Within the past few decades, medicine, especially critical care medicine, has progressed spectacularly as a result of medical research and the participation of patients in clinical trials. The Declaration of Helsinki and international guidelines require that the patient give informed consent before participating in a study. This informed consent is essential to respect the autonomy of patients and protect them against abuse 123.

Therefore, to respect the ethical principles of clinical research, informed consent for research demands to be held at a high standard because the patient does not always benefit directly from the research results and might suffer some risks for the good of the community. Keeping with this high standard, informed consent must fulfil three mandatory conditions as described in the Belmont report: adequate information about the study with a complete disclosure of the risks and benefits, the patient's comprehension, and the voluntariness 4567. This means that the patient should be able to freely decide their participation in the study without any external pressure and that they could resign from the study at any time without any consequence to their care. To respect this requirement and the principle of autonomy, it is essential that patients comprehend and recall knowledge of the study components as well as their participation in the study. However, patients in the intensive care unit (ICU) may be in an inadequate condition to give their informed consent for a research study because their capacity for comprehension is questionable. However, lighter sedation of ICU patients permits an increasing number of them to remain conscious. These patients seem to be able to understand the information presented and to decide freely if they would like to participate in the study. However, there is no available consensus on the capacity required to give consent 8. As objective criteria regarding the decision-making capacity of patients are lacking in international and national directives, investigators usually use their clinical judgement and the Glasgow Coma Scale (GCS) as guidance. Previously, we showed that 22% of clinical trial participants could not recall their participation in the clinical trial despite the fact that informed consent was obtained in an ideal situation, namely before their admission to the ICU 9. Furthermore, 25% of patients were unable to recall the clinical trial purpose and its related risks. In view of these results, we hypothesized that informed consent for research obtained from a patient during their ICU stay is associated with an even worse ability for them to recall their participation in a clinical trial and its components (purpose and risk).

Between May 2000 and January 2001 we included 48 patients, of whom four died during the ICU stay (Figure 1). We therefore analysed the data on 44 patients who signed the informed consent and were alive at 10 ± 2 days; at this time all patients presented with a GCS score of 15, were fully oriented and were judged competent by the investigator.

Of the 44 patients, 35 (80%) recognized they had participated in a clinical trial; 14 of the 44 patients (32%) could recall their clinical trial participation as well as the clinical trial components, namely the clinical trial purpose and the related risk (Figure 1). These 14 patients were assigned to the 'complete recall' group.

Patients with complete recall did not differ from patients with incomplete recall with regard to their demographic, educational, and admission characteristics (Table 1). The lengths of mechanical ventilation and ICU stay were similar in both groups.

When informed consent was obtained, the clinical and laboratory values of the patients did not differ between the two groups (Table 2). The medications administered during the 24 hours before and after the informed consent were similar in the two groups (Table 2). More patients with complete recall had read the informative leaflet or had asked at least one question before signing the informed consent (13 out of 14 (93%) versus 18 out of 30 (60%); p = 0.03). Asking at least one question was associated with a complete recall of the clinical trial (p = 0.03). The sensitivity, specificity, positive predictive values, and negative predictive values of 'asking one question' and 'reading the informative leaflet or asking one question' to differentiate patients with complete recall from patients with incomplete recall are shown in Table 3.

The first investigator included 15 (34%) patients in the clinical trial during the first three months of the inclusion period, and the second investigator added 29 (66%) patients during the last six months. Patient characteristics, attitude during the informed consent procedure, and rates of recall did not differ significantly between investigators (Table 4).

A great majority of ICU patients who had consented to participate in a clinical trial during their ICU stay recalled their clinical trial participation after 10 ± 2 days. Other studies, including those on patients with acute myocardial infarction, have reported clinical trial participation recall rates similar to ours 141516. Our rate is also similar to the rate we found in a study in which informed consent was obtained in an ideal situation, namely before ICU admission 9.

Although this may seem very encouraging, only 32% of the patients who gave consent during their ICU stay recalled clinical trial components; this is in contrast to our 'ideal situation' previous study in which 75% of patients had a complete recall 9. The low rate of complete recall in the present study could be explained by the difficulty of some ICU patients to process information given the stress of the acute phase and possibly experiencing feelings of dependence and anguish 17. Our routine clinical evaluation might not detect this potential cognitive defect 18. The reasons why some patients are able to recall whereas others cannot therefore remain unclear.

The severity of disease, the neurological status of the patients, and the medications received in the ICU when informed consent was obtained and during the 24 hours after the informed consent procedure were similar in both groups of our patients.

Inadequate information disclosure to some patients that could explain our results can reasonably be excluded because the information procedure was well standardized. In contrast with previous reports, we found that neither age nor educational status influenced the ability to recall clinical trial components 1920. This might reflect the fact that the information presented was not difficult to understand.

