Symptoms, care needs and diagnosis in palliative cancer patients in acute care hospitals: A 5-year follow-up survey

Abstract

Introduction. Palliative cancer care in acute hospitals is scarcely studied. We therefore described and compared symptoms, care needs and types of cancer sites in 2002 compared to 2007 and analysed the relationships between these factors. Methods. The study was population-based with a cross-sectional design and was carried out in medical, surgical and oncology wards in two acute care hospitals with no advanced palliative home care service. In 2002, 82 one-day-inventories were done (1 352 patients) compared to 142 one-day-inventories in 2007 (2 972 patients). Symptoms, care needs and cancer site were registered according to a questionnaire. Multiple logistic regression models were used to analyse associations between symptoms, care needs and cancer site. Results. The proportion of palliative cancer patients had decreased during a five year period (14% vs. 11%, p<0.01). The patients were older in 2007 (74 vs. 70 years, p<0.001) and had more symptoms and care needs per patient (2.6 vs. 1.6, p<0.001). The most common symptoms were pain and deterioration and the most common cancer sites were prostate and colorectal cancer in both samples. Associations between symptoms, care needs and cancer site were mostly weak. Deterioration was associated with colorectal cancer, whereas pain was not associated with any specific cancer site. In haematological malignancies there was a high occurrence of infections and a high need of blood transfusions and infusions. Stomach/oesophagus cancers were significantly associated with nausea, nutritional problems and need of infusions while unknown primary malignancies were associated with abdominal surgery and infusions. Discussion. Although we do not know all the causes for hospitalization, this study indicates that more focus should be on the symptoms instead of the specific cancer diagnosis. The findings also indicate that many palliative cancer patients’ problems would be suitable for advanced palliative home care instead of acute hospital care.

Although it has been proposed that palliative care should be available wherever the patients are [1], many palliative cancer patients are still cared for in acute care hospitals oriented towards curative interventions [2]. During the last decades the demographics of dying has changed and has affected both home care and hospital care [3]. In Sweden, the availability of palliative care services varies [4] and the reduction in number of hospital beds increase the strain and workload for both acute care hospitals and home care services [5].

Several studies describe the symptoms and care needs of palliative patients referred to advanced palliative care units or hospices [6–8]; but the symptoms and care needs for palliative patients in acute care hospitals are rarely studied. The existing studies often have methodological limitations with limited patient samples [9] and retrospective designs [10]. The aim of this study was to describe and compare symptoms, care needs and types of cancer sites in 2002 compared to 2007, and to analyse the relationships between these factors in hospitalized palliative cancer patients.

Method

Design and setting

This population-based study had a cross-sectional design and was carried out in medical, surgical and oncology wards in two acute care hospitals in a medium-sized county serving 180 000 individuals in southern Sweden during 2002 and 2007.

The palliative care context

During the study period (2002–2007) the studied area had no 24-hour palliative care services. Instead, a palliative counselling team served the population under daytime Monday to Friday. Between 2002 and 2007 the team increased its staff, from three nurses in 2002 to six nurses and one consulting physician working halftime in 2007. The palliative counselling team was affiliated with the hospitals, and especially the oncological department. It worked as a link between the hospitals and the community based nursing home care and supported the healthcare professionals.

Questionnaire

A questionnaire consisting of 14 items was designed by a multiprofessional panel in palliative care and the final form was distributed to the participating wards and completed by the nurses and the physicians at each unit. The first eight items concerned the organisation, e.g. clinic, ward, responsible healthcare professional, number of hospital beds and number of inpatients. Next four items covered the gender, year of birth, diagnosis, and place of residence of the palliative cancer patients. The last item had multiple choices and concerned their symptoms and care needs during hospitalization. A total of 21 different symptoms and care needs were available in the list, and more than one could be chosen. There was also space to write comments. The WHO palliative care definition was printed on the back of the questionnaire and it was used as a basis for the identification of palliative care patients [11]. The questionnaire was tested both in 2002 and 2007 for face and content validity by an expert panel of two physicians and four nurses, focus group interviews with nurses at the wards, and finally through a pilot study, followed by minor adjustments.

