The incidence of cancer is increasing, and the disease is responsible for 25% of overall mortality in Spain.1 Improvements in active treatments and supportive care collaborate in reducing mortality associated to cancer. In this regard, patients with neoplasms represent a population that is currently susceptible to admission to the Intensive Care Unit (ICU) as part of cancer treatment, for the treatment of intercurrent medical and surgical processes, or as a key tool for the management of toxicity caused by the different oncological therapies.2

Such improvement in survival makes it necessary to reconsider the role of intensive care medicine in this scenario. Different scientific associations are interested in understanding this role; as a result, contacts have begun to be established among scientific groups interested in this subject.3 In Spain, a collaboration agreement was signed between the Spanish Society of Intensive and Critical Care Medicine and Coronary Units (Sociedad Española de Medicina Intensiva, Crítica y Unidades Coronarias [SEMICYUC]) and the Spanish Society of Medical Oncology (Sociedad Española de Oncología Médica [SEOM]), contemplating the creation of a registry to generate information on the epidemiology and factors related to mortality among oncological patients requiring admission to the ICU.4

In recent years, studies have been made of survival in the ICU among different highly selected groups of oncological patients5–9 (involving mainly hematological malignancies), and recommendations have been made in relation to the different therapeutic strategies.2,10 Studies have also been made on the evolution of cancer patients subjected to certain therapeutic procedures inherent to intensive care, such as mechanical ventilation,11,12 high-flow oxygen therapy,13 renal replacement techniques14 or even extracorporeal oxygenation.15

However, few general epidemiological data are available on the impact of cancer patient admission to the ICU. In multicenter studies, the proportion of patients with cancer admitted to the ICU ranges between 13%–20%.16–21 Nevertheless, this population is extremely heterogeneous — a fact that may have an impact upon the care they require. It is clear that a close multiprofessional and multidisciplinary approach is the only possible way to improve the prognosis of critical patients with cancer.4,22–26

The present study was carried out to know the epidemiology, reasons for admission, the therapeutic resources used, and mortality among patients with malignant disease according to the causes giving rise to their admission to the ICU.

An observational study was carried out based on the ENVIN registry (which we refer to as the “ENVIN database”). All the patients admitted between 1 April and 30 June 2018 were included. Expansion of the data on the cancer patients was made by combining a new database referred to as the “ONCOENVIN database”. Three common variables (age, date of admission and gender) served to generate a common identifier allowing data linkage between the two databases.

In the year 2018, the full ENVIN period (April–June) had recorded 27,514 patients admitted to some of the 219 Units that participated that year in the registry. A total of 4247 patients (15.4%) had a history of cancer. Of these, 60% (2547 patients) contributed full data of the ONCOENVIN database. Since participation was voluntary and open, we considered that those Units which in ONCOENVIN had entered fewer than 80% of the cases that had been declared in ENVIN should be excluded. A total of 92 Units completed over 80% of the cases, and their data were therefore considered valid for the epidemiological study. One hundred and twenty-two patients (5.7%) of these Units intending to participate did not have the oncological information. On the other hand, 238 patients were excluded due to defective identification of the date of diagnosis of the malignancy or an age of under 18 years. Finally, we analyzed a total of 1786 patients with cancer disease and a complete registry (42% of the patients with cancer) versus 10,010 patients without cancer belonging to the Units intending to participate (43% of the patients without cancer). Thus, in this selected population, the cancer patients represented 15.1% of the total of patients. Fig. 1 shows the patient screening process for the epidemiological study.

Figure 1.

Patient screening diagram.

(0.34MB).

The percentage of cancer patients with respect to the total admissions in the 92 selected Units was highly variable, with a median of 17 patients (p25–p75: 10–24.5) per Unit, but with a proportion of between 1%–48% of cancer patients with respect to the total per Unit. This variation reflects the diversity of the characteristics of the participating ICUs, which included coronary Units (obviously with a very low percentage of cancer patients) and small postsurgery Units with very high percentages of cancer patients. Overall, 87% of the Units were considered to be polyvalent, while 3.3% were exclusively postsurgery Units, and the same proportion consisted of trauma Units. The geographical distribution was uniform, with representation of all the Spanish regions (Autonomous Communities), and a total of 38 provinces.

In comparison with the patients without cancer, those subjects with a recent history of cancer were generally older, came from hospital wards, with a predominance of surgical antecedents, greater severity upon admission, a shorter stay, and greater in-ICU mortality (12.3% versus 8.5%; p < 0.001). These data are shown in Table 1. The Kaplan–Meier curves reflecting the survival time differences are shown in Fig. 2 (log-rank test, p < 0.001).

Figure 2.

Kaplan–Meier survival curves between patients with or without a history of cancer.

(0.1MB).

