CLINICAL EPIDEMIOLOGY www.jasn.org

Stroke and the “Stroke Belt” in Dialysis: Contribution of Patient Characteristics to Ischemic Stroke Rate and Its Geographic Variation

† ‡ James B. Wetmore,* Edward F. Ellerbeck, Jonathan D. Mahnken,§ Milind A. Phadnis,§ | Sally K. Rigler, ¶ John A. Spertus,** Xinhua Zhou,§ Purna Mukhopadhyay,§ and †‡ Theresa I. Shireman

*Department of Medicine, Division of Nephrology and , † The Kidney Institute, ‡Department of Preventive Medicine and , §Department of Biostatistics, |Department of Medicine, and ¶The Landon Center on Aging, University of Kansas School of Medicine, Kansas City, Kansas, and **St. Luke’s Mid America Heart Institute, University of Missouri-Kansas City, Kansas City, Missouri

ABSTRACT Geographic variation in stroke rates is well established in the general population, with higher rates in the South than in other areas of the United States. ESRD is a potent risk factor for stroke, but whether regional variations in stroke risk exist among dialysis patients is unknown. Medicare claims from 2000 to 2005 were used to ascertain ischemic stroke events in a large cohort of 265,685 incident dialysis patients. A Poisson generalized linear mixed model was generated to determine factors associated with stroke and to ascertain state-by-state geographic variability in stroke rates by generating observed-to-expected (O/E) adjusted rate ratios for stroke. Older age, female sex, African American race and Hispanic ethnicity, unemployed status, , hypertension, history of stroke, and permanent atrial fibrillation were positively associated with ischemic stroke, whereas body mass index .30 kg/m2 was inversely associated with stroke (P,0.001 for each). After full multivariable adjustment, the three states with O/E rate ratios .1.0 were all in the South: , , and Oklahoma. Regional efforts to increase primary prevention in the “stroke belt” or to better educate dialysis patients on the signs of stroke so that they may promptly seek care may improve stroke care and outcomes in dialysis patients.

J Am Soc Nephrol 24: 2053–2061, 2013. doi: 10.1681/ASN.2012111077

Stroke is a catastrophic health event and a leading patients across the United States have consistent cause of disability. It represents a particularly heavy access to insurance and frequent contact with health burden for the long-term dialysis population, in care providers, who routinely measure their BP, whom stroke rates are substantially higher than in irrespective of geographic location. Second, the the general population.1 In the general population, nature of vascular disease differs between dialysis there is substantial geographic variability in stroke and nondialysis patients, so different pathophysio- rates, with the southeastern United States having logic mechanisms may be operative in the two long been recognized as a “stroke belt” of higher populations.5 Third, dialysis patients fundamentally stroke mortality rates.2–4 However, whether a stroke belt of increased ischemic stroke incidence exists in dialysis patients has not been formally Received November 12, 2012. Accepted July 8, 2013. studied. Published online ahead of print. Publication date available at Althoughone might suspect that the same factors www.jasn.org. contributing to ischemic stroke risk in the general Correspondence: Dr. James B. Wetmore, Division of Nephrol- population also apply to dialysis patients, there are ogy, Hennepin Medical Center, 701 Park Avenue, Min- several reasons to posit that this might not be the neapolis, MN 55415. Email: [email protected] case. First, unlike the general population, dialysis Copyright © 2013 by the American Society of Nephrology

J Am Soc Nephrol 24: 2053–2061, 2013 ISSN : 1046-6673/2412-2053 2053 CLINICAL EPIDEMIOLOGY www.jasn.org represent a “survivor cohort” relative to individuals with (pre- sex, African American race and Hispanic ethnicity, and being dialysis) CKD and its attendant cardiovascular disorders, sug- unemployed at the time of dialysis initiation were associated gesting that epidemiologic trends evident in one population with ischemic stroke (P,0.0001 for all). A body mass index might not be found in the other.6 Accordingly, it is uncertain of $30 kg/m2 was associated with a significantly lower rate of whether there is substantial geographic variation in stroke risk ischemic . In terms of comorbid conditions, perma- among dialysis patients and what factors might, in part, explain nent atrial fibrillation (AF) (treated as a time-dependent vari- such a finding. able), diabetes, hypertension, and history of a cerebrovascular To address this gap in knowledge, we constructed a large accident were also associated with ischemic stroke (P,0.0001). cohort of incident dialysis patients to determine whether State of residence was also associated with ischemic stroke in- ischemic stroke rates vary by geography and how differences in dependent of other factors, as described below. stroke rates might be explained by patient characteristics. We reasoned that uncovering the existence of geographic variabil- Geographic Factors Associated with Stroke ity in the stroke rates of dialysis patients might provide Figure 2 and Table 3 demonstrate geographic variation in is- direction for focused health care efforts in regions at elevated chemic stroke rates under various modeling strategies. Figure risk, such as screening new dialysis patients for symptoms that 2A shows observed-to-expected (O/E) ratios after adjustment might be referable to old strokes, lowering the threshold for only for age; Figure 2B, after adjustment for age and sex; Figure investigating cerebrovascular disease, or educating dialysis 2C, after adjustment for age, sex, and race; and Figure 2D, after patients on the importance of seeking immediate care for full adjustment for factors listed Table 2. Figure 2A demon- stroke-type symptoms. strates that stroke rates are highest predominantly in the southern and southeastern United States, although the rate is also high in ; eight states had O/E ratios significantly RESULTS above unity, seven of which were in the South. After additional adjustment for sex (Figure 2B), all seven of the states with the Cohort Characteristics rates significantly .1.0 were in the South, and after further Figure 1 shows the construction of the Medicare-eligible co- adjustment for race (Figure 2C), five of six states were in the hort. There were a total of 265,685 Medicare-eligible individ- South. After full adjustment (Figure 2D), three states re- uals who initiated dialysis between January 1, 2000, and mained, all of which are in the South: The O/E ratio for North October 2, 2005, and survived at least 90 days before our final Carolina was 1.15 (99% confidence interal [CI], 1.04 to 1.27); date of December 31, 2005. for Oklahoma, 1.16 (99% CI, 1.01 to 1.34); and for Missis- The characteristics of the Medicare-eligible cohort are sippi, 1.18 (99% CI, 1.03 to 1.34). Table 3 demonstrates the shown in Table 1. Mean age 6 SD was 64.7615 years; 52.9% of same phenomenon, facilitating comparison between states as the patients were male; and whites made up the largest group the various modeling strategies were undertaken; it shows that at 54.9%, followed by at 30.0%. Diabetes, the effect of state is modified by more thorough statistical at 47.1%, was the leading cause of ESRD. In terms of comorbid adjustment. Of note, employment rate in the eight states conditions, 84.4% of patients had hypertension, 33.2% had with O/E ratios .1wassignificantly, but modestly, lower heart failure, and 10.3% had a history of a cerebrovascular than in the remaining states (4.2% versus 5.0%; P,0.0001). accident upon dialysis initiation. More than 93% were under- In the unadjusted model, only New Mexico had an O/E ratio going in-center hemodialysis. The cohort was followed for a ,1 (0.74 [99% CI, 0.58 to 0.95]) while in the fully adjusted mean of 2.0 years. Bivariate analyses between individuals who model, only had an O/E ratio ,1 (0.90 [99% CI, did and did not have strokes during the observation period 0.82 to 0.98]), demonstrating that, overall, variation .1.0 was revealed significant differences (P,0.01) for all covariates far more common than variation ,1.0. examined with the exception of hemoglobin (P=0.05). Sensitivity Analyses Stroke Events To assess the rigor of our analysis, we performed multiple Of 265,685 individuals, 13,073 (4.9%) experienced at least one sensitivity analyses. First, we performed identical modeling stroke. Total ischemic stroke events numbered 14,240: Of save elimination of the adapted Liu comorbidity index; final individuals with a stroke, 91.9% had one stroke, 7.3% had two, results were identical, with North Carolina, Mississippi, and and 0.8% had three or more. Total follow-up time was 431,049 Oklahoma again being the only states with O/E ratios for patient-years, resulting in a rate of 33 ischemic strokes per 1000 ischemic stroke significantly .1.0. Next, we used a more sen- patient-years. Stroke ratesweregenerallyquite stable over time, sitive definition for ischemic strokes in which an additional as shown in Supplemental Table 1. 20% of strokes were included. Five (, Mississippi, North Carolina, Oklahoma, ) of the seven Person-Level Factors Associated with Stroke (New Jersey, Indiana) states with O/E ratios significantly After multivariable adjustment, factors independently associ- .1.0 were in the South; the same general trend of progressive ated with ischemic stroke are shown Table 2. Older age, female attenuation with greater adjustment was observed, starting

