Author + information
- Received July 21, 2018
- Revision received October 23, 2018
- Accepted October 29, 2018
- Published online February 4, 2019.
- Srikanth Yandrapalli, MDa,∗ (, )
- Christopher Nabors, MDb,
- Abhishek Goyal, MD, MPHc,
- Wilbert S. Aronow, MDa and
- William H. Frishman, MDb
- aDivision of Cardiology, Department of Medicine, Westchester Medical Center and New York Medical College, Valhalla, New York
- bDepartment of Medicine, Westchester Medical Center and New York Medical College, Valhalla, New York
- cDivision of Internal Medicine, Department of Medicine, Brigham and Women’s Hospital, Boston, Massachusetts
- ↵∗Address for correspondence:
Dr. Srikanth Yandrapalli, Division of Cardiology, Westchester Medical Center and New York Medical College, 100 Woods Road, Macy Pavilion, Valhalla, New York 10595.
Background Modifiable risk factors (RFs) play an important role in the development and prognosis of acute myocardial infarction (AMI).
Objectives This study sought to study the prevalence rates of modifiable RFs during a first AMI, sex/race differences, and temporal trends in U.S. young adults.
Methods This was a retrospective cohort analysis of the U.S. National Inpatient Sample years 2005 and 2015 to identify adults 18 to 59 years of age hospitalized for a first AMI. Prevalence rates, race and sex differences, and temporal trends of hypertension, diabetes mellitus, obesity, smoking, dyslipidemia, and drug abuse were analyzed in these patients.
Results The authors’ study included 1,462,168 young adults with a first AMI (mean age 50 ± 7 years, 71.5% men, 58.3% white) of whom 19.2% were 18 to 44 years of age, and 80.8% were 45 to 59 years of age. In the 18- to 44-year group, smoking (56.8%), dyslipidemia (51.7%), and hypertension (49.8%) were most prevalent, and 90.3% of patients had at least 1 RF. In the 45- to 59-year group, hypertension (59.8%), dyslipidemia (57.5%), and smoking (51.9%) were most prevalent, and 92% patients had at least 1 RF. Significant sex and racial disparities were observed in the prevalence of individual RFs. Women had a higher prevalence of diabetes mellitus, hypertension, and obesity, and men had a higher prevalence of dyslipidemia, drug abuse, and smoking. The prevalence of all these RFs increased temporally except for the rate of dyslipidemia, which decreased more recently. Trends were generally consistent across sex and racial groups.
Conclusions During a first AMI in young adults in whom preventive measures are more likely to be effective, modifiable RFs were highly prevalent and progressively increased over time. Significant sex and racial disparities were observed for individual RFs.
Coronary heart disease (CHD) is a major cause of morbidity and mortality, the development of which is related in large measure to the presence of modifiable risk factors (RFs) (1–4). Hypertension, dyslipidemia, smoking, obesity, and diabetes mellitus (DM) are important modifiable RFs for the development of an acute myocardial infarction (AMI) (3–7). Numerous studies have evaluated the prevalence of these RFs during a first or any episode of AMI and have found high prevalence rates of at least 1 RF (approximately 85% to 90%) (3,6,8). Drug abuse has also been described as a contributor to the development and prognosis of AMI (9–14). Differences in outcomes after an AMI have been observed across sex and racial groups, and these have been attributed to the differential baseline RF profiles (15–19). Although rates of AMI in the United States have decreased recently, the prevalence of these modifiable RFs during an AMI are actually increasing (2,20). These epidemiological data form an important basis for the implementation of primary and secondary preventative strategies to reduce the burden of CHD in appropriate patient populations.
There are, however, few studies that focus on recent temporal trends in the prevalence of these RFs during a first AMI in young adults in the U.S. population. In this group of patients, preventive measures are likely to be particularly effective. For that reason, we sought to determine the overall prevalence, race and sex differences, and respective temporal trends of modifiable RFs in young U.S. adults between the years 2005 and 2015.
Data were obtained from the U.S. Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) from January 2005 through September 2015. The NIS contains around 8 million unweighted discharge records per year and is a 20%-stratified sample of all-payer inpatient admissions from acute care hospitals (21). Stratified sampling is done on the basis of geographic region, location, teaching status, ownership, and bed size. Discharge weights were provided for each record, which were used to obtain national-level estimates. The NIS was redesigned in 2012 using a new sampling strategy to provide more precise estimates by reducing sampling error. Redesign discharge weights (TRENDWT) were provided for the years before 2012 to facilitate patient-level analysis when comparing data before and after 2012.
