Palliative Care Units in The Netherlands: Changes in Patients Functional Status and Symptoms

Transcription

1 Vol. 28 No. 3 September 2004 Journal of Pain and Symptom Management 233 Original Article Palliative Care Units in The Netherlands: Changes in Patients Functional Status and Symptoms Michael A. Echteld, PhD, Luc Deliens, PhD, Gerrit van der Wal, MD, PhD, Marcel E. Ooms, MD, PhD, and Miel W. Ribbe, MD, PhD Institute for Research in Extramural Medicine (EMGO Institute) (M.A.E., L.D., G.v.d.W., M.E.O., M.W.R.), Department of Nursing Home Medicine (M.A.E., M.E.O., M.W.R.), and Department of Social Medicine (M.A.E., L.D., G.v.d.W.), VU University Medical Center, Amsterdam, The Netherlands; Pain Expertise Center (M.A.E.), Erasmus Medical Center, Rotterdam, The Netherlands; and End-of-Life Care Research Group (L.D.), Department of Medical Sociology and Health Sciences, Vrije Universiteit Brussel, Brussels, Belgium Abstract Although efforts have been made to define optimal terminal care in palliative care units (PCUs), comprehensive longitudinal evaluations of care outcomes in PCUs at the end of life are scarce. In this study, changes in functional status (assistance needed for walking, and toilet use) and symptoms (pain, nausea, shortness of breath, depression, and anxiety) were assessed in all patients (n 355) admitted to 10 PCUs in Dutch nursing homes. Outcomes were measured at 24 hours, 48 hours, one week, and two weeks before death, and at PCU admission. Results show that functional status deteriorated from admission to one week before death, but most symptoms did not worsen in the last three weeks before death. Decreases in pain, anxiety, and nausea were observed. The results suggest that the care provided in the PCUs stabilized the symptom levels. Patients who die between two and four weeks appeared to have more favorable symptom change patterns than patients who die within two weeks, which supports the recommendation to admit eligible patients in earlier phases of their disease. Limitations include the use of proxy measures and some forms of selection bias, which may lead to underestimation of symptom levels. J Pain Symptom Manage 2004;28: U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Key Words Palliative care, terminal care, symptoms, cancer Address reprint requests to: Michael A. Echteld, PhD, Pain Expertise Center, Erasmus Medical Center, Dr. Molewaterplein 40, 3015 GD Rotterdam, The Netherlands. Accepted for publication: December 13, U.S. Cancer Pain Relief Committee Published by Elsevier Inc. All rights reserved. Introduction This article reports on the evaluation of palliative care units (PCUs) in terms of changes in patient outcomes functional status and symptoms. In The Netherlands, patients who cannot remain at home at the end of their lives due to the high demands of the care or the /04/$ see front matter doi: /j.jpainsymman

2 234 Echteld et al. Vol. 28 No. 3 September 2004 lack of informal helpers are mostly referred either to hospices, hospitals, or nursing homes. Although these care facilities in The Netherlands are often staffed with nurses and physicians, knowledge and skills in psychosocial and spiritual care and symptom management at the end of life are often lacking. In addition, most hospitals and nursing homes do not offer privacy for dying patients and their relatives, and as a rule do not offer overnight stay facilities and bereavement care for relatives. 1,2 In order to meet this lack of specialized care, Dutch nursing homes started PCUs for short-term terminal care for non-residents in the early 1990s. In March 2003, 62 nursing homes had a PCU, adding to the total capacity of 211 beds, ranging from 1 to 14 beds. 3 The PCUs have specialized multidisciplinary care teams, have separate housing or are situated in separate nursing homes wards, and provide individual care for both patients and their informal helpers and next of kin. 2 However, no national requirements for quality of care for these PCUs presently exist in The Netherlands. The Center for the Development of Palliative Care Amsterdam installed by the Dutch Ministry of Public Health started a project group for the development of requirements for care in 10 units for short-term terminal care in nursing homes. These PCUs were meant for patients from outside the nursing homes, thus fulfilling the need for high-care replacements for domestic care. Fourteen requirements were formulated based on existing care protocols 4 (see Appendix). In a study determining the extent to which the requirements were met, it was found that in these 10 PCUs, 71% of the requirements for PCU design and organization of care, 58% of the requirements for complement of staff and team functioning, and 77% of the requirements for staff expertise and training were met (69% overall). 