Gregersen et al. conducted a single-centre (Danish) randomised controlled trial comparing a so-called “restrictive” (Hb threshold of 9.7 g/dL) and a liberal (Hb threshold of 11.3 g/dL) transfusion strategy in elderly (age >65 years) hip fracture patients (N = 284) in the postoperative period. The primary endpoint was physical recovery (assessed using 3 different scales), and secondary endpoints were 30- and 90-day mortality. Analysis was done according to type of residence (nursing home or sheltered housing).

The primary outcome did not differ between the 2 groups: patients’ physical abilities were equally impaired postoperatively (at day 10), without any difference in recovery between the 2 groups whatever the scale used. Mortality was not different in the ITT analysis (both at D30 and D90), but the per-protocol analysis found a higher 30-day mortality in the restrictive strategy group (HR 2.4 [1.1-5.5], P = 0.03). Subgroup analysis showed a higher 90-day mortality with restrictive strategy in nursing home patients (HR 1.9 [1.0-3.4], P = 0.04). The authors conclude that some hip fracture patients require more blood than is specified in current guidelines.

When one first reads it, this study may cast some doubt on the benefit of restrictive transfusion strategies; however, it has to be underscored that the “restrictive” strategy in this study is similar to the liberal strategy in many studies (in particular the FOCUS study; Carson JL et al. N Engl J Med 2011;365:245-62) and that a significant difference in mortality difference is observed in some post-hoc analyses but not in the whole study population. In addition, a vast majority of patients were transfused in both groups, whereas restrictive transfusion strategies usually reduce the proportion of transfused patients (in the FOCUS study, 59% vs. 3% of patients were not transfused). It is thus possible that both groups were harmed by transfusion (the cause of death was mainly pneumonia). Indeed, there is no clear explanation for this increase in mortality, since most of the patients had a postoperative Hb >10 g/dL and that the main causes of death did not seem to be linked to anaemia. These differences may have just been observed by chance.

Nonetheless, this study focuses on very frail patients, not assessed in the FOCUS study. It is thus also possible that these patients may benefit from much higher transfusion triggers. Unfortunately, this study provides little help in better defining how to transfused these patients, since the triggers used are much higher than those recommended in all recent guidelines. It should, however, be remembered that Hb alone may not be the only trigger for transfusion. This study underscores that 1) any transfusion should be decided on a patient basis (not only according to numbers!) and that 2) treating perioperative anaemia (with or without transfusion) is probably an important goal.

– Sigismond Lasocki