Population

The clinical effectiveness and safety review suggests that different sedation techniques are suitable for different population groups (see chapter 6 on clinical effectiveness of sedation techniques). We developed models for the following common procedures:

• Dental procedures:

  • tooth extraction in children
  • tooth extraction in adolescents
• Short painful procedures: manipulation of forearm fracture
• Painless imaging: CT scan
• Endoscopy:

  • Oesophago-gastroscopy
  • Colonoscopy

A description of these groups is given in section 6.12, ‘Further evidence to recommendations: clinical interpretation of evidence by setting’.

Interventions

The techniques are those for which the evidence suggests are clinically effective and safe (see chapter 6 on clinical effectiveness of sedation techniques). The GDG wanted the techniques on the list to capture the majority of techniques routinely used in the NHS. They advised that the techniques in the six population groups below should be evaluated in the model. In each group, the sedative techniques should be compared to general anaesthesia as this is a common alternative to using sedation in the NHS.

The model

The health outcome measure that NICE prefers for cost-effectiveness analysis is quality-adjusted life years (QALYs). It is not likely that the use of sedation techniques will lead to significant differences in QALYs as changes in health related quality of life will only occur over a short period of time. Sedation techniques may be associated with side effects but the GDG suggested that the events observed in the evidence review are not expected to lead to long-term effects that will result in significant QALY differences across different techniques. We therefore carried out a cost-minimisation analysis, that is, we assumed that the quality-adjusted life years would be the same for all treatment strategies.

The success rate of achieving a complete procedure with each technique was not assumed to be equivalent: in the event that a sedation technique fails it is assumed that the procedure would be rescheduled and conducted using general anaesthesia.

We have assessed costs from the perspective of the NHS and personal social services. In economic evaluation it is usual to put a lower weight on costs occurring in the future to reflect both the interest rate and people's time preference – a process known as discounting. However, in the case of this model, all of the included costs occur over a short time horizon and consequently there was no need to discount. The outcome of the analysis was the cost per patient for the whole pathway eventually leading to a successful completion of the procedure, so it includes the cost of the initial procedure plus the cost of any additional procedures required as a result of initial treatment failure.

The cost of sedation includes the time cost of personnel required for the induction and recovery from sedative drug or GA, as well as time cost of the personnel during the procedure. The cost of a strategy also includes the unit cost of drugs for sedation and GA, and the cost of consumables for administering them. We have not included the cost of equipment as it is assumed that these are already available at the point of service delivery and are used for other varied purposes. It would be difficult to estimate the fraction of the cost of equipment attributable to use of sedative drugs or GA.

Some strategies are associated with certain complications and the treatment of complications could result in additional costs.

In the model the expected cost of each strategy is conditional on the strategy's success rate and complication rate as well as the cost of the intervention itself. This can be represented by a decision tree; we present a separate decision tree for each population (see below).

The model was constructed using the best available evidence. Clinical and safety evidence was taken from a systematic review (see chapter 6 on clinical effectiveness and safety review). When the evidence was weak or absent the GDG expert opinion was used to determine the input parameters of the model. The assumptions made in the model and the uncertainties in the input parameters are described explicitly. These were considered by the GDG when interpreting the model results. The impact of uncertainties in the model structure and input parameters were explored through deterministic sensitivity analyses. We did not do a probabilistic sensitivity analysis as the estimates for a number of key input parameters were ascertained by expert opinion. The limitations of the model are discussed.

Cost-effectiveness criteria

The technique with the lowest cost per patient is considered to be the optimal strategy from a cost-effectiveness perspective.

6.2.1. Methods

Decision tree: The decision tree for the five strategies compared in this group is shown below (Figure 115). The use of any of the four sedative drugs (nitrous oxide plus oxygen, nitrous oxide plus sevoflurane, nitrous oxide plus iv midazolam, nitrous oxide plus sevoflurane plus iv midazolam) in a cohort of patients would lead to a successful completion of procedure in some patients. This is described as “success” on the decision tree. In other patients the drug would fail and the procedure would not be completed. In the event that the procedure was not completed, the patient would be given GA on a different occasion to enable the procedure to be completed. The sedative drugs are compared to GA. The GDG suggested that GA leads to completion of procedure in all the patients. Apart from N₂O plus iv midazolam, the GDG assumed that the sedative drugs are associated with vomiting in some patients. They GDG also assumed that the GA strategy is not associated with any complication. The basis for this assumption was that most side effects of GA in children are minor and that many safety measures are in place to minimise the risk of complications. The vomiting event at the branch of the tree for patients who failed to complete the procedure (failure), and who were eventually given GA, reflects the fact that the sedative drug leads to vomiting regardless of whether the procedure is completed (success) or not (failure).

Figure 115

A decision tree of four sedative drugs compared to general anaesthesia in dental procedures in children.

Clinical data on success rate, complication rate and duration: The success rate of sedative drugs and GA are described in Table 1. There were two studies that assessed the use of sevoflurane and nitrous oxide in children^10,23.

Table 1

Success rate of sedative drugs and general anaesthesia in dental procedures in children.

