Vitreo-retinal surgery produces, highly variable levels of stimulation ranging from the trivial to the extreme and changing minute by minute. It is often difficult to titrate anaesthesia and analgesia with sufficient feedback precision in such a dynamic system. Consequently, this may result in large swings in blood pressure or an increased opioid requirement, which might contribute to postoperative nausea and vomiting. Patients undergoing vitreo-retinal surgery are often elderly and frequently have co-morbidity, e.g. diabetes mellitus, hypertension and ischaemic heart disease. It would seem reasonable to assume that the use of a supplementary local block would be beneficial in this group. The commonest local technique is the peribulbar block, which is used extensively in cataract surgery. There is growing concern that the incidence of globe perforation with this technique, although small, may swing the balance of risk and benefit away from its routine use . One might also argue that, since the principal 'benefit' of any local anaesthetic technique is that it allows the patient to avoid the (possibly greater) risk of general anaesthesia, this 'benefit' is lost when the local block is used as a supplement to general anaesthesia. Of greater concern is that the population of patients undergoing vitreo-retinal surgery has a high proportion of 'high myopes', for whom blind needle techniques carry a greater risk and are relatively contraindicated. By contrast, sub-Tenon's anaesthesia, described by Hansen and colleagues  and Stevens , provides rapid effective anaesthesia and akinesia without the risks associated with the retrobulbar and peribulbar techniques. Sub-Tenon's anaesthesia offers a significantly reduced risk of complications, such as scleral perforation, retrobulbar haemorrhage, optic nerve injury and injection of anaesthetic solution into the subarachnoid space, because no sharp instrument enters the orbit [4,5]. Sub-conjunctival haemorrhage and mild chemosis occupying one or two quadrants of the bulbar conjunctiva were originally the only complications reported with this technique , although recently reports of other complications which might be due to sub-Tenon's block, including posterior vitreous detachment have been described . In as far as we can currently judge, sub-Tenon's anaesthesia may be safely applied to this population of patients as a supplement to general anaesthesia. The aim of our study was to investigate whether this block, widely used in surgery of the anterior chamber, offered any benefits over placebo in vitreo-retinal surgery with respect to intraoperative opioid requirements, cardiovascular stability and postoperative pain and nausea.
The study was designed using a 'clamp' technique. This type of servo control technique is classically used in contexts where it is inappropriate to allow the dependent variable (in our case, perceived surgical stimulus) to vary uncontrollably. In this paradigm, the proposed dependent variable is held constant by an intervention. The magnitude of this intervention (as a function of time) then becomes a proxy for the time-varying degree of system perturbation which would have occurred in the unclamped system.
Arterial pressure and surgical stimulus can vary greatly in vitreo-retinal surgery and we assume variations in the former are indicative of inadequate analgesia with respect to the latter. We sought to control arterial pressure within narrow limits by ad libitum administration of alfentanil.
After approval from our Institutional Research Ethics Committee, 43 patients scheduled for vitreo-retinal surgery were randomized to either sub-Tenon's block with bupivacaine 0.75% or a sham block with saline. All patients were premedicated with ranitidine 150 mg and received a standardized general anaesthetic. Each patient received midazolam 30 μg kg−1 intravenously (i.v.) 2 min before induction of anaesthesia, with a target-controlled infusion of propofol to a concentration of 4.5 μg mL−1. Following muscle relaxation with atracurium 0.5 mg kg−1, a reinforced laryngeal mask airway was inserted and the lungs were ventilated with 50% oxygen in nitrous oxide. The target concentration of propofol was reduced to 2.0 μg mL−1 after 2 min of ventilation, during which time the radial artery was cannulated. The patient was transferred to the operating room and baseline arterial pressure was recorded. The sub-Tenon's block was performed by the operating surgeon (one of two surgeons) at the earliest point using the inferonasal one-quadrant as follows: the conjunctival sac was anaesthetized with topical amethocaine (1%). The conjunctiva and the Tenon's capsule were then grasped with Moorfield's forceps, 4-5 mm inferonasal to the limbus. A small incision was then made in the conjunctiva through the Tenon's fascia, down to bare sclera, using blunt-tipped Wescott's spring scissors. At this point, a blunt Southampton curved cannula was passed into the sub-Tenon's space over the bare sclera, until the tip was behind the equator. Then, 4-5 mL of the test solution was injected slowly with gentle advancement of the cannula posteriorly. After injection, the cannula was withdrawn.
