.6. Statistical analysis Continuous variables were presented as mean and standard deviation or median and interquartile range when their distribution departed from the normal (e.g., glycemia, triglycerides). Prevalences were presented as proportions and incidences as rates with their 95% CI. The PPV of asymptomatic Q-wave to diagnose Q-wave UMI was calculated by dividing the number of actual Q-wave UMIs confirmed in imaging tests by the number of asymptomatic Q-waves in the ECG. Chi-square test was used to compare proportions. A p-value b0.05 was considered statistically significant.
2.2. Electrocardiogram assessment
3. Results Both studies followed the same methods for data collection [14] and used the MONICA protocols [15]. Briefly, examinations were performed by trained nurses and interviewers using standard and validated questionnaires and measurement methods. A standard 12-lead ECG was recorded for each participant in the cross-sectional study and ECG results at baseline and at the follow-up visit were recorded for each participant in the cohort study. For all ECGs, we used a digital electrocardiograph (CARDIOLINE® Delta 60 Plus) with a 50-Hz notch filter and a 40-Hz high-frequency cutoff of the band-pass filter. All the ECGs were interpreted blindly (with no clinical information about the participant) by the same senior cardiologist using the Minnesota Code classification system [16]. The presence of asymptomatic Q-waves in individuals with no history of MI was recorded when ECGs were classified as major Q-wave (Q or QS pattern 1–1–X) or minor Qwave (1–2–X or 1–3–X).
2.3. Imaging techniques To confirm myocardial necrosis, all the participants with major or minor Q-waves in the ECG were invited to undergo a series of imaging tests: echocardiography, cardiac magnetic resonance imaging (MRI), and myocardial perfusion single-photon emission computerized tomography (SPECT). Final classification was established by consensus in an expert committee composed of one expert in each of the imaging techniques and a clinical cardiologist. Echocardiograms were done by a trained cardiologist using a Sonos 5500 echocardiograph (Philips®), aiming to detect the presence of akinetic or dyskinetic segments. With the patient in the supine position, 40 ml of gadolinium (GD-DTPA-BMA) was injected using a five-element phased-array cardiac coil. The cardiac MRI was done using a 1.5 T MRI scanner (Gyroscan Intera, Philips Medical Systems, Best, The Netherlands) with a 25 mT/m gradient system. The presence of subendocardial or transmural late gadolinium enhancement images was considered diagnostic of myocardial necrosis [17]. Myocardial perfusion imaging with SPECT was performed on two different days. On the first day, an ergometric bicycle test was completed. It consisted of 99 m Technetium sestamibi injection (925 MBq) at the maximum peak of exercise, with image acquisition 45 min after the injection. On the second day, the same dose of the radionuclide drug was injected, and images corresponding to the resting phase were acquired. Myocardial perfusion images during the stress and rest phases were then compared. The absence of perfusion at stress and rest, with akinesia was considered diagnostic of myocardial necrosis.
3.1. Prevalence of asymptomatic Q-wave and prevalence of actual Q-wave UMI A diagram showing participant selection for the study of prevalences is presented in Fig. 1 (panel A). Characteristics of the sociodemographic and clinical risk factors of the 5580 participants are shown in Table 1. Participants had a high prevalence of risk factors (hypertension, 42.0%; diabetes, 13.5%; current smoking, 23.7%) but low 10-year CHD risk (mean: 4.0%; 95% CI: 3.9–4.1). Thirty-seven of all the ECGs reviewed had asymptomatic Q-wave (14 major Q-wave, 23 minor Q-wave), a prevalence of 0.67% (95% CI: 0.48–0.91). Of these individuals, 33 (89.2%) agreed to undergo additional imaging diagnostic tests. Q-wave UMI was confirmed in 10 cases (4 major Q-wave, 6 minor Q-wave), a prevalence of 0.18% (95% CI: 0.10 to 0.33). Since 114 participants had clinically recognized MI, the frequency of asymptomatic Q-wave, as a proportion of prevalent MI, was 24.5% (95% CI: 18.3–32.0); but actual Q-wave UMI represented 8.1% (95% CI: 4.4 to 14.2) of all prevalent MI. This percentage was particularly high in participants with 10-year CHD risk ≥ 10% (16.7%; 95% CI: 6.7– 35.8), although the differences did not reach statistical significance. 3.2. Incidence of asymptomatic Q-wave and incidence of actual Q-wave UMI Participant selection for the incidence study is diagrammed in Fig. 1 (panel B). Of 2509 eligible participants, 234 were lost to follow-up (died or moved away). Of the 2275 remaining, 1811 agreed to participate in the follow-up (participation rate: 79.5%). The median (1st quartile– 3rd quartile) follow-up was 7.8 (7.4–8.9) years. Table 1 shows the characteristics of these participants.
