The cost-effectiveness of physician staffed Helicopter Emergency Medical Service (HEMS) transport to a major trauma centre in NSW, Australia

Injury, Int. J. Care Injured 43 (2012) 1843–1849

Contents lists available at SciVerse ScienceDirect

Injury journal homepage: www.elsevier.com/locate/injury

The cost-effectiveness of physician staffed Helicopter Emergency Medical Service (HEMS) transport to a major trauma centre in NSW, Australia Colman Taylor a,b,*, Stephen Jan a, Kate Curtis a,c,d, Alex Tzannes d,e, Qiang Li a, Cameron Palmer g,h, Cara Dickson d, John Myburgh a,d,f a

The George Institute for Global Health, Australia Sydney Medical School, The University of Sydney, Australia Sydney Nursing School, University of Sydney, Australia d St George Hospital, Australia e The Ambulance Service of NSW, Australia f University of NSW, Faculty of Medicine, Australia g Trauma Service, The Royal Children’s Hospital Melbourne, Australia h Department of Epidemiology & Preventive Medicine, Monash University, Melbourne, Australia b c

A R T I C L E I N F O

A B S T R A C T

Article history: Accepted 19 July 2012

Background and context: Helicopter Emergency Medical Services (HEMS) are highly resource-intensive facilities that are well established as part of trauma systems in many high-income countries. We evaluated the cost-effectiveness of a physician-staffed HEMS intervention in combination with treatment at a major trauma centre versus ground ambulance or indirect transport (via a referral hospital) in New South Wales (NSW), Australia. Methods: Cost and effectiveness estimates were derived from a cohort of trauma patients arriving at St George Hospital in NSW, Australia during an 11-year period. Adjusted estimates of in-hospital mortality were derived using logistic regression and adjusted hospital costs were estimated through a general linear model incorporating a gamma distribution and log link. These estimates along with other assumptions were incorporated into a Markov model with an annual cycle length to estimate a cost per life saved and a cost per life-year saved at one year and over a patient’s lifetime respectively in three patient groups (all patients; patients with serious injury [Injury Severity Score > 12]; patients with traumatic brain injury [TBI]). Results: Results showed HEMS to be more costly but more effective at reducing in-hospital mortality leading to a cost per life saved of $1,566,379, $533,781 and $519,787 in all patients, patients with serious injury and patients with TBI respectively. When modelled over a patient’s lifetime, the improved mortality associated with HEMS led to a cost per life year saved of $96,524, $50,035 and $49,159 in the three patient groups respectively. Sensitivity analyses revealed a higher probability of HEMS being costeffective in patients with serious injury and TBI. Conclusion: Our investigation confirms a HEMS intervention is associated with improved mortality in trauma patients, especially in patients with serious injury and TBI. The improved benefit of HEMS in patients with serious injury and TBI leads to improved estimated cost-effectiveness. ß 2012 Elsevier Ltd. All rights reserved.

Keywords: Helicopter Emergency Medical Services Primary scene response Cost-effectiveness Health economics Trauma care Trauma system

Background and context Helicopter Emergency Medical Services (HEMS) are an established component of many health systems, particularly in highincome countries. Their use may be broadly grouped into two

* Corresponding author at: P.O. Box M201, Missenden Rd, Camperdown, NSW 2050, Australia. Tel.: +61 2 9657 0300. E-mail address: [email protected] (C. Taylor). 0020–1383/$ – see front matter ß 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.injury.2012.07.184

response types, primary responses (direct to the scene of an incident) and secondary inter-facility transfers (of trauma and other critically ill patients). In primary scene responses, HEMS have several potential advantages for trauma patients over usual care provided by ground transport networks. These include access to areas where road infrastructure is limited, faster transport to hospital in rural areas1 and faster access to definitive care through the provision of advanced interventions (such as endotracheal intubation or fluid resuscitation) by a physician or trained paramedic at the scene. It is hypothesised that the advantages

C. Taylor et al. / Injury, Int. J. Care Injured 43 (2012) 1843–1849

1844

provided by HEMS in primary scene responses may improve the chances of survival for seriously injured trauma patients in comparison to care provided by ground ambulance networks. Previous systematic reviews2,3 and annotated reviews4–8 suggest that HEMS may improve trauma patient mortality in comparison to patients transported by ground ambulance. However, clear differences between studies exist in the context in which the HEMS operates, the staffing of HEMS and ground ambulance comparators, the patient population used as well as the methods used in the evaluation. As a healthcare intervention, HEMS are an expensive alternative compared to ground transport and any potential benefits provided by HEMS need to be considered in context of the associated costs. Few studies have evaluated the cost-effectiveness of HEMS in trauma and due to regional variability previous results are likely to be system specific.9 In the New South Wales (NSW), Australia, HEMS operate under unique geographical conditions which include large distances servicing a sparse population. In this environment, the predominant staffing model for HEMS is physicians trained in specialties such as emergency medicine and anaesthesia with qualified paramedics.10 Despite results supporting the use of physicianstaffed models of trauma care11,12 (which are predominantly delivered via HEMS), the benefit and value of physician staffed HEMS in NSW is unknown. In the context of scarce resources and competing alternatives for government funding there is an imperative to investigate the cost-effectiveness of physicianstaffed HEMS primary scene responses to trauma.

