Recurrent Chronic Subdural Hematoma After Burr-Hole Surgery and Postoperative Drainage: A Systematic Review and Meta-Analysis

BACKGROUND AND OBJECTIVE: Reported recurrence rates of chronic subdural hematoma treated by burr-hole surgery with postoperative drainage vary considerably in the literature. We performed a systematic review and meta-analysis to define the recurrence rate of burr-hole surgery with postoperative drainage. METHODS: PubMed and EMBASE were searched, and Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines were followed. We used the Newcastle-Ottawa scale and Cochrane risk-of-bias tool for quality assessment of included studies and the random-effects model to calculate pooled incidence rates in R with the metaprop function if appropriate. RESULTS: The search yielded 2969 references; 709 were screened full text, and 189 met the inclusion criteria. In 174 studies (34 393 patients), the number of recurrences was reported as per patient and 15 studies (3078 hematomas) reported the number of recurrences per hematoma, for a pooled incidence of 11.2% (95% CI: 10.3-12.1; I2 = 87.7%) and 11.0% (95% CI: 8.6-13.4; I2 = 78.0%), respectively. The pooled incidence of 48 studies (15 298 patients) with the highest quality was 12.8% (95% CI 11.4-14.2; I2 = 86.1%). Treatment-related mortality (56 patients) has a pooled incidence of 0.7% (95% CI 0.0-1.4; I2 = 0.0%). CONCLUSION: The recurrence rate of chronic subdural hematoma treated by burr-hole surgery and postoperative drainage is 12.8%.

O ver the past 3 decades, the incidence of chronic subdural hematoma (cSDH) has nearly tripled for patients older than 80 years, and because of the aging population, it is expected to become the most common cranial neurosurgical condition among adults by 2030. 1,2 Although burr-hole surgery is the first treatment of choice for symptomatic cSDH, [3][4][5] this procedure inherently carries a risk of a recurrent hematoma. Reported recurrence rates vary considerably (0%-33%), probably because surgical strategies also vary widely, from 1 or 2 burr holes with or without postoperative drainage to subdural or epidural location of drains. [6][7][8] In recent years, less invasive, nonsurgical options to treat cSDH have been explored, such as medical treatment (steroids or tranexamic acid) or embolization of the middle meningeal artery, but randomized controlled trials are still lacking or could not prove any benefit. 9,10 To accurately calculate how many patients need to be included in such randomized controlled trials (RCTs), sample size calculations should be based on reliable outcome measures in the control group, treated according to modern standards.
As the use of postoperative drainage has been shown to be clearly superior to burr-hole surgery without postoperative drainage, 5 we performed a systematic review and meta-analysis to define a reliable rate of recurrence after burr-hole surgery with postoperative drainage in patients with a cSDH.

METHODS Search Strategy and Study Selection
The results of this systematic review are reported in accordance with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses checklist. 11 This review was not registered, and a protocol was not prepared. We searched Medline (ovid) and EMBASE (embase.com) on 4th June 2021 from inception, using terms on subdural hematoma and burr hole, to identify all studies from 1946 reporting on recurrent cSDH after burr-hole surgery (see Supplementary Table 1, http://links.lww.com/ONS/A922, entitled "Syntax search").
Articles with less than 25 patients, certain publication types (letter to the editor, commentary, survey, narrative review, study protocol), or records containing other language than English, German, French, Spanish, or Dutch were excluded. Systematic reviews and meta-analyses were only taken into consideration for forward and backward snowballing to identify any additional relevant articles. Records without abstracts were automatically passed into the full-text screening phase.
Studies were eligible for inclusion if (1) patients were 18 years or older; (2) were diagnosed with a cSDH, acute-on-chronic subdural hematoma, or when a study described both cSDH and subdural hygroma; (3) treated with burr-hole surgery and postoperative drainage; and (4) the number of recurrences or recurrence rate was explicitly reported. In Figures 1-3, we present 3 images of a cSDH, acute-on-chronic subdural hematoma, and a subdural hygroma. Postoperative drainage had to be performed in a minimum of 95% of cases to be included. A definition of a cSDH recurrence was not mandatory for inclusion, but the study had to describe recurrences or reoperations. When a study consisted of multiple treatment groups with different strategies and the patient group of interest was distinguishable (burr-hole surgery with additional drainage), the data regarding the number of patients, recurrences, and postoperative drainage of that specific subgroup of patients were collected. In case a study described that patients did not receive a postoperative drain because of brain expansion during surgery, the study was included.
Exclusion criteria were (1) medical treatment for cSDH or embolization of the middle meningeal artery before, or after, burr-hole surgery; (2) other type of surgery than burr-hole surgery; (3) enlarged and endoscopy-assisted

