sign in sign up
<<
Home , Subdural hematoma

CLINICAL ARTICLE J Neurosurg 130:302–311, 2019

Outcomes of chronic subdural in patients with liver cirrhosis

*Ching-Chang Chen, MD,1 Shao-Wei Chen, MD,2 Po-Hsun Tu, MD,1 Yin-Cheng Huang, MD, PhD,1 Zhuo-Hao Liu, MD,1 Alvin Yi-Chou Wang, MD,1 Shih-Tseng Lee, MD,1 Tien-Hsing Chen, MD,3 Chi-Tung Cheng, MD,4 Shang-Yu Wang, MD,4 and An-Hsun Chou, MD5

Departments of 1Neurosurgery and 5Anesthesiology and Divisions of 2Thoracic and Cardiovascular Surgery and 4Trauma and Emergency Surgery, Department of Surgery, Chang Gung Memorial Hospital, Linkou Medical Center, Chang Gung University; and 3Department of Cardiology, Chang Gung Memorial Hospital, Keelung Branch and Linkou Medical Center, Taoyuan City, Taiwan

OBJECTIVE Burr hole craniostomy is an effective and simple procedure for treating chronic subdural hematoma (CSDH). However, the surgical outcomes and recurrence of CSDH in patients with liver cirrhosis (LC) remain unknown. METHODS A nationwide population-based cohort study was retrospectively conducted using data from the Taiwan Na- tional Health Insurance Research Database. The study included 29,163 patients who underwent first-time craniostomy for CSDH removal between January 1, 2001, and December 31, 2013. In total, 1223 patients with LC and 2446 matched non-LC control patients were eligible for analysis. All-cause mortality, surgical complications, repeat craniostomy, ex- tended , and long-term medical costs were analyzed. RESULTS The in-hospital (8.7% vs 3.1% for patients with LC and non-LC patients, respectively), frequen- cy of hospital admission, length of ICU stay, number of blood transfusions, and medical expenditures of patients with LC who underwent craniostomy for CSDH were considerably higher than those of non-LC control patients. Patients with LC tended to require an extended craniotomy to remove subdural in the hospital or during long-term follow-up. The surgical outcome worsened with an increase in the severity of LC. CONCLUSIONS Even for simple procedures following minor head trauma, LC remains a serious comorbidity with a poor . https://thejns.org/doi/abs/10.3171/2017.8.JNS171103 KEY WORDS chronic subdural hematoma; liver cirrhosis; burr hole; craniostomy; vascular disorders

hronic subdural hematoma (CSDH) is a common patients present with LC because the incidence of and tends to occur in elderly patients af- hepatitis and LC is higher in Taiwan and East Asia than in ter minor head trauma.6 Burr hole craniostomy is other regions.9,24 LC and its associated complications have theC simplest and most common treatment for the initial been identified as significant risk factors for increased stages of CSDH. Despite recurrence, CSDH has a favor- morbidity and mortality in trauma patients,7,12,13,17 and they able outcome and prognosis.6,32,33,38 However, in our pre- have a considerable effect on poor outcomes of major sur- viously published data, the mortality and complication gery. 8,9,13,16,23,35 To our knowledge, there is no published rates associated with CSDH in patients with liver cirrhosis report on the surgical risk of CSDH in patients with LC. (LC) were as high as 33.3% (5 of 15 patients).7 Because Conclusions regarding the effects of LC on minor head the number of cases in our previous study was small, our trauma and the outcomes of a simple surgical procedure, results were inconclusive. In our clinical practice, many such as CSDH, are tentative. Therefore, to understand the

ABBREVIATIONS CCI = Charlson Comorbidity Index; CIC = catastrophic illness certificate; CSDH = chronic SDH; CTP = Child-Turcotte-Pugh; ICD-9-CM = International Classification of Diseases, 9th Revision, Clinical Modification; LC = liver cirrhosis; NHI = National Health Insurance; NHIRD = NHI Research Database; SDH = subdural hematoma; TBI = traumatic injury. SUBMITTED May 3, 2017. ACCEPTED August 1, 2017. INCLUDE WHEN CITING Published online February 2, 2018; DOI: 10.3171/2017.8.JNS171103. * C.C.C. and S.W.C. contributed equally to this work.

302 J Neurosurg Volume 130 • January 2019 ©AANS 2019, except where prohibited by US copyright law

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC C. C. Chen et al.

FIG. 1. Enrollment of the study population. actual risks and outcomes of CSDH in patients with LC, national Classification of Diseases, 9th Revision, Clinical we used patient data from a large national database, the Modification (ICD-9-CM). This study was completely ex- Taiwan National Health Insurance Research Database empt from ethics review by the ethics institutional review (NHIRD). In this longitudinal cohort study, NHIRD was board of Chang Gung Memorial Hospital. used to analyze the outcomes of using craniostomy to treat CSDH in patients with LC. Study Population In this study, we included all hospitalization records of Methods patients who were admitted for CSDH and had undergone a craniostomy for CSDH treatment. Index hospitalization Data Source was defined as the date when the patient was admitted for We conducted this nationwide cohort study using data first-time craniostomy to remove CSDH between January from NHIRD. NHIRD is an administrative database of 1, 2001, and December 31, 2013. Patients were identified Taiwan’s National Health Insurance (NHI) program, using the diagnostic ICD-9-CM codes for subdural hem- which is a compulsory, universal health insurance pro- orrhage (432.1, 852.2, and 852.3) and the presence of NHI gram that covers nearly 100% of the 23.7 million resi- reimbursement procedure code 83038C, which is unique dents of Taiwan.19 The database contains comprehensive for craniostomy to “remove chronic subdural hematoma” health-related information, such as diagnostic codes, performed by neurosurgeons. Because subdural medical procedures, and prescription details. As of 2015, hematoma (SDH) should typically be treated with crani- more than 3000 research articles based on NHIRD data otomy (NHI code 83037C), our study population included had been published. NHIRD uses information protection only nonacute patients. Otherwise, patients younger than measures, including an extremely strict data access pro- 20 years were excluded from our study, as shown in Fig. cedure and encryption of all personal information with 1. The study population was divided into LC and non-LC anonymous 32-digit identification numbers. In NHIRD groups according to the diagnosis of cirrhosis (ICD-9-CM data, disease diagnoses are coded according to the Inter- codes 571.2, 571.5, and 571.6).