Setting In NSW, eight HEMS currently operate from various locations around the state, performing primary scene responses and secondary inter-facility transfers as part of the NSW state trauma plan.10 In the Sydney metropolitan area, HEMS physicians and paramedics can also travel via road to incidents in close proximity to the base of operations. Patients meeting major trauma criteria13 are transported to designated major trauma centres which have the full spectrum of care available including surgery, radiology, intensive care and rehabilitation.14 St George Hospital is a designated major trauma hospital located in south eastern Sydney that receives trauma patients from its catchment area transported by the following modes: paramedic-crewed ground ambulance; physician-staffed HEMS or ground ambulance (90% via HEMS); fixed-wing aircraft; international retrievals; other transport modes such as private vehicles and walk-ins.

HEMS + treatment at major trauma centre Paents meeng SGH trauma criteria

Aim The aim of this health economic investigation was to evaluate the cost-effectiveness of direct physician-staffed transport (termed ‘HEMS’) from the scene to a level one trauma centre, for adult trauma patients versus a ground transport or indirect transport (via a referral hospital) alternative (termed ‘non-HEMS’). Methods This study was approved by the South Eastern Sydney and Illawarra Area Health Service Human Resources Ethics Committee. The evaluation was undertaken from the perspective of the health care funder. A Markov model incorporating two health states (alive or dead) and an annual cycle length was developed. Fig. 1 describes the patient event pathway model; patients surviving the indexhospitalisation were modelled over their lifetime. The Injury Severity Score (ISS) is an anatomical scoring system that provides an overall score for patients with multiple injuries (range: 1–75, with higher scores associated with higher mortality). ISS combines the Abbreviated Injury Scale injury scores (AIS; range: 1–6), which are assigned across six body regions. The Trauma Score (TS) combines the patient’s initial Glasgow Coma Score (GCS), capillary refill and respiratory effort (range: 1–16, with lower scores associated with higher mortality). Based on previous research that suggests a HEMS intervention is associated with a higher probability of survival following serious injury(s)15 or traumatic brain injury (TBI),16 we calculated estimates in all patients meeting St George Hospital trauma triage criteria (Model #1; Appendix I shows trauma criteria) as well as patient subgroups with serious injury (Model #2) and traumatic brain injury (TBI) (Model #3). Serious injury was defined according to the NSW definition of major trauma which included an ISS greater than 12.14 Study sample Model event probabilities were derived from trauma patients arriving at St George Hospital during a reference period of 11 years from January 2000 to December 2010. Patients were included in the cohort if they sustained any form of trauma during the reference period and met the St George Hospital trauma triage activation criteria (Appendix I). ISS values calculated using different AIS versions were adjusted using AIS mapping tools.17 The following patients were excluded from the analysis: paediatric patients (16 years), patients who arrived at St George Hospital greater than 24 h after their injury, patients not arriving by

Survive injuries

Survive Die

Die Survive

Non-HEMS + treatment at major trauma centre

Survive injuries

Die

Fig. 1. Patient event pathways in model.

