How best to allocate limited economic resources is becoming an ever increasing problem in health care. Two factors mainly influence the increased demand in this field: the technological progress, that has allowed industry to offer more and more sophisticated instruments to health care operators, and the increased life expectancy. Increased demand and finite resources call for the imposition of controls to avoid wasting resources. Many health care situations require an economic evaluation, i.e. an analysis that helps patients, payers, and providers make rational medical care-related choices based on a better insight into the effect of these choices on the patient's life and on the resource allocation. In recent years the literature on this issue has grown exponentially. However, methods of gathering and synthesizing data on health care outcomes and effectiveness are underdeveloped, as are systems of providing the available information to physicians, patients and health care organizations.
According to a fairly well established terminology, studies in this field aiming at comparing alternative options in terms of both efficacy and resource allocation, may be classified into four categories:
In this proposal we will refer to any of the above types of analyses by using the term economic evaluation, while the term cost evaluation will indicate the activity of gathering and analyzing data with the aim of calculating the monetary costs of a given health care program.
The above four categories of economic evaluation analyses differ only on the basis of the measure of the consequences. Thus it is feasible to develop generic models of economic evaluation problems, which can be specialized into one of these categories by defining the needed measure of consequences, without changing their structure (consequently, in the following, if no further detail is given, the term cost/effectiveness ratio, abbreviated as C/E ratio, will indicate either cost/benefit, or cost/effectiveness or cost/utility ratio).
Cost evaluation task can be viewed 1) in association with economical evaluation, as both a preliminary step necessary for providing reliable probability distribution on cost involved by alternative decisions, and a continuous process necessary to monitor costs, that could change in time following behavioural changes suggested by the economic analysis, and 2) as a ``stand-alone'' task, that can be useful simply for assessing the cost of management of a class of patients or of a given procedure, without necessarily involving a comparison with alternative procedures.
Application domains of economic evaluations in health care concern the design and assessment of screening programs [1, 5], clinical trials [4], the comparison of therapeutic interventions [2, 3], and the purchase of expensive instruments [6]. Typical problems deriving from these domains are the following:
- to decide which cases to enroll for a screening program, for example by finding a treshold value for some patient's characteristics, in order to focus the screening on a population that is likely to yield a good C/E ratio;
- to decide which tests to use for implementing a diagnostic screening, in order to balance predictive power with costs;
- to know how many cases are needed in a clinical trial for demonstrating a clinically significant difference between two treatment strategies: it is important to avoid over-sizing of trial arms, because of the risk that trial cost outweigh the potential benefit;
- to build predictive models based both on past experience and literature results for estimating outcomes of one of more treatments, and for combining outcomes with someone (patient, physician, familiars,...) preferences;
- to include economic models for assessing the convenience of an instrument purchase, for example for knowing the minimum number of patients needed for amortizing the instrumentation cost investment.
Several problems have emerged concerning the interpretation of the results reported in this type of studies, which are mainly due to lack of methodological standardization in the area. As Drummond et al. outline in their book [7], studies on the same issue may greatly differ for cost evaluation, point of view of the analysis, and evaluation of the consequences of the programs being investigated. Of course, it is true that a complete standardization is impossible in this field because of differences among health care organizations in different countries, and because some aspects of evaluation are still very controversial (e.g. the discount of future life years, the use of different utility measures and the inclusion of indirect costs and benefits, which can be difficult to measure). However, in a recent paper [8], Drummond discusses the need for 'some' standardization, which is mainly based on three motivations: 1) maintaining methodological standards, 2) facilitating the comparison of the results of economic evaluations for different health care interventions, and 3) facilitating the transfer of experiences between different settings. The standardization issue is becoming very important, especially because methodologies such as meta-analysis provide tools for pooling different studies, with the aim of reaching more reliable results. (The European Community has also recently shown interest in standardization, in a slightly different field, by diffusing a European Guideline for clinical trials [9]).
Two additional points must be considered. The first one is the multidisciplinarity of these studies. While it is typically the case that physicians and health economists are involved, also statisticians, in particular epidemiologists, often play a role in assessing the efficacy of health programs. Moreover psycologists and sociologists can also be involved in trying to assess scales for the evaluation of life quality. Most studies do not highlight the roles of the different professional figures, and do not describe the basic tools used for their inferences (i.e.data base, statistical packages, source of cost descriptions, etc.). We believe that making these aspects more explicit could help the communication among the different health care professionals. The recognition of the multidisciplinarity of economic evaluation studies will be a fundamental assumption of the project and will be reflected in the architecture of the software and the design of the ontologies.
The second point is the high rate of obsolescence of these studies. As a matter of fact, the rapid progresses in medicine and technology often produce variations in clinical practice that invalidate the hypotheses on which studies were previously based. Therefore, it is essential to ground these analyses on a general architecture incorporating models that are allowed to change, both in structure and quantification, in such a way to follow evolutions of the real setting.
In the following a set of applications is outlined, each one addressing different aspects of economic evaluations as listed above.
