Timely Detection of Adverse Advents in Therapeutic Trial
A probiotics trial on trial: the problem of timely detection of adverse advents in therapeutic trials

https://doi.org/10.1016/j.jclinepi.2008.12.012Get rights and content

Introduction

The outcome of a therapeutic trial may be frustrating. In a recent Dutch randomized, double-blind, placebo-controlled multicenter trial, the effects of probiotics were studied to investigate their potential to diminish infectious complications in patients with predicted severe acute pancreatitis [1]. Desolately, the results of this study were quite contrary to expectations: there was no diminution of infectious complications, and furthermore, patients taking the probiotics had more than double the relative mortality risk. Finally, 33 of the 297 patients included in this trial died with an excess of 15 deaths occurring in the treatment group (Table 1).

These figures were made public in a press conference and drew much media attention in The Netherlands. Some statisticians and methodologists subsequently publicly expressed their strong condemnation, particularly stating that the data monitoring and safety committee (DMSC) had been negligent in its control function, missing an early trend of increased mortality in the treatment group at interim analysis using the wrong statistic [2]. If appropriate statistical tools had been used, they claim, this trial should have been stopped earlier resulting in fewer deaths. Was the DMSC really neglectful in its monitoring task? Could this not be a conclusion with the benefit of hindsight?

After this public criticism, several relatives of deceased trial participants announced that they would initiate a lawsuit for compensation because of this supposed negligence. To triumph in such a tort claim, the plaintiff must, among others, prove that the DMSC breached its duty by failing to adhere to generally accepted authoritative guidelines. As legal professionals will lack sufficient detailed knowledge on the planning, execution, and monitoring of a randomized clinical trial, any judicial ruling will invoke an expert medical opinion.

When scientific conduct is formally scrutinized, it invokes a particular—and perhaps novel—dimension of clinical epidemiology. But how? By its very nature, this basic medical science is a normative endeavor, a task which is essentially prospective: how to reach most favorable outcomes from medical interventions. A judgment of the conduct of a DMSC is different: it is retrospective, and the analysis and subsequent conclusions are now to be understood in a legal context. For this evaluation of the conduct of a DMSC, we first need a preset standard of its tasks and processes for this type of trial. This benchmark not only must be realistic and meet the essential medical requirements, but also be legally understandable and usable. Secondly, we need an established distinguishing method for objective retrospection, because it is always easy to be wise after the event [3]. In the course of this enquiry, we meet two interrelated forms of bias: outcome and hindsight bias [4]. In this article, I will comment how careful appraisal of the work of a DMSC can be done in relation to the controversies that relate to both statistical issues and decision making [5].

Section snippets

Detecting harm: a challenging task for the data monitoring and safety committee

As trial safety is of paramount importance, a DMSC should focus principally on issues of harm [6]. Are there convincing arguments, while carefully balancing benefits and harms, to stop this trial because of harmful side effects? An increased all-cause mortality in the intervention group is of course a principal reason for such a decision. But how and when? For evaluating differences in mortality in a comparative trial, we can opt between two essentially different routes: the use of traditional

Tackling uncertainties and the play of chance

Trials are there to combat therapeutic ignorance [13]. This disputed trial was intended to confront the uncertainties about the effect of probiotics in a defined class of patients. However, for reliable definite answers on research questions, we must pass a phase of insecurity. Especially in the incipient stage of the trial with small numbers of patients, chance may well account for differences in outcomes between groups. With increasing numbers of participants, the limits of oscillations will

Troubling torts

Human subject research is increasingly confronted with litigation [15]. A bad outcome always is the prime reason for such a liability action. But a well-thought-out legal opinion must take into account both the processes and outcomes of the work of the DMSC. However, while looking back, the horrific ending gives rise to outcome bias and some sort of blinded review of the process is to be preferred [3]. As the psychologist Edwards formulated it more than 20 years ago: “A good decision cannot

Living forward

There is much at stake here. Chalmers recently urged to increase our efforts to protect patients from unproven but used treatments. Despite the tremendous progress of medicine, there are still so many therapeutical questions to be answered. Changing attitudes and growing restrictions may curtail therapeutic research [13]. But how to respond to the problem of minimizing the risk for participants in trials? As discussed, the play of chance may well limit our abilities to avert harm. There has

First page preview

First page preview
Click to open first page preview

References (19)

There are more references available in the full text version of this article.

Cited by (0)

View full text