Nausea at the start of specialized palliative care and change in nausea after the first weeks of palliative care were associated with cancer site, gender, and type of palliative care service—a nationwide study

Nausea is a common and distressful symptom among patients in palliative care, but little is known about possible socio-demographic and clinical patient characteristics associated with nausea at the start of palliative care and change after initiation of palliative care. The aim of this study was to investigate whether patient characteristics were associated with nausea at the start of palliative care and with change in nausea during the first weeks of palliative care, respectively. Data was obtained from the nationwide Danish Palliative Care Database. The study included adult cancer patients who were admitted to palliative care and died between June 2016 and December 2020 and reported nausea level at the start of palliative care and possibly 1–4 weeks later. The associations between patient characteristics and nausea at the start of palliative care and change in nausea during palliative care, respectively, were studied using multiple regression analyses. Nausea level was reported at the start of palliative care by 23,751 patients of whom 8037 also reported 1–4 weeks later. Higher nausea levels were found for women, patients with stomach or ovarian cancer, and inpatients at the start of palliative care. In multivariate analyses, cancer site was the variable most strongly associated with nausea change; the smallest nausea reductions were seen for myelomatosis and no reduction was seen for stomach cancer. This study identified subgroups with the highest initial nausea level and those with the least nausea reduction after 1–4 weeks of palliative care. These latter findings should be considered in the initial treatment plan.


Introduction
In palliative care, the primary goal is to relieve symptoms in patients with life threatening illness [1]. In Denmark, palliative care teams in hospitals and hospices provide specialized palliative care and the staff is mainly nurses and physicians [2,3] (in the rest of the paper, specialized palliative care will be referred to as "palliative care"). A common symptom among patients in palliative care is nausea and patients often describe this as being distressful [4] and problematic [5]. The etiology of nausea is often multifactual [4,6]. A study among patients referred and/or admitted to palliative care found that 66% of the patients experienced nausea and/or vomiting [7] and studies assessing nausea separately have found that 6 to 38% experienced nausea [8][9][10][11][12]. Mean levels of nausea/vomiting of 12-36 on a 100 point scale (100 worst score) have been reported [7,[13][14][15][16] and studies reporting nausea separately found mean levels of 1-2 on a 10 point scale (10 worst score) [16][17][18].
Some research has been conducted on the association between patient characteristics and nausea at the start of palliative care. An increased risk of nausea (or higher levels of nausea) at the start of palliative care has been found among women [6,7,12,13], younger patients in some [6,12,13] but not all cases [7,9], and among patients with certain diagnoses such as gynecological cancers [13] and cancers in the digestive system [10,13,19] although the evidence is inconsistent [7]. Furthermore, nausea has in some studies been associated with survival time for patients admitted to palliative care [20][21][22]. Knowledge on the association between nausea at the start of palliative care and other patient characteristics, e.g., social characteristics, is sparse.
Previous studies investigating how nausea changes following initiation of palliative care have in most cases shown a reduction [23,24] but the evidence is inconsistent [25]. The evidence on how patient characteristics such as gender, age, and cancer site are associated with change in nausea among patients in palliative care is sparse [26][27][28].
Large, nationwide studies are needed to gain more robust knowledge on which patient groups have an increased risk of nausea at the start of palliative care. This may help clinicians working outside palliative care to identify patients in high risk of nausea who might benefit from a referral to palliative care. Knowledge from large studies is also needed to identify patient groups for whom it is most difficult to obtain a nausea reduction during palliative care because it may help clinicians working in palliative care to select the most efficient treatment strategies from the beginning or to identify the patient groups requiring the most intensive follow-up and for whom changes to their nausea treatment should be considered early.
The aims of this study were to answer the following research questions: (1) are any patient characteristics associated with nausea level at the start of palliative care? and (2) how does the level of nausea change from the start of palliative care until 1-4 weeks later, and (3) is the change associated with any patient characteristics?

Setting
This study includes cancer patients admitted to any Danish specialized palliative care service. In 2020, there were 43 palliative care services in Denmark (19 hospices and 24 palliative care teams) [29]. Approximately two-thirds of patients who die from cancer in Denmark are referred to palliative care and approximately half of those who die from cancer are admitted [29]. The admittance is often late (on average 39 days before death) [29].

