a b s t r a c t

Introduction: intravenous alteplase reduces disability and improves functionality among acute ischemic stroke patients. Two decades after its approval, only a small fraction of patients get the treatment, and demonstrating its impact on mortality may make a strong case for its wider use. This study assessed the impact of Thrombolytic treatment by alteplase on 1-year mortality and readmission among acute ischemic stroke patients.

Method: The 2008-2013 Georgia Coverdell Acute Stroke registry data were linked with the 2008-2013 hospital discharge and the 2008-2014 death data in Georgia. Multiple imputation was applied; a propensity score mea- suring the probability of receiving intravenous alteplase was calculated and used for matching. A conditional lo- gistic regression was applied to compare 1-year mortality and readmission among propensity score matched pairs.

Results: Overall, 20.3% of 9620 acute ischemic stroke patients died and 22.4% were readmitted in one year. The multivariable regression result showed that patients who did not receive IV alteplase had a 1.49 (95%CI: 1.09- 2.04; p-value = 0.01) times higher odds of dying at one year than those who were treated with the thrombolytic agent. Among patients discharged home, no statistically significant difference was documented in the odds of being readmitted at least once within 365 days post-stroke discharge.

Discussion and conclusion: After accounting for patient differences and missing value, intravenous alteplase is as- sociated with reduction in long-term mortality. The results of this study suggest that patients who are identified as eligible for intravenous alteplase need to be offered the treatment.

(C) 2017

1. Introduction

Intravenous alteplase was approved in 1996 by the Food and Drug Administration as a thrombolytic agent in ischemic stroke because pa- tients who received the treatment are 30% more likely to have minimal or no disability at three months [1] and one year [2] after the index in- cident. Twenty years later, fewer than 10% of ischemic stroke patients receive the treatment [3,4]. Delay in stroke identification, lack of swift transport to stroke ready facilities, contraindications and warning symptoms, and indecision to provide intravenous (IV) alteplase by healthcare providers [5,6] may be associated with the low rates of alteplase use.

Demonstration of IV alteplase’s impact on long-term mortality post- stroke rather than on disability could make a strong case for its wider

* Corresponding author at: Epidemiologist, Georgia Department of Public Health, Division of Health Protection, 2 Peachtree Street, NW, Suite 14-483, Atlanta, GA 30303- 3142, United States.

E-mail address: [email protected] (M.S. Ido).

use. To date, the few studies designed to assess stroke patient outcomes of mortality have shown limited positive results, possibly because of shorter duration of follow-ups, differential effect of IV alteplase based on Stroke severity, and small sample sizes [2,7-10]. Stroke registries help to monitor and measure patient outcomes in the long-term and provide data to support the needed research [11]. This study was con- ducted to evaluate the impact of receiving intravenous thrombolytic treatment on 1-year mortality of stroke patients cared for by hospitals participating in the Georgia Coverdell Acute Stroke Registry.

1. Methods

We conducted a passive follow-up of acute ischemic stroke patients treated by hospitals participating in the Georgia Coverdell Acute Stroke Registry (GCASR). The Georgia Coverdell Acute Stroke Registry is part of the National Paul Coverdell Acute Stroke Program and is designed to monitor the quality of stroke care in the state [12]. The registry has the goal of reducing stroke disability, premature mortality and preventing recurrent stroke through improvement in quality of stroke

http://dx.doi.org/10.1016/j.ajem.2017.07.092

0735-6757/(C) 2017

M.S. Ido et al. / American Journal of Emergency Medicine 36 (2018) 262-265 263

care in collaboration with the participating hospitals and other stake- holders. Because the data were collected for public health surveillance purpose, this study has qualified for the Georgia State University Institu- tional Review Board exemption.

Data sources and data linkage

We used the 2008-2013 GCASR data, the 2008-2013 Georgia Dis- charge Data System (GDDS) and the 2008-2014 Georgia death data for this study. The GDDS has information on patients discharged from non-federal acute care hospitals in Georgia. The death data are collected from death certificates and provide information on deaths of Georgians including those who passed away in other states. The three data sources were linked stepwise using the Fine-Grained Record Integration and Linkage software program version 2.1.5 [13]. First, we linked the GCASR data with GDDS data using a hierarchical deterministic proce- dure; the output was, subsequently, linked with the death data using a probabilistic data linkage approach. We used hospital code, admission and discharge date, age, race and gender for matching the GCASR data with GDDS data. Additional information on residence (ZIP code and county) and a 15 digit alphanumerical code (LONGID - longitudinal ID), created from letters of the first and last names, birthdate and gen- der, were added for the probabilistic linkage.