J Neurosurg Volume 130 • January 2019 303

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC C. C. Chen et al.

TABLE 1. Baseline characteristics of the study population Before Matching After Matching Variable LC Patients Non-LC Patients p Value LC Patients Non-LC Patients p Value No. of patients 1225 27,938 1223 2446 Age, mean ± SD, yrs 63.7 ± 15.3 70.5 ± 13.9 <0.001 63.7 ± 15.3 63.6 ± 15.9 0.949 Age in yrs 20–39 85 (6.9) 1092 (3.9) <0.001 85 (7.0) 177 (7.2) 0.946 40–59 412 (33.6) 4412 (15.8) 410 (33.5) 809 (33.1) 60–79 535 (43.7) 15,154 (54.2) 535 (43.7) 1059 (43.3) ≥80 193 (15.8) 7280 (26.1) 193 (15.8) 401 (16.4) Sex 0.009 0.840 Male 968 (79.1) 21,181 (75.8) 967 (79.1) 1941 (79.4) Female 256 (20.9) 6750 (24.2) 256 (20.9) 505 (20.6) Comorbid conditions Diabetes mellitus 470 (38.4) 6982 (25.0) <0.001 469 (38.3) 922 (37.7) 0.700 Hypertension 566 (46.2) 14,039 (50.3) 0.006 566 (46.3) 1150 (47.0) 0.674 Hyperlipidemia 126 (10.3) 2231 (8.0) 0.004 125 (10.2) 255 (10.4) 0.848 119 (9.7) 1632 (5.8) <0.001 119 (9.7) 249 (10.2) 0.669 Coronary disease 226 (18.4) 4344 (15.5) 0.006 226 (18.5) 439 (17.9) 0.694 History of 29 (2.4) 749 (2.7) 0.505 29 (2.4) 52 (2.1) 0.634 Peripheral arterial disease 24 (2.0) 424 (1.5) 0.219 24 (2.0) 59 (2.4) 0.388 Prior ischemic 202 (16.5) 4240 (15.2) 0.211 202 (16.5) 394 (16.1) 0.752 Prior hemorrhagic stroke 113 (9.2) 1822 (6.5) <0.001 113 (9.2) 220 (9.0) 0.807 Prior seizure 126 (10.3) 876 (3.1) <0.001 126 (10.3) 230 (9.4) 0.386 Dialysis 51 (4.2) 543 (1.9) <0.001 51 (4.2) 96 (3.9) 0.721 Chronic obstructive pulmonary disease 180 (14.7) 2596 (9.3) <0.001 179 (14.6) 360 (14.7) 0.947 Malignancy 186 (15.2) 1944 (7.0) <0.001 185 (15.1) 391 (16.0) 0.500 CCI, mean ± SD 5.1 ± 2.8 1.9 ± 2.2 <0.001 5.1 ± 2.8 2.6 ± 2.8 <0.001 CCI score 1 99 (8.1) 15,546 (55.6) <0.001 98 (8.0) 1121 (45.8) <0.001 2–3 295 (24.1) 7472 (26.7) 295 (24.1) 650 (26.6) 4–5 329 (26.9) 2897 (10.4) 329 (26.9) 322 (13.2) ≥6 502 (41.0) 2023 (7.2) 501 (41.0) 353 (14.4) Hospital level <0.001 0.793 Regional/district hospital 736 (60.1) 14,121 (50.5) 735 (60.1) 1481 (60.5) Tertiary teaching hospital 489 (39.9) 13,817 (49.5) 488 (39.9) 965 (39.5) Length of follow-up, yrs 2.8 ± 3.0 4.5 ± 3.7 <0.001 2.8 ± 3.0 4.2 ± 3.6 <0.001 Data are shown as the number of patients (%) unless indicated otherwise.

Comorbidities and Outcome Measurements versal Child-Turcotte-Pugh (CTP) classification was un- We extracted the baseline characteristics, including known. Therefore, we modified the CTP classification and age, sex, hospital level, and clinical comorbidities, of the scored the severity of LC according to the aforementioned patients at the index hospitalization and prior hospitaliza- 5 main categories of LC-related complications. CTP class- tions (Table 1). Charlson Comorbidity Index (CCI) scores es A, B, and C were modified and defined as scores of 0–1 were used to assess overall systemic health.30 In patients (A), 2–3 (B), and 4–5 (C), respectively (Table 2). Further- with LC, special clinical characteristics, such as alcohol- more, under Taiwan’s NHI program, certain severe dis- ism, viral hepatitis, and previous bleeding history, were eases, such as LC, can be identified using the Registry for identified. Five major cirrhosis-related complications were Catastrophic Illness Patient Database. Patients with one of identified: , ascites and related the following LC-related complications are eligible for a complications (ascites and spontaneous bacteria peritoni- catastrophic illness certificate (CIC): uncontrolled ascites, tis), esophageal varices bleeding, coagulopathy, and hypo- hepatic coma, varices bleeding, and obstructive jaundice. albuminemia. Because we used claims data from NHIRD, CIC holders are exempt from paying insurance premiums the severity of LC in each patient according to the uni- and copayments. Therefore, all certification applications

304 J Neurosurg Volume 130 • January 2019

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC C. C. Chen et al.

