INTRODUCTION
A major objective of inpatient stroke care is the prevention of medical and neurologic complications. Though rare, hemorrhagic transformation (HT) of an ischemic stroke (IS) can cause neurologic deterioration and is associated with an increased risk of death.1 If HT can be reliably identified in administrative data, it could become a component of hospital quality benchmarks. Prior studies have used administrative data to define hemorrhagic transformation of ischemic stroke; however, the accuracy of ICD9 coding for this condition is unknown.2,3,4 Coding algorithms used in prior studies have varied but all have required a discharge diagnosis of IS, defined by ICD9 code or Clinical Classification Software code, and a discharge ICD9 for intracranial hemorrhage. We aimed to determine the accuracy of ICD9 coding for HT after IS using the Greater Cincinnati/Northern Kentucky Stroke Study (GCNKSS).
METHODS
The purpose of the GCNKSS is to ascertain all strokes among a 5 county region that includes Cincinnati. Study personnel identified potential cases from January 1st, 2005 through December 31st, 2005 by screening emergency department encounters and inpatient admissions at 17 acute care hospitals for discharge ICD9 codes for stroke (430–436). Data are collected for all cases of IS, intracerebral hemorrhage (ICH), and subarachnoid hemorrhage (SAH). Potential cases were excluded if they had a discharge/autopsy diagnosis or neuroimaging with stroke but no clinical history consistent with stroke or if they had a clinical diagnosis of stroke and died within 24 hours of symptom onset without focal neurologic deficit or confirmatory neuroimaging/autopsy. Detailed methods of the GCNKSS have been published previously.5
For all potential cases, a research nurse abstracted history, physical examination, and diagnostic testing data, including neuroimaging. Study physicians reviewed the abstracted data and determined whether a stroke occurred, type of stroke, stroke mechanism, and whether HT was present. HT was characterized as asymptomatic or symptomatic (any neurologic deterioration).
For each HT case, hospital discharge ICD9 codes were examined. Our primary expectation was that HT cases would have both an ICD9 code for ischemic stroke (433.x1, 434.x1, 436) and for ICH (431). We looked separately at cases which were treated with rt-PA and evaluated the combination of ICD9 codes for IS with any ICD9 code for intracranial hemorrhage (430, 431, 432).
RESULTS
The GCNKSS identified 2,159 cases of ischemic stroke through hospital discharge screening, of which 102 had HT (symptomatic or asymptomatic). Among HT cases, 26 (25%) had an ICH code (431), and only 3 (3%) had codes for both IS (433.x1, 434.x1, 436) and ICH. HT was symptomatic in 38 patients, of whom 12 (32%) had a code for ICH, but none had both IS and ICH codes. Among the 102 cases of HT, 15 were treated with recombinant tissue plasminogen activator (rt-PA). There were 3 (20%) with an ICH code but only 1 (7%) had both IS and ICH codes. From the 38 cases of symptomatic HT, 9 were treated with rt-PA. One of these patients had an ICH code; this patient did not have an IS code.
A broader definition for hemorrhagic transformation, which used all of the intracranial hemorrhage ICD9 codes (430, 431, 432), identified 34 (33%) of the HT cases, although only 8 (8%) also had an IS code. The broader definition identified 17 (45%) of the symptomatic HT cases, although only 3 (8%) also had an IS code. Full results are presented in Table 1.
Table 1.
Accuracy of ICD9 Coding for Hemorrhagic Transformation of Ischemic Stroke
| ICD9 Codes | Any HT (n=102) |
Symptomatic HTM (n=38) |
|---|---|---|
| Ischemic stroke code only (433.x1, 434.x1, or 436) |
68 (67%) | 21 (55%) |
| 430 (SAH), no ischemic code | 2 (2%) | 2 (5%) |
| 430 (SAH), with ischemic code | 5 (5%) | 3 (8%) |
| 431 (ICH), no ischemic code | 23 (23%) | 12 (32%) |
| 431 (ICH), with ischemic code | 3 (3%) | 0 (0%) |
| 432 (intracerebral hemorrhage, NOS), no ischemic code |
1 (1%) | 0 (0%) |
| 432 (intracerebral hemorrhage, NOS), with ischemic code |
0 (0%) | 0 (0%) |
HT=hemorrhagic transformation, SAH=subarachnoid hemorrhage, ICH=intracerebral hemorrhage, NOS=not otherwise specified
DISCUSSION
Hemorrhagic transformation after ischemic stroke is an appealing quality metric. HT can cause neurologic deterioration and death.1 High rates of HT are associated with rt-PA protocol deviations, and HT rates can be lowered by quality improvement initiatives.6, 7 In order to identify HT in administrative datasets ICD9 codes for both IS and ICH must be present. Although ICD9 codes from administrative databases have been used to identify HT after ischemic stroke in prior studies, the accuracy of these codes was not verified. Our results demonstrate that ICD9 codes cannot accurately identify HT after ischemic stroke. Three-fourths of patients with HT of an ischemic stroke did not receive an ICH ICD9 code at hospital discharge, and few of the HT cases received ICD9 codes for both IS and ICH. Current ICD9 guidelines do not address coding HT after IS. Improved identification of HT will likely require the creation of a dedicated code, or modifier code, as well as physician and coder education.