5 The PCUs differed in the number of requirements that were met: the percentages of requirements that were met ranged from 53% to 85%. Given the rapid decline of physical functioning in the last weeks of life, the absence of change in symptoms can be considered a favorable outcome. A number of studies seem to suggest that this goal is feasible. In a study of patients admitted to specialized palliative care facilities, the mean overall quality-of-life score was 7 out of a maximum score of 10. In addition, most symptoms did not change from admission to two weeks before death. 7 Another study showed that most patients attending a palliative daycare clinic had stable depression scores. 8 A study by Cohen et al. 9 showed that various outcome measures may improve due to the care in PCUs: overall quality of life and physical, psychological, and existential well-being scores improved from PCU admission 7 to 8 days later. Patient interviews corroborated the data and indicated a high degree of satisfaction with care. Typically, PCUs provide whole-person care, and have been acknowledged as the appropriate approach to the care of the terminally ill. 6 However, longitudinal evaluations of functional, physical, and emotional care outcomes in PCUs at the end of life have not been documented before. The present study investigates changes in functional status (the amount of assistance needed for walking and toilet use) from PCU admission to one week before death. Because the PCUs in this study aim to manage symptoms towards the end of life as well as is possible, changes in symptoms were determined. Symptoms generally seen as an important source of patient burden (pain, nausea, shortness of breath, depression, and anxiety) were assessed. Methods Setting Patient data were collected in 10 palliative care units in the west, central, and north of The Netherlands between January 2001 and July The PCUs were established by nursing homes that were members of a governmental project group involved in establishing requirements warranting optimal care in PCUs. The nursing homes volunteered to become members of the project group because they either had plans for starting a PCU or already had started one. Fourteen requirements were drafted by the project group 4 (see Appendix). Using these guidelines, seven units were established between August 2000 and June 2001; the remaining three units had been previously established between 1995 and The project group also facilitated evaluation of the units, resulting in the present study. Characteristics of the PCUs are presented in Table 1.

3 Vol. 28 No. 3 September 2004 Palliative Care Units in The Netherlands 235 Table 1 Characteristics of the Palliative Care Units (PCUs) in the Study, Data Collection Duration, and the Number of Included Patients Urban/Rural Number of Separate PCU Housing Data Collection Duration Numer of Patients PCU No. Location Beds with 24/7 Staff (months; period) in the Study 1 Urban 4 No 4 (8/01 11/01) 10 2 Urban 4 No 12 (1/01 12/01) 31 3 Rural 8 No 4 (11/01 2/02) 13 4 Urban 4 No 10 (9/01 6/02) 37 5 Urban 4 No 16 (1/01 4/02) 34 6 Urban 5 No 15 (4/01 6/02) 55 7 Rural 8 Yes 16 (4/01 7/02) 42 8 Rural 6 Yes 14 (8/01 6/02) 47 9 Urban 5 No 9 (10/01 6/02) Urban 5 No 11 (7/01 5/02) 42 Total: 355 Data Collection The PCUs were asked to participate in data collection for a year. All PCUs participated for about a year, but five PCUs participated for less than five months due to lack of personnel. Four PCUs were willing to collect data for longer than a year in order to compensate for the other PCUs. The average data collection period was 11.1 months (range 4 16). The PCUs appointed a data collection coordinator and staff members to perform the data collection. These staff members were nursing home physicians, nurses, or psychosocial care providers, and received training for data collection from one of the researchers (MAE, psychologist). Three or four training sessions were conducted for each unit; additional training sessions were conducted for new staff members who were appointed while data collection was ongoing. Data were collected by means of observation instruments and by consulting colleagues and patient records only. In addition, data collectors were instructed to mark items as missing when they did not feel confident assigning scores. Measures Symptoms were assessed using the Edmonton Symptom Assessment Scale (ESAS). 10,11 The ESAS consists of nine 10-cm visual analog scales rating pain, fatigue, nausea, depression, anxiety, drowsiness, appetite, well-being, and shortness of breath. Scores ranged from 0 10 (higher scores reflect higher symptom severity). Functional status was assessed using the RAI MDS-PC. 12 The amount of help needed for six activities of daily living (bed mobility, walking, eating, toilet use, personal hygiene, and bathing) was assessed using a seven-point rating scale (1 independent; 2 setup help only; 3 supervision; 4 limited assistance; 5 extensive assistance, 6 maximal assistance; 7 total dependence). In this study, only walking and toilet use were used, because these were the functional status variables having sufficient variability at