Similar to a previous study, in which the probabilistic linkage be- tween hospital discharge and death data was seen to have a very good accuracy with 92% sensitivity and 100% specificity [14], the determinis- tic linkage in this study had a sensitivity of 87% and positive predictive value of 96% (details are provided in the appendix). There were 45,727 records with a clinical diagnosis of acute ischemic stroke in the initial GCASR data list. Of these, 41,216 were linked with GDDS data but only 39,331 had LONGID, the data element critical for linking with the death data and for determining readmissions from index admission. After excluding the readmissions, 36,043 subjects were eligible for fol- low-up (Fig. 1).

Inclusion and exclusion criteria

In this study, we included only subjects who were eligible for IV alteplase administration. Based on AHA’s guidelines for the early man- agement of patients with acute ischemic stroke [15], patients with con- traindications and warning symptoms and those with spontaneous stroke symptom recovery or a National Institute of Health (NIH) stroke scale score of zero were considered ineligible for IV alteplase. We ex- cluded from analysis patients with in-hospital stroke, symptom onset (last known to be well) to hospital arrival time N 270 min or no symp- tom onset time documented, and patients from hospitals with no IV alteplase treated patients during the study period.

Statistical analysis

A total of 9620 patients were included in the analyses. The main out- come of interest was death from any cause within 365 days after index patients’ stroke admission. Readmission for any cause within 365 days post discharge among patients discharged home was also assessed as a secondary outcome. Treatment with IV alteplase documented in the registry was the main predictor of outcome.

Thirty-eight percent of the observations had missing value at least for one variable. Therefore, we performed multiple imputation assum- ing a general missing pattern and missing at random with twenty repli- cations on variables considered to have relation with receiving IV alteplase and patient’s outcome [16]. A propensity score measuring the probability of receiving IV alteplase was calculated using variables that could be related to the decision of administering thrombolytic. Age, gender, race, Medicare as major source of health insurance, trans- port to hospital by EMS, duration of last known well to hospital arrival time, history of other illnesses and medication, severity of patient con- dition as documented by NIH stroke scale score and nothing per mouth status, hospital bed size, hour, day and year of admission were treated as predictors. Hospital bed size and NIH Stroke Scale score were categorized based on their distribution; hospital bed size was

Fig. 1. Flow diagram for patients included in the analysis.

264 M.S. Ido et al. / American Journal of Emergency Medicine 36 (2018) 262-265

classified as small (b 100 beds), medium-small (101-249 beds), medi- um-large (250-400 beds) and large (N 400 beds) while the NIH Stroke Score was classified in quartiles (0-4, 5-8, 9-14, and N 14). Hospitals

Table 1

Characteristics and outcome of acute ischemic stroke patients included for imputation, Georgia Coverdell Acute Stroke Registry, 2008-2013.

were classified geographically as metropolitan (codes 1-3) and non- metropolitan (codes N 3) based on the Rural-Urban Commuting Area classification (version 2.0) of location [17].

Age, years, median(IQR)	67 (56, 79)	69 (58, 80)	0.0008
Female, n(%)	1276 (51.5)	3717 (52.1)	0.5977
Whites, n(%)	1491 (60.1)	4294 (60.1)	0.9867
Medicare as the principal health	1381 (55.7)	4270 (59.8)	0.0003
insurance coverage, n(%)
Brought by EMS, n(%) 683 (69.1)		1660 (85.7)	b 0.0001
NIH Stroke scale score, unit, 11 (6, 18)		5 (2, 11)	b 0.0001
median(IQR)
Persistent or Paroxysmal Atrial 567 (22.9)		1322 (18.5)	b 0.0001
Fibrillation/Flutter, n(%)

We used the propensity score for variable reduction and matching [18]. Patients treated with IV alteplase were matched one-to-one with those eligible but didn’t receive the treatment using a SAS macro pro- gram [19]. Alteplase-treated patients were first matched to controls on 8 digits of the propensity score. Those that did not match were then matched to controls on 7 digits of the score, and we continued the matching down to a 1-digit match. The average difference in pro-

Characteristics Treated with IV alteplase

Not Treated with IV alteplase

p-Value^a

pensity score between the matched pairs was b 0.001.