The Lahoud study²³ assessed the efficacy of this drug combination in dental children and the De Sanctis Briggs study¹⁰ assessed the safety in children undergoing MRI. The data on success rate was taken from the Lahoud study and the study has been described fully elsewhere (see chapter 6 the clinical effectiveness). It was an RCT of 411 anxious children undergoing dental procedures randomised to either 0.1 – 0.3% sevoflurane in 40% of N₂O or 40% of N₂O. The group that received sevoflurane plus nitrous oxide had significantly higher completion rate of 89% and this evidence was of moderate quality. There was only one study that assessed the efficacy of nitrous oxide plus sevoflurane plus intravenous midazolam in children³. The study has been described fully elsewhere (see chapter 6 on clinical effectiveness). It was an RCT of 697 anxious children undergoing dental procedures. Study participants were randomised to one of the three arms: 0.3% sevoflurane plus 40% nitrous oxide plus intravenous midazolam, or 40% nitrous oxide plus iv midazolam, or medical air plus intravenous midazolam. The sevoflurane plus nitrous oxide plus midazolam group had a significantly higher completion rate of 93.3% and this was used in the model. The combination strategy, nitrous oxide plus intravenous midazolam was also taken from the Averley study³ and this combination was associated with a higher completion rate of 79.7% when compared to the medical air plus intravenous midazolam group. The evidence from the Averley study³ was of moderate quality. There were a number of RCTs that assessed the efficacy of nitrous oxide and oxygen^{13,34,41,46-50}. The Fauroux study¹³ reported a completion rate but the evidence was low quality. The GDG felt that in clinical practice the patients receiving this sedative drug will have at least 50%. We used the success rate of 52.4% reported in the Lahoud study²³ for patients that received 40% nitrous oxide. The GDG also felt that the patients in the trials are not typical and the selection pattern may not be representative of clinical practice. If patients are assessed and selected for this strategy, success rate could be as high as 95%. We have therefore used 95% in sensitivity analysis. General anaesthesia was assumed to have a success rate of 100%.

The evidence from the systematic review on the timings for induction, procedure and recovery for the sedative drugs and GA was not complete, and when available, it was inconsistent with the GDG's clinical experience. They considered the timings reported in the review and suggested alternative plausible timings to be used in the model and this is shown in Table 2.

Table 2

Timings and vomiting rate for sedative drugs and GA in dental procedures in children.

Vomiting rates were reported in the systematic review but these were also inconsistent and could not be used in a comparative way. We assumed a conservative a rate of 2% should be used for all the sedative drugs.

NHS staff required for application of strategy: The GDG suggested that the following NHS staff would be required during the induction, procedure and recovery phases of different strategies (Table 3. NHS staff required to apply sedative drugs and general anaesthesia in dental procedures in children^*). We used £23 as the cost per hour for a nurse and anaesthetic assistant. This was based on the median full-time equivalent basic salary for “Agenda for Change Band 5 of the October-December 2007 NHS Staff Earnings” estimates for qualified nurses⁴⁰. The rate for consultant dentist and anaesthetist was assumed to be equivalent to the average consultant (physician) earnings at the NHS and we used a rate of £122 per hour⁴⁰.

Table 3

NHS staff required to apply sedative drugs and general anaesthesia in dental procedures in children.

Cost of drugs, consumables and complications: The unit cost of drugs is listed in table 4. We could not identify the cost of nitrous oxide and sevoflurane from the British National Formulary (BNF). The cost of nitrous oxide was estimated at £10 per patient by one of the GDG members using data from their primary care facility, and the additional cost of sevoflurane was £1 per patient. This was for gasses only and excludes the cost of the equipment to deliver the gasses, for scavenging or maintenance. The cost of intravenous midazolam was estimated at £0.87 assuming a maximum dose of 7.5mg (BNF: 5mg/mL, 2mL amp = 58p).

Table 4

Unit cost of drugs used in the model for dental procedures in children.

General anaesthesia was assumed to be induced with propofol and maintained with sevoflurane and nitrous oxide. Induction dose was 2.5mg per kilogram and a child of 25kg would require 62.5mg for induction. This would cost £0.73 (BNF prices: 1% injection (emulsion), 10mg/mL, net price 20-mL = £2.33). Maintenance would be 0.1 – 0.3% sevoflurane in 40% nitrous oxide and this would cost £11. The total cost of GA was therefore £11.73.

The GDG produced a list of consumables required for the administration of sedative drugs and GA. We have included the cost of these in the model. The list is shown below in Table 5 alongside their unit costs. The cost data were taken from the NHS purchase and supply chain catalogue³⁷. Apart from the strategy, nitrous oxide plus oxygen, and nitrous oxide plus iv midazolam, all sedative drugs and GA would require all the consumables listed in the table. The GDG advised that the application of nitrous oxide plus oxygen and nitrous oxide plus iv midazolam would not require intravenous capnography and electrocardiographic electrodes but would require the other consumables in the table. We assumed that the treatment of vomiting would require 30 minutes of nurse's time.

Table 5

Type and unit cost of consumables included in the model for dental procedures in children.

6.2.2. Results

The total cost per patient of each of the five strategies compared in the base case analysis for this population is given in

Table 6. Base case analysis: Cost per patient of different sedation strategies compared with general anaesthesia for dental procedures in children below. N₂O + iv midazolam was the least expensive strategy at £213 per patient.

