[Effect models and meta-analysis]

The effect models are defined as the simple or complex relation that the risk in the treatment group follows when the risk in the control group varies. The standard statistical methods of meta-analysis are based on simple effect models. The use of these methods could induce inaccurate or erroneous results in more complex situations. In this case, it is necessary to adopt a more appropriate effect model, such as the linear effect model. The properties of this kind of model allow the possibility that a treatment can be beneficial and harmful at the same time, in function of the risk without treatment. From this observation, it is advisable to be careful with the use of simple effect models in meta-analysis which can submerge interesting information in the synthesis.

Clinical trial registries and ethics review boards: the results of a survey by the FICHTRE project

Clinical trial registries could make an important contribution to research by playing a role in the avoidance of trial duplication as well as serving as a source of information on the existence of clinical trials for those wishing to perform overviews. In order to fulfil their role efficiently, these registries should contain information about all trials. However, resources of existing registries are not exhaustive. Currently, all protocols for clinical trials must be submitted to an ethics review board for approval-thus these boards could provide an invaluable source of information about what trials are being performed. One of the objectives of our project was to assess the attitudes of the ethics review boards (ERBs) to providing this type of information to clinical trial registries. We received 115 replies from 281 questionnaires sent to ERBs in seven European countries. More than 70% replied that they would be willing to communicate information about the trials they review, and there seemed to be general agreement that clinical trial registries should be funded and supported by governments. Although ERBs could provide a useful source of information, only an official body such as the Ministry of Health or a drug regulatory body could establish and run a clinical trials registry efficiently.

[Role of meta-analysis in the definition of target population in therapy]

The efficacy of a drug is a quantitative concept rather than a qualitative one. This quantity is expressed by several efficacy indices. None of them meet all the requirements. However, that of absolute benefit is especially suitable for the patients because it tells them the exact gain they can expect from taking the treatment. The absolute benefit varies according to patients' profiles because it interacts with some components of these profiles. In theory, such interactions can be used to predict the size of the absolute benefit for each patient, as well to describe better than with the current tools the therapy target population. We explain why meta-analysis and effect models are means of improving the prediction of the size of the effect and the definition of the therapy target population.

Effects on bone mass of long term treatment with thyroid hormones: a meta-analysis

Osteoporosis is the main cause of spine and hip fractures. Morbidity, mortality, and costs arising from hip fractures have been well documented. Thyroid hormones (TH) are widely prescribed, mainly in the elderly. Some studies (but not all) found a deleterious effect of suppressive TH therapy on bone mass. These conflicting data raised a controversy as to the safety of current prescribing and follow-up habits, which, in turn, raised major health-care issues. To look for a detrimental effect on bone of TH therapy, we performed a meta-analysis (by pooling standardized differences, using a fixed effect model) of all published controlled cross-sectional studies (41, including about 1250 patients) concerning the impact of TH therapy on bone mineral density (BMD). Studies with women receiving estrogen therapy were excluded a priori, as were studies with a high percentage of patients with postoperative hypoparathyroidism, when no separate data were available. We decided to stratify the data according to anatomical site, menopausal status, and suppressive or replacement TH therapy, resulting in 25 meta-analysis on 138 homogeneous subsets of data. The main sources of heterogensity between studies that we could identify were replacement or suppressive TH therapy, menopausal status, site (lumbar spine, femoral neck, Ward's triangle, greater trochanter, midshaft and distal radius, with various percentages of cortical bone), and history of hyperthyroidism, which has recently been found to impair bone mass in a large epidemiological survey. To improve homogeneity, we excluded a posteriori 102 patients from 3 studies, who had a past history of hyperthyroidism and separate BMD data, thus allowing assessment of the TH effect in almost all 25 subset meta-analyses. However, controls were usually not matched with cases for many factors influencing bone mass, such as body weight, age at menarche and at menopause, calcium dietary intake, smoking habits, alcohol intake, exercise, etc. For lumbar spine and hip (as for all other sites), suppressive TH therapy was associated with significant bone loss in postmenopausal women (but not in premenopausal women), whereas, conversely, replacement therapy was associated with bone loss in premenopausal women (spine and hip), but not in postmenopausal women. The detrimental effect of TH appeared more marked on cortical bone than on trabecular bone. Only a large long term prospective placebo-controlled trial of TH therapy (e.g. in benign nodules) evaluating BMD (and ideally fracture rate) would provide further insight into these issues.

