This study reports that for complicated proximal femoral fractures that require a re-operation, total treatment cost and length of stay roughly double. Further, mortality and return to own home were significantly worse in complicated cases.
Previous and fairly recent papers on the subject have indicated comparable findings. This is therefore not an original study, but it addresses an important problem.
The authors use a clever matching method for a large cohort: each complicated case was matched based on eight relevant factors (age, ASA grade, type of fracture and surgery, etc) with two control cases taken from the same prospectively collected dataset. A five-year period was analysed with a total of 2257 patients (2360 fractures, both intra- and extra-capsular). Of these, 144 patients had a re-operation as a result of a complication, 53 during their initial admission, 91 as a re-admission.
The complication categories were: failure of fixation (37/144, 26%), superficial infection (17/144 (12 %), deep infection (29/144, 20%), dislocation (32/144, 22%), haematoma (15/144, 10%), and other (14/144, 10%).
Overall, infection was therefore the most frequent complication in 46/144 (32%) of patients. Not only was infection the most frequent complication, it was also the most costly. In patients with deep infection, total treatment cost was 2.9 times that of uncomplicated cases. For fixation failure this ratio was 2.2, for dislocation 2.1. Eleven patients (38%) with deep infection underwent a Girdlestone arthroplasty. The percentage of patients ending with a Girdlestone is not given for the other complication categories, but is likely to be lower. Given the debilitating effect of a Girdlestone, a full health economic analysis is likely to show an even further increase in socioeconomic impact and cost for infection compared with other complications.
Logistic regression analysis was done using 15 independent variables such as smoking, use of anticoagulants or steroids, pre- or post-operative blood transfusion, consultant vs registrar as surgeon, cardiovascular disease etc. Duration of surgery was not included, but may have been interesting with regards to infection. The full model containing all 15 variables was statistically significant, meaning it could distinguish between control and complication cases. Further, the model correctly identified 69% of complicated cases.
Only two independent variables made a unique, statistically significant contribution to the model: 1) post-operative blood transfusion (odds ratio 2.54), and 2) use of enteral steroids (odds ratio 2.41). However, only 6 of 144 patients of the complicated cases used enteral steroids, versus 63/144 who received post-operative blood transfusion. Correspondingly, the 95% confidence interval lower limit is only 1.06 for enteral steroids. This effectively makes post-operative transfusion the only variable that can be interpreted with confidence as an important risk factor for development of complications. This is further borne out in the subgroup analysis for fixation failure, deep infection and dislocation. Only post-operative blood transfusion had a 95% confidence interval lower limit above 2. The number of patients for the other variables, again, was small among the 144 complicated cases: smoking (6), a history of CVA (7) or diabetes (7), for dementia the number is not reported. Thus, only the effect of post-operative blood transfusion is a robust one, and caution must be used when interpreting the results of the other variables. For deep infection, risk factors were dementia (odds ratio = 6.72, 1.16 – 38.8, p = 0.03), smoking (odds ratio = 23.3, 1.45 - 373.64, p = 0.03) and post-operative blood transfusion (odds ratio = 33.13, 5.69 – 192.73, p < 0.01). For dislocation, the operation not being performed by a consultant (odds ratio = 5.46, 1.14 – 26.32, p = 0.03) and post-operative blood transfusion (odds ratio = 7.33, 2.16 – 24.9, p = 0.01) were linked to dislocation.
Overall, there was a more than fourfold increase in both superficial and deep infection in those who received a post-operative transfusion compared with those who did not. Of the total of 2360 fractures, 28% received a post-operative transfusion (631/2216, data were available in 94% of cases). In the complicated cases 44% (63/144) of patients received a blood transfusion. In patients receiving a blood transfusion the deep and superficial infection rates were 2.85% (18/631) and 1.74% (11/631) respectively, compared with 0.69% (11/1585) and 0.38% (6/1585) in patients not receiving a transfusion. From the 63 patients in the complications group (n = 144) that required a transfusion post-operatively, 35 were associated with infection and haematoma.
In conclusion, in this study on the cost of complications and re-operation in proximal femoral fractures, infection is the single most important complication to avoid.
The authors suggest the relationship between a post-operative transfusion and infection supports an immunomodulatory effect of transfusion, i.e. a causative role of transfusion itself. I feel however, that as surgeons, we should at least consider an alternative explanation. In the absence of robust data substantiating an influence of patient related risk factors for infection, the relation between transfusion and infection can also signify a relationship between surgical technique and infection. Transfusion requirement is likely to indicate the presence of a large haematoma and devitalised tissue, both conducive to the development of a post-operative infection. The immunomodulation of blood transfusion is then only an adjunctive mechanism.
From this type of well-organised and well-executed large, labour intensive study, surgeons should actively search lessons that directly affect our care for this patient group. Looking at large numbers of patients using financial outcome measures may tend to obscure the underlying causes of complications. A further example for the importance of surgical technique is the relation between dislocation and blood transfusion. Post-operative blood transfusion had an odds ratio of 7.3 for prosthetic dislocation. There is no “dislocation-modulatory effect of blood transfusion”. Conversely, the need for blood transfusion indicates a large haematoma and muscle damage related to the operative approach. With proximal femoral fractures being the leading cause of trauma admission, surgeons should use surgical approaches and techniques producing to the least soft tissue damage, combined with strict aseptic technique as for elective total joint implants in younger patients.
Department of Orthopaedics, Haga Hospital. The Hague, The Netherlands