Free
Perioperative Medicine  |   September 2012
Cost-effectiveness Analysis of Goal-directed Hemodynamic Treatment of Elderly Hip Fracture Patients: BeforeClinical Research Starts
Author Affiliations & Notes
  • Erzsebet Bartha, M.D., Ph.D.
    *
  • Thomas Davidson, Ph.D.
  • Ami Hommel, R.N., Ph.D.
  • Karl-Göran Thorngren, M.D., Ph.D.
    §
  • Per Carlsson, Ph.D.
  • Sigridur Kalman, M.D., Ph.D.
    #
  • *Research Assistant, Karolinska Institute, CLINTEC, Division of Anesthesiology, Karolinska University Hospital, Stockholm, Sweden. Assistant Professor, Center for Medical Technology Assessment, Linköping University, Linköping, Sweden. Assistant Professor, Department of Health Sciences, Lund University, Lund, Sweden. §Professor, Department of Orthopedics, Lund University Hospital, Lund, Sweden. Professor, Center for Medical Technology Assessment, Linköping University. #Professor, Karolinska Institute, CLINTEC, Division of Anesthesiology, Karolinska University Hospital.
Article Information
Perioperative Medicine / Cardiovascular Anesthesia / Geriatric Anesthesia / Trauma / Burn Care
Perioperative Medicine   |   September 2012
Cost-effectiveness Analysis of Goal-directed Hemodynamic Treatment of Elderly Hip Fracture Patients: BeforeClinical Research Starts
Anesthesiology 9 2012, Vol.117, 519-530. doi:10.1097/ALN.0b013e3182655eb2
Anesthesiology 9 2012, Vol.117, 519-530. doi:10.1097/ALN.0b013e3182655eb2
What We Already Know about This Topic
  • Clinical trials are undertaken to assess the benefits of new treatments

  • Cost-effectiveness is usually not considered at the time of trial design; thus even if the new treatment is effective, costs may prevent implementation

What This Article Tells Us That Is New
  • This article modeled that the goal-directed hemodynamic therapy for patients with hip fracture is cost-effective and a trial should be undertaken

MEDICAL care resources are limited. In many countries, decisions to adopt, reimburse, or issue specific guidance on use of new medical treatments are increasingly based on cost-effectiveness. Stakeholders in Australia,1 Canada,2 the United Kingdom,3 and the United States4 first used this approach. In 1997, a law integrated cost-effectiveness consideration into Sweden's medical care system's prioritization processes.1
In contrast, funding on applied clinical research decisions is usually not linked to health economic factors, even if research projects are costly and funded by public resources and if resources are scarce.5 This is particularly striking when research projects consider medical treatments that presumably cannot be adopted in the future due to limited resources and poor cost-effectiveness. In such situations, cost-effectiveness analyses of unproven medical technologies may be reasonable before commissioning clinical research projects. This is an issue for policy makers and clinical researchers.
In cost-effectiveness analyses at least two alternative interventions are compared in terms of costs and changes in patients' health, usually using a long-time perspective. When relevant data are unavailable, stochastic (probabilistic) decision analytic models are used to apply best (or next best) evidence combined with reasonable assumptions.6  8 
This analysis is tightly linked to the design and launch of a clinical research project on hemodynamic optimization of elderly patients with hip fracture; we use an analytic method that is increasingly being applied in medical care policy decisions.
Each year, approximately 20,000 patients have hip surgery in Sweden. Four-month mortality is 15% for females and 20% for males, and only 50% of these patients are discharged to their original housing.2In other surgical patients, perioperative fluid overload or deficit may influence postoperative outcome, and in past decades, many fluid treatment protocols were studied.9  15 One is goal-directed hemodynamic treatment (GDHT) that is targeted to increase global blood flow. As per meta-analyses,12  14 GDHT is beneficial for high-risk surgical patients. In elderly patients with hip fracture, current evidence suggests that GDHT might reduce hospital stay.16,17 However, length of hospital stay is a surrogate endpoint. For a policy decision on GDHT in elderly patients evidence is required on clinical effectiveness, on patient-oriented benefits, and on the used resources. The authors planned a clinical trial (n = 460) for the actual population to find evidence on clinical benefit (postoperative complications). But commissioning and funding such a trial could be questioned because it has been suggested that that only “25 to 60% of the mortality will be potentially susceptible to the intervention” due to high age and comorbidities.18 In addition, if GDHT cannot be used in the future, due to limited resources, then clinical trials on this vulnerable patient group would be inappropriate for economic and ethical reasons.
We aimed to estimate cost effectiveness before commissioning a clinical trial on GDHT in elderly patients with hip fracture to guide researchers and those who set research priorities—if a future GDHT trial for the elderly is potentially meaningful.
Materials and Methods
This section describes this investigation's perspective, the decision analytic model,6 and various analysis phases.
Perspective
Cost-effectiveness analyses always compare alternative treatment strategies. In this investigation, routine fluid treatment is compared with GDHT for hypothetical individuals with hip fracture (older than 80 yr). Routine fluid treatment represents current clinical practice in Sweden. Blood pressure and heart rate guide administered fluid volumes. GDHT represents a treatment protocol to be targeting Shoemaker's19 proposed objectives (oxygen delivery more than 600 ml · min−1· m−2, cardiac index more than 4.5 l · min−1· m−2) when using the Lithium Dilution Cardiac Output monitor (LiDCO®, LiDCO Ltd., Sawston, Cambridge, United Kingdom). Our analysis takes a medical care perspective3on costs and it follows effects of interventions, 5 yr postoperatively. Model output is the estimated incremental cost-effectiveness ratio (ICER).