2054 Journal of the American Society of Nephrology J Am Soc Nephrol 24: 2053–2061, 2013 www.jasn.org CLINICAL EPIDEMIOLOGY

differences in patient characteristics could account for such variability. After adjust- ment for age alone (as the single most important factor associated with stroke), there was a distinct clustering of ischemic strokes in southern states, suggesting the presence of a “stroke belt” in long-term di- alysis patients (of the states with increased risk, only Indiana was not in the South). Viewed another way, compared with the five states with the lowest stroke rates (none of which were in the South), the five states with the highest rates (all of which were in the South) had a 28% in- crease in strokes. However, as successive layers of adjustment were introduced (namely sex; race; and a comprehensive set of variables encompassing demo- graphic, functional status, and comorbidity factors), the effect of geography was se- quentially diminished. After full adjust- ment, the three states with significantly increased stroke rates (North Carolina, Mississippi, and Oklahoma) were all in the . Thus, while pa- tient characteristics largely explain the in- creased risk of stroke in dialysis patients, a significantly increased risk is associated with residing in the stroke belt. Our main findings were robust to numerous sensitivity analyses. The phenomenon of a stroke belt was first described as far back as 19654 and has been a continuing focus of public health Figure 1. The study cohort as derived from the master sample. Exclusion flowchart research in the United States in the decades demonstrating the creation of the study cohort. VA, Veterans Affairs. hence.2,3,7–10 These studies primarily ex- amined stroke mortality, rather than inci- from 14 (predominantly southern) states with O/E ratios sig- dence, and have varied considerably in their definitions of the nificantly .1.0 when adjustment for age alone was under- stroke belt: For example, some studies have included taken. Finally, we modeled ischemic strokes (using the specific and even Indiana,10 while others8,11 have not. We posited that definition for ischemic strokes and including the adapted Liu because stroke mortality is probably related to stroke rates, index) in the dually eligible (Medicare and Medicaid) popu- there might well be increased stroke rates in dialysis patients lation; no states had O/E ratios significantly .1.0, probably residing in the southern United States. However, because there due part to the more limited power of a sample size that was was variability in both analytical approaches and findings in only 28.6% as large, but there was a clear trend toward higher theliterature,wehadnoapriorihypothesis as to precisely point estimates, on average, for states in the stroke belt com- which states might have higher-than-expected rates. Further, pared with those outside it. For example, Mississippi, North we did not necessarily expect that patterns in the general pop- Carolina, and Virginia had the highest point estimates in both ulation would be explicitly replicated in the dialysis popula- the fully adjusted model and the model adjusted for age alone. tion. This study was merely designed to determine whether an ischemic stroke signal in dialysis patients would emerge from the southern United States. DISCUSSION Of note, evidence for a clustering of other cardiovascular disorders, such as congestive heart failure,12 has also been de- In this study, we sought to determine whether there was geo- scribed in the southern and southeastern United States. In- graphic variability in stroke rates by state and to what degree deed, these issues are of sufficient public health interest such

J Am Soc Nephrol 24: 2053–2061, 2013 Geographic Variation in Stroke in Dialysis 2055 CLINICAL EPIDEMIOLOGY www.jasn.org

Table 1. Descriptive characteristics of the Medicare-eligible cohort Characteristic All Stroke No Stroke P Valuea Cases (n) 265,685 13,073 252,612 Age (yr) 64.7615.1 69.4611.9 64.5615.2 ,0.0001 Male, n (%) 140,599 (52.9) 5406 (41.4) 135,193 (53.5) ,0.0001 Race/ethnicity, n (%) ,0.0001 African American 79,693 (30.0) 4336 (33.2) 75,357 (29.8) White 145,768 (54.9) 7030 (53.8) 138,738 (54.9) Hispanic 27,738 (10.4) 1248 (9.6) 26,490 (10.5) Other 12,486 (4.7) 459 (3.5) 12,027 (4.8) BMI category, n (%) ,0.0001 , 20 kg/m2 25,710 (9.7) 1321 (10.1) 24,389 (9.7) 20–24.9 kg/m2 84,154 (31.7) 4330 (33.1) 79,824 (31.6) 25–29.9 kg/m2 76,043 (28.6) 3840 (29.4) 72,203 (28.6) 30+kg/m2 79,778 (30.0) 3582 (27.4) 76,196 (30.2) Smoker, n (%) 14,893 (5.6) 608 (4.7) 14,285 (5.7) ,0.0001 Substance abuser, n (%) 5,370 (2.2) 157 (1.2) 5573 (2.2) ,0.0001 Unemployed, n (%) 252,749 (95.1) 12,846 (98.3) 239,903 (95.0) ,0.0001 Unable to ambulate, n (%) 11,443 (4.3) 745 (5.7) 10,698 (4.2) ,0.0001 Unable to transfer, n (%) 4075 (1.5) 295 (2.3) 3780 (1.5) ,0.0001 In-center HD, n (%)b 247,584 (93.2) 12,258 (93.8) 235,326 (93.2) 0.007 Hemoglobin , 11.0 g/dl, n (%) 176,664 (72.9) 8759 (73.6) 167,905 (72.83) 0.05 Comorbid conditions, n (%) AF 38,619 (14.5) 2863 (21.9) 35,756 (14.2) ,0.0001 Hypertension 224,159 (84.4) 11,410 (87.3) 212,749 (84.2) ,0.0001 Diabetes mellitus 141,067 (53.1) 8050 (61.6) 133,017 (52.7) ,0.0001 Congestive heart failure 88,226 (33.2) 4990 (38.2) 83,236 (33.0) ,0.0001 Coronary artery disease 74,734 (28.1) 4302 (32.9) 70,431 (27.9) ,0.0001 Peripheral vascular disease 40,662 (15.3) 2431 (18.6) 38,231 (15.1) ,0.0001 Prior cerebrovascular accident 27,304 (10.3) 3055 (23.4) 24,249 (9.6) ,0.0001 Liu comorbidity index score 5.162.8 5.862.8 5.0162.8 ,0.0001 Cause of ESRD, n (%) ,0.0001 Diabetes 125,220 (47.2) 7178 (54.9) 118,042 (46.7) Hypertension 70,117 (26.4) 3480 (26.6) 66,637 (26.4) GN 22,817 (8.6) 604 (4.6) 225,213 (8.8) Other 47,531(17.9) 1811 (13.9) 465,720 (18.1) aFor the comparison between individuals who did and did not experience a stroke. bIn-center hemodialysis is contrasted to self-care dialysis, which consists of peritoneal dialysis and home hemodialysis. that the Centers for Disease Control and Prevention maintains epidemiology.”13,14 In addition, patients undergoing long- an interactive website that can calculate risk of cardiovascular term dialysis have close contact with the health care system, events, adjusted for age, sex, and race, at the state and even which includes routine measurement of BP and provides am- county level (http://apps.nccd.cdc.gov/DHDSPAtlas). How- ple opportunity to assess for the presence of stroke-related ever, much of the pioneering work on the stroke belt was symptoms. Nevertheless, our findings were generally concor- undertaken without full consideration of how patient-level dant with previous work showing that regional differences in characteristics might be responsible for geographic variation patient characteristics are responsible for the stroke belt in the in stroke. We sought not only to determine the degree of geo- general population.10 As was done by those investigators, we graphic variability present in stroke rates in dialysis patients elected not to model merely the characteristics directly related butalsotodiscerntowhatdegreepatientfactorsmightbe to stroke but also a broad range of potentially contributing responsible for this phenomenon. factors.10 This was done to minimize confounding by dialysis The present report appears to be the first such focused study patients’ clinical characteristics that might differ across states. on geographic variability in stroke rates in the dialysis pop- It is important to note that although the signal of a stroke belt ulation. That stroke would be more common in dialysis diminishes with successive layers of statistical adjustment, pa- patients residing in the United States was not a foregone con- tient characteristics are likely to be inherent to various geo- clusion because the epidemiology of graphic regions and may therefore be only weakly modifiable, in dialysis patient differs substantially from that of the general if at all. In this sense, less adjusted models may be more in- population, a phenomenon controversially known as “reverse sightful in helping to formulate public health strategies