302
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305
Fig. 1. Participant selection flowchart.
Individuals who participated in the follow-up were slightly younger, less likely to be current smokers or to have diabetes or hypertension, and had lower cardiovascular risk than those who were lost to followup. These differences were not clinically relevant (data not shown). Seventeen participants showed a new Q-wave in the follow-up ECG (5 major Q-wave, 12 minor Q-wave), representing an accumulated incidence of 0.94% (95% CI: 0.59–1.50) in the 7.8 years median follow-up, or an incidence rate of 115.0 per 100,000 person-years. Of these, 15 individuals (88.2%) agreed to the imaging protocol. Necrosis was confirmed in 4 cases (2 major Q-wave, 2 minor Q-wave), representing an actual Qwave UMI incidence of 0.22% (95% CI: 0.09 to 0.57) in the 7.8 year median follow-up and an incidence rate of 27.1 Q-wave UMI per 100,000 person-years. Since 22 participants had clinically recognized MI, the proportion of Q-wave UMI with respect to all incident MI was 15.4% (95% CI, 6.2 to 33.5).
3.3. Positive predictive value of Q-wave in ECG to detect Q-wave UMI Among the 48 participants with asymptomatic Q-wave who participated in the imaging technique protocol, myocardial necrosis was confirmed in 14 individuals, a 29.2% (95% CI: 18.2–43.2%) positive predictive value (PPV). No other relevant cardiac abnormalities were identified. The characteristics of patients with actual Q-wave UMI compared to those with asymptomatic Q-wave but without myocardial necrosis are shown in Table 2. Patients with Q-wave UMI were more likely to be men and to have a 10-year CHD risk ≥ 10%. Therefore, PPV was higher in men than in women, and especially high in participants with 10-year CHD risk ≥10% (75%; 95% CI: 40.9–92.9%), compared to those with lower risk. A comparison of PPV for all categories of all the other variables did not show any statistically significant differences (Fig. 2).
4. Discussion The prevalence and incidence of actual Q-wave UMI were 3.7 and 4.3 times lower, respectively, than those estimated by the presence of asymptomatic Q-wave in the ECG alone. This finding yielded insufficient PPV to consider asymptomatic Q-wave equivalent to Q-wave UMI, suggesting that ECG alone should not be trusted to detect Q-wave UMI in our population with low cardiovascular risk. In a subpopulation with greater known coronary risk (e.g., men, smokers), the PPV value was higher but still not very reliable. Only in participants with 10-year CHD risk ≥10% did the PPV value of an ECG exceed 50%. Previous studies in which Q-wave UMI was assessed using ECG have reported a proportion of asymptomatic Q-wave (compared to all MI) similar to that observed in our study [9,20–23]; however, the PPV of asymptomatic Q-wave in the ECG to detect Q-wave UMI has not been well reported. In the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS) study, 248 randomly chosen 70-year-old participants were recruited and examined with both an ECG and a cardiac MRI. In accordance with our data, the PIVUS study found 10 (4.03%) patients with asymptomatic Q-waves, and Q-wave UMI was confirmed by MRI in only 3 of them (30%, 95% CI: 10.8–60.3). A pathological Q-wave was found in 15 (6.05%) participants, of which 8 presented myocardial scars and only 3 were asymptomatic [24]. In the Reykjavik Study, 47 individuals were found to have an asymptomatic Q-wave in the ECG, and only 19 (40%) were confirmed by the MRI (PPV: 41.3, 95% CI: 28.3–55.6) [3]. These results have both epidemiological and clinical implications. The way we consider Q-wave UMI in epidemiological and clinical studies could have a great impact on epidemiological data, affecting how we identify new MI biomarkers, determine causal associations, perform coronary risk functions, and assess the effectiveness of therapeutic
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305 Table 1 Description of participant characteristics and presence of cardiovascular risk factors in both the prevalence and incidence studies.
