Bleeding is a frequent complication following surgery that can lead to substantial morbidity and mortality [1, 2]. During surgery, or when managing trauma, surgeons may be faced by mild to severe bleeding necessitating simple or intricate procedures according to anatomical location, the proximity of adjacent structures, and type of tissue harbouring the bleeding source. In such situations surgeons must decide how best to stem the bleeding using available products and techniques together or sequentially.

Insufficient haemostasis can increase operative time and when excessive may need whole blood or blood constituent transfusions. Further complications include longer hospitalisation, risk of infection, and ventilation requirement. Haemostasis is usually performed mechanically (sutures, ligatures, clips, staples) or with various electrically generated coagulation systems (electrocautery, argon beam lasers, ultrasound) [3, 4].

Blood transfusions and reoperations are not uncommon sequellae when excessive bleeding follows surgery, even in France where blood transfusions have dramatically decreased because of AIDS and hepatitis risks [5, 6]. Between 1992 and 2001 the French Blood Public Organisation [7] reported a fall of 2 million (23.6%) blood transfusion units administered per year. In 2009 the National Blood Bank delivered a total of 2.2 million such units in France, with 25% administered to surgical patients [8] mainly for trauma, orthopaedic surgery and cardiac surgery.

Post-surgical bleeding complications can create a significant economic impact. Boucher and Hannon [9] estimated that haematocyte transfusions in the US cost from $1,840 to $2,760 per unit, plus operating theatre time of $1,890 to $3,150, intensive care unit supervision, and ventilator support, amounting in total to $4,810.

Little information is currently available on bleeding complication rates and related costs in European countries. Hence the availability of a fairly comprehensive hospital database in France provides a unique opportunity to examine this important issue. For the present study we assumed that bleeding episodes could be identified in the database. Such information would permit a comparison of hospitalisation costs both with and without blood loss.

The primary objectives of the present French hospital database analysis were: (1) to estimate the frequency of surgical DRGs associated with blood or blood constituent transfusions administered to ≥10% of WB patients; and (2) to compare the estimated costs of bleeding episodes and transfusions required by these patients versus the costs of WoB patients drawn from similar surgical DRGs.

Our results indicate that bleeding events during major surgery in France (2008) requiring blood or constituent transfusions were relatively common, although frequencies varied between DRG classes. The three DRG combinations most frequently associated with blood transfusions were heart transplants (25.2%), major hip/knee and multiple orthopaedic interventions (24.7%), and pulmonary transplants (24.0%).

In general, surgical patients requiring blood transfusions were more frequently female, elderly and liable to more secondary diagnoses. Also DRGs involving numerous surgical procedures were more often associated with blood transfusions requirements. After adjustment for age, gender, number of procedures and secondary diagnoses, longer hospital admissions and higher costs characterised surgical DRGs, with ≥10% of patients given blood or blood constituent transfusions. When assessing the LOS durations of WB patients we were obliged to perform indirect analyses based on LOS frequencies in excess of DRG averages, so as to accommodate the French algorithm determining DRG codes. Hospital LOS is an important factor when determining DRG codes. The presence of bleeding complications can lead to more severe and more costly DRGs on account of secondary diagnoses and longer LOS durations. Although longer hospitalisations automatically lead to more severe DRGs, this does not imply a direct linkage between hospital LOS and surgical procedure costs, with or without bleeding complications. Indeed, a rapid and fatal complication could direct a patient’s diagnosis to a less severe DRG because the LOS was insufficiently long.

Very few papers have explored the clinical and economic consequences of severe bleeding during surgery. Most have focussed on how to reduce the need for blood transfusion by screening for anaemia and treating affected patients with erythropoietin stimulating agents, as before elective orthopaedic surgery. We identified a recent US analysis by Morton et al. [13] which estimated the increase in average LOS at 2.5 days and found that costs were $17 194 higher for a transfused cohort compared to controls (p < 0.0001). However, there are many reasons for caution, e.g., transfusion data in the US study may have been more complete than reported here for France, i.e. the US difference was based on all hospitalisations, instead of selected hospitalisations in our study which used a different design; differences between the US and French Healthcare systems also exist in terms of medical practice, hospitalisation durations, and costs. Nonetheless, both studies point to similar conclusions in terms of higher costs associated with bleeding complications.

The present study has several limitations. First, in the absence of a specific directive and hospital record code, whole blood and blood constituent charges are not formally included in the database, hence it is probable that transfusion frequency and costs were under-estimated. This is supported by the following estimate for the 77,521 surgical patients in the present study. A ‘one day’ national survey in 2008 identified 742 surgical patients transfused with a total of 5,765 erythrocyte units, i.e. 7.8 units / patient [8]. Accordingly, the 77,521 surgical patients in the present study would have received a total of 7.8 × 77,521 = 604,663 erythrocyte units. The latter number, however, amounts to just one-third of the 2,013,863 erythrocytes units transfused in French hospitals during 2006 [7]: i.e. 2,013,863 / 604,663 = 3.3.

This bias could be important but should in fact underestimate the difference in costs and LOS as a part of patients with bleeding but not coded so and were included in the WoB group.

Second, because our analysis was retrospective, unreported factors may have confounded relationships between surgical bleeding consequences, LOS, and costs. Third, the PMSI database does not provide cost information, hence costs were estimated from the National costs scales and an application of average costs to surgical DRGs associated with severe blood loss. This limitation did not allow us to take into consideration the specific distributions of costs and LOS.

Fourth, French hospital networks are providing pooled information on their hospitals that does not allow taking into account a centre effect and to study relationship of hospital characteristics and costs with bleeding rates.

Fifth, the methods that we selected could have some impact on our results. Our age groups were large, especially for the younger patients. Even, if paediatric cases were very rare (less than 1%) younger adults could present less frequently bleeding and require less transfusions middle age patients and this could have biased our estimates. Lastly, we did not analyse the influence of the sector on the costs consequences of bleeding because we thought it could be redundant with the patient’s characteristics.

Then, it appears that our study can be considered as a pilot study because of the possible important bias that could have underestimate of overestimate our LOS and costs results. Further and more precise analyses are needed to confirm our results in France as similar studies could also be performed in other countries where DRG systems and costs could be different and lead to other conclusions.

In conclusion, this study estimated increases of hospital LOS and costs for patients with severe surgical bleeding complications in France. Hospitalisations longer than the average LOS across all DRGs indicated significant costs associated with severe post-surgical bleeding. Awareness of the clinical and economic consequences of severe post-surgical bleeding provides an important framework when evaluating blood conservation strategies.