[Presentation of OCTAVE II. A current epidemiologic study in France of the added prognostic value of ambulatory blood pressure monitoring]

The diagnosis of "white coat" hypertension, the measurement of indices of 24 hour variability of the blood pressure and increased accuracy of blood pressure estimation are some of the advantages of ambulatory blood pressure monitoring. They are part of the reason why the method has been adopted by ambulatory blood pressure monitoring is really useful in the treatment of hypertension is fragile: although the correlation with target organ complications seems better than with conventional blood pressure measurement, it remains to be shown that this information helps to predict and then improve the prognosis of hypertension. With respect to this first stage of demonstration of the added prognostic value attributable to ambulatory blood pressure monitoring, several studies have been undertaken, one of which is OCTAVE II. The goals were to analyse the predictive values of the indices obtained by ambulatory blood pressure monitoring in terms of cardiovascular morbidity and mortality after having described the correlations between these indices and the characteristics of individual patients. Two hundred and sixty six cardiologists, members of the French College of Cardiology, included 3,569 patients in whom an indication for ambulatory blood pressure monitoring had been retained, over a period of 10 months in 1991. These patients were 56.4 year old on average, with 52.6% men, hypertensive or not. The 5 year follow-up should end at the beginning of 1997.

AREVA: multicenter randomized comparison of low-dose versus standard-dose anticoagulation in patients with mechanical prosthetic heart valves

BACKGROUND

Moderate anticoagulation may be proposed to reduce the risk of hemorrhage for certain patients with a mechanical prosthesis, but the consequences for risk of thromboembolism are debated.

METHODS AND RESULTS

The purpose of the AREVA trial was to compare moderate oral anticoagulation (international normalized ratio [INR] of 2.0 to 3.0) with the usual regimen (INR of 3.0 to 4.5) after a single-valve replacement with a mechanical prosthesis, either Omnicarbon or St Jude. Patients included were between 18 and 75 years old, in sinus rhythm, and with a left atrial diameter < or = 50 mm on the time-motion echocardiogram. Patients were randomized for INR after surgery. From 1991 to 1994, 433 patients underwent valve replacement (aortic, 414; mitral, 19) with 353 St Jude and 80 Omnicarbon prostheses; 380 patients were randomized for INR: 188 for INR 2.0 to 3.0 and 192 for INR 3.0 to 4.5. Mean follow-up was 2.2 years (1 to 4 years). Analysis of 18001 INR samples showed that the mean of the median of INR was 2.74 +/- 0.35 in the 2.0 to 3.0 group and 3.21 +/- 0.33 in the 3.0 to 4.5 group (P < .0001). Thromboembolic events, as assessed from clinical data and CT brain scans, occurred in 10 patients in the 2.0 to 3.0 INR group and 9 patients in the 3.0 to 4.5 INR group (P = .78). Hemorrhagic events occurred in 34 patients in the 2.0 to 3.0 INR group and 56 patients in the 3.0 to 4.5 INR group (P < .01), with 13 and 19 major hemorrhagic events, respectively (P = .29).

CONCLUSIONS

In selected patients with mechanical prostheses, moderate anticoagulation prevents thromboembolic events as effectively as conventional anticoagulation and reduces the incidence of hemorrhagic events.

[Example of a phase IV trial involving several physicians and aiming at answering a scientific question: EOL]

The aim of this trial was to test the hypothesis that a reduced number of doses improves compliance in current medical practice. Compliance with twice a day dosage was compared with compliance with three doses a day. Two bioequivalent presentations of nicardipine were used, the regular presentation (t.i.d.) and the slow-release (b.i.d.). The trial was controlled, randomized, open, in two parallel groups: (1)'t.i.d.' group: one tablet of regular nicardipine, 20 mg, three times a day, three months; (2) 'b.i.d.' group: one capsule of slow-release nicardipine, twice a day, three months. 2651 general practitioners randomized 7274 hypertensive patients. The primary criterion was documented in 93.7 per cent of the cases at the end of the trial. The remaining 6.3 per cent comprised treatment withdrawal (2.8 per cent) and patients lost to follow-up (3.5 per cent). The primary criterion study was compliance, assessed by a self-questionnaire filled in by the patient and a standardised interview by the physician. Compliance was slightly better in the b.i.d. group than in the t.i.d. group (p < 0.001). Remaining pill count was also used but it was a failure. A random sample of investigators made on-site visits. Discordant data were infrequent and were limited to dates of visits. Difficulties with on-site visits were mostly due to a rather frequent lack of source records.

