Professional Documents
Culture Documents
ORIGINAL ARTICLE
caused by the focal brain spinal cord or retinal ischemia, with- procedures such as cerebral angiography (88.41), mechanical ven-
out acute infarction [2]. TIA is an important predictor of stroke tilation (9672), transfusion (99.04), intubation (96.04), gastros-
and subsequent death within the next 3 months and thereafter tomy (431.1431.9) and thrombolytic administration (99.10).
[35]. According to one population level study, the risk of The admitting hospitals were classied as teaching or non-
stroke within 2 days is 5% after an episode of TIA [6]. Another teaching and were further characterized into small, medium and
study from the same data set has shown that the in-hospital large on the basis of available hospital beds. Outcome measures in-
mortality after TIA in the general population during the years cluded discharge disposition and hospital charges. Discharge dis-
200608 was 0.4% [7]. In another population level study [8], position is categorized into routine, home healthcare, short-term
renal failure was associated with a higher length of stay (OR: 1.5, hospital and other facility including intermediate care and skilled
95% CI: 1.41.5, P < 0.0001) in TIA patients. However, this nursing home or death in the NIS data les. We categorized rou-
study did not elaborate on the duration of renal failure or tine discharge as none or minimal disability, any other discharge
dialysis status. Overall, there is a paucity of data regarding the status as moderate-to-severe disability as previously described [10].
in-hospital outcomes of TIA in patients with CKD and ESRD.
Statistical analysis
We aimed to study (i) degree of disability mainly functional
status derived from discharge destination, (ii) length of stay, The SAS 9.3 software (SAS Institute, Cary, NC, USA) was
(iii) charges of hospitalization, and (iv) mortality, by using used to convert the NIS database data into weighted counts to
the Nationwide Inpatient Sample (NIS) for TIA patients with generate national estimates, following Healthcare Cost and
CKD and ESRD. Utilization Project recommendations. We performed univari-
ate analysis, chi-square for categorical variables and t-test for
continuous variables to identify differences in study variables
M AT E R I A L S A N D M E T H O D S and end points between two groups ( patients with ESRD and
CKD) and to do comparison with the general population. We
We used the data les from NIS from 2005 to 2011 for our also performed multivariate logistic regression analysis to assess
analysis. A comprehensive synopsis on NIS data is available the outcome of patients with ESRD and CKD. Two logistic re-
at http://www.hcup-us.ahrq.gov. gression models were created. Model 1 included all patients,
The Medicare severity diagnosis-related group (DRG) code and stepwise logistic regression analysis was used to identify
524 was used to identify the patients admitted with TIA between the association between ESRD and odds of in-hospital mortal-
2005 and 2007 and DRG code 069 for 200811 period. The diag- ity. Model 2 included patients who were discharged alive and
nosis of TIA was further conrmed by International Classication logistic regression analysis was used to identify the association
of Diseases, Ninth Revision, Clinical Modication (ICD-9-CM) between ESRD and odds of moderate-to-severe disability. Lo-
435 either as primary or secondary discharge diagnosis. We gistic regression model was adjusted for age (as a continuous
used ICD-9-CM secondary diagnosis codes of CKD III (585.3), variable), sex (as a categorical variable), and confounding fac-
IV (585.4), V (585.5) and unspecied (585.9) in the CKD tors. Confounding factors were dened as all the factors
Transient ischemic attack in chronic kidney disease and maintenance dialysis patients 129
signicant in univariate analysis based on a type-I error of 0.05. Table 1. Patients demographic and clinical characteristics, in-hospital
Similar methodology was used to compare the outcomes of TIA procedures, hospital characteristics and discharge outcomes of TIA
patients with ESRD and CKD.
in CKD and ESRD patients with the general population.