N Age a b
Women (%) Hypertension (%) Diabetes (%) Smoking Current or former smoker ≤1 year (%) Former smoker N1 year (%) Never smoker (%) Systolic blood pressurea Diastolic blood pressurea Glycemiab Total cholesterola HDL cholesterola LDL cholesterola Triglycerides mg/dlb Total cholesterol N250 mg/dl HDL cholesterol b46 mg/dl (W)/b40 mg/dl (M) LDL cholesterol N130 mg/dl Triglycerides N200 mg/dl Hypertension treatment Diabetes treatment Lipid-lowering treatment ABI b0.9 10-year CHD risk 10-year CHD risk ≥10% (%)
Prevalence study
Incidence study
5580
1811
54.0 (10.9) 54 (45.0–63.0) 53.6% 42.2% 13.5%
53.1 (10.6) 53 (44.0–63.0) 53.3% 46.1% 10.1%
23.9% 25.2% 51.0% 124.6 (22.7) 77.9 (13.0) 93.0 (85.0–102.0) 209.0 (46.8) 52.4 (13.9) 136.1 (36.4) 93.0 (68.0–131.0) 15.6% 31.9%
23.2% 19.1% 57.7% 132.0 (20.1) 82.2 (10.1) 99.0 (92.0–107.0) 223.9 (41.6) 52.7 (14.7) 149.5 (37,0) 91.0 (69.0–126.0) 22.9% 25.1%
54.8% 7.9% 18.4% 5.9% 10.6% 3.3% 4.0 (3.8) 7.0%
61.5% 5.1% 13.8% 4.7% 6.8% NA 4.2 (3.7) 8.5%
UMI, unrecognized myocardial infarction; HDL, high-density lipoprotein; LDL, low-density lipoprotein; ABI, ankle-brachial index; CHD, coronary heart disease. a Mean (standard deviation). b Median (1st quartile–3rd quartile).
interventions. Our findings indicate that asymptomatic Q-waves found in the ECG cannot be directly considered as Q-wave UMI in clinical trials or epidemiologic studies, as has been suggested [13]; however, we cannot ignore them because of the noteworthy proportion of total MI classified as Q-wave UMI [9] and the similarity between the clinical Table 2 Characteristics of patients with confirmed Q-wave unrecognized myocardial infarction, compared to those with asymptomatic Q-wave but without myocardial necrosis.
N Agea Women (%) Hypertension (%) Diabetes (%) Smoking Current or former smoker ≤1 year (%) Former smoker N1 year (%) Never smoker (%) Systolic blood pressurea Diastolic blood pressurea Glycemiab Total cholesterola HDL cholesterola LDL cholesterola Triglycerides mg/dlb 10-year CHD risk ≥10% (%) Major Q-wave Q-wave location on the ECG Anterior Lateral Inferior
UMI
No UMI
p value
14 61.3 (9.4) 7.1% 71.4% 35.7%
34 60.9 (10.5) 38.2% 67.6% 29.4%
0.91 0.03 0.79 0.67
28.6%
8.8%
28.6% 42.8% 142.4 (24.2) 86.7 (10.8) 113.5 (96.7–135.0) 216.1 (42.6) 43.2 (12.9) 143.4 (36.7) 110.0 (91.3–225.0) 42.9% 42.9%
29.4% 61.8% 140.2 (25.9) 81.6 (12.9) 96.0 (91.0–116.0) 206.6 (40.6) 50.9 (15.0) 133.5 (34.3) 100.5 (76.8–130.3) 5.9% 29.4%
14.3% 7.1% 78.6%
8.8% 29.4% 61.8%
303
implications and prognosis of Q-wave UMI and those of symptomatic MI [3–8]. Furthermore, non-Q-wave UMI, known to be a large proportion of all UMIs, is not detected by ECG. In the PIVUS study, cardiac MRI identified UMI in 19.8% of the elderly participants, and 80% of them had no Q-waves on the ECG [24]. In another study, the prevalence of non-Q-wave UMI in patients with suspected coronary artery disease was 27.0% (95% CI: 21.1–33.8), compared to the 8% (95% CI: 5.0–12.9) prevalence of Q-wave UMI [25]. In the Multi-Ethnic Study of Atherosclerosis (MESA), cardiac MRI identified UMI in 7.9% of the participants, 78% of them were not detectable by ECG or clinical evaluation [26]. Our data clearly reinforce the need for an imaging study to discard a false positive when an asymptomatic Q-wave is detected by ECG in clinical practice. This need is all the more urgent because UMI has such important therapeutic and prognostic implications for the individual: it increases the risk of death [3], atrial fibrillation [4], stroke [5], heart failure [6], and new episodes of MI [7]. In addition, cardiac necrosis cannot be confirmed or discarded on the basis of individual concomitant risk factors when an asymptomatic Qwave is found in an apparently healthy individual. But as previously described [27], our results show that 10-year CHD risk is markedly associated with the presence of UMI. Then, only male sex or 10-year CHD risk ≥10% justifies increased urgency in the diagnostic process; once again, an imaging study could contribute critical information needed to start preventive treatment immediately in these patients. One surprising