admission and one week before death. The ESAS was administered at PCU admission, at weekly intervals after admission, and retrospectively, hours before death and 24 0 hours before death. Functional status was assessed at PCU admission and subsequent weekly measurements. Ethics The VU University Medical Center Institutional Review Board allowed using a no objection procedure for data collection, which entailed that staff members of the PCU collected proxy data without providing the researchers with information with which the patients could be identified. At PCU admission, patients or their relatives were informed about the study, and were given the opportunity to object to data collection. None of the patients or their relatives refused participation at PCU admission or any other time during data collection. Obtaining informed consent was not required. Statistical Analyses The ESAS outcome variables were screened for normality using the skewness measure and the Kolmogoroff Smirnoff test. Outliers were identified as Z-scores 3or 3. Descriptive statistics were used to characterize the patients. Mean changes in outcome variables were

4 236 Echteld et al. Vol. 28 No. 3 September 2004 assessed retrospectively, that is, the reference point for the analyses was death. Paci et al. 13 showed that the retrospective approach to change assessment is more appropriate than the prospective approach. Changes in outcomes were determined within three groups of patients: Group 1: patients who died within one week after PCU admission (n 92); Group 2: patients who died between one and two weeks after PCU admission (n 71); Group 3: patients who died between two and four weeks after PCU admission (n 75; there were no significant differences in mean symptom and ADL scores between patients with 2 3 and 3 4 weeks survival times). The remaining patients were not used in the within-group analyses because of the following reasons: 1) The survival times of the remaining patients were widely distributed, which made combining these patients into one group difficult to justify; 2) within the remaining patients, significant differences were found in mean symptom and ADL scores between patients with shorter and longer survival than the median of 42 days. Longer surviving patients had better walking, toilet use, nausea, shortness of breath, and anxiety scores. Including this group would thus lead to overly optimistic (short survivors) or pessimistic (long survivors) end-of-life conclusions within this group. Distinguishing between outcomes of short and long survivors in this group is not possible. Within each group separately, scores of all measurement times (up to three weeks) including the last measurement before death were compared. In addition, comparisons between data collected at PCU admission and 48 0 hours before death were made in all patients. Paired sample t-tests were conducted to test for mean changes over measurement times in the ESAS symptom scores, which have an interval level of measurement. Kruskall-Wallis and Wilcoxon signed rank tests were used for analyses involving the physical functioning variables, which are of an ordinal level of measurement. No outliers were found in the ESAS variables. All ESAS variables deviated from the normal distribution (Kolmogorov Smirnov Z ; P 0.01). However, the skewness values of these variables ranged from , indicating fairly symmetrical distributions, which means that the deviations from the normal distribution were not due to the skewness of the distributions of the variables. The t-test is robust against deviations from the normal distribution that do not entail asymmetry. Square root transformations did not improve skewness or Kolmogorov Smirnov values in the ESAS variables, and were, therefore, not performed. All analyses were performed using SPSS for Windows version Results First, patient survival, demographics and diagnoses, and other results of the patient recruitment process will be presented. Next, missing data are presented in order to investigate whether systematic relationships exist between the rate of missing values and patient outcomes. In the last two sections, levels and changes in functional status and symptoms are presented. Patient Sample and Characteristics For the period the PCUs agreed to participate, all patients admitted for terminal care and with a life expectancy of three months or less were included. Data were collected from 355 patients. Twenty patients who died in the PCU lived longer than three months after PCU admission; patient survival times in the PCU are plotted in Figure 1. Thirty-six patients died outside the PCU: 3 patients returned to their homes because they preferred to die there, 4 patients died in the hospital while being treated, and the remaining 29 patients were transferred to other care facilities because they were no longer considered terminally ill. Ten Fig. 1. Survival times of patients who died in the PCUs. Median survival time was 14 days (SD 32.7).

