Advice
Evidence review
Evidence review
Clinical and technical evidence
Five clinical studies were identified from the literature which investigated the use of End-tidal Control in Aisys anaesthesia delivery systems. These studies included 1 randomised trial (Potdar et al. 2014), 2 prospective observational studies (Tay et al. 2013, Lucangelo et al. 2014), 1 service evaluation (Singaravelu and Barclay 2013) and 1 audit (Kennedy and French 2014).
The randomised trial by Potdar et al. investigated the cost-saving potential and environmental impact of End-tidal Control (n=100) compared with manually‑controlled anaesthesia (n=100). Their primary economic findings are described in the published economic evaluation section of this briefing. Analysis of secondary clinical outcomes found significant differences in the time needed to achieve an end-tidal concentration of sevoflurane of 1.5%, maximum inspired concentration of sevoflurane achieved, and the number of adjustments needed to maintain the depth of anaesthesia between End‑tidal Control and manual-control groups. A summary of these results is reported in table 1.
The prospective observational study by Tay et al. was a before-and-after study in a single Australian teaching hospital, which evaluated the manual control of end-tidal gases and automated End-tidal Control after planned replacement of anaesthesia systems. Primary outcomes and parameters measured included volatile agent costs, greenhouse gas emissions, carbon dioxide absorbent costs, and fresh gas costs. Secondary outcomes included results from voluntary case reports completed by anaesthetists describing when End‑tidal Control had been used, and their reasoning when they had decided against its use. The author described results as 'cases' and not 'patients' to take account of patients who had more than 1 episode of general anaesthesia during the study, that is, multiple cases. Despite 1865 cases of general anaesthesia (having manual control of end-tidal gases in this study, only 1036 cases (of the potential 1810 general anaesthesia cases for whom automated gas control was considered suitable) were explicitly confirmed to have had anaesthesia with End-tidal Control. The study reports a 44% reduction in greenhouse gas emissions when using End-tidal Control; however, fresh gas flow was different between groups and this may have affected these results. In the voluntary case reports, anaesthetists described four reasons for not choosing to use End-tidal Control which are listed in table 2. The authors discussed several concerns about automated End-tidal Control in children. These included: circle system resistance, dead space, the safety profile of low-flow anaesthesia for children, the possibility of gas leaks associated with uncuffed endotracheal tubes triggering a safety check, exit from the automated control mode, and End-tidal Control defaulting to a high fresh gas flow rate of 6 litres per minute. A summary of the clinical outcomes and results is reported in table 2.
The prospective observational study by Lucangelo et al. aimed to compare oxygen, air and anaesthetic consumption during manual and End-tidal Control anaesthesia, using the same anaesthetic system and identical fresh gas flow between groups. The study included 80 consecutive patients having elective abdominal surgery who were assigned to anaesthesia rooms equipped with anaesthesia systems with (n=40) or without (n=40) End-tidal Control, as determined by scheduling availability. The study found no difference in anaesthetic agent or fresh gas consumption between manual and End-tidal Control groups. However, in the manual‑control group the researchers found that a total of 137 interventions were needed by the anaesthetist (including 50 for undershoot and 87 for overshoot, which are transient changes in dosing levels in which the closed-loop control systems overcompensate) to stabilise the end-tidal anaesthetic agent concentration, and 107 interventions to stabilise the end-tidal oxygen concentration. No interventions were reported for the End-tidal Control group. A summary of the results is reported in table 3.
The service evaluation by Singaravelu and Barclay was a UK single-centre study using retrospective information from the event log files stored in Aisys anaesthetic systems. This information was used to compare fresh gas flow rates, inhalational anaesthetic use and the need for user intervention between 168 patients having manual control and 321 patients having End‑tidal Control anaesthesia. One retrospectively applied exclusion criterion removed patients from the study who had anaesthesia for less than 10 minutes. This was because insufficient data prevented a full analysis of system performance in the maintenance phase of anaesthesia. The study reported a reduction in average anaesthetic use of 40‑55% in the End‑tidal Control group, and a reduction in the average number of key presses per patient, from 13.6 key presses with manual control to 6.5 with End-tidal Control. A summary of the results is in table 4.
