Nononcological Advanced Chronic Disease and Palliative Needs: Survival Analysis

Background: Professionals who care for patients with advanced chronic nononcological disease need accurate prognostic tools. There are validated prognostic indices for nononcological pathology in the elds of internal medicine (PALIAR and PROFUND indices) and palliative care (PPI indices, PaP score, ECOG, PPS, KI). The objective of this study is to describe survival and analyzsed factors associated with mortality in advanced chronic nononcological disease patients in palliative care services in the Community of Madrid. Methods: Multilongitudinal observational study of a prospective cohort with a 6-month follow-up. Sociodemographic, clinical, analytical, service use, functionality, and prognostic indices were measured. Survival was analysed at 3 and 6 weeks and at 1, 2, 3, 4, 5, and 6 months through Kaplan–Meier curves. After the bivariate analysis, a Cox proportional-hazards multivariate regression analysis was performed. Results: 217 patients were included. The mean age (SD) was 78.8 (12.6), and 47.5% were women. Some 129 patients died. Mean survival (SD) at 6 months was 146.12 (130.14) days, median (IQR) survival 111.5 (17.50-254.50). All prognostic indices (PALIAR p<0.001, PROFUND p<0.005, PPI p<0.016, PaP Score p<0.001, ECOG p<0.002, PPS p<0.018, and KI p<0.016) predicted mortality at 6 months. The variables that explained survival at less than 3 months were PPS (HR (CI) 0.96 (0.95-0.98), p<0.000), leukocytes (HR (CI) 1.06 (1.02-1.10), p<0.000), delirium at the last admission (HR (CI) 1.79 (1.02-3.09) p<0.030), and ≥ 4 hospitalization in the last year (HR (CI) 1.82 (1.16-2.88) p<0.010). The variables that explained 6-month survival were PPS (HR (CI) 0.96 (0.94-0.97), p<0.000), leukocytes (HR (CI) 1.06 (1.03-1.09), p<0.000), and haemoglobin (HR (CI) 0.88 (0.82-0.97) p<0.005). Conclusions: Clinical and resource use variables were predictors of mortality in survivals shorter than 3 months, but not in survivals longer than 3 months. identied that the risk of mortality (95% CI) was 1.60 (1.24-2.06) (p<0.001) in women with anaemia and 2.29 (1.60-3.26) (p<0.001) in men. In both sexes, the risk of mortality increased with lower haemoglobin concentrations. The Leiden 85-plus prospective study (48), with a sample of 562 healthy people older than 85 years, identied that both prevalent and incident anaemia were associated with a higher risk of death, even after adjusting for sex, education level, and and predicted premature death in institutionalized elderly individuals (49). Multiple studies have shown an association between anaemia and increased mortality (47,48,50,51). Chaves et al. (52) identied, in a prospective study of 686 women than 65 years with moderate to severe disability, that haemoglobin below 11.0 g/dL was associated with higher mortality (HR [95% CI]: 1.2 [1.1-1.4]), while levels and g/dL a lower risk (HR [95% CI]: 0.6 [0.63-0.92]).

Survival was evaluated taking the date of registration in the clinical history as the start date and the date of the death or last follow-up as the end date. Mortality from any cause recorded in the clinical history was included.
Qualitative variables, expressed as absolute frequency (n) and percentage (%), and quantitative variables, expressed as mean and standard deviation (SD) or median and interquartile range (IQR), are described, depending on whether the variable follows a normal distribution. Qualitative variables were compared by the chi-squared test with Yates's correction or by Fisher's exact test. Quantitative variables were compared by Student's t test.
Survival was analysed at 3 and 6 weeks and at 1, 2, 3, 4, 5, and 6 months with Kaplan-Meier curves, as performed by Morita (22)(23)(24) at 6 months. To identify factors associated with mortality, a multivariate Cox proportional-hazards regression analysis was performed, with survival time as the dependent variable and as independent variables those that had p<0.20 and/or were considered of clinical relevance. The hazard ratio (HR) and its 95% con dence interval (CI) were calculated.
For the statistical treatment and graphical representation of the data, the statistical package SPSS v.26.0. and Microsoft O ce Excel 2007 were used. The con dentiality and privacy of the data were rigorously maintained, the data extraction was independent of the analysis, and an anonymized database was built.