Study population and outcomes
We identified hospitalizations for AMI in young adults 18 to 59 years of age by identifying International Classification of Diseases-Ninth Revision-Clinical Modification (ICD-9-CM) codes 410.x1 in the primary discharge diagnosis position. First AMI cases were identified by excluding cases with secondary diagnoses of prior myocardial infarction (MI) (412), prior percutaneous coronary intervention (V45.82), prior coronary arterial bypass grafting (V45.81), post-AMI syndrome (411.0), chronic ischemic heart disease (414.8, 414.9), heart transplant recipient (V421), and coronary arterial disease of bypass grafts or in transplanted hearts (414.02–414.07). Cases with missing values for age and sex were excluded. Cases without death as an outcome, but with a length of stay of 0 days, were excluded to eliminate cases that might not have been true AMI hospitalizations. Patients were categorized into 2 age groups: 18 to 44 years and 45 to 59 years. Race was categorized as white, black, Hispanic, and Asian/Pacific Islander. RFs were identified as current or prior smoker (305.1, V15.82), dyslipidemia (272.0–272.4). DM, hypertension, obesity, and drug abuse data were extracted from the Elixhauser comorbidity data in the NIS. Drug abuse was identified as dependence on or nondependent abuse of cannabis, cocaine, hallucinogens, amphetamines, opioids, sedatives, antidepressants, or combinations of these. The type of AMI was identified as ST-segment elevation MI (STEMI) (410.x1 excluding 410.71) and non-STEMI (NSTEMI) (410.71). Mechanical revascularization was identified as either percutaneous coronary intervention (00.66, 36.01, 36.02, 36.05, 36.06, 36.07) or coronary arterial bypass grafting (36.1x).
Primary outcomes of interest were the overall prevalence of the RFs in the age subgroups, sex and race differences in RF prevalence, and respective temporal trends during the study period.
Weighted data using the redesign trend weights were used for all analyses. Differences in RF prevalence across sex and racial groups were compared using the survey-specific Rao-Scott chi-square test for categorical variables, which takes the survey design (strata, clusters, and redesign discharge weights) into account. Trends in prevalence rates of RFs were analyzed using the Mantel-Haenszel test for trend (22). Categorical variables are expressed as proportions and continuous variables as mean ± SD. In the tables, SEs are reported for the RF proportions to provide a measure of precision of the calculated proportion. All reported p values are 2-sided. Statistical analysis was performed using IBM SPSS Statistics version 25.0 (IBM, Armonk, New York).
Prevalence of RFs
Our study included 1,462,168 cases of first AMI hospitalizations in young adults. Mean age was 50 ± 7 years, and the majority (71.5%) were men. Most patients were white (58.3%). The presentation of AMI was more frequently NSTEMI (54.9%). Of the overall sample, 19.2% (n = 280,975) patients were 18 to 44 years of age (mean age 39 ± 5 years), and 80.8% (n = 1,181,282) were 45 to 59 years of age (mean age 53 ± 4 years). Baseline characteristics of the overall sample and age subgroups are presented in Table 1. Overall, hypertension (57.9%) followed by dyslipidemia (56.8%) and smoking (52.8%) were the most prevalent RFs, and 91.7% of patients had at least 1 RF (Central Illustration). Baseline characteristics categorized by the type of AMI are presented in Online Table 1. Patients presenting with NSTEMI had a higher prevalence of all individual modifiable RFs except for smoking, which was more prevalent in STEMI patients. The proportion of patients with at least 1 RF was higher with NSTEMI (92.6%) than with STEMI (90.6%). Mechanical revascularization was more frequently performed in patients with STEMI (84.9%) than with NSTEMI (58.3%) (Online Table 1).
In the 18- to 44-year group, smoking (56.8%) followed by dyslipidemia (51.7%) and hypertension (49.8%) were the most prevalent RFs, and 90.3% of patients had at least 1 RF. Obesity and DM were present in 1 of 5 patients, and drug abuse in 1 of every 10 cases with a first AMI (Table 1). In the 45- to 59-year group, hypertension (59.8%) followed by dyslipidemia (57.5%) and smoking (51.9%) were the most prevalent, and 92% patients had at least 1 RF. DM was present in 1 in 4, obesity in 1 in 6, and drug abuse in 1 in 20 cases at the time of a first AMI. Minor differences in the prevalence rates of individual RFs were observed between the types of AMI, and these data are presented in Online Table 1.