result was the lack of association between the presence of major Q-wave and higher probability of UMI. A possible explanation is the higher proportion of major Q-wave ECG in the group with UMI, which had limited statistical power due to the sample size. It is likely that sample size also explains the failure to confirm our hypothesis that pathological ABI would be associated with higher PPV, compared to normal ABI. The incorporation of imaging techniques into clinical studies appears to be the best option to assess UMI. Along this line, delayed enhancement cardiac MRI showed 99% sensitivity to detect Q-wave MI, and 91% to detect non-Q-wave MI [28]. The use of imaging techniques could contribute several benefits: a) reduction of sample size, duration, and ethical concerns in clinical trials, avoiding unnecessary prolonged patient exposure to potential harms from experimental therapies; b) the opportunity to exclude patients with asymptomatic MI from primary prevention studies; c) the potential to assess a therapy's effectiveness in reducing UMI in intervention studies; and d) potential improvement in the efficient use of research funding resources [13]. 4.1. Study limitations
0.10 0.78 0.17 0.47 0.08 0.37 b0.01 0.36
0.24
HDL, high-density lipoprotein; LDL, low-density lipoprotein; CHD, coronary heart disease; ECG, electrocardiogram. a Mean (standard deviation). b Median (1st quartile–3rd quartile).
In observational studies, the response rate in recruitment and willingness to participate in follow-up could seriously affect the study population and introduce selection bias. In the present study, the participation rate was high in both contexts (73.8% in the cross-sectional study and 70% in the follow-up cohort study), which guaranteed external validity and a population representative of the study area. A limitation is the small number of asymptomatic Q-waves observed in our population, which limited the power to detect statistically significant differences between subgroups. Another limitation could be the 40-Hz high-frequency cutoff used in the ECG recording, as this might eliminate high-frequency signals that could be important from a clinical point of view [29]. However, a recent study [30] concludes that a 40-Hz high-frequency cutoff of the band-pass filtering yields a higher percentage of optimal quality ECGs and fewer non-interpretable traces, and allows detection of the same percentage of significant Q-waves (suggestive of myocardial necrosis), compared to the 150-Hz filtering. No inter-observer analysis was done for REGICOR ECG performance, but all ECGs were performed by a small team of highly experienced and trained nurses, thereby minimizing any variability. Subjectivity in ECG interpretation by only one cardiologist was another potential limitation. However, interpretation by experienced cardiologists has been demonstrated to be the gold standard [31]. In our
304
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305
Fig. 2. Positive predictive value of asymptomatic Q-wave in the electrocardiogram to diagnose necrosis-confirmed unrecognized myocardial infarction, by participant characteristics. PPV: positive predictive value; ABI: ankle-brachial index; CHD: coronary heart disease; LDL: low density lipoproteins; HDL: high density lipoproteins.
study, all ECGs were interpreted by the same senior cardiologist, who has done all ECG interpretation for the REGICOR study for 35 years. The final classification by expert consensus also has the potential to introduce some subjectivity, but could also be considered a strength of the study. The participation in this process by experts in each of the imaging techniques and a clinical cardiologist allowed us to take advantage of all
available information and to mirror the care process used in clinical practice. Furthermore, we did not include participants with a normal ECG in this study and therefore could not assess the sensitivity, specificity, or negative predictive value of ECG to detect Q-wave UMI. Finally, a significant proportion of UMIs could present as non-Q-wave and were not identified in this study [3,24,25].