Table 6

Base case analysis: Cost per patient of different sedation strategies compared with general anaesthesia for dental procedures in children.

Drug costs and consumable costs varied little between strategies. Complication costs were negligible because the incidence was low for all strategies. The biggest component of cost was staff time (especially dentist and anaesthetist time). The cost of second line treatment also varied substantially between strategies, decreasing as the success rate increases.

N₂O+O₂ was more costly in the base case but this was because we had taken a very conservative approach to estimating efficacy (using the rate from a trial of very anxious children, 52%). If instead we assume a success rate of greater than 59% then it becomes cost saving in the model compared to GA – the GDG felt that a rate of 95% was more plausible. Another sedation strategy (Sevoflurane plus nitrous oxide plus iv midazolam) was more expensive than GA regardless of the success rate assumed. This was because it required a sedationist dentist in addition to the operating dentist.

The results of one-way sensitivity analyses are presented in Table 7 Sensitivity analyses on the cost per patient of using different sedation strategies compared with general anaesthesia in dental procedures in children^† below. We started by varying the success rate of the sedative drug strategies to determine the point at which they become cost-saving compared to the GA strategy. For example, the strategy, nitrous oxide plus oxygen, was cost saving as long as the success rate of the sedative drug is equal to or greater than 59%. Sevoflurane plus nitrous oxide plus iv midazolam was not cost-saving even at a success rate of 100%.

Table 7

Sensitivity analyses on the cost per patient of using different sedation strategies compared with general anaesthesia in dental procedures in children.

When the success rate of all sedation techniques was increased to 95% for all strategies, N2O became the lowest cost strategy. Otherwise the results were robust to sensitivity analysis.

6.3.1. Dental procedures in adolescents

Decision tree: The decision tree for the two strategies compared in this group is shown below (Figure 116. A decision tree of iv midazolam compared to general anaesthesia in dental procedures in adolescents). The application of intravenous midazolam in a cohort of patients would lead to successful completion of procedure in some patients. In other patients it would fail and the procedure would be completed using GA as a second line option. This strategy is compared with using GA as a first line option. General anaesthesia leads to completion of procedure in all the patients and is assumed not to be associated with any complications. Intravenous midazolam is associated with oxygen desaturation of less than 90%. The oxygen desaturation event at the branch of the tree for patients who failed to complete the procedure (failure), reflects the fact that intravenous midazolam leads to oxygen desaturation regardless of whether the procedure is completed or not.

Figure 116

A decision tree of iv midazolam compared to general anaesthesia in dental procedures in adolescents.

Clinical data on success rate, complication rate and duration: The success rate of intravenous midazolam and GA are given in Table 8. There was no directly applicable evidence from the review on the success rate for intravenous midazolam. Success rates of 95.2%, 78.9% and 100% were reported in three heterogeneous studies. The first figure was from a study of oral midazolam in children undergoing intravenous insertion²⁷. The second estimate was from a study of intranasal midazolam in children undergoing venipuncture insertion¹⁴. The third estimate was from a study of oral and intranasal midazolam in children undergoing suture and laceration repair⁸. GDG consensus was that a success rate of 95% be used in the model for this group.

Table 8

Success rate of sedative drugs and general anaesthesia in dental procedures in adolescents.

There was no applicable evidence on the duration of the strategies. The GDG considered the existing evidence from the clinical effectiveness review and made timing estimates that reflect their clinical experience. They suggested that the following estimates should be used in the model (Table 9).

Table 9

Timing for sedative drugs and GA in dental procedures in adolescents.

NHS staff required for application of strategy: The GDG suggested that the following NHS staff would be required during the induction, procedure and recovery phases of the two strategies (Table 10). The unit cost of time spent by the nurse, dentist, anaesthetist and anaesthetist assistant has been described above in the section on “NHS staff required for application of strategy” under “Dental procedure in children”.

Table 10

NHS staff required to apply sedative drug and general anaesthesia in dental procedures in adolescents.

Cost of drugs, consumables and complications: The cost of intravenous midazolam and GA used in the model was £0.87 and £11.73 respectively. We have described how these were arrived at in the section on ‘Cost of drugs, consumables and complications’ under ‘Dental procedures in children’. The GDG advised that the application of intravenous midazolam would not require iv capnography and electrocardiographic electrodes but would require the other consumables in Table 5 above. The cost of consumables for intravenous midazolam was estimated at £31, and for GA, £32. The cost of GA includes the cost of all consumables listed above in Table 5. Oxygen desaturation that is less that 90% is a complication associated with midazolam. Some other interventions considered in this economic analysis are also associated with this complication. The GDG decided that this was unlikely to be associated with a treatment cost.

6.3.2. Results

We have compared two strategies in this group and the total cost per patient in the base case analysis for each of them is shown in Table 11 below. Midazolam was less expensive at £248.

Table 11

Base case analysis: Cost per patient of using iv midazolam compared with general anaesthetia in dental procedures in adolescent.

The cost of consumables was similar for both strategies but the cost of drugs was more for the GA strategy. The biggest component of cost was staff time (especially dentist and anaesthetist time).

We have described the results of one-way sensitivity analyses in Table 12 below. The cost per patient of the midazolam remained lower than the cost of the GA as long as the success rate of midazolam is not below 63%. Midazolam remained associated with lower costs for all the sensitivity analyses conducted.

