TRENDS IN HOSPITALIZATIONS DUE TO ACUTE CORONARY SYNDROMES AND STROKE IN PATIENTS WITH SYSTEMIC LUPUS ERYTHEMATOSUS, 1996 TO 2012

Abstract

Objectives

Cardiovascular disease (CVD) has been recognized as a major cause of morbidity in patients with systemic lupus erythematosus (SLE), but it is not clear if increased awareness of these risks has translated into improvements in CVD morbidity at the population level. To determine this, we examined national trends in rates of hospitalizations for CVD events in patients with SLE from 1996 to 2012.

Methods

We used the Nationwide Inpatient Sample to estimate rates of hospitalizations for acute myocardial infarction, unstable angina, and ischemic stroke from 1996 to 2012 in patients with SLE. We compared these trends to those in the general population.

Results

During the study years, there were an estimated 31,012 hospitalizations for acute myocardial infarction, 4,160 hospitalizations for unstable angina, and 26,144 hospitalizations for ischemic stroke among patients with SLE. Rates of hospitalizations for acute myocardial infarction and ischemic stroke increased over time in patients with SLE, while rates for unstable angina decreased. Rates for all three conditions decreased in the general population over these years, with hospitalization rates for unstable angina decreasing faster in the general population than among patients with SLE.

Conclusion

Increased awareness of the burden of CVD in patients with SLE has not yet translated into decreases in hospitalizations for acute myocardial infarction or stroke. This may be due to barriers in implementation of CVD risk factor modification, or SLE-specific risks that have not yet been identified or effectively targeted.

Over the past 20 years, there has been a large increase in awareness of, and clinical focus on, the importance of cardiovascular disease (CVD) in patients with systemic lupus erythematosus (SLE). Patients with SLE have a 2–10 fold increased risk of CVD compared to the general population, with higher relative risks among younger patients (1). This increased risk is thought to be due to both traditional CVD risk factors and lupus-associated factors, including systemic inflammation and endothelial dysfunction, corticosteroid use, and in some patients, a prothrombotic state due to antiphospholipid antibodies. Early identification and management of cardiovascular risk factors is important to reduce the prevalence of CVD. However, it is not known if increased attention to the risk of CVD in SLE has resulted in greater use of risk factor modification and, in turn, lower CVD morbidity (2). Because the majority of patients with acute coronary syndromes or stroke are hospitalized, hospitalizations for these conditions represent a measure of the burden of CVD. Whether the prevalence of hospitalizations for acute coronary syndromes and stroke in patients with SLE has changed over recent decades at the population level, potentially as a result of increased awareness and use of primary and secondary prevention, has not been reported.

Our aim was to examine trends in hospitalization rates due to acute myocardial infarction (AMI), unstable angina, and stroke in a nationally representative sample of adults with SLE in the United States from 1996 to 2012. We hypothesized that rates of hospitalizations for these conditions in adults with SLE have decreased over time, mirroring trends in the general population (3,4). In addition, we hypothesized that events in older women might be more related to traditional CVD risk factors compared to events in younger women, and therefore might follow the pattern in the general population more closely.

METHODS

Data source

We used data from the Nationwide Inpatient Sample (NIS), compiled by the Agency for Healthcare Research and Quality's Healthcare Cost and Utilization Project (5). NIS is the largest publicly available all-payer (Medicare, Medicaid, private insurance, and uninsured) database of inpatient care in the United States. It contains data from approximately 8 million hospitalizations annually from over 1000 hospitals, representing a 20% stratified sample of community acute-care hospitals, excluding rehabilitation and long-term care hospitals. NIS contains information on patient characteristics, discharge diagnoses and inpatient procedures, length of stay, patient disposition, and hospital characteristics, but does not include data on medications or laboratory test results. Unique patient identifiers are also not included, so the unit of analysis is the hospitalization. Diagnoses and procedures were coded using the International Classification of Diseases, 9th Revision-Clinical Modification (ICD-9). NIS was redesigned in 2012 to sample hospitalizations among all eligible hospitals, rather than including all hospitalizations in selected hospitals, with the goal of providing more stable estimates (5).

We included data from 1996 to 2012, the most recent year of data available. Data were provided through a data use agreement. This study was exempted from human subjects review by the National Institutes of Health Office of Human Subjects Research Protection.

Study population and variables

We used ICD-9 codes to identify hospitalizations of patients aged 18 years or older for which AMI, unstable angina, or stroke was the primary discharge diagnosis. For AMI, we used ICD-9 codes 410.xx (excluding 410.x2), which had a positive predictive value of 83%–94% in recent validation studies using Medicare or other large US administrative databases (6). We used ICD-9 code 411.1 for unstable angina. Ischemic stroke, excluding transient ischemic attack, was identified by ICD-9 codes 433.x1, 434.x1, 436.xx, 437.1x, or 437.9x. These codes had positive predictive values of 88%–96% in validation studies (7).