Die

C. Taylor et al. / Injury, Int. J. Care Injured 43 (2012) 1843–1849

ambulance or helicopter and patients who were transported from outside the St George Hospital catchment area (including 5 regional hospitals which is reflective of where primary scene responses transferred to St George Hospital emanate from). For this evaluation, a HEMS patient included patients receiving a physician intervention pre-hospital and then being directly transported (from the scene by road or HEMS) to St George Hospital. HEMS patients were retrospectively identified by the transport mode on arrival at hospital as well as ambulance case sheet numbers. A non-HEMS patient included patients arriving at hospital via a paramedic-staffed ground ambulance or via a referral hospital (including regional hospitals and urban non-designated trauma centres). Estimating effectiveness As part of a sequential analysis we estimated the adjusted probability of survival until hospital discharge for HEMS and nonHEMS patients. Given the imbalance in patient illness severity between the HEMS and non-HEMS patient groups, we calculated adjusted estimates of in-hospital mortality using logistic regression. Based on clinical judgement and previous research,2 the following covariates were used in the final model: age (4 levels), sex, year of arrival to hospital (11 levels), mechanism of injury (5 categories), use of endotracheal intubation pre-hospital, scene Trauma Score, TBI, serious injury (ISS > 12 in Model #1; ISS used as continuous variable in Models #2 and #3), rural status, admission to ICU, emergency operation performed. TBI was defined as an anatomical injury to the head (AIS codes beginning with 1 coded to the ISS head region) with an AIS score  3. Patients were considered ‘rural’ if the postcode of injury was greater than 1 h estimated travel time via road. Emergency operation was defined as patients being transferred directly from the emergency department to the operating room. All co-variates were entered into the model along with the patient group variable (HEMS and non-HEMS). Model discrimination was tested through the receiver operating characteristic [ROC] area and calibration through the Hosmer–Lemeshow (H–L) statistic. To model survival over a lifetime we incorporated the estimated probability of in-hospital mortality for HEMS and non-HEMS patients into a Markov model with an annual probability of death. In Model #1 we used the probability of death published by the Australian Bureau of Statistics18 (based on the average age of the cohort) that assumes that patients have a normal life expectancy beyond the index hospitalisation. In Models #2 and #3 we adjusted the probability of death for the higher risk of death following traumatic brain injury19 (SMR: 1.51; 95% CI: 1.25–1.78).

1845

intensive care, allied health and rehabilitation) incurred by each patient in a state-wide uniform database (Trendstar Decision Support System1). Patient costing, including overheads (e.g. staffing, floor space) was conducted in accordance with 2008–2009 NSW Program and Product Data Collection Standards.21 For patients who were discharged from emergency department (and therefore patient level costs were unavailable) we applied the average cost of an emergency department presentation ($396 [2009 AUD]).22 Costs for hospital separations that crossed financial years were excluded from the analysis due to inappropriate allocation of DRG cost weights. To calculate a unit cost of the index hospital treatment for HEMS and non-HEMS patients we estimated a predicted cost using a generalised linear model, incorporating a gamma distribution and log link.23 Total cost was adjusted for differences in patient case mix between transport groups using the same covariates identified in the analysis of effectiveness. Residuals were examined to assess model fit and identify potential outliers, with sensitivity analyses performed excluding any identified outliers to assess their impact on the predicted costs. To incorporate pre-hospital transport costs, we utilised the average cost estimate per case for emergency road missions ($546 per case [2006 AUD]) and helicopter missions published by the Independent pricing and Regulatory Tribunal of NSW ($5786 per case [2006 AUD]).24 For the HEMS transport group we calculated an adjusted cost according to the proportion of patients arriving via ground and HEMS. In the non-HEMS transport group, we adjusted the transport cost according to the proportion of inter-hospital transfers and the transport mode utilised during inter-hospital transfer. To estimate cost of treatment at a referral hospital, we applied the standard cost of an emergency department presentation ($396 [2009 AUD]).22 To account for transport to the referral hospital we added the average cost of ground transport as estimated above. This is based on the assumption that patients arriving via a referral hospital (within 24 h post injury) would have been transferred to the referral hospital via ground transport and would have spent the majority of their time in the referral hospital Emergency Department before being transferred to St George Hospital. Post discharge, we assumed an average rate of annual patient healthcare expenditure over a lifetime for survivors in Model #1, based on the Australian Institute of Health and Welfare Report (estimated government contribution: $3617 per annum [2009 AUD]).25 For patients with serious injury or TBI (Models #2 and #3) we assumed an annual health expenditure that was five-fold higher per year over a lifetime based on the government contribution to the financial cost of traumatic brain injury in Australia (estimated government contribution: $18,008 per annum [2009 AUD]).26

Estimating costs Sensitivity analyses Treatment costs were estimated in four phases of care: treatment in a referral hospital, transport to a trauma centre, treatment during the index hospitalisation and treatment following discharge. Treatment costs during the index hospitalisation were calculated on a per patient basis and averaged whereas estimates were sourced for the other phases of care. To adjust for differential timing all cost information was adjusted to 2010 equivalent values using the health specific consumer price index published by the Australian Bureau of Statistics.20 To estimate index hospitalisation costs, medical record numbers and dates of admission from the trauma data were provided to the St George Hospital Casemix Unit, with a report query template to ensure the same data recovery. These units record the actual costs (including emergency department, theatre,

We undertook multiple one and two-way sensitivity analyses to assess the effect of varying key assumptions across plausible ranges. Age at injury was varied across the inter-quartile range of the patient cohort. Index hospitalisation costs were varied across the 95% CI of the point estimate and unadjusted costs was also tested. We also tested the effect of higher ongoing treatment costs for serious injury and traumatic brain injury survivors based on the government contribution to spinal cord injury costs in Australia (estimated government contribution: $51,499 per annum [2009 AUD]).26 In terms of the survival probabilities incorporated into the models, we varied values across the 95% CI of the point estimates. We also tested the effect on survival estimates of excluding patients estimated to be dead on arrival, excluding inter-hospital transfers and omitting variables from the model that may have