Table 12

Sensitivity analyses on the cost per patient of using iv midazolam compared with general anaesthesia in dental procedures in adolescents .

6.4.1. Methods

Decision tree: The decision tree for the three strategies compared in this group is shown below (Figure 117). The application of intravenous ketamine or intravenous fentanyl plus propofol in a cohort of patients would lead to successful completion of procedure in some patients. In others the drug would fail and the procedure would be completed using GA as a second line option. These strategies are compared to using GA as a first line option. General anaesthesia leads to completion of procedure in all the patients and is assumed not to be associated with complications. Intravenous ketamine is associated with vomiting, and both of the sedative drug strategies compared in this group are associated with hypotension and respiratory complications as well as with oxygen desaturation less than 90%.

Figure 117

A decision tree of two sedative drugs compared to general anaesthesia in short painful procedures.

Clinical data on success rate, complication rate and duration: The success rates of the sedative drugs and GA are described in Table 13. There was no evidence on the appropriate success rate to apply in the model for intravenous ketamine. The GDG was of the view that up to 1% of procedures are not successfully completed under ketamine sedation. They advised that a success rate of 99% should be used in the model. They suggested that the 100% reported in Cechvala 2008⁷ for intravenous fentanyl plus propofol was clinically credible, and this rate was used in the model.

Table 13

Success rate of sedative drugs and general anaesthesia in short painful procedures.

The Cechvala study⁷ was an RCT carried out in 22 children undergoing lumbar puncture for diagnosis of acute leukaemia or lymphoma. It compared intravenous fentanyl (1mcg/kg) plus intravenous propofol (1-2mg/kg/min) plus oxygen supplementation plus topical anaesthesia with placebo (normal saline) plus intravenous propofol (1-2mg/kg/min) plus oxygen supplementation plus topical anaesthesia. All study patients completed the procedure and this evidence was judged as moderate quality. General anaesthesia was assumed to have a success rate of 100%. Vomiting and oxygen desaturation rate less than 90% were reported for ketamine in several heterogenous studies included in the systematic review of efficacy and the GDG advised that a rate of 6.65% for vomiting and 0.9% for oxygen desaturation rate less than 90% should be taken from the study with the largest sample size¹⁶. They also suggested from their clinical experience that ketamine would be associated with up to one percent rate of hypotension and respiratory intervention. Hypotension and respiratory intervention rate of 18% was reported in only the Cechvala study⁷ for intravenous fentanyl plus propofol, and this rate was used in the model. The rate of oxygen desaturation less than 90% was reported as 5% in one study⁴. These studies have been described in the sections on the efficacy and safety of sedation techniques.

After considering the limited evidence from the review the GDG provided the following estimates as the timings for the three strategies (Table 14).

Table 14

Timings and vomiting rate for sedative drugs and GA in short painful procedures.

NHS staff required for application of strategy: The GDG suggested that the following NHS staff would be required during the application of the three strategies compared here (Table 15). The unit cost of the time spent by the personnel has been described above (dental procedure in children).

Table 15

NHS staff required to apply sedative drug and general anaesthesia in short painful procedures.

Cost of drugs, consumables and complications: We assumed a median dose of 30mg for ketamine⁴². This would cost £0.76 (BNF: 10mg/mL, 20-mL vial = £5.06). The dosage in Cechvala 2008⁷ for intravenous fentanyl was 1mcg/kg. For a 25 kg child requiring 25mcg, it would cost £0.14 (BNF: 50mcg/mL, net price 2-mL amp = 54p). The dosage for propofol in Cechvala 2008⁷ was 1-2mg/kg/min infusion. We assumed that 25kg child would require 38mg for one minute. The child would require about 4mL which would cost £0.46. (BNF: 1% injection (emulsion), 10mg/mL. net price 20-mL amp = £2.33). The total cost of administering this combination therapy would therefore be £0.60. The cost of GA used in the model was £11.73, and the cost of consumables for all strategies was £32. A description of how these were arrived at has been given above (dental procedure in children). The cost of consumables includes the cost of all consumables listed above in Table 5.

Oxygen desaturation that is less than 90% is a complication associated with the sedative drugs compared in this group but there would be no additional treatment cost for this. We assumed that 30 minutes of nurse's time would be required both for the treatment of vomiting and for hypotension and respiratory interventions.

6.4.2. Results

The average cost of the strategies compared in this model population in the base case analysis is given below in Table 16. Ketamine was the least expensive strategy at £155, and GA was the most expensive strategy at £224.

Table 16

Base case analysis: Cost per patient of using sedation strategies compared with general anaesthesia in short painful procedures.

The cost of consumables for the three strategies was the same but the cost of the GA drugs was higher than the cost of the sedative drugs. The highest cost component was the cost of staff time, particularly the cost of physician and anaesthetist time. Fentanyl plus iv midazolam was actually more expensive than ketamine because it required a sedationist dentist in addition to an operating dentist for its administration.

The complication costs associated with ketamine were low because of low incidence while the cost of complications associated with fentanyl plus propofol was slightly higher because of higher incidence.

The results of one-way sensitivity analyses are presented in Table 17 below. We varied the success rate of the sedative drug strategies to determine the point at which they become cost-saving compared to GA strategy. Ketamine remained cost saving as long as the success rate of using it is not below 69%. The combination drug, fentanyl plus propofol remained cost-saving as long as the success rate of the drug combination is not below 95%.

