Current National Patterns of Comorbid Diabetes among Acute Ischemic Stroke PatientsTowfighi A.a, c · Markovic D.b · Ovbiagele B.d
aDepartment of Neurology, University of Southern California, and bDepartment of Biomathematics, University of California at Los Angeles, Los Angeles, Calif., cDepartment of Neurology, Rancho Los Amigos National Rehabilitation Center, Downey, Calif., and dDivision of Neurosciences, University of California at San Diego, San Diego, Calif., USA Corresponding Author
Background: Type 2 diabetes rates in the general population have risen with the growing obesity epidemic. Knowledge of temporal patterns and factors associated with comorbid diabetes among stroke patients may enable health practitioners and policy makers to develop interventions aimed at reducing diabetes rates, which may consequently lead to declines in stroke incidence and improvements in stroke outcomes. Methods: Using the Nationwide Inpatient Sample (NIS), a nationally representative data set of US hospital admissions, we assessed trends in the proportion of acute ischemic stroke (AIS) patients with comorbid diabetes from 1997 to 2006. Independent factors associated with comorbid diabetes were evaluated using multivariable logistic regression. Results: Over the study period, the absolute number of AIS hospitalizations declined by 17% (from 489,766 in 1997 to 408,378 in 2006); however, the absolute number of AIS hospitalizations with comorbid type 2 diabetes rose by 27% [from 97,577 (20%) in 1997 to 124,244 (30%) in 2006, p < 0.001]. The rise in comorbid diabetes over time was more pronounced in patients who were relatively younger, Black or ‘other’ race, on Medicaid, or admitted to hospitals located in the South. Factors independently associated with higher odds of diabetes in AIS patients were Black or ‘other’ versus White race, congestive heart failure, peripheral vascular disease, history of myocardial infarction, renal disease and hypertension. Conclusions: Although hospitalizations for AIS in the US decreased from 1997 to 2006, there was a steep rise in the proportion with comorbid diabetes (from 1 in 5 to almost 1 in 3). Specific patient populations may be potential targets for mitigating this trend.
© 2012 S. Karger AG, Basel
Type 2 diabetes is among the top ten leading causes of death in the USA and confers a high risk of stroke . Furthermore, the presence of diabetes is common among stroke patients and is associated with poor outcomes after stroke [2,3,4]. The occurrence of diabetes in the general US population is steadily rising , and mounting obesity rates are expected to lead to an additional surge in the number of persons diagnosed with diabetes . Knowledge about whether diabetes rates are increasing among acute ischemic stroke (AIS) patients and the types of factors that may be associated with diabetes in the context of a recent stroke may provide insight into avenues for stroke prevention and promoting better outcomes in AIS patients.
In this study, we first aimed to evaluate recent time trends in the prevalence of type 2 diabetes among unselected persons hospitalized with AIS in the USA. We subsequently investigated the independent links between sociodemographic, hospital and clinical factors and the presence of diabetes in patients hospitalized for AIS.
Data were obtained from the Nationwide Inpatient Sample (NIS), a single-stage stratified cluster sample of hospitals, which approximates a stratified 20% sample of all non-Federal, short-term, general and specialty hospitals serving adults in the USA . Detailed information on the design of the NIS is available at http://www.hcup-us.ahrq.gov. NIS captures discharge level information on several million discharges each year. From 1997 to 2006, the number of states contributing to the NIS increased from 22 to 38 [7.] A unique hospital identifier allows for linkage of discharge data to an NIS data set with hospital characteristics. The NIS provides discharge level weights that can be applied to the data to calculate estimates representative of the US population [7.]
