Skip to main content
. 2015 Apr 21;2015(4):CD009699. doi: 10.1002/14651858.CD009699.pub2

Carson 2011.

Methods Randomised controlled trial
Number of centres: 47 in Canada and USA and 1 centre in UK (see Palmer 1998)
Dates enrolled: 19 July 2004 to 28 February 2009
Follow‐up: 60 days
Participants 2016 participants, of which 24% were men
Inclusion criteria: people aged ≥ 50 years who were undergoing primary surgical repair of a hip fracture and who had clinical evidence of a history of ischaemic heart disease, ECG evidence of previous myocardial infarction, a history or presence of congestive heart failure or peripheral vascular disease, or a history of stroke or transient ischaemic attack) or risk factors for cardiovascular disease (see below) were eligible if they had a haemoglobin concentration of < 10 g/dL within 3 days after surgery
After December 2005, people with any of the following cardiovascular criteria were eligible regardless of haemoglobin concentration:

  1. history of or treatment for hypertension

  2. diabetes mellitus

  3. hypercholesterolaemia (cholesterol level > 200 mg/dL or a low density lipoprotein cholesterol level > 130 mg/dL)

  4. current tobacco use

  5. creatinine level > 2.0 mg/dL


Exclusion criteria:

  1. inability to walk without human assistance before hip fracture

  2. declining blood transfusions

  3. having multiple trauma (defined as having had or planning to undergo surgery for non‐hip‐related traumatic injury)

  4. having a pathological hip fracture associated with cancer

  5. history of clinically recognised acute myocardial infarction within 30 days before randomisation

  6. having previously participated in the trial with a contralateral hip fracture

  7. having symptoms associated with anaemia (e.g. ischaemic chest pain)

  8. actively bleeding at the time of potential randomisation


Mean (SD) age in the liberal transfusion threshold: 81.8 (8.8) years
Mean (SD) age in the restrictive transfusion threshold: 81.5 (9.0) years
Number of men per intervention: 250/239
Interventions Liberal transfusion threshold vs. restrictive transfusion threshold. Participants were randomised when their postoperative haemoglobin fell to < 10 g/dL within the first 3 days post operation
Liberal transfusion threshold: "Patients ..... received 1 unit of packed red cells and additional blood as needed to maintain a haemoglobin level of 10g or more per deciliter. An assessment of the haemoglobin level after transfusion was required and an additional unit of blood was transfused if the patients blood was below 10 g per deciliter" (page 2454 of manuscript)
Restrictive transfusion threshold: "Patients ... were permitted to receive transfusions if symptoms or signs of anaemia developed or at the discretion of their physicians if the haemoglobin level fell below 8 g per deciliter" (page 2454 of manuscript)
Number of people randomised per intervention: 1007/1009
Number of people included in the analysis of the primary outcome: 995/1000
Outcomes Primary outcomes:
Mortality
Mobility and functional recovery
Postoperative morbidity
Secondary outcomes:
quality of life
Length of stay in hospital
Notes Email communication with the main author on 7 March 2013 was successful in identifying the discharge protocols for the US and Canadian participants and the time points for the reported postoperative morbidity.
This trial changed its inclusion criteria in 2005 from including people with a history of cardiovascular disease to people with a history of cardiovascular disease OR risk factors as outlined.
An ancillary study (the FOCUS Cognitive Ancillary Study), with enrolment in the last few months of the trial (April 2008 to February 2009) of 139 participants, that collected delirium outcomes was reported in Gruber‐Baldini 2013. Our reading of the articles for the ancillary study was that the population of this study was a subgroup of that of FOCUS
Risk of bias
Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Low risk Study used an "automated telephone randomisation system" (page 2454 of manuscript) with randomisation schedules for each site using randomly ordered block sizes of 2, 4, 6 and 8
Allocation concealment (selection bias) Low risk Randomisation was undertaken at a central location "staff members at the data co‐ordinating centre" (page 2454 of manuscript)
Blinding of participants (performance and detection bias) High risk "Patients, clinical‐site staff members & clinicians were aware of study group assignment after randomisation" (page 2454 of manuscript)
Given that the participants themselves were involved in assessing outcomes (quality of life, functional mobility), knowledge of treatment allocation may influence outcome measurement, hence an assignment of high risk of bias
Blinding of personnel (performance bias) High risk "Patients, clinical‐site staff members & clinicians were aware of study group assignment after randomisation" (page 2454 of manuscript). The blinding of clinicians would have been difficult anyway as the clinicians themselves determined whether a participant met the requirements for a red blood cell transfusion. 147 protocol violations were reported. In addition, given that the clinicians themselves were involved in assessing outcomes (postoperative morbidity), knowledge of treatment allocation may have influenced outcome measurement
Blinding of outcome assessment (detection bias) 
 All outcomes Low risk "Study investigators who classified cardiovascular events and those who did follow‐up telephone assessments (to assess outcomes [quality of life, mobility & functional recovery] after hospital discharge) were unaware of study group assignments" (page 2455 of manuscript)
Subjective outcomes were measured following discharge (at 30 and 60 days after randomisation) and by people blinded to treatment assignment, hence an assignment of low risk of bias
Incomplete outcome data (attrition bias) 
 All outcomes Unclear risk A small number of participants were not included in most analyses: mortality and mobility at 30 (n = 21) and 60 days (n = 17) and postoperative morbidity (n = 3).
The manuscript reported the number of withdrawals and losses to follow‐up, but did not state when these happened.
ECG was used to measure myocardial infarction incidence: ECG results (after randomisation) were incomplete in 13.4% (n = 135) in the liberal transfusion threshold group and 12.9% (n = 130) in the restrictive transfusion threshold group of participants. Of note also is that there was no blood sample for troponin testing in 17.9% (n=180) in the liberal transfusion threshold group and in 17.3% (n=175) in the restrictive transfusion threshold group of participants.
A large number of people were not included in the quality of life assessments at 30 days (45.6% in the liberal transfusion threshold group vs. 45.9% in the restrictive transfusion threshold group missing) and at 60 days (54% vs. 52% missing). At each time point, and for each assessment tool, the percentage of participants not included in the analysis was similar between the 2 transfusion threshold groups. The manuscript documented the reasons for participants not being included in these analyses as either inability to perform the physical assessment or an incomplete data set per assessment
Selective reporting (reporting bias) Low risk Study protocol was available (on clinical trials.gov) and all of the study's pre‐specified primary, secondary and composite outcomes are reported in the pre‐specified way
Other bias High risk There was a statistical difference in the number of major protocol violations post randomisation between the 2 transfusion threshold groups: 9% in the liberal transfusion threshold and 5.6% in the restrictive transfusion threshold group (P value = 0.003)
The difference between the 2 transfusion threshold groups in the number of transfusions received before randomisation was not significant (P value = 0.07) but may be of interest in the overall context of this review. Full details are available in Table 3