Author reply

In ‘Recognising and managing dying patients in the acute hospital setting: can we do better?’, Gunasekaran et al. in the Internal Medicine Journal report a medical record audit of end-of-life care at a tertiary hospital. They reviewed the files of 201 deceased patients; 70% had been medical patients, 70% were living at home prior to admission and at least 60% had been hospitalised within the 12 months preceding their final admission. Notwithstanding the fact that progress medical notes do not reflect the totality of discussions with patients or the nuances of evolving clinical assessments, over 80% of the patients had a documented diagnosis of dying prior to their death. However, this documentation often occurred close to death, and 66% of patients received a medical intervention in their final 48 h of life. The authors state that consequent to ‘most healthcare professionals having limited formal training in end-oflife care’, there is a ‘tendency towards continued interventions and treatment in the last days of life’. While non-palliative interventions for patients in their terminal phase are to be avoided, in the absence of such a diagnosis, interventions given in the last 48 h of life may be appropriate, or inappropriate, depending on individual clinical scenarios. The data presented do not distinguish between these possibilities. At admission, many frail medical patients exhibit some ‘characteristics of dying’. Death may not be unexpected, but for the individual patient, the outcome is often uncertain. In an Australian study using a protocolised clinical assessment designed to identify inpatients at high risk of dying within 12 months, the positive predictive value of a ‘high-risk’ designation was only 38% (sensitivity 78%). A retrospective file audit of deceased patients may tempt the conclusion that death tends to be diagnosed too late, missing opportunities to de-escalate and refocus care. However, without a comparison group of patients who did not die (in whom medical interventions may have contributed to that outcome), the discriminative performance of such ‘characteristics of dying’ cannot be discerned. One-fifth of the patients audited had a rapid response review, and one-third of those had alterations to resuscitation plans consequent to that review. An interpretation might be that these patients experienced an avoidable escalation of care that should have been prevented with earlier, more conservative, resuscitation decisions. However, rapid response reviews are prompted by a deterioration in the patient’s condition (in this cohort of deceased patients, perhaps due to the onset of the terminal event). The rapid response review may therefore occur at a watershed moment at which re-evaluation of resuscitation status is appropriate. By only including patients who have died, this audit may have found an inflated rate of changes to resuscitation status during rapid response reviews. Gunasekaran et al.’s study provides a useful snapshot of current care and of clinician attitudes, but their data do not prove a remediable deficiency in the recognition of dying patients. As the authors acknowledge, recognition of dying is ‘inherently uncertain’.