Hip surgery: to transfuse or not to transfuse?

1 March 2012

CLINICALLY APPRAISED TOPIC (CAT): What impact does restricting access to blood transfusions have on post-hip surgery patients?

Clinical bottom line:

Restricting transfusions to only the high-risk hip fracture patients with low haemoglobin levels (less then 80g per litre of blood or showing symptoms for anaemia) did not affect patients mortality levels or their ability to walk independently post-surgery.

Clinical scenario:

Your hospital has moved to restrict blood transfusions in all patients with the slogan “Why use two when one will do”. Working in orthopaedics, you are unconvinced because so many are older people with hip fractures who need to mobilise as soon as possible. Your area often transfuses two units when patients have haemoglobin of less than 100 g per litre (100g/L) of blood. You resolve to look at the evidence.


Among patients with hip fractures, do restrictive transfusion policies increase mortality and/or morbidity compared to less restrictive policies?

Search strategy:

PubMed - Clinical queries (therapy/narrow specific search): blood transfusion AND hip fracture.


Carson JL, Terrin ML, Noveck H, et al. Liberal or restrictive transfusion in high-risk patients after hip surgery. New Engl J Med 2011;365:2453-62.

Study summary:

A two-arm parallel-group randomised controlled trial conducted at 47 sites across the United States and Canada between July 2004 and February 2009. Inclusion criteria were age 50 years or more undergoing primary repair of hip fracture, evidence of cardiovascular disease or risk factors for cardiovascular disease (including history of hypertension, diabetes, hypercholesterolemia or current tobacco user), and had haemoglobin level of less than 100g/L blood within three days of the procedure. Exclusion criteria included patients unable to walk without assistance, patients declined transfusions, multiple trauma, pathologic hip fracture associated with cancer, heart attack within previous 30 days, previous participation in the trial with the other hip, had symptoms associated with anaemia, or were bleeding at randomisation.

Restrictive therapy (n=1009):

Patients permitted to receive 1 unit of packed red cells if signs and symptoms of anaemia (chest pain, congestive heart failure, tachycardia, or hypotension unresponsive to volume replacement) developed – or at discretion of surgeon if haemoglobin fell below equivalent of 80g/L blood. Patients with dementia received transfusion when haemoglobin level fell below threshold as may not have been able to report anaemia accurately.

Liberal therapy (n=1007): Patients received 1 unit of packed red cells and additional transfusions as required to maintain haemoglobin level over 100g/L blood.


The primary outcome measure was either death or self-reported inability to walk three metres without assistance at 60 days. The secondary outcome was death or self-reported inability to walk three metres without assistance at 30 days combined with in-patient heart attack, unstable angina, or all-cause death plus residential status, survival, functional outcomes and fatigue.

Study validity:

A randomised trial with the method of sequence generation and concealment reported. There was complete follow-up of 99.2 per cent of the study participants. ‘Intention-to-treat’ analysis was not reported but because of virtually complete follow-up of all participants, it was unlikely to have affected outcomes. The collection of post-discharge data was either via social security record (to confirm deaths) or ‘blinded’ phone calls where assessors did not know which therapy had been used. Self-reported walking was verified in 814 participants. The baseline groups were similar in make-up and received equal treatment. Overall, it was a high quality study.


Overall 14,438 patients were screened and 2016 (14%) met inclusion criteria and were randomised. The mean age of the study population was 81.6 years and 76% were female. 63% had prior cardiovascular disease. There were no significant differences between the groups in terms of the combined outcomes or mortality at 60 days or earlier (see table) or in terms of ability to walk three metres independently, functional scores, fatigue scores, or residential placement at 60 days or earlier. Patients in the restricted policy group had a median of 0 units transfused whereas patients in the liberal policy group had a median of 1 unit transfused.



Pragmatic trial so similar to real world clinical practice.

Using high-risk patients (those more susceptible to mortality or anaemia) would increase the likelihood of showing a clear effect. The presence of no clear benefit for liberal use of transfusions in high risk patients strongly suggests no benefit likely in low-risk patients.

Effect of different policies on hospital length of stay not investigated.

Reviewer: Dr Andrew Jull, RN PhD, Nurse Advisor – Quality, Auckland District Health Board & Associate Professor (School of Nursing).