5 Vol. 28 No. 3 September 2004 Palliative Care Units in The Netherlands 237 patients who were admitted to the PCUs to temporarily relieve the burden of domestic care for relatives were included in the study. Two of these patients actually returned to their homes the remaining 8 patients died in the PCUs. Within the total number of 355 patients, limited data collection was performed retrospectively in 46 patients due to temporary lack of personnel in combination with high complexity of required care only PCU admission date, date of death, primary diagnosis, demographic data, and social support data were collected. Changes in patient functional status and symptoms, therefore, could be assessed in the remaining 309 patients (87%). Patient characteristics are reported in Table 2. Patients were mostly elderly, were diagnosed with cancer, and 98.2% had at least one informal caretaker. Most patients were admitted to the PCU only for a short time. Only 1.8% of the patients were comatose one week after admission. However, 51.4% of the patients were rated as having fluctuating levels of consciousness. Level of consciousness was worse in patients in Group 1 than compared to the other groups (Kruskall Wallis χ ; df 3; P 0.01). Missing Data The number of missing data within each valid case differed over variables and measurement points, due to the choices of data collectors to mark variables as missing if they did not feel confident assigning a score. Percentages of missing data at different measurement points for functional status and symptom variables are presented in Table 3. The table shows that the largest proportion of missing data was found in depression, and the lowest in shortness of breath; the largest proportion of missing data was obtained at 24 hours before death, and the lowest one week before death. To test whether the missing data were systematically related to patient characteristics, the median-dichotomized walking and toilet use variables were related to missing not missing dichotomized depression and anxiety variables. The likelihood of higher missing values in depression and anxiety scores was only significant in patients in need of extensive assistance with toilet use at PCU admission (χ ; df 1; P 0.01, and χ ; df 1; P 0.05, respectively). No significant relationships were found one week before death. Table 2 Demographics and Diagnosis of Included Patients at Admission to the Palliative Care Unit (PCU) n 355 Age (M) 73.4 (28 103; (SD 12.5) Men (%) 45.4 Religious affiliation (%) Protestant 34.0 Roman Catholic 27.8 Other 5.3 No religious affiliation 33.0 Marital status (%) Widowed 43.7 Married/partner 35.3 Never married 11.3 Divorced 9.6 Primary informal caretaker (%) Child 43.8 Spouse 24.9 Other family 15.8 Friend 7.3 Parent 1.2 No primary informal 1.8 caretaker Primary diagnosis (%) Cancer 79.5 Lung 19.9 Gastrointestinal 13.2 Liver & pancreas 11.0 Urogenital (men) 10.7 Head and neck 5.8 Urogenital (women) 4.3 Breast 4.3 Other 10.3 Heart failure 3.4 Stroke 3.4 COPD 1.8 Neurodegenerative 1.5 Kidney failure 0.9 Infections 0.9 Other 8.6 Physician s estimate of life expectancy at PCU admission (%) 1 week week 1 month month 6 months 36.7 Unknown or longer 16.0 Not terminally ill 2.0 Medication at PCU admission (%) Analgesics 63.6 Psychoactive drugs 45.4 Laxatives 36.3 Medication for 25.8 heart disease Diuretics 16.9 Residence before PCU admission (%) Hospital 48.6 Home without 20.3 domestic care Home with 20.3 domestic care Nursing home 7.9 Other 2.9

6 238 Echteld et al. Vol. 28 No. 3 September 2004 Table 3 Percentages of Missing Values Due to Rater Uncertainty About Assigning Scores Measurement Points Admission One Week before Death hours before Death 24 hours before Death Functional status Walking a a Toilet use a a Symptoms Pain Nausea Shortness of breath Depression Anxiety a Functional status was not measured at the hour and 24 hour death intervals. Levels of Functional Status and Changes in Functional Status One week before death, 54.9% of the patients needed extensive assistance or more assistance with walking; 36.5% was totally dependent regarding walking. Again one week before death, 72.8% of the patients needed extensive assistance or more with toilet use, and 50.0% was totally dependent regarding toilet use. Functional status changes from PCU admission to the week before death within three groups (all patients, Group 2, and Group 3) are shown in Table 4. Within each group, and for walking and toilet use separately, mean ranks were calculated and tested. Patient s functional status was decreasing: more