SEARCH

SEARCH BY CITATION

Keywords:

  • Internal Medicine;
  • intervention;
  • prospective randomized trial;
  • recommendation;
  • thromboprophylaxis

Abstract.

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. Investigators
  10. References
  11. Appendix

Objective.  To assess the effectiveness of providing doctors with written thromboprophylaxis prescription aids based on current recommendations.

Design.  A prospective trial of specific anticoagulant prescription forms: a 1-day survey before and after the intervention in each centre.

Setting.  30 Internal Medicine departments of Assistance Publique-Hôpitaux de Paris.

Subjects.  All inpatients were included, except those who were either admitted or discharged on the day of the survey, and those receiving curative anticoagulant treatment.

Interventions.  The study included three parts: (i) a 1-day baseline survey; (ii) over the following 3-month period, departments were randomized into two groups: all practitioners in wards allocated to the intervention group were required to systematically use specific anticoagulant prescription forms, whilst doctors in the control group continued prescribing according to their usual practices and (iii) a 1-day postintervention survey.

Main outcome measure.  The proportion of prescriptions in accordance with the recommendations.

Results.  1469 patients were included. The intervention produced a significant reduction in the frequency of over-prescriptions, from 25% to 14% of the patients who did not meet the guideline criteria (adjusted OR: 0.3; 95% CI: 0.1–0.8). Using specific forms did not improve under-prescription of anticoagulants. A little over 60% of the patients who met guideline criteria for thromboprophylaxis were prescribed anticoagulants in both intervention and control wards, either at baseline or after intervention.

Conclusions.  Multitargeted interventions using a variety of means and strategies should still be considered to improve prescriptions that may have a significant impact on health expenses and, most importantly, on patients outcomes.


Introduction

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. Investigators
  10. References
  11. Appendix

There is now sufficient evidence that low-molecular weight heparins (LMWH) reduce the risk of thromboembolic events in medical patients. Mismetti et al. estimated in a meta-analysis that heparins reduced the risk of asymptomatic deep venous thrombosis (DVT) by 56% and the risk of pulmonary embolism by 52% in patients with acute medical illnesses [1]. The MEDENOX (enoxaparin), PREVENT (dalteparin) and ARTEMIS (fondaparinux) studies consistently found a relative risk reduction of 50% for proximal or symptomatic thromboembolic events in medical patients, whilst the risk of major bleeding was minimal (a nonsignificant excess of 1% in active treatment groups) [2–5]. Nevertheless, thromboprophylaxis is misused in current medical practice, including both overuse and underuse [6–10]. This is what Chopard et al. [6] observed in a recent cross-sectional study of 1372 patients in eight Swiss hospitals. According to explicit criteria, 44.9% of the patients who should have been treated were not, and 41.3% of the patients receiving prophylaxis were unnecessarily treated [6]. Both underuse and overuse of thromboprophylaxis for medical patients have also been reported in Italian [9] or Spanish [10] hospitals. A series of explanations have been suggested in the medical literature: venous thromboembolism (VTE) is not perceived as a significant hazard; risk assessment is somewhat difficult; there is insufficient awareness of guidelines/studies; there are no prospective studies of untreated patients; and, most importantly, medical patients are very heterogeneous. More specifically, underuse of thromboprophylaxis in medical patients may be attributed to insufficient recognition of the risk of VTE for these patients, whilst overuse may be due to an erroneous assessment of the actual risk associated with specific clinical conditions and patient characteristics. Several authors concluded that explicit thromboprophylaxis guidelines should be developed for medical patients to improve the appropriateness of prescriptions.

The primary objective of our study was to assess the effectiveness of providing doctors with written prescription aids, based on current recommendations, on the appropriateness of thromboprophylaxis prescriptions.

