The sitting position presents several unique challenges to the anesthesiologist, particularly for hemodynamic management. Physiologic changes that accompany the head-up position during general anesthesia include the vasodilating and negative inotropic effects of anesthetic drugs, as well as the redistribution of venous blood volume into dependent extremities.1–4 Although the upright position is the normal position for humans during the majority of waking life, as well as the normal reference position for obtaining outpatient arterial blood pressure (BP) measurements, the addition of general anesthesia makes the maintenance of normal BP without vasopressor or intravascular volume support difficult in this position.5,6
Ischemic brain or cervical spinal cord injury are among the most feared complications of the sitting position.7–14 Four cases of intraoperative stroke associated with shoulder surgery in the sitting position were highly publicized and initiated a broad discussion of appropriate hemodynamic management. An additional single case has been reported since then, as well as a self-reported orthopedic surgery database of neurologic injuries during surgery in the beach chair position.15,16 With very few hemodynamic details offered from the 4 original cases, however, it is difficult to either glean lessons or draw future management conclusions from their report. For example, 1 of the 4 patients had a BP cuff positioned on her dependent calf, no adjustment was made to a normal heart-level reference, and even with that departure from normal practice, systolic pressures were 70 to 80 mm Hg for some portion of the case.
It is likely that the anesthesia community would benefit from the addition of a reference group of patients undergoing surgery and general anesthesia in the sitting position without catastrophic outcomes, and, in particular, from a careful examination of the hemodynamic management of those patients. In the absence of such data, we are left with case reports, anecdotes, and expert opinion alone to guide us.
A report of >4000 patients who underwent orthopedic shoulder surgery in the sitting position managed with modest deliberate hypotension under regional anesthesia, and without gross neurological events, has helped to provide a reference cohort.17 Our study adds an even greater number of patients to that database, and in a slightly different clinical context: neurosurgical as well as orthopedic patients, all managed with a normotensive, rather than hypotensive strategy, and all managed under general anesthesia. Together with that study, we provide a reference group of almost 10,000 patients, complete with detailed hemodynamic data, to weigh alongside the small number of bad outcomes with minimal hemodynamic data.
The first aim of this study was to produce a descriptive statistic for intraoperative BP changes from baseline in a large group of patients who underwent surgery in the sitting position without gross neurological events in the immediate postoperative period. The second specific aim was to compare the intraoperative BP changes from preoperative baseline in neurosurgical patients undergoing surgery in the sitting position, comparing BPs that were referenced to heart level versus head level. As a retrospective but large-scale study, the goal was to provide insight, not into the intended or theoretical goals of hemodynamic management in sitting patients, but rather into the hemodynamics actually achieved in a high volume and carefully recorded practice.
The study protocol was reviewed and approved by the IRB. The requirement for written informed consent was waived by the IRB. Patients who underwent either orthopedic shoulder surgery or neurological surgery in the sitting position at Mayo Clinic Rochester between January 1, 2002 and December 31, 2009 were included. All patients underwent surgery with general anesthesia and controlled ventilation, and all BPs were monitored intraoperatively with an arterial line pressure transducer positioned at either head or heart level, or with a BP cuff positioned at the upper arm (approximately heart level). We excluded pregnant patients and patients <18 years of age.
Demographic data from each patient were retrieved. The medical history was also reviewed, focusing on the presence or absence of hypertension (HTN) (identified by either a formal preoperative diagnosis of HTN or the use of antihypertensive medications).