TABLE 2. Clinical characteristics of the patients with LC (i.e., repeat craniostomy and craniotomy to remove SDHs) Variable No. of Patients (%) between the LC and non-LC groups by using a compet- ing risk survival model in which mortality was consid- No. of patients 1225 ered a competing factor.15 The survival curves of these Alcoholic cirrhosis 476 (38.9) 2 long-term outcomes were plotted using the cumulative Viral hepatitis incidence function under competing risk adjustment. We Hepatitis B 217 (17.7) also compared the admission cost during the first 5 years Hepatitis C 231 (18.9) between groups by using the Mann-Whitney U-test. Fur- thermore, we employed the log-rank test to evaluate the History of gastrointestinal bleeding 682 (55.7) correlation between LC severity (i.e., modified CTP score History of major bleeding 326 (26.6) and CIC certificate) and mortality risk during follow-up Complication of cirrhosis by using samples before matching. Data analysis, includ- Hepatic encephalopathy 283 (23.1) ing propensity score matching, was conducted using SAS Ascites and related complications 297 (24.2) (version 9.4, SAS Institute). Esophageal varices bleeding 196 (16.0) Coagulopathy 250 (20.4) Results Hypoalbuminemia 250 (20.4) Study Population Characteristics CIC A total of 29,163 adult patients who underwent first- No 1047 (85.5) time craniostomy to remove CSDH between January 1, Yes 178 (14.5) 2001, and December 31, 2013, were identified. Among these patients, 1223 patients with LC and 2446 non-LC Modified CTP score matched control patients were analyzed. Table 1 presents 0–1 (A) 891 (72.7) the baseline characteristics of the patients. Before match- 2–3 (B) 241 (19.7) ing, the patients with LC were younger, predominantly 4–5 (C) 93 (7.6) male, and had a significantly higher prevalence of almost all comorbidities (except history of myocardial infarction, peripheral arterial disease, and prior ischemic stroke) than non-LC patients. However, the prevalence of hypertension are carefully reviewed by Taiwan’s NHI bureau and can was lower in the LC group. After matching, the 2 groups be considered an accurate severity index for LC. were similar in terms of age, sex, or comorbidities. The Outcomes of interest for this study were postoperative mean ± SD follow-up of the patients with LC was con- complications, in-hospital mortality, all-cause mortality, siderably lower than that of the non-LC patients (2.8 ± repeat craniostomy to remove CSDH, and further cra- 3.0 years vs 4.2 ± 3.6 years). Because the required pro- niotomy to remove SDH. Mortality data were collected cedures are relatively simple, nearly 60% of craniostomy from the NHIRD registry data when the patients were 39 operations were performed at regional hospitals. The CCI withdrawn from the NHI program because of death. The scores of the patients with LC were higher than those of medical expenditures of any admission and readmission non-LC controls before and after matching because LC were directly collected from NHIRD. was a component of the CCI score. Statistical Analysis Characteristics of Patients With LC Before comparing outcomes between the LC and non-LC groups, we performed propensity score match- Table 2 lists the clinical characteristics of all patients ing to ensure that the baseline characteristics between with LC (n = 1225). Among these patients, 38.9% had al- the 2 groups would be comparable. The selected covari- coholic LC, 17.7% had hepatitis B, and 18.9% had hepatitis ates for calculating the propensity scores included age, C. More than half (55.7%) of the patients with LC had a sex, comorbidities, hospital level where the operation was history of gastrointestinal bleeding before their first cra- performed, and year of enrollment. CCI was not selected niostomy. The patients exhibited 5 main cirrhosis-related because LC was a component of CCI. Each patient with complications: hepatic encephalopathy (23.1%), ascites LC was matched with 2 non-LC control patients to ensure (24.2%), esophageal varices bleeding (16.0%), coagulopa- higher precision and lower bias.5 thy (20.4%), and hypoalbuminemia (20.4%). In addition, The baseline patient characteristics were compared be- 178 patients (14.5%) with advanced LC had a CIC. Ac- tween the LC and non-LC groups by using the chi-square cording to the modified CTP score, patients were catego- test for categorical variables and the 2-sample t-test for rized according to LC severity. Accordingly, 891 (72.7%), continuous variables. We compared the risk of binary 241 (19.7%), and 93 (7.6%) patients were categorized with in-hospital complications and outcomes (i.e., in-hospital modified CTP scores of 0–1, 2–3, and 4–5, respectively. mortality) between the 2 groups through logistic regres- sion. Continuous outcomes (e.g., the number of ventilator Perioperative Surgical Outcomes of CSDH in Patients days) between groups were compared through linear re- With and Without LC gression. The mortality rate during follow-up was com- Table 3 lists the in-hospital outcomes of the LC and pared between the groups by using Cox regression analy- non-LC patients after craniostomy. Compared with the sis. We further compared the risk of long-term outcomes non-LC group, the LC group had a higher postoperative

J Neurosurg Volume 130 • January 2019 305

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC C. C. Chen et al.