The study by Kennedy and French described data from an audit that monitored fresh gas flow rates within a single department in a New Zealand hospital. This study compared data retrospectively exported from Datex (now acquired by GE Healthcare) anaesthesia delivery units (from 2001, 2006 and 2009) with detailed event logs retrospectively downloaded from Aisys anaesthesia systems, which had End-tidal Control installed (at 3 specified time periods: June 2011, December 2011, June 2012). A voluntary survey of anaesthetists using the Aisys systems was also done in 2012). Data from 2 other New Zealand hospitals (from 2007 and 2008) also using the Datex anaesthesia delivery units were described and compared. The number of patients included in the audit was not stated. The study reported a general reduction in mean fresh gas flow rates using the anaesthesia delivery units over time: 2.05 litres per minute in 2001, 1.43 litres per minute in 2006 and 1.26 litres per minute in 2009, and that fresh gas flow rates were similar for all 3 New Zealand hospitals described. However, on introduction of the Aisys systems with End-tidal Control, the mean fresh gas flow rate initially increased to 1.50 litres per minute, but dropped to 1.09 litres per minute after 12 months. A summary of the results is in table 5.
Table 1 Summary of the Potdar et al. (2014) randomised trial
|
Study component |
Description |
|
Objectives/hypotheses |
Hypothesis: End-tidal Control anaesthesia is an effective and safe system that would reduce consumption of gases, thus reducing cost and also environmental pollution. |
|
Study design |
Prospective, randomised, single-blind study. Randomisation was conducted using a chit-pull system, in which odd numbers were allocated to the manual‑control group and even numbers allocated to the End‑tidal Control group. |
|
Setting |
Single centre (Indian hospital). |
|
Inclusion/exclusion criteria |
Inclusion criteria: patients having laparoscopic abdominal and pelvic surgery, aged 15–75 years, ASA classification of physical health of 1 or 2, surgical procedure with a minimum of 30 minutes and a maximum of 4 hours under general anaesthesia, patient intubated with endotracheal tube and with controlled ventilation, patients maintained only on sevoflurane and not on any other agents such as propofol, midazolam or sedative infusions. Exclusion criteria: general anaesthesia with laryngeal mask airway, face mask, and spontaneous respiration, patients having cardiac, renal, and respiratory diseases, neurological or psychological illness that may interfere with entropy monitoring, ASA classification of 3 or 4, emergency surgery, patients having haemodynamic instability intraoperatively, a variation of pulse or blood pressure more than 20% of baseline or entropy values of <40 and >60 in the maintenance period of anaesthesia for more than 5 minutes. |
|
Primary outcomes |
Time needed to achieve end-tidal concentration of sevoflurane of 1.5%. Maximum inspired concentration of sevoflurane. Number of adjustments needed to maintain depth of anaesthesia (targeting entropy values between 40 and 60, monitored via an additional device). |
|
Statistical methods |
Initial sample size was not pre‑determined. To ensure adequate sample size, a power calculation was performed retrospectively based on the difference in total cost of anaesthesia per hour between groups. Correlations among different measurements were assessed using Pearson's correlation coefficients. A p value <0.05 was considered statistically significant. A general linear model (ANOVA) was used to investigate and model the effect of various parameters with costs. |
|
Participants |
200 patients randomly assigned to End-tidal Control of inhalational agent (n=100), or manual control (n=100). |
|
Results |
Cost saving potential of End-tidal Control results are presented in the published economic evaluation section. Consumption of oxygen, nitrous oxide and sevoflurane gases Consumption of nitrous oxide was significantly less in the End‑tidal Control group (0.70 litre/minute) than in the manual-control group (0.83 litre/minute), p=0.001. Consumption of sevoflurane was statistically significantly less in the End-tidal Control group than in the manual-control group (0.17 litre/minute vs 0.20 litre/minute), p=0.0001. Oxygen consumption was also less in the End-tidal Control group than in the manual-control group (1.74 litre/minute vs 1.83 litre/minute) but was not statistically significantly different, p=0.21. Time needed to achieve end-tidal concentration of sevoflurane of 1.5% There was a statistically significant difference between the 2 groups (3.08 minutes for End-tidal Control vs 13.40 minutes for manual control), p=0.0001. Maximum inspired concentration of sevoflurane There was a statistically significant difference between the 2 groups (2.66% for End-tidal Control vs 2.11% for manual control), p=0.0001. Number of adjustments needed to maintain the depth of anaesthesia The number of drug delivery adjustments was 3 per patient in the End-tidal Control group. The number of adjustments in the manual-control group varied from 5 to 12. There was a statistically significant difference between the average number of adjustments between the 2 groups, p=0.0001. |