Data collection

We presented the study in 2002 and 2007 at information meetings where we discussed with the responsible nurses from the different wards how to assess a patient as palliative and how to fill in the questionnaire. A palliative patient was defined as a patient whose disease was not responsive to curative treatment and where the prognosis was limited. It was of most importance that each responsible nurse knew how to assess the patients as palliative, so we could get comparable assessments independently from the wards.

We then did one-day-inventories in the wards. This means that on a specific day each month the responsible nurses together with the physicians at the wards assessed which of the inpatients were palliative cancer patients and filled in the questionnaires. In 2002 we did 10 one-day-inventories in 10 wards: one oncology ward (20 beds), four surgical wards (104 beds) and five medical wards (119 beds). In all 82 one-day-inventories were analysed in 2002 (data missing for 18 days, i.e. 18%). In 2007 we did 12 one-day-inventories in 12 wards: one oncology ward (17 beds), four surgical wards (89 beds) and seven medical wards (153 beds). In all 142 one-day-inventories were analysed in 2007 (data missing for two days, i.e. 1.4%). We did one reminder call or sent one e-mail reminder both in 2002 and 2007. In total, 1 352 were registered and assessed in 2002, and 2 972 patients in 2007.

Statistical analysis

The computer based statistical programme SPSS, version 16.1 (SPSS Inc, Chicago, Ill), was used for statistical analyses. We did multiple logistic regression analyses (backward stepwise, likelihood ratio) to analyse associations between each symptom and care need and cancer sites. Prior to these analyses we assessed bivariate associations between each symptom and care need and the different cancer sites. Cancer sites that were associated with the respective symptom and care need with a p-value ≤0.20 were included in the logistic regression model. All analyses were performed controlling for age. In order to evaluate the strength of the association between symptoms and care needs with type of cancer site, also the Nagelkerke R-square statistics were calculated. Data from 2002 and 2007 were used for general comparisons, but only data from 2007 was used in the regression analyses in order to focus on the current situation (2007), as associations are dependent on changes/improvements in therapy, and such changes between 2002 and 2007 cannot be out ruled. P-values <0.05 were regarded as statistically significant.

Ethics

This study was carried out in accordance with the Declaration of Helsinki of the World Medical Association [12]. Swedish law and praxis consider this study part of normal clinical improvement procedures since the study did not involve any identifiable patient data. Approval by The Regional Ethics Committee was thus not considered necessary. Ethical approval was instead granted by the operations managers and the heads of the departments at both hospitals. Written information about the study was sent out to the head nurses of wards who selected a responsible nurse for the study at the ward.

Results

Characteristics and situational data of the palliative cancer patients

In 2002, 1 352 patients were registered and 192 were assessed as being palliative; in 2007, 2 972 patients were registered and 328 were assessed as being palliative, i.e. a minor decrease (14% vs. 11%, p<0.01). The oncology ward was the only ward that had significantly less palliative cancer patients during 2007 (71% vs. 58%, p<0.05), while there were no significant difference in the assessment between the years at the surgical wards (16% vs. 14%, NS) or at the medical wards (5% vs. 4%, NS). Even though the total proportion of palliative cancer patients had decreased and the oncology ward registered less palliative cancer patients, there were no differences of how the total number of the palliative patients was divided in the wards. Yet there was a decrease of men at surgical wards between the years (56% vs. 37%, p<0.01) (Table I). The patients were significantly older in 2007 (74 vs. 70 years, p<0.01) (Table I).

Table I. Characteristics and situational data of the palliative cancer patients (n=520).