All patients with a primary diagnosis of AIS by the International Classification of Diseases, ninth revision, procedure codes (ICD9-CM) were included (codes 433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91 and 436) whether they were alive or dead at the time of discharge. The patient sample included patients with both first and recurrent strokes. We computed the weighted proportion of AIS hospitalizations in patients with comorbid type 2 diabetes across the 10-year study period. Patients with type 2 diabetes were identified using ICD9-CM codes 250.0–250.9 with a fifth digit 0 or 2 (type 2 diabetes). Codes with a fifth digit 1 or 3 (type 1 diabetes) were excluded from the analyses . We assessed the proportion of AIS hospitalizations with comorbid type 2 diabetes by region, race and socioeconomic status. For the analysis by socioeconomic status, Medicaid insurance or lack of insurance was used as a proxy variable to define low socioeconomic status.
To assess for linear trends across time, we included year as a continuous variable in the logistic regression models while adjusting for the NIS survey design. Univariate logistic regression (adjusted for the survey design variables, i.e. stratum, cluster and weight variables) was used to evaluate sociodemographic, hospital level and clinical predictors one at a time (unadjusted analysis). To evaluate the independent association of these factors with comorbid type 2 diabetes, we used multivariable logistic regression modeling after adjusting for the survey design variables. The first multivariable model adjusted for sociodemographic (age, sex, race and primary payer) and hospital factors (hospital region, volume of acute strokes per year, bed size and hospital location/teaching status). The second multivariable model also adjusted for comorbid conditions (congestive heart failure, peripheral vascular disease, myocardial infarction, dementia, chronic pulmonary disease, mild or moderate/severe liver disease, cancer, metastatic carcinoma, renal disease, valvular disease, atrial fibrillation and hypertension).
In the aforementioned models, the effect of time was modeled using five 2-year time intervals from 1997–1998 to 2005–2006. Since there was a nonlinear/nonmonotone relation between age and the odds of coexisting type 2 diabetes, we used 5 age categories in 10-year increments (<55, 55–64, 65–74, 75–84, >84 years) rather than including age as a continuous variable. Also, since preliminary unadjusted analyses indicated that the effect of sex differed by age, age by sex interaction effects were taken into account by including the sex · age group interaction terms in the models. Odds ratios for the effects of sex and age were computed using linear contrasts under the above models. We also assessed interaction effects between time and each of the other variables by including the appropriate interaction terms in the above multivariable models. All data analyses were conducted using SAS (version 9.1; SAS Institute Inc., Cary, N.C., USA). Statistical hypotheses were tested using p < 0.05 as the level of statistical significance.
Over the 10-year study period, total primary AIS hospitalizations in the US declined, with 81,388 fewer patients (17% relative decrease) in 2006 compared to 1997, p < 0.01 (online suppl. table 1; for all online suppl. material, see www.karger.com/doi/10.1159/000334192). The number of AIS patients with comorbid type 2 diabetes, however, generally rose (online suppl. table 1), with 26,667 more patients (27% relative rise) in 2006 compared to 1997. The proportion of AIS patients with comorbid diabetes rose from 20% in 1997 to 30% in 2006 (p < 0.01; fig. 1).
|Fig. 1. Overall trends in the proportion of AIS patients with comorbid diabetes in the USA from 1997 to 2006: trend p value <0.001.|
Compared to 1997, AIS patients in 2006 were younger, more likely to be ‘other race’, to have Medicaid or ‘other’ insurance, to be admitted to urban teaching hospitals, to be admitted from the emergency department, and to have several comorbid conditions, namely myocardial infarction, congestive heart failure, peripheral vascular disease, chronic pulmonary disease, mild liver disease, renal disease and cancer (table 1). They were less likely to be White, to have Medicare, and to be admitted to rural or urban nonteaching hospitals. Finally, the hospital stays were shorter in 2006 compared to 1997.
|Table 1. Descriptive summary table for persons hospitalized with a primary diagnosis of AIS with comorbid diabetes in the USA: 1997 versus 2006|
Analysis of trends in the proportion of individuals hospitalized with AIS who had comorbid type 2 diabetes by region revealed that the proportion with diabetes increased at a higher rate in the South compared to other regions (table 2). The proportion of individuals with AIS with comorbid diabetes was higher among patients of Black/other race than among White patients. In addition, the relative increase in diabetes prevalence over time tended to be highest among Blacks/other race and lowest among Whites. The relative increase in diabetes prevalence among Medicaid/uninsured patients was higher than among insured/non-Medicaid patients (table 2).