This study has several limitations. HT was rare, and the number of symptomatic cases was small. ICD9 codes performed so poorly, it is unlikely that adding cases would change the conclusion. In this study we do not know the order of ICD9 codes and are unable to differentiate between principal and secondary diagnosis codes. Some prior studies of HT have required IS to be the principal discharge diagnosis and ICH to be a secondary diagnosis. This added constraint would likely make ICD9 coding less accurate for identifying HT. ICD9 coding may vary across institutions. All of the data in the present study come from hospitals affiliated with the GCNKSS. Because this study includes cases from 17 different hospitals, the results still have substantial generalizability. We are unable to assess the accuracy of ICD10 coding. As there is no specific code for HT in ICD10, it is also likely to be unreliable.
SUMMARY/CONCLUSIONS
ICD9 codes cannot accurately identify HT in ischemic stroke patients. HT cannot be reliably identified in administrative datasets.
ACKNOWLEDGMENTS
SOURCES OF FUNDING
Woo, Moomaw (R01NS30678; R01NS36695; 1U01NS069763), Khoury (RO1NS30678), and Alwell (R01NS03678) research support from NIH. Kissela research support: NIH (RO1NS30678), consultant to Allergan and NexStim, medicolegal case review, and adjudication committee Reata Pharmaceuticals. Flaherty (R01NS30678; R01NS36695; 1U01NS069763; 2P50NS044283) research support: NIH, consultant to Boehringer-Ingelheim and CSL Behring; advisory board for CSL Behring; medicolegal case review; PI of an NIH funded study with study drug supplied by Novo Nordisk. Khatri research support: NIH/NINDS, Penumbra and Genentech; consultant to Jannsen, Lake Biosciences; honorarium from Medical Dialogues. Adeoye: Genentech speaker’s bureau. Kleindorfer (R01NS30678) research support from NIH; Genentech speaker’s bureau (modest); medicolegal case review.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
CONFLICT(S) OF INTEREST/DISCLOSURES
Other authors have no disclosures to report.
REFERENCES
- 1.Tanne D. Markers of increased risk of intracerebral hemorrhage after intravenous recombinant tissue plasminogen activator therapy for acute ischemic stroke in clinical practice: The multicenter rt-pa acute stroke survey. Circulation. 2002;105:1679–1685. doi: 10.1161/01.cir.0000012747.53592.6a. [DOI] [PubMed] [Google Scholar]
- 2.Bateman BT, Schumacher HC, Boden-Albala B, Berman MF, Mohr JP, Sacco RL, Pile-Spellman J. Factors associated with in-hospital mortality after administration of thrombolysis in acute ischemic stroke patients: An analysis of the nationwide inpatient sample 1999 to 2002. Stroke; a journal of cerebral circulation. 2006;37:440–446. doi: 10.1161/01.STR.0000199851.24668.f1. [DOI] [PubMed] [Google Scholar]
- 3.Dubinsky R, Lai SM. Mortality of stroke patients treated with thrombolysis: Analysis of nationwide inpatient sample. Neurology. 2006;66:1742–1744. doi: 10.1212/01.wnl.0000218306.35681.38. [DOI] [PubMed] [Google Scholar]
- 4.Qureshi AI, Chaudhry SA, Rodriguez GJ, Suri MF, Lakshminarayan K, Ezzeddine MA. Outcome of the 'drip-and-ship' paradigm among patients with acute ischemic stroke: Results of a statewide study. Cerebrovascular diseases extra. 2012;2:1–8. doi: 10.1159/000335097. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kleindorfer DO, Khoury J, Moomaw CJ, Alwell K, Woo D, Flaherty ML, Khatri P, Adeoye O, Ferioli S, Broderick JP, Kissel BM. Stroke incidence is decreasing in whites but not in blacks: A population-based estimate of temporal trends in stroke incidence from the greater cincinnati/northern kentucky stroke study. Stroke; a journal of cerebral circulation. 2010;41:1326–1331. doi: 10.1161/STROKEAHA.109.575043. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Katzan IL, Furlan AJ, Lloyd LE, Frank JI, Harper DL, Hinchey JA, Hammel JP, Qu A, Sila CA. Use of tissue-type plasminogen activator for acute ischemic stroke - the cleveland area experience. Jama-J Am Med Assoc. 2000;283:1151–1158. doi: 10.1001/jama.283.9.1151. [DOI] [PubMed] [Google Scholar]
- 7.Katzan IL, Hammer MD, Hixson ED, Furlan AJ, Abou-Chebl A, Nadzam DM. Utilization of intravenous tissue plasminogen activator for acute ischemic stroke. Arch Neurol. 2004;61:346–350. doi: 10.1001/archneur.61.3.346. [DOI] [PubMed] [Google Scholar]