TABLE 3. In-hospital outcomes LC vs Non-LC Patients Variable LC Patients Non-LC Patients OR or B (95% CI)* p Value No. of patients 1223 2446 Categorical outcome In-hospital mortality 106 (8.7) 75 (3.1) 3.00 (2.21–4.07) <0.001 Postoperative infection 9 (0.7) 14 (0.6) 1.29 (0.56–2.98) 0.555 New-onset 61 (5.0) 83 (3.4) 1.50 (1.07–2.10) 0.020 De novo dialysis 28 (2.3) 22 (0.9) 2.58 (1.47–4.53) 0.001 Repeat craniostomy 235 (19.2) 425 (17.4) 1.13 (0.95–1.35) 0.172 Craniotomy to remove hemorrhage SDH 61 (5.0) 84 (3.4) 1.48 (1.05–2.07) 0.024 EDH 10 (0.8) 14 (0.6) 1.43 (0.64–3.23) 0.387 ICH 45 (3.7) 63 (2.6) 1.45 (0.98–2.13) 0.063 Continuous outcome Ventilator use, days 4.2 ± 9.2 3.1 ± 8.5 1.12 (0.52–1.72) <0.001 Packed red blood cells, units 1.6 ± 3.1 0.7 ± 2.0 0.88 (0.71–1.04) <0.001 Fresh frozen plasma, units 3.0 ± 7.5 0.8 ± 3.3 2.25 (1.90–2.60) <0.001 Platelets, units 7.7 ± 16.5 1.3 ± 7.8 6.39 (5.60–7.17) <0.001 Total transfusion, units 12.2 ± 22.6 2.8 ± 10.6 9.46 (8.39–10.53) <0.001 ICU duration, days 6.0 ± 9.3 4.4 ± 8.0 1.63 (1.05–2.21) <0.001 Hospital stays, days 18.0 ± 16.9 16.1 ± 17.1 1.96 (0.79–3.13) 0.001 In-hospital cost, NTD × 104 19.2 ± 24.9 15.9 ± 30.2 3.26 (1.30–5.22) 0.001 B = regression coefficient; EDH = ; ICH = ; NTD = New Taiwan dollars. Values are shown as the number of patients (%) or mean ± SD unless indicated otherwise. * B values are shown for continuous variables, and OR values are shown for categorical variables. in-hospital mortality rate (8.7% vs 3.1%, p < 0.001) and 3-month (18.9% vs 7.8%), 1-year (32.4% vs 14.5%), 5-year a higher incidence of new-onset seizure and de novo di- (56.5% vs 30.7%), and last follow-up (65.3% vs 38.3%) ex- alysis. Notably, the rate of repeat craniostomy for CSDH aminations. removal was not higher in the patients with LC (19.2% vs Figure 2A shows long-term mortality in both groups. 17.4%, p = 0.172). However, a higher proportion of patients Figure 2B and 2C depicts the cumulative incidence (esti- with LC required further craniotomy to remove SDH after mated using competing risk survival modeling) of repeat their first burr hole craniostomy (5.0% vs 3.4%, p = 0.024). craniostomy and the need for craniotomy to remove SDH Overall, the LC group had significantly more blood trans- in the 2 groups. Figure 3 show higher admission costs in fusions; longer durations of ICU stay, ventilator use, and the LC group than in the non-LC group through the first hospital stay; and higher total in-hospital costs. 5 years of follow-up. Figure 4 depicts the relationship be- tween the risk of all-cause mortality and the severity of Long-Term Surgical Outcomes of CSDH in Patients With LC. As expected, a higher modified CTP score was as- and Without LC sociated with a higher risk of mortality (Table 5). Patients Table 4 presents the postcraniostomy follow-up out- with advanced LC who had a CIC also exhibited a higher comes. Because of the relatively low mortality rate, most risk of mortality than the other patients. patients were discharged after surgery for CSDH removal and observed closely in the outpatient department. Dur- ing the observation period, 24.4% of patients in the LC Discussion group and 21.9% of patients in the non-LC group required Recurrence of CSDH repeated craniotomy to remove CSDH. Most Although burr hole surgery follows a simple operative were performed within 1 month in both groups (21.7% procedure and has more favorable outcomes than surgical vs 19.2%) with no significant difference in the number procedures for acute SDH or other brain injuries, the re- of repeated craniotomies between groups. However, sig- currence rate is relatively high. In the literature, the recur- nificantly more patients in the LC group underwent cra- rence rate of CSDH after burr hole surgery varies widely niotomy to remove an SDH either within 3 months (1% from 3% to 30%.2,3,10,14,20,21,28,33,34,37,40 A repeat operation vs 0.2%; p = 0.002) or during the last follow-up (1.7% vs increases morbidity and mortality.26,29,34 Furthermore, re- 0.7%; p = 0.004). The LC group had a considerably high- peat surgery poses a considerable financial and physical er all-cause mortality rate than the non-LC group at the burden for patients, and the risk of complications due to

306 J Neurosurg Volume 130 • January 2019

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC C. C. Chen et al.

