Dr Pierre Chopard, Service of Quality of Care, University Hospitals of Geneva, CH-1211 Geneva 14, Switzerland. (fax: +41-22-372-99-21; e-mail: firstname.lastname@example.org).
Aim of the study. To examine the frequency and adequacy of thromboprophylaxis in acutely ill medical patients hospitalized in eight Swiss medical hospitals.
Methods. A cross-sectional study of 1372 patients from eight Swiss hospitals was carried out. After exclusion of patients (275) given therapeutic anticoagulation, 1097 patients were audited. The adequacy of thromboprophylaxis was assessed by comparison with predefined explicit criteria.
Results. Of 1097 patients, 542 (49.4%) received thromboprophylaxis. According to the explicit criteria, 644 (58.7%) should have been on prophylaxis (P < 0.001, when compared with the rate observed). The rate of prevention differed widely between hospitals (from 29.4 to 88.6%) with no difference between teaching and nonteaching hospitals. According to the explicit criteria, a substantial proportion (44.9%) of the patients who should have been treated were not. Conversely, 41.3% of the patients were unnecessarily treated.
Conclusions. Even though the appropriateness of the explicit criteria used could be challenged, our data suggest that the current practice is associated with important uncertainty leading to both overuse and underuse of thromboprophylaxis in patients hospitalized in medical wards. More efforts are urgently needed to develop new or endorse existing explicit, evidence-based criteria and guidelines for thromboprophylaxis in this population of patients.
Deep vein thrombosis and pulmonary embolism (PE) are potentially serious conditions caused by some combination of genetic and environmental factors, including immobilization and surgery. The need for systematic thromboprophylaxis following surgery is based on the high prevalence of postoperative deep venous thrombosis (DVT) and PE, the clinically silent nature of thromboembolic disease, and the potential for major adverse clinical outcomes . However, this practice cannot be transferred to acutely ill patients hospitalized in medical wards except for certain groups of patients (myocardial infarction, paralytic stroke) [2, 3]. Indeed, the risk seems less in these medical patients and there is no clear consensus. Several studies on small patient samples did not provide consistent results to support systematic prophylaxis [4–10]. More recently, the MEDENOX study  showed a highly significant reduction (63%) in the risk of venous thromboembolism by thromboprophylaxis with low-molecular weight (LMWH) enoxaparin for acutely ill medical patients. Moreover, the rate of DVT (15%) in the placebo group was similar to the rates observed after general abdominal surgery . Only few studies have already evaluated the adherence of prevention of venous thromboembolism to established criteria in clinical practice in acute ill medical patients. They suggested both underuse of thromboprophylaxis in surgical patients  as well as underuse and overuse in medical patients [13–16] with potential differences between teaching and nonteaching hospitals. Importantly, some authors found that clinical guidelines can improve adequacy of thromboprophylaxis [16, 17].
Because information about the use of thromboprophylaxis in hospitalized medical patients is scarce and a preliminary audit had shown considerable heterogeneity of practice amongst three medical wards of the same institution , we carried out a national survey in Switzerland aiming at (i) establishing the proportion of patients receiving thromboprophylaxis in medical wards; (ii) assessing the appropriateness of thromboprophylaxis according to explicit criteria; (iii) studying potential differences between teaching and nonteaching hospitals.
A cross-sectional 1-day survey was performed in eight Swiss hospitals (teaching and nonteaching hospitals). On the survey day, all patients hospitalized in medical wards of eight hospitals (noted hereafter as A–H) were included.
Data were collected by independent observers on a especially designed standard sheet from medical and nursing records. The doctors in charge of patients were not informed of the aims and schedule of the cross-sectional audit in order to avoid bias because of previous information. The appropriateness of medical diagnoses was not evaluated. Obesity was accepted if the body mass index (BMI) was equal or higher than 30 (for man and woman without distinction). Whenever the BMI could not be calculated but where the diagnosis was noted, it was accepted.
The standard sheet was filled out for each patient present in the wards at the given day.