Table 17

Sensitivity analyses on the cost per patient of using different sedation strategies compared with general anaesthesia in short painful procedures .

Ketamine remained the cost-saving compared with the other strategies when the GA induction time is 15 minutes or the cost of nitrous oxide is £5. Unlike ketamine, the other two strategies require physician sedationist in addition to operating physician and this makes it less expensive. When we assumed that sedation was administered by a nurse, fentanyl plus propofol became cost-saving when compared with ketamine and GA.

6.5.1. Methods

Decision tree: The decision tree for the two strategies compared in this group is shown below (Figure 118). The use of high dose chloral hydrate as a sedative drug in a cohort of patients would lead to successful completion of procedure in some patients, and in others it would fail. In the event of failure, the procedure would be completed using GA as a second line treatment option. This strategy is compared to using GA as a first line option to enable completion of procedure. General anaesthesia is assumed to lead to completion of procedure in all the patients and would not to be associated with any complication. High dose chloral hydrate is associated with vomiting.

Figure 118

A decision tree of chloral hydrate compared to general anaesthesia in painless imaging procedures.

Clinical data on success rate, complication rate and duration: The success rate of oral chloral hydrate was reported in two studies^18,32. The Marti-Bonmati study³² was carried out in children undergoing MRI and the Houpt study¹⁸ was in children undergoing dental procedure. The GDG felt that the success rate reported in the former study should be used as it is a more applicable study for this model group. The Marti-Bonmati study³² has been described before in the section on clinical effectiveness and safety. In the study, high dose chloral hydrate (96mg/kg) was compared to intermediate dose (70mg/kg). It was reported that high dose chloral hydrate had a completion rate of 100% and we have used this rate in the model. The study was judged to be of moderate quality. We have assumed the success rate of GA to be 100%.

Table 18

Success rate of sedative drugs and general anaesthesia in painless imaging procedures.

After considering the evidence on the timings reported in the review the GDG suggested that it would be more clinically realistic to use the following timings in the model.

Table 19

Timing for sedative drugs and GA in painless imaging procedures.

NHS staff required for application of strategy: The GDG also suggested that the following NHS staff would be required during the different phases of applying the two strategies (Table 20). The unit cost of time spent by the personnel has been described above (dental procedure in children). We used £29 as the cost per hour for a radiographer. This was based on the median full-time equivalent basic salary for “Agenda for Change Band 5 of the October-December 2007 NHS Staff Earnings” estimates⁴⁰.

Table 20

NHS staff required to apply sedative drug and general anaesthesia in painless imaging procedures*.

Cost of drugs, consumables and complications: The maximum dose of chloral hydrate in the BNF is 2g (BNF, cloral betaine 707mg (=chloral hydrate 414mg): net price 30-tab pack =£7.90). A maximum of five tablets would cost £1.32. The cost of GA used in the model was £11.73 and we have described elsewhere how we arrived at this figure (dental procedure in children). The cost of consumables for each of the two strategies compared here was £32. This included the cost of all consumables listed above in Table 5. The treatment cost of vomiting was assumed to be equivalent of 30 minutes of nurse's time.

6.5.2. Results

We compared two strategies in this population and the result of the base case analysis showed that GA was less expensive at £224 than high dose chloral hydrate (Table 21). This was not surprising as the administration of the sedative drug requires a physician unlike the administration of GA.

Table 21

Base case analysis: Cost per patient of high dose chloral hydrate compared with general anaesthesia in painless imaging.

The highest cost component of these strategies remained the cost of staff time especially physician and anaesthetist time. The cost of complication was low because of low incidence. The cost of consumables for the two strategies was the same but the cost of GA drugs was higher.

The results of one-way sensitivity analyses are presented in Table 22 below. We changed the success rate of high dose chloral hydrate and, at 95% this strategy was even more expensive. Other results of the sensitivity analysis suggest that the GA strategy would be associated with less cost. The sedative drug strategy became less expensive only when the nurse was the only personnel that will apply the sedative drug.

Table 22

Sensitivity analyses on the cost per patient of using high dose chloral hydrate compared with general anaesthesia in short painless imaging .

6.6.1. Methods

Decision tree: We compared intravenous midazolam and GA and the decision tree is the same as the one used to compare intravenous midazolam and GA in dental procedures in adolescents (Figure 116. A decision tree of iv midazolam compared to general anaesthesia in dental procedures in adolescents). The use of intravenous midazolam in a cohort of patients would lead to a successful completion of the procedure in some patients but would fail in others. In the patients where it failed, GA would be used to complete the procedure. The use of GA as a first line option would lead to completion of procedure in all patients. Intravenous midazolam is associated with oxygen desaturation level less than 90% and GA is assumed not to be associated with complications.

