Objective: The U.K. NHS Economic Evaluation Database (EED) project is commissioned to identify papers on economic evaluations of health technologies and to disseminate their findings to NHS decision makers by means of structured abstracts that are available through a public database and the Cochrane Library. This paper discusses current issues relating to the economic aspects of producing NHS EED abstracts.

Methods: A review of NHS EED was undertaken between 1994 and 1999 to determine the methodologies adopted and issues that influence the usefulness of economic evaluations. Methods adopted to improve the quality of NHS EED abstracts are also reported.

Results: Eighty-five percent of NHS EED abstracts are cost-effectiveness analyses (CEAs), 9.3% are cost-utility analyses (CUAs), and only 1.4% are cost-benefit analyses (CBAs). Of the total abstracts, 65.9% are based on single studies, 19.5% on reviews, 3.9% on estimates of effectiveness, and 10.7% on combinations of these sources. Models are utilized in 16.7% of CEAs, 60.2% of CUAs, and 20% of CBAs. Analyses of CBA studies reveal a degree of misuse of well-established definitions. NHS EED internal control mechanisms are reported that provide a means of ensuring that abstracts are based on sound academic principles.

Conclusions: Most economic evaluations are conducted by means of CEA, followed by CUA, while CBA accounts for an extreme minority of cases. Single studies form the principal source of effectiveness data, although models are widely used, principally in CUA. The structure of NHS EED abstracts provides decision makers with the principal results and an interpretation of the relative strengths and weaknesses of economic evaluations.

The NHS Health Technology Assessment (HTA) Programme runs an annual process of identifying suggestions for health technology assessment. The objective of this paper is to describe and evaluate the relative importance of the different sources used by the program in 1998 to identify potential priorities. There were four different sources: a) a widespread consultation of healthcare commissioners, providers and consumers; b) research recommendations from systematic reviews; c) reconsidering previous research priorities which had not been taken forward for funding; and d) horizon scanning. Collectively, the four sources generated just over 1,100 HTA suggestions. By far the largest source of suggestions and priorities was the widespread consultation. However, the success rate of this source, in terms of being commissioned, was low. Research recommendations from systematic reviews provided the second largest source of priorities and the best success rate of all sources. Value was found from different sources for different healthcare areas.

Objective: To estimate the cost-effectiveness (CE) of interferon beta-1b (IFNβ-1b) in slowing disability progression in persons with relapsing-remitting multiple sclerosis (RRMS).

Methods: Treatment program costs and health outcomes are modeled for cohorts of 1,000 females and 1,000 males followed 40 years from onset. Fifteen scenarios model MS natural history progression, treatment efficacy, direct treatment costs, and MS healthcare costs. A single randomized placebo-controlled trial of IFNβ-1b found reduced disease activity by MRI, reduced frequency and severity of exacerbations, and a tendency toward slower disability progression.

Disability years avoided are modeled as the primary health outcome analyzed. A ministry of health (MOH) perspective is adopted, using Nova Scotia population-based data. Annual IFNβ-1b direct treatment costs (Can $16,685) are high relative to both MOH healthcare costs per person with MS (Can $2,000) and estimated MOH costs avoided.

Results: Given “reference case” assumptions for women with RRMS, treatment reduces lifetime disability years by 10%. Cost per disability year avoided before discounting is Can $189,230 (US $124,892), and Can $274,842 (US $181,395) after discounting at 5%. Estimates for alternative scenarios vary greatly, leaving main findings unchanged.

Conclusions: Using the Expanded Disability Status Scale, cost per disability year avoided due to interferon beta-1b treatment in RRMS is quite high. Comparable CE estimates, using MS-specific or generic health-related quality-of-life outcome measures, are even higher. Further research is required to better measure treatment effects, modification of MS natural history, and net societal costs of IFNβ-1b in RRMS.