Preoperative hemodynamic variables were obtained from the most recent preoperative evaluation. For the minority of patients whose BPs were not recorded in a preoperative evaluation clinic, the first recorded intraoperative BP, measured in the supine position, before anesthetic induction, was used as a baseline value. The same physiological variables were retrieved from the intraoperative period, at 5-minute intervals for noninvasive measurements and at 2-minute intervals for arterial line measurements. All intraoperative hemodynamic data were retrieved from the ChartPlus™ electronic database, which serves as our intraoperative electronic anesthetic record. At our institution, editing or changing hemodynamic data after the fact is not allowed and would be discoverable on a permanent audit trail that cannot be deleted. The electronic system that is used in the operating room (OR) is “Anesthesia Manager” and is a vended product (not developed by Mayo) from Picis (http://www.picis.com/). This system does allow for both manual charting/editing of the data as well as deleting the data if they are deemed incorrect. Anesthesia Manager is a recording system, and it has been designed to connect to a monitoring device (i.e., cardiac monitor) and collect the data automatically as long as there is the connection, the monitor is gathering physiologic data, and the software is running in the computer. Our anesthesia providers never proceed with an OR case unless the data collection system is operating and are not allowed, by policy, to “postedit” data without a justification given in the electronic medical record itself. In the authors’ experience in >10 years with this system, such an event is very rare, is discouraged, and is discoverable after the fact through an electronic audit trail. For patients who underwent neurosurgical procedures, all of whom had arterial lines, the placement of the arterial pressure transducer (head or heart level) was recorded.
Definitions of Outcome of Interest
Referenced to previous studies, we defined “clinically significant hypotension” as either a reduction of systolic BP by >20% below baseline10,18 or a reduction of mean arterial BP (MAP) at the reference level, either head or heart level, chosen by the clinician by >10 mm Hg below baseline.19 In the Pohl and Cullen7 report of 4 patients who developed cerebral ischemia after shoulder surgery in the sitting position, all had intraoperative BP reductions between 28% and 42% at the level of the BP cuff.7 Thus, we chose to use an intraoperative BP reduction of ≥40% below baseline as the definition of a “notable hypotensive episode.” Any patient who had at least 1 intraoperative BP measurement that met or exceeded the reduction cut point was counted as an incident case. Each episode was quantified in minutes to determine the overall duration of intraoperative notable hypotension.
Statistical Power and Sample Size Calculation
We identified >5000 patients from the ChartPlus tool by using keywords of surgical position and surgical procedure. This number of patients provided a statistical power of >99% to detect a difference between preoperative and intraoperative BP of 0.1 SD units. For the second aim, we anticipated a sample size of approximately 500 patients in each group with respect to pressure transducer placement. To be considered clinically meaningful, we considered that the mean difference would need to be at least 5 mm Hg. Under the assumption that the standard deviation of BP change from baseline is at most 20 mm Hg, this sample size would provide statistical power of >99% to detect a difference of 4 mm Hg between those with transducer placed at the head versus heart level.
The purpose of the study was to assess intraoperative BP changes from baseline for patients who undergo surgery in the sitting position. In all cases, separate analyses were performed for neurosurgical patients and orthopedic patients. Because patients were not in the sitting position for anesthetic induction, tracheal intubation, and initiation of monitoring, we excluded data collected for the first 45minutes after patients entered the OR. Recordings obtained after this time were assumed to occur while the patient was positioned in the sitting position. The intraoperative BP was summarized for each 2-hour interval (starting from minute 46 to minute 165) through the entire operation, limited to a maximum 10 hours. We sampled and assessed the BP change during the first 2-hour interval for 100 patients. For each patient, the values obtained during this period were reviewed graphically, and no substantial departures from normality or time trends were identified. Thus, we decided to use the MAP over the first 2-hour interval, expressed as a change from baseline as the analysis variable for primary comparisons. Analyses were performed with change from baseline expressed as an absolute change (mm Hg) and also as a percentage change from baseline.
All data are presented as mean ± SD unless otherwise specified. The MAP over the first 2-hour interval of intraoperative measurements was compared with preoperative baseline by using the paired t test. Linear regression analysis was used to assess whether the methods of BP measurement or the presence of preexisting HTN were associated with intraoperative hemodynamic changes from baseline. Episodes of notable intraoperative hypotension were identified using all data starting from minute 46 through the entire operation, limited to a maximum 10 hours. The percentage of patients who experienced ≥1 episodes of notable hypotension (either systolic BP or MAP reduction from baseline ≥40%) was compared between patients with and without preexisting HTN using the χ2 test. Among patients who experienced notable hypotension, the total duration of hypotension was compared between groups using the 2-sample t test. In all cases, 2-tailed P < 0.05 were considered statistically significant.