1846

C. Taylor et al. / Injury, Int. J. Care Injured 43 (2012) 1843–1849

been associated with the intervention (endotracheal intubation; ICU admission; emergency operation performed). Finally, we performed probabilistic sensitivity analyses on the three costeffectiveness models using Monte Carlo Simulations incorporating gamma distributions for costs and beta distributions for probabilities.27 Estimating cost-effectiveness We derived two estimates of cost-effectiveness from our cost and effect assumptions, the incremental cost per life saved at one year and the incremental cost per life year saved over a patient’s lifetime. Estimates were derived for the whole patient group (Model #1) as well as the pre-specified patient sub-groups (Models #2 and #3). Future costs and benefits were discounted at a rate of 5%.28 To represent the uncertainty in the estimates, we derived cost-effectiveness acceptability curves (CEAC) for each of the three models.27

Results

Table 2 One year and lifetime treatment costs for HEMS and non-HEMS patients.

One year All patients Seriously injured patients TBI patients Life time All patients Seriously injured patients TBI patients

HEMS

Non-HEMS

$47,230 $66,335 $66,326

$29,650 $47,714 $42,560

$100,584 $313,784 $312,765

$82,371 $285,901 $276,861

all patients, seriously injured patients and head injured patients. One year treatment costs were predominantly influenced by the cost of the index hospitalisation. Residuals produced from estimating index hospitalisation costs in the a priori patient groups indicated good model fit and the removal of potential outliers from the models did not substantially affect the estimates (<10% change in predicted costs). Costs were uniformly higher for patients in the HEMS group compared to patients in the non-HEMS group.

Study sample Estimated effectiveness and cost-effectiveness Table 1 shows a total of 13,992 patients were included in the original cohort. After exclusions 10,180 patients with complete records remained (a priori exclusions n = 3546 patients; incomplete records n = 266). The final cohort included 1869 seriously injured patients (18.4%) and 1067 patients with traumatic brain injury (10.5%). Three hundred and ninety one patients (3.8%) were included in the HEMS group and 9789 patients (96.2%) in the nonHEMS group. Unadjusted in-hospital mortality was 2.9% and 6.1% for non-HEMS and HEMS patients respectively. A summary of the baseline characteristics of the HEMS and non-HEMS groups for the pre-specified patient groups is provided in Appendix II. After imputation of emergency department (ED) costs (for patients discharged directly from the ED), cost data were available in 88% of the final patient cohort (N = 8940). Cost data were missing due to unavailability of records (11%; N = 1107) and cross over between financial years (1%, N = 133). The unadjusted cost of the index hospitalisation was $8717 and $28,118 for non-HEMS and HEMS patients respectively. Estimated costs Table 2 reports the modelled one-year and life time treatment costs for HEMS and non-HEMS transport groups for

All models showed that adjusted probability of in-hospital survival in the HEMS group was significantly higher than patients in the non-HEMS group (p < 0.05). Removing covariates thought to be associated with the intervention led to more accurate survival estimates in both sub-group patient models without affecting model discrimination and calibration, and therefore these estimates were utilised in the base case costeffectiveness estimates for Models #2 and #3. After adjustment for patient imbalance between groups, the odds of in-hospital death in the non-HEMS group were between 3.2 and 3.8 times higher compared to the odds of death in the non-HEMS group (all patients: 3.80 [95% CI: 1.85–7.83; p = 0.0003]; patients with serious injury: 3.00 [95% CI: 1.45–6.21; p = 0.0031]; patients with TBI: 3.20 [95% CI: 1.23–8.34; p = 0.0173]). All models revealed excellent discrimination (AUC: 0.97 [all patients]; 0.90 [serious injury]; 0.91 [TBI]) and calibration (H–L p-value: 0.40 [all patients]; 0.73 [serious injury]; 0.96 [TBI]). As shown in Table 3, the improved survival in the HEMS group translated into a cost per life saved at one year between $519,787 and $1,566,379. The survival difference led to an additional 0.2–0.7 life years per patient across the a priori patient groups and a cost per life year saved between $49,159 and $96,524.

