Evaluation of clinical coding data to determine causes of critical bleeding in patients receiving massive transfusion: a bi‐national, multicentre, cross‐sectional study

    loading  Checking for direct PDF access through Ovid


Blood transfusions are one of the most commonly performed hospital procedures, and in Australia, the costs associated with blood supply are rapidly approaching $1 billion each year (National Blood Authority, 2011a). According to international data, massive transfusions (MT) account for 1·3% of all patient transfusion episodes and 10% of total blood products used (Rose et al., 2009). Apart from the financial burden and organisational challenges that critical bleeding patients pose on treating clinical, laboratory and transfusion services, recently published evidence‐based MT clinical guidelines reveal significant evidence gaps (National Blood Authority, 2011b). Clinical trials are inherently difficult in this setting due to the heterogeneity of the patients and bleeding indications, protocol compliance in the setting of acute haemorrhage and ethical issues. Moreover, outside the trauma setting, few local or international studies have examined transfusion practice and the clinical profile of MT patients (Rose et al., 2009; Sinha & Roxby, 2011; Halmin et al., 2016), and none have been conducted at a national level. To address the need for data on transfusion practice and patient outcomes in Australia and New Zealand (ANZ), the bi‐national Massive Transfusion Registry (MTR) was established, prospectively collecting comprehensive data on all eligible patients at each participating institution from time of entry to the MTR. As individual patient chart review was likely to be prohibitively costly due to the anticipated number of patients, we sought to use routinely collected data, including administrative and clinical data systems, to determine the cause of critical bleeding for the registry.
Previous local and international population‐based studies investigating clinical indications for red cell (RBC) transfusion have chiefly used International Classification of Diseases primary codes (Titlestad et al., 2001) or the diagnosis‐related groups (DRGs) (Syrjälä et al., 2001; Allden et al., 2011) as a source of clinical information to link with transfusion data. Good to excellent coding quality has been demonstrated for the principal procedure and principal diagnosis code (Campbell et al.2001; Quan et al., 2004; Henderson et al., 2006). However, as expected, these individual codes alone do not always reflect the indication for transfusion (Llewelyn et al., 2009). For example, studies examining indication for RBC transfusion using DRG data identified that patients with high transfusion requirements are often assigned the tracheostomy DRG (Whyte & Brook, 1998; Syrjälä et al., 2001). Recent studies using algorithms with combinations of diagnosis codes and the principal procedure code to determine reasons for transfusion are promising (Llewelyn et al., 2009; Wells et al., 2009). However, their ability to identify causes of critical bleeding in patients who receive MT is unknown, and previous transfusion validation studies have not been performed in ANZ.
Extending the methods used by Llewelyn et al (2009), we developed a novel algorithm based on diagnostic codes classified according to the Australian modification of the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD‐10‐AM) and Australian Classification of Health Interventions (ACHI). The aim of this study was to examine the accuracy and reliability of this algorithm for determining causes of critical bleeding in patients who receive MT using expert chart review as the reference standard.
    loading  Loading Related Articles