	2002			2007
	All (n=192)#	Male (n=107)	Female(n=80)	All (n=328)	Male (n=182)	Female (n=146)
Age; mean, (quartiles)	70 (62-71-79)	71 (62-73-78)	70 (61-69-79)	74 (67-76-83)**	74 (76-66-84)*	74 (69-76-82)*
Oncology ward (%) ŧ	31	25	40	37	36	38
Surgical ward (%) ŧ	55	56	50	40	37**	43
Medical ward (%) ŧ	14	19	10	23	27	19
Number of symptoms and care needs per patient; mean, median (range)	1.6, 1 (1-5)	1.7, 1 (1-4)	1.5, 1 (1-5)	2.6, 2 (1-9)***	2.6, 2 (1-6)***	2.7, 2 (1-9)***

# = gender not given in five patients.

ŧ = the percentage describes the distribution of the palliative cancer patients between the wards, but not the percentage within the ward.

Significant difference between the years, *p=<0.05, **p=<0.01, ***p=<0.001.

Symptoms and care needs

There was a difference in the mean number of symptoms and care needs for each patient between the years (1.6 in 2002 vs. 2.6 in 2007, p<0.001) (Table I). Overall, the three most common symptoms and care needs during hospitalization were pain 39%, deterioration 27% and nausea 17% in 2002, compared to pain 44%, deterioration 44%, and infections 25% in 2007. Deterioration here means a successively increasing weakness. The detailed distribution for men and women is presented in Table II.

Table II. Prevalence of symptoms and care needs for palliative cancer patients during hospitalization.

	2002			2007
Symptoms and care needs	All % (n=192) #	Male % (n=107)	Female % (n=80)	All % (n=328)	Male % (n=182)	Female % (n=146)
Abdominal surgery	6	4	8	5	3	8
Blood transfusion	8	9	8	13	13	12
Deterioration	27	30	24	44 a***	42♂*	47♀**
Diagnostic procedure	9	8	10	24 a***	25♂***	23♀*
Infection	7	8	6	25 a***	27♂***	23♀**
Infusion	3	4	3	13 a***	10	17♀**
Nausea	17	22	13	19	14	25♀*
Nutrition	10	11	9	3 a**	2♂**	4
Pain	39	41	35	44	44	45
Radiotherapy	7	8	5	17 a**	20♂**	13
Social issues	6	6	8	17 a**	16♂**	17
Surgical stenting	3	3	3	10 a**	ll♂*	9
Other	4	6	3	13 a**	14♂**	11♀**

# = gender not given in five patients.

a = significant difference between the years for all, ♂ = significant difference between the years for male, ♀= significant difference between the years for female. *p=<0.05, **p=<0.01, ***p=<0.001.

Cancer sites

The three most common cancer sites in 2002 were prostate 18%, colorectal 15% and lung 12%, and in 2007 prostate 13%, colorectal 13%, lung and haematological malignancies 12% (Table III). The proportion of patients with haematological malignancies increased between 2002 and 2007, which was the only statistically significant change between the two observed years. The detailed distribution for men and women is presented in Table III.

Table III. Prevalence of cancer sites in palliative cancer patients.

	2002			2007
Cancer site	All % (n=192) #	Male % (n=107)	Female % (n=80)	All % (n=328)	Male % (n=182)	Female % (n=146)
Breast	4	-	10	6	-	13
Colorectal	15	15	15	13	12	14
Gynaecological	10	-	21	6	-	13
Haematological	5	5	4	12 a**	14 ♂*	11
Lung	12	15	9	12	14	10
Pancreas	10	12	6	8	6	11
Prostate	18	31	-	13	24	-
Stomach/oesophagus	5	6	4	6	6	7
Unknown primary	5	1	11	6	3	10
Urological	11	13	8	8	11	5
Other	5	2	12	10	10	6

# = gender not given in five patients.

a = significant difference between the years for all, ♂= significant difference between the years for male, ♀ = significant difference between the years for female. *p=<0.05, **p=<0.01, ***p=<0.001.