TIA P-value
ESRD CKD
R E S U LT S Overall number (%) 18 316 67 256
Age mean (SD) 68 28 77 26 <0.0001
A total of 18 316 ESRD and 67 256 CKD patients were admitted Women 9931 (54.2) 33 461 (49.8) <0.0001
with TIA in the study period ( 20052011). Patients with CKD Race/ethnicity
were older (mean age 77 26 versus 68 28, P 0.0001). Dia- White 7273 (47.8) 40 736 (72.1) <0.0001
lysis patients had a higher percentage of women than non- African Americans 4981 (32.7) 9912 (17.5)
Hispanic 2105 (13.8) 3505 (6.2)
dialysis patients (54.2 versus 49.8%, P 0.0001). Co-morbid Other 872 (5.7) 2403 (4.2)
conditions such as hypertension and diabetes mellitus were sig- Co-morbid conditions
nicantly higher in ESRD patients (Table 1). Prevalence of at- Hypertension 17 126 (93.5) 59 943 (89.1) <0.0001
rial brillation, dyslipidemia, congestive heart failure, chronic Diabetes mellitus 6017 (33.0) 20 732 (30.8) 0.03
lung disease, history of stroke and alcohol abuse were higher Atrial brillation 2643 (14.4) 14 463 (21.5) <0.0001
Dyslipidemia 1332 (7.2) 6515 (9.7) <0.0001
in the CKD patients. Moreover, in-hospital complications Congestive heart failure 3443 (18.8) 13 506 (20.1) 0.09
such as deep venous thrombosis and sepsis were seen in a high- Chronic lung disease 2739 (15.0) 12 461 (18.5) <0.0001
er percentage age for patients with ESRD (Table 1), and they Stroke 1712 (9.3) 7979 (11.9) <0.0001
were also more likely to receive procedures such as intubation. Nicotine dependence 862 (4.7) 3884 (5.8) 0.007
Average length of stay (4 6 versus 3 5, P 0.0001) and Alcohol abuse 147 (0.8) 790 (1.2) 0.03
In-hospital complications
mean hospital charges ($24 386 49 136 versus $19 075 32 Pneumonia 257 (1.4) 1019 (1.5) 0.6
076, P 0.0001) were higher in patients with ESRD when com- Deep venous thrombosis 110 (0.6) 72 (0.1) 0.0001
pared with those with CKD. Pulmonary embolism 5 (0.03) 60 (0.1) 0.1
On multivariate analysis, patients with ESRD had higher Sepsis 137 (0.7) 122 (0.2) 0.0002
Myocardial infarction 179 (1.0) 463 (0.70) 0.10
ORIGINAL ARTICLE
Outcomes Unadjusted Adjusted for age and gender Adjusted for age, gender and
potential confoundersa
Odd ratio (95% CI) P-value Odd ratio (95% CI) P-value Odd ratio (95% CI) P-value
Model 1 Analysis comprising all patients
Discharged alive Reference Reference Reference
In-hospital mortality 1.83 (0.973.52) 0.07 2.74 (1.335.61) 0.007 2.84 (1.356.35) 0.009
Model 2 Analysis comprising alive patients
None-to- minimal disability Reference Reference Reference
Moderate-to-severe disability 1.04 (0.921.14) 0.7 1.42 (1.351.66) <0.0001 1.54 (1.321.73) <0.0001
Race/ethnicity, hypertension, diabetes, atrial brillation, dyslipidemia, chronic lung disease, nicotine dependence, alcohol abuse, stroke, hospital bed size and hospital teaching status.
a
Table 3. Disability and mortality in ESRD and CKD patients in comparison with the general population
ESRD/CKD versus general population ESRD/CKD (%) General P-value Adjusted for age, gender and potential
population (%) confoundersa
ORIGINAL ARTICLE
patients have a higher incidence of in-hospital complications (e.g. patients who were discharged to nursing homes were
and suggest that ESRD patients presenting with TIA may war- deemed to have moderate-to-severe disability and those going
rant inpatient admission due to their higher rates of complica- home were categorized to have none-to-minimal disability).
tions rather than outpatient or emergency room management. This methodology is well supported by the previous literature
We observed more in-hospital complications such as DVT [12]. We cannot infer from this data set if the patients with
and sepsis in ESRD patients. These ndings warrant clinicians ESRD were more disabled at baseline.
to pay particular attention to the DVT prophylaxis, and symp- Interestingly, TIA-related mortality was similar in CKD
toms such as fever, chills, constitutional symptoms or signs patients and the general population. However, higher odds of
such as catheter site erythema to prevent infection-related com- disability were seen in the CKD group. There was no difference
plications. In our observational cohort, ESRD patients also had in mortality and disability among different stages of CKD.
a higher mean length of stay when compared with those with While it is plausible that CKD patients may be prone to a stroke
CKD. This could also be explained by higher in-hospital com- transformation after a TIA episode, our ndings need further
plications observed in the ESRD group. We also acknowledge studies to identify a precise mechanistic explanation. ICD-9
that the social situation (such as socioeconomic status and codes for CKD are very specic but less sensitive and may have
marital status) of an individual patient may also have affected inuenced the results as well [17]. Odds of disability and mortality
the length of hospital stay and discharge destination and cannot were much higher in the ESRD group than the general popula-
be accounted for in our analysis. tion. These are novel ndings, and to our knowledge, they have
ESRD patients also had higher mean hospital charges pre- not been reported previously. Higher chances of TIA-related mor-
sumably due to a more complicated hospital course. However, tality and disability in ESRD patients warrant an in-patient admis-
these ndings need to be interpreted with the understanding sion rather than an emergency room/TIA clinic management. It
that in this data set total hospitalization charges represent should be noted that we have intentionally excluded patients with
the amount that hospitals billed for services, but not how acute kidney injury from our analysis to increase the reliability of
much hospital services actually cost or the specic amounts re- CKD and ESRD codes. In our previously published study [13],
ceived in reimbursement. These charges, therefore, include TIA patients with AKI had higher odds of moderate-to-severe
hospital overhead costs, charity care and bad debt, among disability (OR: 1.3, 95% CI: 1.21.4, P < 0.0001) and death (OR:
other costs, but do not include physicians professional fees. 4.2, 95% CI: 3.06.1, P < 0.0001) when compared with patients
The mean hospital charges were derived without any adjust- without acute kidney injury.