|Table 2. Temporal trends in the proportion of AIS patients with comorbid diabetes by region, race and socioeconomic status|
Individuals with AIS in 2005–2006 had over 1.5 times greater odds of having type 2 diabetes than their counterparts in 1997–1998 (odds ratio 1.59, 95% confidence interval 1.56–1.64; table 3). This relationship persisted after adjustment for sociodemographic variables, hospital characteristics and comorbid conditions. Comparison of sex differences in comorbid diabetes revealed that below the age of 74 years, women were more likely than men to have comorbid diabetes, whereas over the age of 75 years, men were more likely than women to have comorbid diabetes (table 3). After adjustment for sociodemographic, hospital and clinical factors, there was no sex difference in odds of having comorbid diabetes in the 65- to 74-year age group (table 3).
|Table 3. Factors associated with comorbid diabetes in patients hospitalized primarily for AIS in the USA between 1997 and 2006|
Increases in type 2 diabetes prevalence among AIS patients were seen in all age groups regardless of sex, but were most pronounced among individuals <55 years of age (online suppl. tables 2, 3). Analysis of the effects of age on comorbid diabetes in both male and female AIS patients revealed an inverted U-shaped relationship, with the odds of diabetes being highest in individuals aged 55–74 years and being lower in both older and younger individuals (online suppl. tables 2, 3).
Individuals who were of Black, ‘other’ or ‘unknown’ race were all more likely than Whites to have comorbid diabetes (table 3). These findings persisted after adjustment for sociodemographic, hospital and clinical factors. After adjustment for the aforementioned factors, private and other insurance (vs. Medicare) were associated with lower odds of diabetes. Midwest (vs. northeast) geographic location was associated with a slightly higher odds of associated diabetes after adjusting for sociodemographic, hospital and clinical factors. With regard to hospital factors, small bed size was associated with higher odds of comorbid diabetes, while volume of acute strokes per year had no effect. Finally, comorbid conditions associated with an increase in the odds of diabetes were congestive heart failure, peripheral vascular disease, myocardial infarction, renal disease and hypertension (table 3).
The analysis of interaction effects between time and other factors indicated that although the odds of having diabetes increased across time regardless of the other factors, the magnitude of the effect significantly differed by age, race, payer type, hospital location/teaching status, geographical region and comorbid condition (online suppl. table 4). Specifically, the association was more pronounced for younger patients regardless of sex, for persons of Black race, Medicaid patients, rural and urban nonteaching hospitals, hospitals located in the South, and patients with comorbid congestive heart failure, chronic pulmonary disease or hypertension.
In this nationally representative study, we observed that from 1997 to 2006 there was a 10% absolute rise in the prevalence of type 2 diabetes among patients hospitalized for AIS in the USA. This increase occurred in the setting of a 17% reduction in the overall number of AIS hospitalizations. The increase in the proportion of individuals with AIS with comorbid diabetes disproportionately affected patients who were younger, non-White, of low socioeconomic status and those admitted to hospitals in the South. In addition, the presence of hypertension and renal disease among AIS patients, two medical conditions strongly related to diabetes , also rose substantially over the decade, and had the highest odds of being associated with diabetes in our study. These results highlight the importance of (1) intensifying efforts at reducing the incidence of diabetes in our population, particularly among underserved minority populations, and (2) identifying and treating diabetes in patients with AIS.
Diabetes rates have also increased in the general population, likely as a result of the obesity epidemic. Analysis of cross-sectional samples of the US population revealed that the crude prevalence of diagnosed diabetes in people aged ≥20 years rose from 5.1% in 1988–1994 to 7.7% in 2005–2006 (p< 0.0001) . Similar to the findings in our study, non-Hispanic Blacks and Mexican Americans were more likely to have diabetes compared to Whites . Intensifying efforts at preventing, screening for and treating diabetes will be an important step towards lowering the incidence of stroke and improving outcomes after stroke.