Decision Analytic Model
The analysis is done using a decision analytic model. It applies mathematical relationships that illustrate consequences of both treatment strategies. The model consists of two parts, a decision tree7 that was developed for short-term postoperative outcome and a Markov structure6 that was developed for long-term outcome (fig. 1A and B). The next section briefly describes various analysis phases. A more comprehensive description of the model structure, data collection, and data incorporation (into the model) are available elsewhere.4
Fig. 1. A and B Model structure. (A  ) A decision tree. The short-term model starts with the decision between alternative fluid strategies (rectangle  ) followed by arrows  that represent transitions toward selected postoperative complications. (B  ) The Markov structure. The long-term model: upon entering the model, hypothetical individuals have quality of life that is associated with postoperative outcome. The individuals make transitions along the arrows  among health states or stay in the same state during 1 yr, that is, one cycle. During this cycle, survival decreases and survivors' quality of life declines by decrements. One cycle is repeated five times, representing 5 postoperative years.
Image Not Available
Fig. 1. A and B Model structure. (A  ) A decision tree. The short-term model starts with the decision between alternative fluid strategies (rectangle  ) followed by arrows  that represent transitions toward selected postoperative complications. (B  ) The Markov structure. The long-term model: upon entering the model, hypothetical individuals have quality of life that is associated with postoperative outcome. The individuals make transitions along the arrows  among health states or stay in the same state during 1 yr, that is, one cycle. During this cycle, survival decreases and survivors' quality of life declines by decrements. One cycle is repeated five times, representing 5 postoperative years.
×
Analysis Phases
Decision Tree Development.
The decision tree is used to estimate the short-term costs and postoperative outcome (fig. 1A and B). It starts with a decision between two fluid treatment strategies (rectangle). A chance node (circle) follows the decision, where various events may occur by chance, and hypothetical individuals may make transition (arrows) toward one of the selected postoperative outcomes (triangles in fig. 1A and B): uncomplicated recovery, cardiovascular complications, stroke, other complications (i.e.  , pulmonary and urinary tract infections, postoperative confusion, kidney insufficiency, wound infection, and pulmonary embolism), and death. Each postoperative outcome (triangles) is accounted for with estimated occurrence probability, health-related quality of life (QoL) index, and cost. So the model translates selected postoperative complications into health states (exemplified by health-related QoL) that are measured by the EQ-5D instrument.20 The decision tree's timeline is 4 months after the operation.
Model Inputs Into the Decision Tree.
For routine fluid treatment, probabilities for each postoperative outcome were generated with data drawn from a Swedish trial (n = 402; 100% follow-up rate).21 For GDHT, probabilities for each postoperative outcome were generated using estimates of relative risk (GDHT vs.  routine fluid treatment) for mortality and morbidity. These were extracted from the scientific literature. Appendixes 1 and 2 display the search strategy and results. No data were found on the clinical benefit of the actual GDHT protocol on the actual population, so next-best data were used. The relative risk for postoperative morbidity was directly calculated from findings of Venn et al.  17 (actual population but another GDHT strategy). The relative risk for mortality was extracted from meta-analysis13 (actual strategy but younger population). Table 1lists clinical effect estimates.
Table 1. Model Inputs with Mean Estimates, Confidence Intervals, and Distribution
Image Not Available
Table 1. Model Inputs with Mean Estimates, Confidence Intervals, and Distribution
×
Prefracture and postfracture QoL indices were derived from the age-matched nonfractured Swedish population22,23 and a longitudinal Swedish clinical trial.24 Decrements of QoL (difference between prefracture and postfracture QoL indices) were calculated and used for the analysis (table 1).
Short-term medical care costs consisted of fluid treatment costs including medical devices—monitoring with LiDCO® and human resources during the perioperative period (table 1). Hospital costs for each postoperative complication and for uncomplicated recovery included hospital stay length, cost per one bed-day, plus laboratory, microbiology, radiology, and operations expenses. Fluid treatment cost data were calculated at the Karolinska University Hospital, Huddinge, Sweden. Hospital cost data were based on individual patient-specific cost data at University Hospital in Lund, Sweden.
Markov Structure Development.
A Markov structure was developed for modeling long-term survival, medical care costs, and QoL. After hospital discharge, hypothetical individuals are categorized into health states in the Markov structure (circles in fig. 1A and B). QoL, which is aligned with each postoperative outcome, exemplifies these health states. Hypothetic individuals may make transitions along the arrows (fig. 1A and B) among health states or stay in the same state during 1 yr, i.e.  , one cycle. Note that this model simplifies real life, because it allows for recovery only from the “other” complications state. So after cardiovascular complications or stroke, hypothetical individuals continue to live with consequences of these complaints. These factors describe the health states (circles): estimated survival probabilities, QoL, and medical care costs. During one cycle, survival decreases and survivors' QoL declines by decrements. One cycle is repeated five times—representing 5 postoperative yr. Here, a QoL index was multiplied by the time spent in the current health state (1 yr), which generated the number of quality-adjusted life years (QALYs). Each life-year of a hypothetical individual is associated with medical care costs. Costs and QALYs were aggregated, which yields the expected, estimated mean costs and QALYs of both treatment strategies.
Model Input Into the Markov Structure.
