Discussion
Given the lack of definitive evidence available in the literature, there is uncertainty over the impact of the timing of surgery for CSDH on outcome. This observational cohort study, involving 656 patients with CSDH, investigated whether the time interval between referral and surgery impacted on functional outcome at discharge (mRS), complication rates, recurrence requiring reoperation, length of stay, and survival. While there was variation between units, the median time to surgery was 1 day, so in general it appears that most patients receive surgery in a timely manner.
Time to surgery demonstrated a positive linear relationship with length of stay. We found that a number of variables were associated with a longer time interval between referral and surgical intervention. These included increasing patient age, lower mRS, higher GCS, use of antiplatelet agents, presence of comorbidities, and presence of bilateral CSDH (p<0.05 for all variables).
Although interactions between age and comorbidities were tested and not found to be significant when building the model for this study, large epidemiological studies have demonstrated increasing comorbidities with age.15 16 Published reports by the National Confidential Enquiry into Patient Outcome and Death have highlighted the importance of multidisciplinary decision-making in surgery in the elderly where there is an increased risk of perioperative morbidity and mortality.17 Integrated care pathways have been effectively developed in Trauma and Orthopaedics for patients with a fractured neck of femur, with best practice tariffs if the patient is admitted under the joint care of a consultant orthopedic surgeon and a consultant geriatrician, and is postoperatively cared for by a geriatrician-directed multiprofessional rehabilitation team.18 Such pathways and systemic incentives have only recently started to be developed for elderly patients with CSDH, and seem to show early promising results including better preoperative optimization of patients.19 20
As antiplatelet agents are usually stopped prior to surgery, it is common practice to delay surgery for 5-10 days, depending on the agent, or transfuse platelets if surgery needs to be undertaken on an urgent basis.21 22 Notably, class III evidence demonstrates that waiting 3 days is sufficient to proceed to surgery after cessation of antiplatelet therapy with no recurrence after this time point.23
Evidence that time to surgery does not have a substantial adverse impact on clinical outcomes, as suggested in this study, could allay pressure from the managing team and encourage optimization for surgery. Our findings are in line with those of a recent retrospective cohort study from Sweden showing that increased time from CT scan to surgical evacuation for CSDH did not negatively impact outcomes, when surgery was performed within a clinically appropriate time frame.24 It appears that patients at risk of deterioration or those who are neurologically impaired are already being prioritized, as we found that a more unfavorable initial mRS, and lower GCS were associated with a shorter time to surgery. It would be ideal for the patient to undergo surgery as soon as practically feasible to begin the process of recovery and resolution of deficit(s), and we advocate for this to remain the ultimate aim in clinical practice. However, this has to be weighed against the importance of comprehensive preoperative assessment and optimization.
The trend between a time interval of 7 days or more and unfavorable outcome at discharge means that one cannot completely rule out a relationship between increasing time to surgery and worse functional outcome at discharge. Due to a paucity of studies in this field and risk of a type II error, further research should be carried out to investigate this potential relationship. We believe that decisions on when to intervene should always be guided by clinical assessment and the patient’s condition. When interpreting the results of this study, it must be remembered that this was an observational study without any randomization. Therefore, residual confounding and confounding by indication may still exist.
The linear relationship between time to surgery and length of stay has important service provision implications. The management of CSDH already represents a significant burden to neurosurgical service provision, with an estimated incidence of 8.2/100 000/year after 70 years of age.25 This is expected to increase in the context of increasing use of antiplatelet and anticoagulant therapy in an aging population.26 There are broader implications, with head-injured patients representing a significant source of financial deficits for healthcare organizations in the UK.27 Consequently, there is an important consideration to optimize neurosurgical service provision by reducing the length of stay for patients with CSDH. There is also limited evidence to suggest an increased length of stay is associated with a worse prognosis in CSDH.28
Limitations and future work
Time to surgery was determined based on the time point of referral to neurosurgical services; a study looking at the time from the onset of neurological deterioration (eg, onset of hemiparesis) or time from diagnosis (CT scan) to surgery would be interesting, but these data were not available. Other pertinent data not available included the time of day the surgery was completed and whether or not the patient required a planned or unplanned admission to an intensive care unit. Discharge data also did not specify how long the patient remained in another hospital, such that overall in-hospital length of stay postoperatively could not be determined. Only patients undergoing burr-hole evacuation were included; however, this is the most common surgical procedure used to treat CSDH in the UK.8 29 It would be important to consider whether our findings are applicable to other patient groups such as those undergoing different procedures. Additionally, it would be important to establish whether risk stratification should be used when prioritizing patients for surgery.
Although this is a large sample of 656 patients, there is a risk of a type II error particularly given the conclusions derived from this study. This is exacerbated by selection bias due to the individual surgeon decision-making on optimal timing of surgery. The outcomes investigated were relatively short-term, with four out of five being assessed at discharge. The median length of stay in neurosurgical units was 7 days in the original study.8 The outcome with the longest follow-up time was recurrence rate, which was assessed for 60 days postoperatively; evidence suggests this is the highest risk time window and thereafter the risk of recurrence reduces.30 Conclusions about the longer-term outcome cannot be drawn at this stage. There is scope to expand this in further studies to lengthen follow-up to 6 months or longer. This would be helpful to understand the disease progression and impact on function,31 especially since previous studies highlighted that CSDH is comparable to hip fracture as a sentinel event for underlying systemic pathology with increased 1 year mortality.31,32 This is particularly relevant for an elderly patient group; in this context, 6 month or 1 year outcomes are useful tools when completing holistic/comprehensive assessments. Additionally, complications following CSDH could be further classified into mild, moderate and severe groups based on clinical relevance, to ascertain if any of these groups are affected individually, as described in a recent population-based study investigating predictors of recurrence and complications following CSDH.33 This was not possible in this paper, as such data were not collected.
Despite its limitations, this is a large UK wide study including 656 patients treated across 26 neurosurgical centers. Therefore, it is likely to be representative of current UK neurosurgical practice. As previously stated, whether timeliness of surgical intervention in CSDH has an impact on patient outcome has not been studied widely, and uncertainty remains. This study adds to the existing knowledge and prompts a new set of questions. The findings of the study apply to a state-funded health system where universal health coverage is provided, and therefore the conclusions may not necessarily be generalizable to different healthcare systems.
Although our study did not evaluate a specific novel intervention, it is still useful to classify it according to the IDEAL (Idea, Development, Exploration, Assessment, Long-term study) framework.34 It can be viewed as an IDEAL stage 2b study, as exploratory analyses published as part of the primary paper8 identified a number of modifiable factors associated with better outcome, including use of two burr-holes and early mobilization. These are now being considered for evaluation with an IDEAL stage 3 study. Additionally, it can be viewed as an IDEAL stage 4 study with regards to the use of subdural drains, an intervention which was found to be beneficial in a previous high-quality randomized controlled trial.35 Our prospective, multicenter, observational study demonstrated that the national recurrence rate was 9%, very similar to that observed in the drain arm of the aforementioned trial. More importantly, the multivariate analysis showed that not using a drain independently predicted recurrence and unfavorable functional outcome, validating the effectiveness of subdural drains in a real-world setting.