Ninety-five percent upper confidence limits for the incidence of ischemic brain or spinal cord injury were calculated based on equation 10.7 of Feller, which relates binomial probabilities to tail area of the β distribution (W. F. Feller . An Introduction to Probability Theory and Its Applications, Volume 1, Chapter 6, Wiley).
Patient Recruitment and Demographics
During the time period from January 1, 2002 to December 31, 2009, we identified 5601 cervical laminectomy, suboccipital craniotomy, or arthroscopic shoulder surgeries performed in the sitting position. Of these, 43 were excluded because the patient was <18 years of age, 302 were excluded because the duration of surgery was <45 minutes, and 79 were excluded because of incomplete data or duplicate observations. Thus, 5177 (950 neurosurgery; 4227 orthopedic surgery) patients were included. There were no patients who experienced ischemic brain or spinal cord injury (0/5177, exact upper 95% confidence limit = 7 per 10,000; orthopedic surgery 0/4227, exact upper 95% confidence limit = 9 per 10,000; neurosurgery 0/950, exact upper 95% confidence limit = 39 per 10,000). Within each surgical group, demographics and baseline patient characteristics were similar between groups defined by the method used for intraoperative BP monitoring (Table 1).
Intraoperative BP Change from Preoperative Baseline
Among neurosurgical patients, baseline BP measurements were obtained in the Preoperative Evaluation Clinic (POE) for 728 and intraoperatively preinduction for 222. Among orthopedic patients, baseline measurements were obtained in the POE for 3827 and intraoperatively preinduction for 400. For both types of surgery, the MAP measurements for patients with measurements obtained in the POE were not found to differ significantly compared with patients with measurements obtained intraoperatively preinduction (all P > 0.07). MAP during the first 2 hours of surgery was reduced significantly from baseline for both neurosurgical and orthopedic patients regardless of the method of BP measurement (Table 2).
Among neurosurgical patients, mean systolic BP over the first 2 hours of surgery evidenced a decline from baseline of 12.1% ± 13.5% for those monitored with an A-line placed at the heart level and a decline of 12.1% ± 12.4% for those monitored with an A-line at the head level corresponding to a difference between groups of 0.0% (95% confidence interval [CI], −1.7% to +1.6% for head versus heart). Mean MAP over the first 2 hours of surgery evidenced a decline from baseline of 17.6% ± 11.5% for those monitored at heart level and a decline of 19.7% ± 10.7% for those monitored at head level, corresponding to a difference between groups of −2.1% (95% CI, −3.5% to −0.6% for head versus heart) indicating less of a decline if MAP was monitored at heart level. Figure 1 shows the mean ± SD BP in neurosurgical patients at baseline and for each intraoperative 2-hour interval over the first 10 hours according to the level of transducer placement. Similar to what was found in the analysis restricted to the first 2-hour interval, MAP measurements obtained after the first 2 hours tended be slightly higher when monitored at heart level compared with head level.
The Effect of Preexisting HTN on Intraoperative BP Changes from Baseline
Baseline and MAP over the first 2 hours of surgery are presented according to method of monitoring and presence of preexisting HTN in Table 3 (orthopedic patients) and Table4 (neurosurgical patients). In all cases, mean systolic BP and MAP were significantly higher at baseline in patients with preexisting HTN (134 ± 18 mm Hg vs 127 ± 17 mm Hg in orthopedic patients, and 132 ± 17 mm Hg vs 127 ± 15 mm Hg in neurosurgical patients) and those patients were observed to have larger reductions from baseline over the first 2 hours of surgery (−16.6% ± 12% vs −12.1% ± 12.9% for orthopedic patients, and −14% ± 12% vs −10.1% ± 12.4% for neurosurgical patients). Intraoperative mean systolic BP and MAP did not differ significantly between those with and without preexisting HTN.