Age-adjusted standard mortality.5was used for hypothetical individuals with no postoperative complication. For those with cardiovascular complications or stroke, yearly mortality was estimated using age- and disease-related mortality from the Swedish National Stroke Registry6and the Swedish National Registry on Secondary Prevention in Cardiac Intensive Care7(Kalle Spångberg, Ph.D., section manager, Uppsala Clinical Research Center, Uppsala University, Uppsala, Sweden, written communication: May 15, 2009). Mean inpatient and outpatient long-term medical care-cost data came from the Swedish National Board of Health and Welfare for patients who received hip fracture surgery in 2007 and were hospitalized in 2008 (table 1) (Leif Forsberg, statistician, Department of Statistics, Monitoring and Evaluation, Swedish National Board of Health, Stockholm, Sweden, written communication: December 7, 2009).
Assumptions Used in the Model.
For the base case analysis, the following assumptions were made:
  1. GDHT may influence each of the selected postoperative complications.

  2. GDHT may influence postoperative mortality in elderly patients.

  3. Prefracture QoL is equivalent to the nonfractured, aged-matched population.

  4. Postfracture QoL—associated with postoperative stroke and cardiovascular events—is equivalent to nonfractured patients with stroke and cardiovascular diseases.

  5. Postfracture QoL associated with “other” complications is equivalent to QoL reported for healing complications after hip fracture.

Cost-effectiveness Analysis.
Data uncertainty was accounted for by defining probability distributions for all model inputs that were applied in the base case analysis (table 1). First the model was run with average values of all model inputs that yield average costs and QALYs. Then a second-order Monte Carlo simulation was performed, and the cohort was simulated through five cycles (years). In each simulation, input data values were randomly drawn from the defined probability distributions; the simulation was performed 1,000 times—generating 1,000 estimates of aggregated costs and QALYs. The model was programmed and analyzed using Microsoft Excel (Microsoft Corporation 1985–2001, version 12.0.6554.5003, Redmond, WA). Costs and effects (QALYs) were discounted by 3% annually. The ICER was calculated dividing the difference between costs (incremental cost) by the difference between the QALYs (incremental effect) for the two strategies. The ICER is an estimate of additional costs for getting one additional life-year with full health when the two alternatives are compared. The ICER is always related to one possible threshold value that society is willing to pay for one additional life-year with full health (cost-effectiveness threshold, λ). In Sweden, there is no fixed official threshold or range accepted, but a cost of between €20,000 and €50,000 is discussed depending of the severity of the condition (in exceptional cases even higher). Standards within Sweden's medical care system guided this analysis, which applied Swedish hospital costs that are converted to euros using this exchange rate: €1 = SEK (Swedish Krona) 9.41.
Sensitivity Analyses of Uncertain Data.
Probabilistic sensitivity analyses were done to account for uncertainty of model assumptions and to address variability in data that were used. One-way sensitivity analyses were performed over upper and lower limits, respectively, of 95% CI for the model inputs. The model was also run using stepwise, increased estimates of relative risk for morbidity and mortality that represent lower expected clinical effect, compared with the base case analysis.
Results
Analysis of the Average ICER
The base case analysis compared QALYs and costs for a hypothetical cohort of hip fracture patients, age 80 yr and older, who were treated with preoperative GDHT or routine fluid treatment. The analysis accounted for 5 postoperative yr (table 2). On average the GDHT leads to reduced costs by €−1,882 and to increase of QALYs by 0.344, which yields a negative average ICER (when the ICER is negative it should not be expressed). Consequently the GDHT is dominant8
Table 2-a. Average Costs and QALYs, Incremental Costs and QALYs (Δ cost, Δ effect), and Incremental Cost-effectiveness Ratio (ICER)
Image Not Available
Table 2-a. Average Costs and QALYs, Incremental Costs and QALYs (Δ cost, Δ effect), and Incremental Cost-effectiveness Ratio (ICER)
×
Table 2-b. Average Costs and QALYs, Incremental Costs and QALYs (Δ cost, Δ effect), and Incremental Cost-effectiveness Ratio (ICER)
Image Not Available
Table 2-b. Average Costs and QALYs, Incremental Costs and QALYs (Δ cost, Δ effect), and Incremental Cost-effectiveness Ratio (ICER)
×
One-way Sensitivity Analyses
The cost-effectiveness analysis was generally robust to changes of model inputs within ranges of 95% CIs (fig. 2)—with one exception. The ICER was sensitive to relative mortality and morbidity risk-value changes. Relative risk influence was separately tested with stepwise increased values by 25 to 90% (appendix 3). When clinical effect is reduced via  a 90% relative risk increase, then GDHT dominance disappears, which yields an ICER of €383 per gained QALY (fig. 3).
Fig. 2. One-way sensitivity analyses. One-way sensitivity analyses using upper and lower level of a 95% CI of the selected model inputs, respectively. The selected model inputs were the relative risk for mortality and morbidity, prefracture values of quality-of-life index (QoL index), the decrements of quality-of-life (QoL decrements) associated with postoperative complications and hospital costs.
Image Not Available
Fig. 2. One-way sensitivity analyses. One-way sensitivity analyses using upper and lower level of a 95% CI of the selected model inputs, respectively. The selected model inputs were the relative risk for mortality and morbidity, prefracture values of quality-of-life index (QoL index), the decrements of quality-of-life (QoL decrements) associated with postoperative complications and hospital costs.
