Neurosurgery and coronavirus: impact and challenges—lessons learnt from the first wave of a global pandemic

On 18th March 2020, our unit entered the acute “COVID-19” phase, demarcated by the day of the first pre-operatively suspected COVID-19 infection in a neurosurgical patient, and when neurosurgical rota and service changes were adopted. The phase officially ended on 15th May (59 days later) when our hospital, in line with the NHS directive, entered the recovery phase. During the COVID-19 period, data were collected prospectively on all patients referred to the acute neurosurgical service, patients who were admitted electively, and patients referred to subspecialty multidisciplinary teams (MDT). These data were then compared to those obtained in the immediately preceding 59-day period (18th January to 17th March), the “pre-COVID-19”, to assess the impact of the pandemic. The manuscript was written following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist [24].

Data on the acute neurosurgical referrals were obtained through an online Patient Care System (PCS) and data on referrals to the MDTs were obtained from the relevant coordinators. Emergency and elective operating lists were sourced through the software “Galaxy Operating Theatres”. Electronic patient records were accessed to capture the following: age, gender, diagnosis, type of procedure (emergency vs. elective, adult vs. paediatric, cranial vs. spinal), subspecialty (functional, neuro-oncology, trauma, neurovascular, skull base, spinal, paediatric neurosurgery), COVID-19 infection (pre-operatively vs. post-operatively), and post-operative complications. For COVID-19-infected patients, data on ethnicity, co-morbidities, perioperative complications, ITU/hospital stay, discharge destination, and mortality were also included. Additionally, referral data included type of referral (new vs. follow-up), pathology, MDT outcome, and treatment delay/recommendation.

Rota and service provision changes were put in place during the pandemic on 18th of March 2020. Given the rapidly evolving Public Health England [20], NHS England [15], and the Society of British Neurological Surgeons guidelines [22], a local KCH Neurosurgery working group was established to actively review the evidence and synthesize a Guidance [12] for adjustments to elective and emergency operations, theatre ventilation, and use of personal protective equipment (PPE).

KCH had its first two COVID-19-positive patients on 04th of March 2020 (Fig. 1—day 0) and reached its peak on the 8th of April 2020 with 517 COVID-19-positive inpatients of which 96 were in ITU. The first change to our neurosurgical practice was put in place on 18th March 2020 where the rota was revised from a subspecialty team-based system (consisting of 1–2 neurosurgical trainees working for a specific consultant) to a twilight rota with 1 “green” team (no COVID-19 patient contact), 1 “red” team (COVID-19-positive/suspected patient contact), and 1 stand-by team in case other team members fell ill and had to consequently self-isolate. Consultant and trainee rotas both switched to a 24-h shift with one on-site consultant at all times and another on-call in reserve.

The three daily elective operating lists were condensed into one list per day from the 23rd of March 2020. Throughout the pandemic, one emergency neurosurgical theatre remained active. The allocation of theatre resources was undertaken by the Executive board of the hospital. The neurosurgical lists were arranged/triaged by the on-call neurosurgery consultant on the day. Trust guidelines recommended that ventilation in both laminar flow and conventionally ventilated theatres should remain fully switched on during surgical procedures where patients may have COVID-19 infection. All operations were performed with full PPE, including either FFP3 mask (for aerosol-generating procedures) or fluid-resistant mask (non-aerosol-generating procedures), hat, visor, gloves and fluid-resistant disposable gown. Staff training for proper donning and doffing was mandatory.

Our inpatient neurosurgical service usually consists of one purely elective 31-bedded ward, three wards for emergency and additional elective neurosurgical patients (~ 50–60 beds), and one dedicated 12-bedded neurosurgery high dependency unit (HDU). During COVID-19, out of these, one emergency admission ward and its adjacent HDU were closed to neurosurgery admissions to provide capacity for the hospital’s general COVID-19 patients. The HDU was relocated to replace one of the other neurosurgical wards with level 2 bed numbers expanded to 32. The remaining two neurosurgical wards were divided into “dirty” (COVID-19 positive) and “clean” (non-COVID-19) wards (Fig. 2).