2056 Journal of the American Society of Nephrology J Am Soc Nephrol 24: 2053–2061, 2013 www.jasn.org CLINICAL EPIDEMIOLOGY

Table 2. Patient-level factors associated with ischemic other medical risk factors. Additionally, the role of body mass stroke index does not seem to have been previously explored. While Variable ARR (99% CI) P value an inverse relationship exists between body mass index and 13 Age, per year 1.02 (1.01 to 1.02) ,0.0001 mortality in the dialysis population, it was unclear to us Female sex 1.50 (1.44 to 1.57) ,0.0001 whether this would be the case for ischemic stroke. That we Race/ethnicity found this to be the case suggests that the phenomenon of White 1 (reference) reverse epidemiology might affect stroke risk as well. African American 1.28 (1.22 to 1.35) ,0.0001 In terms of comorbid conditions, diabetes15–18 and a his- Hispanic 1.15 (1.06 to 1.24) ,0.0001 tory of a cerebrovascular accident16,18 seem to be reliably Other 0.86 (0.77 to 0.97) 0.0016 associated with stroke risk, concordant with our findings. , BMI category 0.0001 However, the potential role of other traditional risk factors , 2 20 kg/m 0.99 (0.92 to 1.07) 0.76 is somewhat unclear. Hypertension, for example, has been 20–24.9 kg/m2 1 (reference) 15 18 2 associated with stroke in some but not all studies, and 25–29.9 kg/m 0.96 (0.91 to 1.01) 0.027 fi 17,19–21 $30 kg/m2 0.82 (0.77 to 0.86) ,0.0001 even the ndings on the role of AF are discordant, al- Smoker 1.04 (0.94 to 1.15) 0.31 though the most comprehensive study to date reported that, in Substance abuser 0.87 (0.72 to 1.04) 0.05 the setting of AF, dialysis patients had a 1.8-fold increased risk Unemployed 1.66 (1. 42 to 1.95) ,0.0001 of stroke compared with non–dialysis-dependent individu- Inability to ambulate 0.92 (0.82 to 1.03) 0.06 als.22 Such discrepancies in the literature suggested that even Inability to transfer 1.05 (0.89 to 1.25) 0.45 basic relationships between common comorbid conditions In-center hemodialysisa 1.09 (1.01 to 1.19) 0.0060 and stroke risk are not fully understood in the dialysis popu- Comorbid conditions lation. We suggest that our study, the largest of its kind to date, , AF 1.40 (1.33 to 1.48) 0.0001 may help clarify these relationships. , Hypertension 1.25 (1.16 to 1.35) 0.0001 Our findings should be interpreted in the context of Diabetes mellitus 1.13 (1.06 to 1.21) ,0.0001 important limitations. First, our outcomes were based on Congestive heart failure 0.95 (0.88 to 1.01) 0.035 Coronary artery disease 0.96 (0.92 to 1.01) 0.064 claims, rather than clinical data such as degree of BP control. Peripheral vascular disease 1.03 (0.97 to 1.09) 0.28 Although our claims-based approach is imperfect, it seems Prior cerebrovascular 2.47 (2.34 to 2.61) ,0.0001 unlikely that this would introduce bias in a way that varies by accident geography. Second, consistent with many approaches,23 we Liu comorbidity score primarily employed stroke codes that used only the first three 0–3 1 (reference) digits. This could result in inclusion of events that were not 4–6 1.00 (0.94 to 1.07) 0.92 infarcts. However, .98% of the codes used in our primary $7 0.96 (0.86 to 1.07) 0.34 definition did not have the fifth position “0.” It therefore likely ARR, adjusted rate ratio; CI, confidence interval. that the overwhelming majority of events captured were is- aIn-center hemodialysis is contrasted to self-care dialysis, which consists of peritoneal dialysis and home hemodialysis. chemic strokes, especially when our analytic safeguards based on length of stay and the presence of a carotid endarterectomy are considered. Furthermore, our sensitivity analysis, which designed to address risk factors that are found more com- included code 433, was deliberately designed to be more lib- monly in the southern United States. eral; the same general pattern of increased strokes in the south Relatively few studies have explicitly examined factors was found. We did not examine hemorrhagic strokes, which associated with ischemic strokes. As with our study, most are also common in the dialysis population; whether a stroke investigators15–17 (but not all18) show age to be a significant belt exists for hemorrhagic strokes should also be a subject of risk factor associated with stroke, a finding that would be ex- future study. pected from the general population. The role of sex has been Additionally, our primary analysis used only individuals less fully explored; one study showed no association between who were Medicare-eligible from the start of dialysis, which sex and stroke,15 whereas another suggested a trend in which means they were likely to be older than individuals who ac- female sex was associated with stroke.16 The role of race is yet quired Medicare at a later date; our results may not be more difficult to characterize: Seliger et al.15 showed that generalizable to the general dialysis population. Another African American race was associated with new strokes limitation is that our claims-based algorithm was likely capable (ischemic and hemorrhagic combined) only in the setting of of detecting only individuals with permanent, as opposed to previously established cardiovascular disease; in individuals paroxysmal, AF. Because paroxysmal AF, which constitutes a without preexisting disease, African America race was associ- stroke risk similar to that provided by permanent AF, is much ated with lower stroke rates. Sozio et al.16 also showed that African harder to ascertain from claims data, our analysis is likely to American race was inversely associated with stroke risk. These underestimate the stroke risk conferred by all types of AF. AF findings, combined with our own, suggest that the role of race was treated with special scrutiny, as it was treated as a time- in stroke risk may be nuanced and dependent on a milieu of dependent covariate; in reality, patients accumulate comorbid