Relationships between symptoms and care needs with cancer site in 2007

Several regression models were tested. Gender related cancers (gynaecological, breast, and prostate cancer) strongly influenced the overall model and overshadowed certain associations. Therefore, the gender related cancers were excluded in the overall model (Table IV); instead separate models/analyses were performed for men and women (Table V).

Table IV. Relationships between symptoms and care needs with cancer sites for palliative cancer patients in 2007.

	All (n=328)^α
Symptoms and care needs	Cancer site	p-value	OR	95% CI	Nagelkerke R-Square
Abdominal surgery	Colorectal	.001	7.2	2.3-22	.18
	Unknown primary	.007	7.8	1.8-34
Blood transfusion	Haematological	<.001	14	6.7-31	.27
Deterioration	Colorectal	.035	2.1	1.0-4.1	.06
Diagnostic procedure	___
Infection	Haematological	<.001	10	5.0-22	.19
Infusion	Haematological	.025	2.7	1.1-6.4	.07
	Stomach/oesophagus	.005	4.7	1.6-14
	Unknown primary	.032	3.4	1.1-10
Nausea	Stomach/oesophagus	.012	3.8	1.3-11	.13
Nutrition	Stomach/oesophagus	.001	15	3.1-68	.22
Pain	___
Radiotherapy	___
Social issues	___
Surgical stent	Pancreas	.001	4.9	1.9-13	.08

OR = Odds Ratios; CI = Confidence Interval.

α = Gender specific cancer sites not included.

As mentioned, the three most common symptoms and care needs for all patients in 2007 were pain, deterioration and infection. Deterioration was weakly associated with colorectal cancer and infection with haematological malignancies, whereas pain was not associated with a specific cancer site in the overall statistical model (Table IV). Patients with haematological malignancies had a higher occurrence of infections and a higher need of blood transfusions and infusions compared to other patients. Patients with stomach and oesophagus cancers had significantly more reports of nausea, nutritional problems and infusions. Patients with unknown primaries had more reports of abdominal surgery and infusions (Table IV).

In the gender-specific model, the associations were similar to that of the overall model, but with the following differences (Table V): in men, pain and diagnostic procedures were related to a diagnosis of disseminated prostate cancer and the probability of deterioration was significantly higher not only for colorectal cancer (as in the general model), but also for pancreatic and lung cancer. In women, colorectal and gynaecological cancers were related to abdominal surgery and in this model the probability of delivering palliative radiotherapy had OR (odds ratios) of 5–10 for breast, lung and stomach/oesophagus cancers (Table V).

Table V. Relationships between symptoms and care needs with cancer sites for men and women in 2007.

	Male (n=182)					Female (n=146)
Symptoms and care needs	Cancer site	p-value	OR	95% CI	Nagelkerke R-Square	Cancer site	p-value	OR	95% CI	Nagelkerke R-Square
Abdominal surgery	Unknown primary	.004	36	3.2-422	.26	Colorectal	.002	17	2.9-95	.28
						Gynaecological	.006	13	2.0-85
Blood transfusion	Haematological	<.001	25	8.0-70	.35	Haematological	<.001	12	3.6-42	.24
	Stomach/oesophagus	.008	8.3	1.8-40
Deterioration	Colorectal	.047	2.0	1.0-3.9	.16	___
	Pancreas	.003	9.4	2.2-41
	Lung	.002	4.4	1.7-11
Diagnostic procedure	Prostate	.043	2.2	1.0-4.9	.08	Unknown primary	.013	4.4	1.4-14	.11
Infection	Haematological	<.001	13	4.8-36	.24	Haematological	<.001	7.8	2.5-25	.19
Infusion	Haematological	.001	6.5	2.0-21	.17	___
	Stomach/oesophagus	.014	7.0	1.5-33
Nausea	Stomach/oesophagus	<.001	17	3.7-79	.19	___
Nutrition	___					Stomach/oesophagus	.004	19	3.2-111	.37
Pain	Prostate	.007	2.8	1.3-5.9	.14	___
Radiotherapy	___					Breast	.017	5.1	1.3-20	.22
						Lung	.003	8.8	1.8-37
						Stomach/oesophagus	.009	10	1.8-58
Social issues	___					Breast	.016	3.9	1.3-12	.09
Surgical stent	___					Pancreas	.004	6.9	1.9-26	.13

OR = Odds Ratios; CI = Confidence Interval.