×
Fig. 3. Deterministic analyses with relative-risk point estimates for postoperative mortality and morbidity. Differences in quality-adjusted life-years (Δ QALY) are plotted against cost differences (Δ costs, €). The model was run using the baseline, and the stepwise increased the baseline values (by 25–90%) of relative risk. Appendix 3 displays the stepwise increased estimates for relative risks.
Image Not Available
Fig. 3. Deterministic analyses with relative-risk point estimates for postoperative mortality and morbidity. Differences in quality-adjusted life-years (Δ QALY) are plotted against cost differences (Δ costs, €). The model was run using the baseline, and the stepwise increased the baseline values (by 25–90%) of relative risk. Appendix 3 displays the stepwise increased estimates for relative risks.
×
Probabilistic Analysis
Figure 4shows the Monte Carlo simulation results; here, differences between costs are plotted against differences of effect for all simulated values. Figure 4demonstrates how the combined uncertainty of model inputs is translated into the uncertainty of model outputs. For 96.4% of the simulated values (right lower quadrant), GDHT for hypothetical individuals (ages 80 yr and older) is dominant; it is less costly and more effective as measured by QALYs on a 5-yr time line—compared with traditional fluid treatment. For 3% of the values (right upper quadrant), GDHT is more costly and more effective; here, GDHT may still be cost- effective. The governing factor in this quadrant is a threshold value of how much society is willing to pay for one additional life-year with full health for the target population. The slope of the dotted line (λ) represents one possible cost-effectiveness threshold value. Values below the dotted line represent simulations when GDHT is cost-effective. So the combined uncertainty of model inputs and the cost-effectiveness threshold value determine the probability of cost-effectiveness.
Fig. 4. Results of the Monte Carlo simulation. Simulated values of incremental costs and effects (Δ costs, Δ quality-adjusted life-years, QALYs) of goal-directed hemodynamic treatment (GDHT) compared with routine fluid therapy. The slope of the dotted line  (λ) represents one possible threshold value that indicates how much society is willing to pay for one additional life-year with full health for the target population (cost-effectiveness threshold).
Image Not Available
Fig. 4. Results of the Monte Carlo simulation. Simulated values of incremental costs and effects (Δ costs, Δ quality-adjusted life-years, QALYs) of goal-directed hemodynamic treatment (GDHT) compared with routine fluid therapy. The slope of the dotted line  (λ) represents one possible threshold value that indicates how much society is willing to pay for one additional life-year with full health for the target population (cost-effectiveness threshold).
×
The 95% CI of the ICER is calculated by the upper and lower limit of 95% of the simulated values of incremental costs (€−3,043 to €+239) and QALYs (0.082- 0.492). It yields a negative ICER at the lower limit and a cost per QALY gained at €2,915 at the upper limit.
Probabilstic Sensitivity Analysis
Figure 5illustrates probabilistic sensitivity analyses; here, probabilties for GDHT being cost-effective are plotted against varying values of the cost-effectiveness threshold. The model was run for the base case analysis (red line) and two alternative scenarios—to test two main model assumptions. Scenario 1: GDHT does not influence most of the selected posotperative complications that constitute the group of “other” complications (blue line). Scenario 2: GDHT does not influence the group of “other” complications and mortality (green line).
Fig. 5. Probabilistic sensitivity analyses. Probabilities for cost effectiveness are plotted against the cost-effectiveness threshold for the base case analysis, when the goal-directed hemodynamic treatment (GDHT) does not influence the major group of complications (other complications) and when the GHDT does not influence mortality and other complications. Other complications are listed in the section on the decision tree. QALY = quality-adjusted life-years.
Image Not Available
Fig. 5. Probabilistic sensitivity analyses. Probabilities for cost effectiveness are plotted against the cost-effectiveness threshold for the base case analysis, when the goal-directed hemodynamic treatment (GDHT) does not influence the major group of complications (other complications) and when the GHDT does not influence mortality and other complications. Other complications are listed in the section on the decision tree. QALY = quality-adjusted life-years.
×
For the base case analysis and both scenarios the probabilty of being cost-effective is above 0.975 at a cost effectiveness threshold of €10,000.
Discussion
Our main finding is that compared with routine treatment for patients, ages 80 yr and older, GDHT yields gained QALYs at lower medical care costs over a 5-yr time line (in 96.5% of the simulations). In health economic terms the GDHT is the dominant strategy. The analysis is most sensitive for changes in morbidity and mortality relative risk values. Although with very modest clinical effect size values (relative risk for mortality/morbidity 0.92/0.84), GDHT may be cost-effective. The influence of postoperative complications on postfracture QoL is probably understated because these were extracted from a nonfractured population. But in the one-way sensitivity analyses, the ICER remained negative when using the QoL decrements within ranges of 95% CIs (fig. 2). The analytic model is used to estimate incremental cost effectiveness ratio—with existing data, in a position of uncertainty considering the benefit of the GDHT on the actual population, and before a planned clinical trial. Our cost-effectiveness analysis provides support for commissioning a clinical trial. This analysis was not intended for guiding GDHT implementation in routine clinical practice. Introduction of GDHT should await evidence-based data from future randomized controlled trials, which demonstrate that the technology conveys net benefit.
The presented model is a dynamic framework, and it can be updated either when new evidence comes up on the clinical effect size for GDHT or when initially high costs of new technologies decrease over time.
Clinical effect size (relative risk) constitutes the most important variable in this analysis, because when the model was run by the upper and lower limits of 95% CI of relative risk the ICER has changed substantially.
Why Cost Effectiveness?