Patients and methods

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. Investigators
  10. References
  11. Appendix

Thromboprophylaxis prescription guidelines were derived from the literature by a group of thrombosis and internal medicine specialists [7, 8]. According to these guidelines, thromboprophylaxis was recommended if a patient presented one of the three following conditions (see Annex): (i) at least one major risk factor (e.g. recent myocardial infarction, stroke with recent limb paresis, congestive heart failure, severe acute respiratory insufficiency); or (ii) at least two simultaneous conditions associated with increased risk of a thromboembolic event (e.g. recent immobilization because of an acute medical situation, severe dehydration, active cancer); or (iii) the association of one such condition and at least one patient characteristic associated with a moderately increased risk (e.g. age ≥70 years or history of venous thromboembolic disease).

The study took place during one rotation period of residents, and included three parts: (i) in January, a 1-day baseline survey was performed in 30 Internal Medicine departments of Assistance Publique-Hôpitaux de Paris; (ii) over the following 3-month period Internal Medicine departments were randomized into two groups: all practitioners in wards allocated to the intervention group were required to systematically use specific anticoagulant prescription forms that featured the recommended prescription criteria, whilst doctors in the control group continued prescribing according to their usual practices and (iii) a 1-day postintervention survey was repeated in April.

In each ward, baseline and postintervention surveys were performed on days randomly selected within 1-month periods. All inpatients were included, except those who were either admitted or discharged on the day of the survey, and those receiving curative anticoagulant treatment. The doctor in charge of each patient was asked to answer an anonymous questionnaire including the patient's characteristics (age, sex, body weight, date of admission), the presence or absence of a series of venous thromboembolic risk factors, and his or her evaluation, on a 10 cm visual analogical scale, of the patient's risk of anticoagulation haemorrhagic complications.

Analysis

Under-treatment was defined as the patient not being prescribed anticoagulants although thromboprophylaxis was recommended according to the guidelines; over-treatment as a patient receiving prophylaxis although it was not recommended. The effectiveness of the intervention on reducing anticoagulant prescriptions when not recommended and on increasing the likelihood of anticoagulant prescriptions when recommended was assessed in two logistic regression models controlling for the type and number of risk factors and for the haemorrhagic risk evaluated by the doctor. Data were analysed with stata 8.0®, using appropriate multilevel survey analysis procedures to take into account possible associations between prescription decisions for patients treated by the same practitioner.

Results

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. Investigators
  10. References
  11. Appendix

The overall design and primary findings of this study are illustrated in Fig. 1. A total of 939 patients (502 in control wards and 437 in intervention wards) were included in the baseline survey, and 863 (445 in control wards and 418 in intervention wards) in the postintervention survey. Between 14% and 21% of these inpatients were receiving curative anticoagulant treatments and were excluded from the study. Overall, the proportion of patients who were prescribed anticoagulant thromboprophylaxis decreased from 40% (144 of 353) to 32% (115 of 360) in the intervention group, whilst this proportion remained stable in the control group: 38% of the patients (150 of 397) at baseline, and 36% of the patients (133 of 359) at the postintervention survey. In parallel, the proportion of patients eligible for thromboprophylaxis according to our guidelines decreased in both groups, from 41% (145 of 353) to 36% (129 of 360) in the intervention group, and from 39% (155 of 397) to 30% (107 of 359) in the control group. Median age, median duration of hospitalization at the date of the survey and sex ratio of the patients included did not differ significantly between intervention and control groups or between surveys (Table 1).

image

Figure 1.  Overall design and primary findings.

Download figure to PowerPoint

Table 1.   Characteristics of included patients
 BaselinePostintervention
Control groupIntervention groupControl groupIntervention group
N397353359360
Median age (years)75717470
Median duration of hospitalization (days)10799
Sex ratio (male, %)51495251