ment for ination. Our study is observational and based on the NIS data set.
In our analysis, higher odds of disability and mortality were Limitations of the NIS data set have been described previously
observed after the adjustment of confounders. We report the [14]. This data set depends on the accuracy of diagnoses and
degree of disability derived from the discharge destination procedure codes. In a previous study, sensitivity and positive
Transient ischemic attack in chronic kidney disease and maintenance dialysis patients 131
predictive values for primary diagnosis of ICD-9-CM code 435, 3. Lisabeth LD, Ireland JK, Risser JM et al. Stroke risk after transient ischemic
TIA, were 75 and 80%, respectively [15]. A small proportion of attack in a population-based setting. Stroke 2004; 35: 18421846
4. Lovett JK, Dennis MS, Sandercock PA et al. Very early risk of stroke after a
patients with TIA mimics are coded as TIA. The DRG system rst transient ischemic attack. Stroke 2003; 34: e138e140
has been used in previous studies pertaining to costs and length 5. van Wijk I, Kappelle LJ, van Gijn J et al. Long-term survival and vascular
of stay of stroke patients [16]. We think that using a combin- event risk after transient ischaemic attack or minor ischaemic stroke: a co-
ation of DRG and ICD-9-CM codes would result in higher ac- hort study. Lancet 2005; 365: 20982104
curacy for TIA patient identication. Our study has several 6. Chandratheva A, Mehta Z, Geraghty OC et al. Population-based study of
risk and predictors of stroke in the rst few hours after a TIA. Neurology
limitations that are inherent to administrative data sets [10]. 2009; 72: 19411947
The specicity of administrative codes for CKD [17] is 93% 7. Chaudhry SA, Tariq N, Majidi S et al. Rates and factors associated with ad-
with a median of 98%. Negative predictive values are also gen- mission in patients presenting to the ED with TIA in the united states-2006
erally high, that is, a median of 84%. Sensitivity of CKD codes is to 2008. Am J Emerg Med 2013; 31: 516519
low; (median, 41%; range, 388%). ICD-9 codes of AKI have 8. Qureshi AI, Adil MM, Zacharatos H et al. Factors associated with length of
hospitalization in patients admitted with transient ischemic attack in Uni-
high specicity , making our data more accurate, although it ted States. Stroke 2013; 44: 16011605
may underestimate the prevalence of CKD. Our study was 9. Waikar SS, Wald R, Chertow GM et al. Validity of international classica-
able to assess only in-hospital outcomes. Long-term outcomes tion of diseases, ninth revision, clinical modication codes for acute renal
could be extrapolated from discharge destination as it strongly failure. J Am Soc Nephrol 2006; 17: 16881694
correlates with modied Rankin scale at 3 and 12 months [18]. 10. Saeed F, Adil MM, Khursheed F et al. Acute renal failure is associated with
higher death and disability in patients with acute ischemic stroke: analysis
In summary, we observed higher odds of moderate-to-severe of nationwide inpatient sample. Stroke 2014; 45: 14781480
disability and in-hospital mortality among TIA patients with 11. Johnston SC, Rothwell PM, Nguyen-Huynh MN et al. Validation and re-
ESRD. Our ndings suggest having a lower threshold to admit nement of scores to predict very early stroke risk after transient ischaemic
ESRD and CKD patients presenting with TIA as they are more attack. Lancet 2007; 369: 283292
likely to have a complicated disease course than the general 12. Qureshi AI, Chaudhry SA, Hassan AE et al. Thrombolytic treatment of pa-
tients with acute ischemic stroke related to underlying arterial dissection in
population. the United States. Arch Neurol 2011; 68: 15361542
13. Saeed F, Adil MM, Malik AA et al. Worse in-hospital outcomes in patients
with transient ischemic attack in association with acute kidney injury:
C O N F L I C T O F I N T E R E S T S TAT E M E N T
ORIGINAL ARTICLE