Nearly all face-to-face outpatient appointments were cancelled and essential appointments were conducted via phone consultations. The MDT referral process remained unchanged; however, meetings took place virtually via Microsoft® Teams (Redmond, USA) from 01st April 2020.

Pre-COVID-19, the median number of new acute referrals was 31 (range 17–45) per 24 h. During COVID-19, this decreased to 21 (range 10–34) per day. There was a statistically significant reduction in the overall number of referrals from 1847 to 1227 (Table 1; p < 0.05).

Subspecialty emergency referrals changed to proportionately fewer skull base and spinal referrals, but proportionately increased trauma, vascular, oncology, and paediatric referrals (Table 1; p < 0.01). Approximately 10% of patients referred as an emergency pre-COVID-19 (n = 190) and during COVID-19 (n = 119) were accepted for emergency transfer. There was no change to the definition of what constituted an emergency during or pre-COVID-19, namely being a condition that was life or limb threatening within a matter of days if left untreated. There was no significant change in the proportion of patients with neurological deficit, GCS ≤ 8, or age > 65 years being transferred to our neurosurgical centre (p > 0.05).

The total number of operations decreased from n = 453 (pre-COVID-19) to n = 206 (Table 1; p < 0.0001) with the daily median number of operations decreasing from 8 to 3. A higher percentage of emergency operations was performed during COVID-19 (75.2% of total, n = 155) compared to pre-COVID-19 (n = 198, 43.7% of total, p < 0.00001). There was no significant change in the proportion of cranial versus spinal operations (Table 1). Overall, significantly fewer patients aged > 65 years underwent an operation during COVID-19 (p < 0.01). The operations for adults and paediatrics per neurosurgical subspecialty changed significantly (p < 0.01) with subspecialties with a high proportion of elective work, such as functional, skull base, and spinal neurosurgery, being affected the most.

Tables 2 and 3 summarize the data on patients undergoing neurosurgery in adult and paediatric cases, respectively. The total number of adult operations performed dropped from n = 408 to n = 173 during COVID-19, with a significant amount of functional and degenerative spinal neurosurgical work being deferred or cancelled (p < 0.01; Table 2). The number of operations amongst the emergency subspecialties, such as trauma and vascular neurosurgery, also decreased during COVID-19 by approximately 50%; however, the case mix remained similar. The most common traumatic pathologies requiring emergency operation were chronic subdural hematoma (pre-COVID-19 n = 26, COVID-19 n = 19), vertebral fracture (pre-COVID-19 n = 11, COVID-19 n = 5), and acute subdural hematoma (pre-COVID-19 n = 9, COVID-19 n = 3). Although all the numbers decreased, the smallest drop in cases was amongst surgeries for chronic subdural haematomas, possibly related to their relatively more chronic presentation. Similarly, vascular operations decreased with fewer aneurysms being clipped during COVID-19 (n = 1) compared to pre-COVID-19 (n = 9).

Within the neuro-oncology service, the overall number of operations decreased from n = 60 (14.7% of total) to n = 31 (17.9% of total) during COVID-19. Similarly, the number of craniotomies for high-grade gliomas decreased from n = 31 (7.6% of total) to n = 12 (6.9% of total) during the pandemic.

Our skull base service was severely affected during COVID-19 with only 3 operations for pituitary adenoma/apoplexy being performed during COVID-19 (1.7% of total) from a previous number of 16 operations pre-COVID-19 (3.9% of total). No operations for trigeminal neuralgia, vestibular schwannoma, or chiari malformation were performed during COVID-19.

In functional neurosurgery, no new implantations for deep brain stimulation (DBS), spinal cord stimulation, or occipital nerve stimulation were performed. The battery change service for patients with movement disorders, however, continued albeit in the day case setting (DBS battery change n = 5 in both periods). All spinal surgeries decreased during COVID-19; however, notably, operations for cauda equina syndrome (pre-COVID-19 n = 26, COVID-19 n = 14) and myelopathies (pre-COVID-19 n = 28, COVID-19 n = 8) were reduced by ≥ 50% during COVID-19, whereas operations within spinal oncology category remained stable (pre-COVID-19 n = 14, COVID-19 n = 15).