Data
Data in this study was obtained from the Danish Palliative Care Database, which includes data on all patients referred to palliative care in Denmark since January 2010 [30]. At the time of this study, validated data was available for patients who had died in 2020 or earlier. The database includes sociodemographic and clinical data as well as patient reported symptoms. Since 2010, symptom data were collected at the start of palliative care and since June 14, 2016, also 1-4 weeks later. Symptoms were measured by The European Organisation for Research and Treatment of Cancer Quality of Life Questionaire-Core-15-Palliative Care questionnaire (EORTC QLQ-C15-PAL).

Population
Adult cancer patients reporting their level of nausea at the start of palliative care (i.e., from 3 days before the day palliative care is started) who were admitted to palliative care and died between June 14, 2016, and December 31, 2020, were included in the analyses of nausea at the start of palliative care (henceforth referred to as the "baseline population"). Patients in the baseline population who also reported their nausea level at 1-4 weeks later were included in the analyses of change in nausea ("follow-up population").
If patients were admitted to more than one palliative care service, only data from the first admittance was included.

Symptom assessment
The EORTC QLQ-C15-PAL is a shortened version of EORTC QLQ-C30 adapted to patients in palliative care [31]. The questionnaire assesses the level of nine symptoms and functions (including nausea) and overall quality of life. In the EORTC QLQ-C30 questionnaire, two items assess nausea and vomiting in the past week on a four-point scale ("not at all," "a little," "quite a bit," and "very much"). Using the EORTC QLQ-C30 scoring manual [31,32], the two items can be converted into a single 0-100-point nausea/vomiting score. When the EORTC QLQ-C30 questionnaire was shortened and adapted to patients in palliative care, the vomiting question was omitted and the response to the nausea item was still converted into a 100-point nausea/vomiting scale (where 100 is the worst score) using an algorithm maximizing compatibility with the original QLQ-C30 score [31,32]. Throughout the paper, the term "nausea score" is used although the nausea item is converted into a nausea/vomiting score.

Data analysis
Descriptive statistics were used to present characteristics for patients included and not included in the baseline population and follow-up population, respectively (see patient characteristics in Table 1). Chi-square test, t-test, and Log-Rank test were used to test if the included and not included patients differed.
Average nausea level at the start of palliative care was computed overall and according to patient characteristics for the baseline population (this will be called "analysis 1"). Multiple regression analyses were performed to test if differences in nausea level according to patient characteristics were significant when controlling mutually for the effect of other patient characteristics.
For the follow-up population, mean change in nausea during the first 1-4 weeks of palliative care was computed overall and according to patient characteristics (analysis 2). This provides the observed "raw" change in nausea in each group.
The potential for obtaining a nausea reduction is higher for patients with high nausea levels at the start [33]; therefore, a "fair" comparison of changes should take nausea level at the start of palliative care into account. This was done in analysis 3 by analyzing the association between each of the eight patient characteristics and nausea change when controlling for start nausea level using multiple linear regression. Finally, a "full" multiple linear regression model including all eight patient characteristics and nausea at the start of palliative care simultaneously was performed (analysis 4). Thus, this analysis assessed the association between each patient characteristic and change in nausea while mutually controlling for several possible confounders. If analysis 4 resulted in non-trivial changes in the associations between a patient characteristic and nausea change compared to analysis 3, this indicated that one or more of the other characteristics were important confounders. If the associations changed only trivially, it was only important to control for nausea level at the start of palliative care. Backward stepwise selection was performed to identify the patient characteristics significantly associated with change in nausea using a significance level of 0.05.
When interpreting the results from the regression analyses, a mean difference > 0 reflects less nausea reduction after 1-4 weeks of palliative care compared to the reference group whereas a mean difference < 0 reflects more nausea reduction after 1-4 weeks of palliative care compared to the reference group.
The analyses were performed using SAS statistical software version 9.4.

Patient characteristics
Of the 34,247 adult Danish cancer patients admitted to palliative care in 2016-2020 who died in the same period, 23,751 patients (69%) reported their level of nausea at the start of palliative care and were included in the baseline population ( Fig. 1). The included patients were on average 70.2 years old and half were women (Table 1). Of the 34,247 patients, 28,724 survived at least 1 week after admittance to palliative care and of these patients, 8037 (28%) reported their level of nausea at the start of palliative care and 1-4 weeks later (Fig. 1). These patients were included in the follow-up population and were on average 70.1 years old and half were women ( Table 2).
Patients included in the baseline population differed significantly from patients who were not included on several patient characteristics (Table 1). Significant differences for several patient characteristics were also found for patients included and non-included in the follow-up population ( Table 2). In both cases, most differences between included and non-included patients were, however, small, but survival time from the start of palliative care was longer for included patients (Tables 1 and 2).