Objectives: Interferon beta-1b has recently become available for the treatment of secondary progressive multiple sclerosis (SPMS). This study aims at estimating the cost-effectiveness of this new treatment that has been shown in a clinical trial to reduce the progression of the disease. Effectiveness is measured as the number of quality-adjusted life-years (QALYs) gained from the reduction in progression. Because the clinical trial period will only capture part of the treatment's effect in terms of QALYs gained, since benefits achieved during the trial will have an effect beyond it, the cost-effectiveness analysis involves modeling over the longer term using complementary data.

Methods: A Markov model with states based on disability expressed by EDSS scores was used. Transition probabilities were calculated directly from clinical trial data for the first 3 years and then extrapolated to 10 years. Mean costs and utilities for each Markov state were calculated from a population-based cross-sectional study in Sweden.

Results: The incremental cost per QALY is SEK 342,700 (US $39,250;US $1 = SEK 8.73, March 10, 2000) when all costs (direct, informal care, and indirect) are included (discounted 3%). When indirect costs are excluded, the cost per QALY is SEK 542,000 ($62,100).

Conclusions: Cost-effectiveness analysis in SPMS requires that the effect of treatment beyond clinical trials be included. Also, analysis should be done from a societal perspective, since many of the costs occur outside the healthcare system. The cost-utility ratios estimated in this analysis are at or below the mean threshold value indicated in a recent survey of health economists ($60,000).

Objective: To analyze the rates and influences on the adoption of three selected health technologies: statins, coronary stents, and magnetic resonance imaging (MRI).Methods:A retrospective diffusion study using primary care prescribing data and questionnaire responses from acute hospital trusts in the West Midlands region (population 5.3 million or 10% of England).

Results: The selected technologies had markedly different diffusion curves. Statins diffused rapidly soon after launch. Coronary stents were initially used 6 years after first availability, but within 2 years all responding hospitals reported using them. MRI scanners were initially purchased 6 years after first availability with a subsequently slow rate of diffusion, and are still absent from some hospitals. Influences on the adoption of each technology were different. Commercial marketing was reported as a major influence on the diffusion of statins but not at all on MRIs. Cost impact was a major negative influence on the diffusion of MRI scanners and statins, whereas enthusiastic individuals were key to the diffusion of stents.

Conclusions: Influences on adoption and consequent diffusion rates are very different for different health technologies. It is not at all clear that such diffusion patterns relate well to an optimum timing rate. This has important implications for technology gatekeepers in health care.

Objectives: This paper considers the methodological problems that arise in conducting cost analyses in economic evaluations where only observational (rather than experimental) data are available and where the technology being evaluated is diffuse, such that the unit of analysis has to be the institution rather than the patient.

Methods: A case study is reported that concerns the application of computer technology in radiology: picture archiving and communication systems (PACS). A range of different approaches were used to estimate changes in running costs, including time series analyses of routine data and direct observation of resource use.Results:The analysis illustrates some of the difficulties involved in costing the introduction of a diffuse technology. Nevertheless, it provides a firm indication that, overall, the introduction of PACS was found to be associated with a significant increase in hospital costs, suggesting that the initial expectations of financial savings were unduly optimistic.

Conclusions: The research demonstrates that, using multiple methods, it is possible to estimate cost changes within a single hospital. In addition, the paper discusses the nature of the uncertainties in such analyses and possible ways of representing such uncertainty in terms of confidence intervals.

Objectives: Guidelines for colorectal cancer screening and surveillance in people at average risk and at increased risk have recently been published by the American Gastroenterological Association. The guidelines for the population at average risk were evaluated using cost-effectiveness analyses.

Methods: Since colorectal cancers primarily arise from precancerous adenomas, a state transition model of disease progression from adenomatous polyps was developed. Rather than assuming that polyps turn to cancer after a fixed interval (dwell time), such transitions were modeled to occur as an exponential function of the age of the polyps. Screening strategies included periodic fecal occult blood test, flexible sigmoidoscopy, double-contrast barium enema, and colonoscopy. Screening costs in 1994 dollars were estimated using Medicare and private claims data, and clinical parameters were based upon published studies.

