Do death certificates accurately record deaths due to bloodstream infection?

To the Editor: Sepsis and bloodstream infection (BSI) are associated with significant morbidity and mortality although they are amenable to targeted intervention in the hospital setting. The Australian Bureau of Statistics (ABS) reported 1463 deaths due to septicaemia in Australia in 20121 — a number that is likely an underestimate.2

These statistics are based on reporting of cause of death on death certificates by medical practitioners. However, previous studies in Australia have found inaccuracies in reporting, with major revision of cause of death required in half of cases referred for coronial review.3 The need for education of reporting doctors and supervision by senior clinicians has been highlighted.4

A review of BSI at our institution during 2012 identified 571 patient episodes of BSI. We were able to analyse death certificates for 65 of 73 patients who died within 30 days of a clinically significant blood culture. The mortality from BSI reported in the literature is high, with a 30-day mortality rate of up to 23% for BSI with any organism, and up to 43.9% for BSI involving methicillin-resistant Staphylococcus aureus.5,6We thus assumed that BSI was likely to have contributed to mortality in deaths where there was evidence of BSI within 30 days of death. Our observed mortality rate of 13% was lower than reported figures, suggesting that BSI was likely contributory in these deaths.

In our series, BSI was reported as an underlying cause of death in 15 of 65 cases (23%), while only 39 death certificates (60%) identified the presence of BSI in any field. Seventeen death certificates (26%) did not reflect the presence of BSI at all. The BSI events not recorded as the underlying cause of death may not have been captured in ABS statistics as contributing to cause of death. Trends in the incidence of and mortality from BSI may therefore not be detected. This is a particular concern in an era of increasing global antibiotic resistance, when rising mortality due to BSI may signal increasing antibiotic resistance in the community.

Several systemic deficiencies likely contribute to underreporting. Events leading to death, but reported as following from the recorded underlying cause of death, are not reflected in cause of death data collected by the current system, which focuses on reporting the underlying cause of death. This approach fails to identify potentially preventable or reversible intermediary events, such as BSI, during complex and advanced medical care. Inaccuracies may be compounded by lack of training and experience and poor supervision of junior doctors who complete death certificates.7 As a result, it is likely that national death statistics are heavily biased towards single events such as road traffic accidents, or “upstream” diagnoses, like malignancy. We agree that increased and structured supervision and training of medical students and junior doctors in this area appears to be required. However, it may also be necessary to revise the reporting process itself to identify potentially preventable intermediary causes of death, that may then be more effectively targeted with research funding, awareness campaigns and clinical care bundles.