R. Ramos et al. / International Journal of Cardiology 225 (2016) 300–305
5. Conclusions In the population aged 35–74 years, Q-waves identified opportunistically in a routine ECG overestimated the actual Q-wave UMI prevalence, which was shown to be 8% to 13% of all MI. This overestimation questions the PPV of this finding in the general population. Only in participants with a high 10-year CHD risk did the PPV exceed 50%. Therefore, identifying a pathologic Q-wave in asymptomatic patients requires confirmation by imaging tests, both in clinical practice and in epidemiological studies, to confirm or rule out myocardial necrosis. Funding sources This project was supported by clinical research grants from the Ministerio de Salud (PI20471); Spain's Ministry of Science and Innovation through the Carlos III Health Institute, co-financed with European Union ERDF funds (Network for Prevention and Health Promotion in primary Care) [RedIAPP RD12/0005]; the Red de Investigación Cardiovascular [RD12/0042/0061, RD12/0042/0013], and Miguel Servet Contract (CP12/03287); and by the Departament de Salut, Generalitat de Catalunya, Agency for Health Technology Assessment [AATRM 034/33/ 02; AATRM 078/28/06] and Agency for Management of University and Research Grants [2014 SGR 240; 2014 SGR 902]. Conflict of interest No author has reported any potential conflicts of interest involving the work under consideration for publication. Acknowledgments The authors are grateful to participants in the REGICOR and HERMES studies and to Susana Tello, Martina Sidera and Isabel Ramió for field work. The authors also appreciate the revision of the English text by Elaine Lilly, Ph.D., of Writer's First Aid. References [1] M.J. Leening, M. Kavousi, J. Heeringa, F.J. van Rooij, J. Verkroost-van Heemst, J.W. Deckers, F.U. Mattace-Raso, G. Ziere, A. Hofman, B.H. Stricker, J.C. Witteman, Methods of data collection and definitions of cardiac outcomes in the Rotterdam study, Eur. J. Epidemiol. 27 (2012) 173–185. [2] A. Linnersjö, N. Hammar, A. Gustavsson, C. Reuterwall, Recent time trends in acute myocardial infarction in Stockholm, Sweden, Int. J. Cardiol. 76 (2000) 17–21. [3] E.B. Schelbert, J.J. Cao, S. Sigurdsson, T. Aspelund, P. Kellman, A.H. Aletras, C.K. Dyke, G. Thorgeirsson, G. Eiriksdottir, L.J. Launer, V. Gudnason, T.B. Harris, A.E. Arai, Prevalence and prognosis of unrecognized myocardial infarction determined by cardiac magnetic resonance in older adults, JAMA 308 (2012) 890–896. [4] B.P. Krijthe, M.J. Leening, J. Heeringa, J.A. Kors, A. Hofman, O.H. Franco, J.C. Witteman, B.H. Stricker, Unrecognized myocardial infarction and risk of atrial fibrillation: the Rotterdam study, Int. J. Cardiol. 168 (2013) 1453–1457. [5] M.A. Ikram, M. Hollander, M.J. Bos, et al., Unrecognized myocardial infarction and the risk of stroke: the Rotterdam study, Neurology 67 (2006) 1635–1639. [6] M.J. Leening, S.E. Elias-Smale, J.F. Felix, J.A. Kors, J.W. Deckers, A. Hofman, B.H. Stricker, J.C. Witteman, Unrecognised myocardial infarction and long-term risk of heart failure in the elderly: the Rotterdam study, Heart 96 (2010) 1458e1462. [7] R.Y. Kwong, H. Sattar, H. Wu, G. Vorobiof, V. Gandla, K. Steel, S. Siu, K.A. Brown, Incidence and prognostic implication of unrecognized myocardial scar characterized by cardiac magnetic resonance in diabetic patients without clinical evidence of myocardial infarction, Circulation 118 (2008) 1011e1020. [8] A. Dehghan, M.J. Leening, A.M. Solouki, E. Boersma, J.W. Deckers, G. van Her pen, J. Heeringa, A. Hofman, K. JA, O.H. Franco, M.A. Ikram, J.C. Witteman, Comparison of prognosis in unrecognized versus recognized myocardial infarction in men versus women N55 years of age (from the Rotterdam study), Am. J. Cardiol. 113 (2014) 1–6. [9] S.E. Sheifer, T.A. Manolio, B.J. Gersh, Unrecognized myocardial infarction, Ann. Intern. Med. 135 (2001) 801–811.