J Am Soc Nephrol 24: 2053–2061, 2013 Geographic Variation in Stroke in Dialysis 2057 CLINICAL EPIDEMIOLOGY www.jasn.org

Further studies should investigate whether regional differences in treatment ap- proaches, which seem likely to exist, are potentially responsible for our findings. These limitations are probably counter- balanced by the large sample size, the richness of the data used (i.e., the Centers for Medicare & Medicaid [CMS] 2728 form combined with claims data), and our use of multiple sensitivity analyses, which gener- ally supported the findings of the primary analysis. We suggest that the analytic co- hort we constructed is likely to be broadly similar to more contemporary dialysis pa- tients. For example, data from the U.S. Renal Data System1 suggest that racial composition of dialysis patients is generally stable. Hypertension as a cause of ESRD has become slightly more common, and GN slightly less common. Other subtle changes are that the mean age of dialysis Figure 2. Strokes are generally more common in the southern United States. States initiation has increased slightly, as has the with O/E adjusted odds ratios significantly .1 for new ischemic stroke, after successive mean estimated GFR at initiation. Given adjustments. (A) Adjusted for age. (B) Adjusted for age and sex. (C) Adjusted for age, general stability in the United States dialy- sex, and race. (D) Full multivariable adjustment. sis population, our findings are likely to persist into the near future at least. How- events continuously. However, such an analysis would be ever, caution should also be used in attempting to extrapolate extraordinarily complex and would require robust code-based out findings to non–United States populations because stroke algorithms for every comorbidity in order to reliably identify rates appear to vary substantially by country.24 them. Finally, we did not censor at change of dialysis modality, In conclusion, we found significant geographic variability but misclassification due to this is likely to be small, since 95% in stroke rates in long-term dialysis patients. The highest of long-term dialysis patients use in-center HD. rates were generally in southern states. Adjustment for One major potential factor that our analysis was unable patient-level characteristics accounted for the majority of to account for is regional differences in treatment, such as this geographic variability, suggesting that the burden of hypertension treatment, anemia management, and, in the stroke risk factors is higher in the southern states, but strokes case of persons with permanent AF, treatment with . rates remained significantly higher than average in a few southern states. These findings suggest that regionally targeted stroke-related health care efforts, such as screening new di- Table 3. States with an O/E adjusted rate ratios, for alysis patients for symptoms of previous strokes, alerting ischemic stroke remaining significantly .1.0, after successive providers to a have a high index of suspicion for investigating adjustments possible cerebrovascular disease, or educating dialysis patients State Unadjusted Age Age, Sex Age, Sex, Race Fulla about the importance of seeking emergency care for stroke MS 1.27 1.36 1.32 1.19 1.18 symptoms, might be beneficial and should be a focus of future NC 1.28 1.35 1.32 1.22 1.15 study. OK 1.18 1.22 1.20 1.21 1.16 AL 1.21 1.27 1.25 1.16 – SC 1.24 1.31 1.27 1.15 – CONCISE METHODS IN 1.16 1.15 – 1.14 – GA 1.14 1.20 1.17 –– LA 1.15 1.20 1.18 ––Study Design and Data Sources for Analysis We performed a retrospective cohort analysis of incident, Medicare- Significance was maintained for each state above using 99% confidence in- tervals. Note that Indiana is considered as being outside the “stroke belt” in eligible long-term dialysis patients. We also performed a secondary some studies, but has been included in it by others. MS, Mississippi; NC analysis in a subcohort of dually eligible (Medicare and Medicaid) North Carolina; OK, Oklahoma; AL, Alabama; SC, South Carolina; IN, In- long-term dialysis patients; these patients are a particularly vulnerable diana; SC, South Carolina; IN, Indiana; GA, ; LA, . aAdjusted for age, sex, and race, as well as for the comorbidity and other group who have similar socioeconomic status, which enables us to factors shown in Table 2. minimize the potential confounding from this factor. Medicare is a