However, none of the final models were exhaustive. Only two models had a Nagelkerke R-square over 0.3; for men blood transfusion was associated with haematological malignancies and stomach/oesophagus cancer and for women nutrition was associated with stomach/oesophagus cancer (Table V).

Discussion

In this 5-year repeated population-based study of palliative cancer care in acute hospitals pain and deterioration were the most common symptoms over time. The number of hospitalized palliative cancer patients had slightly decreased, they were on average older, and had more symptoms and care needs. There were no significant differences over the five year period regarding the diagnostic panorama except for an increase of haematological malignancies. When investigating how symptoms and care needs were associated with cancer sites, we found weak associations to specific cancer sites. Pain was the most prevalent symptom, but evenly distributed and only related statistically to prostate cancer in men, while pain did not relate to any specific cancer site in women. On the other hand: certain malignancies were associated with typical symptoms and care needs that need to be attended in clinical practice, for example, haematological malignancies were significantly associated with the need of blood transfusions, infusions and infections, factors that need to be taken into consideration when planning for advanced home care.

The study had the same design during 2002 as 2007, with the same questionnaire, and a continuity of responsible nurses at each ward. We do not know why or for how long time patients were hospitalized, but we do know what kind of symptoms and care needs they had during hospitalization. The study was conducted in two acute care hospitals in an area without any advanced palliative care services. The current results may therefore differ from hospitals in regions where advanced palliative care services are available on a 24/7 basis. Between 2002 and 2007 the palliative counselling team serving the population under study increased its staff. This increased the palliative care competence of the area and may be a reason why the prevalence of hospitalized palliative cancer patients decreased in the study period. Cancer patients’ symptoms can indeed be improved by hospital based palliative care teams [13] and palliative home care can reduce the need for hospitalization in acute care hospitals if the care is provided by trained personnel [14].

Otherwise the health care organisation in the county was stable during the five year study period, with only some hospital wards being reorganised. In 2007 in-patients with haematological malignancies were treated at the oncology ward instead of at the medical department. More medical wards were included in the study in 2007, but this did not affect the total number of palliative cancer patients. The proportion of palliative cancer patients decreased from 14% to 11% and this decrease was seen at the oncology ward, while the number in the medical and surgical wards was the same. In a Swedish study from 2001, 18% of the patients in a university hospital had incurable cancer and were assessed as being in need of palliative care. The authors were surprised by the low number of hospitalized palliative cancer patients and concluded that it reflected “a well-functioning hospital-based home-care unit” [9]. Yet, we consider that even 11% palliative cancer patients in acute care settings is a number that could be lowered with improved advanced palliative care. As many as 85% of healthy Swedes would prefer advanced home care in a palliative care situation, provided that the standards of the service is comparable with hospital care, and that the service is offered on a 24/7 basis [15].

The number of symptoms and care needs per palliative cancer patient, as assessed by nurses and physicians, increased significantly over time from 1.6 to 2.6 in our study. Yet, compared to other studies of symptoms assessed by patients, this number is low. A probable reason for this is that we did only register symptoms and care needs relevant for the actual hospitalization, not total symptom occurrence. Using patient-administered screening instruments, Tsai et al. [16] found a median number of 11 symptoms per patient on admission to a palliative care unit in Taiwan, but this list also included symptoms that are normally not reasons for hospitalization, e.g. taste alterations or dry mouth. Likewise, in a British study of cancer outpatients, the majority reported a mean of 10 symptoms [17]. In another British study palliative patients referred to community services and hospice had an average of seven symptoms, assessed retrospectively by chart analysis, while patients referred to hospital or outpatient clinics had three symptoms per patient which is comparable with our results [7].