We compared the characteristics of IV alteplase treated and non- treated patients, among those included for imputation, using chi-square test for categorical variables and Wilcoxon test for continuous variables. We compared outcomes, on imputed data, among propensity score matched pairs using conditional logistic regression. Moreover, we ap- plied a generalized estimating equation accounting for correlation at hospital level and taking hospital as a random variable. An extended Cox proportional hazard model for matched data was used to check for consistency of the results. All applications of the propensity score gave similar estimates, and we reported results from the conditional lo- gistic regression analyses. To assess for bias from imputation, we ana-

lyzed the data with complete information as well. The analyses were done using the SAS(R) statistical software version 9.3 (SAS Institute, Inc.).	Hospital bed size, n(%) b100 beds 100-249 beds	37 (1.5) 337 (13.6)	257 (3.6) 1167 (16.3)
	250-399 beds	605 (24.4)	1570 (22.0)
3. Results	>= 400 beds Admitted on, n(%)	1501 (60.5)	4146 (58.1)

Nothing per mouth status 410 (16.5) 668 (9.4) b 0.0001 Medication prior to admission, n(%)

Antihypertensive 1619 (65.4) 4996 (70.0) b 0.0001

Lipid lowering drug 900 (36.3) 2763 (38.7) 0.0334

Medical history of, n(%)

Dyslipidemia 873 (38.1) 2771 (41.6) 0.0035

Diabetes mellitus 635 (27.7) 2484 (37.3) b 0.0001

Hypertension 1834 (80.1) 5606 (84.1) b 0.0001

Coronary artery disease (MI) 556 (24.3) 1767 (26.5) 0.0346

Heart failure 246 (10.7) 648 (9.7) 0.162

Atrial fibrillation/flutter 443 (19.3) 1110 (16.7) 0.0034

Smoking 520 (22.7) 1473 (22.1) 0.5536

Last known well to hospital arrival 60 (40, 87) 180 (93, 225) b 0.0001 time, minutes, median(IQR)

b 0.0001

The final analytical dataset had 9620 patients from 48 hospitals. Hos- pitals certified as stroke centers by nationally recognized accreditation

Weekend 696 (28.1) 1986 (27.8) 0.8114

Day time 1602 (64.6) 4547 (63.7) 0.4145

Calendar Year, n(%) b 0.0001

agencies treated 89% of the patients. Patients’ median age was	2008	191 (7.7)	988 (13.8)
68 years (IQR: 57, 80 years). More than half (52%) were female, 60%	2009	282 (11.4)	973 (13.6)

were White, 59% had Medicare as their main source of health insurance coverage, and 25.8% received IV alteplase. The average overall mortality rates documented were 4.6% at 7 days, 10.4% at 30 days, 13.5% at 90 days and 20.3% at one year. Moreover, 22.4% of the patients were readmitted within one year post discharge. Patients treated with IV alteplase had statistically significantly different features including age, health insur- ance coverage, means of transport to hospital, NIH stroke scale score, NPO status and medication intake prior to admission compared to those who were eligible but didn’t receive the treatment (Table 1).

The one year mortality among patients with missing value in a pre- dictor variable was 20.5% compared to the 19.9% among those with complete data. The aggregate conditional logistic regression result from imputed data showed that patients who did not receive IV alteplase had a 1.49 (95% CI: 1.09-2.04) times higher odds of dying at one year than those who were treated with the thrombolytic agent (Table 2). The lowest and highest odds ratios observed in the twenty replicates were 1.22 and 1.70, respectively. Moreover, the analysis on complete records showed a result in similar direction albeit statistically not as significant (p-value = 0.37) as the one on imputed data (Table 2). Among patients discharged home, no statistically significant difference (OR = 0.90, 95% CI: 0.59, 1.39) was documented in the odds of being readmitted at least once within 365 days post discharge (Table 2).

4. Discussion

Acute ischemic stroke patients who were treated with intravenous thrombolysis had better odds of survival at one year post discharge. Those who were eligible but didn’t receive intravenous alteplase had a 49% higher odds of dying at one year than those who received the thrombolytic agent. The data, however, didn’t show evidence that treat- ment with intravenous thrombolysis reduces 1-year readmission among patients discharged home.

2010	398 (16.0)	1082 (15.2)
2011	438 (17.7)	1226 (17.2)
2012	537 (21.7)	1463 (20.5)

2013 634 (25.6) 1408 (19.7)

One year outcome, n(%)

Mortality	526 (21.2)	1423 (19.9)	0.1719
Readmission	566 (22.8)	1590 (22.3)	0.5689

IQR: Interquartile range.

a: ?² and Wilcoxon tests were applied for nominal and quantitative variables, respectively.