Clinical data on success rate, complication rate and duration: There was no directly applicable evidence from the review on the success rate for intravenous midazolam in patients undergoing oesophago-gastroscopy. Indirect evidence from three heterogeneous studies was considered by the GDG^8,14,27. The first study was on oral midazolam in children undergoing intravenous insertion, and reported a success rate of 95.2%. The second was on intranasal midazolam in children undergoing venipuncture insertion, and reported a rate of 78.9%. The last study was on oral and intranasal midazolam in children undergoing suture and laceration repair, and reported a rate of 100%. The GDG agreed that a rate of 95% be used in the model. A success rate of 100% was used for GA. There was also no directly applicable evidence on the duration of the strategies for this group. The GDG considered other estimates reported in the review and made timing estimates that reflect their clinical experience. They suggested the estimate in the table below should be used (Table 23).

Table 23

Timings for sedative drugs and GA in oesophago-gastroscopy.

NHS staff required for application of strategy: The GDG suggested that the following NHS staff would be required during the application of the strategies (Table 24). The unit cost of the time spent by the staff is described above (dental procedure in children).

Table 24

NHS staff required to apply sedative drug and general anaesthesia in oesophago-gastroscopy .

Cost of drugs, consumables and complications: The cost of intravenous midazolam and GA used are £0.87 and £11.73 respectively and a description of how we arrived at these estimates is given above (dental procedure in children). The cost of consumables for the two respective strategies is £31 and £32. The GDG advised that the application of intravenous midazolam would not require intravenous capnography and electrocardiographic electrodes but would require the other consumables in Table 5 above. The cost of consumables for GA includes the cost of all consumables listed above in Table 5. Oxygen desaturation less than 90% would not be associated with additional treatment cost.

6.6.2. Results

There were two strategies compared in this population and the total cost per patient in the base case analysis is given in Table 25 below. Midazolam was less expensive at £122.

Table 25

Base case analysis: Cost per patient of using iv midazolam compared with general anaesthesia in oesophago-gastroscopy.

The cost of consumables was similar but drug cost was higher for GA. The highest cost component was cost of staff time particularly physician and anaesthetist time.

The results of one-way sensitivity analyses are described in Table 26 below. The cost per patient of the iv midazolam strategy remained lower than the cost of the GA strategy as long as the success rate of midazolam strategy is not below 75%. The midazolam strategy remained associated with lower costs for all the sensitivity analyses conducted.

Table 26

Sensitivity analyses on the cost per patient of using iv midazolam compared with general anaesthesia in oesophago-gastroscopy .

6.7.1. Methods

Decision tree: The decision tree that was used for the model for this group is shown below (Figure 119). The use of the combination technique, intravenous midazolam plus intravenous fentanyl in a cohort of patients would lead to a successful completion of the procedure in some patients but would fail in others. In the patients where it fails, GA would be used to complete the procedure. The use of GA as a first line option would lead to completion of procedure in all patients. The combination technique is associated with vomiting and oxygen desaturation less than 90%.

Figure 119

A decision tree of a combination sedation technique compared to general anaesthesia in colonoscopy.

Clinical data on success rate, complication rate and duration: There was no directly applicable study in the systematic review that reported the success rate for this drug combination. Indirect evidence from one study was considered²⁹. The study compared intravenous fentanyl plus midazolam with intravenous midazolam plus ketamine in 57 children undergoing placement of intravenous line. All patients were reported to have completed the procedure. The consensus was that a rate of 95% is a clinically realistic rate and should be used in the model. A success rate of 100% for GA was assumed. There were a number of heterogeneous studies on the safety of the combination sedation option and the GDG advised that we use rates from the study with largest sample size. A rate of 5.22% was reported for vomiting³⁸, and 2.56% for oxygen desaturation less than 90%³¹.

There were no directly applicable timing estimates for the strategies and the following estimates were made based on the clinical experience of the GDG (Table 27).

Table 27

Timings for sedative drug and GA in colonoscopy.

NHS staff required for application of strategy: The following NHS staff in Table 28 below was suggested by the GDG to be required for the application of the strategies. The unit cost of time spent by the personnel has been described above (dental procedure in children).

Table 28

NHS staff required to apply sedative drug and general anaesthesia in colonoscopy.

Cost of drugs, consumables and complications: The cost of midazolam plus fentanyl was estimated based on the dosage reported in Lucas da Silva 2007²⁹ (midazolam, 0.15mg per kg; fentanyl, 1μg per kg). We assumed a maximum dose of 7.5mg reported in the BNF for midazolam which would cost £0.87. For a child 25kg, 25μg fentanyl would cost £0.14 (BNF for fentanyl: 50mcg/mL, net price 2-mL amp = 54p; BNF for midazolam: 5mg/mL, 2mL amp = 58p, 7.5mg would cost 87p). The total cost of this drug combination used in the model was therefore £1.01. The cost of GA was £11.73 and we have described how we arrived at this (dental procedure in children). The cost of consumables for each of the respective strategies was £32. This includes the cost of all consumables listed above in Table 5. The treatment cost of vomiting was assumed to be equivalent of 30 minutes of nurse's time.

6.7.2. Results

The total cost per patient for each of the two strategies compared in this population in the base case analysis is given in Table 29 below. The combination strategy, iv midazolam plus fentanyl, was less expensive at £215.

Table 29

Base case analysis: Cost per patient of using iv midazolam plus fentanyl compared with general anaesthesia in colonoscopy.

The cost of GA drug was higher but the cost of consumables for both strategies was the same. The greatest cost component was the cost of staff time especially anaesthetist and physician time. The cost of complication was low because of low incidence.

