Intensive Care Medicine

, Volume 34, Issue 8, pp 1393–1400

Bedside adherence to clinical practice guidelines in the intensive care unit: the TECLA study


    • Service de Réanimation MédicaleHôpital Bocage
    • Commission d’Epidémiologie et de Recherche CliniqueSociété de Réanimation de Langue Française
  • Hervé Mentec
    • Commission d’Epidémiologie et de Recherche CliniqueSociété de Réanimation de Langue Française
  • François Feihl
    • Commission d’Epidémiologie et de Recherche CliniqueSociété de Réanimation de Langue Française
  • Djillali Annane
    • Commission d’Epidémiologie et de Recherche CliniqueSociété de Réanimation de Langue Française
  • Christian Melot
    • Commission d’Epidémiologie et de Recherche CliniqueSociété de Réanimation de Langue Française
  • Philippe Vignon
    • Commission d’Epidémiologie et de Recherche CliniqueSociété de Réanimation de Langue Française
  • Christian Brun-Buisson
    • Commission d’Epidémiologie et de Recherche CliniqueSociété de Réanimation de Langue Française
  • TECLA Study Group

DOI: 10.1007/s00134-008-1059-y

Cite this article as:
Quenot, J., Mentec, H., Feihl, F. et al. Intensive Care Med (2008) 34: 1393. doi:10.1007/s00134-008-1059-y



To assess adherence to clinical practice guidelines for three interventions routinely used in critical care medicine.

Design and setting

Multicenter, 1-day cross-sectional observational study in 44 intensive care units in four countries.


A total of 419 patients hospitalized in participating ICUs on the day of the survey.

Measurements and results

Red blood cell transfusion (n = 29) was performed appropriately in 22 patients (76%), while among the 390 patients who received no transfusion 4 (1%) had a valid indication. Setting of tidal volume in acute respiratory distress syndrome, assessed in 45 patients, was deemed appropriate in 37 cases (82%). Prescription of stress ulcer prophylaxis (n = 128) was appropriate in only 24 patients (19%), while among the 268 patients who were not treated 28 (10%) had an indication.


The implementation of recommendations varies across different domains of care. While the adherence to current recommendations in routine practice is acceptable as regards tidal volume settings in acute respiratory distress syndrome, it is suboptimal for blood transfusion and prevention of upper gastrointestinal bleeding. Practice surveys are useful to inform strategies currently developed to assess practices of health-care professionals and develop strategies for more effective dissemination of medical knowledge.


TransfusionPractice guidelinesRespiratory distress syndromeStress ulcer prophylaxisIntensive care


Clinical research studies are an indispensable tool for improving medical knowledge and improving practice in intensive care units (ICUs) [1, 2]. However, translating the results of clinical research into actual clinical practice is a complex process [3]. The first step in this process is the publication of data generated from clinical studies. However, only one-third of completed clinical trials are eventually published, predominantly those reporting positive results [4, 5] (Clinical trials gov, “Linking patients to medical research,”; Current controlled trials, “Featuring the metaregister of controlled trials,” Although the implementation of research results into clinical practice can be facilitated by evaluating the impact of applying new approaches through large-scale epidemiological surveys [68], the quantity of scientific information published is so huge that it renders valuable information inaccessible for most physicians. For this reason official health organizations, scientific societies, and international collaborative groups such as the Cochrane Collaboration provide summaries of the best evidence on specific interventions, and these summaries are of paramount importance for disseminating medical knowledge. Finally, even when the information is delivered to the target physicians, several hurdles limit the uptake of specific recommendations into routine practice, including difficulty in changing behavior, an inadequate environment, and factors related to cost or cultural issues.

A number of interventions in critical care medicine have recently been scrutinized, and several clinical trials performed to assess their clinical impact. For example, red blood cell transfusions are a cornerstone of critical care practice, but there are divergent views on the risks of anemia and the benefits of transfusion in this setting. A recent controlled trial of red blood cell transfusion in critical care showed that a restrictive strategy is at least as effective as (and possibly superior to) a more liberal strategy [9], and several clinical practice guidelines have been issued to recommend a conservative transfusion policy, restricting transfusion to patients having hemoglobin levels below 7 g/dl.