patients were worsening from admission to one week before death than improving from admission to one week before death, except for toilet use within Group 2. Differences between groups at PCU admission were significant with respect to walking (Kruskall Wallis χ ; df 3; P 0.01) and toilet use (Kruskall Wallis χ ; df 3; P 0.01), indicating that patients with short survival times had worse functional status scores than patients with long survival times. Differences in functional status between groups one week before death were not significant. Table 4 Mean Ranks of Patients with Deteriorating Walking and Toilet Use Scores (Negative Ranks) and Improving Walking and Toilet Use Scores (Positive Ranks) from PCU Admission (T0) to One Week Before Death (T1) Group a Functional Status Variable Rankings b n Mean Ranks Wilcoxon Z All patients Walking Negative ranks Positive ranks c Ties 41 Toilet use Negative ranks Positive ranks c Ties 55 Group 2 (death between one Walking Negative ranks and two weeks) Positive ranks c Ties 16 Toilet use Negative ranks Positive ranks Ties 15 Group 3 (death after two weeks) Walking Negative ranks Positive ranks d Ties 10 Toilet use Negative ranks c Positive ranks Ties 25 a Group 1 was not included because in this group, functional status measures were only measured once (at admission). b Lists the number of negative ranks (the number of times the rank at T0 is less than the rank at T1), positive ranks (the number of times the rank at T0 is greater than the rank at T1), and ties. c P d P 0.05.

7 Vol. 28 No. 3 September 2004 Palliative Care Units in The Netherlands 239 Levels of Symptoms and Changes in Symptoms The maximum range of scores of pain, nausea, shortness of breath, anxiety, and depression was 0 to 10. Figure 2 showed that the highest mean symptom score was 5.1 (pain 48 hours before death in Group 2). The levels of all symptom scores were thus mid-level or lower. Changes in symptoms were calculated within the three different groups of patients and are plotted in Figure 2. Most symptoms did not change significantly within the last two weeks before death including PCU admission, but a number of increases and decreases in symptoms were observed. In all patients, pain scores increased from admission to 48 hours before death, but decreased from 48 to 24 hours before death. In Group 2 (death within two weeks), pain scores increased from admission to one week before death; no other significant changes in pain were observed. Nausea scores decreased in all patients, from one week to 48 hours before death in Group 2, and from two to one week and 48 to 24 hours before death in Group 3. Shortness of breath scores increased from admission to 48 hours before death in all patients, and from 48 to 24 hours before death in Group 1. An increase in shortness of breath scores was observed from admission to one week before death in Group 2; all other changes in shortness of breath were not significant. Depression scores did not change significantly, and anxiety scores only decreased from 48 to 24 hours before death in Group 1. A number of changes had P-values between 0.10 and 0.05, and can be considered as trends. Anxiety improved from 48 to 24 hours before death in all patients (t 1.82; P 0.07) and in Group 2(t 1.89; P 0.07). Trends are not indicated in Figure 2. Discussion In The Netherlands, requirements for PCUs were drafted in order to ensure the best possible replacement for home terminal care. 4 The PCUs in this study succeeded in meeting most requirements, 5 and the present study investigated whether the care provided in these PCUs led to favorable patient outcomes at the end of life. The results of this study show that functional status deteriorated from admission to one week before death. Except for shortness of breath in patients who die within one week, outcomes did not worsen in the last three weeks before death. In addition, decreases in pain, anxiety, and nausea were observed. Furthermore, the mean outcomes were not higher than mid-level. The results indicate that end-of-life care provided in the PCUs stabilized the symptom levels. Comparing the results of this study with other studies is difficult, because outcome measures, measurement times, and patient groups reported elsewhere are often incompatible with the current study or insufficient data are reported to perform an adequate comparison. In many studies in PCU patients, no outcome Fig. 2. Mean scores of pain, nausea, shortness of breath, depression, and anxiety for all patients and within three patient groups. *t-test, P 0.05; **t-test, P 0.01.