RF differences across sex categories are presented in Table 2. In sex-specific analysis, men and women had a similar age at first AMI (50 ± 7 years in both). In the 18- to 44-year group, smoking was most prevalent in both sex categories, followed by hypertension in women and dyslipidemia in men. During a first AMI, women 18 to 44 years of age had a higher prevalence of DM (29.8% vs. 19.9% in men; p < 0.001; rate difference 9.9%), obesity (26.9% vs. 18.6% in men; p < 0.001; rate difference 8.3%), and hypertension (51% vs. 49.3% in men; p < 0.001; rate difference 1.7%). Men had a higher prevalence of dyslipidemia (54.6% vs. 44.1%; p < 0.001; rate difference 10.5%), smoking (58.1% vs. 53.4%; p < 0.001; rate difference 4.7%), and drug abuse (10.1% vs. 8.4%; p < 0.001; rate difference 1.7%) than did women.
In the 45- to 59-year group, hypertension followed by dyslipidemia were the most prevalent RFs in women, whereas dyslipidemia followed by hypertension were the most prevalent RFs in men. During a first AMI, women 45 to 59 years of age had a higher prevalence of DM (34.8% vs. 26% in men; p < 0.001; rate difference 8.8%), obesity (22.3% vs. 15.1% in men; p < 0.001; rate difference 7.2%), and hypertension (63% vs. 58.6% in men; p < 0.001; rate difference 4.4%). Men had a higher prevalence of dyslipidemia (58.9% vs. 54.1%; p < 0.001; rate difference 4.8%), smoking (52.8% vs. 49.6%; p < 0.001; rate difference 3.2%), and drug abuse (4.8% vs. 4%; p < 0.001; rate difference 0.8%). In RFs for which men had higher prevalence rates, the rate differences were lower in the 45- to 59-year group than in the 18- to 44-year group.
RF differences across racial groups are presented in Table 3. Age of first AMI was similar across groups. In both age groups, DM was more prevalent in Hispanics compared with other groups; dyslipidemia was more prevalent in patients of Asian/Pacific Islander race; hypertension, obesity, and drug abuse were more prevalent in blacks; and smoking was more prevalent in whites compared with other racial groups (p < 0.001 for individual RF rate comparison across racial groups). In the 18- to 44-year group, dyslipidemia was the most prevalent RF during a first AMI in patients of white (54.1%), Hispanic (50.2%), and Asian/Pacific Islander (55.8%) races, whereas hypertension (63.8%) was the most prevalent RF in blacks. In the 45- to 59-year group, dyslipidemia was the most prevalent RF in patients of white (59.1%) and Asian/Pacific Islander (61.6%) races, whereas hypertension was the most prevalent RF in blacks (74.4%) and Hispanics (64.8%).
We compared sex differences in RF prevalence in individual racial groups to identify any potential sex–race interaction contributing to observed sex differences in the prevalence of individual RFs across age groups. Results are presented in Online Figure 1 and Online Table 2 for both age groups combined, in Online Table 3 for the 18- to 44-year age group, and in Online Table 4 for the 45- to 59-year age group. Overall, observed sex differences across racial groups were generally consistent with those observed without race stratification.
Temporal trends in RF prevalence
Compared with 2005, the prevalence of all these 6 modifiable RFs increased in 2015 among young adults during a first AMI (ptrend < 0.001) (Central Illustration). In the 18- to 44-year group, between 2005 and 2015, rates of hypertension increased from 42.7% to 57.2%, dyslipidemia from 47% to 51.3%, smoking from 54.5% to 60.3%, DM from 18.8% to 26.1%, obesity from 14.6% to 28.9%, and drug abuse from 8.8% to 11.9% (ptrend < 0.001 for all RFs). In the 45- to 59-year group, between 2005 and 2015, rates of hypertension increased from 53.9% to 65.9%, dyslipidemia from 50.5% to 60.3%, smoking from 46% to 56.9%, DM from 26.1% to 31.7%, obesity from 12.2% to 22.1%, and drug abuse from 3.1% to 6.2% (ptrend < 0.001 for all RFs). The greatest relative increase in prevalence rate between 2005 and 2015 was observed for obesity (98% increase) in the 18- to 44-year group, and for drug abuse (100%) followed by obesity (73%) in the 45- to 59-year group. In both the age groups, the rate of dyslipidemia increased through 2012 and 2013, and then gradually decreased in 2015 (Figure 1, Central Illustration).