TABLE 4. Long-term outcomes after burr hole surgery for CSDH removal LC vs Non-LC Patients Variable LC Patients Non-LC Patients HR (95% CI) p Value No. of patients 1223 2446 Repeat craniostomy at follow-up* 1 mo 266 (21.7) 470 (19.2) 1.14 (0.99–1.30) 0.070 3 mos 277 (22.6) 498 (20.4) 1.12 (0.98–1.28) 0.107 6 mos 283 (23.1) 512 (20.9) 1.11 (0.97–1.27) 0.123 1 yr 285 (23.3) 516 (21.1) 1.11 (0.97–1.27) 0.124 Last 299 (24.4) 535 (21.9) 1.12 (0.99–1.28) 0.081 Craniotomy to remove SDH at follow-up 3 mos 12 (1.0) 4 (0.2) 6.02 (1.94–18.67) 0.002 6 mos 14 (1.1) 8 (0.3) 3.52 (1.48–8.38) 0.005 Last 21 (1.7) 16 (0.7) 2.62 (1.37–5.02) 0.004 All-cause mortality at follow-up 3 mos 231 (18.9) 190 (7.8) 2.55 (2.11–3.09) <0.001 1 yr 396 (32.4) 354 (14.5) 2.49 (2.15–2.87) <0.001 5 yrs 691 (56.5) 751 (30.7) 2.28 (2.06–2.53) <0.001 Last 799 (65.3) 937 (38.3) 2.36 (2.14–2.59) <0.001 Values are shown as the number of patients (%) unless indicated otherwise. * HRs were estimated by considering all-cause mortality as a competing risk factor by using the Fine and Gray method.

FIG. 2. All-cause mortality (A), cumulative incidence of repeat craniostomy (B), and proportion of extensive craniotomy to remove SDH (C) during follow-up in the LC and non-LC patients. Figure is available in color online only.

J Neurosurg Volume 130 • January 2019 307

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC C. C. Chen et al.

FIG. 3. Admission costs in the first 5 years of follow-up in patients with LC and non-LC patients. NTD = New Taiwan dollars. the nature of CSDH and older age is high. Many risk fac- tively accurate data of the incidence, timing, and mortality tors for recurrence of CSDH have been reported.1,3,10,20,21,​ ​ rates of repeat craniostomy and extensive craniotomy after 2 8 , ​3 2 – 3 4 , ​3 7 , 4 0 However, the definition of recurrence differs in the first craniostomy for CSDH. Most repeat procedures each study. The time interval between the initial opera- were performed within 1 month. Notably, the percentag- tion and recurrence varies from 3 months1,10,20,28,40 to more es of patients who underwent repeat craniostomy did not than 6 months.33,36 The laterality of recurrent hematoma differ between the LC and non-LC groups. Clinically, if does not differ considerably among related studies. Most coagulation function in the LC group is poor, we would studies enrolled patients with ipsilateral hematomas,10,20,​28,​ probably attempt to repeat burr hole drainage or just adopt 33,36 but several enrolled patients with both ipsilateral and conservative treatment rather than craniotomy in the case contralateral hematomas.27,31 However, after recurrence, of recurrence. However, the number of patients in the LC the operation strategy might differ depending on the deci- group who required extensive craniotomy to remove SDH sions of different surgeons. Some surgeons perform repeat was significantly higher on hospital admission and dur- craniostomy once or twice, whereas others use extensive ing follow-up. Possible reasons for the higher ratio of cra- craniotomy to directly remove CSDHs. In our study, be- niotomy for SDH are as follows: 1) occurrence of acute cause we used claims data from NHIRD, the ICD-9-CM SDH after craniostomy; 2) failure of repeat craniostomies and NIH procedure coding system cannot distinguish the requiring further extensive craniotomy; and 3) new-onset laterality of the first craniostomy. Furthermore, the coding SDH resulting from a new . Irrespective of system does not indicate whether repeat craniostomy was these reasons, patients with LC who underwent simple required for ipsilateral, contralateral, or bilateral hemato- craniostomy to remove a CSDH were vulnerable and were mas. Therefore, we could not define the true recurrence of at a higher risk of postoperative complications or needed CSDH. However, our large database cohort provides rela- further extensive craniotomy to remove the CSDH. Oth-

FIG. 4. All-cause mortality stratified by modified CTP scoresA ( ) and CIC (B) for patients with LC before propensity score match- ing. Figure is available in color online only.

308 J Neurosurg Volume 130 • January 2019

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC C. C. Chen et al.