We have described the results of one-way sensitivity analyses in Table 30 below. We varied the success rate of the combination strategy to determine the point at which it becomes cost-saving compared to GA strategy. The combination strategy is cost saving as long as the success rate of using it is equal to or greater than 68%. The combination strategy remained cost saving compared to the GA strategy for all the sensitivity analyses conducted here.

Table 30

Sensitivity analyses on the cost per patient of using iv midazolam plus fentanyl compared with general anaesthesia in colonoscopy .

Martinez 2002³³

Martinez 2002³³ was a randomised double blind study comparing diazepam with midazolam as a premedication administered in conjunction with meperidine prior to procedural sedation with propofol in children having upper endoscopy. It is considered to be a partial economic evaluation as the only costs reported were the costs of the study drugs themselves which was $25.95 for midazolam and $0.92 for diazepam. It is therefore not useful for decision making as it does not estimate the overall resource use and costs of the alternative sedation strategies. For example, it does not consider the cost of treating adverse events.

Iannalfi 2005¹⁹

Iannalfi 2005¹⁹ was a randomised controlled trial comparing moderate sedation with general anaesthesia in children having lumbar puncture and/or bone marrow aspiration. It only enrolled 31 children and therefore there were less than 20 patients in each arm. RCTs with less than 20 patients in each arm are excluded from the clinical effectiveness reviews as the groups are not sufficiently large for randomisation to provide groups who are reliably comparable for known and unknown confounders. We have therefore not considered it any further as the clinical effectiveness outcomes are potentially open to bias.

Lee 2000²⁶ and Jameson 2007²⁰

These two studies were model based cost minimisation studies which estimated the cost per patient treated and assumed that the health benefits would be equivalent^20,26. In both cases the studies compare sedation with anaesthesia for patients undergoing dental treatment. After considering the clinical review evidence, the GDG agreed that it is not likely that the use of sedation techniques will lead to significant changes in quality-adjusted life years as changes in health-related quality of life will only occur over a short period of time. The GDG also suggested that the adverse events observed in the clinical review are not expected to lead to long-term effects that will result in significant QALY differences across different techniques. However, the results of these studies could not be used as the GDG wanted to compare four different sedation strategies with GA in children undergoing dental procedure.

Pershad 2006³⁹

The final model based evaluation³⁹ used clinical evidence from RCT and non-RCT sources to compare four different procedural sedation and analgesia (PSA) techniques for use in children requiring manipulation of a forearm fracture in the emergency department (ED). The four techniques were:

• Deep sedation with ketamine / midazolam (K/M)
• Deep sedation with propofol / fentanyl (P/F)
• Deep sedation with fentanyl / midazolam (F/M)
• Axillary approach to brachial plexus regional block with midazolam premedication (ABRA/M)

The model incorporated evidence on adverse event rates, duration of sedation, and likelihood of PSA failure. The clinical effectiveness and adverse effects data were derived from published literature following a systematic literature search, but the methods for selecting papers has not been explicitly reported. Some additional data from an unpublished trial undertaken in the author's institution were also incorporated in the analysis. The methods described in the paper suggest that the estimates obtained from the RCTs were synthesised in a way which did not maintain randomisation. The adverse events considered in the model were emesis, recovery agitation, respiratory depression requiring assisted ventilation and lidocaine toxicity. It was assumed that deep sedation with P/F would be used when axillary block failed. It was assumed that deep sedation would be 100% successful for all three techniques based on existing data showing that success rates are between 98% and 100% with K/M and F/M.

Resource use included medication costs for sedation and analgesia techniques, staffing costs for administering sedation and treating adverse events, and ED overhead costs based on duration of ED stay which was assumed to vary according to the total sedation time. Duration of ED stay was used as the clinical effectiveness outcomes so that the cost-effectiveness was reported as the cost per hour of time in the ED avoided. Unit costs were reported for staff time, ED overheads and medication costs. Costs were calculated from the hospital's perspective and were reported in US$, but the price year was not reported. Uncertainty was examined deterministically using one-way and two way sensitivity analysis. A probabilistic sensitivity analysis was used to consider the importance of parameter uncertainty but the authors simply report that the model was “robust” through 1000 iterations.

P/F was found to be the dominant strategy as it had the lowest cost and the shortest ED stay which was the sole effectiveness outcome considered. However this conclusion was sensitive to several key assumptions. The conclusions would be different if the rate of respiratory depression for P/F were to increase from 1.1% to 6.9%, if the rate of lidocaine toxicity were to be reduced from 2.5% to less than 1%, or if the rate of failure of axillary block were to be reduced from 6.8% to less than 2%. Small increases (e.g 3 mins) in the duration of physician time required to administer deep sedation would result in axillary block being the lowest cost option, which is quite possible given that this duration was not well defined by the evidence base. This economic evaluation is considered to be only partially applicable as it is a US based study and the assumptions regarding resource use and unit costs that have been used to populate the model may not be relevant in a UK NHS setting. It is also not clear whether the PSA regimens compared are equivalent in terms of reducing pain and discomfort for patients or whether the main outcome measure, length of emergency department stay, is an important outcome for patients and their families and carers. It is considered to have potentially serious limitations due to uncertainty around the selection and synthesis of effectiveness data and the sensitivity of the conclusions to key assumptions regarding physician time.