Similar increases in prevalence rates were noted for all the RFs in the sex-specific analysis across age groups (Figures 2 and 3). Prevalence rates of dyslipidemia increased from 2005 through 2012 and 2013, and then gradually decreased in 2015 in both men and women across age groups. In young adults 45 to 59 years of age, the rate difference between men and women narrowed for hypertension, smoking, and drug abuse during the study period (Figure 3). For hypertension, women 45 to 59 years of age had a 6.1% higher prevalence rate in 2005 and only 1.9% higher rate in 2015, compared with men. For smoking, men had a 5.8% higher rate in 2005 and only 0.6% higher rate in 2015, compared with women. For drug abuse, men had a 0.9% higher rate in 2005 and 0.2% higher rate in 2015.
Temporal changes in the prevalence rate of RFs in race subgroups are presented in Figures 4 and 5. Although the prevalence rates of all RFs increased over time in the 45- to 59-year group, in patients 18 to 44 years of age, the rates of drug abuse in blacks (ptrend = 0.43) and Hispanics (ptrend = 0.246), and smoking in patients of Asian/Pacific Islander race (ptrend = 0.99), remained relatively stable. Across age groups, consistently higher prevalence rates were noted for hypertension, drug abuse, and obesity in blacks, smoking in whites, and DM in Hispanics between 2005 and 2015. Prevalence of dyslipidemia in patients of Asian/Pacific Islander race was consistently higher than in other race subgroups in the 45- to 59-year-old patient group. Across age groups, consistently lower prevalence rates were noted for drug abuse and obesity in patients of Asian/Pacific Islander race, DM in whites, and dyslipidemia in blacks.
In this contemporary study in a large, nationally representative sample of first AMI events in young adults in the United States, the prevalence rates of major modifiable RFs were very high, with 90% patients having at least 1 RF (Central Illustration). These rates progressively increased between 2005 and 2015, except for dyslipidemia, which decreased modestly in more recent years. Sex differences in the rates of certain RFs were clearly evident, with men having higher rates of smoking, dyslipidemia, and drug abuse, whereas women had higher rates of DM, hypertension, and obesity. Importantly, sex differences in the rates of certain RFs narrowed with increasing age and over time. Important racial differences exist in the prevalence of certain RFs, with blacks having higher rates of hypertension, obesity, and drug abuse, whites having higher rates of smoking, Hispanic having higher rates of DM, and patients of Asian/Pacific Islander race having higher rates of dyslipidemia. These differences persisted over time.
RFs play an important role in the development of and progression of CHD, thus necessitating strategies to address the leading modifiable RFs to reduce the burden of CHD (2,5,23). Annual AMI rates and mortality from AMI decreased recently in the United States (20,24–26), but the prevalence of modiﬁable RFs, especially obesity, DM, and hypertension, has increased in these patients in recent times (2,20,26). In our study of first AMIs in a relatively young portion of the U.S. population, we demonstrated a substantial burden of modifiable RFs with almost 92% patients having at least 1 RF at the time of development of AMI. Although younger patients (18 to 44 years) with AMI typically have a strong family history, 90% of such patients in our study had at least 1 modifiable RF, most commonly dyslipidemia. Of first AMI patients 45 to 59 years of age, more than one-half had hypertension, dyslipidemia, or smoking.
The high prevalence of certain RFs in the 18- to 44-year patient group, particularly hypertension (49.8%) and dyslipidemia (51.7%), is very concerning. These numbers are substantially higher than the recent (2013 to 2016) estimates of hypertension (men 12.8%; women 9.4%) and dyslipidemia (men 13.1%; women 8.2%) in the general U.S. adult population 20 to 44 years of age (27). Similarly, the rate of DM in the 18- to 44-year patient group in our study was 22.6%, with a temporal trend toward an increase, which is substantially higher than the 4% DM prevalence rate in U.S. adults 20 to 44 years of age (2011 to 2014 estimate) (27). These findings highlight the alarming rates of modifiable RFs in a subset of U.S. young adults who are at a very high risk for CHD and atherosclerotic cardiovascular disease.