Results: Cost per life-year saved was $12,636 for flexible sigmoidoscopy every 5 years and $14,394 for annual fecal occult blood testing. The assumption made for polyp dwell time critically affected the attractiveness of alternative screening strategies.

Conclusions: Sigmoidoscopy every 5 years and annual fecal blood testing were the two most cost-effective strategies, but with low compliance, occult blood testing was less cost-effective. Lowering colonoscopy costs greatly improved the cost-effectiveness of colonoscopy every 10 years.

Objective: Economic evaluations such as cost-effectiveness and cost-utility analyses generally fail to incorporate elements of intangible costs and benefits, such as anxiety and discomfort associated with the screening test and diagnostic test, as well as the magnitude of utility associated with a reduction in the risk of dying from cancer. This paper seeks to include all costs and effects incurred by introducing mammography screening through the application of discrete ranking modeling.

Methods: In the present analysis, 207 women were interviewed and asked to rank, according to priority, a number of alternative breast cancer screening setups. The alternative programs varied with respect to number of tests performed, risk reduction obtained, probability of a false-positive outcome, and extent of copayment. Using discrete ranking modeling, the stated preferences were analyzed and the relative weighting of the program attributes identified. For a range of hypothetical breast cancer programs, relative utilities and corresponding willingness-to-pay estimates were derived.

Results: A comparison of cost and willingness to pay for each of the programs suggested that net benefits are maximized when screening person aged 50–74 years biennially. More intensive screening produces lower or similar levels of utility at a higher cost.

Conclusion: Discrete ranking modeling can aid decision making by identifying inferior healthcare programs, i.e., programs that are more costly but less beneficial.

Objective: To define the relative benefits of screening for diabetes and improved treatment programs and ways of improving the efficiency of screening for a population-based cohort derived from the Third National Health and Nutrition Examination Survey (NHANES III).

Methods: A Markov decision model is used to estimate microvascular benefits of glucose control for four different screening and treatment scenarios, including either universal screening or improved glucose control of known diabetic subjects, neither, or both.

Results: A population cohort of subjects with recent onset of diabetes (<5 years) was derived from NHANES III (of whom close to half were unaware that they had diabetes). In this population-based cohort, the total benefit achievable by universal screening and improved treatment (limiting HbA1c to less than 9%) is a reduction of about 30,000 cases of blindness over the lifetime of the cohort. Screening alone results in 7% of this benefit, and improved treatment alone provides 65%. Screening a targeted group of patients with three or more risk factors for developing diabetes would reduce the number of required fasting glucose measurements needed by 82% and provide 50% of the total benefit of screening the entire population with a fasting glucose measurement.

Conclusions: Morbidity from type 2 diabetes can be most effectively reduced by developing ways to modestly improve the glycemic control of known diabetic subjects, particularly those with high A1c's and early onset of disease. Targeting can significantly reduce the number of persons who need to be screened with a fasting blood test while preserving a large component of the benefit of screening.

Objectives: The evaluation of the diagnostic value of endoscopic procedures usually lacks a gold standard when performed in cross-sectional studies. The objective is to demonstrate an easily applicable method to assess the possible range of sensitivity, specificity, and predictive values of endoscopic procedures in the absence of a gold standard method.

Methods: Data from a study of 328 endoscopies comparing two different methods to diagnose superficial bladder cancer were used as a numerical example. Both endoscopic procedures were performed in the same patients in one session. Under the assumption of a systematic misclassification process, a model to correct sensitivity estimates is developed.

Results: The lowest possible sensitivity estimate for a new fluorescence endoscopy technique (FE) was 78%, the maximum 97.5%. Depending on realistic assumptions made upon the misclassification, a reasonable estimate for sensitivity was 93.4% (95% confidence interval [CI]: 90%–97.3%) for the FE technique. The sensitivity of the traditional white-light endoscopy method ranged from 47.2% to 53%, with a reasonable estimate of 46.7% (95% CI: 39.4%–54.3%).