The effectiveness of the intervention in reducing over-treatment can be seen in the reduction, relative to the baseline survey, of the proportion of patients who were prescribed anticoagulants amongst those who did not meet the guideline criteria for thromboprophylaxis (Fig. 2). This proportion decreased by 44% (from 25% to 14% of the patients) in the intervention wards whilst it increased by 17% (from 22% to 26% of the patients) in the control wards. The reduction in unwarranted anticoagulant prescriptions in the intervention wards specifically affected patients who did not present any recognized risk factors (an 86% reduction, from 8% to 1% of the patients), or presented only one risk factor or clinical condition (a 60% reduction, from 28% to 11% of the patients). However, anticoagulant prescriptions increased by 19% (from 50% to 59%) for patients who presented two or more characteristics associated with moderately increased risk, although this status did not warrant thromboprophylaxis in our recommendations. In contrast, the proportion of patients over-treated increased by 17% (from 22% to 26%) in control wards, with increases in all three patients subgroups.

image

Figure 2.  Effects of the intervention on over-treatment. Over-treatment is defined as the proportion of patients who were prescribed anticoagulants amongst the patients who did not require thromboprophylaxis according to our guidelines. The percentage increase or decrease in over-treatment is computed as the difference between postintervention and baseline over-treatment levels, divided by the baseline level. It is computed first for all the patients who did not require thromboprophylaxis, then for each of the three subgroups defined by the type of criteria associated with anticoagulant prescription.

Download figure to PowerPoint

The results of a logistic regression analysis confirmed the statistical significance of the effectiveness of the intervention in reducing over-treatment (Table 2). Patients who did not meet thromboprophylaxis criteria according to our guideline were significantly less likely to be prescribed anticoagulants in the postintervention survey in intervention wards. This analysis also confirmed that the numbers of clinical conditions and individual risk factors present for each patient were associated with an increased probability of anticoagulant prescription, whilst the perceived risk of haemorrhagic complications played a moderating role.

Table 2.   Predictors of anticoagulant prescriptions for patients who did not meet criteria for thromboprophylaxis: logistic regression analysis (generalized estimating equation)
 Odds ratio (95% confidence interval) P-value
Number of clinical conditions associated with increased risk2.2 (1.3–3.8)0.004
Number of patient characteristics associated with increased risk3.4 (2.8–4.2)0.000
Risk of haemorrhagic complications of anticoagulants (0–10)0.9 (0.9–1)0.01
Intervention versus control group1.4 (0.7–2.6)0.3
Postintervention survey versus baseline1.4 (0.8–2.5)0.3
Intervention group postintervention0.3 (0.1–0.8)0.015

On the other hand, using specific forms did not affect anticoagulant prescriptions for patients who met the guideline criteria for thromboprophylaxis. Between 62% and 64% of these patients were prescribed anticoagulants in both intervention and control wards, either at baseline or after intervention. In both intervention and control wards, we found similar reductions of under-treatment of patients with major risk factors (from 38% to 30% in the control group and from 38% to 29% in the intervention group), whilst under-treatment increased slightly for patients with combinations of clinical conditions and patient characteristics associated with increased risk (from 38% to 42% in control wards and from 36% to 40% in intervention wards), resulting in a minor overall reduction of under-treatment in both groups (Fig. 3). Logistic regression analysis (Table 3) confirmed that the intervention was not associated with any significant change in anticoagulant prescriptions for these patients, but confirmed that prescriptions were influenced by the number of risk factors and by the doctor's assessment of the risk of haemorrhagic complications, which was consistent with what was found for patients who did not meet thromboprophylaxis criteria.

image

Figure 3.  Effects of the intervention on under-treatment. Under-treatment is defined as the proportion of patients who were not prescribed anticoagulants amongst the patients who did require thromboprophylaxis according to our guidelines. The percentage increase or decrease in under-treatment is computed as the difference between postintervention and baseline under-treatment levels, divided by the baseline level. It is computed first for all the patients who required thromboprophylaxis, then for each of the two subgroups defined by the type of criteria warranting anticoagulant prescription.