8 240 Echteld et al. Vol. 28 No. 3 September 2004 change data are reported, because only crosssectional data were collected. Paci et al. 13 found that pain and distress increases towards death in patients admitted to PCUs in Italy. However, the survival time of the patients reported here (median 37.9 days) is much longer than in the current study (median 14.0 days), suggesting that patients cannot be compared well. A study in terminally ill patients with comparable survival times (mean 20.9 days), who were admitted to palliative care wards in Hong Kong hospitals, 7 showed that physical well-being, meaning of existence, and satisfaction with oneself deteriorated from admission to (on average) 5.6 days prior to death. Cohen et al. 9 showed that existential well-being in patients admitted to PCUs improved from admission to a week later, but this group is not representative of the 92% of other patients who could not be interviewed. The results in the current study compare favorably with these studies in terms of symptom control, even though direct comparisons are not possible. A number of additional observations regarding care outcomes can be made. From PCU admission to 48 hours before death, physical outcome variables (pain and shortness of breath) got worse in all patients. These differences are not significant in Group 1, most likely because all patients in this group were very close to death, and, therefore, a ceiling effect may have caused the absence of change. Worsening of pain and shortness of breath from admission to one week before death was visible only in patients in Group 2. Apparently, patients who die within one week and who die after two weeks are better off in terms of symptom change than patients who die within two weeks. Affect and nausea show favorable changes; nausea and depression are low; apparently, pain and nausea are the most difficult symptoms to control. Limitations Proxy Measures. The most salient limitation of this study is the use of proxy data instead of self-report data. Proxy and self-report data tend to agree relatively well when the information is concrete and easily observable, but in some cases, systematic biases exist. In a comprehensive review, McPherson and Addington-Hall 15 found that agreement between patients and proxies is high in, for example, shortness of breath, but relatively poor in pain, depression, and anxiety. Moreover, proxies view pain as more severe and frequent than patients, and tend to overestimate the severity of depression, anxiety, and physical symptoms. More specifically, Nekolaichuk et al. 16 found that on average, physician symptom ratings were lower than patient ratings, but nurse ratings agreed more closely with patient ratings. Klinkenberg et al. 17 found that at a group level, retrospective assessments of symptoms one week before death by significant others corresponded with physician s symptoms assessments. However, at an individual level, significant other and physician ratings did not correspond well. These studies suggest that the absolute levels of outcomes found in the current study cannot be compared with self-report data reported in other studies. However, there is no reason to assume that the changes in outcomes found in this study are not valid and comparable to selfreported changes. Another disadvantage resulting from the proxy rating approach was the large number of missing data. PCU staff members were instructed to rate the patients, unless they felt the rating could not be performed reliably, for example, when communication with the patients is limited. It appeared to be especially difficult to rate patient depression scores. More missing values were assigned at PCU admission than one week before death, probably because it takes some time to get acquainted with patients in order to reliably judge their abilities and symptom levels. In addition, the results indicate that the number of missing values was partly dependent on the level of patient functioning: patients who needed more toilet use at PCU admission were assigned more missing data. This result may indicate a slight underestimation of depression and anxiety scores at PCU admission due to auto selection caused by patient functioning. It should be mentioned that an important advantage of collecting proxy data is the collection of data on all patients admitted to the units within a certain time interval, thus providing a complete assessment of patients admitted to PCUs. In addition, collecting self-report data leads to auto selection in patients at the end of their lives, because many patients cannot be interviewed due to symptoms or limited consciousness.

