Admission route and use of invasive procedures during hospitalization for acute myocardial infarction: analysis of 2007-2011 National Health Insurance database

OBJECTIVES The goal of this study was to investigate trends in admission to the emergency department and the use of invasive or surgical procedures during hospitalization for acute myocardial infarction (AMI) in Korea. METHODS The National Health Insurance (NHI) claims database between 2007 and 2011 was used. We identified all admission claims that included codes from the tenth revision of the International Classification of Diseases beginning with I21 as the primary or secondary diagnosis. Information about the admission route, admission date, discharge date, and whether coronary artery angiography, angioplasty, or bypass surgery was performed was also obtained from the NHI database. RESULTS Of the 513,886 relevant admission claims over the five years encompassed by this study, 295,001 discrete episodes of admission for AMI were identified by analyzing the date and length of each admission and the interval between admissions. The number of AMI admissions gradually decreased from 66,883 in 2007 to 47,656 in 2011. The number and proportion of admissions through the emergency department also decreased from 38,118 (57.0%) in 2007 to 24,001 (50.4%) in 2011. However, during the same period, admissions in which invasive or surgical procedures were performed increased from 15,342 (22.9%) to 17,505 (36.7%). CONCLUSIONS The reported numbers of emergency department visits and admissions for AMI decreased from 2007 to 2011. However, only a small portion of the patients underwent invasive or surgical procedures during hospitalization, although the number of admissions involving invasive or surgical procedures has increased. These findings suggest that simply counting the number of admission claims cannot provide valid information on trends in AMI occurrence.


INTRODUCTION
Acute myocardial infarction (AMI) is one of the leading causes of death worldwide [1,2]. According to National Health Insur ance (NHI) claims data, medical service uses and costs associat ed with AMI in Korea have continuously increased over the last few decades [3]. The NHI claims data have been used to assess disease burden, because their database encompasses almost all medical usage in Korea under the mandatory universal health insurance system [4]. The NHI claims data have also been used to investigate disease prevalence and incidence [57]. However, since compensation for medical services is the main purpose of NHI claims, the NHI claims data do not include the informa tion necessary for estimating disease frequency. For example, diagnoses in the NHI claim database do not perfectly match clinical diagnoses [8]. Moreover, multiple claims can be issued for a single disease event [5,6,9]. Therefore, it is necessary to identify and validate disease episodes in order to estimate dis ease frequency correctly [57].
Since early revascularization has been proven to improve the clinical outcome of AMI patients, the revascularization rate has increased in many countries, including Korea [10,11]. Shorten ing the initiation time for revascularization is also associated with improved outcomes in AMI patients [12]. Studies have re ported that using emergency medical services shortened the initiation time for revascularization compared to patients who selftransported to a hospital [13,14]. However, no nationally representative study has investigated the utilization of medical resources by AMI patients. In particular, previous research has not addressed how many patients are admitted through the emer gency department and how many patients undergo revascular ization procedures during hospitalization. Most previous stud ies of AMI patients in Korea were limited to selected large hos pitals, and only included subjects who were admitted through the emergency department [7,15].
Thus, we used the NHI claims database from 2007 to 2011 to investigate trends in AMI admissions through the emergency department and in AMI admissions involving invasive revascu larization procedures.

MATERIALS AND METHODS
We analyzed entries in the NHI claims database between Jan uary 1, 2007 and December 31, 2011. The database includes date of admission, date of discharge, admission route, diagnosis codes according to the tenth revision of the International Clas sification of Diseases, and details about which medical services were used during hospitalization [16].
We aimed to assess how many patients with a diagnosis of AMI used emergency services and/or underwent invasive re vascularization procedures, rather than attempting to estimate the absolute number of patients with AMI. Thus, we defined an AMI admission as a hospitalization claim with I21.X as the pri mary or secondary diagnosis. We did not include claims with the diagnosis codes I22.X-I25.X, because they represent sec ondary conditions after an AMI event or other ischemic heart diseases. We identified a total of 513,886 claims for AMI admis sions among 269,843 patients during the five years of the study. Multiple claims can be issued for a single AMI episode, for in stance, when a patient is admitted to more than one hospital, or admitted twice or more to the same hospital. Under the Ko rean NHI system, hospitals submit their claims for inpatient service to the Health Insurance Review and Assessment Service monthly. Therefore, multiple claims can be submitted for a sin gle episode, if the hospitalization duration is especially long or happens to encompass multiple reporting periods. Thus, we an alyzed the date of admission, date of discharge, and length of hospitalization in order to identify discrete AMI episodes. Two adjacent admission claims were regarded as the same episode of AMI when they met at least one of the following three con ditions: 1) the admission date of the second hospitalization was within 28 days of the admission date of the first hospitalization; 2) the admission date of the second hospitalization was within three days of the discharge date of the first hospitalization; or 3) the length of the second hospitalization was less than three days ( Figure 1) [9].  We assessed whether the hospitalized patient was admitted through the emergency department. In our analysis, invasive procedures included both diagnostic and therapeutic proce dures involving the coronary arteries. Invasive therapeutic pro cedures included surgical and nonsurgical revascularization techniques involving the coronary arteries, such as percutane ous transluminal coronary angioplasty and coronary artery by pass surgery. Coronary angiography, which is a diagnostic pro cedure, was also included because it is routinely performed to evaluate coronary artery occlusion and is frequently accompa nied by therapeutic procedures.