In order to assess the type of patient admitted to the ICU, and taking into account the prevalence of surgical cases among the cancer patients, a distribution was established according to whether the patients suffered a medical disease condition (n = 585) or had been admitted after elective surgery (n = 834) or after urgent surgery (n = 273). Coronary and trauma cases were excluded because of their scant representation (5.2% overall) and lack of influence in relation to the objective of our study. In general, the patients with a background medical disease condition were more seriously ill, with a greater percentage of comorbidities and of infectious complications, and with significant higher mortality (27%) than the patients subjected to elective surgery (2%) or urgent surgery (13,2%). The rest of the differential characteristics are shown in Table 2. Obviously, the hematological patients were more often admitted due to medical causes (89.6%) than to causes related to elective surgery (3.3%) or urgent surgery (7.1%). In contrast, the patients with solid organ tumors were more often admitted due to overall surgical causes (72.1%) than to medical causes (27.9%).

Another epidemiological approach involves taking into account the reason for admission in relation to the neoplastic disease itself. Three groups were established: no relation to cancer (NRC) (n = 400); postsurgery related to cancer (PRC) (n = 1006); and medical complications related to cancer (MCRC) (n = 379) — with the inclusion of patients as described in the methodology. This latter group consisted of younger individuals, in more serious condition, with predominantly background medical disease, and with a greater use of devices such as mechanical ventilation (46.2%) or extrarenal filtration (9.2%). Table 3 describes the characteristics of these patients, including the location of the tumor, the year of cancer diagnosis, and the antineoplastic treatment received. Of note is the absence of treatment up until the time of admission in most of the PRC patients (49.5%). In contrast, having received treatments different from those of the above groups was more frequent among the NRC patients (41.5%). Neutropenia during admission, chemotherapy in the ICU or tumor lysis syndrome proved more frequent in MCRC patients. A relevant finding in this latter group is the proportion of patients with some form of limitation of therapeutic effort (17.2%) versus the other groups (NRC: 5.8% and PRC: 1.5%; p < 0.001). Probable aspergillosis was diagnosed in 2.9% of the MCRC patients.

The duration of ICU stay was longer in the MCRC patients than in those belonging to the other groups, and in-ICU mortality was also significantly greater (31.4%) than in the NRC (12.8%) or PRC patients (5%) (p < 0.001). In terms of overall survival, the MCRC patients were clearly differentiated from the other two groups (log-rank test, p < 0.001), as can be seen from the Kaplan–Meier curves shown in Fig. 3.

Figure 3.

Kaplan–Meier survival curves between patients according to reason for admission and relation to cancer.

(0.1MB).

The present study provides the first description in Spain of the epidemiology of patients with solid organ tumors or hematological malignancies requiring admission to the ICU. Fifteen percent of all patients admitted to the ICU have cancer, though there is marked variability in this percentage depending on the type of Unit involved — a circumstance that reflects the heterogeneity of the Spanish ICUs. The conduction of this multicenter study, based on the Units participating in the ENVIN registry, has afforded a very precise image of the case-mix of the cancer patients admitted to the ICU, due both to the large number of participating Units and to their widespread geographical distribution. None of the participating Units were monographic cancer patient ICUs. The multicenter and voluntary nature of the registry implied that not all the Units participating in the ENVIN registry completed the data on cancer patients, and thus that not all the possible patients have been recorded. The utilization of the arbitrary criterion of securing the complete registry of at least 80% of the patients with cancer reduced the number of participating ICUs, but in our opinion it affords a quite accurate idea of the epidemiological reality of cancer patients requiring admission to intensive care in our country.

Although it was not the objective of our study to analyze factors related to mortality (an aspect that will be examined by another article), the mortality rate in the ICU among cancer patients was significantly higher than in patients without cancer (12.3% versus 8.5%; p < 0.001). However, due to the heterogeneity of the cancer patients, this observation says little of the factors influencing such mortality. The Kaplan–Meier curves offer a good idea of mortality differences, but do not represent more than a point in time (ICU stay) in the course of the neoplastic disease. They consequently must be correlated to their true value in relation to the number of live patients discharged from the ICU. No in-hospital mortality data were available, as the reporting of such information is not mandatory in the ENVIN registry.

Among the cancer patients, those admitted to the ICU for postsurgery care represented 62% of the total. This figure is the same as that recorded in the European series derived from the SOAP trial,16 and similar to the percentages obtained in the multicenter Dutch18 and Brazilian studies19 (56% and 64%, respectively). In this group of patients, and in the same way as in the global population of critical patients,29 a distinction is made between those who are admitted to the ICU in the postoperative period of elective surgery (less severe cases, with shorter stays and lesser mortality) and those who require admission following urgent surgery (more severe cases, with stays and mortality rates intermediate between those of the elective surgery patients and patients with background medical disease). Some authors have only studied the admission of non-elective cases, which allows for a certain unification of the case-mix,17,20 though the distinction between medical cases (40.7%) and urgent surgical cases (59.3%) remains considerable — with a poorer prognosis among the former.