The Incidence and Duration of Intraoperative Notable Hypotension (an Intraoperative BP Reduction of ≥40% Below Baseline)
Over the entire duration of surgery, 52% (95% CI, 49%–56%) of neurosurgical patients, 51% (95% CI, 47%–55%) of orthopedic patients monitored with an A-line, and 48% (95% CI, 46%–50%) of orthopedic patients monitored with noninvasive BP experienced ≥1 episodes of notable intraoperative hypotension (as defined above). The incidence of notable hypotension, as well as duration of hypotension expressed in both minutes and percent of total surgery, are presented in Tables 5, 6, and 7. Findings are presented both overall and also separately for patients with and without preexisting HTN. In all cases, patients with preexisting HTN tended to have an increased incidence and longer duration of notable intraoperative hypotension.
In the absence of either extreme hypotension or extreme HTN, there is no established “safe” BP for patients in any surgical position while under general anesthesia. The recommendation to maintain intraoperative BP close to the patient’s preoperative baseline would seem to make good sense, but we do not have sufficient empiric data to quantify either subtle or gross neurologic deficits as a function of BP.20–22
The work presented here is intended to contribute a carefully collected and analyzed body of data that can inform the debate surrounding the management of surgical patients in the sitting position. We cannot claim to have generated new guidelines or variables for hemodynamic management, but our data represent the largest contribution to date of a badly needed reference group of carefully managed and recorded cases in the sitting position for which no adverse events occurred. Even a study as large as this cannot establish an empiric “safety margin” guideline. What it can do, however, is to describe the exact physiological variables maintained in >5000 uneventful cases in a conventionally managed and monitored practice which includes: (1) the sitting position; (2) near-normal BPs (within 20% of baseline); and (3) general anesthesia.
In the absence of such data, the anesthesia community is left only with anecdotal cases of dramatic but apparently rare catastrophes that are poorly defined in terms of physiological records. The highlighting of such anecdotes may not only serve to alert our profession to a potential risk, but it also runs the coincident risk of suggesting practice variables that are not evidence-based. Neither can such anecdotal evidence serve to discriminate between competing hypotheses for the most appropriate BP reference levels (heart versus head).
Our results reveal that the average reduction in intraoperative systolic BP ranged from 12% to 19% depending on the type of surgery and method of BP measurement. Although the average reduction was less than the predefined definition of “clinically significant,” a substantial percentage of the individual patients met this criterion. The mean reduction in the intraoperative MAP met criterion for a “clinically meaningful” decline, regardless of the method of BP measurement or the level of transducer placement, in both orthopedic and neurosurgical patients.
We assumed that the maintenance of intraoperative BP for HTN patients was more problematic and challenging compared with normotensive patients because of increased intraoperative BP fluctuation.5,23 Our results support this assumption. HTN patients had a greater magnitude of intraoperative BP reduction compared with normotensive patients, regardless of the method of BP measurement. Although anesthesia providers generally appreciate the importance of preexisting HTN, our study endorses this proposition by providing a quantitative measure of the difference between preexisting HTN and normotensive patients in terms of both the incidence and duration of intraoperative “notable hypotension.”
Assuming that the most reasonable goal of any management strategy is to preserve baseline physiologic variables, we analyzed the separate results of 2 different BP monitoring techniques among neurosurgical patients. We found only minimally different systolic and MAP between heart versus head level arterial line transducer placement (Table 2 and Figure 1). The intraoperative BPs acquired from transducers placed at heart level were statistically higher than those acquired from head level transducer placement, but the absolute difference was small. Our data cannot determine the superiority of one reference level over the other because we have no outcome differences to correlate them to, and because the absolute numerical differences between the 2techniques was so small. Similarly, our study cannot determine, retrospectively, what the exact hemodynamic goal was for each provider for each case; we can only determine the final hemodynamic results.
It is interesting to note, however, that regardless of the real-time intention of the practitioners, the retrospective result was approximately the same: we apparently allow arterial BP to decrease between 15% and 20% below baseline, regardless of whether it is referenced to head or heart level.
