8

[United Blood Services (UBS), Blood Centers of the Pacific and Inland Northwest Blood Center]. Donations at the 15 UBS centres represent more than 80% of these donations and 6% of the US blood supply. Blood Systems serves more than 500 hospitals. We studied LOC reactions at 15 UBS blood centres from 1 January to 31 December 2007. Donor, donation and reaction data (including injury and OMC) acquisition is controlled by Standard Operating Procedure (SOP) and was previously described in detail [4,5]. Donor and donation information was obtained from the blood bank computer system (MAK ⁄ Progesa, Paris, France) by accessing the information through the Blood Systems Data Warehouse using Sagent Information Studio 5.5.1 (Troy, NY, USA). The donor records captured donor demographic information (age, gender, race ⁄ ethnicity, donation history), biometric characteristics (height, weight) and clinical measurements (pulse, blood pressure, temperature) prior to donation. Donation records included phlebotomy start and end time, collection status, collection site, donation type, etc. Additionally, each UBS blood centre provided required information on all reactions including classification (mild, moderate or severe), symptoms, time of reaction, location of reaction, injuries (lacerations, fractures, etc.) and the use of OMC (e.g. emergency medical services or emergency room visit) using a standardized adverse reaction reporting form. These forms captured detailed information on observed and reported reactions. The data recorded on the adverse reaction forms during the study period were entered by hand into a database. For this analysis, we include LOC reactions associated with a presentation to donate or an allogeneic whole blood donation attempt. We use the term VVS and fainting interchangeably to describe donor’s transient LOC of any duration before, during or after blood donation. VVS is associated with arterial hypotension and ⁄ or bradycardia. Clinically, these episodes are preceded by various symptoms including light-headedness, dizziness, nausea, sweating, pallor, unclear thinking and visual disturbances. LOC may have been directly observed by staff, reported by the donor, or reported by a third party and may have occurred on- or off-site. We recorded the start time of a reaction associated with LOC in relation to the end of phlebotomy (T = 0) using a clock at each donation site which is synchronized with the blood establishment computer system clock. We defined the following periods: Period 1 (Registration and eligibility assessment) – from donor presentation at the collection site through medical health screening; Period 2 (Phlebotomy) – from venipuncture until 4 min after phlebotomy end (presumably when the donor stands up). A donor is likely to be recumbent or semi-recumbent during Period 2. SOP requires that donor sits on the side of the bed for 1–3 min after the removal of the needle; and Period 3 (Postphlebotomy) – >4 min after phlebotomy end to the last event reported (in the dataset, 265 min after the needle was removed). A donor is likely to be sitting, walking or standing during this period. Period 3 is divided into two sub-periods: Period 3A – LOC occurring on-site, and Period 3B – LOC occurring off-site, outside the vicinity of the blood drive and without staff involvement for immediate donor management. The donor, donation and reaction data were merged to create a research database. Uneven age group categories were created because younger donors have a higher risk of fainting. Body mass index (BMI) and estimated blood volume (EBV) for each donor were calculated based on self-reported weight, height and gender using published formulas [8,9]. Some merged records had missing information for potentially relevant parameters in the analysis. We did not impute missing values based on other information. If the data were not recorded or available, the individual fainting reaction record did not contribute information to the statistical analysis for a given variable. The data variables with highest frequency of missing values were race (3Æ3%), height (1Æ3%); consequently, EBV and BMI were missing in 1Æ3% of records, and Predonation pulse (1Æ3%). Implausibly extreme values for any parameter were assumed to be data entry errors and were not included in the analysis. The Period 1 dataset contained 956 766 records of registered donors with the intention to donate – donating and deferred, autologous and allogeneic, apheresis and whole blood. When donors register, they do not indicate for what type donation procedure they have come. In addition, the treatment of donors is similar during Period 1 regardless of the intended donation type. For Periods 2 and 3, intended whole blood (complete and incomplete) donations were studied because

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19