From clinical evidence to everyday practice: implementing findings from a cost-effectiveness analysis for endoscopic injection therapy for upper-gastrointestinal bleeding

BACKGROUND

A previous upper-gastrointestinal bleeding trial showed that patients treated with repeated fibrin glue injection for upper-gastrointestinal bleeding have significantly less rebleeding than those treated with polidocanol.

OBJECTIVE

To analyse the cost and effectiveness of repeated fibrin glue injection and to investigate whether these results change physicians' attitudes.

DESIGN

A retrospective random sample of five hospitals from the previous study, collection of cost identification, and follow-up data on 320 patients (155 in the polidocanol group, 165 in the fibrin glue group).

METHODS

An incremental cost-effectiveness analysis and comparison of outcomes was performed using chi-squared tests and Kaplan-Meier survival analysis. A survey was carried out using a questionnaire in the five hospitals on local guidelines for management of ulcer bleeding, and its results were analysed qualitatively. The measure of effectiveness is the number of prevented rebleedings. Further variables were length of hospital stay and length of intensive care unit (ICU) stay.

RESULTS

The cost for the prevention of one additional rebleeding by repeated fibrin glue treatment amounts to 14,316 +/- 4981 euros (incremental cost-effectiveness ratio). There were no significant differences in length of stays in ICU or in hospital. The physicians did not change their management plans for patients with upper-gastrointestinal bleeding. In a survey, it was seen that other factors, such as local guidelines, attitudes towards new treatment options, and ease of handling of drugs, are more important than a result of a single study for a behavioural change of the doctors.

CONCLUSIONS

The study was not designed prospectively to address a pharmacoeconomic question. As relevant variables (e.g. length of ICU stay) could not be reliably ascertained retrospectively, this may lead to biased estimates of the incremental cost-effectiveness ratio.

[Outcome of minimally invasive surgery. Qualitative analysis and evaluation of the clinical relevance of study variables by the patient and physician]

INTRODUCTION

Mechanistic study endpoints, evaluated exclusively by the physician, are mostly used in clinical studies evaluating new treatment modalities (e.g. laparoscopic cholecystectomy). Those endpoints often lack clinical relevance. The patient's opinion concerning the importance of a study endpoint is particularly important in the evaluation of minimally invasive procedures, which place special emphasis on patient comfort.

METHODS

In a first step it was evaluated by meta-analysis, which clinical endpoints have been used for comparison of laparoscopic and conventional cholecystectomy. Furthermore, using a qualitative analysis it was investigated how important the individual study endpoints are for patients and physicians. Ten patients and five surgeons were questioned in a structured interview.

RESULTS

Of all outcome variables used world-wide, approximately one third were hermeneutic study endpoints, depending on the quality of the study, but often the method of evaluation was insufficient. Only three of 215 endpoints (< 2%) were quality of life scores, an integrated concept of outcome was missing completely. The qualitative analysis confirms the claimed difference between isolated and integrated evaluation of treatment goals. The importance of postoperative death is underestimated by patients and physicians; postoperative pain is overestimated. Patients ranked the outcome variable "restoration of full physical fitness" as the most important study endpoint after avoidance of complications and death. It is underestimated in isolated evaluation and has not been used in the world literature at all.

CONCLUSION

The analysis of clinical relevance of study endpoints should be the first and not the last step of studies to evaluate surgical technology. It cannot be based purely on intuition; it must make use of scientifically accepted techniques (e.g. qualitative analysis).

Second step: testing-outcome measurements

Despite worldwide enthusiasm for endoscopic surgery, this new technology is now on the top of McKinlay's "product life circle curve." Critical questions are being asked about its benefits and burdens, but the concepts applied and the methodologies used for technology assessment are in a similar position as endoscopic surgery and need a critical evaluation. (1) There are incorrect and outdated concepts for the scientific basis of surgery (surgical theory) including the basic sciences involved; biomedicine still dominates, but assessment of outcome after operations is no longer possible without clinical epidemiology and social psychology. (2) Based on an outdated scientific theory for surgery, an outdated concept of disease is still propagated. It is denoted as mechanical and is based solely on biomedicine. Human subjects are reduced to biologic machines, and outcomes measurement excludes most dimensions of functioning and well-being. To achieve a valid result for outcome measures, a hermeneutic approach must be combined with the mechanical approach. (3) Based on an outdated model of disease, the outcomes used in endoscopic surgery rely too much on traditional measures, such as mortality rate, complication rate, hospital stay, and especially an endless list of biochemical mediators. Their alterations during the perioperative period have not yet been shown to be related to clinical or hermeneutic outcomes. A new method of assessment for clinical trials in endoscopic surgery and for other surgical problems is outlined, such as for surgical infections and for surgical oncology. It includes an index of recovery and objective health status assessed by the doctor, a quality-of-life self-report by the patient, and the true endpoint concept as a critical weighting of both types of outcome by patients and doctors.

[Prediction of rehabilitation outcome after surgical interventions based on biochemical parameters: a new concept in mediator research]

The perioperative changes in epinephrin-, norepinephrine-, histamin-, C5a- and interleukin-6-levels were studied in 40 patients undergoing cholecystectomy for the diagnosis of acute cholecystitis. All relevant mediator levels could be determined in 38 patients. The outcome was not optimal in 16 of them (42%). In order to evaluate the predictive value of the mediators under investigation for the quality of the patients' outcome, a model based on the Bayes' theorem was developed. Using this model the outcome (optimal vs. not optimal) could be correctly predicted in 30 (79%) of our study patients. This kind of data analysis allows to define states of increased risk for a not optimal recovery based on biochemical parameters.

Effect of gastrin receptor blockade on endocrine cells in rats during achlorhydria

Hyperplasia of the oxyntic enterochromaffinlike cells in response to long-lasting blockade of acid secretion is closely related to hypergastrinemia. In the present study, the effect of a specific gastrin receptor antagonist on proton pump inhibitor-induced changes on serum gastrin levels, mucosal height, as well as gastrin- and enterochromaffin-like cells was investigated in rats. The proton pump inhibitor BY 308 or the vehicle methylcellulose [Methocel (controls)] was administered for 2 weeks in the presence and absence of the gastrin receptor antagonist PD 136450 (CAM 1189). BY 308 significantly increased serum gastrin levels, gastrin cell density, and antral gastrin concentration. Concomitant application of PD 136450 did not alter this response. In the oxyntic stomach, mucosal height, enterochromaffinlike cell density, labeling index of enterochromaffinlike cells, and histamine concentration were elevated after treatment with BY 308. These increases were almost completely abolished by PD 136450. Even in normogastrinemic control rats, PD 136450 significantly decreased mucosal height of the oxyntic part of the stomach and the labeling index of enterochromaffinlike cells. The results show that (a) trophic effects of drug-induced achlorhydria are mediated by gastrin; (b) even in control rats (normogastrinemic), gastrin is a trophic factor for the oxyntic mucosa; and (c) antral gastrin cell hyperplasia in states of chronic achlorhydria is not mediated by gastrin itself.