Particularly when a new treatment strategy is very resource-consuming, and if the strategy accounts for patients with limited life expectancy with uncertain benefit, then a cost-effectiveness analysis may be meaningful (before initiating a costly clinical trial) as input for prioritization of research projects. Several methods exist for setting priorities in clinical research. These include measures of the burden disease,25,26 the expected “payback” from the research,27,28 estimated welfare losses,29 or value of information analysis.5 Before commissioning a large randomized clinical trial, we ran a pilot trial on 40 patients. During the design period and literature search we could determine that patient recruitment is cumbersome due to the acute confusion of the patients under the circumstances of unscheduled surgery, and there is a huge gap between the numbers of GDHT trials in the elderly patients compared with the younger population.
The lack of GDHT trials on elderly patients may be a result of assumptions that such trials are not meaningful due to age and comorbidities. In this analysis, we found that even modest clinical effect may improve health outcomes and decrease medical care costs (fig. 3).
Analysis Strengths and Limitations
This early analysis was done per health economical evaluation standards. We used a two-part model that is commonly used for reimbursement decisions of new, unproven medical technologies when clinical trials are not yet available. Simplification of real life constitutes an analytical model limitation, but it is possible to model the complexity of expected GDHT influence on postoperative complications. It is unlikely that all postoperative complications may be influenced by GDHT; this complexity is partially modeled by probabilistic scenario analyses (scenarios 1 and 2).
External validity of the results (model outputs) depends on model input validity. In our analysis, model inputs have high external validity. Short-term survival and hospital costs of the traditional fluid treatment were obtained from a trial population with a follow-up rate of 100% on patients at University Hospital in Lund, Sweden.21 Data from the Swedish National Registry on Secondary Prevention in Cardiac Intensive Care and the Epidemiologic Centre of the Swedish National Board of Health have high validity, because these national registries have data from all Swedish hospitals (100%). The Swedish National Stroke Registry has data from 83% of all hospitals of Sweden but is still the best available data source for survival after stroke.
Implications for Further Research
Our results show low values of the estimated number needed to treat using GDHT through the entire range of estimated relative risk (appendix 4). The displayed estimates of number needed to treat in appendix 4 are the estimated number of patients needed to treat to prevent one patient with negative outcome (postoperative complication). GDHT is also predicted to have high probability of being cost-effective (over ranges of CI) even if GDHT may require more resources during perioperative care. These results support research funding in the area. A future trial should address clinical effectiveness, patient-oriented benefits (QoL), and cost-effectiveness to support future policy decision on the current large patient population. Given the expected absolute risk for postoperative complications (0.6) and the point estimates of relative risk used in the one-way sensitivity analyses (fig. 3), corresponding sample sizes for future clinical trials are calculated (appendix 4). When the relative risk is between 0.5 and 0.79, then the required sample size is between 84 and 490. When the expected relative risk exceeds 0.88, then the sample size should exceed 1,488, which is probably not realistic to aim for in a clinical trial. In a future trial, during an interim analysis, one stopping rule could be relative risk higher than 0.79, because already this effect size would indicate need for a sample size over 490.
Conclusion
Scientific evidence on clinical benefits of GDHT on elderly patients with hip fracture is scarce. So we addressed the question of whether or not a costly, large trial is meaningful due to increased age and frailty of the patients. Early cost-effectiveness analysis predicts that GDHT may save costs of medical care and may gain QALYs—compared with traditional fluid treatment. Large trials on GDHT for elderly patients should be supported, because even strategies with modest clinical effect promise to be cost effective.
Moreover, when accounting for expected time and cost for a clinical trial, this type of cost-effectiveness analysis was found feasible, meaningful, and recommendable before launch of costly applied clinical research projects in general. Such analyses might be beneficial even in countries in which cost-effectiveness analyses are not accepted for policy decisions. An early pretrial analysis might reduce risk for inefficient use of scarce research resources when anticipated societal or patient benefits from clinical research are low.
The authors thank Kalle Spångberg, Ph.D., Section Manager, Uppsala Clinical Research Center, Uppsala University, Uppsala, Sweden, and Leif Forsberg, Statistician, Department of Statistics, Monitoring and Evaluation, Swedish National Board of Health, Stockholm, Sweden.