Data Extraction
Two investigators (R.L. and M.F.) independently screened title and abstract to identify potential suitable records. Differences in judgment were discussed and resolved with mutual consent. Next, both investigators independently screened full texts, based on the inclusion and exclusion criteria and independently performed forward and backward snowballing of systematic reviews and meta-analyses to identify any additional relevant articles. Systematic reviews and meta-analyses were only taken into consideration for forward and backward snowballing to identify any additional relevant articles. A third rater (W.P.V.) adjudicated any discrepancies.

Data Collection
The following preoperative characteristics were extracted: the number of patients, mean or median age, sex, antithrombotic therapy, history of trauma, Glasgow Coma Scale (GCS) score, hematoma laterality, and total number of cSDHs (bilateral hematomas counted as 2 hematomas). Regarding operative treatment the following characteristics were collected: number of burr-holes made as per standard protocol per side, number of patients receiving irrigation, type of irrigation used, number of patients receiving a postoperative drain, type of drain, number of patients not receiving a drain because of brain expansion during surgery, and postsurgical treatment. Whenever the definition of a recurrent cSDH was stated, this was noted, as well as the number of recurrences, whether the recurrence was measured per patient or per hematoma, number of recurrences treated by surgery, and number of patients in whom the recurrence was detected by computed tomography scan, or clinical symptoms. Furthermore, data on frequency and causes of mortality directly related to cSDH treatment and the occurrence of re-recurrence were collected when possible.

Quality Assessment
The Newcastle-Ottawa scale for cohort studies, which is validated for assessing the quality of observational cohort studies, was used. 12 According to the assigned number of stars, the following subdivision was made: 7-9, high quality; 4-6, high risk of bias; and 0-3, very high risk of bias. In case there was no nonexposed cohort in an included study, the maximum amount of stars to be assigned was 8. Thereby, the subdivision for these studies subsequently was 6-8, high quality; 3-5, high risk of bias; and 0-2, very high risk of bias. In addition, the Cochrane risk of bias for randomized controlled trials was used for studies in which a randomization method was used. 13 Risk-of-bias judgment was determined by the overall result of the 6 domains: low risk of bias (low risk of bias for all domains), some concerns (some concerns in at least 1 domain for this result but not to be at high risk of bias for any domain), or high risk of bias (high risk of bias in at least 1 domain) accordingly. Each paper was graded and assigned a score by 2 authors [R.L. and D.V.]. A subgroup analysis was performed of the studies with the highest quality, meaning studies with Newcastle-Ottawa scale score 7-9 and score 6-8 in case of the absence of a nonexposed cohort and low risk of bias judgment determined on the Cochrane risk of bias for randomized controlled trials.
used to determine single-arm prevalence, whenever possible. In addition, for single-arm prevalence of recurrence, a forest plot was created, and 95% prediction interval (PI) was calculated with the metaprop function in case of at least 10 included studies. 16 The prediction interval helps in the clinical interpretation of the heterogeneity by estimating what true treatment effects can be expected in future settings. 17 Single-arm prevalence analyses of a variable consisting of either 0% or 100% incidences in the included studies was determined by dividing the total number of cases in the included studies by the total number of cases of this variable.
We assessed the presence of bias using a funnel plot. The funnel plot was created with the pooled incidence and random-effects model by plotting the recurrence rate against the standard error.