With that said, it could be argued that a “hydrostatic correction factor” (affected either by subtracting an estimated heart-to-head hydrostatic gradient from numerical cuff pressures obtained at the arm, or by positioning arterial line transducers at head level, both of which accomplish the same thing) might compensate for our apparent tendency to allow intraoperative numerical BPs to decrease by 15% to 20%. It is important to note that, regardless of whether a clinician adopts a view of the cerebral circulation as a closed (“siphon”) or open (“waterfall”) system,22,24–29 neither the authors of this manuscript, nor the clinicians at our institution, ever advocate or use a BP reference point that is below the level of the heart. It is also important to note that, even after the average reductions in intraoperative BP occurred, the absolute BPs achieved at either heart or head level during these cases ranged between 74 and 78 mm Hg.
We defined “notable hypotension” with reference to the prior reports of cerebral ischemia after shoulder surgery in the sitting position. Intraoperative notable hypotension was found in approximately 50% of our cases. The duration of notable hypotension represented the cumulative time of each episode summed over the entire operation. Without specific human data, one can speculate, but not know for certain, that catastrophic events might be related more to the total duration of each episode of extreme hypotension rather than the cumulative duration of all extreme hypotensive episodes throughout the operation. Similarly, the magnitude of intraoperative BP reduction in a single measurement event might vary even more broadly, from 40% to even 80%, below preoperative baseline. Although these postulates are intriguing, they do not find robust support in human data; nor would we be able to correlate any such parameters to outcome in our dataset.
This retrospective study design has several limitations, including the potential of incompletely recorded data and the potential for unmeasured confounding variables including type and dosage of anesthetic drugs, patients’ preoperative intravascular volume status, preoperative or intraoperative medications that might have hemodynamic effects, such as mannitol, hypertonic saline, diuretics, the type of antihypertensive medications, intraoperative blood loss, and intraoperative intravascular volume replacement. The purpose of our study was to examine, in detail, the actual hemodynamics achieved during these surgeries and anesthetics. We cannot determine which of the many possible influences on BP were at play during each segment of each surgery. An additional limitation is the lack of information on long-term outcome after surgery. We would point out, however, that our study was initiated, in part, to address the concern about acute, catastrophic outcomes in the immediate postoperative period after surgery and anesthesia in the sitting position. Finally, although this is a large case series, the statistical power to detect rare events is limited. Even if prior reports are accurate, the incidence of catastrophic complications is almost certainly very low. With 5177 patients, the upper 95% CI if no events are observed is 7 per 10,000. We cannot conclude, therefore, that maintenance of the same hemodynamic variables observed in this study is “safe” beyond that degree of certainty.
Name: Pathomporn Pin-on, MD.
Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.
Attestation: Pathomporn Pin-on has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: Darrell Schroeder, MS.
Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.
Attestation: Darrell Schroeder has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: James Munis, MD, PhD.
Contribution: This author helped design and conduct the study, analyze the data, and write the manuscript.
Attestation: James Munis has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
This manuscript was handled by: Gregory J. Crosby, MD.
1. Dalrymple DG, MacGowan SW, MacLeod GF. Cardiorespiratory effects of the sitting position in neurosurgery. Br J Anaesth. 1979;51:1079–82
2. Smelt WL, de Lange JJ, Booij LH. Cardiorespiratory effects of the sitting position in neurosurgery. Acta Anaesthesiol Belg. 1988;39:223–31