We identified hospitalizations for adults (age 18 or older) with SLE using the ICD-9 code of 710.0 listed as any secondary discharge diagnosis. We have previously validated this approach to identification of patients with SLE (8). We excluded hospitalizations of patients with SLE if these also included discharge codes for rheumatoid arthritis, systemic sclerosis, dermatomyositis, polymyositis, or other connective tissue diseases.

Statistical analysis

We computed hospitalization rates for AMI, unstable angina, and stroke as the weighted number of hospitalizations per population per year. For SLE, we used recent prevalence data to estimate the population with SLE in the US in each year, and used these as the denominators for the rates, as detailed previously (8). For comparison, we also computed rates of each condition in persons without SLE. We used US Census data to estimate intercensal population sizes for use as denominators in rates for non-SLE hospitalizations. We age-standardized both rates to the 2000 Standard Population, and computed relative risks of hospitalization for each condition by year. We repeated the analysis in women age 18 to 49, women age 50 or older, and men to determine if trends in these subgroups differed from those in patients with SLE overall.

We also used a second method to compare trends in CVD hospitalizations in SLE that did not require estimating the denominator population of patients with SLE. This approach used all-cause hospitalizations (excluding maternity-related hospitalizations) among patients with SLE as an internal comparator. Rates for all-cause hospitalizations were determined to be stable over the study years (9). We indexed the age-adjusted rate of CVD hospitalizations to the age-adjusted rate of all-cause hospitalizations among patients with SLE in each year. Changes in the ratio of CVD hospitalizations to all-cause hospitalizations over time would therefore provide an indication of whether the frequency of CVD hospitalizations had changed relative to other conditions.

We used SAS programs version 9.3 (SAS Institute, Cary NC) for all analyses.

RESULTS

Among patients with SLE, there were an estimated 31,012 hospitalizations for AMI, 4,160 hospitalizations for unstable angina, and 26,144 hospitalizations for ischemic stroke during the study years. Most patients were middle-aged or elderly women hospitalized at non-teaching hospitals, most often in southern states (Supplemental table 1).

Rates of hospitalizations for AMI and stroke among patients with SLE increased modestly over the study period (Figure 1). For AMI, the rate was 9.6/1000 patients in 1996 and 14.5/1000 patients in 2012, while for stroke, these rates were 9.0/1000 patients and 14.2/1000 patients, respectively. In contrast, rates of both conditions decreased in those without SLE. Hospitalizations for unstable angina decreased in both patients with SLE and those without SLE, although decreases were more pronounced in those without SLE (Figure 1). Findings were similar in younger (age 18 to 49 years) and older (age 50 and older) women and in men when these groups were analyzed separately (Figure 2). The relative risk of hospitalizations among younger women with SLE compared to younger women without SLE were stable over time, while relative risks increased over time among older women and men with SLE (Supplemental Figure 1).

Figure 1.

Age-adjusted rates of hospitalizations for acute myocardial infarction (AMI), stroke, and unstable angina in adults with systemic lupus erythematosus (A) or without systemic lupus erythematosus (B), by year. Values are three-year moving averages. Error bars are 95% confidence limits. Confidence limits in panel B are very close to the point estimate and difficult to appreciate.

Figure 2.

Age-adjusted rates of hospitalizations for acute myocardial infarction (AMI), stroke, and unstable angina in adults with systemic lupus erythematosus (top row) or without systemic lupus erythematosus (bottom row), by year, stratified by sex and also by age among women. Values are three-year moving averages. Error bars are 95% confidence limits. Confidence limits in the graphs in the bottom row are very close to the point estimate and difficult to appreciate.

Compared to all-cause hospitalizations, rate ratios of hospitalizations for AMI and stroke were stable over time among patients with SLE, again suggesting no change in the rate of hospitalizations for these conditions among patients with SLE (Figure 3). The rate ratio of unstable angina decreased over time in this analysis, suggesting some improvement in rates of this condition.

Figure 3.

Age-adjusted ratios of hospitalizations for acute myocardial infarction (AMI), stroke, and unstable angina to all non-maternity hospitalizations among adults with systemic lupus erythematosus, by year.

DISCUSSION

In this national population-based study, rates of hospitalizations for AMI and stroke in patients with SLE increased from 1996 to 2012. Rates increased in both younger and older women, even though CVD events in older women might be expected to be more closely related to traditional CVD risk factors, and therefore possibly more sensitive to changes in preventive measures. Although hospitalizations for unstable angina decreased among patients with SLE and patients in the general population, the rate of decrease was lower among patients with SLE.

The decrease in CVD hospitalization rates in the general population is consistent with previous evidence of a steady decline in the incidence of acute coronary syndromes and stroke over the past 20 years (3,4). That our analysis corroborated these findings in the general population helps to validate our results. Improvements in these rates in the general population have been attributed to public health programs and medical interventions to control cardiovascular risk factors. The lack of improvement in CVD hospitalization rates in SLE may be due to several possible factors: 1. continued low awareness of CVD risk among physicians caring for patients with SLE; 2. high awareness but low implementation of measures for the management of CVD risk factors in patients with SLE; 3. high implementation of prevention measures but a long lag before these interventions affect atherosclerotic risk and influence the rates of CVD hospitalizations; or 4. high implementation of prevention measures but incomplete effectiveness because SLE-related factors have an additional, or dominant, impact on CVD risk compared to traditional risk factors.