The total number of referrals to the vascular MDT decreased from n = 245 to n = 161 during COVID-19 (p < 0.05). The total number of patients referred with an intracranial aneurysm decreased from n = 185 (75.5% of total pre-COVID-19) to n = 132 (82.0% of total during COVID-19; Table 4). Within that group, referred unruptured symptomatic aneurysms remained approximately stable (~ 2.5 of total). The number of AVMs referred decreased from n = 34 (13.9% of total pre-COVID-19) to n = 8 (5.0% of total during COVID-19).

Sixteen patients underwent emergency treatment pre-COVID-19, 6 of those underwent open surgery and 10 underwent endovascular treatment. During COVID-19, only 1 patient underwent emergency open surgery and 17 patients underwent emergency endovascular treatment. All elective surgery was halted during COVID-19 (Table 4).

The total number of referrals to the neuro-oncology MDT decreased from n = 443 to n = 275 during COVID-19 (p < 0.05) with the median number of referrals per MDT dropping from 53 ± 11.63 to 37 ± 9.58. There was no significant change in the ratio of new to follow-up referrals during these periods. Equally, there was no significant change in the treatment recommendation provided for patients with high-grade gliomas (HGG), low-grade gliomas (LGG), and cerebral metastases (CM) (p > 0.05; Table 5). However, there was a significant treatment delay (surgery or adjuvant therapy), n = 4 patients (0.9% of total) pre-COVID-19 versus n = 32 patients (11.6% of total, p < 0.00001) during COVID-19, with patients with a meningioma affected more severely (n = 16 overall) compared to patients with gliomas or malignant tumours (n = 7 HGG, n = 2 LGG, n = 3 CM). The most common reasons for treatment delay were surgery delay due to COVID-19 because of resource limitations (n = 26), secondly unrelated reasons (n = 7), patient preference due to fear of infection (n = 2) and chemotherapy delay due to COVID-19 (n = 2).

Within the skull base service, the number of referrals was significantly reduced from n = 329 to n = 101 during COVID-19 (p < 0.001). Notably, the overall number of patients referred for a pituitary adenoma reduced from n = 80 to n = 31 (p < 0.001). Out of those, n = 11 were referred with pituitary apoplexy pre-COVID-19 and n = 3 during COVID-19. There was no statistically significant difference in treatment recommendation between patients referred pre-COVID-19 and during COVID-19 for patients with vestibular schwannoma and pituitary adenoma/apoplexy. However, in a higher proportion of patients referred with a meningioma during COVID-19, active treatments such as surgery or SRS, instead of monitoring were recommended, possibly indicating that larger or more clinically symptomatic lesions were being referred during the COVID-19 period (p < 0.05; Table 5). All meningioma cases, where specialist intervention was recommended, were located in the medial sphenoid wing. Surgery was recommended to 3 patients with pituitary adenomas during COVID-19: 1 had pituitary apoplexy, 1 had progressively deteriorating vision, and in 1 patient, the pituitary mass had progressed over a short period of time and turned out to be a metastasis. Surgical intervention was deferred in n = 5 for sphenoid wing meningiomas, and n = 15 for pituitary adenoma.

The total number of referrals to the spinal MDT decreased significantly during COVID-19 from n = 526 to n = 248 (p < 0.001; Table 4). The proportion of patients referred with cauda equina syndrome increased significantly from 4.0% (n = 21) to 10.5% (n = 26, p < 0.001). There was no statistically significant difference between spinal MDT treatment recommendations before and during COVID-19 (Table 4).

The number of functional neurosurgery MDTs was reduced from twice-weekly pre-COVID-19 to 3 during COVID-19. No elective functional neurosurgery took place during COVID-19 although battery replacement for movement disorder patients continued (Table 2).

In the paediatric service, the total number of operations performed was not as severely affected as the adult service (pre-COVID-19 n = 45, during COVID-19 n = 33) but the case load amongst the subspecialties changed (Table 3). In particular, trauma cases increased from n = 2 (4.4% of total pre-COVID-19) to n = 5 (15.2% of total during COVID-19). Oncology operations also increased from n = 7 (15.6% of total) to n = 9 (27.3% of total) during COVID-19. No functional or neurovascular operations were performed during COVID-19 within our paediatric cohort.