Conclusions: This method to determine the theoretically possible range of sensitivity estimates in endoscopic procedures is helpful in cross-sectional studies with a missing gold standard method. It is easily applicable for a variety of endoscopic procedures, including upper and lower gastro-intestinal tract, urogenital tract, or diagnostic laparoscopic surgery.

Objectives: The question addressed here is whether home rehabilitation after stroke is better and/or less expensive than the more conventional alternatives, i.e., rehabilitation during inpatient care, day care, and outpatient visits—alone or in combinations appropriate to disease stage and patient needs. Home rehabilitation is managed by teams of professionals who train patients at home.

Methods: The scientific literature was systematically searched for controlled studies comparing outcomes and costs of home rehabilitation with the more conventional strategies.

Results: The abstracts of 204 papers were evaluated, from which 89 were selected for greater scrutiny. From the 89 studies, we found 7 controlled studies involving 1,487 patients (6 of the 7 were randomized, 4 of the 6 assessed costs). No statistically significant differences, or tendencies toward differences, were revealed as regards the outcome of home rehabilitation versus hospital-based alternatives. Thus, home rehabilitation was neither better nor worse at improving patients' ability to manage on their own or resume social activities. Depression and reduced quality of life were common in all groups of patients and caregivers, irrespective of the rehabilitation strategy. In the four randomized studies that reported on costs, home rehabilitation was found to be less expensive than regular day care, but not less expensive than conventional strategies even though hospital stay was reduced.

Conclusion: The outcomes and costs of home rehabilitation after stroke seem to be comparable to alternative treatment strategies.

Objectives: To compare the cost-effectiveness of a multidisciplinary rehabilitation program with traditional treatment within primary care in terms of the health-related quality of life (HRQOL) in patients with prolonged musculoskeletal disorders (MSD) on the one hand and the costs to society on the other. Predictors of total costs, such as motivation, socio-economic level, age, pain, and working environment, were investigated.

Methods: A prospective, matched, controlled 2-year follow-up study was designed. The main outcome measures were HRQOL using the Nottingham Health Profile (NHP) and patient-specific total costs due to society. The cost-effectiveness was expressed as a quotient of the total costs to society/NHP global score difference value.

Results: Patients with prolonged MSD generate substantial total costs to society, chiefly in the area of indirect costs. Multidisciplinary rehabilitation improved HRQOL more cost-effectively. Motivation was revealed as a predictor of total costs. The relationship in savings in terms of indirect costs between the highly-motivated and the less-motivated patients was calculated at 4:1.

Conclusions: The large group of patients with prolonged MSD generate substantial total costs, and even small reductions in direct and indirect costs could be of importance to society. The multidisciplinary rehabilitation program applied here was more cost-effective as compared with conventional treatment within primary care when it came to improving the patients' perceived HRQOL. Motivation could be a predictor of total costs, which has to be addressed in the process of identifying the patient as a partner in the rehabilitation process.

Objective: To assess the effect of the use of cochlear implants (CI) on the health status of postlingually deaf adults.

Methods: Participants comprised 45 postlingually deaf adult multichannel CI users and 46 deaf candidates on the waiting list for a CI. The latter group acted as control subjects to corroborate the validity of retrospective completion of the questionnaires by the CI recipients. Three HRQOL instruments were used: a) a specially developed CI questionnaire (NCIQ); b) a generic HRQOL questionnaire (SF-36); and c) a health-state classification system (HUI-2) suited to estimate single preference scores.

Results: Retrospectively estimated pre-implant scores in the CI user group corresponded very well with the scores in the control group. Postimplant scores in the CI users were substantially higher in all six domains (p < .001) of the NCIQ than the scores in the control group. Effects due to a CI were also observed with the SF-36 in five of the seven domains (p < .01). Statistically significant differences between the two groups (p = .001) were observed in two of the six domains of the HUI-2.