The prevalence of diabetes among stroke patients in this study was consistent with estimates from various prior studies. Data from international studies of single communities or hospital populations in the 1990s and early 2000s revealed that 21–25% of AIS patients had diabetes mellitus, similar to the rates noted in the late 1990s in our study [2,3,4]. A more recently conducted US nationwide voluntary stroke registry found a diabetes prevalence rate of 31.5% among hospitalized AIS patients, in accord with the diabetes prevalence rate for 2006 noted in our study . Our data suggest that the US diabetes prevalence among AIS patients is approximately 4 times as high as that seen in the general population and over the last decade has risen by relatively the same degree.
The exact factors underlying the rise in comorbid diabetes among patients hospitalized for AIS cannot be easily deciphered from this study, but based on prior data , recent information  and a bit of speculation, the higher diabetes prevalence rates may be related to increasing prevalence of diabetes in the general population, more thorough screening, better diagnostic techniques, improved documentation and prolonged survival.
Studies have shown that the presence of diabetes is associated with worse outcomes after ischemic stroke [2,3,4] and is associated with a higher risk of vascular events . Yet, analyses of recent nationwide AIS registries have revealed evidence of deficiencies in the quality of guideline-recommended diabetes-related care, particularly with regard to glycemic control monitoring and renoprotective antihypertensive agent use. Such deviations from evidence-based diabetes care among ischemic stroke patients likely need to be minimized to enhance outcomes in these patients .
Consistent with previous literature, there was an overall decline in the absolute number of AIS hospitalizations during the study period . The apparent decrease in AIS hospitalizations may be attributed to a decrease in certain vascular risk factors [14,15,16,17] and greater use of stroke-protective medications [14,17]. However, it is conceivable that without a temporal rise in the prevalence of diabetes mellitus, the reduction in AIS hospitalizations during this time period may have been even more substantial.
This study has limitations. First, we cannot exclude a possibility of inaccurate reporting of ICD codes, but any potential reporting errors in this large database were unlikely to have been systematic. Second, the rise in diabetes prevalence could have been due to lowering of the fasting plasma glucose range for diagnosing impaired fasting glucose in 2003 ; however, between the years 1997 and 2006, the rate of increase in prevalence of diabetes among AIS patients was relatively constant, suggesting the change in diagnostic criteria had a minimal effect on diabetes rates in this population. Third, detailed patient level data, such as glycemic biomarkers, vascular biomarkers and medications were not available for our analysis. Fourth, since up to one fourth of all persons with diabetes may actually be undiagnosed , we may have missed patients with diabetes who did not have a premorbid history and were not screened during their hospitalization. Finally, we were unable to capture AIS that did not result in hospitalization. While an increase in out-of-hospital strokes can potentially explain a decline in AIS hospitalizations, it would not explain the rising prevalence of diabetes. The study benefited from its nationwide scope, standardized methodology and clinician-diagnosed incidence data.
Amytis Towfighi: none; Daniela Markovic: none; Bruce Ovbiagele: none.
Amytis Towfighi, MD
7601 E. Imperial Highway
Downey, CA 90242 (USA)
Tel. +1 562 401 7611, E-Mail firstname.lastname@example.org
Received: January 10, 2011
Accepted: September 28, 2011
Published online: March 28, 2012
Number of Print Pages : 8
Number of Figures : 1, Number of Tables : 3, Number of References : 19
Additional supplementary material is available online - Number of Parts : 1
Vol. 33, No. 5, Year 2012 (Cover Date: May 2012)
Journal Editor: Hennerici M.G. (Mannheim)
ISSN: 1015-9770 (Print), eISSN: 1421-9786 (Online)
For additional information: http://www.karger.com/CED