2058 Journal of the American Society of Nephrology J Am Soc Nephrol 24: 2053–2061, 2013 www.jasn.org CLINICAL EPIDEMIOLOGY federally funded program for which nearly all adults with ESRD are For the geographic analysis, the southern states were considered to entitled, regardless of age; although not all individuals receiving long- be those from three of the nine official U.S. Census Bureau geographic term dialysis are Medicare enrollees, most are. Medicaid, a public divisions: Oklahoma, , , and Louisiana from the “west insurer funded jointly by federal and state governments, provides south central” region; Mississippi, , Alabama, and Ken- broad medicalcare benefitsincluding prescriptiondrugsforlow-income tucky, from the “east south central” region; and the eight southeast- or medically needy patients. The data sources and strategy used for ern states of the “south Atlantic” region. linking Medicare and Medicaid patients have been previously described25,26 and are provided in more detail in Supplemental Appendix 1. Stroke Outcomes Our primary outcome was ischemic stroke rate (taking into account Study Cohort and Rationale for Analytic Approach that some patients had multiple strokes). We used recent information The cohorts consisted of individuals .18 years of age who initiated on the sensitivity and specificity of stroke-related International long-term dialysis on or after January 1, 2000; survived at least 90 Classification of Diseases, Ninth Revision (ICD-9), claims to identify days after dialysis initiation; and were continuously enrolled in Medi- ischemic strokes from Medicare data.23 Aspecific approach, in which care (primary analysis) or both Medicare and Medicaid (secondary only the codes with higher specificities for ischemic strokes were analysis) from dialysis initiation. Last date of enrollment was October used, was the primary approach. A sensitivity analysis was performed 2, 2005 (to permit a minimum of 90 days of possible follow-up). in which we used a broader range of stroke codes. Briefly, to ensure that all Medicare claims were observable, we studied For the primary approach for identifying ischemic strokes, we used only individuals who were Medicare-eligible from the time of dialysis thestrategyofGoetal.28 The appearance of codes 434 or 436 in the initiation and for whom Medicare was the primary payer. Individuals primary position of an inpatient claim was first assessed; if these were were censored at the time that they lost Medicare eligibility, if this not present, the appearance of code 362.3 was sufficient to make the occurred. For the dually eligible cohort (Medicare plus Medicaid) diagnosis of an ischemic stroke. If the code 434 or 436 was present substudy analysis, individuals were censored when they lost Medicare and accompanied by a fatal hospitalization, an ischemic stroke was or Medicaid coverage. Patients enrolled in any form of managed care attributed. If 434 or 436 was present and the hospitalization was plan (e.g., those in Arizona or Tennessee) or in the Department of nonfatal, the length of the hospitalization was considered. If the hos- Veterans Affairs health system were also excluded because claims data pitalization duration was $48 hours, an ischemic stroke was assessed; were not available. Of note, persons undergoing long-term dialysis if the hospitalization was ,48 hours, the presence of a carotid end- were generally not enrolled in Medicare managed care plans prior to arterectomy (ICD-9 code 381.2) was determined. If a carotid endar- 2006. Additional criteria for censoring were receipt of a kidney trans- terectomy was not present an ischemic stroke was assessed, while the plant or death on or before December 31 2005. presence of a carotid endarterectomy meant that the stroke was not assessed. The sensitive approach differed only by treating code 433 Covariates and Descriptive Variables analogously to 434 and 436. Demographic and clinical variables were drawn from the CMS 2728 dialysis intake form, a mandatory requirement of Medicare- Statistical Analyses subsidized dialysis that describes patients’ clinical history. A variety We generated descriptive statistics (means and SDs for continuous of covariates were considered as potential risk factors for ischemic variables and frequencies for categorical variables) to illustrate how stroke (Supplemental Appendix 1). Because previous stroke is a individuals who experienced ischemic strokes differed from those strong predictor of future stroke, we examined the cohort for stroke who did not. Bivariate analyses comparing each of the explanatory claims in the first 90-day run-in period before the start of the ob- variables by use versus nonuse were performed by using the Pearson servation window; an individual was considered to have a preexist- chi-squared test or t test, as appropriate. To identify independent ing stroke if either a history of a cerebrovascular accident (either factors associated with ischemic stroke, we generated a Poisson gen- ischemic or hemorrhagic) was declared on the CMS 2728 form or eralized linear mixed model29 with stroke rate (number of strokes per if a stroke claim appeared during the 90-day run-in period; strokes unit of exposure time) being regressed simultaneously on all apriori occurring after this period (i.e., during the observation window) selected explanatory variables as fixed effects and with state modeled were therefore considered “incident” strokes. Additionally, we sup- as a random effect. Although cause of ESRD was not included among plemented the medical information from the CMS 2728 form with a these aprioriselected variables, we included, in modified form, the modified form of the Liu comorbidity index27 (a summary measure Liu comorbidity index,27 which incorporates cause of ESRD. The of comorbidity burden), as described in Supplemental Appendix 1. time-dependent nature of permanent AF as an explanatory variable Because permanent AF is a unique risk factor for stroke that is was handled by creating two separate rows of data for patients who poorly captured in the CMS 2728 form, requires complex algo- experienced AF, with the first row contributing person-time to the rithms to capture from claims data, and develops over time in a no-AF group and the second row contributing person-time to the AF substantial number of dialysis patients, we treated AF as a time- group. The Poisson model fit was assessed by ensuring that the ratio dependent covariate, an approach that permitted us to determine of the generalized chi-squared test statistic to its degrees of freedom the onset of the disorder relative to new ischemic strokes captured in was close to 1, indicating that there were no major concerns with the observation window; details in this are described in Supplemen- overdispersion in the model (i.e., the observed variability in stroke tal Appendix 1. rates was close to the variability expected under the Poisson model).

J Am Soc Nephrol 24: 2053–2061, 2013 Geographic Variation in Stroke in Dialysis 2059 CLINICAL EPIDEMIOLOGY www.jasn.org

After developing the Poisson generalized linear mixed model, we author(s) and in no way should be seen as an official policy or in- examined the variability in stroke rates by state of residence and terpretation of the U.S. government. compared states. For each state, we determined whether the observed number of incident strokes, called the observed (O) value, was above or below the expected (E) value given the total exposure time for DISCLOSURES persons belonging to that state. The random effect estimates for each None. state calculated by our model facilitated the O/E rate ratio compar- isons. Specifically, we obtained the estimates of the random effects for each state because these variables modify each state’s log-rates of REFERENCES ischemic stroke from the overall cross-state (fixed) model effects. Exponentiation of these estimates generated state-specific observed 1. United States Renal Data System: USRDS 2012 Annual Data Report: versus expected (O/E) adjusted rate ratios adjusted for the effect of Atlas of End-Stage Renal Disease in the United States, Bethesda, MD, other covariates. Using the estimated SEMs of the predictions, we National Institutes of Health, National Institute of Diabetes and Di- gestive and Kidney Diseases, 2012 estimated confidence intervals for these state-specific O/E rate ratios. 2. Lanska DJ: Geographic distribution of stroke mortality in the United fi Because of the large sample size of our cohort, statistical signi cance States: 1939-1941 to 1979-1981. Neurology 43: 1839–1851, 1993 was inferred only for P,0.01. All statistical analyses were done with 3. Howard G, Labarthe DR, Hu J, Yoon S, Howard VJ: Regional differences SAS software, version 9.2 (SAS Institute, Inc., Cary, NC ). We then in African Americans’ high risk for stroke: The remarkable burden of incorporated an approach used by others to illuminate patient char- stroke for Southern African Americans. Ann Epidemiol 17: 689–696, acteristics associated with states’ stroke rates:3 sequentially adjusted 2007 fi 4. Borhani NO: Changes and geographic distribution of mortality from for age; then age and sex; then for age, sex, and race; and, nally, for all cerebrovascular disease. Am J Public Health Nations Health 55: 673– factors based on our generalized linear mixed model, which accoun- 681, 1965 ted for individual-level characteristics. 5. Covic A, Kanbay M, Voroneanu L, Turgut F, Serban DN, Serban IL, Goldsmith DJ: Vascular calcification in chronic kidney disease. Clin Sci (Lond) 119: 111–121, 2010 Sensitivity Analyses 6. Hsu CY, Vittinghoff E, Lin F, Shlipak MG: The incidence of end-stage To examine the robustness of our results, we performed several renal disease is increasing faster than the prevalence of chronic renal sensitivity analyses. First, as stated above, we repeated the analysis insufficiency. Ann Intern Med 141: 95–101, 2004 with a more sensitive method for identifying ischemic stroke. Second, 7. Lackland DT, Bachman DL, Carter TD, Barker DL, Timms S, Kohli H: The we performed the analysis both with and without the modified Liu geographic variation in stroke incidence in two areas of the south- comorbidity index as a summary measure of overall illness burden. eastern stroke belt: The Anderson and Pee Dee Stroke Study. Stroke 29: 2061–2068, 1998 Third, we performed an analysis in the cohort of dually eligible 8. Cushman M, Cantrell RA, McClure LA, Howard G, Prineas RJ, Moy CS, (Medicare & Medicaid) dialysis patients, which creates a more ho- Temple EM, Howard VJ: Estimated 10-year stroke risk by region and mogenous cohort based upon socioeconomic status. race in the United States: Geographic and racial differences in stroke risk. Ann Neurol 64: 507–513, 2008 9. Glymour MM, Kosheleva A, Boden-Albala B: Birth and adult residence Compliance and Protection of Human Research in the Stroke Belt independently predict stroke mortality. Neurology Participants 73: 1858–1865, 2009 The research protocol was approved by the institutional review board 10. Liao Y, Greenlund KJ, Croft JB, Keenan NL, Giles WH: Factors ex- at the University of Kansas Medical Center (KUMC). The work was plaining excess stroke prevalence in the US Stroke Belt. Stroke 40: undertaken in accordance with the principles of the Declarations of 3336–3341, 2009 Helsinki. Data Use Agreements between KUMC and the U.S. Renal 11. Yang D, Howard G, Coffey CS, Roseman J: The confounding of race Data Service and CMS were in place. and geography: How much of the excess stroke mortality among Afri- can Americans is explained by geography? Neuroepidemiology 23: 118–122, 2004 12. Mujib M, Zhang Y, Feller MA, Ahmed A: Evidence of a “heart failure belt” in the southeastern United States. Am J Cardiol 107: 935–937, ACKNOWLEDGMENTS 2011 13. Kalantar-Zadeh K, Block G, Horwich T, Fonarow GC: Reverse epide- The authors thank Connie Wang, MD and Amanda Gellhaus for miology of conventional cardiovascular risk factors in patients with chronic heart failure. J Am Coll Cardiol 43: 1439–1444, 2004 technical assistance with manuscript preparation. 14. Kalantar-Zadeh K: What is so bad about reverse epidemiology anyway? Funding for this study was provided by National Institutes of Semin Dial 20: 593–601, 2007 Health (National Institute of Diabetes and Digestive and Kidney 15. Seliger SL, Gillen DL, Tirschwell D, Wasse H, Kestenbaum BR, Stehman- Diseases) grants K23 DK085378 (J.B.W.) and R01 DK080111 (T.I.S.), Breen CO: Risk factors for incident stroke among patients with end- by a National Kidney Foundation Young Investigator Award (J.B.W.), stage renal disease. J Am Soc Nephrol 14: 2623–2631, 2003 and by a Sandra A. Daugherty Foundation Grant (J.B.W.). 16. Sozio SM, Armstrong PA, Coresh J, Jaar BG, Fink NE, Plantinga LC, Powe NR, Parekh RS: Cerebrovascular disease incidence, character- The data reported here have been supplied by the U.S. Renal Data istics, and outcomes in patients initiating dialysis: The choices for System (DUA#2007-10 & 2009-19) and CMS (DUA#19707). The healthy outcomes in caring for ESRD (CHOICE) study. AmJKidneyDis interpretation and reporting of these data are the responsibility of the 54: 468–477, 2009