The choice of conditioning regimen for bone marrow transplantation
In patients with acute myeloid leukemia, busulfan and total body irradiation (TBI) can be considered alternatives. The trade off can be summarized as follows: busulfan is very much cheaper than TBI (about 100 ECU vs 800-1300 ECU), but it can produce early adverse effects, whose treatment is very expensive. The efficacy of the two treatments can be considered similar, even if incidence of severe infections in the two cases requires further investigation. In addition, the problem must take into account two different scenarios: in the first one, the TBI can be performed within the hospital where the patient has been admitted (this implies the existence of an accelerator and adequate structures and equipes); in the second one, the TBI treatment requires to transport the patient in another hospital. The two scenarios differ very much for what concern cost components. In addition, one useful outcome of the analysis would be an assessment of the conditions needed in order to make convenient the purchase of TBI equipment when it is not available yet (for example, it could be useful to know how many treatments must be performed per year in order to amortisize the TBI cost within a given period).
Screening for an occult cancer
It is known that patients with idiopathic deep venous thrombosis (IDVT) show an increased risk, with respect to the general population, of manifesting a cancer within an interval lasting about from six months to two years after IDVT diagnosis. This suggests that IDVT may be caused by an occult cancer. Given the relatively old age of these patients (median age is 64 years), and given that there is not a preferred tumor site, the question is ``is it worthwhile, in terms of survival, submitting these patients to a screening for occult cancer, how many types of cancers have to be searched for, in which sequence, and by which tests? Which is the cost per each life year gained, if any?'' Effectiveness of screening programs depends on several variables, for example prevalence of the diseases, benefits of early diagnosis, test availability. The choice of the tests sequence is one of the main problems to face when more than one test is available for investigating a disease. As a matter of fact, tests differ for invasiveness, cost, sensitivity and specificity, thus their sequence may affect the final outcome. When the screening program concerns multiple diseases, as in the case proposed here, the sequence of their investigation is another choice variable.
Clinical trial design
The number of research proposal for clinical trials submitted to health care agencies is rising faster than the money available to fund them. Decision about funding trials must weigh the potential gains in clinical benefit against the cost of the trial implementation. The cost is highly dependent on the number of subjects enrolled in the trial and in the number of follow-up visits for each subject. Thus, a decision analysis could have the purpose of 1) deciding how many subjects are to be enrolled in order to achieve a statistically significant difference (where the significance level may be deicided by the decision maker) between the outcomes in the different trial arms and 2) deciding the timing of subjects' follow-up . The model will take into account probabilistic parameters such as the number of years citizens will benefit from the trial results, prevalence of diseases to which trial results could be applied, the cost of therapeutic changes involved by the trial outcomes, etc. As an application field, we propose the ``splenectomy in mild hereditary spherocytosis'': in this hematological disease, often splenectomy is performed with the only purpose of reducing the risk of developing gallstones (the risk is due to the high amount of bilirubin coming from the incresed red cell catabolism). However, there is no general agreement on the convenience of such behaviour.
Use and Misuse of Antibiotics
The mortality for patients with serious bacterial infections is 20-40or not the antibiotics treatment, initiated during the first days of the infection, is covering. i.e. if it is effective against the pathogen which is causing the infection. The mortality of patients who do not receive a covering treatment with antibiotics is twice as high as the mortality of those patients who do [12,13, 14]. In spite of the great advantage which can be obtained by use of a covering treatment, a treatment which does not provide maximum coverage is often chosen. This may be traced back to the fact that the costs of an antibiotics treatment with maximal coverage exceed the benefits of treatment. The dominating contribution to cost is not economical, but rather ecological: Consistent use of broad-spectred antibiotics which provide a high probability of coverage promotes the development of resistant bacterial strains. This compromises the possibility of providing effective antibiotic treatment for future patients.
The balancing of therapeutic benefit against the ecological cost yeilds very different results in different countries. In the Scandinavian countries, the antibiotics policy is very conservative with severe restrictions on the use of broad-spectred antibiotics, while the policy is more liberal in Souther Europe and in particular in Israel. This gives very different profiles of resistant microorganisms, but despite these differences 35-45patients in all countries receive inappropriate treatment, either due to the fact that the pathogenic microorganism is resistant to the chosen treatment, or that the chosen treatment is unnecessarily broad-spectred and consequently has an unnecessarily high ecological cost.
The purpose of the project is to develop statistical models for the resistance of microorganisms, and for the mortality of patients with severe bacterial infections. The statistical models can be adjusted to regional epidemiological profiles. Large data bases with microbiological data are available for several European countries and from Israel. Such models make it possible to evaluate the cost/benefit of different national antibiotics policies.
Use of the PET in hearth failure
Patient suffering from heart failure may benefit from Positron Emission Tomography (PET). PET allows to study, in addition to blood flow, also the heart tissue metabolism, and this is very useful to decide further interventions. As a matter of fact, efficacy of revascularization surgery, such as bypass, depends greatly on tissue residual vitality. When tissue is no longer vital, alternative treatments, such as farmacological treatment and heart transplantation, may be considered. The ability of PET of addressing patients to the most appropriate treatment has been demonstrated for well-defined classes of patients, characterized by low left ventricular ejection fraction and wall motion. However, PET technology is complex and it requires highly specialized operators, instrumentation (mainly the cyclotron) requires adequate rooms and safety measures, and cost is very high. In addition, parallel development of other technologies, such as SPECT, less accurate, but with wider application spectrum and lower cost, imposes very careful economic evaluation.
Next: Market situation and prospects Up: User needs and application Previous: User needs and application