Brandt HE, Deliens L, Ooms ME, van der Steen JT, van der Wal G, Ribbe MW. Symptoms, Signs, Problems, and Diseases of Terminally Ill Nursing Home PatientsA Nationwide Observational Study in the Netherlands. Arch Intern Med. 2005;165(3):314-320. doi:10.1001/archinte.165.3.314
Nursing homes (NHs) are less well studied than hospices or hospitals as a setting for terminal care. For more targeted palliative care, more information is needed about the patient characteristics, symptoms, direct causes and underlying diseases, and incidence of terminally ill NH patients. These aspects are examined in this study.
Prospective observational cohort study in 16 NHs representative of the Netherlands. All long-term care patients assessed by an NH physician to have a life expectancy of 6 weeks or less were enrolled in our study.
The terminal disease phase was marked with symptoms of low fluid and food intake, general weakness, and respiratory problems or dyspnea. Direct causes of these conditions were diseases of the respiratory system (mainly pneumonia) and general disorders (eg, cachexia). The 2 main underlying diseases of the terminal phase were mental and behavioral disorders and diseases of the circulatory system. Cancer was the underlying disease in only 12% of the patients. Patients with cancer showed a different pattern of symptoms than those without cancer. Per 100 beds per year, 34 NH patients entered the terminal phase. Most patients (82.9%) died within 7 days of inclusion.
For patients without cancer in Dutch NHs, the terminal disease phase is difficult to predict, and once diagnosed, patient survival time is short. A better identification of the symptom burden might improve the prognostication of life expectancy in elderly patients.
The number of people living to an old age in Western countries is rapidly increasing. Consequently, the pattern of diseases that cause the most morbidity and mortality is also changing.1 In the near future, the nursing home (NH) will increasingly be the site of terminal care and the place of death for many elderly people and people with dementia.2- 5 Before death, NH patients enter a terminal phase in which optimal palliative care should be provided. Traditionally, palliative care has focused on patients with cancer, but this type of care is needed for a much wider range of terminal illnesses and should be integrated more broadly across the health care services, as was recently stated by the World Health Organization.1
Nursing homes have been less well studied than hospices or hospitals as a setting for terminal care.6 We do not know whether NHs provide good care for the dying or how comfortably people die in a NH. Moreover, little is known about the symptoms and disorders that are associated with the terminal phase of these patients’ lives. To our knowledge, there are no reports on the incidence of NH patients who require palliative terminal care. Knowledge about the number of patients who will enter the terminal phase is required for the planning and organization of palliative terminal care. As the role of NHs in providing end-of-life care continues to expand, these facilities will be challenged to meet residents’ and families’ expectations and to achieve adequate outcomes of care.7 Knowledge of the symptoms and disorders that are associated with the terminal phase of NH patients’ lives will help to improve the quality of dying. Physicians who, for example, do not realize how little time is left for the patient may miss the chance to devote more effort to improve the quality of the patient’s life in the remaining period.8 This study examined patient characteristics, symptoms, direct causes and underlying diseases, and incidence of the terminal phase in terminally ill patients in the Netherlands. In Dutch NHs, curative, supportive, or palliative care is provided for all patients according to the course of their illness.
A prospective observational study was conducted in 16 Dutch NHs during a 16-month period from November 2001 until March 2003. Unlike most other countries, the Netherlands employs specially trained physicians to provide medical care in NHs.9 Three types of NHs can be distinguished: those for physically ill patients, those for psychogeriatric patients, and combined types with separate wards for each category. Both the physically ill and the psychogeriatric patients benefit from the different nursing home functions, mainly rehabilitation and long-term care. This type of care can be characterized as continuous, long-term, systematic, and multidisciplinary.10 Most (96%) of the psychogeriatric patients have dementia and will stay in the NH until they die.11 In 2000, there were 334 NHs in the Netherlands with a total of 58 778 beds, which accounted for approximately 18% of the total mortality of Dutch inhabitants.12,13
The NHs in the present study were purposefully sampled to be representative of all Dutch NHs. First, the proportion of the 3 types of NHs was calculated.14 This proportion was 3:1:1 in order of combined, those for solely physically ill patients, and those for solely psychogeriatric patients. Therefore, the aim was to include 9 combined NHs, 3 NHs for physically ill patients, and 3 NHs for psychogeriatric patients. Second, the mean number of beds for physically ill and psychogeriatric patients per NH was calculated, and the NHs were arranged in order of this calculated mean. Finally, the NHs closest above and below the mean were invited to participate in the study.
The beginning and end dates of the data collection varied per NH. Each participating NH included all patients as soon as the estimated life expectancy became 6 weeks or less; follow-up was until death, with a maximum of 12 weeks. Each patient included was given a unique identification number by the local research coordinator, who supervised the data collection.
The treating physician was responsible for the inclusion of patients in the study if they met the following criteria: (1) remaining life expectancy of 6 weeks or less; (2) admitted for long-term care; and (3) admitted for rehabilitation, but during their admittance it became obvious that the patient would not leave the NH. The exclusion criteria were (1) admitted for rehabilitation and expected to be discharged from the NH and (2) sudden and unexpected death (ie, with no clear terminal palliative phase).