[Methodology of preparing a list of educational objectives: example of application to pharmacology. Groupe Objectifs Pédagogiques de L'Association des Enseignants de Pharmacologie]

Using a structured approach to categorize pharmacological knowledge and a systemic analysis of prescribing practice, we identified the knowledge needed to optimally prescribe and manage treatments with drugs. The approach consisted in finding the branched chains of knowledge beginning with each operation required to solve each problem which arises in prescribing and managing drugs at the most elementary level. This elementary knowledge is then transformed into educational objectives. The next step is to share the educational objectives between basic medical training, continuing medical education and acquisition of therapeutic knowledge. The method could be applied in other medical teaching domains.

Platelet-aggregation inhibition and hemodynamic effects of beraprost sodium, a new oral prostacyclin derivative: a study in healthy male subjects

The antiaggregation and hemodynamic effects of the new prostacyclin analogue beraprost sodium were investigated in a randomized, placebo-controlled, double-blind clinical trial of Latin-square design. Twelve healthy Caucasian males randomly received 8-day oral treatments of 20, 40, and 60 micrograms of beraprost sodium and a placebo. One-week washout periods followed each treatment. Pharmacokinetic and pharmacodynamic measurements were performed on days 1 and 8 for each period of treatment. All three doses of beraprost sodium significantly inhibited platelet aggregation on day 8 (compared with placebo) during the 1st h after drug intake. Incubation of the 60-micrograms beraprost sodium samples with ADP (2, 5, and 10 microM) and collagen (1.25 micrograms/mL) decreased platelet aggregation by 10, 19, 16, and 6 +/- 4% (mean +/- SE), respectively, compared with placebo. No significant hemodynamic effects on blood pressure, heart rate, and digital pulse were observed. The 60-micrograms dose of beraprost sodium did significantly decrease the IRZ index (which may reflect the left ventricular pre-ejection period) on days 1 and 8. Some subjects experienced headache and facial flushing, effects that were dose dependent and reversible. Beraprost sodium at 20- to 60-micrograms doses exerts platelet antiaggregation (day 8 of therapy) and slight hemodynamic (days 1 and 8 of treatment) effects in Caucasian males. Beraprost sodium hemodynamic effects and potential benefits in patients with cardiovascular disease should be explored further.

Thrombolytic therapy with streptokinase in acute ischemic stroke

BACKGROUND

In patients with acute ischemic stroke, early treatment with thrombolytic agents is thought to permit reperfusion of ischemic neurons and to promote recovery of function. The Multicenter Acute Stroke Trial-Europe (MAST-E) was designed to assess the efficacy and safety of streptokinase in patients with acute ischemic stroke.

METHODS

Patients with moderate-to-severe ischemia in the territory of the middle cerebral artery were randomly assigned to receive streptokinase (1.5 million units over a period of one hour) or placebo within six hours after the onset of stroke. The primary efficacy outcome was a binary criterion combining mortality and severe disability at six months, with severe disability defined as a score of 3 or higher on the Rankin scale. The primary safety outcomes were mortality at 10 days and cerebral hemorrhage.

RESULTS

All randomized patients (156 in the streptokinase group and 154 in the placebo group) were evaluated at six months. The incidence of the primary efficacy outcome was similar in the two groups (124 patients in the streptokinase group and 126 in the placebo group died or had a Rankin score > or = 3). However, the mortality rate at 10 days was significantly higher in the streptokinase group than in the placebo group (34.0 percent vs. 18.2 percent, P = 0.002). The higher rate in the streptokinase group was mainly due to the hemorrhagic transformation of ischemic cerebral infarcts. At six months, more deaths had occurred in the streptokinase group than in the placebo group (73 vs. 59, P = 0.06).

CONCLUSIONS

In patients with acute ischemic stroke, treatment with streptokinase resulted in an increase in mortality. The routine use of streptokinase cannot be recommended in acute ischemic stroke.

A dose-effect study of beraprost sodium in intermittent claudication

We compared the efficacy and safety of three doses of beraprost sodium, an epoprostenol analogue, with placebo in the treatment of intermittent claudication (Fontaine's stage II). One hundred sixty-four patients were randomized to receive either placebo, 20 micrograms beraprost sodium (BPS60 group), 40 micrograms beraprost sodium (BPS120 group), or 60 micrograms beraprost sodium (BPS180 group) three times daily administered orally in a double-blind manner for 12 weeks. Treadmill exercise tests were performed twice during an initial selection phase (D-28 and D0) at week 10 (at trough beraprost concentration) and week 12 (at peak beraprost concentration) of the treatment phase. At week 10, all groups showed an increase in pain-free walking distance, and this distance was greatest in the BPS60 and BPS120 groups (p = 0.055). At week 12, a similar pattern was observed, and the difference was significant between the groups (p = 0.023). The most frequent adverse events reported were gastrointestinal disorders, headaches, skin disorders, and flushes. Patients who received either 60 or 120 micrograms of beraprost sodium daily had an increased pain-free walking distance. Further studies are required to investigate why the highest dose used (180 micrograms daily) showed lower efficacy. Having both vasodilating and antiplatelet properties and being able to increase pain-free walking distance in the short term, beraprost sodium is a promising drug for the treatment of intermittent claudication.