After a first AMI in adults above 45 years of age, the 5-year risk of recurrent AMI or fatal CHD events is as high as 17% to 20%, and an increased risk of such events has been associated with the presence of particular modifiable RFs (2,28). Thus, the very high prevalence of these RFs in young patients puts them at a substantial risk for future CHD and heart failure, thereby contributing to the increased morbidity, mortality, and health care costs associated with CHD. In a pooled survival analysis of 905,115 person-years of data, the lifetime risk estimates for total cardiovascular disease were high (>30%) for all individuals, even those with optimal RFs in middle age (29). However, maintenance of optimal RF levels in middle age was associated with substantially longer morbidity-free survival and 14 years longer survival free of total cardiovascular disease compared with those with 2 or more RFs (29). The increasing prevalence of such RFs during a first AMI as noted in our study is concerning and has important health care implications with regard to planning appropriate secondary preventative strategies to mitigate the increased risk of recurrent CHD in these high-risk patients (23).
In our study, the prevalence of obesity has almost doubled between 2005 and 2015 in young adults with a first AMI. In patients 18 to 44 years of age, rates of obesity surpassed those of DM in 2013. These findings suggest an overall increase in the metabolic risk in younger adults in the United States, apart from that associated with DM, that could further contribute to CHD events (30). Encouragingly, rates of dyslipidemia have decreased slightly between 2012 and 2015, but this trend will require confirmation through analysis of more recent data.
Sex differences in the presence of CHD RFs were extensively studied and have been linked to the increased mortality in women from AMI (4,15,16,31). In the INTERHEART study of 27,098 participants (25% women) with a first AMI enrolled in 52 countries, women had a first AMI 9 years later than men (median age 65 years vs. 56 years; p < 0.0001), a finding that was attributed to higher levels of RFs in men before the age of 60 years compared with women (4). In our study in which there was no clinically meaningful difference in the mean age between young men and women hospitalized for a first AMI, we demonstrated a higher prevalence of certain modifiable RFs such as DM, hypertension, and obesity in women, rates of which remained higher during the time period studied. Importantly, among RFs for which men had higher prevalence rates (smoking, dyslipidemia, drug abuse), the rate differences across sex categories were almost 50% lower in patients 45 to 59 years of age than those 18 to 44 years of age. Also, over time, the sex gap narrowed for smoking, hypertension, and drug abuse. These findings suggest that the baseline risk in younger women having a first AMI event is increased with age and also over time, highlighting the need for appropriate primordial, primary, and secondary prevention strategies in women. Additionally, findings from the VIRGO (Variation in Recovery: Role of Gender on Outcomes of Young AMI Patients) study showed that young women were less likely than men to discuss risk-modification strategies with health care providers after a first AMI (3), further stressing the need for increased RF preventative education in young women.
Racial differences were prominent among prevalence rates of CHD RFs. Blacks had higher prevalence rates of most studied RFs compared with other racial groups in our study. Certain racial differences observed in our study were consistent with previously reported findings of higher rates of hypertension in blacks, DM in Hispanics, and smoking in whites (17,18,32–34). In our study, rates of hypertension were particularly high among blacks, with 3 of 4 patients 45 to 59 years of age having hypertension. In our study, the prevalence rate of dyslipidemia was higher in patients of Asian/Pacific Islander race, and drug abuse was substantially higher in blacks, during a first AMI in young adults. The differences in prevalence of RFs observed among racial groups in this study may be of use in the development of more effective and tailored risk prevention strategies. Importantly, observed sex differences in the prevalence of individual RFs were generally consistent across racial groups, underlining the importance of developing tailored and holistic approaches to cardiovascular risk prevention and management. Health disparities in CHD is a topic of great interest as such differences can influence the incidence, health care delivery, and outcomes of CHD in different population groups (31). Emerging data should facilitate the understanding of the pathophysiological and sociodemographic differences responsible for such disparities, to improve overall cardiovascular outcomes.