Download figure to PowerPoint

Table 3.   Predictors of anticoagulant prescriptions for patients who met criteria for thromboprophylaxis: logistic regression analysis (generalized estimating equation)
 Odds ratio (95% confidence interval)P-value
Number of major risk factors1.6 (1.2–2.2)0.004
Number of clinical conditions associated with increased risk1.3 (0.9–1.8)0.2
Number of patient characteristics associated with increased risk2.5 (1.8–3.3)0.000
Risk of haemorrhagic complications of anticoagulants (0–10)0.8 (0.7–0.8)0.000
Intervention versus control group1.6 (0.8–3.1)0.19
Postintervention survey versus baseline1.4 (0.7–3)0.35
Intervention group postintervention0.7 (0.2–1.8)0.44

Discussion

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. Investigators
  10. References
  11. Appendix

Our baseline survey confirmed the misuse of thromboprophylaxis in medical patients, demonstrated in both unnecessary and insufficient anticoagulant prescriptions.

It should be noted that the present study was not designed to assess the validity of the recommendations provided to the practitioners, but their impact on prescription practices. In this context, our study demonstrates the feasibility of designing prescription forms incorporating practical recommendations and their effectiveness in reducing unwarranted prescriptions.

Nevertheless, this effectiveness was limited by the fact that doctors did not adhere fully to the proposed guidelines. Whilst there was a marked reduction in anticoagulant prescriptions for patients with a single or no risk factor, this reduction did not affect patients with multiple risk factors, although this condition had not been included in the recommendations for thromboprophylaxis. It could be argued that doctors erred, if they did, on the side of caution, balancing the limited risks associated with LMWHs with persistent uncertainties regarding the actual thromboembolic risk associated with multiple risk factors. Overall, the results of the logistic regression analyses suggest that doctors did balance thromboembolic and haemorrhagic risks in making prescription decisions.

On the other hand, incorporating thromboprophylaxis guidelines in anticoagulant prescription forms did not reduce under-prescription. Data from the International Medical Prevention Registry of Venous Thromboembolism (IMPROVE) showed that only 39% of acutely ill hospitalized patients enrolled in the registry received thromboprophylaxis [11]. Whilst inclusion criteria in the MEDENOX study were considered as warranting thromboprophylaxis, only 54% of the patients included actually received anticoagulants [2]; 44.9% of the patients surveyed by Chopard et al. in Swiss hospitals who should have been treated were not [6]. Whilst almost all the patients included in a multicentre chart audit of 29 Canadian hospitals had indications for thromboprophylaxis, only 16% received appropriate prophylaxis [12].

The lower frequency of under-treatment observed in our study (36–38% of the patients who should have received thromboprophylaxis did not) might be explained by the diffusion of DVT prevention practices using LMWH in France [13]. Thromboprophylaxis with LMWH is more frequently prescribed in France than in the USA or the UK, and LMWH is relatively inexpensive for French hospitals. The results of the regression analyses also suggest that some under-prescriptions may be explained by the perceived risk of haemorrhagic complications of anticoagulant prophylaxis.

The fact remains, nevertheless, that our intervention did not reduce under-prescription. These findings should be contrasted with those of interventions using computer-based reminders, linked to patient databases, that have been shown to reduce thomboprophylaxis under-prescription in both medical and surgical settings [14, 15].

More comprehensive implementation strategies, focusing additional efforts on communicating the rationale underlying the guidelines and actively involving clinical opinion leaders with locally recognized expertise in the dissemination process, can also improve compliance with thromboprophylaxis guidelines. In a before and after study carried out in Italy, Scaglione et al. found that an active guidelines implementation strategy produced a significant increase in appropriate prophylaxis prescriptions for high-risk patients (from 25% to 66.7%, P = 0.0075); however, this intervention also induced a small increase in thromboprophylaxis prescriptions for low- and medium-risk patients (from 13.7% to 20.0%) [9]. In a retrospective review of patients admitted to the medical service of a tertiary care centre, Stinnett et al. reported a 67% increase in the proportion of high-risk patients receiving thromboprophylaxis (from 43% to 72% of the patients) following interventions designed to increase awareness of VTE [16].