[Integratable message: central problem of communication for therapeutic information]

The aim of therapeutic information is that people who need it have access to data of the highest level of evidence. In this context, a message is defined as the medium of information. The conditions for an ideal message to the prescriber are described: its content, the qualities it must meet, and the functional issues it will tackle. An example illustrates the importance of the form of a message. The difference between recommendation and message is highlighted.

[The Cochrane Collaboration: the need for international collaboration]

The Cochrane Collaboration has undergone an unexpected growth since its inauguration during an international meeting in 1993. This collaboration was established in response to the criticisms of a British physician, Archie Cochrane, who criticised the medical profession because it had not organised a critical summary of all relevant randomised controlled trials. The driving force of the collaboration is the collaborative review groups who collect, sort and synthesize data concerning treatment efficacy in their particular area of interest. The Cochrane Centres and the Field Coordinations help them to do this and together the produce the Cochrane Database of Systematic Reviews. The need for international cooperation is obvious because the information necessary to this system is produced throughout the world, and only some of this information is published.

[Consequences of discrepancies between current data of the science and medical practice]

The target population for drug treatment is composed of all patients likely to benefit from it. It is defined on the basis of all the data collected during the clinical evaluation of the therapy (on the studied population). The dissemination population comprises those patients who are given the therapy once it has been released. Several surveys have shown that these two population do not coincide, although they should. The consequences of this gap can theoretically be measured. However, in order to do such an evaluation, we need relevant indices representing the consequences in terms of lack of expected benefit for the patients. We constructed models that represent the gap and its impact on life span and explored them by simulation. It came out that the consequences can be assessed in practice provided that outcome measurement technology is applied.

[Note on the necessary definition of therapeutic objectives for optimal prescription]

Prescription of a therapy depends on the patient, his/her condition, and the therapeutic goal(s) the doctor has identified. In turn, a therapeutic goal depends on the availability of an effective therapy. Without a clearly defined objective, a prescription is likely not to be appropriate. The patient may not appreciate the importance of good compliance if he/she does not know what it is for. The pathway of specialised investigations can include unnecessary tests if it is not built up around the objective of identifying the therapeutic goal(s). If there is no therapy available, the diagnosis is of no use. The arguments for and against a therapy cannot be appropriately assessed and compared if they are not put in the perspective of the therapeutic goal(s).

[Hierarchies according to the level of evidence of source data, before their integration in the synthesis, in the matter of therapeutic efficacy]

The discipline therapeutic information uses the concept of the level of evidence for source data concerning therapeutic efficacy, and its ordering before integration in syntheses. In this paper we will start by considering the problems raised by the definition of the level of evidence in terms of the dimensions it covers. We have differentiated three components: clinical pertinance of the question asked, the methods used to reply and the quality of the data collected. Second, we will examine the different criteria important for each of these three dimensions. There are many criteria possible which do not all have the same weight, and thus for any non-arbitrary tool developed to enable the level of evidence to be ordered, it is necessary to know the weight of the different criteria. Thirdly, we will present the techniques used for working with multicriteria situations in econometrics which represent a methodology we propose using to apply in our context. To do so we need to build a 'reference' base for the level of evidence using 'experts' opinions which will help us to examine the weights of the different criteria. This approach, in conjunction with some epistemological and sociological considerations, may contribute to a better understanding of the different dimensions of this concept.

Comparison between a bid and a tid regimen: improved compliance with no improved antihypertensive effect. The EOL Research Group

OBJECTIVES

To compare compliance with an antihypertensive treatment administered either twice daily or three times daily. The two formulations of the antihypertensive treatment used (nicardipine) "regular tablets" (t.d.) and "slow release tablets" (b.d.) are bioequivalent at the daily dosage used in the study.

STUDY DESIGN

Open, controlled, parallel designed study with centralised, randomised allocation to the treatment groups: TID group: A nicardipine 20 mg tablet, three times daily for 3 months. BID group: A capsule of slow release (SR) nicardipine, 50 mg twice daily for 3 months.