Nausea level at the start of palliative care
Among the 23,751 patients who reported their nausea level at the start of palliative care, 52.8% experienced some level of nausea and the average nausea score was 23.3 on a 100point scale ( Table 3, analysis 1). Only small differences in the level of nausea at the start of palliative care were seen according to most patient characteristics. There were, however, large differences according to cancer site. The highest average nausea levels were seen for gynecological cancers, i.e., cervical, ovarian, and endometrial cancers (mean 28.9-36.0); stomach cancer (mean 34.8); and pancreatic cancer (mean 30.7) ( Table 3, analysis 1). A higher average nausea level was found for women compared to men (27.0 vs. 19.7), and for inpatients in palliative care services and non-palliative care hospital departments (28.5 and 28.8) compared to patients in palliative outpatient clinics and patients receiving palliative home care (20.4 and 21.6). In multiple regression analyses where the associations between each patient characteristics and nausea were mutually controlled for the effect of other patient characteristics, cancer site was the variable most strongly associated with nausea (data not shown). Significant, but small, associations were also found between the other patient characteristics (except for children status) and nausea level (data not shown).

Change in nausea level the first weeks of palliative care
Among the 8037 patients who reported their level of nausea at the start of palliative care and 1-4 weeks later, the average time between the first and the second nausea reporting was 15.5 days (data not shown). After 1-4 weeks of palliative care, nausea was on average reduced by 7.5 points, but differences were seen according to patient characteristics ( Table 3, analysis 2). Except for cancer site, the average nausea change varied less than five points on a 100-point scale according to patient characteristics, e.g., the gender difference was only 3.5 points (women: − 9.2 points vs. men: − 5.7). The nausea reduction varied between 1.8 and 18.2 points across cancer site. Reductions of at least 10.0 points were seen for cervical, endometrial, small intestine, and ovarian cancer, respectively. Often larger reductions were seen for cancer sites with the highest nausea levels at the start of palliative care (Table 3, analysis 2).
When associations between patient characteristics and nausea change were controlled for nausea level at the start of palliative care, the associations were small (< 3 points), except for cancer site (Table 3, analysis 3). For most cancer sites, the nausea reduction was similar in size to the overall  reduction of 7.5 (± 3 points). The largest reductions were seen for patients with cervical cancer, lymphatic malignancies, or cancer in the brain and CNS, respectively (Table 3, analysis 3). The smallest reductions were seen for patients with myelomatosis or ovarian cancer. For stomach cancer patients, no reduction was seen since the average change was almost zero, i.e., 0.9 points on the 100-point scale (mean difference in nausea change 8.4 − 7.5 = 0.9). The associations between each patient characteristic and nausea changed only a little when the associations were controlled not only for nausea level at the start of palliative care but also for other patient characteristics (Table 3, analysis 4). Most associations remained significant, but who referred the patient to palliative care and children status were no longer significantly associated with nausea change (Table 3, analysis 4).

Main findings
In this study, approximately half of the patients experienced nausea at the start of palliative care and the average level of nausea was 23.7 on a 100-point scale. Nausea was on average reduced 7.5 points during the first weeks after palliative care was initiated. Several patient characteristics were significantly associated with nausea at the start of palliative care and nausea reduction after the first weeks of palliative care, respectively. Cancer site was the characteristic with the strongest association to nausea at the start of palliative care and nausea change, respectively.