In the setting of acute respiratory distress syndrome (ARDS) the mortality rate is approx. 40–50% [10, 11]. Mechanical ventilation using large tidal volumes can cause the disruption of pulmonary epithelium and endothelium, lung inflammation, atelectasis, hypoxemia, and the release of inflammatory mediators [12, 13]. There is some evidence from the medical literature that a lower tidal volume (6–8 ml/kg) than is traditionally used (10–12 ml/kg) may reduce mortality [14]. These studies demonstrate that the adoption of target values in clinical practice results in beneficial outcomes to patients. On the other hand, while the use of stress ulcer prophylaxis is very common in critical care medicine, there is little evidence that this intervention influences patients' outcomes, and its use is recommended only in selected groups of patients with multiple risk factors for bleeding [15].

To assess the translation of clinical research into routine critical care practice we assessed the current use of these three interventions—red blood cell transfusion, tidal volumes in ARDS, and stress ulcer prophylaxis—through a multicenter, cross-sectional study. Preliminary results were presented at the 32th Congress of the Société de Réanimation de Langue Française (SRLF, French-speaking Society of Intensive Care) in Paris, January 2005.

Materials and methods

Study design

The TECLA survey was a multicenter, cross-sectional, observational study conducted in ICUs in France, Belgium, Canada, and Switzerland. The study consisted of a 1-day cross-sectional recording of all interventions performed and treatments administered to all patients hospitalized in the participating ICUs on 11 May 2004. A total of 630 critical care physicians representing 120 ICUs were invited to participate by means of a mail request through the SRLF website. The recruitment of participating ICUs was on a voluntary basis, from a mailing list, with no financial incentive. A total of 44 units (37%) agreed to participate in the study, involving in total 103 physicians in 38 hospitals (see Acknowledgements). The participating ICUs were university and/or regional hospitals (n = 17, 39%), general (nonacademic) hospitals (n = 25, 57%), or semiprivate hospitals (n = 2, 4%). There were 36 (82%) mixed medicosurgical, 4 (9%) surgical, and 4 (9%) medical ICUs. Six hospitals had two separate ICUs participating in the study. Three units (8%) had 6 beds or fewer, 32 (71%) had between 7 and 15 beds, and 9 (21%) had more than 15 beds. The median number of admissions per year in the participating ICUs was 437 (258–1215). There was no difference between participating centers and nonresponders as regards size, academic status, and main activities. Since this observational study required no deviation from routine medical practice, institutional review board approval was not required. The study was approved by the SRLF Ethics Committee. Physicians collecting the data were blinded to the main objective of the study.

Patient population

All 419 adult patients present in the ICUs of the participating centers on the data collection day were eligible except those with a do-not-resuscitate order or any other decision to withdraw or withhold therapy in the ICU. Patients who were intubated and receiving mechanical ventilation were included in the evaluation of tidal volume settings if they had clinical and radiographic features of ARDS, i.e., an acute decrease in the ratio of partial pressure of arterial oxygen to fraction of inspired oxygen (PaO2/FIO2 ratio) of 250 or less, bilateral pulmonary infiltrates on chest radiography consistent with the presence of edema, and no clinical evidence of left arterial hypertension or a pulmonary-capillary wedge pressure of 18 mmHg or less (if measured). Patients were excluded if they had increased intracranial pressure, pregnancy, vasculitis with diffuse alveolar hemorrhage, or if they had received a bone marrow or lung transplant. The predicted body weight (PBW) was calculated as 50 +0.91 × [height (cm) −152.4] in men and as 45.5 +0.91 × [height (cm) −152.4] in women. All patients were included in the evaluation of blood transfusion practices according to hemoglobin concentrations except those having chronic anemia, pregnancy, bone marrow transplantation, or leukemia. Patients were excluded from the evaluation of stress ulcer prophylaxis if they were being previously treated for gastrointestinal (GI) ulcer before ICU admission or had had a gastrectomy.