TABLE 5. All-cause mortality of the patients with LC stratified by and multiple-organ involvement. The CTP classification CTP score has been shown in multiple studies to correlate with peri- operative mortality in patients who undergo various types CTP Score CTP Score CTP Score 7,8,11,16,35 Variable 0–1 2–3 4–5 of extrahepatic major surgery and trauma. In acute TBI, the rates of complications, hemorrhage, and mortality No. of patients 891 241 93 increase gradually with the severity of LC, as indicated by All-cause mortality the progression of the CTP classification from CTP class 7 In-hospital mortality 63 (7.1) 27 (11.2) 16 (17.2) A to C. Although we could not obtain detailed data from At follow-up NHIRD, we attempted to modify the scoring system of the CTP classification according to similar complication cate- 3 mos 132 (14.8) 66 (27.4) 33 (35.5) gories. As expected, prognosis worsened with the severity 1 yr 240 (26.9) 104 (43.2) 53 (57.0) of LC, as indicated by the modified CTP scores (Fig. 4A 5 yrs 461 (51.7) 160 (66.4) 71 (76.3) and Table 5). Similarly, patients with CIC had more severe Last 542 (60.8) 180 (74.7) 78 (83.9) complications, more frequent postoperative complications, and higher mortality than other patients (Fig. 4B). Values are shown as the number of patients (%). Finally, in some studies, seizure has been considered an independent risk factor for CSDH recurrence.10,40 Seizure erwise, patients with LC who could undergo a cranioto- is frequently associated with brain atrophy and repeated my might have a more favorable preoperative condition, head trauma, which play crucial roles in the recurrence relatively normal coagulation function, and strong family of CSDH. In the present study, preoperative seizure his- support. Therefore, there may be numerous patients who tory did not differ between the LC and non-LC groups. required craniotomies but did not receive craniotomies. Of However, postoperative seizure was considerably higher in course, for those patients who accepted nonsurgical treat- the LC group. Postoperative seizure might directly result ment after recurrent CSDH, the clinical outcome must be from additional craniotomy to remove SDH and LC-relat- devastating, which explains the increased mortality rate ed hepatic encephalopathy. Frequent might imply in the LC group. These results are highly consistent with the use of antiepileptic agents, such as sodium valproate, clinical observations. which has also been reported as a risk factor for CSDH recurrence.10,40 Effect of LC in Patients With CSDH Nevertheless, to date, no strict guidelines are avail- For patients with acute (TBI), able for blood transfusion before brain surgery and sei- LC is a strong risk factor for postoperative complica- zure prophylaxis in patients with LC and CSDH. Despite tions.7,22 In a 5-year study based on data from the National an improved survival rate due to intensive preoperative Trauma Databank in America,22 patients with LC and and perioperative management that included correcting TBI required significantly more ventilator days and ex- coagulation parameters, improving nutritional status, and hibited 2-fold higher mortality than did non-LC patients controlling preoperative ascites, the mortality rate remains with TBI. In our previous retrospective study,7 the rates high in patients with acute TBI and patients undergoing of postoperative complication, rebleeding, and mortality trauma surgery. This also appears to be the case for pa- were 84.4%, 68.8%, and 37.5%, respectively. However, in tients undergoing minor head surgeries to treat conditions cases involving mild head injury or minor surgery, such as such as CSDH. burr hole craniostomy for CSDH, the results and prognosis remain unknown. In some previous studies, drug-related Limitations coagulopathy and the use of antiplatelet agents have been Despite robust results, this study had several limitations. reported to contribute to postoperative complications and The strict review and investigation of the preoperative have been associated with CSDH recurrence.4,10,32 Cirrho- brain through CT by NHI ensures that the surgical indica- sis-related complications may be associated with acquired tions are accurate. However, the registry data of NHIRD thrombocytopenia and coagulopathy resulting from hy- and ICD-9-CM do not include detailed CT findings, such persplenism, impaired liver function with decreased fi- as laterality of CSDH, width of the hematoma, and density brinogen and increased , damaged systemic and internal architecture of the hematoma. These imag- vessel walls and deficient platelet aggregation, induction of ing findings could reveal potential factors for effectively hypertension, and activation of the clotting cascade.18,25 In predicting recurrence. In Taiwan, most neurosurgeons the present study, the number of postcraniostomy bleeding perform craniostomy to treat CSDH by using burr hole episodes was not significantly higher, except in the case of surgery with subdural drainage. However, the number of SDH because of the higher number of blood transfusion burr holes and length of indwelling placement for procedures. In contrast to cases of acute TBI, intensiv- drainage cannot be determined from the coding system. ists usually have more time to correct bleeding tendency Thus, factors such as level and functional before craniostomy in CSDH cases, and recurrence may status before operation remain unknown, and hence the develop weeks or months later. However, other complica- functional outcomes (i.e., modified Rankin Scale score, tions and the mortality rate were still considerably high, Barthel Index, or neurological sequela after procedures) either while in the hospital or during follow-up. Coagu- cannot be compared. Therefore, we employed admission lopathy and poor prognosis were related to the severity of and readmission costs to reflect the long-term outcomes. LC and were due to associated systemic vascular insults Similarly, many biochemical indices could not be deter-