Table 1 Study sample characteristics. Initial cohort N Excluded [N; %]

13,992 3812 (27.2%)

Final cohort

All patients

Serious injury

TBI

N Died [N; %] Unadjusted cost of index hospitalisation HEMS N Died [N; %] Unadjusted cost of index hospitalisation Non-HEMS N Inter-hospital [N; %] Died [N; %] Unadjusted cost of index hospitalisation

10,180 309 (3.0%) $23,257

1869 288 (15.4%) $32,553

1067 211 (19.8%) $30,740

391 24 (6.1%) $28,118

182 23 (12.6%) $50,075

84 15 (17.9%) $59,513

9789 502 (5.1%) 285 (2.9%) $8716.99

1687 283 (16.8%) 265 (15.7%) $30,607

983 195 (19.8%) 196 (19.9%) $28,348

C. Taylor et al. / Injury, Int. J. Care Injured 43 (2012) 1843–1849

1847

Table 3 Incremental cost per life saved at one year and cost per life year saved.

All patients Seriously injured patients TBI patients

All patients Seriously injured patients TBI patients

% 1-year survival HEMS

% 1-year survival non-HEMS

Net lives saved per 100 transports

Cost per life saved

94.71 93.20 92.82

93.59 89.71 88.25

1.12 3.49 4.57

$1,566,379 $533,781 $519,787

Mean life expectancy HEMS

Mean life expectancy non-HEMS

Mean life year gain per patient

Cost per life year saved

16.43 15.39 15.33

16.24 14.83 14.60

0.19 0.56 0.73

$96,524 $50,035 $49,159

showed a higher likelihood of HEMS being cost effective in patients with serious injury and traumatic brain injury.

Sensitivity analysis results One and two-way sensitivity analyses revealed the incremental cost effectiveness ratios (ICERs) for cost per life year saved were sensitive to a number of input assumptions in Model #1 and less sensitive in Models #2 and #3. For patients in Model #1 ICERs ranged from $48,169 to $242,463 with a high discount rate (10%), unadjusted costs and a two-way analysis of the upper estimates of 95% CI for in-hospital mortality (for both groups) producing higher ICERs. In Models #2 and #3 ICER estimates ranged from $32,645 to $82,516 and $33,105 to $81,640 respectively. These models were sensitive to the same inputs as Model #1 as well as higher ongoing treatment costs. In all models the ICER estimates were not sensitive to excluding patients who were estimated to be ‘dead on arrival’. Excluding inter-hospital transfers from the analysis led to a much higher ICER in Model #1 (OR for death in non-HEMS versus HEMS: 2.3 [95% CI: 1.0–5.2]; $242,463 per life year saved) and similar ICERs in Models #2 (OR: 3.6 [95% CI: 1.4–9.2]; $50,657 per life year saved) and #3 (OR: 4.3 [95% CI: 1.2–15.2]; $45,360 per life year saved). Fig. 2 shows the results of the probabilistic sensitivity analysis, displaying the likelihood of the ICER for cost per life year saved being below increasing willingness to pay thresholds. Results

Discussion In NSW, physician-staffed Helicopter Emergency Medical Services (HEMS) are an expensive resource with undefined health economic benefits. Our results show a HEMS intervention combined with treatment at a major trauma centre is associated with improved mortality leading to an estimated cost per life saved between $519,787 and $1,566,379 and an estimated cost per life year saved between $49,159 and $96,524. The estimated costeffectiveness of HEMS improved in patients with more serious injuries and in patients with traumatic brain injury. This evaluation was a novel assessment of the value provided by HEMS in NSW in primary scene responses. The hospital selected for this investigation is classified in the highest tier of trauma hospital in NSW and has a large urban and rural catchment area which is broadly representative of many high-income trauma systems. Our effectiveness estimates were sourced from registry data which are prospectively collected and comprehensively capture patients meeting hospital trauma criteria. We also utilised individual patient cost data from hospital accounting systems which captured

1 0.9

Probability of being cost-effecve

0.8 0.7 0.6 0.5 ALL PATIENTS

0.4

SERIOUS INJURY

0.3

TBI

0.2 0.1

$96,000.00

$100,000.00

$92,000.00

$88,000.00

$84,000.00

$80,000.00

$76,000.00

$72,000.00

$68,000.00

$64,000.00

$60,000.00

$56,000.00

$52,000.00

$48,000.00

$44,000.00

$40,000.00

$36,000.00

$32,000.00

$28,000.00

$24,000.00

$20,000.00

$16,000.00

$8,000.00

$12,000.00

$-

$4,000.00

0

Willingness to pay per life-year saved Fig. 2. Cost effectiveness acceptability curves for cost per life-year saved in all patients, patients with serious injury and patients with traumatic brain injury.