9 Vol. 28 No. 3 September 2004 Palliative Care Units in The Netherlands 241 Selection Bias. Two forms of selection bias may have influenced the results in this study: 1. Bias by unequal distribution of PCU data collection duration. As can be seen in Table 1, the contribution duration of the PCUs varied from 4 to 16 months. No apparent relationships between contribution duration and urban/rural location and number of beds are visible. The two PCUs with separate housing and 24/7 staff both had long data collection contribution times, reflecting the flexibility benefits of a 24/ 7 team. We found no indication for the relationship between bias by unequal distribution of data collection duration and care outcome measures, namely, symptom levels. 2. Bias by selection of respondents. Limited retrospective data of 46 patients were collected. The reason for this, among other things, is that these patients required highly complex care. This form of selection bias may have lead to slight underestimation of mean symptom levels, which is an additional reason to pay more attention to symptom changes than symptom levels. Instruments. Validation research on the measures was not performed in the Dutch language. Although back translation procedures were performed, future comparisons with Dutch language validated versions of the measures used in this study may not be valid. In addition, caution is needed when interpreting the informal caretaker measure, which includes counts of people who visit patients regularly or act as formal representatives of the patients. The measure does not guarantee that the informal caretakers actually have good relationships with the patients. Informal conversations with staff members confirm that a number of patients are regularly visited by family members for business reasons only. Conclusions and Recommendations Given the differences in patterns of symptom change between Group 2 and Group 3, it is advised to admit terminally ill patients in earlier phases of their disease than what was found in the current study. Travis et al. 18 found that movement into palliative care of residents in long-term care facilities in the U.S. is often delayed too long. The admission duration of the patients in the current study was relatively short (the median was 14 days), which could be indicative of delay. As a comparison, the median survival time in Italian PCUs with comparable care goals was 37.9 days. 13 Informal information from PCU staff members consistently indicated that many patients and their caretakers regret that they were referred to the PCUs only in a late stage of the patient s disease. Unless the patients explicitly indicate their preference for domestic care, it can be advised to pursue admitting patients in earlier stages in order to achieve longer admission durations. These decisions may be supported by symptom change data in the last, for example eight, weeks of life. This study did not show these data, because there were insufficient patients with long survival times to form groups. Longer lasting future studies may produce such data. Independent, small-scale PCUs with specialized staff appear to be a good concept of care for the terminally ill in terms of patient outcomes at the end of life. Starting a PCU using the requirements drafted for the PCUs in the current study 4 can be recommended, but the requirements still need to be prioritized, as was recommended elsewhere. 5 Therefore, it is advised to learn from the experience of already established PCUs when starting a PCU. In order to evaluate the PCUs further, it is also advised to investigate which specific care acts lead to favorable end-of-life outcomes. Specific care acts may be monitored by asking caregivers to write short reports in a log each time these acts are performed. Lastly, it is advised that instruments are used in clinical practice for monitoring patient outcomes, which may be used for care planning. Ellershaw et al. 19 demonstrated the benefits of implementing a method of measuring patient outcomes for care planning. The ESAS 10 and the Palliative Care Outcome Scale 20 have been mentioned as useful tools in clinical practice, but there are concerns about the total symptom scores generated by the ESAS at the end of life. 21 References 1. Carney MT, Meier DE. Palliative care and endof-life issues. Anesthesiol Clin North America 2000; 18:

10 242 Echteld et al. Vol. 28 No. 3 September Francke AL. Palliative care for terminally ill patients in The Netherlands. The Hague, The Netherlands: Ministry of Health, Welfare and Sport, Netwerk Palliatieve Zorg voor Terminale Patiënten Nederland [Dutch Palliative Care Network for Terminally Ill Patients]. Accessed 3/03; www. palliatief.nl. 4. Bokhoven RFM, Ooms ME, Ribbe MW, et al. Een plan voor de opzet van units voor kortdurende terminale zorg in bij een verpleeghuis. Tijdschrift voor Verpleeghuisgeneeskunde 2001;1: Echteld MA, Deliens L, Ooms ME, et al. Bevorderende en belemmerende factoren bij de implementatie van units voor kortdurende terminale zorg in verpleeghuizen [Facilitators and barriers in implementing units for short-term terminal care in nursing homes]. Tijdschrift