|
Conclusions |
The authors concluded that End-tidal Control is a good system for conserving the consumption of gases, and reducing the number of adjustments needed to maintain depth of anaesthesia. |
|
Abbreviations: ANOVA, analysis of variance; ASA, American Society of Anesthesiologists. |
|
Table 2 Summary of the Tay et al. (2013) prospective before-and-after observational study
|
Study component |
Description |
|
Objectives/hypotheses |
Compared with the conventional practice of using manual control in the delivery of volatile agents, the automated control of end‑tidal anaesthetic gases in a clinical setting would produce a significant difference in volatile agent consumption cost and the rate of greenhouse gas emissions. |
|
Study design |
Prospective before-and-after observational study. |
|
Setting |
A single tertiary hospital (teaching hospital in Australia), which included a 12 week manual phase (January to April 2011), followed by a preparation and education phase of 2 months (April to May 2011) to introduce the Aisys Carestation with End-tidal Control to all medical, nursing and technical assistance staff. A 12 week End‑tidal Control phase (July to October 2011) was then implemented for comparison. |
|
Inclusion/exclusion criteria |
Inclusion criteria: all patients needing elective or emergency surgery involving general anaesthesia with a volatile agent. Exclusion criteria: patients needing cardiac or neuro surgery, total intravenous anaesthesia, electroconvulsive therapy, sedation and regional anaesthesia without a volatile agent general anaesthetic. |
|
Primary outcomes |
Voluntary case report from anaesthetists in End-tidal Control phases. Greenhouse gas emissions. |
|
Statistical methods |
Patient baseline characteristics and categorical variables were compared by the chi-squared test and continuous variables were tested for normality and compared by a 2‑sample Wilcox on rank‑sum (Mann–Whitney) test. Mean differences and 95% confidence intervals were reported. P values <0.05 were considered statistically significant. |
|
Participants |
3675 cases of general anaesthesia. Of these, 1865 were in the manual phase (age range: 2 months to 94 years), and 1810 in the End-tidal Control phase (age range: 6 months to 91 years). Of the 1810 cases in the End-tidal Control phase, 1169 had voluntary case report forms returned, which indicated End-tidal Control was used in 1036 cases (and not used in 133 cases). |
|
Results |
Volatile agent cost per hour, carbon dioxide absorbent use/costs and fresh gas costs are presented in the published economic evaluation section. Voluntary case reports Reasons reported for not using End-tidal Control included anaesthetists not being aware that End-tidal Control was available to them, quick surgical cases, leaks in the breathing system often as a result of inadequately placed airway device and use in children 6 years or younger. Greenhouse gas emissions The rate of greenhouse gas emissions was 13.0 kg/hour (SD 6.2) in the End-tidal Control phase and 23.2 kg/hour (SD 10.8) in the manual phase, an absolute reduction of 10.2 kg/hour (95% CI: 2.7 to 17.7 kg/hour, p=0.0179) or a relative reduction of 44% when using End-tidal Control. |
|
Conclusions |
The authors concluded that the use of End-tidal Control increases participation in low-flow anaesthesia with environmental benefits. |
|
Abbreviations: CI, confidence interval; SD, standard deviation |
|
Table 3 Summary of the Lucangelo et al. (2014) prospective observational study
|
Study component |
Description |
|
Objectives/hypotheses |
To compare oxygen, air and anaesthetic consumption during manual and End-tidal Control low-flow anaesthesia provided by the same anaesthetic machine using identical fresh gas flow (1 litre/min). |
|
Study design |
Prospective observational study of consecutive patients admitted to operating rooms that either had the End-tidal Control feature present or absent on the anaesthetic machine. |
|
Setting |
Operating rooms in a single hospital (hospital name and dates of study were not stated in the paper). |
|
Inclusion/exclusion criteria |
