Article Text

Download PDFPDF

Original research
Real-world effects of late window neurothrombectomy: procedure rates increase without night-time bias
  1. Michelle Marie Williams1,
  2. Thabele Leslie-Mazwi2,
  3. Joshua A Hirsch2,
  4. Carol Kittel1,
  5. Alejandro Spiotta3,
  6. Reade De Leacy4,
  7. J Mocco4,
  8. Felipe C Albuquerque5,
  9. Andrew F Ducruet5,
  10. Nitin Goyal6,
  11. Adam S Arthur7,
  12. Peter Kan8,
  13. Maxim Mokin9,
  14. Travis M Dumont10,
  15. Alan Reeves11,
  16. Stacey Q Wolfe1,
  17. Kyle Fargen1
  1. 1 Neurological Surgery, Wake Forest Baptist Medical Center, Winston-Salem, North Carolina, USA
  2. 2 Interventional Neuroradiology, Massachusetts General Hospital, Boston, Massachusetts, USA
  3. 3 Neurosurgery, Medical University of South Carolina, Charleston, South Carolina, USA
  4. 4 Neurosurgery, Mount Sinai Hospital, New York City, New York, USA
  5. 5 Neurosurgery, Barrow Neurological Institute, Phoenix, Arizona, USA
  6. 6 Semmes-Murphey Neurologic and Spine Institute, Memphis, Tennessee, USA
  7. 7 UT Dept Neurosurgery/Semmes-Murphey Clinic, Memphis, Tennessee, USA
  8. 8 Neurosurgery, Baylor College of Medicine, Houston, Texas, USA
  9. 9 Neurosurgery, University of South Florida, Tampa, Florida, USA
  10. 10 Neurosurgery, University of Arizona/Arizona Health Science Center, Tucson, Arizona, USA
  11. 11 Neuroendovascular Division, Department of Radiology, University of Kansas Medical Center, Kansas City, Kansas, USA
  1. Correspondence to Dr Michelle Marie Williams, Neurological Surgery, Wake Forest Baptist Medical Center, Winston-Salem, NC 27103, USA; williamsm0912{at}


Introduction With the expansion of the interventional time window for stroke from emergent large vessel occlusion (ELVO), the rate of mechanical thrombectomy (MT) is expected to rise, potentially causing higher burnout rates and requiring hospitals to develop strategies for adequate coverage of these procedures.

Methods Neurointerventional physicians at 10 participating stroke centers prospectively recorded time requirements for all MT consultations over 30 consecutive 24-hour call periods, including both false positive consultations and MT procedures, during mid to late 2018. Consult start time, procedure start and end time, and data regarding commute to the hospital and delay in scheduled procedures were collected and compared with those from an identical prospective study performed in 2017.

Results Data were collected from a total of 300 days of call. A total of 166 procedures were performed (mean 0.55 per day), an increase from 0.32 per day in 2017. Overall mean MT direct time burden during each 24-hour call was 124 min (compared with 85 min in 2017). The percentage of consultations for thrombectomy varied based on time of day, with 87% of consults between the hours of midnight and 04:00 proceeding to thrombectomy compared with 37% between the hours of 16:00 and 20:00.

Conclusions MT procedural volumes have increased from one every 5 days in 2016 to one every 2 days in 2018. The highest percentage of consults leading to thrombectomy occur in the early morning hours after midnight. Compared with similar data from 2016 and 2017, call demands continue to escalate, representing a significant demand on neurointerventional teams.