Conclusions: All three questionnaires detected improvements in HRQOL due to CI use. To make a detailed assessment of the effect of a CI on functional outcomes and well-being, a special purpose HRQOL instrument is far more adequate than a general HRQOL instrument. This study also showed that a CI affects several other health domains besides auditory performance. The effect of CI use on general functioning and well-being proved to be considerable.

Objectives: To compare outcome attributes of three available technologies for stem cell transplantation—bone marrow transplantation (BM), peripheral blood cell transplantation (PB), and cord blood transplantation (CB). To compare the cost per recipient for the three technologies, incorporating all donor and antigen matching functions, as well as the transplantation procedures.

Methods: Review of relevant literature using the CancerLit and HealthSTAR databases (plus earlier searches of MEDLINE and Embase) to summarize outcome differences. Estimations of cost per transplant using models drawing on Canadian experience.

Results: The literature review indicated that PB may have a modest advantage over BM in terms of outcomes. The literature indicated that CB has longer engraftment times (although this may be due to inadequate controls). Our modeling exercise indicated that costs for PB and BM are similar. CB has a higher cost per recipient, due in large part to the higher testing costs for every donated unit.

Conclusions: Current literature is inadequate to provide an outcomes comparison of CB, PB, and BM. Viewed from a system-wide perspective, CB is a higher cost technique. However, PB and CB are emerging as technologies, and so our results should be viewed alongside the notion that there may be a steep learning curve in these technologies.

Objectives: To examine determinants of use of cardiac procedures after acute myocardial infarction and identify variation factors.

Methods: Observational prospective cohort study of 2,519 patients in 48 centers with a two-level logistic-regression analysis.

Results: Elderly patients were less likely to undergo pre- and inhospital thrombolysis (odds ratios, 0.71 and 0.64; 95% CI, 0.62–0.81 and 0.58–0.69, respectively). The elderly, females, and patients with heart failure on admission were less likely to undergo noninvasive tests (0.74, 0.62, and 0.51; 95% CI, 0.67–0.81, 0.46–0.83, and 0.38–0.68, respectively) and coronary angiography (0.38, 0.53, and 0.67; 95% CI, 0.34–0.42, 0.38–0.74, and 0.52–0.86, respectively) but not revascularization. Hospital factors were more difficult to interpret.

Conclusions: Elderly, women, and heart failure patients underwent fewer cardiac procedures than lower-risk patients. Physicians should change their attitude toward these groups and use advanced procedures, bearing in mind the patients' needs rather than good procedural outcomes.

Objectives: Since 1988, four government-funded health technology assessment (HTA) agencies have been established in Canada. This paper is a descriptive review of reports issued by these organizations during the period from 1988 to 1998.

Methods: Publications from the national and three provincial HTA agencies in Canada were obtained and reviewed. Only the 117 assessment reports that were reported to have undergone external review were included in this analysis. Each report was classified on a standard abstraction form according to criteria such as technology type(s), assessment focus, whether a specific policy question was clearly stated and relevant decision maker(s) identified, description of search strategy and selection criteria, sources of data and assessment methods used, whether recommendations or conclusions were made, and duplication or overlap of reports. The trends in these qualities over the 10-year period were also examined.

Results: Therapeutic technologies have received the most attention from all four agencies, although the focus on devices, drugs, and procedures varied between agencies. The policy question under investigation was clearly identified in 82% of reports, and 71% clearly identified the decision maker toward whom the assessment was targeted. Efficacy or effectiveness was examined in 81% of reports, and costs were assessed in 65% of studies. These were the two most frequently examined aspects. Almost all assessments were descriptive literature reviews; 9% included meta-analyses and 32% had cost analyses or economic evaluations. Forty-four percent of reports had a clear description of the literature search strategy, and selection criteria were clearly specified in 38% of studies, but there was considerable variation among agencies in the level of description of these methods. Conclusions were clearly stated in 83% of the assessments' conclusions, and 13% had recommendations. When analyzed longitudinally, it is apparent that the quality of reports has improved markedly during the past decade. This was determined by examining the clarity of specifying the policy question(s) under investigation, the identification of the target audience of decision makers for the information, and by evaluating the thoroughness of the description of the methods used in the assessment.