Inclusion criteria: 18–80 years of age, ASA classification of physical health of 1 or 2, expected duration of surgery exceeding 1 hour. Exclusion criteria: BMI exceeding 30, chronic use of opioids, contraindication to any component of the anaesthesia protocol, neurological disorders, and arterial hypertension. |
|
Primary outcomes |
Anaesthetic machine characteristics. Amount of consumed gases. Oxygen and sevoflurane efficiencies. Number of interventions by the anaesthetist. |
|
Statistical methods |
Normality was assessed by the Kolmogorov–Smirnov–Lilliefors test. Non-normal data were described as medians [IQR]. The Mann–Whitney test was used to compare data between groups. The adjusted p value was calculated according to Dineen and Blakesley method. The significance level was set at 5%. |
|
Participants |
80 consecutive patients admitted to the operating room in need of elective abdominal surgery under general anaesthesia, 40 of whom were anaesthetised with the End-tidal Control feature present on the anaesthetic machine and 40 with End-tidal Control absent. No difference in patient characteristics was found between groups. No patient was excluded from the trial. |
|
Results |
No clinical complications were observed. Anaesthetic machine characteristics Tidal volume, respiratory rate, duration of anaesthesia, sevoflurane delivery, and awakening time did not differ significantly between End-tidal Control and manual-control anaesthesia groups. The median [IQR] time to reach target end-tidal anaesthetic agent concentration was 145 [130–171] seconds with End-tidal Control and 71 [43–98] seconds with manual control (P<0.00001). The median [IQR] time to maintain steady end-tidal anaesthetic oxygen concentration was 145 [130–171] seconds with End-tidal Control and 360 [278–531] seconds with manual control (P<0.00001). Amount of consumed gases The median [IQR] oxygen delivery was 87 [48–120] litres with End-tidal Control, and 74 [52–105] litres with manual control. The median [IQR] sevoflurane delivery was 15 [11–23] ml with End‑tidal Control and 17 [12–23] ml with manual control. The median [IQR] oxygen uptake was 260 [231–275] ml/minute with End-tidal Control and 252 [226–277] ml/minute with manual control. The median [IQR] sevoflurane uptake was 3.7 [2.3–4.4] ml/minute with End-tidal Control and 3.8 [3.0–4.4] ml/minute with manual control. The delivery and uptake of oxygen and sevoflurane were not significantly different between manual and End-tidal Control groups. Oxygen and sevoflurane efficiencies The median [IQR] oxygen efficiency was 47 [34–60] % with End‑tidal Control and 51 [44–62] % with manual control. The median [IQR] sevoflurane efficiency was 21 [12–39] % with End-tidal Control and 22 [14–40] % with manual control. The oxygen and sevoflurane efficiencies were not significantly different between manual and End‑tidal Control groups. Number of interventions by the anaesthetist To reach the pre-established end-tidal anaesthetic agent concentration, the median number of interventions in the manual‑control group was 4 (with a total of 137, including 50 for undershoot and 87 for overshoot of end‑tidal anaesthetic agent concentration). No interventions were needed for the End-tidal Control group. To maintain the end-tidal oxygen concentration, 107 interventions were needed in the manual‑control group with all patients needing at least 1 intervention. No interventions were needed for the End‑tidal Control group. |
|
Conclusions |
The authors concluded that low-flow anaesthesia delivered with an anaesthetic machine able to automatically control end‑tidal anaesthetic and oxygen concentrations provided the same clinical stability as that of manually-controlled anaesthesia. Similar oxygen and sevoflurane consumption was reported between groups; however End-tidal Control avoids the continuous manual adjustment of delivered sevoflurane and oxygen concentrations. |
|
Abbreviations: ASA, American Society of Anesthesiologists; CI, confidence interval; IQR, interquartile range
|
|
Table 4 Summary of the Singaravelu and Barclay (2013) service evaluation
|
Study component |
Description |
|
Objectives/hypotheses |
To evaluate End-tidal Control in clinical practice by measuring inhalation anaesthetic use and the need for user intervention and comparing this with contemporaneous surgeries done using manual control of fresh gas flow. |
|
Study design |
Service evaluation. |
|
Setting |
Gynaecology theatres within a single UK centre (Liverpool Women's Hospital) between June and October 2010. |
|
Inclusion/exclusion criteria |
Because of the study design, initial inclusion and exclusion criteria were not explicitly described in the paper. Subsequent exclusions, applied retrospectively, included patients with duration of anaesthesia of less than 10 minutes. |
|