  • stroke
  • thrombectomy
  • angiography
  • intervention
View Full Text

Statistics from


Mechanical thrombectomy (MT) for patients with stroke from emergent large vessel occlusion (ELVO)1 is among the most efficacious and validated procedures in medicine.2–11 The number needed to treat to reduce disability by at least one level on the modified Rankin scale is 2.6 for patients presenting within 6 hours of onset of symptoms.12 Recent results of the DAWN and DEFUSE 3 trials have confirmed similar benefit to MT in selected candidates presenting up to 24 hours after symptom onset.13 14 As a result, increased numbers of consultations and procedures for MT are anticipated. This increase in volume is expected to have significant impacts on hospitals and providers, including worsened physician or staff burnout. Knowledge of these trends is therefore imperative in the identification, design and implementation of protective strategies, which may include increasing provider complements, 'shiftwork' call schedules, or developing paradigms for fair financial compensation.

A previous prospective study, performed during 2017 prior to the publication of trials demonstrating the efficacy of late window thrombectomy,13 14 was carried out at 10 geographically diverse stroke centers in the USA with data collected from a total of 270 days of call. This study demonstrated a significant burden of MT call on neurointerventional physicians. On average, physicians spent an average of 85 min addressing MT call over each 24-hour call period, with MT procedures occurring approximately every 3 days.15 Importantly, MT procedural frequency had increased substantially from a retrospective study performed 1 year before involving eight of the same stroke centers, where MT cases were performed once every 5 days on average.16

Following the publication of late window thrombectomy trials, we hypothesized that stroke consultations and thrombectomy procedures have increased. Additionally, we sought to evaluate whether a night-time bias exists and hypothesized that the ratio of thrombectomy to overall consultations would be lowest during early morning hours. To do so, we sought to re-evaluate the MT call burden at the same 10 centers to understand the influence of these recent trials on MT consultation and procedural incidence.


The methods for data collection were identical to the previous prospective study performed in 2017 and have been described in detail elsewhere.15 16 Institutional review board (IRB) approval was obtained at the 10 participating stroke centers (table 1). Neurointerventional physicians at the centers prospectively recorded time requirements for all thrombectomy consultations over 30 consecutive 24-hour call periods, including both false positive consultations (consults that ultimately did not proceed to MT) and MT procedures. All consults were included based on the on-call physician’s interpretation of the reason for contact. This resulted in data from 300 total days of call (30 consecutive days at 10 institutions) between the months of May and November 2018 based on differing approval times of each unique IRB. Each on-call neurointerventional physician was responsible for recording all data prospectively in real time during their call periods. These data were then de-identified and transmitted to the primary center for analysis. The following data were collected: (1) start time for each consultation, defined as time of initial contact of the neurointerventional physician (in some cases physicians were contacted ahead of patient arrival to the facility, and thus this metric incorporates time when the physician was waiting on stand-by for the patient to be transported for thrombectomy or to undergo diagnostic imaging; (2) groin puncture time (for thrombectomy procedures only); (3) groin closure time (for thrombectomy procedures only); (4) end time for the consultation (for false positive consultations, end time was defined as the time when the decision was made not to intervene; for MT procedures occurring during work hours, end time was defined as the time when all orders and all family and service provider discussions had been completed; for those procedures occurring outside of work hours, end time was defined as the time when all orders and discussions had completed and the physician arrived back at home); (5) whether the procedure resulted in a delay of ≥30 min to scheduled procedures or clinic visits; (6) whether an alternate physician was required to cover the procedure; and (7) whether the physician had to commute from home to perform the procedure.

Table 1

Data collected from multiple geographically diverse stroke centers between May and November 2018

'Peak work hours' were defined as 07:00 to 16:59 Monday through Friday, while 'non-peak work hours' were defined as 17:00 to 6:59 Monday through Friday and the entirety of Saturday and Sunday. Additionally, the number of thrombectomies and false positive consults were evaluated based on time of day, with each 24-hour period being divided into six 4-hour intervals, and the percentage of total consultations that proceeded to thrombectomy were calculated.

Statistical analysis

All analyses were conducted using R: A Language and Environment for Statistical Computing, Version 3.5.1 (R Foundation for Statistical Computing, Vienna, Austria) and RStudio: Integrated Development for R, Version 1.1.456 (RStudio, Boston, Massachusetts, USA). Descriptive statistics were calculated such that mean (SD) were used for normally distributed variables and median (range) for non-parametric data. For all analyses, two-tailed hypothesis testing was used with p<0.05 interpreted for statistical significance.