The main outcome variable was the appropriateness of thromboprophylaxis, assessed by comparison with a predefined set of explicit criteria (Table 1). Although these criteria did not undergo any validation procedure, they are based on the presently available evidence and on the results of the largest study performed in this setting . The presence of one major criterion, two minor criteria or one minor with one additional criterion was required to justify prophylactic anticoagulation. We also conducted an additional analysis on the basis of risk factors recently derived from the MEDENOX study  complemented by the two risk conditions identified in the 2004 ACCP Consensus on prophylaxis of venous thromboembolism as requiring thromboprophylaxis (grade 1A recommendation) , namely stroke and acute myocardial infarction. The observed treatment, according to the decision of the doctor in charge, has been called ‘implicit’ throughout the paper, as opposed to the explicit treatment derived from the predefined criteria.
Table 1. Explicit criteriaa for thromboprophylaxis appropriateness in hospitalized medical patients
Indication for thromboprophylaxis was given in the presence of at least one major, or two minor, or one minor plus one additional criterion .
aRisk factors identified in the complementary analysis of the MEDENOX study are: previous venous thromboembolic event, acute infectious disease, age more than 75 years and cancer. Stroke and myocardial infarction are two risk factors of grade 1A recommendation in the 2004 ACCP Consensus .
Acute myocardial infarction
Acute respiratory failure
Acute cardiac failure
Acute illness + past history of venous thromboembolism
Acute illness + hypercoagulable state
Malignancy/evolutive myeloproliferative disorder
Hormonal treatment (contraceptive or substitutive)
Age more than 60 years
Confined to bed
Chronic venous insufficiency
First, we compared the eight hospitals in terms of mean patients age (anova), sex and use of thromboprophylaxis (chi-squared tests). Then, we adjusted comparison of hospitals in terms of thromboprophylaxis for patients age and sex in logistic regression models. We obtained adjusted odds ratios of thromboprophylaxis use in comparison with a reference hospital. Then, for all patients, we determined whether thromboprophylaxis was indicated based on explicit criteria. We studied overuse amongst patients who did not have an indication and underuse amongst patients who did have one. The prevalence of underuse and overuse was examined across patients characteristics (sex, age groups) and hospitals and, for underuse only, presence of major or minor criteria. A scatter plot of these data was used to investigate the association between underuse and overuse amongst hospitals. This association was measured by the correlation coefficient (r).The level of statistical significance was set at P ≤ 0.05. Data were analysed by means of spss (SPSS-windows, SPSS Inc., Chicago, IL, USA, version 11.0).
A total of 1372 patients were hospitalized on the day of the survey. Amongst them, 275 patients were given therapeutic anticoagulation. Thus, 1097 patients were included in the audit. Their characteristics are displayed in Table 2.
Table 2. Demographic data of the included patients and variability of thromboprophylaxis amongst the institutions surveyed
aTwo missing data for age.
bNine missing data for gender.
cAdjusted for age, comorbidities and gender (95% CI).
The P-values refer to the overall comparison between the institutions.
Mean age (±SD)a
Proportion of patients who received a thromboprophylaxis
Proportion of patients who should have received a thromboprophylaxis
Risk (odds ratio) to receive a thromboprophylaxisc
Frequency of thromboprophylaxis
Overall, 542 patients (49.4%) were receiving prophylactic anticoagulation with quite varying rates amongst hospitals (P < 0.001; Table 2). Patients treated with thromboprophylaxis were 6.6 years older than those without prophylaxis (P < 0.001). The odds ratios for thromboprophylaxis were different between hospitals even after adjustment for patient's characteristics (Table 2). Hospital D had an odds ratio of 3.1, indicating a patient hospitalized in this hospital was almost three times more likely to receive thromboprophylaxis than in hospital G.
Appropriateness of thromboprophylaxis
According to the predefined explicit criteria, 644 patients (58.7%) should have been given thromboprophylaxis (Table 2). The difference between the overall rate of thromboprophylaxis observed with the rate established by means of explicit criteria was statistically significant (P < 0.001). According to the explicit criteria, 289 (44.9%) who should have been treated were not, whereas 187 patients (41.3%) were unnecessarily treated. Agreement between observed treatment rate (implicit criteria) and explicit criteria was very poor (κ-coefficient was 0.13). Amongst the 379 patients showing at least one major risk factor according to the explicit criteria, 152 (40.1%) did not receive any prophylaxis (Table 3). A total of 137 patients (51.7%) with two minor criteria or one minor criterion plus one additional were also not treated.