Patients' median age was 61 years (range 16–92), and median Simplified Acute Physiology Score (SAPS) II on ICU admission was 38 points (range 0–112; Table 1). Medical patients represented 66% of the patient population, surgical 29%, and trauma 5%. By the McCabe classification, 57% of patients had no preexisting underlying disease likely to affect prognosis, while 13% had a rapidly (< 1 year) fatal disease.
Table 1

Clinical characteristics of the study population (n = 419)

Age, median (years; range)

61 (16–92)


Gender: M/F



Admission category



277 (66%)


Unscheduled surgery

85 (20%)


Scheduled surgery

37 (9%)



20 (5%)


Transfer to ICU from


Emergency room

141 (34%)


Medical ward

135 (33%)


Surgical ward

78 (19%)


Operating/recovery room

37 (9%)


Another ICU

28 (7%)





35 (8.3%)


Metastatic cancer

18 (4.3%)


Hematological malignancy

1 (0.2%)


Acquired immunodeficiency syndrome

5 (1.2%)


Chronic respiratory insufficiency

52 (12%)


Chronic heart failure

44 (10%)


Type I diabetes

20 (4.7%)


Chronic liver failure

9 (2%)


Chronic renal failure

11 (2.6%)


McCabe's classification


No underlying disease affecting prognosis

238 (57%)


Ultimately fatal (< 5 years)

128 (30%)


Rapidly fatal (< 1 year)

53 (13%)


SAPS II score at admission, median (range)

38 (0–112)


Data collection

The following data were recorded: age, weight, height, admission diagnosis, SAPS II score on admission [16], vital signs and standard laboratory data on the day of survey, details of any intervention or treatment including administration of drugs, blood products and nutrition, or ventilatory, circulatory, and renal support, noting the indication for each intervention. One physician in each participating ICU was in charge of completing the data collection.

The standards for each of the three interventions under study relied on data from published original papers and from available guidelines generated by official health organizations or scientific societies. The pivotal TRICC study [9] and recommendations from both the Agence Française de Sécurité Sanitaire des Produits de Santé (French Health Products Safety Agency, “Transfusions de globules rouges homologues: produits, indications, alternatives,” and the SRLF consensus conference [“XXIII conférence de consensus en réanimation et médecine d'urgence: transfusion érythrocytaire en réanimation (nouveau-né exclu),” 2003,] were used to establish standards of care for blood transfusion in the critically ill. For stress ulcer prophylaxis we used recently published original articles [1721] and the revised guidelines of the consensus conference of the SRLF [15]. For the setting of tidal volume in ARDS we established our own standard of care partly on the basis of the pivotal ARDS NetWork study [14] and the recommendations of the SRLF expert group (J.C. Richard, C. Girault, S. Leteurtre, F. Leclerc, Groupe d'expert de la SRLF, “Prise en charge ventilatoire du syndrome de détresse respiratoire aiguë de l'adulte et de l'enfant (nouveau-né exclu),” 2005, which were under development at the time of study. The appropriateness of the prescription of the three therapeutic interventions under study was assessed according to algorithms developed by the SRLF Commission for Epidemiology and Clinical Research (Commission d'Epidemiologie et de Recherche Clinique) (Figs. 123).
Fig. 1

Algorithm for red blood cell transfusion. HR, Heart rate
Fig. 2

Algorithm for tidal volume in ARDS
Fig. 3

Algorithm for stress ulcer prophylaxis. ACT, Activated clotting time; ISS, Injury Severity Score; GCS, Glasgow Coma Score

Data evaluation and quality control

Data were collected in each ICU by experienced physicians using standardized electronic case report forms and automatically entered into a computer program. The program included reliability checks based on ranges for physiological and biological data, and logical checks for inconsistencies and missing data. Extensive data cleaning was carried out by the Steering and the Writing Committees, including queries addressed to the investigators for questionable or missing data. The results are expressed as proportional values for qualitative variables.