Studies from SLE specialty centers reported low rates of assessment and treatment of traditional CVD risk factors before 1996, which improved subsequently (10). While awareness has likely increased over time, no national surveys have assessed if awareness of CVD risks in SLE is widespread among primary care providers. Recent studies suggest ongoing deficiencies in awareness or other barriers to risk factor assessment. Studies using data from cardiovascular health questionnaires and medical records have shown that 58% of patients reported never having received CVD counseling (11), while serum lipid profiles and body weight were recorded in only one third of patients, and physical activity and family CVD history were recorded in less than 10% of patients (12,13). Demas and colleagues reported that only 3% of patients with SLE treated at an academic center had annual assessment of five risk factors previously proposed as quality indicators for CVD (smoking, blood pressure, body mass index, diabetes, and serum lipids), while 26% had four risk factors assessed (14). In a population-based cohort study from Wisconsin which examined patients with SLE from 1991 to 2009, lipid testing was performed in only 66% of patients over a 7-year mean follow-up period (2). Among those with hyperlipidemia, less than 20% were treated with statins. Risk factor control can also be impacted by poor adherence to recommended preventive measures (13). These findings indicate issues with either awareness or appropriate implementation of prevention measures.

The time needed for prevention measures to diffuse into clinical care and affect CVD risks may delay changes in rates of CVD hospitalizations. In the Atherosclerosis Risk in Communities cohort, stroke incidence decreased from 1987 to 2011, following an increase of anti-hypertensive medication use from 29% in 1987–1989 to 43% in 1996–1998 (4). In a recent large study of Medicare beneficiaries, the incidence of ischemic stroke decreased by 40% between 1988 and 2008, with a lag of 3–4 years after the use of antihypertensive drugs and statins started to increase (15). These findings suggest that the effects of interventions on CVD events can be seen over relatively short periods at the population level. Prospective studies examining the impact of the management of traditional risk factors on CVD morbidity in SLE would help in the development of strategies for the modification of CVD risk in patients with SLE.

Independently of the role of the traditional risk factors, SLE likely has an important impact on CVD risk. However, the specific pathogenetic mechanisms are not clear (1). Rates of CVD-related hospitalizations may not have decreased because the SLE-associated factors affecting cardiovascular risk have not changed over time, despite overall advances in the management of patients. Improved control of inflammation, better management of hypertension and nephrotic syndrome, minimization of cumulative corticosteroid doses, and prophylactic measures against thrombosis in patients with antiphospholipid antibodies, may help reduce CVD events in SLE, but these measures alone may not be sufficient or may not impact the key targets.

Rates of hospitalizations for unstable angina decreased among patients with SLE. These decreases may reflect improvements in the care of patients with CVD or stable angina, but the significance of any improvements is unclear in light of increases in rates of AMI (16). Alternatively, the decreases may be due in part to changes in the diagnostic criteria for AMI and use of more sensitive biomarkers for AMI, which may lead more patients to be diagnosed with AMI rather than unstable angina, and due to expanded use of observation care units rather than hospitalization for patients with unstable angina at low risk for AMI (17).

The strengths of this study include the large population-based sample and many years of data. A limitation of the NIS is the lack of clinical data on risk factors, including laboratory test results. The likelihood of underascertainment of CVD events is low because these conditions require inpatient care, although our study would not capture out-of-hospital CVD deaths or events. However, there is no evidence to suggest that a change in the proportion of in-hospital versus out-of-hospital events has occurred that could mask a decrease in in-hospital event rates in patients with SLE. Rates were based on hospitalizations, rather than unique patients. It is possible that the lack of a decrease in rates among patients with SLE was due in part to a higher likelihood of recurrent CVD events in patients with SLE than in the general population, or a lower case fatality rate among patients with SLE and CVD events, but we are unaware of data to support these possibilities. Because national data on SLE prevalence are not available, we used data from two metropolitan areas to estimate the number of SLE patients in the US (8). However, findings were similar in an analysis using an internal comparison that did not rely on external prevalence data, suggesting that any inaccuracies in the SLE prevalence estimates would not affect the conclusions.

Despite recognition of the importance of CVD in patients with SLE, rates of hospitalizations for CVD events have not decreased. While issues related to screening and treatment of traditional CVD risk factors in patients with SLE may be contributory, it is not clear if control of these risk factors alone will be sufficient to reduce CVD events, or if interventions targeting SLE-specific risks need to be developed. Until the dominant mechanisms underlying CVD risk in SLE are known, efforts to optimize screening for and modification of known risk factors remain essential.