1848

C. Taylor et al. / Injury, Int. J. Care Injured 43 (2012) 1843–1849

the true cost of the index hospitalisation. To evaluate the cost-effectiveness of a HEMS intervention we used an a priori statistical analysis plan which included a comparison of relevant health care strategies29 and robust outcomes. However, due to the differences in emergency medicine organisation and funding arrangements between regions, our results are likely to have limited generalisability outside Australia. Our estimates of the index hospitalisation cost and mortality in both groups relied on the robustness of our modelling approach. Covariates were selected based on clinical judgement and previous research.2 Removing covariates which were potentially associated with a HEMS intervention (such as ICU admission) did improve the accuracy of estimates in our sub-group models but the overall effect was not substantially different. In order to capture uncertainty we utilised the full 95% confidence intervals of the cost and effect estimates in our probabilistic sensitivity analysis. Beyond in-hospital survival, our estimates of life years saved depended on the age related probability of death for the general population in both groups. This assumes conservatively that there is no ongoing effect of HEMS transport beyond the index hospitalisation. Despite data being collected prospectively, our analysis is retrospective in nature and therefore the possibility of unobserved confounding cannot be excluded. A recent article highlighted the utility of propensity score matching in examining the benefit of HEMS,30 which resulted in lower estimated absolute risk reduction attributable to HEMS compared to traditional logistic regression. As our study utilised logistic regression, the benefit attributable to HEMS may have been overestimated although our estimates of HEMS benefit were similar to previous literature.2,31 In defining comparators we used a tasking agency perspective, where usual care (termed non-HEMS) included ether direct transport via ground to St George Hospital or ground transport to a regional hospital followed by inter-hospital transfer to St George Hospital. In this scenario, we hypothesised that the true benefit of HEMS was its ability to bypass regional hospitals in rural areas. By using these definitions we may have introduced a selection bias (where a comparable non-HEMS patient may have died before reaching St George Hospital), however inter-hospital transfer numbers were small and removing inter-hospital transfers did not substantially affect the cost-effectiveness estimates in Models #2 and #3. Despite the above limitations of our analysis, this investigation is one of few methodologically rigorous economic evaluations of HEMS9 and the results add to a paucity of economic literature in the field of out-of-hospital emergency care.32 A previous evaluation of HEMS from the perspective of the service provider in the US found a HEMS intervention in trauma patients led to cost per life year saved of $2454 USD33 (2010 AUD $4444). Our study adopted a broader perspective, including the ongoing medical cost of survivors (as is appropriate in the Australian context34), and found a higher level of costs. A more recent study found a HEMS intervention led to a cost per quality adjusted life year saved of s28,32735 (2010 AUD $59,140) in seriously injured patients. Although our study did not adjust for quality of life, the costeffectiveness of HEMS in seriously injured patients is similar. The cost-effectiveness of HEMS transport appears to improve with increasing injury severity and in patients with traumatic brain injury. This is consistent with other studies demonstrating benefit associated with HEMS.15,36 Our estimates showed HEMS are cost-effective in these populations despite survivors incurring higher ongoing treatment costs.26 Previous studies have shown that HEMS are poorly targeted to the patients that can benefit most due to the difficulty in assessing injury severity on-scene as well as a desire to keep under-triage to a minimum.37 The difference in cost-effectiveness in all patients versus the two specific subgroups (seriously injured and TBI) reflects the importance of accurate

dispatch criteria, to ensure HEMS are well targeted. A previous review highlighted a paucity of evidence supporting current dispatch criteria31 and given the potential improvements in costeffectiveness, our results support the need for further attention in this area. To judge the acceptability of our cost-effectiveness estimates, our results can be viewed in the context of the value of human life and society’s willingness to pay to avoid fatalities. A previous report38 reviewed the value of a statistical life (VSL) from 17 Australian and 227 international studies. A meta-analysis was undertaken of the higher quality studies yielding an average VSL of $6.0 million (2010 AUD $6.7m). The report also found the average value of a statistical life year (VALY) in Australia to be $124,095 (2010 AUD $138,767). More recently, a stated choice experiment was used to estimate the value of risk reduction (VRR) in car occupants in NSW,39 finding participants were willing to pay between $6.3 and $6.4 million (2010 AUD $6.9–7.0m) to avoid fatalities. Evidence from overseas has also confirmed individuals are willing to pay for a HEMS.40,41 Put into the context of such findings, our ratios appear within the range of acceptable values.

Conclusion There have been many evaluations of the benefit of HEMS in trauma, both in NSW and internationally. However, few evaluations have estimated the cost-effectiveness of HEMS relative to ground or indirect transport (via a referral hospital). Our evaluation shows that a HEMS intervention is associated with reduced in-hospital mortality. When modelled over a lifetime, this association leads to reasonable cost-effectiveness estimates, in the context of a single funder health care system, particularly in patients with serious injuries and traumatic brain injury. These data highlight the importance of further research into appropriate patient selection for a HEMS intervention.