References
Commonwealth Department of Health HaCS: Guidelines for the pharmaceutical industry on preparation of submissions to the Pharmaceutical Benefits Advisory Committee, including submissions involving economic analyses, Canberra: Australian Government Public Service, 2002, pp 27–42
Health Ontario Mo: Ontario guidelines for economic analysis of pharmaceutical products, Queen's printer for Ontario, 1994, pp 2–11
Claxton K, Sculpher M, Drummond M: A rational framework for decision making by the National Institute For Clinical Excellence (NICE). Lancet 2002; 360:711–5
Langley PC: Formulary submission guidelines for Blue Cross and Blue Shield of Colorado and Nevada. Structure, application and manufacturer responsibilities. Pharmacoeconomics 1999; 16:211–24
Claxton KP, Sculpher MJ: Using value of information analysis to prioritise health research: Some lessons from recent UK experience. Pharmacoeconomics 2006; 24:1055–68
Briggs A CK, Schulper M: Decision Modelling for Health Economic Evaluation, 2ndedition. Oxford, United Kingdom, Oxford University Press, 2007
Drummond MF: Methods for the Economic Evaluation of Health Care Programmes, 3rd edition. Oxford, United Kingdom, Oxford University Press, 2005
NICE: Guide to the methods of technology appraisal, National Institute for Clinical Excellence, 2004, pp 19–29
Bundgaard-Nielsen M, Secher NH, Kehlet H: ‘Liberal’ vs. ‘restrictive’ perioperative fluid therapy–a critical assessment of the evidence. Acta Anaesthesiol Scand 2009; 53:843–51
Bundgaard-Nielsen M, Ruhnau B, Secher NH, Kehlet H: Flow-related techniques for preoperative goal-directed fluid optimization. Br J Anaesth 2007; 98:38–44
Abbas SM, Hill AG: Systematic review of the literature for the use of oesophageal Doppler monitor for fluid replacement in major abdominal surgery. Anaesthesia 2008; 63:44–51
Kern JW, Shoemaker WC: Meta-analysis of hemodynamic optimization in high-risk patients. Crit Care Med 2002; 30:1686–92
Poeze M, Greve JW, Ramsay G: Meta-analysis of hemodynamic optimization: Relationship to methodological quality. Crit Care 2005; 9:R771–9
Boyd O: Optimisation of oxygenation and tissue perfusion in surgical patients. Curr Anaesth Crit Care 2003; 13:206–14
Price JD, Sear JW, Venn RM: Perioperative fluid volume optimization following proximal femoral fracture. Cochrane Database Syst Rev 2004; CD003004
Sinclair S, James S, Singer M: Intraoperative intravascular volume optimisation and length of hospital stay after repair of proximal femoral fracture: Randomised controlled trial. Bmj 1997; 315:909–12
Venn R, Steele A, Richardson P, Poloniecki J, Grounds M, Newman P: Randomized controlled trial to investigate influence of the fluid challenge on duration of hospital stay and perioperative morbidity in patients with hip fractures. Br J Anaesth 2002; 88:65–71
Foss NB, Kehlet H: Mortality analysis in hip fracture patients: Implications for design of future outcome trials. Br J Anaesth 2005; 94:24–9
Shoemaker WC, Appel PL, Kram HB, Waxman K, Lee TS: Prospective trial of supranormal values of survivors as therapeutic goals in high-risk surgical patients. Chest 1988; 94:1176–86
EuroQol–a new facility for the measurement of health-related quality of life. The EuroQol Group. Health Policy 1990; 16:199–208
Hommel A, Ulander K, Bjorkelund KB, Norrman PO, Wingstrand H, Thorngren KG: Influence of optimised treatment of people with hip fracture on time to operation, length of hospital stay, reoperations and mortality within 1 year. Injury 2008; 39:1164–74
Burstrm K, Johannesson M, Diderichsen F: Swedish population health-related quality of life results using the EQ-5D. Qual Life Res 2001; 10:621–35
Burstrm K, Johannesson M, Diderichsen F: Health-related quality of life by disease and socio-economic group in the general population in Sweden. Health Policy 2001; 55:51–69
Tidermark J, Zethraeus N, Svensson O, Trnkvist H, Ponzer S: Femoral neck fractures in the elderly: Functional outcome and quality of life according to EuroQol. Qual Life Res 2002; 11:473–81
Gross CP, Anderson GF, Powe NR: The relation between funding by the National Institutes of Health and the burden of disease. N Engl J Med 1999; 340:1881–7
Michaud CM, Murray CJ, Bloom BR: Burden of disease–implications for future research. Jama 2001; 285:535–9
Townsend J, Buxton M: Cost effectiveness scenario analysis for a proposed trial of hormone replacement therapy. Health Policy 1997; 39:181–94
Davies L, Drummond M, Papanikolaou P: Prioritizing investments in health technology assessment. Can we assess potential value for money? Int J Technol Assess Health Care 2000; 16:73–91
Phelps CE, Parente ST: Priority setting in medical technology and medical practice assessment. Med Care 1990; 28:703–23
Boyd O, Hayes M: The oxygen trail: The goal. Br Med Bull 1999; 55:125–39
Rahbari NN, Zimmermann JB, Schmidt T, Koch M, Weigand MA, Weitz J: Meta-analysis of standard, restrictive and supplemental fluid administration in colorectal surgery. Br J Surg 2009; 96:331–41
Wilson J, Woods I, Fawcett J, Whall R, Dibb W, Morris C, McManus E: Reducing the risk of major elective surgery: Randomised controlled trial of preoperative optimisation of oxygen delivery. Bmj 1999; 318:1099–103
Takala J, Meier-Hellmann A, Eddleston J, Hulstaert P, Sramek V: Effect of dopexamine on outcome after major abdominal surgery: A prospective, randomized, controlled multicenter study. European Multicenter Study Group on Dopexamine in Major Abdominal Surgery. Crit Care Med 2000; 28:3417–23
Lobo SM, Salgado PF, Castillo VG, Borim AA, Polachini CA, Palchetti JC, Brienzi SL, de Oliveira GG: Effects of maximizing oxygen delivery on morbidity and mortality in high-risk surgical patients. Crit Care Med 2000; 28:3396–404