Standard anaesthetic monitoring as well as invasive arterial pressure were recorded with a Datex AS/3® monitor and downloaded to a PC with a refresh rate of 6 s. Deviation in arterial pressure beyond our 'set-point' tolerance was defined as a rise in arterial pressure >25% of baseline and which persisted for >3 min. 'Baseline' is defined as the arterial pressure after induction and after stabilization, before the start of surgery. Deviation from this was assumed to be indicative of inadequate analgesia and was treated with a bolus dose of alfentanil 250 μg. According to the protocol, if a 'deviation' continued despite three contiguous doses of alfentanil, the target concentration of propofol was increased by 25%. A 'bradycardic episode' was defined as a heart rate (HR) <40 beats min−1 and was treated by temporary cessation of the surgical stimulus and atropine 300 μg.
A postoperative questionnaire was completed at 12 and 24 h by the nursing staff, who recorded the worst pain and nausea scores in the first and second 12 h postoperatively. Scores were nominal and as follows: 'None or almost none', 'Mild', 'Moderate' and 'Severe'. Postoperative pain relief was prescribed to be taken as required. This comprised an oral acetaminophen (paracetamol) 500 mg/codeine 30 mg combination and morphine 10 mg. The consumption of analgesics and antiemetics was recorded.
The patient characteristics of the two groups are shown in Table 1. There was no significant difference in patient age or duration of operation, or in the proportions of patients with hypertension, ischaemic heart disease or diabetes between groups. There was no significant difference between the groups with respect to additional surgical procedures undertaken. The mean axial length (±2 SD) for both combined groups was 24.22 (±2.8) mm. All but three postoperative data sheets were returned completed.
The distribution of data is shown as a series of box plots in Figures 1-5 (SPSS® for Windows v.9 (SPSS Inc, Chicago, ILL, USA) showing median, interquartile range and outliers). Due to the experimental design, the frequency of intraoperative 'blood pressure set-point deviations' and the number of doses of alfentanil were identical, the latter serving as a proxy for the former. These data were analysed with the U-test (SPSS® for Windows v.9) and all of the differences depicted in Figures 1-5 are significant at the level indicated below each plot.
Table 2 shows the frequencies of episodes of 'breakthrough' hypertension experienced by the two groups. Also shown is the postoperative consumption of oral analgesia (acetaminophen/codeine), which is significantly reduced in the sub-Tenon's group. Only five patients required intramuscular (i.m.) morphine in the first 24 h postoperatively. All were in the control group.
In order to determine which surgical interventions produced the greatest stimulation, the incidence of arterial pressure set-point deviations occurring within 2 min of the various surgical events were analysed for the control group (Table 3). The 'risk' of a response to each of these is expressed as an odds ratio (logistic regression analysis, SPSS® for Windows v. 9). Explant fixation and cryotherapy/scleral indentation are associated with the greatest propensity to stimulation, whereas laser photocoagulation and gas instillation are associated with the least.
Arterial pressure set-point deviation episodes
Our results show that the use of a sub-Tenon's bupivacaine block, in addition to general anaesthesia, increases the stability of our control loop, and effectively moderates surgical stimulus with consequent reduction in the need for supplementary opioid analgesia or increase in anaesthetic depth. This is demonstrated in Figure 1, which shows a median value of five blood pressure set-point deviations in patients in the control group compared to zero in the sub-Tenon's group. There is of course some scatter in these data, particularly in the control group in which the greatest outlier was 14 episodes. Overall, 85% patients in the control group experienced at least one blood pressure set-point deviation, by our criteria, compared to 22% of patients in the sub-Tenon's group. This agrees qualitatively with data from Williams and colleagues' non-randomized study of peribulbar anaesthesia compared to a retrospective general anaesthesia control group . Here the incidence of at least one hypertensive episode was 14% in the control group and 4% in the block group. The qualitative difference is probably related to differences in technique, study design and their broader tolerance for defining a 'hypertensive episode'.