If the informed consent satisfies the three criteria specified by international guidelines, the consent is deemed valid 1221. However, if the clinical trial has already begun and the patient is unaware of the purpose, related risks, and their participation in the clinical trial, they are obviously unable to decide mindfully whether to continue in the clinical trial or to withdraw from it at any time. In this case, we might question the respect of the participant's autonomy offered by the informed consent. This leads us to view the informed consent as a process rather than a simple procedure. The informed consent process requires, to our mind, multiple conversations on several occasions while the research is conducted.

We found simple but important factors associated with complete recall of the clinical trial components. We tried to identify factors that were easy to observe by an investigator such as 'asking questions' and 'reading the informative leaflet'. In our previous study, we found that more patients who were able to mention all clinical trial components had read the leaflet and had asked at least one question. 'Asking questions' increased the chance of recall in critically ill patients. We could speculate inversely that asking no questions could increase the potential for poor initial comprehension of the clinical trial components.

Our results are in line with Flory and Emanuel's findings 22, which concluded that person-to-person interaction with clinical trial participants may be the most effective way of improving their understanding.

This finding revives the debate about informed consent for research in the ICU. In the past, deferred consent 23 or waiving of consent for research in the emergency setting 2425 was considered acceptable in particular research situations. However, during the past few years, legislation has tried to enhance protection for incompetent patients. Barriers to the inclusion of ICU patients in research studies have increased, especially across Europe 226. We fully support the idea that oversight is necessary to ensure the uniform application of ethical standards 11, but the most dependable safeguard for the research subject is investigator understanding and respect of the ethical requirements of clinical research.

Our study presents some limitations. First, the small number of patients enrolled was dependent on the clinical trial. This might have impeded the detection of other potential factors that could influence the recall of the clinical trial. In addition, the small clinical trial size precluded the possibility of a multivariate analysis. Second, we did not investigate the recall of patients who had refused to participate in the clinical trial on inflammation primarily because there was no consent to investigate. Perhaps the degree of recall might have been different in patients who had refused participation. Third, we did not assess patients' cognitive capacity. There is good evidence that the cognitive capacity of ICU patients, or even sick patients in general, is impaired 2728. However, spending more time with patients to test their cognitive capacity, and interacting with them for much longer than for a 'standard' informed consent procedure, would have biased our results. It is for this same reason that we did not measure the patients' memory. Fourth, scores of severity of illness and laboratory values may suggest that patients with incomplete recall were more ill than those with complete recall. Although not statistically significant, these results do not allow for the possibility that the severity of the disease might act on recall. However, the fundamental message remains that a poor rate of recall of a clinical trial is present in ICU patients. Finally, as our patients were presumed fully competent, the study faced the best possible situation in ICU patients. We are aware that these patients do not represent the majority of ICU patients and our results may therefore not be generalizable. However, it is reasonable to speculate that the rate of recall may be even worse in usual patients in the ICU.

To respect the principle of autonomy in the informed consent procedure, patients should be able to decide freely without any external pressure whether they agree to continue to participate in a clinical trial at any time. In concordance with our previous paper, about 80% of ICU patients are able to recall their clinical trial participation. However, as hypothesized, the rate of recall of the clinical trial components is very low. We presume that it could be much lower than in patients informed before being admitted to the ICU. Our results reinforce the importance of viewing informed consent as a process and the need for revisiting informed consent several times during an ongoing clinical trial. We suggest that investigators, while the research is conducted., repeatedly repeat clinical trial information to the participants even after their ICU stay. We therefore propose to reassess regularly whether continued participation in a clinical trial is desired by the patient, as recommended previously 11. Whether encouraging patients to read the informative leaflet and ask questions about the clinical trial improves the informed consent procedure remains to be fully investigated.

• The rate of recall of the clinical trial components (purpose and risks) in patients informed after being admitted to the ICU is very low. It is presumably much lower than in patients who gave their informed consent outside the ICU setting.

• More patients who could recall their clinical trial participation and the clinical trial components had read the informative leaflet or had asked at least one question before signing the informed consent.

• Reconsidering the informed consent procedure repeatedly during an ongoing clinical trial could be useful to respect patients' rights.

• Informed consent to participate in a clinical trial should be considered an ongoing process rather than a single procedure.

GCS = Glasgow Coma Scale; ICU = intensive care unit.

The authors declare that they have no competing interests.

CC participated in the design of the study and the collection, analysis and interpretation of the data and wrote the manuscript. PM contributed to the conception and design of the study and to the analysis and revision of the manuscript. SL was involved in the design of the study and the collection of data. BR conceived and coordinated the study and was involved in the interpretation of data and in revision of the manuscript. All authors read and approved the final manuscript.