3. Geevarghese KP, Garretson HD. Positioning of the patient for neurological surgery. Int Anesthesiol Clin. 1977;15:283–95
4. Marshall WK, Bedford RF, Miller ED. Cardiovascular responses in the seated position–impact of four anesthetic techniques. Anesth Analg. 1983;62:648–53
5. Gale T, Leslie K. Anaesthesia for neurosurgery in the sitting position. J Clin Neurosci. 2004;11:693–6
6. Soeding PF, Wang J, Hoy G, Jarman P, Phillips H, Marks P, Royse C. The effect of the sitting upright or ‘beachchair’ position on cerebral blood flow during anaesthesia for shoulder surgery. Anaesth Intensive Care. 2011;39:440–8
7. Pohl A, Cullen DJ. Cerebral ischemia during shoulder surgery in the upright position: a case series. J Clin Anesth. 2005;17:463–9
8. Hitselberger WE, House WF. A warning regarding the sitting position for acoustic tumor surgery. Arch Otolaryngol. 1980;106:69
9. Wilder BL. Hypothesis: the etiology of midcervical quadriplegia after operation with the patient in the sitting position. Neurosurgery. 1982;11:530–1
10. Matjasko J, Petrozza P, Cohen M, Steinberg P. Anesthesia and surgery in the seated position: analysis of 554 cases. Neurosurgery. 1985;17:695–702
11. Haisa T, Kondo T. Midcervical flexion myelopathy after posterior fossa surgery in the sitting position: case report. Neurosurgery. 1996;38:819–21 discussion 821–2
12. Nitta H, Yamashita J, Nomura M, Igarashi N. Cervical spinal cord infarction after surgery for a pineal region choriocarcinoma in the sitting position: case report. Neurosurgery. 1997;40:1082–5 discussion 1085–6
13. Morandi X, Riffaud L, Amlashi SF, Brassier G. Extensive spinal cord infarction after posterior fossa surgery in the sitting position: case report. Neurosurgery. 2004;54:1512–5 discussion 1515–6
14. Lam AM, Baldwin G. Blood pressure and adverse perioperative neurologic outcomes: an uncomfortable position. Anesth Analg. 2012;114:1156–9
15. Drummond JC, Lee RR, Howell JP Jr. Focal cerebral ischemia after surgery in the “beach chair” position: the role of a congenital variation of circle of Willis anatomy. Anesth Analg. 2012;114:1301–3
16. Friedman DJ, Parnes NZ, Zimmer Z, Higgins LD, Warner JJ. Prevalence of cerebrovascular events during shoulder surgery and association with patient position. Orthopedics. 2009;32:256
17. Yadeau JT, Casciano M, Liu SS, Edmonds CR, Gordon M, Stanton J, John R, Shaw PM, Wilfred SE, Stanton M. Stroke, regional anesthesia in the sitting position, and hypotension: a review of 4169 ambulatory surgery patients. Reg Anesth Pain Med. 2011;36:430–5
18. Black S, Ockert DB, Oliver WC Jr, Cucchiara RF. Outcome following posterior fossa craniectomy in patients in the sitting or horizontal positions. Anesthesiology. 1988;69:49–56
19. Young ML, Smith DS, Murtagh F, Vasquez A, Levitt J. Comparison of surgical and anesthetic complications in neurosurgical patients experiencing venous air embolism in the sitting position. Neurosurgery. 1986;18:157–61
20. Drummond JC. The lower limit of autoregulation: time to revise our thinking? Anesthesiology. 1997;86:1431–3
21. Kirby RR, Cullen DJ. Lower limit of cerebral autoregulation questioned. APSF Newsletter. 2009;24:5
22. Drummond J, Hargens AR, Patel PM. Hydrostatic gradient is important: blood pressure should be corrected. APSF Newsletter. 2009;24:6
23. Trentman TL, Fassett SL, Thomas JK, Noble BN, Renfree KJ, Hattrup SJ. More hypotension in patients taking antihypertensives preoperatively during shoulder surgery in the beach chair position. Can J Anaesth. 2011;58:993–1000
24. Cullen DJ, Kirby RR. Beach chair position may decrease cerebral perfusion: Catastrophic outcomes have occurred. APSF Newsletter. 2007;22:25–7
25. Cucchiara RF. Hazards of beach chair position explored. APSF Newsletter. 2008;22:81
26. Lanier WL. Cerebral perfusion: Err on the side of caution. APSF Newsletter. 2009;24:1
27. Munis JR, Lozada LJ. Giraffes, siphons, and starling resistors. Cerebral perfusion pressure revisited. J Neurosurg Anesthesiol. 2000;12:290–6
28. Munis JR. The problems of posture, pressure, and perfusion. APSF Newsletter. 2008;22:88
© 2013 International Anesthesia Research Society
29. Hicks JW, Munis JR. The siphon controversy counterpoint: the brain need not be “baffling”. Am J Physiol Regul Integr Comp Physiol. 2005;289:R629–32