Table 3. Rate of thromboprophylaxis by explicit criteria
Observed rate of thromboprophylaxis (N)
Major factor (at least one)
Acute myocardial infarction
Acute respiratory failure
Acute cardiac failure
Acute illness + past history of thromboembolism
Acute illness + hypercoagulable state
Minor or additional factors (at least two factors)
Rates of underuse and overuse were also different across gender, age and hospitals (Table 4). Differences across hospitals persisted despite adjustment for patients characteristics (not shown). Interactions between age, gender and institutions were not significant. Finally, hospitals that tended to have a high rate of overuse had only little underuse and inversely (Fig. 1).
Table 4. Underuse and overuse of thromboprophylaxis according to explicit criteria (see Table 1)
Patients needing thromboprophylaxis
Yes (N = 644)
No (N = 453)
aPatients needing thromboprophylaxis and not receiving it.
bPatients not needing thromboprophylaxis and receiving it.
cSeven missing data underuse/two missing data overuse.
NA, not applicable.
Major (≥1 criteria)
Minor (≥2 criteria)
The analysis made on the basis of risk factors identified in the MEDENOX study yielded similar results in terms of variability (results not shown).
Role of the teaching or nonteaching status of the institution
The rate of overuse or underuse of thromboprophylaxis was not statistically significantly different between teaching and nonteaching institutions (Table 4).
Currently, prevention of venous thromboembolism regimens of acutely ill patients in medical wards are left to the appreciation of doctors, except for stroke and acute myocardial infarction which represent established indications for thromboprophylaxis in medical patients. Our results suggest that current practice is associated with important uncertainty leading to both substantial overuse and underuse of thromboprophylaxis. Some institutions are restrictive in practice (underuse) whilst other has too broad indications (overuse).
These results confirm and extend results obtained in previous studies in the setting of thromboprophylaxis in both surgical [12, 17] and medical services [13–16]. They clearly underscore the inadequacy and heterogeneity of current practice as thromboprophylaxis was indicated and not prescribed in about 45% (range: 11.7–67.7%) of cases and not indicated but prescribed in about 40% (range: 13.0–88.9). Based on predefined criteria, prophylaxis was justified in only half of the cases and agreement between implicit and explicit criteria was extremely poor. Even more worrying was the fact that almost 50% of the patients not given prophylaxis should have been receiving it according to the criteria mentioned above, including a substantial number of patients with active malignant diseases. Of some concern is also the difference in the rate of thromboprophylaxis in the eight hospitals surveyed. Although some specificity of the patients could partly explain the different rates, they are likely to be due to arbitrary attitudes. Indeed, these differences tend to disappear with the use of explicit criteria.
In spite of these clear-cut findings, our survey has two main limitations. First, the explicit criteria used have never been validated before. However, they are compatible with the only large-scale randomized-controlled study , and the use of risk factors identified in the MEDENOX study gave similarly poor results. Secondly, contraindications to thromboprophylaxis were not considered, which may overestimate the underuse. However, thromboprophylaxis is rarely contraindicated in nonsurgical patients and in the study by Aujesky et al.  in which this item was taken in consideration, the rate of underuse was very similar (44%). Thirdly, quality of data collection did not undergo specific assessment.
In conclusion, decision-making for thromboprophylaxis in current clinical practice seems too random in acutely ill medical patients. This may reflect large uncertainty amongst the medical community. Further research is needed to develop or implement presently available explicit evidenced-based criteria and to standardize the indications for prevention of venous thromboembolism in these patients.
Conflict of interest
Authors have no conflict of interest, except Dr Rubino who was employed by Aventis at the time of the survey. The article addresses the adequacy of thromboprophylaxis and not the efficacy of an Aventis compound.
The excellent cooperation of Dr G. Mombelli (Ospedale La Carità, Locarno) and the critical reading of the manuscript by Dr T. V. Perneger and Dr P. Bovier (University Hospital of Geneva) are gratefully acknowledged.
Grant support: the survey was supported by an unrestricted educational grant from Aventis Pharma (Zurich, Switzerland).