A post hoc analysis was carried out using pooled thrombectomy data from the previously published prospective and retrospective data.15 16 Friedman’s test was conducted to test whether there was a significant difference in distribution of MT rate across three studies (2018, 2017, 2016). The Conover test with Bonferroni correction for multiple comparisons was used for post hoc pairwise testing of the thrombectomy rate between studies. A six-sample test for equality of proportions with Bonferroni correction for multiple pairwise comparisons was made to test differences in the proportion of thrombectomies across time of day.


A total of 294 MT consultations (mean 0.98 per day), including 128 false positive consultations (mean 0.43 per day) and 166 MT procedures (mean 0.55 per day), were prospectively recorded at the 10 centers.

Time requirement

One hundred and sixteen (39%) recorded 24-hour call periods had no MT consultations and therefore zero minutes of MT call burden. The median total time for each thrombectomy consultation, including both false positives and those progressing to intervention, was 117 min (IQR 45, 184). Median time for false positive consultations was 38 min (IQR 15, 70). Median time for MT consultations from start to end was 158 min (IQR 120, 221), with median procedural time (groin puncture to groin closure) being 50 min (IQR 31, 69). The median overall time burden per 24-hour call period was 61 min (IQR 0, 216), with a mean of 124 min per day.

Procedure frequency

MT procedures most frequently occurred between 15:00 and 16:00 hours. When compared with a previous prospective study carried out in 2017,15 the total number of procedures performed increased from 0.32 to 0.55 per day, which represents a continued increase from a rate of 0.21 per day in 2016.16 Additionally, the ratio of MT procedures to false positive consultations was 1.3, compared with a false positive ratio of 0.65 in the previous prospective study.

Procedure rate

Table 2 reports the percentage of total consultations that were taken for thrombectomy based on time of day. Each 24-hour call period was divided into 4-hour segments. There was a statistically significant difference between the percentage of consults taken for thrombectomy between midnight and 04:00 hours (86.7%) and the percentage taken between 16:00 and 20:00 hours (36.5%; p=0.030). The greatest total number of consults occurred between the hours of noon and 16:00, and the least between the hours of midnight and 04:00.

Table 2

Percentage of MT consults proceeding to thrombectomy based on time of consultation for 2018 data alone and pooled 2017 and 2018 data

Out of a total of 294 MT consultations, 130 (44.2%) occurred during the peak hours of 07:00 and 17:00, Monday through Friday and 164 (55.8%) occurred during non-peak hours, either on the weekends or Monday through Friday between the hours of 17:00 and 07:00. For procedural consults (eliminating false positive consults), 80 (48.2%) occurred during peak hours and 86 (51.8%) occurred during non-peak hours.

Combined 2017 and 2018 data are shown in table 2. When data were pooled in order to evaluate a larger sample of times that consultations occur, there remains a significant difference between the percentage of consultations taken for thrombectomy based on time of day (p=0.007). Pairwise post hoc analysis with Bonferroni correction for multiple comparisons demonstrates a significant difference between 00:00–03:59 hours and 16:00–19:59 hours (p=0.046).

Delays and commute

A delay of 30 min or more in elective schedules or procedures was caused by 38 of the 166 thrombectomy procedures (22.9%). For consultations requiring thrombectomy, an alternative surgeon was required to cover 13 cases (7.8%). The physician was required to commute from outside of the hospital for 79 cases (47.6% of all consults that resulted in thrombectomy).

Pooled neurothrombectomy procedural times

Time from groin puncture to groin closure for 179 thrombectomy procedures from the previously published retrospective study16 were pooled with similar data from the 84 procedures previously published in the prospective study15 and the 166 procedures from the current study (429 total procedures), as shown in figure 1. The total number of MT consults as well as the rate per day in each study are shown in (table 3). Data were pooled in order to investigate time of day that consults and procedures most frequently occur, which would not be influenced by temporal trends. The distribution of procedural times for the 429 procedures is shown in figure 1.