305
[10] J. Herlitz, M. Dellborg, T. Karlsson, M.H. Evander, A. Berger, R. Luepker, Epidemiology of acute myocardial infarction with the emphasis on patients who did not reach the coronary care unit and non-AMI admissions, Int. J. Cardiol. 128 (2008) 342–349. [11] D. Tsiachris, C. Tsioufis, C. Thomopoulos, D. Syrseloudis, V. Antonakis, L. Lioni, I. Kallikazaros, T. Makris, V. Papademetriou, C.I. Stefanadis, New-onset diabetes and cardiovascular events in essential hypertensives: a 6-year follow-up study, Int. J. Cardiol. 153 (2011) 154–158. [12] K.A. Ammar, J.A. Kors, B.P. Yawn, R.J. Rodeheffer, Defining unrecognized myocardial infarction: a call for standardized electrocardiographic diagnostic criteria, Am. Heart J. 148 (2004) 277e284. [13] Y.B. Pride, B.J. Piccirillo, C.M. Gibson, Prevalence, consequences, and implications for clinical trials of unrecognized myocardial infarction, Am. J. Cardiol. 111 (2013) 914–918. [14] M. Grau, I. Subirana, R. Elosua, P. Solanas, R. Ramos, R. Masiá, F. Cordón, J. Sala, D. Juvinyà, C. Cerezo, M. Fitó, J. Vila, M.I. Covas, J. Marrugat, Trends in cardiovascular risk factor prevalence (1995–2000–2005) in northeastern Spain, Eur. J. Cardiovasc. Prev. Rehabil. 14 (2007) 653–659. [15] Manual of the MONICA Project, World Health Organisation, Geneva, 2000 (http:// www.thl.fi/publications/monica/manual/index.htm, (accessed 27.06.16)). [16] P.W. Macfarlane, Minnesota coding and the prevalence of ECG abnormalities, Heart 84 (2000) 582–584. [17] H. Thiele, M.J.E. Kappl, S. Conradi, J. Niebauer, R. Hambrecht, G. Schuler, Reproducibility of chronic and acute infarct size measurement by delayed enhancementmagnetic resonance imaging, J. Am. Coll. Cardiol. 47 (2006) 1641–1645. [18] J. Marrugat, I. Subirana, E. Comín, C. Cabezas, J. Vila, R. Elosua, et al., for the VERIFICA Investigators, Validity of an adaptation of the Framingham cardiovascular risk function: the VERIFICA study, J. Epidemiol. Community Health 61 (2007) 40–47. [19] J. Marrugat, R. D'Agostino, L. Sullivan, R. Elosua, P. Wilson, J. Ordovas, et al., An adaptation of the Framingham coronary heart disease risk function to European Mediterranean areas, J. Epidemiol. Community Health 57 (2003) 634–638. [20] A. de Torbal, E. Boersma, J.A. Kors, G. van Herpen, J.W. Deckers, D.A. van der Kuip, B.H. Stricker, A. Hofman, J.C. Witteman, Incidence of recognized and unrecognized myocardial infarction in men and women aged 55 and older: the Rotterdam study, Eur. Heart J. 27 (2006) 729–736. [21] M.R. MacDonald, M.C. Petrie, P.D. Home, M. Komajda, N.P. Jones, H. Beck-Nielsen, R. Gomis, M. Hanefeld, S.J. Pocock, P.S. Curtis, J.J. McMurray, Incidence and prevalence of unrecognized myocardial infarction in people with diabetes: a substudy of the Rosiglitazone Evaluated for Cardiac Outcomes and Regulation of Glycemia in Diabetes (RECORD) study, Diabetes Care 34 (2011) 1394–1396. [22] S.E. Sheifer, B.J. Gersh, N.D. Yanez III, P.A. Ades, G.L. Burke, T.A. Manolio, Prevalence, predisposing factors, and prognosis of clinically unrecognized myocardial infarction in the elderly, J. Am. Coll. Cardiol. 