The principal findings of our study were significantly higher need for blood transfusion postoperatively and higher first-year mortality in patients using LdAA before major hip surgery. The mortality risk in the preoperative LdAA users compared to non-users was increased nearly five times during the first month after hip fracture and was thereafter threefold higher up to one year after fracture.
Prophylactic treatment with LdAA for cardiovascular disease, both primary and secondary, is common in older people. An Irish study in 2006 showed that 42% of hip fracture patients were regular users of acetylsalicylic acid before the fracture . Similarly, 46% of the patients in our study used LdAA, which can be compared to 17% of the whole population 80 years or older in Sweden during 2006 (according to the Swedish Prescription Drug Register).
At the time of our study national guidelines regarding prophylactic use of acetylsalicylic acid for cardiovascular disease recommended dosages between 75 mg to 320 mg but recent international recommendations have proposed a dose of 75 mg for most indications . This can be of importance as most side effects of acetylsalicylic acid are dose-related and differences in effects on bleeding time and fibrinogenic activity have been shown with doses of 75 mg and 160 mg . However, we did not find any significant difference in 1-year mortality between patients treated with ≤75 mg LdAA compared to those treated with ≥150 mg. Since acetylsalicylic acid in the blood is 50 to 80 percent protein-bound, metabolized for the most part in the liver and excreted mainly by the kidneys, the amount of active drug depends mainly on renal function and concomitant drug use . Also, sex-related difference in excretion of acetylsalicylic acid has been shown, with women more likely to have higher drug concentrations in the blood . Although the effect of preoperative use of LdAA on 1-year mortality in our study was higher in women than in men the difference was not statistically significant. This question may need to be investigated in a larger sample. Additionally, due to age-related decline in kidney function, higher concentrations of free salicylate can lead to increased risk of bleeding [15, 16].
Preoperative discontinuation of LdAA in hip fracture patients is not an option as surgery should be done without unnecessary delay . In doses of 75 to 160 mg irreversible blockade of cyclooxygenase lasts through the platelets' life span of 8 to 10 days. Because cardiac disease is common in hip fracture patients, withdrawal of LdAA for a few days may also increase the risk of cardiovascular complications. Rebound phenomena connected to promotion of prothrombogenic factors is also a risk factor for thromboembolic complications which in itself makes withdrawal questionable .
In our original randomized trial we did not include patients with cervical fractures treated with internal fixation with hook pins, a less traumatic surgical procedure than hemiarthroplasty and usually not requiring blood transfusion. This group of patients usually includes both younger healthier patients and very old frail persons with limited walking ability or with an expected short life span. Because the three surgical procedures used in this study are major procedures that may differ in the amount of related blood loss we chose type of surgery rather than type of fracture in our analysis of bleeding and transfusion outcomes. However, using type of fracture in the analysis gave essentially similar results both in the original study and in this analysis (data not shown).
Both surgical and medical management of patients with hip fracture have improved during the last decade, but mortality has not decreased as might have been expected . Despite that several studies have shown high mortality following hip fracture few studies have involved interventions to decrease it. The most frequently proposed solution has been to focus on fall prevention and not on interventions to diminish postoperative mortality. Doubts about the effectiveness of fall prevention on long-term mortality have been raised in a study comparing all-cause mortality in hip fracture patients with that of a general population . Fields that need to be further studied include bleeding complications, blood transfusions, nutrition, congestive heart failure, hydration, chest infections, pharmaceutical treatment, cognitive function, and frailty.