What this study adds
Any level of nausea reported at the EORTC QLQ-C15-PAL is considered clinically relevant to patients [34,35] and 52.8% experienced at least "a little" nausea in this study. The average nausea level found in this study of 23.7 points lies within the range found in previous studies where the average nausea levels among patients referred to palliative care ranged 12-36 on a 100-point-scale [7,[13][14][15][16]. It has been found that to be clinically relevant, a nausea reduction should be at least 3-10 points on the 100-point EORTC nausea scale [36,37], which means that the nausea reduction of 7.5 in this study may be clinically relevant.
It is important that clinicians working outside palliative care are aware of patients who could benefit from a referral to palliative care because of high nausea levels and this study found women and patients with certain a P-value < 0.05 for the difference between patients who were and were not included in the follow-up population on different covariates. P-values were obtained from chi-square test (categorical variables), t-test (age), or log-rank test (survival time)    cancers (pancreatic cancer, ovarian cancer, or stomach cancer) had particularly high levels at referral-maybe there is a need for additional attention towards detecting patients with nausea in these subgroups. The higher nausea level among women compared to men at the start of palliative care in this study was also found in previous studies among cancer patients referred to palliative care [7,12,13,38]. This may be because women are at higher risk of chemotherapy-related nausea and vomiting [39,40] or of nausea in general compared to men [41] or because anti-emetic drugs are less effective in women [42]. In this study, patients with pancreatic cancer, ovarian cancer, and stomach cancer experienced the highest nausea levels at the start of palliative care. Previous studies among patients referred to palliative care also found the highest nausea levels among gynecological, esophageal, and gastro-intestinal cancers [10,19], although one study found no significant association between cancer site and nausea [7]. The position of the tumors in gastrointestinal cancer can cause disruptions in the digestive system, e.g., bowel slowdown or constipation [43,44], which might cause nausea. Pancreatic cancer patients may have digestive problems because of surgery or imbalance in digestive hormones and enzymes [45] which may cause nausea. This study also found a higher risk of nausea among inpatients compared to patients seen in outpatient clinics or at home which can probably be explained by that inpatients are often closer to death and thus experiencing higher symptom levels [29,46]. The largest reductions in nausea were often seen for cancer sites with the highest nausea levels at the start of palliative care and thus differences in nausea level at the start of palliative care explain some of the variations in nausea reduction seen across cancer sites. After controlling for nausea level at the start of palliative care, no nausea reduction was seen for patients with stomach cancer. The reason why a nausea reduction was most difficult to obtain among patients with stomach cancer is probably that the cancer itself (i.e., tumor growth in the stomach) causes the nausea.
Future studies of how nausea is most effectively treated in palliative care should in particular examine novel strategies among patient with cancers where nausea reductions are the hardest to obtain. Testing the effectiveness of different nausea treatments in randomized controlled trials is, however, difficult because it is difficult to recruit enough patients in a very sick nauseated patient group, and international collaborations may be needed. Population-based registry studies including longitudinal nausea assessments during palliative care combined with detailed information on medication would be another way to gain more knowledge on how nausea is most effectively treated in palliative care.  Association between patient characteristic and change in nausea is significant when controlling for nausea at the start of palliative care or b after controlling for baseline nausea, gender, age, cancer site, type of service, living situation, and survival time. Referral unit and children were removed from the fully controlled regression model in the stepwise backwards selection. c Difference in nausea change compared to the reference group = mean change minus mean change in reference group. In the regression analyses, a mean difference > 0 reflects a lower reduction in nausea from baseline to follow-up compared to the reference group whereas a mean difference < 0 reflects a higher reduction in nausea from baseline to follow-up compared to the reference group

Strengths and limitations
The main strength of this study is the large and nationwide study population and the fact that symptoms were patient reported and systematically collected at the start of palliative care and 1-4 weeks later. A limitation of this study is the proportion of patients reporting their level of nausea at the start of palliative care (69%) and the proportion surviving minimum 1 week from the start of palliative care with two nausea assessments (28%), respectively, because it means the included patients may be selected. The main explanation of non-response is probably that the study population is very sick. The patients included in the study did survive longer than non-included, suggesting that the sickest patients were not included. Another limitation is that we did not have information on whether the patients were undergoing chemotherapy or radiation therapy and information was not available either on which medications the patients received. Therefore, it was not possible to study how nausea was associated with these factors.

Conclusion
Approximately half of the patients starting in palliative care experienced nausea. Significant variations in baseline levels were found according to several patient characteristics. After the first weeks of palliative care, nausea decreased on average 7.5 points. Significant differences in nausea reduction were found according to cancer site. In unadjusted analyses, the cancer sites with high nausea levels at the start of palliative care often exhibited the largest reductions. When controlling for nausea level at the start of palliative care, the smallest nausea reductions were seen for patients with myelomatosis and ovarian cancer and no reduction was seen for stomach cancer.