Anticancer therapy within the last 30 days of life: results of an audit and re-audit cycle from an Australian regional cancer centre

There have been significant recent developments in the treatment of cancer with many new treatments demonstrating clinical evidence for improved survival and quality of life (QOL). Systemic anticancer therapy (SACT) now includes cytotoxic chemotherapy, endocrine or hormonal agents, targeted or biologic agents and immune checkpoint inhibitors. Non-chemotherapy treatments are often associated with simpler routes of administration, less but not negligible adverse effect profiles and the potential of profound and durable clinical responses. This has made the decision making process for commencing, continuing and ceasing SACT more complex and requires a careful consideration of key factors, specifically disease biology, patient and family expectations, and clinician biases.

Earle et al. [1] have proposed several indicators for the assessment of quality of care near the end of life including the rate of chemotherapy administration; emergency department (ED) presentation, hospitalisation and intensive care unit (ICU) admission; and lack of or late referral to palliative care and hospice services. Over the last few decades, there is a trend towards more aggressive care with US registry studies finding an increase in patients receiving chemotherapy within 14 days of death; and increased rates of ED presentation, hospitalisation and ICU admission in the last month of life [2]. Use of chemotherapy is associated with higher rates of cardiopulmonary resuscitation and mechanical ventilation, late hospice referral, death in ICU, and death in a non-preferred place [3]. In the current context of immune checkpoint inhibitors, use near the end of life is associated with poorer performance status, lower hospice enrolment and higher rates of death in hospital [4].

The rationale for SACT with palliative intent is primarily to improve or maintain quality of life. Despite this objective, the quality of life of patients as assessed by psychological and physical distress in the final week of life has been found to not improve in patients with moderate or poor performance status who received chemotherapy and in fact worsens in patients with good or excellent performance status who received chemotherapy [5].

There is increasing evidence of the benefit of early involvement of palliative care for patients with cancer. The seminal randomised control study by Temel and colleagues [6] demonstrated early palliative care consultation for patients with non small cell lung cancer (NSCLC) improved QOL, mood and depressive symptoms, and survival by more than two months. A secondary analysis of this study demonstrated that palliative care did not affect the number of chemotherapy regimens administered, but that chemotherapy near the end of life was reduced and hospice enrolment was higher [7]. Both the American Society of Clinical Oncology (ASCO) and the European Society of Medical Oncology (ESMO) have made recommendations in clinical practice guidelines for the concurrent use of SACT and early involvement of palliative care services for patients with advanced cancer [8, 9].

The first large scale report of mortality within 30 days of chemotherapy emanated from the National Confidential Enquiry into Patient Outcome and Death (NCEPOD) conducted in the United Kingdom [10]. The reported rate of mortality within 30 days of SACT was 2% and has become the historical benchmark. Subsequently, several centres have published data related to SACT within the last 30 days of life [4, 11‐28]. A selection of studies are summarised in Table 1 with focus on recent publications and the Australasian context. Comparisons between these studies are difficult for several reasons; studies differed with regard to: the included and excluded tumour types, treatment with curative and palliative intent, and treatment modality. Only three reports included immune checkpoint inhibitors. In addition, the studies reported different outcome measures most commonly the number of deaths within 30 days of treatment as a proportion of all patients who received treatment and, less commonly, as the number of deaths within 30 days of treatment as a proportion of all deaths.

Subsequent to NCEPOD, Christie Cancer Centre in the United Kingdom implemented its key recommendation to review all deaths within 30 days of SACT at a morbidity and mortality meeting and reassess progress through an audit process. Over a four year period, this practice did not reduce the rate of deaths within 30 days of SACT and had a minor but statistically insignificant reduction in the rate of treatment related deaths [22]. In contrast, Wilson et al. reported two audits performed at Auckland Hospital six years apart [19]. Mortality within 30 days of treatment with chemotherapy fell minimally with rates of 2.8% in 2009 and 2.2% in 2015. They proposed a series of clinical interventions that have informed this improvement implementation plan.

The aim of the study was to identify the rates of SACT within the last 30 days of life at the institution in order to compare with published benchmarks. We examined the use of the different types of SACT to observe any changing trends in practice given the development of new therapies, especially in the contemporary paradigm of immune checkpoint inhibitors. The audit also assessed other quality of care and aggressiveness of care parameters. Results from the initial audit informed the implementation of a service improvement plan which was then followed by a re-audit to assess any effect this improvement plan had on clinical practice. This study is novel in the emerging era of immune checkpoint inhibitors and contributes to our understanding of quality use of SACT, aggressiveness of care near the end of life and institution based interventions to improve the quality of patient care.

Clinical audit is an essential part of clinical governance used to assess current performance and an important tool for practice improvement [29]. Audits have been found to have a small but potential important impact on professional practice [30]. This study reviewed 440 patient deaths and examined for patterns in SACT and other parameters of aggressiveness of care. A multifaceted quality improvement implementation plan was implemented and a re-audit of a further 75 patient deaths was conducted to assess the effect of this intervention. Components of the intervention have previously been proposed by other authors [10, 19]. The intervention was simple and can easily be replicated at other centres. The mechanism of the intervention is multifaceted. The initial education session raises awareness of this issue and reports on current performance. Regular reviews of all cases at morbidity and mortality meetings maintains awareness and allows repeated feedback to clinicians. Mandated repeat assessment of performance status and discussion of borderline cases at treatment commencement provide a decision point which can break up treatment inertia. Finally, simplifying referral processes to palliative care and hospice services supports patients, families and clinicians in symptom management, maximising the use of SACT and transitioning to the end of life period.