Overall, 30-day perioperative mortality remained low during COVID-19 (n = 4, 2.0%) compared to pre-COVID-19 (n = 5, 1.1%; Table 1). Within emergency operations, 30-day perioperative mortality was lower during COVID-19 (1.9%, n = 3) compared to pre-COVID-19 (2.5%, n = 5), partly reflecting the process of patient selection with higher threshold for transfer and surgery in critically ill patients during the COVID-19 period. The single elective mortality in the COVID-19 period related to a 28-year-old patient with a solitary CM who subsequently passed away due to leptomeningeal disease (Table 1; 30-day perioperative elective mortality during COVID-19 n = 1, 2.2%).

There were 17 neurosurgical patients who were diagnosed with COVID-19 either pre-operatively (n = 4) or post-operatively (operation pre-COVID-19 n = 7, operation during COVID-19 n = 6; all patients tested negative before surgery; Table 1), representing 2.6% of total neurosurgical operations. Out of these 17 patients, 6 (35.3%) were from a black and minority ethnic (BAME) background (Table 6) and one of these BAME patients died of post-operative COVID-19 infection (accounting for 20.0% of all deaths after emergency operation). This was an 86-year-old Asian man with hypertension who underwent burr hole drainage of a chronic subdural hematoma but developed COVID-19 infection 15 days post-operatively and died 6 days later of COVID-19-related pneumonia. This was the only single mortality of a neurosurgical patient with COVID-19 infection within our cohort. There was no difference in the ethnic mix of our patients between the pre-COVID and COVID periods. The median age amongst these 17 patients was 63 ± 15.44 years and male:female ratio was 10:7. Overall, n = 4 patients (23.5%) were admitted to ITU because of COVID-19-related complications. The majority of patients who were infected with COVID-19 had underlying co-morbidities such as hypertension and diabetes mellitus, and all patients admitted to ITU had underlying health problems. Out of the 13 patients who developed post-operative COVID-19 infection, 53.8% (n = 7) had suffered from a post-operative complication (n = 6 wound infection, n = 1 hematoma, n = 1 CSF leak) with a median time to post-operative infection of 18 ± 9.5 days. The median length of stay for the 17 patients diagnosed with COVID-19 was 36 ± 23.97 days; 4 (23.5%) were discharged to a rehabilitation unit, and 11 (64.7%) were discharged home.

Overall, we saw a reduction in acute referrals during COVID-19 by approximately 33.6% and in the number of operations performed by approximately 55.6%. This is comparable to the published literature where a reduction of more than 50% has been described by 226 respondents from more than 60 countries [9]. Mathiesen et al. demonstrated in a European snapshot that in 80% of respondents (20 neurosurgical departments), neurosurgical beds and neuro-intensive care beds were rationalized by postponing elective surgery, fewer acute traumatic brain injuries and subarachnoid haemorrhages admissions, and changing surgical indications in order to ration resources [14]. Although we did not see a statistically significant difference in patients with neurological deficit, GCS ≤ 8, or age > 65 years being transferred to our neurosurgical unit during COVID-19, there was a trend towards admitting fewer elderly patients with depressed GCS. This may represent the tendency to protect the ventilated ITU bed capacity by limiting the admission of patients with extremely poor prognosis. Additionally, referrals to our neurosurgical subspecialist MDTs were decreased although subspecialties with a more elective case mix (skull base, spine, functional) were worse affected than those with a more urgent case mix (neuro-oncology, neurovascular, paediatrics).

Hecht et al. described a similar percentage reduction for neurosurgical emergency admissions although across all subspecialties (p = 0.0007) during COVID-19 [6]. In keeping with their findings, we observed a reduction in the total numbers of patients presenting with sub-/epidural hematomas, traumatic vertebral fractures, and hydrocephalus (as described in Table 2). Interestingly, operations for cauda equina syndrome and spinal myelopathy decreased during COVID-19, which could be either due to a delay in presentation to the emergency department (patient-related factors) or due to a delay in referral through the general practitioner as many had their practice disrupted during COVID-19 (physician-related factors). In contrast though, we observed a similar number of patients with neurovascular emergencies (i.e. ruptured intracranial aneurysm) presenting during and pre-COVID-19 with a higher proportion undergoing endovascular treatment during COVID-19; the treatment decision was not based on resource allocation but was merely a result of the type of aneurysms presenting during the COVID-19 period. However, this also helped to reduce theatre usage and length of hospital stay.