2060 Journal of the American Society of Nephrology J Am Soc Nephrol 24: 2053–2061, 2013 www.jasn.org CLINICAL EPIDEMIOLOGY

17. Sánchez-Perales C, Vázquez E, García-Cortés MJ, Borrego J, Polaina M, 24. Wetmore JB, Ellerbeck EF, Mahnken JD, Phadnis M, Rigler SK, Gutiérrez CP, Lozano C, Liébana A: Ischaemic stroke in incident dialysis Mukhopadhyay P, Spertus JA, Zhou X, Hou Q, Shireman TI: Atrial fi- patients. Nephrol Dial Transplant 25: 3343–3348, 2010 brillation and risk of stroke in dialysis patients. Ann Epidemiol 23: 112– 18. Power A, Chan K, Singh SK, Taube D, Duncan N: Appraising stroke risk 118, 2013 in maintenance hemodialysis patients: A large single-center cohort 25. Wetmore JB, Mahnken JD, Mukhopadhyay P, Hou Q, Ellerbeck EF, study. AmJKidneyDis59: 249–257, 2012 Rigler SK, Spertus JA, Shireman TI: Geographic variation in car- 19. Wiesholzer M, Harm F, Tomasec G, Barbieri G, Putz D, Balcke P: In- dioprotective antihypertensive medication usage in dialysis patients. cidence of stroke among chronic hemodialysis patients with non- Am J Kidney Dis 58: 73–83, 2011 rheumatic atrial fibrillation. Am J Nephrol 21: 35–39, 2001 26. Wetmore JB, Mahnken JD, Rigler SK, Ellerbeck EF, Mukhopadhyay P, 20. Genovesi S, Pogliani D, Faini A, Valsecchi MG, Riva A, Stefani F, Spertus JA, Hou Q, Shireman TI: The prevalence of and factors asso- Acquistapace I, Stella A, Bonforte G, DeVecchi A, DeCristofaro V, ciated with chronic atrial fibrillation in Medicare/Medicaid-eligible di- Buccianti G, Vincenti A: Prevalence of atrial fibrillation and associated alysis patients. Kidney Int 81: 469–476, 2012 factors in a population of long-term hemodialysis patients. Am J Kidney 27. Liu J, Huang Z, Gilbertson DT, Foley RN, Collins AJ: An improved co- Dis 46: 897–902, 2005 morbidity index for outcome analyses among dialysis patients. Kidney 21. Wizemann V, Tong L, Satayathum S, Disney A, Akiba T, Fissell RB, Kerr Int 77: 141–151, 2010 PG, Young EW, Robinson BM: Atrial fibrillation in hemodialysis pa- 28. Go AS, Hylek EM, Chang Y, Phillips KA, Henault LE, Capra AM, tients: Clinical features and associations with anticoagulant therapy. Jensvold NG, Selby JV, Singer DE: Anticoagulation therapy for stroke Kidney Int 77: 1098–1106, 2010 prevention in atrial fibrillation: How well do randomized trials translate 22. Olesen JB, Lip GY, Kamper AL, Hommel K, Køber L, Lane DA, into clinical practice? JAMA 290: 2685–2692, 2003 Lindhardsen J, Gislason GH, Torp-Pedersen C: Stroke and bleeding in 29. McCulloch C, Searle S: Generalized, Linear, and Mixed Models,New atrial fibrillation with chronic kidney disease. NEnglJMed367: 625– York, John Wiley & Sons, Inc., 2001 635, 2012 23.AndradeSE,HarroldLR,TjiaJ,CutronaSL,SaczynskiJS,DoddKS, Goldberg RJ, Gurwitz JH: A systematic review of validated methods for identifying cerebrovascular accident or transient ischemic attack using ad- This article contains supplemental material online at http://jasn.asnjournals. ministrative data. Pharmacoepidemiol Drug Saf 21[Suppl 1]: 100–128, 2012 org/lookup/suppl/doi:10.1681/ASN.2012111077/-/DCSupplemental.

J Am Soc Nephrol 24: 2053–2061, 2013 Geographic Variation in Stroke in Dialysis 2061 Supplementary Appendix 1

Details on Methods

Details on data sources and linking strategy

Data for these analyses were assembled from two primary sources. First, we utilized the

USRDS, a national system that collects data on virtually all patients undergoing chronic dialysis

in the U.S. From the USRDS, we received standard patient records that included demographics,

comorbidites, functional status, and dialysis modality (from the Medical Evidence Form, known

as “CMS 2728”) at the time of dialysis commencement. The USRDS also incorporates data on

inpatient and outpatient medical claims paid by Medicare, which provides insurance coverage for the vast majority of dialysis patients. The Medicare claims files contain International

Classification of Diseases – 9th Revision (ICD-9) codes for each date of service.

To make possible the study of dually-eligible individuals, the USRDS performed a deterministic match of these Medicaid beneficiaries against the core USRDS files to identify dually-eligible individuals on chronic dialysis. This permitted us to link USRDS data with

Centers for Medicare & Medicaid Services (CMS) Medicaid prescription drug billing claims, in the form of the Medicaid Analytic eXtract Personal Summary Files and the final action prescription drug claims files. These sources were linked using previously-described methodology to permit identification of dually-eligible dialysis patients in 2000-05.