Conclusions: Canadian government agencies have contributed a considerable quantity of health technology assessments. There has been very little duplication of technologies evaluated, and the quality of the assessment reports has markedly improved during the past decade.

Though the desire to live is fundamental and it is never easy to accept death, a full acceptance is based on the understanding that death is an inherent part of all life and is necessary for the continuation and vitality of species. (1)

Objectives: To describe the types of report that are cited by protocols and reviews included in the Cochrane Database of Systematic Reviews.

Methods: The citation for each reference included in the Cochrane protocols and reviews published in the Cochrane Database of Systematic Reviews in issue 1, 1999 of The Cochrane Library were categorized by reference type (e.g., journal article, conference proceeding, book chapter, personal communication, Cochrane review).

Results: Of a total of 24,913 citations, 21,694 (87.1%) were references to journal articles. There was a significant difference between the proportion of references to studies that were journal articles (12,348 of 13,472; 91.7%) and the proportion of other references in this category (9,346 of 11,441; 81.7%).

Conclusion: The great majority of studies included in Cochrane reviews at the beginning of 1999 had been published as journal articles.

Objective: The rise in the number of implantations of cardiac pacemakers is of some concern to decision makers in the health sector. We assessed the intrinsic and relative clinical efficacy of cardiac pacemakers in current clinical indications to find out whether scientific or clinical arguments might justify differences in market prices.

Methods: We retrieved papers on cardiac pacing (January 1993–April 1998) from five databases (MEDLINE, HealthSTAR, EMBASE, Cochrane Library, and PASCAL). The citations in these papers were used to seek further articles. We selected the articles that met the criteria of evidence-based medicine (EBM) (randomized and nonrandomized controlled trials) and classified them according to clinical indication and type of evaluation (either of the intrinsic efficacy of a pacemaker versus a control or of the relative efficacy of different pacing modes).

Results: A total of 542 references were retrieved, but under 10% met our EBM criteria. Very few were comparative studies versus controls; most were recent and tended to use endpoints other than survival. Clinical efficacy was not proven on the basis of EBM criteria, even in common indications (e.g., sick sinus syndrome). Studies comparing different pacing modes were rarely randomized and did not provide consistent evidence for the superiority of any pacing mode in a given indication.

Conclusions: Knowledge of the natural history of the diseases for which cardiac pacing is indicated is scarce. There is an approximately 20-year gap between technological progress and clinical evaluation that cannot be easily bridged because of methodologic difficulties and ethical issues. Current guidelines on pacemaker use either rely on expert opinion or highlight present inadequacies and make recommendations for future work. Available clinical efficacy data do not justify the wide differences in the price of cardiac pacemakers.

[bull ] Many interventions in health and health care are implemented at the level of geographic area or health service organizational unit. Such interventions will be implemented for entire clusters of healthy subjects or patients.

[bull ] Evaluation of cluster level interventions may be difficult because only a small number of clusters may be available for study and the responses of individuals are correlated within clusters.

[bull ] This review synthesized current methodologic recommendations into a checklist for investigators:

1. Recognize the cluster as the unit of intervention or allocation;

2. Justify the use of the cluster as the unit of intervention or allocation;

3. Include a sufficient number of clusters;

4. Randomize clusters wherever possible;

5. In nonrandomized studies include a control group;

6. In single group studies include repeated measurements over time;

7. Allow for clustering when estimating the required sample size;

8. Consider matching or stratifying clusters where appropriate;

9. Choose between repeated cross-sectional and cohort designs in studies with follow-up;

10. Allow for clustering at the time of analysis;

11. Allow for confounding at both individual and cluster level; and

12. Include estimates of intraclass correlation and components of variance in published reports.

[bull ] The present level of adherence to these recommendations is low.