However, none of these studies was properly controlled, and other factors might have been at play. As it stands, our study design allowed us to assess the specific impact of this intervention on the appropriateness of thromboprophylaxis prescriptions. Using randomized-controlled trials methodology allowed us to contrast the reduction in unwarranted prescriptions observed in intervention wards with their general stability in control wards, controlling for the evolution of patient recruitment and other potential confounding factors. In particular, the results of the MEDENOX study were published shortly before this study was undertaken, and may have contributed to the moderate increase in both warranted and unwarranted anticoagulant prescriptions observed in the control group.

Thromboprophylaxis prescription decisions rely on the assessment of the risks incurred by individual patients [5]. Our study clearly demonstrates how locally developed recommendations, made available at the time of each anticoagulant prescription, may assist in the evaluation of thromboembolic risk and induce a significant reduction of unwarranted prescriptions, reducing thus both unnecessary expenses and iatrogenic risks. These prescriptions aids, however, had no significant effect on under-prescription situations. Computer-generated reminders triggered by patient characteristics and clinical conditions may be more effective in that regard, but require the development of extensive electronic clinical records, which is not currently feasible in most French hospitals. Multitargeted interventions using a variety of means and strategies should still be considered to improve prescriptions that may have a significant impact on health expenses, and most importantly, on patient outcomes.

Acknowledgements

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. Investigators
  10. References
  11. Appendix

This study was endorsed as a collective project by the Collegial Federation of Internal Medicine departments and supported by the Department of Evaluation of Assistance Publique-Hôpitaux de Paris.

The authors also wish to thank Mathieu Richon, Frank Soury and Sophie Patris (Department of Public Health, Faculté Bichat) and Azeb Sebatlab for collecting reliable survey data, and Susan Wright, MPH, for carefully reviewing and editing the final manuscript.

Investigators

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. Investigators
  10. References
  11. Appendix
  1. Dpt, department.

Ambroise Paré Hospital: T. Hanslik (Dpt Pr Baglin), M. L. Chabinat (Dpt Pr Dorra)
Antoine Béclère Hospital: R. Fior (Dpt Pr Galanaud)
Beaujon Hospital: N. Belmatoug, S. Stelianides (Dpt Pr Fantin)
Bicêtre Hospital: M. Bard, J. Wicki (Dpt Pr Delfraissy)
Bichat Hospital: C. Carbon (Dpt Pr Carbon)
Boucicaut Hospital: A. Buu Sao (Dpt Pr Patri)
Cochin Hospital: B. Gombert, L. Heripret (Dpt Pr Sicard)
Foch Hospital: O. Lidove (Dpt Pr Bletry)
Henri Mondor Hospital: A. Schaeffer (Dpt Pr Schaeffer)
Hotel Dieu Hospital: G. Grateau (Dpt Pr Capron)
Jean Verdier Hospital: O. Fain (Dpt Pr Thomas)
Laennec Hospital: H. Durand, F. Deudoune, H. Lietaud, V. Jublance, C. Le Jeunne (Dpt Pr Hugues)
Lariboisière Hospital: I. Mahé (Dpt Pr Caulin), J. P. Kevorkian (Dpt Pr Warnet)
Louis Mourier Hospital: C. Campinos, D. Vincent (Dpt Pr Pradalier), A. Grasland, J. Pouchot, Ph. Vinceneux (Dpt Pr Vinceneux)
Paul Brousse Hospital: F. vayre, H. Monsuez (Dpt Pr Passeron)
Pitié Salpétrière Hospital: T. Genereau, A. Texeira (Dpt Pr Herson)
Raymond Poincaré Hospital: M. Khatibi, D. Schaffo (Dpt Pr Remy)
Saint-Antoine Hospital: I. Delevaux, S. Vigne (Dpt Pr Imbert and Pr Cabane), I. Mabro, C. Tournigand (Dpt Pr Krulik)
Saint-Louis Hospital: M. El Guedj, C. Sauve (Dpt Pr Sereni)
Tenon Hospital: C. Arnal (Dpt Pr Cocheton).