Blain 1998⁵

This costing study compares the cost of inhaled sedation (nitrous oxide in oxygen, titrated up to a maximum of 40%) with local anaesthesia to general anaesthesia (intravenous induction with inhalational maintenance) for children having dental extractions from a UK NHS perspective. Treatment was provided in a UK secondary care setting. The costing analysis was restricted to staffing costs during treatment and recovery. If treatment took place over more than one visit then the total duration over multiple visits was used. Staff costs were based on the agreed minimum staffing level for each service and 1994 salary scales. These were used to calculate the ratio of staff costs per minute during treatment and recovery for the two services and overall costs were reported using units that represent one minute of care within the sedation service (see Table 31 below). The duration of treatment and recovery was taken from a case-control study conducted in the UK which was also reported within Blain 1998⁵. Children who were not suitable for treatment with sedation were excluded from both the sedation and anaesthesia cohorts before 265 matched pairs (matched for age and gender) were selected. The mean age was 7.63 (SD 2.45) and 7.54 (SD 2.46) for the sedation and anaesthesia groups respectively. However, there were a much larger number of patients rejected from the sedation group (42% versus 16%) suggesting that the groups may not be comparable. The overall costs were 64.3 units for sedation and 80.8 units for anaesthesia. It is not possible to convert these back to UK£ from the data provided. This study is directly applicable as it takes a UK NHS perspective although its usefulness is limited as it does not report the actual costs and therefore these cannot be uplifted to reflect current prices. The duration of treatment and recovery are key factors in the costing analysis and these have potentially serious limitations as they are based on a case-control study, in which there were considerably more patients excluded from one group.

Table 31

Staffing levels, cost ratios and duration of treatment and recovery associated with sedation and general anaesthesia.

Shaw 1996⁴³

This was a prospective study that evaluated treatment success, assessed parents' and children's satisfaction, and compared the cost of inhalation sedation with that of existing general anaesthesia. It was carried out in children having dental extractions or minor oral surgery in a UK NHS secondary care setting. Treatment was judged as successful by the clinician if the procedure was completed. Data on treatment satisfaction was collected by questionnaire. Cost was based on hospital data and included staff cost only. It excluded the cost of other hospital overheads, such as the equipment, anaesthetic gases and reception staff. Ninety percent of children treated with sedation completed treatment. Thirteen children were treated with general anaesthesia. The cost per patient of providing treatment with sedation was reported to be 30% less than that for outpatient general anaesthesia and 57% less than day-stay general anaesthesia. More detailed cost information was not reported. This study has a number of limitations and should be cautiously interpreted. The number of patients studied for general anaesthesia was small. Cost data included only staff cost and this was not reported in enough details to allow judgement on quality. The study sample was not randomised. There were no sensitivity analyses on the results.

Jameson 2007²⁰

This paper compares the cost of providing advanced conscious sedation in a primary care-based service with the cost of treatment under a dental general anaesthetic (DGA) in a hospital based community dental service. The cost analysis for advanced conscious sedation takes into account the rate of referrals for DGA after initial assessment and the rate of sedation failure, which are estimated from 2,771 patient records. The rate of failure under DGA is not considered and is therefore assumed to be 100%.

The cost of treatment under DGA is presented using both NHS reference costs¹² and a bottom-up costing using local audit data. The bottom-up costing included salary costs for anaesthetists, dental staff and administration staff and the cost of consumables, equipment, portering and the availability of inpatient beds reserved for use by the service. Separate costs were estimated for long and short procedures and an average cost was derived using weighting list data to estimate the ratio of long to short procedures. Using the HRG costs, the cost for short and long procedures was £568 and £616 respectively, with a mean cost of £590.21. The average cost estimate based on the local audit data was much lower at £359.91.

The cost of treatment under sedation was estimated using the patient list data from 205 patients and applying the relevant fees paid to the primary care based sedation service by the NHS, giving a cost per patient of £223.78. Once the additional cost of referring patients who had failed under sedation for a DGA are included, the cost is £245.57 per patient treated.

Sensitivity analyses were conducted on the rate of sedation failures, the rate of referrals for DGA following sedation failure and the rate of referrals for DGA following assessment. The rate of failure would need to increase to 77% before DGA became the lowest cost option, whilst the rate of referral following failure was not found to be a significant factor. If the rate of referrals following assessment at the sedation service were to increase to above 36.32% then DGA would be the lowest cost option, however the current rate is only 4-5%.

It is not clear whether the patients receiving care under the two services are similar. It is not known whether the age profile of the two cohorts was similar or how many patients receiving DGA had special needs meaning that they would not be able to receive treatment in a primary care setting. The fact that 56.7% of those failing under sedation (1.98% of all those receiving sedation) were referred back to their GP as there was insufficient justification for a DGA suggests that the cohorts may not be comparable. This study is considered to have minor limitations as there is uncertainty regarding the comparability of the cohorts being treated in the different settings, but the sensitivity analyses suggest that the conclusions are unlikely to be affected by small differences in the case mix. The results are considered to be directly applicable to the UK NHS dental services as a whole with the caveat that there would need to be sufficient demand within a particular region to meet the upfront costs of establishing a primary care based sedation service such as this as an alternative to DGA.