Results
Characteristics of the study participants 217 patients were included. The mean age was 78.8 (SD: 12.6), and 47.5% were women. Tables 1, 2, and 3 show the sociodemographic characteristics, disease according to NHO criteria, clinical and analytical variables, service use, and functional and prognostic indices strati ed according to 6-month survival.  KI: Karnofsky Index.
The sociodemographic variables, disease characteristics according to NHO criteria, clinical variables, analytical variables, service use, and functional and prognostic indices strati ed according to survival at 3 and 6 weeks and 1, 2, 3, 4, and 5 months can be found in the Supplementary Materials 1.
The mean survival of the patients during the 6 months of follow-up was 146.12 days (SD 130.14), with a median of 111.5 (IQR 17.50-254.50) and a range of 0-420 days. In this period, 129 patients died, for a mortality of 60.19% in men and 58.77% in women. Figure 1 shows the Kaplan-Meier survival curves according to the prognostic index.
The means and medians of survival according to each prognostic index can be visualized in Supplementary Materials 2. Table 4 shows the prognostic factors of survival at 3 and 6 weeks and 1, 2, 3, 4, 5, and 6 months. PPS and haemoglobin were protective factors (HR<1), and leukocytes, 4 or more hospitalizations in the last year, and delirium in the last hospitalization were risk factors for mortality (HR>1). This study reveals factors associated with mortality at 6 months, such as the PPS scale and the number of leukocytes. The PPS scale has powerful predictive value for survival in patients receiving palliative oncological care (8) and has been validated in noncancer patient care (8-15). It is an excellent tool to measure the functional status and progression of the patient, and our ndings once again con rm it as an adequate index and predictor of mortality. In this study, for each 10-points increase in PPS, the probability of death decreased by 5% in all studied periods of survival: at 2 and 3 weeks and at 1, 2, 3, 4, 5, and 6 months.
The number of leukocytes was another predictor of long-term mortality, an effect that is well described for coronary disease (29-31) and cerebrovascular disease (32) but has also been associated with hypertension (33), glucose intolerance (34), and the risk of overall mortality (35).  After the 4th month, haemoglobin <10 g/dL was associated with mortality. For each 1 g/dL increase in haemoglobin, they were 11% more likely to be alive at 4, 5 and 6 months. In a Dutch cohort study of 1016 patients older than 85 years who lived in the community and were followed up for 10 years, Izaks GJ et al. (47) identi ed that the risk of mortality (95% CI) was 1.60 (1.24-2.06) (p<0.001) in women with anaemia and 2.29 (1.60-3.26) (p<0.001) in men. In both sexes, the risk of mortality increased with lower haemoglobin concentrations. The Leiden 85-plus prospective study (48), with a sample of 562 healthy people older than 85 years, identi ed that both prevalent and incident anaemia were associated with a higher risk of death, even after adjusting for sex, education level, and income, and predicted premature death in institutionalized elderly individuals (49). Multiple studies have shown an association between anaemia and increased mortality (47,48,50,51). Chaves et al. (52) identi ed, in a prospective study of 686 women older than 65 years with moderate to severe disability, that haemoglobin below 11.0 g/dL was associated with higher mortality (HR [95% CI]: 1.

Strengths and limitations
This was a prospective multicentre study in which the different palliative care units of the Community of Madrid, acute care units, mid-stay units, and both hospital and home support teams participated. The study re ects the evolution of the disease in the context of routine clinical practice, since it was an observational study in which the only selection criterion was the authorization of the patient and/or their family to participate in the study by granting their informed consent. Having been able to consult the follow-up data of all the patients at 6 months minimized our losses of data. The different scales of functionality and prognosis were exhaustively studied, which allowed us to study many variables.
One of the limitations of this study is that its design could introduce great variability, although to reduce this possible bias, all the researchers went through in-person training, normalizing the data collection as well as making sure the study was carried out in conditions of routine clinical practice, which favoured the involvement of professionals, reduced losses, and ensured good data collection.

Applicability to clinical practice and research
In advanced chronic disease, the type of cardiac, respiratory, digestive, nephrological, and neurological pathology was not associated with an increased risk of mortality. This nding leads us to think that, as in advanced oncological disease, the type of tumour is not a predictor of mortality, nor is the type of chronic pathology, there being a common end of life in both cases.
Few prognostic scales allow us to de ne and estimate survival in patients with nononcological advanced chronic disease. We have speci c prognostic indicators for pathologies, but the NHO criteria, without taking into account the speci c pathology, were published in 1995 (53). They are still the most commonly used guidelines to determine the prognosis in advanced chronic nononcological diseases due to their simplicity. These medical guidelines were developed based on expert opinions, and as demonstrated by Fox et al. (54), up to 70% of patients whose survival estimate was less than 6 months exceeded this period. Some studies even show that these criteria cannot be applied when life expectancy is very short. (55).
In our study, all the evaluated prognostic indices could classify the population into groups with signi cantly different survival at 3 and 6 weeks and at 1, 2, 3, 4, 5, and 6 months.

Conclusions
These ndings support the recommendation of the European Association of Palliative Care to use a prognostic tool and avoid relying only on the clinical impression of the professional, with the aims of improving decisions for patients with advanced disease (56), communicating more realistic expectations, offering treatments tailored to the needs of each patient, avoiding futile therapies, and thereby optimizing the use of health resources (57)(58)(59) Informed consent to participate was obtained from all subjects and/or their legal guardian(s).

Consent for publication
Not applicable Availability of data and materials The datasets generated and/or analysed during the current study are not publicly available because contains sensitive clinical information about patients, so there are ethical and legal restrictions to sharing the data set. The datasets used and analyzed during the current study are available from the author on reasonable request: cmiguel@salud.madrid.org.

Competing interests
The authors declare that they have no competing interests with CMT are in charge of table and gure design. The rst draft was initially written by CMT and CDMS with discussion with IDCG and JCGM. All authors contributed to data interpretation, critically reviewed the rst draft, approved the nal version, and agreed to be accountable for the work.