Figure 1

Pooled procedural times (puncture to closure) for 429 thrombectomy procedures from the previous retrospective study (179 procedures),16 previous prospective study (84 procedures),15 and the current study (166 procedures).

Table 3

Thrombectomy incidence from retrospective study (2016), prospective study (2017), and current study (2018) for all 10 centers that participated in the studies


This prospective study of MT call among neurointerventional physicians at 10 stroke centers after the publication of late window thrombectomy trials demonstrates increasing MT procedural incidence and MT time burden. This study now shows an MT incidence slightly more frequent than every other day (average 0.55 per day), with an average daily time burden addressing MT consultations of over 2 hours (124 min). Procedural incidence has increased from just 1 year prior from once every 3 days. At that time, average time burden was 85 min using identical methodology recorded at the same stroke centers.15 In addition, the present study demonstrates that procedures remain most common between the hours of noon and 18:00, with nearly 50% of procedures requiring physicians to commute to the hospital from home. Interestingly, there is no evidence of a night-time effect on decision-making regarding candidacy for thrombectomy. Although consultations were least frequent during the early morning hours, the highest percentage of consultations proceeding to thrombectomy occurred during the period from midnight to 04:00 hours (87%), which would be expected to be the time period when this bias would be most significant. These results are consistent with a recent physician survey of hypothetical cases that also showed no bias in decision-making during night-time hours.17

The incidence of MT is on the rise. A retrospective study was performed in 2016 to assess the frequency of thrombectomy consults as well as the resultant time burden to neurointerventionalists across 10 centers.16 At that point, thrombectomies were occurring at a rate of approximately one every 5 days. A follow-up prospective study was carried out from May to September 2017, which showed that this rate had already increased to approximately one every 3 days.15 The present study demonstrates that procedural incidence has further escalated to every other day at the same centers. It is important to note that the 10 centers included in the sample were chosen initially because of their varying stroke center certifications, stroke volumes, and geographical locations. However, despite that intentional variability, these centers do represent a uniformity of large academic, tertiary care centers. The increase in thrombectomy volumes captured at these centers may be less generalizable to private hospitals or smaller centers, but the magnitude of the MT increase in the 10 center sample suggests that the present study likely reflects the larger trend of increased thrombectomy across the country.

Patient selection for ELVO has evolved following the publication of the DEFUSE 3 and DAWN trials. Prior to these landmark studies, evidence strongly supported thrombectomy for selected patients presenting within 6–8 hours of the time since last known normal. DEFUSE 3 investigated thrombectomy within 6–16 hours,14 while the DAWN trial expanded the time window even further, evaluating thrombectomy up to 24 hours since last known well.13 Despite small differences in the inclusion criteria in the two trials, MT and medical management was superior to medical management alone for patients presenting with proximal arterial occlusions. There is now evidence supporting MT in patients presenting with wake-up, unknown time of onset, or delayed onset strokes and favorable imaging profiles. It is likely that the increased incidence of MT is at least partially due to a real-world effect of the expanded time criteria from these trials.

Another possible contributing factor to the increasing rate of thrombectomy is more aggressive patient selection for MT, with neurointerventionalists now more willing to intervene on patients with less severe symptoms, more distal occlusions, or larger completed infarcts on presentation. A previous survey on neurointerventionalists showed increasing aggressiveness in selecting patients for MT after publication of the 2015 trials.18 Since that time, studies have demonstrated that MT is safe with a high rate of good outcomes in patients presenting with National Institute of Health Stroke Scale score of <8.19–22 A meta-analysis of seven randomized controlled MT trials showed significantly better functional outcomes with MT compared with medical management for patients presenting with more severe infarct burden as measured by pre-procedure ASPECTS scores as low as 3–6.23 Therefore, more aggressive interventional management of patients with previously marginal indications is another possible contributor to the witnessed increase in MT incidence in this study. A study performed from 2015 to 2016 found false-positive thrombectomy transfers accounted for the majority (54%) of consultations.24 Whether the lower rate of false-positive consultations in the present study (44%) represents an increase in aggressiveness to pursue thrombectomy in the current climate is unclear.