RESULTS
We initially identified 513,886 claims from 269,843 patients for hospitalization with a diagnosis code of I21.X between 2007 and 2011. These figures included 180,597 patients with a single claim and 89,246 patients with multiple claims. From a total of 333,289 claims issued for the 89,246 patients with multiple claims, we identified 114,404 AMI episodes according to the algorithm described in the materials and methods section. The final number of total identified discrete AMI admissions was 295,001 ( Figure 1). Overall, 53.4% of the episodes involved admission through the emergency department. Male and mid dleaged (40 to 69 years) patients were more likely to be admit ted through the emergency department. Altogether, 27.6% of AMI patients underwent invasive or surgical procedures during hospitalization. Male and middleaged (40 to 69 years) patients were more likely to undergo invasive procedures ( Table 1).
The total number of AMI admissions gradually decreased from 66,883 in 2007 to 47,656 in 2011. This trend was consistent across both sexes and all age groups (Table 2). Moreover, the number and proportion of admissions through the emergency   (Table 3). Figure 2 shows the decreasing trend in the reported numbers of hospital admissions and emergency department visits for AMI, as well as the increasing trend in hospital admis sions involving invasive or surgical procedures. Figure 3 shows that the proportion of admissions through the emergency de partment among all AMI admissions decreased, although the proportion of admissions involving invasive or surgical proce dures increased between 2007 and 2011.