35 (2000) 119–126. [23] L.L. Boland, A.R. Folsom, P.D. Sorlie, H.A. Taylor, W.D. Rosamond, L.E. Chambless, L.S. Cooper, Occurrence of unrecognized myocardial infarction in subjects aged 45 to 65 years (the ARIC study), Am. J. Cardiol. 90 (2002) 927–931. [24] C.E. Barbier, T. Bjerner, L. Johansson, L. Lind, H. Ahlström, Myocardial scars more frequent than expected: magnetic resonance imaging detects potential risk group, J. Am. Coll. Cardiol. 48 (2006) 765–771. [25] H.W. Kim, I. Klem, D.J. Shah, E. Wu, S.N. Meyers, M.A. Parker, A.L. Crowley, R.O. Bonow, R.M. Judd, R.J. Kim, Unrecognized nonQ-wave myocardial infarction: prevalence and prognostic significance in patients with suspected coronary disease, PLoS Med. 6 (2009) e1000057. [26] E.B. Turkbey, M.S. Nacif, M. Guo, R.L. McClelland, P.B. Teixeira, D.E. Bild, R.G. Barr, S. Shea, W. Post, G. Burke, M.J. Budoff, A.R. Folsom, C.Y. Liu, J.A. Lima, D.A. Bluemke, Prevalence and correlates of myocardial scar in a US cohort, JAMA 314 (2015) 1945–1954. [27] D. McAreavey, J.S. Vidal, T. Aspelund, G. Eiriksdottir, E.B. Schelbert, O. Kjartansson, J.J. Cao, G. Thorgeirsson, S. Sigurdsson, M. Garcia, T.B. Harris, L.J. Launer, V. Gudnason, A.E. Arai, Midlife cardiovascular risk factors and late-life unrecognized and recognized myocardial infarction detect by cardiac magnetic resonance: ICELAND-MI, the AGES–Reykjavik study, J. Am. Heart Assoc. 5 (2016) e002420. [28] R.J. Kim, T.S. Albert, J.H. Wible, M.D. Elliott, J.C. Allen, J.C. Lee, M. Parker, A. Napoli, R.M. Judd, Gadoversetamide Myocardial Infarction Imaging Investigators, Performance of delayed-enhancement magnetic resonance imaging with gadoversetamide contrast for the detection and assessment of myocardial infarction: an international, multicenter, double-blinded, randomized trial, Circulation 117 (2008) 629–637. [29] J. García-Niebla, P. Llontop-García, J.I. Valle-Racero, G. Serra-Autonell, V.N. Batchvarov, A. Bayés de Luna, Technical mistakes during the acquisition of the electrocardiogram, Ann. Noninvasive Electrocardiol. 14 (2009) 389–403. [30] D. Ricciardi, I. Cavallari, A. Creta, G. Di Giovanni, V. Calabrese, N. Di Belardino, et al., Impact of the high-frequency cutoff of bandpass filtering on ECG quality and clinical interpretation: a comparison between 40 Hz and 150 Hz cutoff in a surgical preoperative adult outpatient population, J. Electrocardiol. 49 (2016) 691–695. [31] M.C. de Bruyne, J.A. Kors, A.W. Hoes, D.A. Kruijssen, J.W. Deckers, M. Grosfeld, G. van Herpen, D.E. Grobbee, J.H. van Bemmel, Diagnostic interpretation of electrocardiograms in population-based research: computer program research physicians, or cardiologists? J. Clin. Epidemiol. 50 (1997) 947–952.