Important strengths of our study are the use of a large, nationally representative sample from the NIS to demonstrate national-level estimates of accurate rates of RFs. Further, we focused only on first AMIs to more fully understand the burden of modifiable RFs that led to the development of an initial CHD event. To the best of our knowledge, this is the first and largest contemporary study to report such findings.
Being an administrative database, NIS does not provide information on clinical or laboratory level data like blood pressure readings, body mass index, or cholesterol levels. Miscoding of diagnostic codes resulting in misclassification bias, coding inconsistencies across hospitals, and upcoding across time might have also contributed to the observed trends in prevalence (35). Different states included in the NIS reported race data variably during the time period of our study (36). We included only cases with available race data when analyzing racial differences. We did not analyze the prevalence of nontraditional risk RFs, or evaluate outcomes based on RF profiles, as these were outside the scope of our study.
Among young adults in the United States hospitalized for a first AMI, the prevalence rates of major modifiable RFs were very high, with over 90% of patients having at least 1 such RF. Significant sex and racial disparities were observed. Prevalence rates progressively increased between 2005 and 2015 except for dyslipidemia, for which a decreasing trend was noted more recently. Modifiable atherosclerotic RFs were highly prevalent in these young patients who had preventive measures are more likely to be effective, and these data can be used to plan appropriate preventative strategies in select populations to help reduce the burden of CHD.
COMPETENCY IN SYSTEMS-BASED PRACTICE: More than 90% of patients presenting with a first myocardial infarction before 60 years of age in the United States have ≥1 modifiable cardiac risk factor, with smoking, dyslipidemia, and hypertension the most prevalent. Although there are significant disparities related to sex and race, the prevalence of these risk factors has been increasing over time.
TRANSLATIONAL OUTLOOK: Emerging data should improve understanding of the sociodemographic factors responsible for sex- and race-based disparities in the prevalence of cardiovascular risk factors that could be used to improve preventive care.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Listen to this manuscript's audio summary by Editor-in-Chief Dr. Valentin Fuster on JACC.org.
- Abbreviations and Acronyms
- acute myocardial infarction
- coronary heart disease
- diabetes mellitus
- Healthcare Cost and Utilization Project
- myocardial infarction
- National Inpatient Sample
- non–ST-segment elevation myocardial infarction
- risk factor
- ST-segment elevation myocardial infarction
- Received July 21, 2018.
- Revision received October 23, 2018.
- Accepted October 29, 2018.
- 2019 American College of Cardiology Foundation
- Benjamin E.J.,
- Blaha M.J.,
- Chiuve S.E.,
- et al.
- Kayani W.T.,
- Ballantyne C.M.
- Leifheit-Limson E.C.,
- D’Onofrio G.,
- Daneshvar M.,
- et al.
- Fryar C.D.,
- Chen T.-C.,
- Li X.
- Shah B.,
- Bangalore S.,
- Gianos E.,
- et al.
- Mittleman M.A.,
- Mintzer D.,
- Maclure M.,
- Tofler G.H.,
- Sherwood J.B.,
- Muller J.E.
- Beck C.A.,
- Southern D.A.,
- Saitz R.,
- Knudtson M.L.,
- Ghali W.A.
- American Heart Association
- DeFilippis E.M.,
- Singh A.,
- Divakaran S.,
- et al.
- Gupta T.,
- Kolte D.,
- Khera S.,
- et al.
- Gupta A.,
- Wang Y.,
- Spertus J.A.,
- et al.
- ↵Healthcare Cost and Utilization Project (HCUP). NIS Overview. August 2018. Agency for Healthcare Research and Quality. Available at:. https://www.hcup-us.ahrq.gov/nisoverview.jsp. 232018.
- Agresti A.
- Burke G.L.,
- Sprafka J.M.,
- Folsom A.R.,
- Luepker R.V.,
- Norsted S.W.,
- Blackburn H.
- Agarwal S.,
- Sud K.,
- Thakkar B.,
- Menon V.,
- Jaber W.A.,
- Kapadia S.R.
- Centers for Disease Control and Prevention (CDC)
- Yandrapalli S.,
- Jolly G.,
- Horblitt A.,
- Sanaani A.,
- Aronow W.S.
- Mensah G.A.,
- Mokdad A.H.,
- Ford E.S.,
- Greenlund K.J.,
- Croft J.B.
- Steinwald B.,
- Dummit L.A.
- Healthcare Cost and Utilization Project (HCUP)