Gan TJ, Soppitt A, Maroof M, el-Moalem H, Robertson KM, Moretti E, Dwane P, Glass PS: Goal-directed intraoperative fluid administration reduces length of hospital stay after major surgery. ANESTHESIOLOGY 2002; 97:820–6
Conway DH, Mayall R, Abdul-Latif MS, Gilligan S, Tackaberry C: Randomised controlled trial investigating the influence of intravenous fluid titration using oesophageal Doppler monitoring during bowel surgery. Anaesthesia 2002; 57:845–9
Sandham JD, Hull RD, Brant RF, Knox L, Pineo GF, Doig CJ, Laporta DP, Viner S, Passerini L, Devitt H, Kirby A, Jacka M, Canadian Critical Care Clinical Trials Group: A randomized, controlled trial of the use of pulmonary-artery catheters in high-risk surgical patients. N Engl J Med 2003; 348:5–14
Pearse R, Dawson D, Fawcett J, Rhodes A, Grounds RM, Bennett ED: Early goal-directed therapy after major surgery reduces complications and duration of hospital stay. A randomised, controlled trial [ISRCTN38797445] Crit Care 2005; 9:R687–93
Noblett SE, Snowden CP, Shenton BK, Horgan AF: Randomized clinical trial assessing the effect of Doppler-optimized fluid management on outcome after elective colorectal resection. Br J Surg 2006; 93:1069–76
Donati A, Loggi S, Preiser JC, Orsetti G, Mnch C, Gabbanelli V, Pelaia P, Pietropaoli P: Goal-directed intraoperative therapy reduces morbidity and length of hospital stay in high-risk surgical patients. Chest 2007; 132:1817–24
Wakeling HG, McFall MR, Jenkins CS, Woods WG, Miles WF, Barclay GR, Fleming SC: Intraoperative oesophageal Doppler guided fluid management shortens postoperative hospital stay after major bowel surgery. Br J Anaesth 2005; 95:634–42
Lobo SM, Lobo FR, Polachini CA, Patini DS, Yamamoto AE, de Oliveira NE, Serrano P, Sanches HS, Spegiorin MA, Queiroz MM, Christiano AC Jr., Savieiro EF, Alvarez PA, Teixeira SP, Cunrath GS: Prospective, randomized trial comparing fluids and dobutamine optimization of oxygen delivery in high-risk surgical patients [ISRCTN42445141]. Crit Care 2006; 10:R72
Lopes MR, Oliveira MA, Pereira VO, Lemos IP, Auler JO Jr., Michard F: Goal-directed fluid management based on pulse pressure variation monitoring during high-risk surgery: A pilot randomized controlled trial. Crit Care 2007; 11:R100
Senagore AJ, Emery T, Luchtefeld M, Kim D, Dujovny N, Hoedema R: Fluid management for laparoscopic colectomy: A prospective, randomized assessment of goal-directed administration of balanced salt solution or hetastarch coupled with an enhanced recovery program. Dis Colon Rectum 2009; 52:1935–40
Mayer J, Boldt J, Mengistu AM, Rohm KD, Suttner S: Goal-directed intraoperative therapy based on autocalibrated arterial pressure waveform analysis reduces hospital stay in high-risk surgical patients: A randomized, controlled trial. Crit Care 2010; 14:R18
Benes J, Chytra I, Altmann P, Hluchy M, Kasal E, Svitak R, Pradl R, Stepan M: Intraoperative fluid optimization using stroke volume variation in high risk surgical patients: Results of prospective randomized study. Crit Care 2010; 14:R118
Appendix 1. Summary of Search Results on Meta-analyses for Postoperative Mortality
Image Not Available
Appendix 1. Summary of Search Results on Meta-analyses for Postoperative Mortality
×
Appendix 2. Summary of Search Results on Clinical Trials that Used Goal-directed Hemodynamic Treatment before Onset of Organ Failure
Image Not Available
Appendix 2. Summary of Search Results on Clinical Trials that Used Goal-directed Hemodynamic Treatment before Onset of Organ Failure
×
Appendix 3. Estimates of Relative Risk (Goal-directed Hemodynamic Treatment Compared with Routine Fluid Treatment) Used in Deterministic Sensitivity Analyses
Image Not Available
Appendix 3. Estimates of Relative Risk (Goal-directed Hemodynamic Treatment Compared with Routine Fluid Treatment) Used in Deterministic Sensitivity Analyses
×
Appendix 4. Given the Expected Absolute Risk for Postoperative Complications (0.6) and the Point Estimates of Relative Risk Used in the One-way Sensitivity Analyses, Corresponding Sample Sizes of a Future Clinical Trials and Number Needed to Treat Values Are Calculated
Image Not Available
Appendix 4. Given the Expected Absolute Risk for Postoperative Complications (0.6) and the Point Estimates of Relative Risk Used in the One-way Sensitivity Analyses, Corresponding Sample Sizes of a Future Clinical Trials and Number Needed to Treat Values Are Calculated
×
Fig. 1. A and B Model structure. (A  ) A decision tree. The short-term model starts with the decision between alternative fluid strategies (rectangle  ) followed by arrows  that represent transitions toward selected postoperative complications. (B  ) The Markov structure. The long-term model: upon entering the model, hypothetical individuals have quality of life that is associated with postoperative outcome. The individuals make transitions along the arrows  among health states or stay in the same state during 1 yr, that is, one cycle. During this cycle, survival decreases and survivors' quality of life declines by decrements. One cycle is repeated five times, representing 5 postoperative years.
Image Not Available
Fig. 1. A and B Model structure. (A  ) A decision tree. The short-term model starts with the decision between alternative fluid strategies (rectangle  ) followed by arrows  that represent transitions toward selected postoperative complications. (B  ) The Markov structure. The long-term model: upon entering the model, hypothetical individuals have quality of life that is associated with postoperative outcome. The individuals make transitions along the arrows  among health states or stay in the same state during 1 yr, that is, one cycle. During this cycle, survival decreases and survivors' quality of life declines by decrements. One cycle is repeated five times, representing 5 postoperative years.