Primary outcomes |
Inhalation anaesthetic use. User intervention. |
|
Statistical methods |
Data were compared using Spearman correlation and t-tests. |
|
Participants |
321 patients were anaesthetised using End-tidal Control (n=181 sevoflurane, n=140 desflurane). 168 patients were anaesthetised using manual control of fresh gas (n=143 sevoflurane, n=25 desflurane). |
|
Results |
Inhalation anaesthetic use Average fresh gas flow during End-tidal Control decreased significantly with increased duration of anaesthesia (Spearman r=−0.88, p=0.0016). When comparing anaesthetics of the same duration, the average volatile anaesthetic use was consistently reduced by 40–55% in the End-tidal Control group. User intervention The mean number of key presses was 6.5 (95% CI 6.0 to 7.0) with End-tidal Control, and 13.6 (95% CI 12.8 to 14.4) with manual control. Secondary outcomes With End-tidal Control, the measured end-tidal concentration was within 10% of the set target for 98% of the total time spent in steady state, allowing 5 minutes for equilibration after each change in the set target. The mean difference between measured end-tidal concentration and target end-tidal concentration using End-tidal Control was 1.47 (95% CI: 1.29 to 1.66) %. |
|
Conclusions |
The authors concluded that automatic implementation of low-flow anaesthesia using End-tidal Control allows the user to set and maintain a desired end-tidal volatile concentration while using less anaesthetic and reducing the number of interventions needed by the clinician. |
|
Abbreviations: CI, confidence interval |
|
Table 5 Summary of the Kennedy and French (2014) audit
|
Study component |
Description |
|
Objectives/hypotheses |
To describe the effect of the introduction of the Aisys anaesthesia machine with automated control of end-tidal vapour concentration on fresh gas flow rates. |
|
Study design |
Audit study. |
|
Setting |
Single theatre suite (comprising 11 operating theatres) in a New Zealand (Christchurch) hospital using Datex ADUs (during 2009), and from Aisys machines with End‑tidal Control (in June 2011, December 2011 and June 2012). Comparative data from the same hospital using ADUs (from 2001 and 2006) were described, as well as data from 2 other major tertiary (Middlemore hospital) and secondary (North Shore hospital) care metropolitan public hospitals in New Zealand also using ADUs (in 2007 and 2008 respectively). |
|
Inclusion/exclusion criteria |
Because of the study design, no inclusion or exclusion criteria were described. However, the authors do state that the theatres in the Christchurch hospital had a broad mix of adult elective and acute surgery, but that data were not collected from operating theatres with significant paediatric practice or from cardiothoracic or neurosurgery operating theatres. |
|
Primary outcomes |
Mean fresh gas flow rates. Online voluntary survey results. |
|
Statistical methods |
Not reported. |
|
Participants |
Population size and demographics not described. |
|
Results |
Mean fresh gas flow rates End-tidal Control: Christchurch Aisys June 2011: 1.50 litre/minute Christchurch Aisys Dec 2011: 1.29 litre/minute Christchurch Aisys June 2012: 1.09 litre/minute Manual gas control: Christchurch ADU 2001: 2.05 litre/minute Christchurch ADU 2006: 1.43 litre/minute Middlemore ADU 2007: 1.24 litre/minute North Shore ADU 2008: 1.27 litre/minute Christchurch ADU 2009: 1.26 litre/minute The overall proportion of time spent in End-tidal Control mode with the Aisys machines was 34% in June 2011, 60% in December 2011 and 61% in June 2012. There is an association between reduction in flow rates and increasing proportion of time spent in End-tidal Control mode. Online voluntary survey results The survey was completed by 68/90 anaesthetists (75%), including 18 trainees and 50 specialist anaesthetists. End-tidal Control was used 'often' or 'most of the time' by 67% of respondents. The reasons most commonly selected for not using End-tidal Control were the need to teach trainees (47.7% 'relevant' or 'very relevant') and when using total intravenous anaesthesia (34.9% of respondents). Major issues reported with End-tidal Control were:
Major advantages reported for End‑tidal Control were:
|
|
Conclusions |
The authors concluded that automatic control of anaesthetic agent concentration can lead to reduction in overall fresh gas flows. |
|
Abbreviations: ADU, anaesthesia delivery units |
|
Costs and resource consequences
The manufacturer stated that as of May 2014, 401 Aisys Carestation units have been sold to 55 UK hospitals and 37 Aisys CS2 units have been sold to 9 UK hospitals, giving a total of 438 systems across the UK. End-tidal Control has been purchased for 425 (97%) of these systems.