Details on covariates and descriptive variables

Demographic and clinical variables, drawn from the CMS 2728 dialysis intake form, included age, sex, race by ethnicity, body mass index, employment status, , substance abuse (alcohol or illicit drugs), ability to ambulate and to transfer, cause of ESRD, and dialysis modality. Comorbidities consisted of diabetes, congestive heart failure, coronary artery disease, cerebrovascular disease, and peripheral vascular disease. Ethnicity was categorized into one of four mutually-exclusive groups: non-Hispanic Caucasians, non-Hispanic African-Americans,

Hispanics, and Others. Body mass index (BMI) was classified into 4 categories: < 20 kg/ m2,

20-24.99 kg/m2, 25-29.99 kg/m2, ≥ 30 kg/m2. Cause of ESRD was categorized as diabetes, hypertension, glomerulonephritis, or other. Because the CMS 2728 form is structured such that diabetes and hypertension may be considered as both a cause of ESRD and/or a “freestanding” comorbidity, for the purposes of the present analysis, these two covariates were considered a comorbidity if they were listed as either the cause of ESRD or as a “freestanding” comorbidity on the CMS 2728 form. Dialysis modality at time of dialysis initiation was categorized as in- center hemodialysis or self-care dialysis (home hemodialysis or peritoneal dialysis). We used a modified form of the Liu Comorbidity Index. This index is a summary measure of disease burden which also includes cause of ESRD; therefore, cause of ESRD was not modeled separately. However, our form of this index used only 90 (rather than 180) days in which to acquire claims since we have previously found (data under peer review) little difference in indices generated using 90 or 180 days of claims data and because we had required that patients have Medicaid and Medicare coverage throughout the first 90 days.

Details on the determination of chronic atrial fibrillation

The ICD-9 code 427.31 was used to identify AF claims using a well-established algorithm designed to determine the presence of nontransient, nonvalvular AF. Individuals who had hyperthyroidism or thyrotoxicosis were eliminated, based on the presence of relevant ICD-9 and/or CPT (Common Procedural Technology) and/or HCPCS (Healthcare Common Procedure

Coding System) codes, or by a prescription at any time for methimazole or propothiouracil. We

next eliminated patients with evidence of valvular heart disease (using ICD-9 codes). Finally, to minimize potential misclassification from perioperative sources of AF (e.g., coronary artery bypass surgery), claims (rather than individuals) were eliminated unless there was a preexisting

(> 30 d) AF claim. This resulted in the elimination of individuals in whom AF claims were only proximally related to cardiac surgery, but allowed inclusion of individuals in whom there was evidence of preexisting AF. To classify individuals as having chronic AF, we initially required a total of 2 (or more) AF claims, separated by ≥30 days, of which no more than 1 was an inpatient claim. Additionally, we expunged all outpatient AF claims within 7 days of a subsequent AF claim-containing admission and within 30 days after an AF claim-containing admission, retaining only the original inpatient claim. Patients were divided into AF-free time and, if they developed AF, AF-time.

Supplementary Table 1. Ischemic strokes per 1000 patient-years, by year examined.

Year Ischemic Stroke Rate1 2000 32.4 2001 33.4 2002 36.0 2003 34.5 2004 33.1 2005 29.8 1Events per 1000 patient-years

Supplementary Table 2. Unadjusted ischemic stroke rates, per 1000 patient-years, and observed:expected adjusted rate ratios, with 99% confidence intervals, by state.