DISCUSSION
Through an analysis of the NHI claims database, we observed that the reported number of AMI admissions decreased from    2007 to 2011. We also observed that the number of emergency department visits for AMI decreased, but the number of AMI admissions involving invasive or surgical procedures increased. Two studies have previously estimated the incidence of AMI in Korea based on analyses of the NHI claims database. One study concluded that the incidence of AMI increased between 1997 and 2007 [5], while another study reported that the inci dence of AMI decreased between 2006 and 2010 [6]. It is inap propriate to directly compare the absolute incidence rates esti mated by these two studies, because they used different meth ods to identify AMI admissions. However, it is still notable that the two studies reported opposite trends in the incidence of AMI, even within a short overlapping period (2006 to 2007). Our study provides at least a partial explanation of these con flicting results. According to other studies that analyzed hospital records, over 90% of patients with AMI were admitted through the emergency department, and over 70% underwent invasive procedures during hospitalization [7,17]. We assume that ma jority of patients with AMI would visit the emergency depart ments when the symptoms occurred, and majority of patients would also undergo invasive diagnostic or therapeutic proce dures during their hospitalization. In our study, we found that almost half of the reported AMI admissions were not through the emergency department, and that only a small portion of the admissions were accompanied by invasive or surgical proce dures. These findings suggest that trends in insurance claims for AMIs may not reflect trends in AMI incidence.
We evaluated trends in total AMI admissions using the corre sponding diagnosis codes in the NHI claims database. Diagno ses in the NHI claims database do not perfectly match clinical diagnoses, resulting in misclassification [8]. The decline in re ported AMI admissions might be due to the decrease of false positives, as diagnostic techniques have improved and health insurance review has become more ubiquitous [5]. The decreas ing numbers of admissions through the emergency department can be partially explained by the decrease of total AMI admis sions. However, the decreasing proportion of admissions through the emergency department among total AMI admissions im plies that there are other factors in play. The proportion of non STsegment elevation myocardial infarction (NSTEMI) among total AMI patients has increased in recent years [1820]. Pati ents with NSTEMI are less likely to present with typical chest pain or discomfort than patients with STsegment elevation myo cardial infarction (STEMI) [21]. The increase in NSTEMI patients might be related to the observed decrease in emergency depart ment admissions, as the severity of symptoms affects the utili zation of emergency medical services [22].
The use of invasive revascularization strategies is increasing among NSTEMI patients [18]. Therefore, the increasing propor tion of NSTEMI cases might be also related to the observed in crease in invasive procedures among total AMI admissions [18]. Invasive revascularization strategies have been proven to im prove the outcome of AMI patients, in comparison with conser vative or thrombolytic therapy [23]. The preference for invasive procedures over noninvasive medical treatment could increase the use of invasive procedures [24]. We also observed a dispari ty between men and women in the proportion of patients who underwent invasive or surgical procedures. This result is consis tent with previous studies, which have reported sex disparities in the distribution of patients who receive revascularization pro cedures [25]. The causes of these disparities are not yet fully understood, but a lower chance of undergoing revascularization procedures may accompany the high frequency of cardiovascu lar disease risk factors and atypical symptoms in females [26]. Findings from previous studies and the present study suggest that the relative prevalence of STEMI and NSTEMI has recent ly changed in Korea [17,19]. An increase in NSTEMI cases may affect the utilization of the emergency department and invasive procedures [1822]. A simple analysis of the NHI claims data base may overestimate the incidence of AMIs. We cannot fully explain why emergency department visits for AMI have decreas ed while invasive procedures have become increasingly com mon. Nonetheless, we would point out that an analysis limited to the diagnosis codes in the NHI claims database cannot pro vide valid information for estimating the incidence of AMIs.
The major advantage of this study is the high coverage level of the NHI [4]. We suggest that the results in this study repre sent trends in emergency department visits for AMIs and the rate of invasive procedures among AMI admissions in the Kore an population. We also developed an algorithm to identify dis crete admission episodes based on the date of admission, date of discharge, and length of admission. It is important to distin guish the number of insurance claims and the number of dis ease episodes, because multiple claims are possible for a single disease event [9]. The discrepancy between insurance claims and disease episodes varies widely depending on the disease in question [8,27]. Therefore universal algorithms cannot be ap plied to various kinds of diseases.
This study also has several limitations that should be address ed. First, the NHI claims database cannot identify patients who did not use medical services [6]. However, this underestimation would not severely influence the proportion of emergency de partment visits or the proportion of patients who underwent invasive procedures. Second, we included coronary angiogra phy in the category of invasive procedures. Since the reasons for undergoing invasive procedures and their results were not available in the NHI claims database, we did not have enough information to determine whether coronary angiography was performed for purely diagnostic purposes or was performed in connection with therapeutic procedures. Therefore, the trend in the number of admission episodes involving invasive procedures may not reflect a real trend in AMI incidence. Third, in our data, the admission route was not recorded for approximately 2.4% of AMI admissions. In addition, no data are available regarding the accuracy of the admission routes recorded in the NHI claims database. Further studies are required to assess the validity and reliability of the NHI claims database. Fourth, we did not have information regarding whether the invasive procedures were performed appropriately. The feasibility of using invasive proce dures or a preference for invasive procedures over noninvasive medical therapy might produce a tendency to overuse invasive procedures, but no information regarding this issue was accessi ble in this study. Finally, including AMI admission episodes cod ed with secondary diagnoses might have led to overestimating newly developed AMI episodes. In most patients with newly developed AMI, AMI would be recorded as the primary diagno sis. AMI admission episodes coded with secondary diagnoses might include patients with a previous history of AMI. The low proportions of emergency department visits and admissions in volving invasive procedures imply this overestimation. Howev er, this overestimation would not affect trends in the utilization of medical resources. Analyzing AMI admission episodes re stricted to primary diagnoses is likely to be more appropriate for evaluating medical usage patterns in newly developed cases of AMI. Further studies incorporating information from medical records or laboratory tests could improve the validity of analy ses based on the claims database [28].
In conclusion, the total number of NHI claims for AMI admis sions decreased between 2007 and 2011. However, AMI admis sions accompanied by coronary angiography and/or coronary revascularization procedures increased during the same period. Moreover, unexpectedly small proportions of AMI admissions were through the emergency department or involved invasive procedures. Our findings suggest that simply counting the num ber of admission claims cannot provide valid information about trends in AMI events. Careful interpretation is required to dis cern the actual trends in AMI admissions based on an analysis of health insurance claims.