PATIENTS

7274 hypertensive patients were investigated by 2.651 general practitioners. Compliance with the nicardipine was assessed by means of standardised interviews with the patients and by a questionnaire for the investigators.

RESULTS

Compliance was slightly higher in the BID than in the TID group; 71.2% and 24.5% of patients in the first group declared their compliance was 100% and 80% compared to 82.3% and 15% in the second group. A statistically significant relationship was shown between compliance with nicardipine and the decrease in blood pressure after three months of therapy. However, no significant difference was noticed between the two groups of patients in the absolute decrease in blood pressure after the treatment period: 25.7/14.7 mm Hg in the TID group compared with 25.9/15.0 mm Hg in the BID group.

CONCLUSIONS

A difference in compliance between the bioequivalent BID and TID formulations of the same active agent was shown in hypertensive patients. However, the difference was not large enough to lead to a difference either in the number of controlled patients or in the decrease in blood pressure. Reducing the number of daily doses does not automatically lead to greater efficacy of treatment.

Myocardial protection during coronary artery bypass graft surgery: a randomized, double-blind, placebo-controlled study with trimetazidine

We conducted a randomized, double-blind, placebo-controlled study to assess the cardioprotective effects of trimetazidine (TMZ), an antiischemic drug, on left ventricular function using transesophageal echocardiography (TEE) after coronary artery bypass grafting (CABG). Forty patients undergoing elective CABG received either TMZ or a placebo (PCB). The primary measures of efficacy were serial measurements of fractional area change (FAC), percent of systolic wall thickening (SWT), and malonedialdehyde (MDA) production. The two groups were similar for the following variables: number of vessels revascularized (2.5 +/- 0.2 in the TMZ group and 2.8 +/- 0.1 in the PCB group), duration of aortic clamping (46 +/- 4 min in the TMZ group and 48 +/- 3 min in the PCB group), and bypass time (63 +/- 4 min in the TMZ group and 70 +/- 4 min in the PCB group). FAC increased by 12% in both groups 20 min after aortic unclamping (P < 0.05) and remained above the initial value at the sixth postoperative hour. SWT was 23.8% +/- 1.6%, 25.4% +/- 1.9%, then 21.6% +/- 1.5% in the TMZ group and 22.8% +/- 1.6%, 23.8% +/- 1.4%, then 22.3% +/- 1.6 % in the PCB group, after induction of anesthesia and 1 and 6 h after aortic unclamping (not significant). MDA increased by 24% in the PCB group and 25% in the TMZ group 20 min after aortic unclamping (P < 0.01). Lactate levels were lower in the TMZ group (P < 0.05) and patients from the TMZ group received less intravenous calcium before aortic clamping (P < 0.02) and less calcium channel entry blocking drugs in the early phase after aortic unclamping (P < 0.01) compared to the PCB group. We conclude that in patients with good preoperative ejection fraction undergoing CABG, TMZ as administered did not demonstrate clinically significant cardioprotective effects on left ventricular performance and lipid peroxidation compared to PCB.

Effects of bisoprolol on heart rate variability in heart failure

Analysis of heart rate variability (HRV) provides a non-invasive index of autonomic nervous system activity. HRV has been shown to be reduced in heart failure. Preliminary data indicate that beta blockers improve clinical status in patients with heart failure, but HRV improvement remains to be demonstrated. Fifty-four patients from the randomized double-blind, placebo-controlled Cardiac Insufficiency Bisoprolol Study were included in the HRV study. The bisoprolol daily dose was 5 mg once daily. We assessed HRV during 24-hour Holter recordings before randomization and after 2 months of treatment. HRV as measured in the time domain by root-mean-square successive differences (rMSSD), the percentage of adjacent RR differences >50 ms (pNN50), and the SD of RR intervals (SDNN), and in the frequency domain by high-frequency (0.16 to 0.40 Hz) and low-frequency (0.04 to 0.15 Hz) power. Most patients were in New York Heart Association functional class III. The mean left ventricular ejection fraction was 27 +/- 7%, and heart failure was idiopathic or ischemic. After 2 months, the patients receiving bisoprolol had a reduced mean heart rate compared with that in placebo patients (p=0.0004). Bisoprolol increased 24-hour rMSSD (p=0.04) and 24-hour pNN50 (p=0.04), daytime SDNN (p=0.05), and daytime high-frequency power (p=0.03) power. Bisoprolol induced a significant increase in HRV parameters related to parasympathetic activity in heart failure. Increased vagal tone may contribute to the protective effect of beta blockers and may have prognostic implications.