How do treatment aims in the last phase of life relate to hospitalizations and hospital mortality? A mortality follow-back study of Dutch patients with five types of cancer

At the end of life, the majority of people prefer to be cared for and die at home [1, 2]. An international study in Belgium, the Netherlands, Italy, and Spain found that 71–90% of the cancer patients whose preferred place of death was known by their GP wanted to die at home, against only 1–6% in hospital [3]. Accordingly, a low percentage of hospitalizations and a high percentage of patients who die at home are being regarded as indicators of high-quality palliative care [4, 5]. To achieve this, the availability of community-based palliative care is very important in order to support patients’ wishes and care needs. The Netherlands is one of the countries that have a generalist palliative care model [6, 7], meaning that for patients at home, GPs are central to the coordination and provision of care at the end of life, together with a community nurse, when needed. The GP remains the treating physician in the palliative and dying phase and can consult a specialist or specialized palliative team in complex cases [8].

One of the most important aspects of GP palliative care is the timely recognition of palliative care needs. There is evidence that having a palliative treatment goal, i.e., a treatment aimed at maximizing a patient’s comfort, leads to favorable outcomes. Abarshi et al. (2010) found that patients whose care in the last phase of life was primarily palliative were less likely to be transferred between care settings and less likely to die in hospital [9]. Many studies point out that cancer patients receive more palliative care and in an earlier stage than non-cancer patients [10‐13]. Cancer patients are also more involved in end-of-life discussions and advance care planning than, for example, patients with dementia or organ failure [14]. These differences are often attributed to cancer patients’ more predictable illness trajectories [10, 14‐16].

However, even for the patient group with the most predictable illness trajectory, GPs seem to face difficulties combining disease management and palliative approaches. A study by Claessen et al. (2013) revealed that the main impetus for Dutch GPs to initiate a palliative approach with cancer patients is information from the medical specialist stating that curative treatments are no longer available, and that not all GPs see added value in discussing palliative care with the patient when cure is still an option [17]. Moreover, the cancer patient population is not a homogeneous one, as different types of cancer vary widely in incidence and possibilities for treatment and cure. As a result, 5-year survival rates vary considerably: from 88 and 87% for prostate and breast cancer patients to 17 and even only 7% for patients with lung and pancreatic cancer [18]. Thus, although the typical illness trajectories for progressive chronic illnesses show that patients with cancer experience rapid decline in the terminal phase [19, 20], the speed and length of the period of functional decline might differ by cancer type as well as by age and gender [21]. These factors could all have consequences for the timely initiation of palliative care.

Until now, no studies have examined aspects of GP end-of-life care by cancer type. For the five most common cancer types causing death in the Netherlands—lung, colorectal, breast, prostate, and pancreatic cancer, together accounting for more than 58% of Dutch cancer decedents [22]—this study aims to (1) describe and compare the aims of treatment during the last week of life, 2 to 4 weeks before death, and 2 to 3 months before death, (2) describe and compare the percentage of patients who were admitted to hospital within the last month of life and died in hospital, and (3) examine which patient characteristics and treatment aims relate to hospital admissions and hospital mortality.

This study revealed large variation with regard to treatment aims, hospital admissions, and hospital mortality (and their interrelatedness) among Dutch patients who died from lung, colorectal, breast, prostate, or pancreatic cancer.

With regard to variation in treatment aims, our study showed that patients with prostate cancer most often and patients with pancreatic cancer least often had a palliative treatment aim in their last month of life. In addition, breast cancer patients more often than all other cancer patients had a curative or life-prolonging treatment aim until the end of life, often combined with a palliative aim as well. It is likely that this variation in treatment aims is related to prognosis and survival. For example, it has been investigated that more than half of pancreatic cancer patients already present with metastatic disease at the time of diagnosis, resulting in a median survival of only 9 to 11 weeks [25] and a 1-year survival rate of only 24% [18]. Although a diagnosis of pancreatic cancer in itself should thus ideally trigger physicians to start palliative care in practice, some patients with this type of cancer may deteriorate and die so abruptly that it is difficult for both the patient and the physician to find the right moment. In contrast to pancreatic cancer, the 5-year survival rate for breast cancer is remarkably high (87%) [18], which may contribute to physicians’ and patients’ tendency to continue with curative or life-prolonging treatments as long as possible. However, we found no indications for such a tendency among patients with prostate cancer, although they have a similar high 5-year survival rate (88%) [18]. Apparently, not only differences in prognosis may lead to variation in treatment aims between cancer types, but also availability of curative and life-prolonging treatment options, experienced symptom burden, and the patient’s age are likely to play a role here. With regard to the latter, the prostate cancer patients in our study were averagely 6.5 years older than those with breast cancer. As earlier studies found that older people are more often treated with a palliative-centered goal [12, 26, 27], our finding that prostate cancer patients most often had a palliative treatment goal may reflect a palliative philosophy in the care of the older old in the Netherlands. However, further qualitative research is recommended to be able to better interpret the variation in treatment aims we found.