Prescription of red blood cell transfusion

Of the 419 patients assessed for eligibility two were excluded because of chronic anemia or leukaemia. Among the 417 patients included in the final evaluation of red blood cell transfusion prescriptions, the average hemoglobin concentration was 10.8 ± 0.9 g/dl. Twenty-nine patients (7%) received at least one pack of red blood cells on the study day. In 19 patients the transfusion was justified by a hemoglobin level less than 7 g/dl (average 6.1 ± 0.5 g/dl) associated with hemorrhagic shock and acute hemorrhage in the context of multiple trauma or GI bleeding. In a further three patients the main reason for transfusion was a hemoglobin level between 7 and 10 g/dl (average 7.9 ± 0.7 g/dl) in the context of acute coronary insufficiency (n = 2) or the early therapy of septic shock (n = 1). According to our predefined algorithm (Fig. 1), transfusion was deemed appropriate in these 22 (76%) patients. Seven remaining patients (24%) received “inappropriate” red blood cell transfusion. In these patients the average hemoglobin level was 7.8 ± 0.3 g/dl, but no specific explanation was found to justify this prescription. Of the 390 patients who did not receive red blood cell transfusion four (1%) had a hemoglobin level below 7 g/dl and should have been transfused but were not; no explanation was given as to why they did not receive blood transfusion (Table 2).
Table 2

Appropriateness of red blood cell transfusion in 417 patients according to treatment algorithm (Fig. 1) developed from the medical literature

Red blood cell transfusion

Yes (n = 29)

No (n = 390)


Indication present

22 (76%)

4 (1%)


No indication present

7 (24%)

386 (99%)


Tidal volume in ARDS

Of the 419 patients assessed for eligibility 229 (55%) were receiving mechanical ventilation on the study day. We evaluated ventilatory settings in 45 of these patients (20%) who had a reported diagnosis of ARDS, and in whom PBW could be calculated (two patients were excluded because one patient presented with elevated intracranial pressure and another patient had a lung transplant). The volume-assist control mode was used for the majority of patients receiving mechanical ventilation and all patients with ARDS. The mean tidal volume in patients with ARDS was 7.2 ± 0.5 ml/kg PBW and mean plateau pressure was 28 ± 5 cmH2O. According to our predefined algorithm (Fig. 2), tidal volume setting was considered appropriate in 37 of 45 patients (82%) with ARDS. In the 8 patients (18%) receiving “inappropriate” tidal volume the average tidal volume was 8.8 ± 0.6 ml/kg PBW and plateau pressure 35 ± 4 cmH2O.

Prescription of stress ulcer prophylaxis

Of the 419 patients assessed for eligibility in the evaluation of prescription of stress ulcer prophylaxis 23 were excluded because they had previously been treated for gastroduodenal ulcer with proton pump inhibitor therapy (11 before admission to the ICU and 12 during ICU stay). Of the 396 remaining patients 128 (32%) received preventive therapy for upper GI bleeding: 85 (67%) a proton pump inhibitor, 27 (21%) an H2-receptor antagonist (ranitidine), and 16 (12%) sucralfate. According to our algorithm (Fig. 3), the prescription was appropriate in only 24 (19%) of these 128 patients, and thus the remaining 104 (81%) patients were treated unnecessarily. Conversely, among the 268 patients who were not treated, an indication for stress ulcer prophylaxis was present in 28 (10%; Table 3).
Table 3

Appropriateness of stress ulcer prophylaxis in 396 patients according to treatment algorithm (Fig. 3) developed from the medical literature

Stress ulcer prophylaxis

Yes (n = 128)

No (n = 268)


Indication present

24 (19%)

28 (10%)


No indication present

104 (81%)

240 (90%)



This is the first study to assess the bedside application of research results from the medical literature for three frequently used interventions in critically ill patients, including transfusion of red blood cells, low tidal volume in ARDS, and preventive therapies for upper GI bleeding. The main finding of our study is the relatively high rate of appropriateness as regards tidal volume settings, but the adherence to recommendations for blood transfusion and stress ulcer prophylaxis was mediocre.

This discrepancy between the adherences to guidelines for these interventions may result from numerous factors. Firstly, the nature of the intervention may impact on its use. Red blood cell transfusion is considered to be potentially associated with serious adverse events, particularly in France since the blood contamination scandal in the 1990s. Since then the use of blood transfusion has been severely restricted by French regulatory authorities, and physicians are reluctant to prescribe blood transfusions. The very low rate of transfused patients bears witness to this reticence. The salient message from the report by Hebert et al. [9] was that a restrictive blood transfusion policy impacts favorably on ICU patients' outcome, which corresponds to physicians' beliefs, making them more inclined to recommend this policy. Furthermore, some hospitals may have a stronger tradition of applying clinical practice guidelines. A recent study [22] conducted among Canadian critical care practitioners evaluated red cell transfusion practice in the critically ill and examined changes in practice over time. In this study there was a significant decrease in transfusion and use of single-unit transfusion over time. Nonetheless, almost one quarter of the transfusions carried out on the day of our study were performed unnecessarily. Increased awareness of the indications for blood transfusion and the potential adverse effects could help to reduce the number of unnecessary transfusions over time.