Conflict of interest statement Co-authors Curtis, Tzannes, Dickson and Myburgh and all paid employees of the NSW Department of Health, which funds Helicopter Emergency Medical Services. The primary author (C. Taylor) has a paid part-time position with Novartis Pharmaceuticals. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.injury.2012.07.184. References 1. Sheperd M, Trethewy C, Kennedy J, Davis L. Helicopter use in rural trauma. Emergency Medicine Australasia 2008;20:494–9. 2. Butler DP, Anwar I, Willet K. Is it the H or the EMS in HEMS that has an impact on trauma patient mortality? A systematic review of the evidence. Emergency Medicine Journal 2010;27:692–701. 3. Ringburg AN, Thomas SH, Steyerberg EW, van Lieshout EMM, Patka P, Schipper IB. Lives saved by helicopter emergency medical services: an overview of literature. Air Medical Journal 2009;28(6):298–302. 4. Thomas SH. Helicopter emergency medical services transport outcomes literature: annotated review of articles published 2000–2003. Prehospital Emergency Care 2004;8(3):322–33. 5. Thomas SH. Helicopter EMS transport outcomes literature: annotated review of articles published 2004–2006. Prehospital Emergency Care 2007;11(4):477–88. 6. Thomas SH, Biddenger PD. Helicopter trauma transport: an overview of recent outcomes and triage literature. Current Opinion in Anaesthesiology 2003;16: 153–8. 7. Thomas SH, Cheema F, Cumming M, Wedel SK, Thomson D. Nontrauma helicopter emergency medical services transport: annotated review of selected outcomes-related literature. Prehospital Emergency Care 2002;6(2):242–55.

C. Taylor et al. / Injury, Int. J. Care Injured 43 (2012) 1843–1849 8. Thomas SH, Cheema F, Wedel SK, Thomson D. Trauma helicopter emergency medical services transport: annotated review of selected outcomes-related literature. Prehospital Emergency Care 2002;6(3):359–71. 9. Taylor CB, Stevenson M, Jan S, Middleton PM, Fitzharris M, Myburgh J. A systematic review of the cost and benefit of helicopter emergency medical services. Injury 2010;41:10–20. 10. Taylor CB, Stevenson M, Jan S, Liu B, Tall G, Middleton PM, et al. An investigation into the cost, coverage and activities of Helicopter Emergency Medical Services in the state of New South Wales, Australia. Injury 2011;42: 1088–94. 11. Botker MT, Bakke SA, Christensen EF. A systematic review of controlled studies: do physicians increase survival with prehospital treatment? Scandinavian Journal of Trauma Resuscitation and Emergency Medicine 2009:17. 12. Garner AA. The role of physician staffing of helicopter emergency medical services in prehospital trauma response. Emergency Medicine Australas 2004;16(4):318–23. 13. Ambulance Service of New South Wales. Protocol T1 – pre-hospital management of major trauma. 2008. 14. NSW Health. Selected specialty and statewide service plans – NSW trauma services. 2011http://www.health.nsw.gov.au/pubs/2009/pdf/trauma_report.pdf. 15. Frankema SP, Ringburg AN, Steyerberg EW, Edwards MJ, Schipper IB, van Vugt AB. Beneficial effect of helicopter emergency medical services on survival of severely injured patients. British Journal of Surgery 2004;91(11): 1520–6. 16. Davis DP, Peay J, Serrano JA, Buono C, Vilke GM, Sise MJ, et al. The impact of aeromedical response to patients with moderate to severe traumatic brain injury. Annals of Emergency Medicine 2005;46(2):115–22. 17. Palmer CS, Franklyn M, Read-Allsopp C, McLellan S, Niggemeyer L. Development and validation of a complementary map to enhance the existing 1998 to 2008 Abbreviated Injury Scale map. Scandinavian Journal of Trauma Resuscitation & Emergency Medicine 2011;19:29. 18. Australian Bureau of Statistics. ABS life tables. 2011http://www.abs.gov.au/. 19. Harrison-Felix CL, Whiteneck GG, Jha A, DeVivo MJ, Hammond FM, Hart DM. Mortality over four decades after traumatic brain injury rehabilitation: a retrospective cohort study. Archives of Physical Medicine and Rehabilitation 2009;90(9):1506–13. 20. Australian Bureau of Statistics. Consumer price index Australia. 2011http:// www.abs.gov.au/. 21. NSW Health Department. NSW program and product data collection standards incorporating subacute and non-acute and national minimum dataset for mental health establishments data collection requirements. North Sydney: NSW Health Department; 2006–2009. 22. NSW Health. NSW costs of care standards. Inter-Goverment & Funding Strategies Branch; 2010http://www.health.nsw.gov.au/policies/owner/igfsm.html. 23. Manning WG, Mullahy J. Estimating log models: to transform or not to transform? Journal of Health Economics 2001;20(4):461–94. 24. Independent Pricing and Regulatory Tribunal of New South Wales. Report to the Minister for Health on Revenue and Charging Structure. Review of financial aspects of the ambulance service in NSW. 2005http://www.sfaq.ipart.nsw.gov.au/files/Review%20of%20Financial%20Aspects%20of%20the%2 0Ambulance%20Service%20of%20NSW%20-%20November%202005.pdf.PDF. 25. Australian Institute of Health and Welfare. Health expenditure Australia 2008– 09. Health and welfare expenditure series, 2010http://www.aihw.gov.au/publication-detail/?id=6442472450.