×
Fig. 2. One-way sensitivity analyses. One-way sensitivity analyses using upper and lower level of a 95% CI of the selected model inputs, respectively. The selected model inputs were the relative risk for mortality and morbidity, prefracture values of quality-of-life index (QoL index), the decrements of quality-of-life (QoL decrements) associated with postoperative complications and hospital costs.
Image Not Available
Fig. 2. One-way sensitivity analyses. One-way sensitivity analyses using upper and lower level of a 95% CI of the selected model inputs, respectively. The selected model inputs were the relative risk for mortality and morbidity, prefracture values of quality-of-life index (QoL index), the decrements of quality-of-life (QoL decrements) associated with postoperative complications and hospital costs.
×
Fig. 3. Deterministic analyses with relative-risk point estimates for postoperative mortality and morbidity. Differences in quality-adjusted life-years (Δ QALY) are plotted against cost differences (Δ costs, €). The model was run using the baseline, and the stepwise increased the baseline values (by 25–90%) of relative risk. Appendix 3 displays the stepwise increased estimates for relative risks.
Image Not Available
Fig. 3. Deterministic analyses with relative-risk point estimates for postoperative mortality and morbidity. Differences in quality-adjusted life-years (Δ QALY) are plotted against cost differences (Δ costs, €). The model was run using the baseline, and the stepwise increased the baseline values (by 25–90%) of relative risk. Appendix 3 displays the stepwise increased estimates for relative risks.
×
Fig. 4. Results of the Monte Carlo simulation. Simulated values of incremental costs and effects (Δ costs, Δ quality-adjusted life-years, QALYs) of goal-directed hemodynamic treatment (GDHT) compared with routine fluid therapy. The slope of the dotted line  (λ) represents one possible threshold value that indicates how much society is willing to pay for one additional life-year with full health for the target population (cost-effectiveness threshold).
Image Not Available
Fig. 4. Results of the Monte Carlo simulation. Simulated values of incremental costs and effects (Δ costs, Δ quality-adjusted life-years, QALYs) of goal-directed hemodynamic treatment (GDHT) compared with routine fluid therapy. The slope of the dotted line  (λ) represents one possible threshold value that indicates how much society is willing to pay for one additional life-year with full health for the target population (cost-effectiveness threshold).
×
Fig. 5. Probabilistic sensitivity analyses. Probabilities for cost effectiveness are plotted against the cost-effectiveness threshold for the base case analysis, when the goal-directed hemodynamic treatment (GDHT) does not influence the major group of complications (other complications) and when the GHDT does not influence mortality and other complications. Other complications are listed in the section on the decision tree. QALY = quality-adjusted life-years.
Image Not Available
Fig. 5. Probabilistic sensitivity analyses. Probabilities for cost effectiveness are plotted against the cost-effectiveness threshold for the base case analysis, when the goal-directed hemodynamic treatment (GDHT) does not influence the major group of complications (other complications) and when the GHDT does not influence mortality and other complications. Other complications are listed in the section on the decision tree. QALY = quality-adjusted life-years.
×
Table 1. Model Inputs with Mean Estimates, Confidence Intervals, and Distribution
Image Not Available
Table 1. Model Inputs with Mean Estimates, Confidence Intervals, and Distribution
×
Table 2-a. Average Costs and QALYs, Incremental Costs and QALYs (Δ cost, Δ effect), and Incremental Cost-effectiveness Ratio (ICER)
Image Not Available
Table 2-a. Average Costs and QALYs, Incremental Costs and QALYs (Δ cost, Δ effect), and Incremental Cost-effectiveness Ratio (ICER)
×
Table 2-b. Average Costs and QALYs, Incremental Costs and QALYs (Δ cost, Δ effect), and Incremental Cost-effectiveness Ratio (ICER)
Image Not Available
Table 2-b. Average Costs and QALYs, Incremental Costs and QALYs (Δ cost, Δ effect), and Incremental Cost-effectiveness Ratio (ICER)
×
Appendix 1. Summary of Search Results on Meta-analyses for Postoperative Mortality
Image Not Available
Appendix 1. Summary of Search Results on Meta-analyses for Postoperative Mortality
×
Appendix 2. Summary of Search Results on Clinical Trials that Used Goal-directed Hemodynamic Treatment before Onset of Organ Failure
Image Not Available
Appendix 2. Summary of Search Results on Clinical Trials that Used Goal-directed Hemodynamic Treatment before Onset of Organ Failure
×
Appendix 3. Estimates of Relative Risk (Goal-directed Hemodynamic Treatment Compared with Routine Fluid Treatment) Used in Deterministic Sensitivity Analyses
Image Not Available
Appendix 3. Estimates of Relative Risk (Goal-directed Hemodynamic Treatment Compared with Routine Fluid Treatment) Used in Deterministic Sensitivity Analyses
×
Appendix 4. Given the Expected Absolute Risk for Postoperative Complications (0.6) and the Point Estimates of Relative Risk Used in the One-way Sensitivity Analyses, Corresponding Sample Sizes of a Future Clinical Trials and Number Needed to Treat Values Are Calculated
Image Not Available
Appendix 4. Given the Expected Absolute Risk for Postoperative Complications (0.6) and the Point Estimates of Relative Risk Used in the One-way Sensitivity Analyses, Corresponding Sample Sizes of a Future Clinical Trials and Number Needed to Treat Values Are Calculated
×