Literature Search
The online search yielded a total of 4140 records. Screening the references of 33 systematic reviews yielded 15 additional articles. Duplicates were removed and yielded 2969 articles. After screening on title and abstract, 2260 records were excluded. The full texts of the remaining 709 articles were assessed for eligibility. Of the remaining full texts, 520 were excluded. In total, 189 articles reporting on 36 971 patients were selected for further analysis (Figure 4). Most studies were of a retrospective nature (n = 166) and otherwise randomized controlled trials (n = 23).

Quality Assessment
The funnel plot shows that included studies do not seem to be symmetrically ranged around the pooled incidence of recurrence ( Figure 5). Smaller studies seem to cause a more scattered array of recurrence rate. The risk-of-bias scores are shown in Table 1.
Regarding the Newcastle-Ottawa scale and adjusted scale for cohort studies, the highest quality was observed in 50 (30.9%) studies, high risk of bias was observed in 104 (64.2%) studies, and very high risk in 8 (4.9%) studies.

Treatment Modalities
Pooled incidences and recurrence rates per irrigation method, type of postoperative drainage, and drainage time are shown in Tables 4-6, respectively. Ringer solution (also titled Hartmann solution) showed a recurrence rate of 21.4%, and irrigation fluid at room temperature (22°C) showed a recurrence rate of 12.0%.

DISCUSSION
This systematic review shows a pooled incidence of recurrence of 12.8% after burr-hole surgery and postoperative drainage in patients with a chronic subdural hematoma. Because this recurrence rate has been accurately determined, it can serve as an outcome measure for power calculations to determine sample sizes for future randomized clinical trials.
Most of the included studies were of a retrospective nature, the majority was judged to carry a high to very high risk of bias, and the RCTs were all considered to be of low quality. The overall pooled incidence is 11.2%. In our subgroup analysis of only those studies of the highest quality, the pooled incidence of recurrence was remarkably higher at 12.8%, which could infer that studies with poor quality systematically underestimate the recurrence rate. Therefore, the most reliable pooled incidence of recurrence is 12.8%. Our analyses of the pooled incidence of recurrence at 3 and 6 months shows that both periods do not accurately approach the overall recurrence rate of cSDH, possibly because of the low number of studies of which data could be derived from.
Recurrence of a cSDH leads to rehospitalization and reoperation, resulting in worse clinical outcome, loss of independency in these often very frail patients, and a significant increase in health care costs. [205][206][207][208] Prevention of recurrences is therefore important, and various nonsurgical treatments are now being explored in randomized clinical trials, including medication (tranexamic acid). 9,209 In a recent RCT, the use of dexamethasone vs placebo demonstrated that fewer recurrences occurred but that less favorable outcomes and more adverse events were noted in the dexamethasone group. 10 This systematic review also provided information on recurrence rates depending on the method of irrigation used. Ringer solution (also titled Hartmann solution) showed a relatively high recurrence rate, as did irrigation fluid at room temperature (22°C) when compared with irrigation at body temperature (37°C), which is in line with a recent study. 210 The pooled recurrence rate of subgaleal drainage was lower compared with subdural and subperiosteal drainage, although subgaleal drainage was used in only 8 studies, making it hard to draw firm conclusions. In the literature, recurrence rates after burr-hole surgery do not appear to be affected by drain location. 211 However, 2 recent systematic reviews and metaanalysis showed that the insertion of a subperiosteal drain resulted in a significantly lower recurrence rate compared with a subdural drain but did not take into account a subgaleal drain. 212,213 Furthermore, we researched whether patient characteristics or treatment modalities could be of influence on the high-quality recurrence rate by performing subgroup analyses. Only the irrigation method, saline/normal saline/isotonic saline/physiological saline, seems to be associated with recurrence because this rate is higher compared with the other determined rates in between treatment modalities. The pooled incidences of recurrence in the high-quality group could therefore be regarded as more stable and should be used as most reliable recurrence rates.