J Neurosurg Volume 130 • January 2019 309

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC C. C. Chen et al. mined. Although we used the diagnosis of coagulopathy, subdural hematoma. Clin Neurol Neurosurg 115:981–984, platelet transfusion, and fresh frozen plasma transfusion 2013 as replacements for bleeding tendency and severity of 5. Austin PC: Statistical criteria for selecting the optimal num- ber of untreated subjects matched to each treated subject the LC, accurate data, such as the international normal- when using many-to-one matching on the propensity score. ized ratio, platelet count, and time of correction, are un- Am J Epidemiol 172:1092–1097, 2010 known. These may be crucial factors in adverse surgical 6. Chari A, Hocking KC, Edlmann E, Turner C, Santarius T, outcomes. Otherwise, patients with LC who could undergo Hutchinson PJ, et al: Core outcomes and common data ele- craniotomy might be selected because they have a more ments in chronic subdural hematoma: a systematic review favorable preoperative condition, normal coagulation of the literature focusing on baseline and peri-operative care function, and strong family support. Therefore, a relatively data elements. J Neurotrauma 33:1569–1575, 2016 7. Chen CC, Hsu PW, Lee ST, Chang CN, Wei KC, Wu CT, et large percentage of patients may require craniotomy but al: Brain surgery in patients with liver cirrhosis. J Neuro- might never receive craniotomy. Despite these limitations, surg 117:348–353, 2012 our study included a large population with a diverse group 8. Chen CC, Huang YC, Yeh CN: Neurosurgical procedures of patients who underwent long-term follow-up at various in patients with liver cirrhosis: a review. World J Hepatol hospitals. Therefore, our findings may have broad gener- 7:2352–2357, 2015 alizability and provide a true reflection of the outcomes, 9. Chen CH, Wang MH, Wang JH, Hung CH, Hu TH, Lee SC, complications, mortality rate, and costs of craniostomy for et al: Aflatoxin exposure and hepatitis C virus in advanced CSDH. liver disease in a hepatitis C virus endemic area in Taiwan. Am J Trop Med Hyg 77:747–752, 2007 10. Chon KH, Lee JM, Koh EJ, Choi HY: Independent predictors Conclusions for recurrence of chronic subdural hematoma. Acta Neuro- chir (Wien) 154:1541–1548, 2012 Craniostomy is a simple and effective method for treat- 11. Christmas AB, Wilson AK, Franklin GA, Miller FB, Rich- ing CSDH. However, when patients with LC undergo cra- ardson JD, Rodriguez JL: Cirrhosis and trauma: a deadly niostomy, the surgical mortality rate increases and they duo. Am Surg 71:996–1000, 2005 tend to require further extensive craniotomy. With in- 12. Dangleben DA, Jazaeri O, Wasser T, Cipolle M, Pasquale M: creasing severity of LC, the mortality rate also increases. Impact of cirrhosis on outcomes in trauma. J Am Coll Surg LC is common in the general patient population, and the 203:908–913, 2006 13. Demetriades D, Constantinou C, Salim A, Velmahos G, Rhee rate of this preexisting condition has remained unchanged P, Chan L: Liver cirrhosis in patients undergoing laparotomy over the past decade. Patients with LC require intensive for trauma: effect on outcomes. J Am Coll Surg 199:538– and aggressive monitoring, careful therapeutic interven- 542, 2004 tion, and corrective coagulation parameters for a longer 14. Ernestus RI, Beldzinski P, Lanfermann H, Klug N: Chronic follow-up period. We recommend that surgeons consider subdural hematoma: surgical treatment and outcome in 104 the relatively high mortality rate, poor outcomes, and cost- patients. Surg Neurol 48:220–225, 1997 effectiveness and convey this information to family mem- 15. Fine JP, Gray RJ: A proportional hazards model for the sub- distribution of a competing risk. J Am Stat Assoc 94:496– bers before surgery. 509, 1999 16. Garrison RN, Cryer HM, Howard DA, Polk HC Jr: Clarifica- Acknowledgments tion of risk factors for abdominal operations in patients with hepatic cirrhosis. Ann Surg 199:648–655, 1984 The study was based on data from NHIRD, which is provided 17. Georgiou C, Inaba K, Teixeira PG, Hadjizacharia P, Chan LS, by the NHI Administration, Ministry of Health and Welfare of Brown C, et al: Cirrhosis and trauma are a lethal combina- Taiwan, and managed by the National Health Research Institutes tion. World J Surg 33:1087–1092, 2009 of Taiwan. However, the interpretation and conclusions contained 18. Grønbaek H, Johnsen SP, Jepsen P, Gislum M, Vilstrup H, herein do not represent these institutions. Drs. CC Chen, SW Chen, Tage-Jensen U, et al: Liver cirrhosis, other liver diseases, and and AH Chou had complete access to all data used in this study risk of hospitalisation for intracerebral haemorrhage: a Dan- and take responsibility for the integrity of the data and accuracy ish population-based case-control study. BMC Gastroen- of the analysis. terol 8:16, 2008 We thank Alfred Hsing-Fen Lin and Zoe Ya-Jhu Syu for assist- 19. Hsing AW, Ioannidis JP: Nationwide population science: les- ing in the statistical analysis. We also acknowledge Wallace Aca- sons from the Taiwan National Health Insurance Research demic Editing for editing this manuscript. Database. JAMA Intern Med 175:1527–1529, 2015 20. Kim J, Moon J, Kim T, Ahn S, Hwang G, Bang J, et al: Risk factor analysis for the recurrence of chronic subdural hema- References toma: a review of 368 consecutive surgical cases. Korean J 1. Abouzari M, Rashidi A, Rezaii J, Esfandiari K, Asadollahi Neurotrauma 11:63–69, 2015 M, Aleali H, et al: The role of postoperative patient posture 21. Koivisto T, Jääskeläinen JE: Chronic subdural haematoma— in the recurrence of traumatic chronic subdural hematoma to drain or not to drain? Lancet 374:1040–1041, 2009 after burr-hole surgery. 61:794–797, 2007 22. Lustenberger T, Talving P, Lam L, Inaba K, Branco BC, Plu- 2. Andersen-Ranberg NC, Poulsen FR, Bergholt B, Hundsholt rad D, et al: Liver cirrhosis and traumatic brain injury: a fatal T, Fugleholm K: Bilateral chronic subdural hematoma: uni- combination based on National Trauma Databank analysis. lateral or bilateral drainage? J Neurosurg 126:1905–1911, Am Surg 77:311–314, 2011 2017 23. Mansour A, Watson W, Shayani V, Pickleman J: Abdominal 3. Arbit E, Patterson RH Jr, Fraser RA: An implantable subdu- operations in patients with cirrhosis: still a major surgical ral drain for treatment of chronic subdural hematoma. Surg challenge. Surgery 122:730–736, 1997 Neurol 15:175–177, 1981 24. McCaughan GW: Asian perspectives on viral hepatitis: 4. Aspegren OP, Åstrand R, Lundgren MI, Romner B: Antico- hepatitis C virus . J Gastroenterol Hepatol 15 agulation therapy a risk factor for the development of chronic Suppl:G90–G93, 2000

310 J Neurosurg Volume 130 • January 2019

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC C. C. Chen et al.