Overall, the patient population was as expected of a regional cancer centre with a predominance of solid tumour types representing the common cancer diagnoses and a broad spread of patient performance status. The rate of SACT within the last 30 days of life was 27% in the initial audit period and remained stable at 25% during the re-audit period. Review of the literature, as summarised in Table 1, did not find a published report to consider all treatment modalities, namely cytotoxic chemotherapy, endocrine and hormonal agents, targeted and biologic agents and immunotherapy. When comparing this figure to other published reports, it is important to note that this statistic is the number of deaths within the last 30 days of life as a proportion of all deaths of patients managed at the Goulburn Valley Health oncology department. A similar statistic in the Australian context was reported by Wein et al. at a rate of 26% but only included patients treated with palliative intent and did not include immunotherapy [21]. Other authors have reported treatment rates as a proportion of patient deaths between 4 and 38% [14, 16, 18, 21‐23, 28].

The more commonly reported statistic is the number of deaths within 30 days of SACT as a proportion of all patients who received SACT. We calculated a comparable statistic where the number of deaths of patients who received chemotherapy within the last 30 days of life expressed as a percentage of all patients who received chemotherapy was initially 8% and fell substantially to 2% during the re-audit period. This compares well with reports from other centres and is in fact the lowest reported rate in Australasian region [11, 13, 19, 24‐27].

This study is one of only a few to include immunotherapy in assessing mortality within 30 days of treatment. A recent study from New Zealand reported a rate of SACT within the last 30 days of life of 5.2% [11]. This included chemotherapy, targeted therapies and immunotherapy, but excluded endocrine / hormonal treatments. It should be noted that this was measured as a proportion of all patients who received SACT. Gilsch et al. conducted a retrospective review of 157 deceased patients treated with immune checkpoint inhibitors and reported that 27% received a dose within the last 30 days of life [4]. This is substantially lower than our rates of 47% in the audit period and 89% in the re-audit period.

SACT within the last 30 days of life with cytotoxic chemotherapy occurred in 27% and in higher proportions in non-chemotherapy treatments, specifically endocrine and hormonal agents 47%, targeted and biologic agents 43%, and immune checkpoint inhibitors 47%. Interestingly, this rate rose to 89% during the re-audit period with regard to immune checkpoint inhibitors whereas rates for endocrine / hormonal agents and targeted/biologic agents remained stable and the rate of chemotherapy use near the end of life fell to 13% (Fig. 1). This should be interpreted with caution due to the small patient numbers and short re-audit period. Possible explanations for these trends include the increasing availability and number of indications for immune checkpoint inhibitors over the recent period. Furthermore, non-chemotherapy treatments have a more tolerable side effect profile and may be more accepted by patients and clinicians when treatment decisions are being made. These factors may contribute to a shift in treatment modalities from chemotherapy towards immune checkpoint inhibitors. Also, the composition of tumour types between the audit and re-audit periods was different (Table 6) and may affect the types of treatments used.

When the other parameters of aggressiveness of care are examined, the re-audit period was notable for an increase in palliative care referrals (78% vs 65%) and decrease in late palliative care referrals (41% vs 58%). This did not seem to affect the other indicators of aggressiveness of care. Similar rates of more than one ED presentation, more than one hospitalisation and ICU admission were seen in the audit and re-audit period. The rate of hospital admission for more than 14 days rose from 14 to 22% after the intervention. This is demonstrated in Fig. 2. The association of treatment within the last 30 days of life and increased rates of more than one hospitalised was observed before and after the intervention but was not maintained with regard to more than one ED presentation or ICU admission.

Limitations to this project should be noted. Data has been collected in a retrospective manner. Data was collected from patient medical records which relies on complete and accurate documentation. Furthermore, patients in regional areas often have shared care between different centres and therefore a number of outcome events may not be captured in this data collection. The re-audit period was relatively shorter than the initial audit period and the observed trends in practice may attenuate over a longer period of time. The re-audit period was considered to have started immediately after the audit period which resulted in a short period of time before clinicians were fully exposed to the improvement implementation plan. The educational meeting and palliative care contacts were enacted within a week of the conclusion of the audit period. However, there may have been a learning curve period as clinicians gained repeated exposure and feedback from the regular case and mortality meetings. Nonetheless, these factors would likely have contributed to an underestimation of the effect of the improvement implementation plan.

This study provides a contemporaneous benchmark for SACT and other parameters of aggressiveness of care within the last 30 days of life in an Australian regional setting. Importantly, the changing treatment paradigm with the increasing use of immune checkpoint inhibitors and other targeted agents is considered. It also establishes the components of a quality improvement implementation plan and demonstrates its impact on use of SACT and palliative care referral practices. Further research is required into the factors which affect the treatment decision making process in order to ensure quality of care.