Ischemic Observed-to-Expected Adjusted Rate Ratios (99% Confidence Intervals) State Stroke Rate1 Unadjusted Age Age, Sex Age, Sex, Race Full2 AL 36.0 1.21 (1.05 − 1.39) 1.27 (1.11 − 1.47) 1.25 (1.09 − 1.44) 1.16 (1.02 − 1.32) 1.12 (0.99 − 1.26) AK 20.9 0.90 (0.64 − 1.26) 0.89 (0.61 − 1.30) 0.90 (0.63 − 1.28) 0.95 (0.72 − 1.25) 0.96 (0.77 − 1.21) AZ 25.9 0.89 (0.75 − 1.06) 0.90 (0.76 − 1.08) 0.91 (0.77 − 1.09) 0.94 (0.80 − 1.11) 0.97 (0.84 − 1.12) AR 37.1 1.12 (0.94 − 1.33) 1.16 (0.97 − 1.39) 1.15 (0.96 − 1.37) 1.09 (0.92 − 1.28) 1.04 (0.90 − 1.21) CA 28.1 0.92 (0.84 − 1.01) 0.95 (0.86 − 1.05) 0.95 (0.87 − 1.05) 0.94 (0.86 − 1.03) 0.96 (0.88 − 1.04) CO 24.8 0.90 (0.72 − 1.13) 0.93 (0.73 − 1.17) 0.93 (0.74 − 1.17) 0.95 (0.77 − 1.16) 0.96 (0.80 − 1.15) CT 31.3 0.99 (0.82 − 1.19) 0.95 (0.78 − 1.15) 0.95 (0.79 − 1.15) 0.96 (0.81 − 1.14) 0.95 (0.81 − 1.11) DE 42.0 1.14 (0.88 − 1.46) 1.15 (0.88 − 1.50) 1.15 (0.89 − 1.49) 1.09 (0.87 − 1.36) 1.07 (0.88 − 1.31) DC 36.5 1.09 (0.85 − 1.40) 1.13 (0.86 − 1.47) 1.11 (0.86 − 1.43) 0.99 (0.80 − 1.23) 1.02 (0.84 − 1.24) FL 32.7 1.00 (0.90 − 1.11) 1.00 (0.90 − 1.12) 1.02 (0.92 − 1.13) 0.99 (0.90 − 1.09) 0.99 (0.90 − 1.09) GA 32.6 1.14 (1.01 − 1.27) 1.20 (1.06 − 1.35) 1.17 (1.04 − 1.32) 1.07 (0.96 − 1.20) 1.04 (0.94 − 1.15) HI 25.5 0.90 (0.70 − 1.17) 0.91 (0.69 − 1.20) 0.91 (0.69 − 1.19) 0.99 (0.78 − 1.25) 0.97 (0.79 − 1.19) ID 20.0 0.84 (0.64 − 1.12) 0.82 (0.60 − 1.12) 0.85 (0.63 − 1.14) 0.91 (0.71 − 1.16) 0.94 (0.76 − 1.16) IL 35.4 1.05 (0.94 − 1.17) 1.03 (0.92 − 1.15) 1.03 (0.92 − 1.15) 1.01 (0.91 − 1.11) 1.01 (0.92 − 1.12) IN3 37.3 1.16 (1.02 − 1.33) 1.15 (1.00 − 1.32) 1.14 (1.00 − 1.31) 1.14 (1.00 − 1.29) 1.11 (0.99 − 1.25) IA 29.9 0.94 (0.77 − 1.15) 0.89 (0.72 − 1.10) 0.89 (0.73 − 1.10) 0.93 (0.77 − 1.12) 0.95 (0.80 − 1.12) KS 26.4 0.87 (0.70 − 1.09) 0.85 (0.68 − 1.07) 0.86 (0.69 − 1.07) 0.88 (0.73 − 1.07) 0.91 (0.76 − 1.09) KY 30.3 0.95 (0.80 − 1.12) 0.96 (0.81 − 1.14) 0.96 (0.81 − 1.13) 0.98 (0.84 − 1.14) 0.97 (0.84 − 1.12) LA 36.3 1.15 (1.01 − 1.31) 1.20 (1.05 − 1.38) 1.18 (1.03 − 1.35) 1.09 (0.96 − 1.23) 1.07 (0.95 − 1.20) ME 28.5 0.95 (0.73 − 1.23) 0.91 (0.69 − 1.20) 0.93 (0.71 − 1.21) 0.97 (0.77 − 1.22) 0.97 (0.80 − 1.19) MD 36.7 1.14 (0.99 − 1.31) 1.15 (0.99 − 1.32) 1.13 (0.98 − 1.30) 1.05 (0.92 − 1.20) 1.07 (0.94 − 1.20) MA 34.5 0.94 (0.80 − 1.10) 0.89 (0.75 − 1.05) 0.90 (0.76 − 1.06) 0.93 (0.80 − 1.08) 0.95 (0.82 − 1.09) MI 36.1 1.11 (0.99 − 1.25) 1.10 (0.98 − 1.24) 1.10 (0.98 − 1.23) 1.06 (0.96 − 1.18) 1.04 (0.94 − 1.15) MN 28.6 0.92 (0.76 − 1.11) 0.88 (0.72 − 1.07) 0.89 (0.74 − 1.08) 0.93 (0.79 − 1.11) 0.93 (0.80 − 1.09) MS 37.5 1.27 (1.09 − 1.47) 1.36 (1.17 − 1.59) 1.32 (1.14 − 1.54) 1.19 (1.03 − 1.37) 1.18 (1.03 − 1.34) MO 37.0 1.13 (0.98 − 1.30) 1.14 (0.98 − 1.32) 1.13 (0.98 − 1.30) 1.10 (0.96 − 1.25) 1.07 (0.94 − 1.21) MT 27.8 0.98 (0.74 − 1.31) 0.98 (0.72 − 1.34) 0.99 (0.73 − 1.33) 1.01 (0.79 − 1.30) 1.02 (0.82 − 1.27) NE 24.8 0.82 (0.65 − 1.05) 0.79 (0.61 − 1.02) 0.80 (0.62 − 1.03) 0.85 (0.69 − 1.06) 0.90 (0.74 − 1.09) NV 30.9 0.96 (0.75 − 1.22) 0.98 (0.76 − 1.26) 0.98 (0.77 − 1.26) 0.99 (0.80 − 1.23) 0.99 (0.82 − 1.19) NH 29.1 0.91 (0.69 − 1.20) 0.87 (0.65 − 1.17) 0.89 (0.67 − 1.18) 0.94 (0.74 − 1.19) 0.95 (0.78 − 1.17) NJ 41.5 1.09 (0.96 − 1.23) 1.06 (0.93 − 1.20) 1.07 (0.94 − 1.21) 1.03 (0.92 − 1.16) 1.04 (0.93 − 1.16) NM 16.0 0.74 (0.58 − 0.95) 0.76 (0.59 − 0.99) 0.77 (0.60 − 0.99) 0.82 (0.66 − 1.01) 0.86 (0.71 − 1.04) NY 30.3 0.90 (0.81 − 1.01) 0.89 (0.80 − 1.00) 0.89 (0.80 − 1.00) 0.87 (0.79 − 0.96) 0.90 (0.82 − 0.98) NC 36.9 1.28 (1.14 − 1.43) 1.35 (1.20 − 1.51) 1.32 (1.18 − 1.47) 1.22 (1.10 − 1.35) 1.15 (1.04 − 1.27) ND 21.9 0.85 (0.63 − 1.15) 0.80 (0.57 − 1.11) 0.82 (0.60 − 1.12) 0.89 (0.69 − 1.15) 0.93 (0.75 − 1.16) OH 39.0 1.11 (1.00 − 1.24) 1.09 (0.97 − 1.22) 1.08 (0.97 − 1.21) 1.07 (0.97 − 1.19) 1.03 (0.94 − 1.14) OK 38.1 1.18 (1.01 − 1.38) 1.22 (1.03 − 1.45) 1.20 (1.02 − 1.42) 1.21 (1.04 − 1.41) 1.16 (1.01 − 1.34) OR 27.0 0.94 (0.76 − 1.18) 0.96 (0.76 − 1.21) 0.96 (0.77 − 1.21) 1.00 (0.82 − 1.22) 0.99 (0.83 − 1.19) PA 35.0 1.00 (0.89 − 1.13) 0.98 (0.87 − 1.11) 0.98 (0.87 − 1.10) 0.98 (0.88 − 1.09) 0.98 (0.88 − 1.09) RI 21.8 0.82 (0.61 − 1.11) 0.78 (0.56 − 1.07) 0.80 (0.59 − 1.08) 0.86 (0.67 − 1.11) 0.91 (0.73 − 1.13) SC 37.5 1.24 (1.09 − 1.41) 1.31 (1.14 − 1.50) 1.27 (1.11 − 1.46) 1.15 (1.01 − 1.30) 1.10 (0.98 − 1.24) SD 19.4 0.87 (0.65 − 1.16) 0.83 (0.61 − 1.14) 0.85 (0.63 − 1.14) 0.92 (0.72 − 1.17) 0.94 (0.76 − 1.16) TN 34.7 1.11 (0.97 − 1.27) 1.15 (1.00 − 1.32) 1.13 (0.99 − 1.30) 1.09 (0.96 − 1.24) 1.06 (0.94 − 1.19) TX 30.0 1.03 (0.94 − 1.12) 1.10 (1.00 − 1.21) 1.08 (0.98 − 1.19) 1.01 (0.93 − 1.10) 0.99 (0.91 − 1.07) UT 28.8 0.93 (0.71 − 1.22) 0.94 (0.71 − 1.26) 0.95 (0.72 − 1.26) 0.99 (0.78 − 1.25) 1.00 (0.82 − 1.23) VT 23.1 0.95 (0.70 − 1.29) 0.93 (0.66 − 1.29) 0.93 (0.68 − 1.28) 0.97 (0.75 − 1.25) 0.98 (0.79 − 1.23) VA 35.6 1.11 (0.98 − 1.26) 1.13 (0.99 − 1.29) 1.12 (0.99 − 1.27) 1.05 (0.93 − 1.18) 1.04 (0.93 − 1.17) WA 28.1 0.91 (0.77 − 1.09) 0.91 (0.76 − 1.09) 0.92 (0.77 − 1.10) 0.96 (0.81 − 1.12) 0.95 (0.82 − 1.11) WV 41.0 1.19 (0.98 − 1.45) 1.20 (0.98 − 1.47) 1.18 (0.97 − 1.44) 1.18 (0.99 − 1.42) 1.13 (0.95 − 1.33) WI 31.4 0.95 (0.81 − 1.11) 0.91 (0.78 − 1.08) 0.92 (0.79 − 1.08) 0.95 (0.82 − 1.10) 0.93 (0.81 − 1.06) WY 18.5 0.90 (0.64 − 1.27) 0.89 (0.61 − 1.30) 0.90 (0.63 − 1.29) 0.94 (0.72 − 1.25) 0.96 (0.76 − 1.21) 1Unadjusted rate, per 1000 patient-years. 2Adjusted for age, sex, and race, as well as for the comorbidity and other factors shown in Table 2. 3The 99% confidence intervals exceed unity at the thousandths place for the Age- and the Age,Sex,Race-adjusted models.