Prior studies with randomized design showed some reduction of 1- year mortality for patients treated IV alteplase but the results were not statistically significant [2,7-9]. Besides sample size, the observed differ- ence might be due to differences in characteristics of study populations. The NINDS study [2] applied exclusion criteria similar to those used in this study, but it included only patients who could be treated within 3 h of symptom onset. The third International Stroke Trial (IST) study [7], on the other hand, included patients who neither had a clear indica- tion nor contraindication for IV alteplase and who could be treated within 6 h of symptom onset. Although assessment at 18-month post discharge didn’t show a meaningful reduction in mortality, a modest re- duction is reported at three-years of follow-up in the IST study [10].

Administration of IV alteplase has its own risks for complication such as bleeding and is not indicated for all patients who present with acute ischemic stroke. In this study, we considered only patients who do not have a condition that may preclude the use of IV alteplase. In a similar observational study [11], where only eligible patients as defined by na- tional guidelines were considered, Schmitz et al. reported a similar find- ing (Hazard Ratio = 1.52; 95% confidence interval, 1.14-2.04) based on the Danish Stroke Registry.

Some shortcomings are worth considering in the interpretation of the result of this study. Aside from the clinical diagnosis, no imaging in- formation was available to differentiate the type and extent of ischemic

M.S. Ido et al. / American Journal of Emergency Medicine 36 (2018) 262-265 265

Table 2

Relative risk of 1-year mortality and readmission by intravenous alteplase treatment sta- tus, Georgia Coverdell Acute Stroke Registry 2008-2013.

Outcome Complete data Imputed data

Disease, Health Behavior and Injury Epidemiology Unit, in preparing this manuscript.

Appendix A. Supplementary data

Odds ratio^a (95% CL)

p-Value Odds ratio^a

(95% CL)

p-Value

Supplementary data to this article can be found online at http://dx.

1-year Mortality 0.37 0.01

doi.org/10.1016/j.ajem.2017.07.092.

Didn’t receive IV alteplase

1.12

(0.83,1.67)

1.49 (1.09,

2.04)

Received IV alteplase Referent Referent

1-year Readmission 0.44 0.64

References

Didn’t receive IV alteplase

0.81 (0.48,

1.38)

0.90 (0.59,

1.39)

1. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med 1995;

Received IV alteplase Referent Referent

CL: Confidence limit. IV: Intravenous.

a: Alteplase treated patients were matched one-to-one with eligible non-treated pa- tients on probability of receiving IV alteplase (propensity score), and conditional logistic regression applied.

insult that may affect patient outcome [20]. The data linkage left out 14% unmatched patient records and could potentially introduce bias. Statis- tically significant difference in 1-year mortality was observed when we applied multiple imputation to get accurate estimates that are general- izable to the study population. Nonetheless, patients with missing data had a 1-year mortality (20.5%) comparable to those with complete data (19.9%). Thus, it is reasonable to assume that missingness were at ran- dom. Although bias could potentially be introduced, analysis of com- plete data produced results similar to the ones generated from imputed data. Estimates for each replicate dataset from the multiple im- putation also indicated an association between IV alteplase treatment and 1-year mortality in the same direction.

This study adds more positive results to the existing literature on the impact of IV alteplase particularly in the context of its effectiveness and hospitals’ effort to improve the overall quality of acute stroke care. The results would help healthcare providers make their decision promptly and save valuable time in the care of stroke patients.

5. Conclusions

This study indicates that, after accounting for patient differences and missing value, intravenous alteplase is associated with reduction in long-term mortality, and strengthens the importance of swift identifica- tion and treatment of eligible patients. Further studies are needed to confirm these observations.

Sources of funding

The authors received no financial support for the research, author- ship, and/or publication of this article. The Georgia Coverdell Acute Stroke Registry (GCASR) has been funded by the Centers for Disease Control and Prevention under agreement# U58DP003960.

Disclosures

None.

Acknowledgments

We acknowledge the help of Cherie L. Drenzek, DVM, MS, Georgia state epidemiologist and Rana A Bayakly, MPH, Director of Chronic

333(24):1581-7.