References

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. Investigators
  10. References
  11. Appendix
  • 1
    Mismetti P, Laporte-Simitsidis S, Tardy B et al. Prevention of thromboembolism in internal medicine with unfractioned or low-molecular-weight heparins: a meta-analysis of randomised clinical trials. Thromb Haemost 2000; 83: 149.
  • 2
    Samama MM, Cohen AT, Darmon JY et al. A comparison of enoxaparin with placebo for the prevention of venous thromboembolism in acutely ill medical patients. N Engl J Med 1999; 341: 793800.
  • 3
    Leizorovicz A, Cohen AT, Turpie AG, Olsson CG, Vaitkus PT, Goldhaber SZ; PREVENT Medical Thromboprophylaxis Study Group. Randomized, placebo-controlled trial of dalteparin for the prevention of venous thromboembolism in acutely ill medical patients. Circulation 2004; 110: 8749.
  • 4
    Cohen AT, Davidson BL, Gallus ASet al.; ARTEMIS Investigators. Efficacy and safety of fondaparinux for the prevention of venous thromboembolism in older acute medical patients: randomised placebo controlled trial. BMJ 2006; 332: 3259.
  • 5
    Leizorovicz A, Mismetti P. Preventing venous thromboembolism in medical patients. Circulation 2004; 110: IV139.
  • 6
    Chopard P, Dorffler-Melly J, Hess U et al. Venous thromboembolism prophylaxis in acutely ill medical patients: definite need for improvement. J Intern Med 2005; 257: 3527.
  • 7
    Haas S. Recommendations for prophylaxis of venous thromboembolism: International Consensus and the American College of Chest Physicians Fifth Consensus Conference on antithrombotic therapy. Curr Opin Pulm Med 2000; 6: 31420.
  • 8
    Bergmann JF, Elkharrat D. Prevention of venous thromboembolic risk in non-surgical patients. Haemostasis 1996; 26(Suppl. 2): 1623.
  • 9
    Scaglione L, Piobbici M, Pagano E, Ballini L, Tamponi G, Ciccone G. Implementing guidelines for venous thromboembolism prophylaxis in a large Italian teaching hospital: lights and shadows. Haematologica 2005; 90: 67884.
  • 10
    Vallano A, Arnau JM, Miralda GP, Perez-Bartoli J. Use of venous thromboprophylaxis and adherence to guideline recommendations: a cross-sectional study. Thromb J 2004; 2: 3.
  • 11
    Tapson VF, Decousus H, Piovella F et al., for the IMPROVE Investigators. A multinational observational cohort study in acutely ill medical patients of practices in prevention of venous thromboembolism: findings of the International Medical Prevention Registry on Venous Thromboembolism (IMPROVE). Blood 2003; 102: 1154 (abstract).
  • 12
    Kahn SR, Panju A, Geerts W et al., for the CURVE Study Investigators. Multicenter evaluation of the use of venous thromboembolism prophylaxis in acutely ill medical patients in Canada. Thromb Res 2006 (Epub ahead of print).
  • 13
    Bergmann JF, Mouly S. Thromboprophylaxis in medical patients: focus on France. Semin Thromb Hemost 2002; 28(Suppl. 3): 515.
  • 14
    Kucher N, Koo S, Quiroz R et al. Electronic alerts to prevent venous thromboembolism among hospitalised patients. N Engl J Med 2005; 352: 96977.
  • 15
    Durieux P, Nizard R, Ravaud P, Mounier N, Lepage E. A clinical decision support system for prevention of venous thromboembolism: effect on physician behavior. JAMA 2000; 283: 281621.
  • 16
    Stinnett JM, Pendleton R, Skordos L, Wheeler M, Rodgers GM. Venous thromboembolism prophylaxis in medically ill patients and the development of strategies to improve prophylaxis rates. Am J Hematol 2005; 78: 16772.

Appendix

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Patients and methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. Acknowledgements
  9. Investigators
  10. References
  11. Appendix

Appendix: Guidelines

Doctors in the intervention group were requested to use the following form whenever prescribing anticoagulant thromboprophylaxis.

inline image