1849

26. Access Economics. The economic cost of spinal cord injury and traumatic brain injury in Australia. Report prepared for the Victorian Neurotrauma Initiative; 2009, http://www.tac.vic.gov.au/upload/vni/The%20economic%20cost%20of% 20spinal%20cord%20injury%20and%20traumatic%20brain%20injury%20in%20 Australia.pdf. 27. Claxton K, Sculpher M, McCabe C, Briggs A, Akehurst R, Buxton M, et al. Probabilistic sensitivity analysis for NICE technology assessment: not an optional extra. Health Economics 2005;14(4):339–47. 28. Australian Government Department of Health Aging. Guidelines for preparing submissions to the Pharamceutical Benefits Advisory Committee (version 4.3). Canberra: Australian Government Department of Health Aging; 2008. 29. Drummond MF, Richardson WS, O’Brien BJ, Levine M, Heyland D. Users’ guides to the medical literature XIII. How to use an article on economic analysis of clinical practice. A. Are the results of the study valid?. Evidence-Based Medicine Working Group. JAMA 1997;277(19):1552–7. 30. Galvagno SM, Elliot RH, Nabeel Zafar S, Millin MG, Efron DT, Koenig GJ, et al. Association between helicopter vs ground emergency medical services and survival for adults with major trauma. JAMA 2012;307(15):8. 31. Ringburg AN, de Ronde G, Thomas SH, van Lieshout EMM, Patka P, Schipper IB. Validity of Helicopter Emergency Medical Services dispatch criteria for traumatic injuries: a systematic review. Prehospital Emergency Care 2009; 13(1):2009. 32. Lerner EB, Maio RF, Garrison HG, Spaite DW, Nichol G. Economic value of out-ofhospital emergency care: a structured literature review. Annals of Emergency Medicine 2006;47(6):515–24. 33. Gearhart PA, Wuerz R, Localio AR. Cost-effectiveness analysis of helicopter EMS for trauma patients. Annals of Emergency Medicine 1997;30(4):500–6. 34. Meltzer D. Accounting for future costs in medical cost-effectiveness analysis. Journal of Health Economics 1997;16(1):33–64. 35. Ringburg AN, Polinder S, Meulman TJ, Steyerberg EW, van Lieshout EMM, Patka P, et al. Cost-effectiveness and quality-of-life analysis of physician-staffed helicopter emergency medical services. British Journal of Surgery 2009;96: 1365–70. 36. Davis DP, Ochs M, Hoyt DB, Bailey D, Marshall LK, Rosen P. Paramedic-administered neuromuscular blockade improves prehospital intubation success in severely head-injured patients. Journal of Trauma-Injury Infection & Critical Care 2003;55(4):713–9. 37. Bledsoe BE, Wesley AK, Eckstein M, Dunn TM, O’Keefe MF. Helicopter scene transport of trauma patients with nonlife-threatening injuries: a meta-analysis. J Trauma 2006;60(6):1257–65. 38. Access Economics. The health of nations: the value of a statistical life. Report prepared for the office of the Australian Safety and Compensation Council. Canberra; 2007, http://www.safeworkaustralia.gov.au/AboutSafeWork Australia/WhatWeDo/Publications/Documents/330/TheHealthOfNations_Value_ StatisticalLife_2008_PDF.pdf. 39. Hensher DA, Rose JM, de Dios Ortuzar J, Rizzi LI. Estimating the willingness to pay and value of risk reduction for car occupants in the road environment. Transportation Research Part A 2009;43:692–707. 40. Olsen JA, Donaldson C. Helicopters, hearts and hips: using willingness to pay to set priorities for public sector health care programmes. Social Science & Medicine 1998;46(1):1–12. 41. Ringburg AN, Buljac M, Stolk EA, van Lieshout EMM, van Beeck F, Patka P, et al. Willingness to pay for lives saved by Helicopter Emergency Medical Services. Prehospital Emergency Care 2009;13(1):37–43.