1. Kwiatkowski TG, Libman RB, Frankel M, Tilley BC, Morgenstern LB, Lu M, et al. Effects of tissue plasminogen activator for acute ischemic stroke at one year. National Insti- tute of Neurological Disorders and Stroke recombinant tissue plasminogen activator stroke study group. N Engl J Med 1999;340:1781-7.
2. Schwamm LH, Ali SF, Reeves MJ, Smith EE, Saver JL, Messe S, et al. Temporal trends in patient characteristics and treatment with intravenous thrombolysis among acute ischemic stroke patients at get with the guidelines-stroke hospitals. Circ Cardiovasc Qual Outcomes 2013;6:543-9.
3. Scott PA, Meurer WJ, Frederiksen SM, Kalbfleisch JD, Xu Z, Haan MN, et al. A multi- level intervention to increase community hospital use of alteplase for acute stroke (INSTINCT): a cluster-randomized controlled trial. Lancet Neurol 2013;12(2): 139-48.
4. Scott PA, Xu Z, Meurer WJ, Frederiksen SM, Haan MN, Westfall MW. Attitudes and beliefs of Michigan emergency physicians toward alteplase use in stroke: baseline survey results from the INSTINCT trial hospitals. Stroke 2010;41(9):2026-32.
5. Brown DL, Barsan WG, Lisabeth LD, Gallery ME, Morgenstern LB. Survey of emergen- cy physicians about recombinant tissue plasminogen activator for acute ischemic stroke. Ann Emerg Med 2005;46(1):56-60.
6. The IST-3 collaborative group. Effect of thrombolysis with alteplase within 6 h of acute ischemic stroke on long-term outcomes (the third International Stroke Trial [IST-3]): 18-month follow-up of a randomized controlled trial. Lancet Neurol 2013;12:768-76.
7. Wardlaw JM, Murray V, Berge E, del Zoppo G, Sandercock P, Lindley RL, et al. Recom- binant tissue plasminogen activator for acute ischemic stroke: an updated systemat- ic review and meta-analysis. Lancet 2012;379:2364-72.
8. Fischer U, Mono M, Zwahlen M, Nedeltchev K, Arnold M, Galimanis A, et al. Impact of thrombolysis on stroke outcome at 12 months in a population: the Bern stroke pro- ject. Stroke 2012;43:1039-45.
9. Berge E, Cohen G, Roaldsen MB, Lundstrom E, Isaksson E, Rudberg AS, et al. Effects of alteplase on survival after ischaemic stroke (IST-3): 3 year follow-up of a randomised, controlled, open-label trial. Lancet Neurol 2016;15(10):1028-34.
10. Schmitz ML, Simonsen CZ, Hundborg H, Christensen H, Ellemann K, Geisler K, et al. Acute ischemic stroke and long-term outcome after thrombolysis: nationwide pro- pensity score-matched follow-up study. Stroke 2014;45(10):3070-2.
11. Centers for Disease Control and Prevention (CDC). Use of a registry to improve acute stroke care-seven states, 2005-2009. MMWR Morb Mortal Wkly Rep 2011;60(7): 206-10.
12. Jurczyk P, Lu JJ, Xiong L, Cragan JD, Correa A. FRIL: a tool for comparative record link- age. AMIA Annu Symp Proc 2008:440-4.
13. Ido MS, Bayakly R, Frankel M, Lyn R, Okosun IS. Administrative data linkage to eval- uate a quality improvement program in acute stroke care, Georgia, 2006-2009. Prev Chronic Dis 2015;12:E05. http://dx.doi.org/10.5888/pcd12.140238.
14. Jauch EC, Saver JL, Adams HP, Bruno A, Connors JJ, Demaerschalk BM, et al. Guide- lines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2013;44:870-947.
15. John W, Graham JW, Olchowski AE, Gilreath TD. How many imputations are really needed? Some practical clarifications of multiple imputation theory. Prev Sci 2007; 8:206-13.
16. WWAMI Rural Health Research Center. RUCA data: code definitions, version 2.0. http://depts.washington.edu/uwruca/ruca-codes.php. (Accessed March 20), 2015.
17. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41-55.
18. Lori S. Parsons. Paper 165-29: Performing a 1:N case-control match on propensity score. Proceedings of the Twenty-Ninth Annual SAS(R) Users Group International Conference. 2004. SAS Institute Inc. Cary, NC. Accessed at http://www2.sas.com/ proceedings/sugi29/165-29.pdf on December 23, 2015.
19. Petty GW, Brown RD, Whisnant JP, Sicks JD, O’Fallon WM, Wiebers DO. Ischemic stroke subtypes: a population-based study of functional outcome, survival, and re- currence. Stroke 2000;31:1062-8.