Cardiovascular instability is likely to have deleterious effects on myocardial oxygen demand and can be considered to be a proxy indicator of surgical stress response. Both of these are thought to adversely affect outcome, particularly in certain patient subgroups, e.g. those with a history of hypertension and ischaemic heart disease [8-10]. In addition, with such dynamic changes in the degree of surgical stimulation as characteristically occur in vitreo-retinal surgery, it is difficult to titrate the level of anaesthesia and analgesia with sufficient feedback precision. This itself can contribute to the observed cardiovascular instability.
Our study was designed to clamp cardiovascular variables within a narrow tolerance, and so there ought to be no difference in these variables between the two groups. However, despite this, a number of patients deviated beyond our set tolerance despite feedback control. We arbitrarily define a period of 'breakthrough' hypertension or tachycardia as being >40% above baseline and observed that this occurred in eight patients in the control group compared to none in the sub-Tenon's group. It could be argued that this phenomenon was a consequence of insensitivity and lag in the control loop. However, cardiovascular variables were measured in real time and invasively we designed the system to have relatively narrow error tolerance. In addition, our 'effector' treatment (alfentanil) is a potent opioid with a rapid onset of action. All of these factors should contribute to a sensitive and rapid servocontroller.
Patients in the sub-Tenon's group also suffered fewer bradycardic episodes. Figure 2 shows that the median number of episodes was one in the control group patients (range 0-3) and zero (range 0-1) in the sub-Tenon's group. Overall, 57% patients in the control group suffered at least one episode of bradycardia, compared to 9% in the sub-Tenon's group. These data agree qualitatively with Williams and colleagues' study , where the incidences were 17% and 2%, respectively, and also with Shende and colleagues' randomized controlled trial of peribulbar block as an adjunct to general anaesthesia . In this latter study, the incidence of at least one episode of bradycardia was 70% in the control group and 30% in the block group. However, Chung and colleagues, in a small randomized trial of peribulbar anaesthesia in addition to thiopental/isoflurane general anaesthesia for vitreo-retinal surgery, were unable to show any difference in the incidence of bradycardia .
Figures 3 and 4 summarize the data on postoperative pain at 12 and 24 h. Sub-Tenon's block appears to confer significant moderation in the severity of pain in the first 24 h, and particularly so in the first 12 h, where the median pain scores corresponded to 'moderate' for the control group and 'none or almost none' for the sub-Tenon's group. Analgesia was provided by oral acetaminophen/codeine in the majority of patients and consumption was significantly less in the sub-Tenon's group. The peribulbar studies of Chung and colleagues , Shende and colleagues  and Williams and colleagues  also show reduction in the incidence and severity of postoperative pain to varying degrees. Chung and colleagues reported incidences of pain in the first 2 h of 64% and 14% (control vs. block) and Shende and colleagues reported incidences of 63% and 23% at 12 h. Williams and colleagues reported a greater incidence in the need for postoperative opioid administration in the control group (41% vs. 3% in the block group). It is difficult to compare directly these data because of the different techniques and methodologies. Shende and colleagues used morphine perioperatively in the control group, which is likely to have a hangover effect on postoperative pain. They also analysed pain quantally, the frequency of 'pain' being the proportion of patients with a verbal numerical score >4 (possible range is 0-10). Chung and colleagues also expressed pain as a frequency. They used a visual analogue scale (VAS) but did not explicitly state what value constituted 'pain' in their analysis. We have analysed the pain scores of individual patients as a quasi-continuous variable in order to observe differences in the severity of pain rather than just its frequency at a given threshold. Not only was the incidence of postoperative pain in the first 12 h (of a severity of 'moderate' or worse) less in the sub-Tenon's group (10% vs. 66% in the control group) but also severity of pain was significantly reduced towards none or almost none. It is not clear, whether this effect is due to persistent local anaesthetic effect or whether de-afferentation of the operative site during surgery contributes in some way to subsequent pain perception.