Early mortality (within 120 days) following hip fracture has been shown to be related to patient age and sex, type of fracture, pre-fracture residence, mobility and ASA score [4, 5, 19]. A possible factor behind the higher mortality in the LdAA treated patients is the increased need for postoperative red blood cell transfusion. Patients treated with LdAA have increased bleeding diathesis and, in our study, they also had somewhat higher APTT and INR values, the reason for which is unclear because these values are not expected to be affected by LdAA. Although the differences are statistically significant, the numeric differences are small and their clinical relevance is uncertain. Other studies have shown up to 20 percent increase in blood loss in acetylsalicylic acid treated patients, but no significant increase in mortality [10, 20]. Although some studies have shown doubled amount of bleeding complications in LdAA treated patients, the clinical significance of these complications has been disputed. Blood transfusions have been shown to both cause and to worsen congestive heart failure through circulatory overload and by causing fever, low blood pressure and pulmonary symptoms [10, 21–23]. Studies showing significant increase in mortality in patients receiving transfusions have, however, mainly been carried out on patients in critical care units and not specifically on orthopaedic patients [23, 24]. The 100 g/L hemoglobin level used as threshold for transfusion in our study may be considered to be relatively high. However, some previous studies have used a similar threshold although others have used a threshold as low as 85 g/L . The indication for red cell transfusions should ideally be identified individually for each patient instead of a set pretransfusion threshold and be based on premorbidity and relevant clinical risk.
One reason for the higher mortality among the preoperative LdAA users may be the higher risk of early postoperative complications often leading to longer time before start of mobilization and rehabilitation . Although postoperative complications were few there was a significantly higher incidence of thromboembolic events among LdAA users than among non-users. The number of patients afflicted with these complications was small but considering that LdAA treated patients had higher APTT and INR values and increased bleeding diathesis this seemingly paradoxical finding, merits further study. Even so it accentuates the need for a risk-benefit analysis when considering the use of antiplatelet therapy in frail elderly patients.
Previous studies of hip fracture patients have also shown an effect of preoperative LdAA on postoperative blood transfusion. Manning et al. studied 89 patients with femoral neck fractures treated with hemiarthroplasty or dynamic hip screw fixation with regard to the effect of preoperative aspirin on blood loss and transfusion requirements . Although the authors showed no significant effect on perioperative blood loss, the 24 LdAA treated patients included in the analysis were significantly more likely to receive postoperative blood transfusion than the 52 patients without LdAA, a finding similar to our study, which also similarly did not show differences in intraoperative blood loss.
Some of the strengths of our study are that it included a representative sample of patients above 50 years with hip fracture, excluding only those with non-displaced cervical fracture treated with pin fixation (as those patients seldom require blood transfusion), that the majority of patients were operated within 24 hours after admission, and that we followed all participants prospectively for up to one year after surgery.
The main limitation of our study is that its primary objective was to evaluate the efficacy of a compression bandage in reducing need for blood transfusion and the mortality analysis is thus secondary. In the initial trial the proportion of LdAA treated patients in the two groups (compression vs non-compression) was similar and the compression bandage did not show any effects on postoperative blood transfusions or hemoglobin. We therefore assumed that the intervention with compression bandage did not influence the results of the present report. This is supported by the finding that adding postoperative use of compression bandage as a covariate to the analyses did not change the results. Another possible limitation is the large number of patients not assessed for eligibility for participation in the original trial by the orthopedic surgeon at the emergency room. Possible reasons for not assessing patients are that the surgeon did not know about the trial, had assessed the patients to be ineligible but failed to document this or was unwilling to recruit patients because of the extra work involved. However, these patients had similar characteristics to the patients included in the present report and since we have described the exact inclusion criteria for the trial we believe the patients in this study are representative of patients fulfilling these criteria.
Increased first-year mortality was seen both in patients using LdAA before fracture and in those with baseline cardiovascular and/or cerebrovascular comorbidity, with the latter having larger effect. The exact causes of the increased mortality among LdAA users are not known but this study does not exclude the possibility of side effects from acetylsalisylic acid as a contributing factor.
In a US study of 8930 patients with hip fracture it was shown that pulmonary complications were as common as cardiac complications in causing early mortality . In our study only data on all-cause mortality were analyzed since we lacked information on cause of death. Finally, more detailed information on other medications used by the patients before surgery could have facilitated a more comprehensive assessment of the patients' risk of bleeding.