Approximately 2.4 million people had general anaesthesia in 2007 in England (NICE 2012), but there are no data available to quantify how many of these general anaesthesia patients had inhalational anaesthetic agents compared with intravenous anaesthesia (noting the potential overlap of patients having both inhalational and intravenous anaesthetic agents). Two specialist commentators have estimated that less than 10% of patients will have total intravenous anaesthesia, leaving approximately 90% having some form of inhalational anaesthesia. However it is difficult to precisely estimate the total UK population for whom End‑tidal Control could be used.
Published economic evaluation
Four of the 5 reviewed studies included an economic evaluation of End-tidal Control compared with manual control of anaesthetic gases. In all studies, this economic evaluation was limited to consumption of gases. Only 1 of these was done in the UK, and only 1 standardised fresh gas flow between groups to eliminate a potential confounding factor.
The service evaluation by Singaravelu and Barclay, the only reviewed study done in the UK, found that for surgery of 20–40 minutes duration, the average cost of volatile anaesthetic per hour was reduced from £14.92 to £6.98 (saving £7.94 per hour) for sevoflurane and £11.91 to £7.08 (saving £4.83 per hour) for desflurane with End-tidal Control.
The prospective observational study by Lucangelo et al. was the only published study specifically designed to use identical fresh gas flow between groups. This reported no significant difference in gas consumption between manual and End-tidal Control, and therefore no difference in cost.
For the 2 studies set in India and Australia, the reported costs have been converted to £GBP (pounds sterling).
The randomised trial by Potdar et al. stated that the total cost of oxygen, nitrous oxide and sevoflurane consumption decreased from 417.76 Indian rupees per hour in the manual-control group (n=100) to 353.95 Indian rupees per hour in the End-tidal Control group (n=100), p=0.0001. This translated to a saving of £0.64 per hour (using exchange rates on 6 June 2014 as stated on XE, because dates of inclusion were not reported in the study).
The prospective observational study by Tay et al. stated that the mean volatile anaesthetic (isoflurane, desflurane, sevoflurane) cost per hour (in Australian dollars) decreased from $18.87 (SD $6.15) in the manual-control group (n=1865) to $13.82 (SD $3.27) in the End-tidal Control group (n=1036), an absolute reduction of $5.05 (95% CI: $0.88 to $9.22, p=0.0243), or a relative reduction of 27%. This translated to an overall cost saving in volatile anaesthetic agent through the use of End-tidal Control of approximately £3.32 (95% CI: £0.58 to £6.06) per hour (using exchange rates on 1 November 2011 as stated on XE). Carbon dioxide absorbent usage was 144 kg in the End‑tidal Control phase ($4050) and 156 kg in the manual phase ($4108); the differences between groups for usage and costs were not statistically significant. Consumption savings of fresh gases (oxygen, air, nitrous oxide) from the medical gas supplier were not clinically significant between groups.
Strengths and limitations of the evidence
Four of the 5 reviewed studies were conducted outside the UK; therefore it is unclear how generalisable the results would be to the UK NHS. Additionally, none of the 5 reviewed clinical studies reported results from a paediatric population, with 1 study describing anaesthetists choosing not to use End-tidal Control in patients under 6 years. Therefore the applicability of End-tidal Control to a paediatric population was not identified from the literature.
The best quality evidence identified by literature review was the randomised single-blind study by Potdar et al. that included 200 patients. This trial appropriately randomised patients using a chit‑pull system of 200 labelled chits (or tickets) with odd numbers allocated to manual control and even numbers to End-tidal Control anaesthesia. Single blinding was appropriate in this study as anaesthetists cannot be blinded to the use of manual or End‑tidal Control of anaesthetic gases. This was the only reviewed study that took into account depth of anaesthesia as an outcome measure of both manual and End‑tidal Control groups. However, the study and its reporting had several weaknesses. None of the figures or table numbers were referred to correctly in the text, incorrect statistical tests were applied, and contradictory p values were stated in the results and discussion sections. The intervention and control arms were uneven at baseline, with patients in the End‑tidal Control group being significantly younger (mean age 38.9 years compared with 43.0 years), suggesting a possible weakness in the chosen method of randomisation. The authors described a power calculation to confirm adequate sample size to detect the difference in total costs between arms, but this was performed retrospectively, rather than prospectively (which would have been more appropriate). This trial may also lack external validity to the NHS in general because of the extensive exclusion criteria, in which only patients having laparoscopic abdominal or pelvic surgery, aged between 15–80 years, with surgery lasting 30 minutes to 4 hours, were considered.