Figure 5 summarizes the data on postoperative nausea. No patients in the sub-Tenon's group complained of nausea of any severity. The range of nausea scores in the control group ranged from none to severe, and a significant difference was demonstrated between the two groups. In terms of frequency of nausea (of severity 'mild' or greater), the incidences were 38% vs. 0% in the control and sub-Tenon's groups. Other studies of peribulbar block [7,11,12] have shown a reduction in nausea and vomiting; however, as we have already discussed in the context of postoperative pain, direct comparison of data is difficult because of the differing general anaesthesia techniques, use of an antiemetic and intraoperative morphine. In our study, arterial pressure set-point deviations were treated with bolus doses of alfentanil. Therefore, by the design of the study, the number of doses of alfentanil was a proxy for the degree of perceived surgical stimulus. Consequently, patients in the control group had a significantly greater total dose of alfentanil; in fact, the abscissa of Figure 1 could have been labelled as the 'number of 250 μg doses of alfentanil'. Clearly from these data, it is not possible to determine absolutely the degree of any negative contribution of alfentanil over any putative positive contribution of the sub-Tenon's block itself to postoperative antiemesis. However, in a pragmatic sense, we can say at the very least that sub-Tenon's block permits a reduction in intraoperative opioid requirement and this may contribute to the observed significant reduction in postoperative nausea in the first 12 h. By 24 h, the incidence of nausea was low in both groups, and no significant difference in severity could be detected.
Our study has shown that sub-Tenon's block, when used in addition to general anaesthesia in vitreo-retinal surgery, has a useful opioid-sparing effect, and can contribute to smooth intraoperative conditions. In addition, it improves postoperative pain and nausea. There were no ocular complications from either the bupivacaine or saline sub-Tenon's injections, and although a sample of this size cannot be used to evaluate complication risk, the data add support to studies elsewhere which suggest that this is a very safe technique .
1. Gillow JT, Aggarwal RK, Kirkby GR. A survey of ocular perforation during ophthalmic local anaesthesia in the United Kingdom. Eye
2. Hansen EA, Mein CE, Mazoli R. Ocular anesthesia for cataract surgery: a direct sub-tenon's approach. Ophthalmic Surg
3. Stevens JD. A new local anaesthetic technique for cataract extraction by one quadrant sub-Tenon's infiltration. Br J Ophthalmol
4. Guise PA. Single quadrant sub-Tenon's block. Evaluation of a new local anaesthetic technique for eye surgery. Anaesth Intens Care
5. Biggs M, Beck S, Esakowitz L. Sub-Tenon's versus peribulbar anaesthesia for cataract surgery. Eye
6. Rüschen H, Bremner FD, Carr C. Complications after sub-Tenon's eye block. Anesth Analg
7. Williams N, Strunin A, Heriot W. Pain and vomiting after vitreoretinal surgery: a potential role for local anaesthesia. Anaesth Intens Care
8. Desborough JP. The stress response to trauma and surgery. Br J Anaesth
9. Rao TL, Jacobs KH, El-Etr AA. Reinfarction following anesthesia in patients with myocardial infarction. Anesthesiology
10. Steen P, Tinker J, Tarhan S. Myocardial reinfarction after anesthesia and surgery. JAMA
11. Shende D, Sadhasivam S, Madan R. Effects of peribulbar bupivacaine as an adjunct to general anaesthesia on perioperative outcome following retinal detachment surgery. Anaesthesia
12. Chung F, Westerling D, Chisholm LD, Squires GW. Postoperative recovery after general anaesthesia with and without retrobulbar block in retinal detachment surgery. Anaesthesia