Concerning tidal volume in ARDS, adoption of the low tidal volume strategy has been slow even in centers having participated in the ARDSNet trial, as demonstrated by Weinert et al. [23]. On the other hand, physicians in Europe [23] may have more easily endorsed the recommendation because patients with ARDS appeared to have been more routinely treated with much lower tidal volumes [2426]. Secondly, the simplicity of the recommendation is also of paramount importance, as shown by the relatively good level of application of standards for blood transfusion and tidal volume in ARDS. Thirdly, the robustness of the recommendation may also influence the level of uptake by physicians, as shown in this study for stress ulcer prophylaxis, where 81% of patients were treated unnecessarily. It should be noted that this intervention differs from the other two in that it concerns drugs for which the potential influence of industry-driven recommendations (for example, for a larger use of proton pump inhibitors) is more likely than for the other two interventions under study. Admittedly, there are few recent data in the literature [1721] concerning this intervention, an important factor that needs to be taken into account when evaluating the appropriateness of prescription. In addition, the low cost of these drugs encourages wider use.

One reason for the poor adherence to recommendations in this area is the inconsistency of data from the literature and the lack of a strong consensus on which ICU patients should receive prophylaxis. In addition, the perceived low consequences of GI bleeding relative to other complications which may occur during ICU care could encourage physicians not to devote much attention to this problem and its prevention. A recent study [27] demonstrated that no single strategy of stress ulcer prophylaxis is preferred when mortality is used as the outcome. Pitimana et al. [28] demonstrated that the implementation of clinical practice guidelines for stress ulcer prophylaxis increased the appropriateness of prescription (from 75.8% to 91.1%) and decreased the cost of care, but without significant impact on clinical outcomes. In the absence of a clinical trial demonstrating survival benefit the individual clinician's assumptions regarding the effect of prophylaxis on GI bleeding and potentially increased risk of pneumonia, as well as consideration of the attributable mortality of pneumonia vs. GI bleeding, are likely to impact substantially on physicians' attitudes and decisions [20, 29]. Lastly, the experienced practitioners may be aware of current research but have defensive objections to treating patients with a protocol that minimizes their ability to adjust therapy based on an individual's physiology, or may believe their patient's condition is not representative of those in clinical trials [30].

There are several potential limitations associated with the methods used in this study, including social responses bias and the response rate. An acceptable proportion (44 centers, 37%) of all centers contacted agreed to participate. Naturally, this relatively good participation rate provides significant potential for bias, as there is a concern that only the high-performing, highly motivated centers will engage in such a study. This would further bias the results towards an overestimation of compliance with guidelines in that the participating units may favor high-performing, quality ICUs. This percentage of participation may be explained in part by the fact that the main objective of the study (namely to evaluate routine practice with respect to three specific interventions) was not stated in the first phase of the study in order to maintain blinding of the investigators to study goals and minimize bias during the practice survey. However, this characteristic likely jeopardized the motivation of ICU physicians to record data for the study, and it would provide no protection against the tendency for only highly motivated centers to engage in such a study. The results of this study cannot be extrapolated to all ICUs in France, and particularly in Switzerland, Belgium, or Canada where there were only one or two participating centers.

This study is in line with efforts at European level to evaluate professional practices, and quantify the differences between what is recommended in clinical guidelines and/or the medical literature, and what actually happens in daily routine practice at the bedside. Lastly, no subgroup analyses were possible (e.g., academic vs. nonacademic hospitals) in view of the low number of participating centers.