The largest study identified was the prospective before-and-after observational study by Tay et al., which included 1865 cases of manually controlled anaesthesia and 1810 cases eligible for End-tidal Control. However, of this group of eligible cases, only 1036 cases (57.2%) were confirmed as using End-tidal Control via voluntary case report forms. The authors noted the intrinsic increased risk of bias caused by non-randomisation of patients to manual or automated control in their study. Although the study included patients of all ages who were having elective or emergency general anaesthesia with a volatile agent, it excluded patients having cardiac surgery (in line with manufacturer's stated contraindications) or neurosurgery procedures, which could increase the risk of selection bias and limit the generalisability of the data. The End-tidal Control and manual-control results were recorded during different time periods, which meant that the results could be influenced by seasonal variation. Whereas the characteristics of the 1810 cases eligible for manual and End-tidal Control were described and were not statistically different, the demographics of the 1036 End-tidal Control cases were not described or compared with the manual-control group. It was unclear whether the results stated in this study referred to the 1810 cases eligible for End-tidal Control, or the 1036 cases confirmed as having End-tidal Control anaesthesia.
The prospective observational study by Lucangelo et al. included 80 consecutive patients. The authors attempted to reduce bias by assigning patients to anaesthetic rooms with or without End‑tidal Control on the anaesthetic machine based on an operating schedule prepared by a surgeon who was unaware of the study design, although randomisation was not attempted. This was the only study reviewed that maintained identical fresh gas flow between groups, thus removing 1 potential confounding factor of analysis. The study was at risk of selection bias by excluding patients with expected surgery duration of less than 1 hour. This may have resulted in an overestimation of End‑tidal Control benefits, because the same study also stated that End‑tidal Control takes longer to reach the target end‑tidal anaesthetic concentration than manual control. Although the study recorded the number of key presses for the manual‑control group, the authors failed to record the initial setting of target end-tidal anaesthetic agent concentration, starting and stopping the End-tidal Control mode as equivalent key presses in the End‑tidal Control arm. This was the only identified study that addressed the safety of End‑tidal Control, by stating there were no clinical complications reported during the study. However this statement must be considered with caution given that only 40 patients had anaesthesia with End‑tidal Control on the anaesthetic machine in this study.
The service evaluation by Singaravelu and Barclay included 321 patients anaesthetised using End-tidal Control and 168 with manual control, and was the only identified UK study. The study may be subject to reporting bias because it retrospectively excluded patients having anaesthesia for less than 10 minutes, and therefore may have overestimated the benefit of End-tidal Control. The authors did not describe the characteristics of the patients involved in this study, so therefore it was not possible to assess the risk of selection bias or study generalisability. The authors identified the lack of information to describe each anaesthetist's reasons for the choice of manual or End-tidal Control mode as a limitation of their study. The proportion of patients having sevoflurane and desflurane was different between the 2 groups (44% of the End-tidal Control group had desflurane compared with 15% in the manual-control group); therefore any difference in average costs would be subject to performance bias.
The audit described by Kennedy and French compared the Aisys anaesthesia system with End-tidal Control used in a hospital in New Zealand, with historical data collected from a different anaesthesia system (the Datex anaesthesia delivery unit with manual control) from 3 hospitals across New Zealand. The difference in anaesthesia systems and time periods used in this study resulted in significant potential for bias in the results. Additionally, there was likely to be selection bias, because paediatric and neurosurgery patients were not included in the End-tidal Control group but were included for the manual-control groups at 2 different hospitals. The number of patients included and their characteristics were not described, which limited interpretation of the study. Results from the audit were analysed at 3 specified time points for the End-tidal Control group only and resulted in a change to default settings of the Aisys system during the audit. It was unclear whether similar feedback was provided to the manual-control group resulting in any change to clinical practice, thus introducing another potential source of performance bias. Although the authors reported an association between a reduction in flow rates and increasing proportion of time spent in End-tidal Control mode, it was unclear how this association was statistically tested.