The present study interestingly identifies an absence of a night-time effect on selecting patients for MT. There is limited evidence suggesting a weekend effect, with worse clinical outcomes following thrombectomy in patients undergoing procedures on weekends compared with weekdays.25 However, there have been no studies evaluating the effect of time of day on selection of patients for MT. Should a bias exist, it would be expected that consultations in the early morning hours, when neurointerventionists are at home presumably sleeping, would be most affected by this bias, resulting in fewer consultations proceeding to thrombectomy. Importantly, this prospective study, in concert with data from the previous prospective study in 2017, clearly demonstrates that a night-time bias does not exist. In fact, consultations occurring between the midnight to 04:00 time period were the most likely to proceed to thrombectomy (>70% in the pooled analysis), indicating, if anything, a bias towards intervention during early morning hours.

At most stroke centers, MT call is continuously covered by a relatively small number of physicians with elective clinical and procedural schedules outside of this specific call. The present study expands on prior published research and suggests that the physician time commitment for MT call, both in terms of total quantity but also the non-work hour responsibilities, has become progressively more burdensome. A recent study evaluating the rate and risk factors of burnout in neurointerventional physicians demonstrated a deleterious effect of covering more than one hospital while on call.26 Interestingly, while frequency of MT call was not a risk factor for burnout in that study, compensation for call was protective. As MT volumes continue to increase, understanding the effects of this increase on call teams, hospital workflow, physician practices, and burnout risk is imperative. This represents an important area of future study and requires expansion of focus to the entire care team, not only the physician component.

There are limitations to this study. We sought to capture the total time that physicians are engaged in MT patient care, but this is likely an underestimation given the additional duties of MT consultation and procedures which are less easily accounted for. These include time recovering from cases, sleep disruption, and lifestyle adjustments (living in proximity to the hospital, for example). The call burden for operating room staff, fellows, technologists, and others directly involved in MT procedures is not captured by these data. Data are sampled during a single month for each center, and therefore variability month to month may affect individual center numbers. However, with 10 moderate volume participating centers, this variability should be mitigated. Additionally, while the participating centers are mostly comprehensive stroke centers that are geographically diverse, there are other models of neurointerventional care that might not be captured in this sample. Finally, demographic data on the patients undergoing thrombectomy were not collected, as the intention of the study was to evaluate the rate of MT procedures and time requirements to physicians while not over-burdening on-call physicians who were required to collect real-time data for this study in a prospective manner.


MT procedural and consultation volumes continue to increase, from approximately one procedure every 5 days in 2016, to one every 3 days in 2017, to now roughly one every 2 days. The average MT call direct time burden has increased to over 2 hours per 24-hour call shift. Consultations also now more frequently result in thrombectomy, with no observed night-time bias in patient selection for intervention. The present study expands on prior published research and suggests that the physician time commitment for neurointerventional call has escalated compared with similar studies from 2016 and 2017. With this confirmation and quantification of increased MT volumes, understanding the deleterious effects on call teams, hospital workflow, physician’s practices, and burnout risk is imperative.


View Abstract


  • Twitter @JoshuaAHirsch, @rdeleacymd, @AdamArthurMD

  • Correction notice Since this paper was published online first, middle initials have been added to Dr Ducruet and Dr Albuquerque names.

  • Contributors Conception and design: KF. Data collection and interpretation: all authors. Statistical analysis: CK. Drafting the article: MMW, KF. Critical revision of article: all authors. Final approval of article: all authors.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.