The rate of use of standard medical therapies at the bedside in the setting of intensive care appears satisfactory, as far as tidal volume in ARDS is concerned. Blood transfusion was performed appropriately in the majority of patients, but a considerable proportion of patients were transfused unnecessarily. Similarly, stress ulcer prophylaxis is also less well implemented. Translation of research results into actual clinical practice at the bedside faces various hurdles which may differ according to specific interventions. Practice surveys are useful to inform strategies currently developed to assess practices of health-care professionals and develop strategies for more effective dissemination of medical knowledge, and ultimately improve management and outcomes of patients.


TECLA Steering Committee members were: F. Feihl, MD, H. Mentec, MD, D. Annane, MD, PhD, C. Brun-Buisson, MD, PhD, C. Melot, MD, PhD, P. Vignon, MD, PhD. Members of the Epidemiology and Clinical Research Commission (Commission d'Epidemiologie et de Recherche Clinique) were: Hervé Mentec (Argenteuil, France), Djillali Annane (Garches, France), Pascal Beuret (Roanne, France), Julien Bohé (Lyon, France), Bernard Bouffandeau (Elbeuf, France), Fabrice Bruneel (Versailles, France), Christian Brun-Buisson (Créteil, France), Yves Cohen (Bobigny, France), Robin Cremer (Lille, France), François Feihl (Lausanne, Switzerland), Bertrand Guidet (Paris, France), Mercé Jourdain (Lille, France), Jean-Michel Liet (Nantes, France), Christian Melot (Brussels, Belgium), Mehran Monchi (Massy, France), Jean-Charles Preiser (Liege, Belgium), Jean-Pierre Quenot (Dijon, France), Jean-Philippe Rigaud (Dieppe, France), Marie-Denise Schaller (Lausanne, Switzerland), Philippe Vignon (Limoges, France), Christophe Vinsonneau (Paris, France). Participants by country (listed alphabetically) were: France: C. Galland, Saint-Omer Hospital; C. Brun-Buisson, Henri Mondor Hospital, Créteil, E. Maury, Saint-Antoine Hospital, Paris; M. Jourdain, R. Salengro Hospital, Lille; P. Beuret Roanne Hospital; S. Moulront, Dunkerque Hospital; N. Milesi-Defrance, General Hospital, Dijon; I. Coquet and J.P. Quenot, Bocage Hospital, Dijon; E. Boulet R. Dubos Hospital, Pontoise; L. Holzapfel Fleyriat Hospital, Bourg en Bresse; D. Annane, Raymond Poincaré Hospital, Garches; S. Jamali, Dourdan Hospital; H. Mentec, Victor Dupouy Hospital, Argenteuil; M. Monchi, J. Cartier Institut, Massy Palaiseau; D. Goldran-Toledano, Sainte Justine Hospital, Gonesse; P. Vignon, Dupuytren Hospital, Limoges; M. Fartoukh, Tenon Hospital, Paris; J.C. Farkas, Saint André Polyclinic; J.P. Rigaud, Dieppe Hospital; J.L. Ricome Poissy Hospital, St Germain en Laye; C. Broux, Michalon Hospital, Grenoble; A. Mercat, Angers Hospital; J. Bohé Lyon Sud Hospital; P.E. Bollaert, Central Hospital, Nancy; D. Villers, Hotel Dieu Hospital, Nantes; A. Combes, Pitié Salpêtrière Hospital, Paris; B. Raynard, G. Roussy Institut, Villejuif; D. Dubois, Arras Hospital; J.F. Loriferne, St. Camille Hospital, Bry sur Marne; Richard C. Bicêtre Hospital; I. Mohammedi, E. Herriot Hospital, Lyon; T. Boulain, Orléans Hospital; P. Tourneux, Amiens Hospital; A. Lepape, Lyon Sud Hospital, Lyon, Switzerland: C. Vannay-Boubiche, Lausanne Hospital; F. Feihl, Lausanne Hospital; MD Schaller Lausanne Hospital Canada: P. Jouvet, Sainte Justine Hospital, Montreal, Quebec. Belgium: V. Fraipont, Citadelle Hospital, Liege; C. Mélot, Erasm Hospital, Brussels. The TECLA Study Group is indebted to the Société de Réanimation de Langue Française, the company SCDMI ( and to the Polytechnic School of the Free University of Brussels. We gratefully thank all the participating members of the study. We thank Fiona Ecarnot for editorial assistance.

Copyright information

© Springer-Verlag 2008