Also, with regard to the percentage of patients who were admitted to hospital and died there, considerable variation was found between cancer types, with prostate cancer patients having the lowest rate for hospital admissions in the last month of life (18.5%) as well as for dying in hospital (3.1%) and lung cancer patients having the highest rates for both outcome variables (41.8% resp. 22.6%). These findings are in line with data from Statistics Netherlands [28] and with results from an international comparative study that found a higher percentage of hospital admissions in the last month of life as well as a higher percentage of hospital mortality among patients with lung cancer as compared to the total cancer population [29]. A study on reasons for hospitalization at the end of life found that acute episodes played a role in 70% of hospitalizations among cancer patients and that respiratory and digestive problems were most common in this patient group [30]. Acute episodes such as shortness of breath, pain, and an altered level of consciousness are indeed common reasons for hospitalization in patients with lung cancer [31] and an important explanation for the fact that lung cancer patients visit the emergency room more often than patients with other types of cancer [32]. Keeping in mind that the frequency of certain acute complications differs by cancer type and that hospitalizations can be beneficial to a patient’s quality of life as well (e.g., when receiving a blood transfusion), a study by De Korte-Verhoef et al. (2014) showed that, according to GPs, nearly a quarter of last hospitalizations in the final 3 months of life of cancer patients could have been avoided in retrospect [33]. They most frequently mentioned “proactive communication with the patient, including talking about withholding curative and life-prolonging treatments and diagnostics” as strategy to avoid hospitalizations. GPs however also stated that it is difficult to find the right moment to discuss such issues, especially with patients who unexpectedly deteriorate quickly or with those who are in a slow deterioration process [33].

This resonates with our finding that having a palliative treatment aim as only important aim and being older were significantly associated with less hospital admissions in the last month of life, and that having a palliative treatment aim as only important aim, having prostate cancer, and dying in a more recent year were associated with less hospital mortality. This is in line with earlier studies on health care transitions in the last phase of life showing that having a palliative treatment goal reduces the risk of being transferred between care settings and of dying in hospital [9], and that curative or life-prolonging treatments were most frequently mentioned as reasons for terminal hospital admissions [34]. The decreasing trend in hospital mortality is also reported elsewhere. In two consecutive measurements, Cohen et al. reported a decreasing trend in hospital mortality in several European countries, among which is the Netherlands. For example, the proportion of Dutch cancer patients dying in hospital decreased from 31% in 2003 to 26% in 2008 [35, 36]. Other research additionally pointed to prostate cancer as one of the diseases with the strongest decline in hospital mortality over the years [37]. Several factors may explain this decreasing trend; it could be a reflection of more community services being available such as the expansion of home care [38], but also of policy changes with regard to hospital admissions and discharge—e.g., when terminally ill patients are nowadays discharged more early to other external palliative care services [37]. All the same, these results and our findings even more emphasize the need for early initiation of GP palliative care to support patients from all cancer types to receive care and die at their preferred location, which presumably is, according to what has been found in earlier studies [1‐3], “at home.”

There is large variation between different types of cancer with regard to treatment aims, hospital admissions, and hospital mortality. As having a palliative treatment aim is associated with less hospital admissions and less hospital mortality, the results highlight the need for early recognition of the palliative phase and timely initiation of GP palliative care to support patients from all cancer types to stay at the place they prefer as long as possible.