Limitations
A relatively high statistical heterogeneity was observed for nearly all calculated pooled incidences in the current systematic review and meta-analysis. This could be due to clinical heterogeneity because clinical diversity in studies, despite the strict inclusion and exclusion criteria, is inevitable. A difference in baseline characteristics of included studies was observed. Some studies for example applied certain age restrictions, and some only included unilateral hematomas or patients with antithrombotic therapy, and across studies, different treatment methods regarding number of burr-holes, type of irrigation, and location of the postoperative drain and different definitions of recurrence were used. Moreover, methodological heterogeneity also attributes to the statistical heterogeneity because retrospective and prospective studies and RCTs are included. When measuring prevalence of a phenomenon in diverse environments, it is expected that highly heterogeneous studies are assembled. 214 Because all included studies present a recurrence rate and therefore essentially measure the same parameter, we think it is worth summarizing despite clinical and methodological heterogeneity. Furthermore, in the literature, a wide range of recurrence rates are used, and the current study is a first step to accurately provide one overall recurrence rate. A subgroup analysis of studies with the highest quality was performed and still showed high heterogeneity meaning that even in these studies clinical and methodological differences are present and cannot be avoided. Bartley et al, 29 2020 Intracerebral hematoma (n = 1), basal ganglion infarction (n = 1) Bellut et al, 30 2012 Prediction interval calculation of studies with the highest quality showed that the recurrence rate in 95% of future studies is expected to range from 2% to 25%. This wide prediction interval is likely caused by the high heterogeneity of studies because the prediction interval is a summary of the spread of underlying effects in the studies included in the random-effects meta-analysis. 16 The Saline/normal saline/isotonic saline/physiological saline and Ringer solution/Hartmann solution are combined to determine the recurrence rate, and warm saline and warm Ringer solution are combined to investigate the effect of normal vs warm irrigation method. Calculated by dividing the number of patient per group by the total number of patients in which the irrigation method was described. In 9 studies, more than 2 drain locations were reported and could be separated per location and was calculated additionally.
b In 42 studies of which drain location was unknown, the first author was emailed to retrieve the drain location. Hereby, we determined 9 drain locations extra. Calculated by dividing the number of patients per group by the total number of patients in which postoperative drain location was described.
"Other" includes studies reporting 2 or more drain locations which could not be separated. created forest plots show a considerable in-between study difference regarding recurrence rate, even in the studies with the highest quality. Consequently, the calculated pooled recurrence rate should be interpreted with some caution. Fortunately, randomized controlled trials are being increasingly conducted in neurosurgery. However, for power calculations to determine sample size, it is crucial to use accurate outcome measures. The current study provides the most accurate known recurrence rate for cSDH treated by burr-hole surgery and postoperative drainage. Therefore, we propose to use the recurrence rate calculated in this study for future sample size calculations. Furthermore, to compare the results of these clinical studies, consistency in terms and definitions is essential, whereas definitions for a recurrent cSDH vary considerably in the literature. We would like to recommend using the following standardized definition of a recurrent cSDH in future studies: A recurrent cSDH is defined as recurring or aggravated clinical symptoms, caused by a radiologically proven, ipsilateral reaccumulation of cSDH within 6 months of prior surgical treatment, for which additional treatment is necessary. It then still remains debatable whether a recurrence on the contralateral side is a recurrence or a newly diagnosed cSDH.

CONCLUSION
The overall recurrence rate of chronic subdural hematoma after burr-hole surgery and postoperative drainage is 12.8%. This rate should be interpreted with some caution because observed heterogeneity of included studies is high. A unified definition of cSDH recurrence after surgical treatment is advocated to ensure reliable comparison between future studies.

Funding
This study did not receive any funding or financial support.