The patients admitted due to medical causes (representing 34.6% of all the patients with cancer) constituted a very heterogeneous group, with differences between those presenting solid organ tumors (72.1%) and those with hematological neoplasms (27.9%). The proportion of hematological patients admitted to the ICU in Spain is greater than that recorded in other series, where the figures range between 14.6%–17%.16,19

The mortality rate among the medical cases with cancer (27.5%) was clearly higher than in the other two groups (2% and 13.2% for elective and urgent surgery, respectively), as has also been reported in other epidemiological series.19 The in-ICU mortality rate in the multicenter Dutch study17 was 30.4% for medical cases and 16.2% for surgical cases. On the other hand, it is notorious that the proportion of patients in which the limitation of therapeutic effort was decided proved clearly higher (14%) in the group of medical patients versus the other patient groups — thus identifying a population of more complex patients with more serious acute disease. It would be interesting to examine the influence of the staging of cancer disease in this decision.

It is of interest to describe the reasons for admission to the ICU and their relation to cancer. This relationship was strong among the surgical patients (94.2% of the elective surgery patients and 76.6% of the urgent surgery cases), and less manifest among the medical cases — with a high percentage of patients being admitted to the ICU due to reasons not directly related to cancer (38%), and which represent intercurrent processes that are seen in both oncological patients and in the non-oncological population.

Respiratory failure (38.5%) and sepsis/septic shock (36.4%) were the most frequent causes of admission to the ICU in the medical complications related to cancer (MCRC) group. This observation is not surprising, since the use of invasive mechanical ventilation is more frequent in these patients, and vasopressor drug administration (not recorded in this study) is probably also more common, as suggested by the literature.16,19,21 The global patients admitted due to reasons related to cancer presented a mortality rate of 31.4%, which demonstrates the importance of organ failure in the clinical course of these patients, which constitute a subgroup of special individuals due to their high mortality and use of resources — generating debate regarding the possible futility of their admission to the ICU.

With regard to the cancer treatment received before admission to the ICU, we recorded a notoriously high percentage of symptomatic treatments (17.8%) or other treatments (41.5%) among those patients admitted to the ICU due to reasons unrelated to cancer. These patients possibly correspond to two extremes: those who receive palliative treatment and those free of active disease who are not receiving specific treatments, and who are admitted to the ICU for other quite different reasons.

The present study has a number of limitations. Assessing cancer stage and evaluating its influence in deciding or not deciding treatments is a complex matter. Although an attempt has been made to encompass all the therapeutic possibilities as a continuity from the patient pending the start of treatment to the treatment of metastatic disease, this aspect is not easy to define at the point in time represented by admission to the ICU. Perhaps more simple definitions for the type of cancer (local or metastatic for solid organ tumors; high or low malignancy for hematological neoplasms) would have allowed for better grouping of the patients and easier analysis,19 but there would always remain a percentage of patients in which the evolutive stage of the disease cannot be determined. It also must be taken into account that what we documented was in-ICU mortality, not in-hospital mortality or mortality at 90 days, which would have been more correct in relation to the course of the neoplastic disease process.

The percentage of patients with cancer and nosocomial infections, including aspergillosis, was low in all the analyzed groups, though we did not calculate the rates related to the use of devices that favor infection (as registered in the ENVIN); no conclusions therefore can be drawn in this regard. In any case, nosocomial infections in these patients do not appear to constitute a problem much different from that seen in the general population without malignant disease.

The choice of 5 years as the limit in defining a personal history of cancer is arbitrary. It is possible that some of these patients were “healed” of their cancer at the time of admission to the ICU. However, we adopted this criterion because it is the definition used in the ENVIN, and due to the difficulties that may be encountered in deciding whether a cancer has healed or not in patients admitted to the ICU for other reasons. It is also complicated to determine whether the reasons for admission to the ICU are related to the neoplasm or not. This seems clear in the extreme scenarios (postsurgery or septic shock in a patient with profound post-chemotherapy neutropenia), but it might not be so clear in intermediate patients. On the other hand, the multiple comparisons made identified statistically significant differences as a consequence of the sample size. Some of these differences are of scant clinical significance, however, and would require more detailed post hoc analysis.

In conclusion, the present multicenter study describes the epidemiology of patients with a recent history of cancer or who are admitted for the treatment of cancer or its complications. There is great variability in the percentage of cancer patients in the different ICUs in Spain, with a predominance of patients in the postoperative period of elective surgery (46.7%). The global mortality rate among the cancer patients was 12.3%, though a more seriously ill patient population with greater mortality is identified. Specifically, the mortality rate among the patients with background medical disease was 27.5%, while in those subjects admitted due to medical complications related to cancer, the mortality rate was even higher (31.4%). More specific studies are needed on the mortality-related factors in these populations.

PMOA, FAL and RGC are coordinators of the ENVIN registry. JGM, FGV and CBZ form part of the SEMICYUC-SEOM collaborative working group. RDN, CRS and IAA are principal investigators at their hospitals and contributed a larger number of cases. The study was designed by the SEMICYUC-SEOM working group. The manuscript was written by PMOA and corrected by the rest of the authors, who agree to its contents.

The ENVIN-HELICS registry was partially funded during the year 2018 by the Spanish Medicines Agency (Agencia Española del Medicamento y Productos Sanitarios [AEMPS]).

The authors declare that they have no conflicts of interest in relation to the present manuscript.