Pain was the most common symptom in both years in our study. This was despite the growth of knowledge regarding pain assessment and new pain treatments during the last decade. In other studies pain was also the most common symptom among palliative cancer patients [[8,[18]. When pain is present it is often impossible to neglect and may therefore overshadow other symptoms of the patient [18]. Hence, other important needs may appear more frequently when pain is eliminated or absent [8]. For example in the county of Östergötland with almost 100% coverage of advanced palliative home care around the clock, data from 431 consecutive patients showed that pain and nausea were well-controlled (mean 2.5 and 1.8 on a 10 point scale). Instead, other symptoms such as poor appetite and fatigue were reported as bothersome symptoms in this population of cancer patients close to death [19].

It is thus considered important to ask for the most troublesome symptom since this better reflects the patient’s perspective. If patients are not asked for their most troublesome symptom 1/3 of the symptoms might be missed [20]. This means that there are probably a number of unrecognised symptoms in our study since the symptom assessments were done by nurses and physicians and not by asking the patients. A general problem with symptom reports is that a distinction seldom is made between symptom prevalence, intensity and perceived distress, three dimensions that are not equivalent [21].

Our study shows that the medical interventions have increased during the 5-year interval and there are significant differences between the years for the following medical interventions: diagnostic procedures, radiotherapy and surgical stenting. The only medical intervention which had significantly decreased was enteral nutrition treatment, whereas the need of infusions increased. Even though the care need “diagnostic procedure” increased in 2007, we could not see any association to a specific cancer site for the total number of patients, only for prostate cancer in men and unknown primary cancer in women. One reason for this could be that prostate cancer was related to pain, which in turn was related to metastases, which eventually requires special diagnostic work-up. As for the women with unknown primary cancer, some specific cancer types in women, such as breast and ovarian cancer, may respond to treatment even when the cancer is disseminated. And this, of course, first requires them to be typed. Medical interventions, both diagnostic and therapeutic, cause suffering, and so called routine procedures in acute care settings are sometimes traumatic for seriously ill patients [2]. Medical interventions are probably sometimes necessary, but the healthcare professionals need to be aware of their consequences before starting an intervention.

Even though the relationships between symptoms and care needs and type of cancer sites were highly significant in the majority of cases, we found low values of the Nagelkerke R-square. This may partly be a problem of statistical power, but the fact that few symptoms and care needs were associated with a specific cancer site, supports the idea that in palliative care focus should shift from diagnosis to symptoms, which often are similar for many cancers [22]. This means that the palliative cancer patients do not always have to be treated for symptoms at his or her “home clinic” in acute hospitals. Instead they could be treated in palliative care units or by advanced palliative care teams. Patients with haematological malignancies may be an exception since they have more complex complications. When patients are cared for at the right care level, both patients, relatives, and the healthcare system will benefit. There has to be close links between oncology services and palliative care to optimize the care of the palliative cancer patients [23]. We agree with the suggestion that palliative care services should be integrated into acute care settings, with a broader discussion of quality of life related goals [2].

Conclusion and implications

In this population-based repeated 5-year study we found that palliative cancer patients have several symptoms and care needs during hospitalization, but relatively few were associated with specific cancer sites. This indicates that the care focus should be on treating symptoms irrespective of the diagnosis in palliative cancer care. Our study indicates that many palliative cancer patients are still cared for in hospitals with an acute care philosophy, while there are evidence that many palliative cancer patients’ problems may be resolved in advanced palliative home care.

Acknowledgments

This study was facilitated by grants from the Kronoberg County Research, Skane County Council’s research and development foundation, Södra regionvårdsnämnden, Cancerstiftelsen i Kronobergs län and the Swedish Cancer Society. Thanks are due to Novko Gavric, MD, who was the initiator of this study.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.