Although there was a reduction in the referrals to our subspecialist MDTs, we developed strategies to avoid critical delays. For example, in neuro-oncology, we actively reached out to our referring centres to encourage continued referral of patients. Importantly, there was no significant change in treatment recommendation with regard to gliomas and CMs in our cohort (p > 0.05; Table 5). In terms of delivery of surgery, we mitigated the effects of reduced theatre space during COVID-19 in our unit by securing additional theatre capacity in the private healthcare sector, contracted through the NHS. This also meant that the operations performed during COVID-19 in the private sector (n = 26) were provided by a “clean” neurosurgical team to a “clean” (COVID-19 negative) cohort of neuro-oncology patients, further reducing the risks. Similar arrangements were used to maintain the delivery of the degenerative spine disease for medically refractory neural compression, with no significant difference in waiting times between pre-COVID-19 and COVID-19 periods, matching the capacity to referrals (p > 0.05, median waiting times 28 and 27 weeks respectively).

There has been little focus on the impact of COVID-19 on neurosurgical training [1]. During the COVID-19, and as part of restructuring the services to release capacity to deal with COVID-19, most of our neurosurgical trainees were redeployed to ITU or COVID-19 wards. This combined with the fact that fewer operations were performed, meant a reduction in training opportunities. There was a move to teach through Zoom (Communications Technology Company, San Jose, CA) conference calls within our unit as elsewhere [19] to keep theoretical knowledge up-to-date; however, this cannot replace actual operative experience. This is further confounded as certain subspecialties have had to modify their operative techniques based on recommendations from various neurosurgical societies, particularly to avoid approaches through the respiratory tract (e.g. trans-sphenoidal surgery) or to limit the use of aerosol-generating instruments (including drills, ultrasonic aspirator) [3, 7].

As part of an effort to reduce physical contacts, in our unit, face-to-face outpatient clinic appointments were almost exclusively changed to telephone. The majority of telephone clinics were for patients being under regular follow-up with stable imaging findings and clinical course. All postponed or cancelled elective patients were equally kept under close telephone follow-up and prioritized for re-scheduling according to disease/symptom severity. Follow-up outcomes included further telephone consultations, repeat imaging, face-to-face assessment, or rescheduled surgery. None of these patients required emergency admission. Only patients who required neurosurgical intervention were seen in face-to-face clinic in order to be pre-assessed for surgery. Overall, this system worked well and may have potential implications on outpatient management for the future [4, 10]. In our experience, remote access platforms such as Attend Anywhere® (Melbourne, Australia) or secure patient online chat-rooms such as The Brain Tumour Charity’s BRIAN [23] provide invaluable tools in keeping contact with patients.

The GlobalSurg reported a 30-day perioperative mortality of 23.8% amongst COVID-19 patients undergoing emergency or elective surgery [2]. In contrast, within our cohort, 30-day perioperative mortality remained low during COVID-19 (2.0%). In fact, none of our four neurosurgical patients that underwent surgery whilst infected with COVID-19 died. We noted that 13 patients (2.0%) who underwent neurosurgery before and during COVID-19 were infected with COVID-19 whilst being an inpatient; however, no single factor could be identified to trace the cause of these post-operative inpatient infections. Importantly, the majority of these patients had underlying co-morbidities or suffered a post-operative complication, hence making them more susceptible to COVID-19 infection [5]. There is new evidence suggesting that people from a BAME background are more severely affected by COVID-19 [8, 11, 17, 18]. 35.3% of our patients with COVID-19 were from a BAME background, of which one died. More data and larger cohorts are needed to further study this aspect.

Patient numbers were limited due to the relatively short time period observed. This was however inevitable and a reflection of the dynamic nature of the pandemic since the last day of recruitment was dictated by a national change in strategy towards the next phase to restore services. Our retrospective data collection for the pre-COVID-19 phase may also be another limitation here. We further did not analyse the impact of COVID-19 on excess/indirect neurosurgical mortality due to lack of presentation to hospitals such as the case maybe, for example, for ruptured aneurysms.