voor Gerontologie en Geriatrie, 2004;35: Ng K, von Gunten CF. Symptoms and attitudes of 100 consecutive patients admitted to an acute hospice/palliative care unit. J Pain Symptom Manage 1998;16: Lo RSK, Woo J, Zhoc KCH, et al. Quality of life of palliative care patients in the last two weeks of life. J Pain Symptom Manage 2002;24: Lloyd-Williams M, Riddleston H. The stability of depression scores in patients who are receiving palliative care. J Pain Symptom Manage 2002;24: Cohen SR, Boston P, Mount BM, et al. Changes in quality of life following admission to palliative care units. Palliat Med 2001;15: Bruera E, Kuehn N, Miller MJ, Slemser P, Mac- Millan K. The Edmonton Symptom Assessment System (ESAS): a simple method for the assessment of palliative care patients. J Palliat Care 1991;7: Chang VT, Hwang SS, Feuerman M. Validation of the Edmonton Symptom Assessment Scale. Cancer 2000;88: Steel K, Ljunggren G, Topinkova E, et al. The RAI-PC: An assessment instrument for palliative care in all settings. Am J Hosp Palliat Care 2003;20: Paci E, Miccinesi G, Toscani F, et al. Quality of life assessment and outcome of palliative care. J Pain Symptom Manage 2001;21: Statistical Package for the Social Sciences Version SPSS Inc; 2001 Nov McPherson CJ, Addington-Hall JM. Judging the quality of care at the end of life: can proxies provide reliable information? Soc Sci Med. 2003;56: Nekolaichuk CL, Bruera E, Spachynski K, et al. A comparison of patient and proxy symptom assessments in advanced cancer patients. Palliat Med 1999; 13: Klinkenberg M, Smit JH, Deeg DJH, et al. Proxy reporting in after-death interviews: the use of proxy respondents in retrospective assessment of chronic diseases and symptom burden in the terminal phase of life. Palliat Med 2003;17: Travis SS, Loving G, McClanahan L, et al. Hospitalization patterns and palliation in the last year of life among residents in long-term care. Gerontologist 2001;41: Ellershaw J, Smith C, Overill S, et al. Care of the dying: setting standards for symptom control in the last 48 hours of life. J Pain Symptom Manage 2001;21: Hearn J, Higginson IJ, Palliative Care Core Audit Project Advisory Group. Development and validation of a core outcome measure for palliative care: The Palliative Care Outcome Scale. Quality in Health Care 1999;8: Rees E, Hardy J, Ling J, et al. The use of the Edmonton Symptom Assessment Scale (ESAS) within a palliative care unit in the UK. Palliat Med 1998; 12:75 82.

11 Vol. 28 No. 3 September 2004 Palliative Care Units in The Netherlands 243 Appendix Requirements for Design of the Palliative Care Unit (PCU), Organization of Care, Complement, Team Functioning, and Staff Expertise A. Requirements for the design of the PCU and organization of care. 1. The unit should have at least four beds, exclusively used for terminal palliative care. 2. The unit should be a small-scale facility in or near a nursing home, physically separated from other wards. 3. The unit should have single rooms that may be furnished and decorated in accordance with the patients wishes. 4. All rooms in the unit should be equipped with a sink, toilet and shower. 5. In or near the unit there should be lodging possibilities for relatives. 6. The unit should have a living room, exclusively for patients in the unit. 7. No limitations should exist for the deployment of high-tech materials needed for terminal palliative care. All usual palliative care services should be available in the unit. 8. Written formal agreements between the unit and external care facilities should exist regarding admission policy, indication, support for medical and nursing care, and consultation. B. Requirements for complement of staff and team functioning. 9. Requirements for complement of medical staff: fte s per bed; registered nurses: 0.02 fte s per bed; licensed practical nurses: 0.97 full-time equivalents (fte) per bed; psychosocial personnel: 0.12 fte s per bed; paramedical personnel: fte s per bed. 10. Unit personnel should form a recognizable team for the colleagues in the nursing home. The unit should organize multidisciplinary support meetings, exclusively for the unit staff. At start of the unit, attention should be given to team building. The unit should provide structural support and guidance for the staff. C. Requirements for staff expertise and training. 11. The unit should focus its attention on coping with the emotional distress patients, their caregivers and the staff may have. In addition, support should be provided for colleagues in coping with their own emotions and emotions of patients and their caregivers. The protocol on quality in terminal palliative care in nursing homes should be used in the unit. 12. Unit personnel should receive training in terminal palliative care. The training should entail courses, internships, and formalized peer learning. In addition, five percent of all staff expenses should be allocated for staff training. 13. Unit staff should structurally provide care for the next of kin from admission of the patient until death, including the period of bereavement. 14. Staff members should be sufficiently proficient in operating materials for terminal palliative care.