25. McCormick PA, Murphy KM: Splenomegaly, hypersplenism ing findings and closed system drainage in the recurrence of and coagulation abnormalities in liver disease. Best Pract chronic subdural hematoma. J Neurosurg 87:870–875, 1997 Res Clin Gastroenterol 14:1009–1031, 2000 37. Wakai S, Hashimoto K, Watanabe N, Inoh S, Ochiai C, Nagai 26. Mellergård P, Wisten O: Operations and re-operations for M: Efficacy of closed-system drainage in treating chronic chronic subdural haematomas during a 25-year period in a subdural hematoma: a prospective comparative study. Neuro- well defined population. Acta Neurochir (Wien) 138:708– surgery 26:771–773, 1990 713, 1996 38. Weigel R, Schmiedek P, Krauss JK: Outcome of contem- 27. Mori K, Maeda M: Surgical treatment of chronic subdural porary surgery for chronic subdural haematoma: evidence hematoma in 500 consecutive cases: clinical characteristics, based review. J Neurol Neurosurg Psychiatry 74:937–943, surgical outcome, complications, and recurrence rate. Neurol 2003 Med Chir (Tokyo) 41:371–381, 2001 39. Wu CY, Chen YJ, Ho HJ, Hsu YC, Kuo KN, Wu MS, et al: 28. Nakaguchi H, Tanishima T, Yoshimasu N: Factors in the Association between nucleoside analogues and risk of hepati- natural history of chronic subdural hematomas that influence tis B virus–related hepatocellular carcinoma recurrence fol- their postoperative recurrence. J Neurosurg 95:256–262, lowing liver resection. JAMA 308:1906–1914, 2012 2001 40. Yamamoto H, Hirashima Y, Hamada H, Hayashi N, Ori- 29. Nayil K, Ramzan A, Sajad A, Zahoor S, Wani A, Nizami gasa H, Endo S: Independent predictors of recurrence of F, et al: Subdural hematomas: an analysis of 1181 Kashmiri chronic subdural hematoma: results of multivariate analysis patients. World Neurosurg 77:103–110, 2012 performed using a logistic regression model. J Neurosurg 30. Needham DM, Scales DC, Laupacis A, Pronovost PJ: A sys- 98:1217–1221, 2003 tematic review of the Charlson comorbidity index using Ca- nadian administrative databases: a perspective on risk adjust- ment in critical care research. J Crit Care 20:12–19, 2005 31. Oh HJ, Lee KS, Shim JJ, Yoon SM, Yun IG, Bae HG: Postop- Disclosures erative course and recurrence of chronic subdural hematoma. The authors report no conflict of interest concerning the materi- J Korean Neurosurg Soc 48:518–523, 2010 als or methods used in this study or the findings specified in this 32. Pang CH, Lee SE, Kim CH, Kim JE, Kang HS, Park CK, et paper. al: Acute intracranial bleeding and recurrence after bur hole craniostomy for chronic subdural hematoma. J Neurosurg Author Contributions 123:65–74, 2015 33. Santarius T, Kirkpatrick PJ, Ganesan D, Chia HL, Jalloh Conception and design: Chou, CC Chen, SW Chen. Acquisition I, Smielewski P, et al: Use of drains versus no drains after of data: Chou, SW Chen, TH Chen. Analysis and interpretation burr-hole evacuation of chronic subdural haematoma: a ran- of data: CC Chen, SW Chen, Tu. Drafting the article: CC Chen. domised controlled trial. Lancet 374:1067–1073, 2009 Critically revising the article: CC Chen, SW Chen, Tu, Huang. 34. Song DH, Kim YS, Chun HJ, Yi HJ, Bak KH, Ko Y, et al: Reviewed submitted version of manuscript: CC Chen, SW Chen, The predicting factors for recurrence of chronic subdural Liu, AYC Wang, Lee. Approved the final version of the manu- hematoma treated with burr hole and drainage. Korean J script on behalf of all authors: Chou. Statistical analysis: Cheng, Neurotrauma 10:41–48, 2014 SY Wang. Administrative/technical/material support: Chou, 35. Thielmann M, Mechmet A, Neuhäuser M, Wendt D, Tossios Huang, Liu, AYC Wang, Lee, TH Chen, Cheng, SY Wang. Study P, Canbay A, et al: Risk prediction and outcomes in patients supervision: Chou, Tu, Huang, TH Chen. with liver cirrhosis undergoing open-heart surgery. Eur J Cardiothorac Surg 38:592–599, 2010 Correspondence 36. Tsutsumi K, Maeda K, Iijima A, Usui M, Okada Y, Kirino T: An-Hsun Chou: Chang Gung Memorial Hospital, Taoyuan City, The relationship of preoperative magnetic resonance imag- Taiwan. [email protected].

J Neurosurg Volume 130 • January 2019 311

Unauthenticated | Downloaded 10/07/21 03:13 AM UTC