The period of 6 weeks is based on prior consultation with NH physicians who participated in our study. Admittance to NHs for terminal care is handled differently in different countries. In the Netherlands, it is set at 3 months because of administrative reasons, such as refunding by health care insurance companies.
For practical reasons, the NH physicians perceived a weekly follow-up registration of 3 months with a maximum follow-up of 6 months as not feasible. We agreed on a weekly follow-up registration of 6 weeks with a maximum follow-up of 12 weeks (3 months). On the basis of their clinical experience, NH physicians decided whether a patient had a life expectancy of 6 weeks or less.
On inclusion, the NH physician completed a questionnaire on basic demographics, illness characteristics (eg, stage of dementia), symptoms, direct cause, and underlying disease of the terminal phase. The stage of dementia was assessed according to the validated Global Deterioration Scale.15 This instrument identifies 3 major clinical phases: forgetfulness phase, confusional phase, and late dementia phase. These phases are further subdivided into 7 clinically identifiable stages, ranging from stage 1 (no cognitive decline) to stage 7 (very severe cognitive decline). Stage 5 represents the phase of early dementia, and patients in this stage can no longer survive without some assistance.
Symptoms were measured according to a list of 25 symptoms (with 1 additional open question). This list was developed in cooperation with experienced NH physicians, who on the basis of their clinical practice included the main common symptoms present in NH patients. On inclusion, the NH physicians were also asked to give a maximum of 3 symptoms in order of magnitude, on which they based their estimated life expectancy of 6 weeks or less for that particular patient (1 indicates most important; 2, second most important; 3, third most important).
The validated Dutch Classification Codes of Diseases for Nursing Home Medicine (CvZ-V)16 were used for registration of the direct cause of the terminal disease phase (disorder that directly caused the symptoms on which the NH physician based an estimated life expectancy of 6 weeks or less) and the underlying disease (disease that was the underlying cause of the disorder on which the NH physician based the limited life expectancy). The CvZ-V is a Dutch standard, based on the International Classification of Diseases, 10th Edition (ICD-10),17 which gives a classification of diseases, intercurrent diseases, disorders, injuries, other problems, and causes of death in NH patients. However, unlike the ICD-10, it contains a general section with codes for diseases, disorders, and problems that supersede one particular organ system.
We aimed to check all eligible patients by screening all deaths that occurred in the participating NHs. At the time of patient death, the NH physicians completed a questionnaire for all long-term care patients, including patients who had not been included in the study, for whom a short questionnaire was completed. In this short questionnaire, we checked the reason for noninclusion: (1) “a sudden and unexpected death (no or very short terminal phase)”; (2) “the patient should have been included earlier”; or (3) “other, namely. . . . ” The patients included under point 2 were “missed” patients who were identified afterward. For these patients, the NH physician was asked in a separate question why these patients were not included earlier. The alternatives could be (1) “forgotten (to be included),” (2) “pressure of work,” (3) “organizational reasons (eg, should have been included during the weekend),” (4) “family had objections for inclusion in our study,” and (5) “other, namely. . . . ”
We went to great lengths to ensure that all eligible patients were included or at least identified. A check was made per NH to ensure that all deaths of long-term care patients were registered. At the end of the data collection, the local research coordinator of each NH was asked in a personal interview for a number or estimate of patients incorrectly omitted. This number included the missed patients who were identified afterward.
All analyses were performed with SPSS statistical software, version 10.1 (SAS Institute Inc, Cary, NC) for Windows. The Pearson χ2 and Fisher exact tests were used to detect statistically significant differences (P<.05) in categorical data between groups.
The Fisher exact test was chosen when the number in the cells was small or zero. The incidence was calculated from 16 selected NHs, most of which participated from January 2002 to March 2003. For the numerator, we counted all patients who entered the terminal phase, whether included or missed and irrespective of outcome (ie, died or survived within the maximum duration of follow-up of 12 weeks, including second inclusions). The denominator was calculated separately for each participating NH, and the results were added together. The total number of beds was multiplied by the mean bed occupancy of the home in 2002 (if the number of beds changed during the inclusion period, the average was calculated) and the inclusion period. The length of the inclusion period varied among the NHs. Because of this, we corrected data to the 1-year period.
The study was approved by the Medical Ethics Committee of the VU University Medical Center. Confidentiality of the data was guaranteed by providing coded information only to the researcher. Informed consent was not necessary, because the physicians and nurses were simply reporting information collected as part of the usual care provided. As required by Dutch law, the patients and their families were informed about the study through an informational flyer and were given the option of refusing the transfer of data.
A total of 544 patients were included. For 516 of these patients, NH physicians completed questionnaires on inclusion, which were used for analysis. An additional 272 patients were found to have been incorrectly omitted from the study. These patients were considered to be missed patients who were identified afterward. The main reasons for noninclusion, given in the questionnaire in order of sequence by the NH physicians, were pressure of work, organizational reasons (eg, should have been included during the weekend), forgotten (to be included), other reasons, and family had objections for inclusion in our study.
Representative sampling in this study was checked against 2 parameters: the distribution of physically ill and psychogeriatric patients and sex. No significant differences were found between the national NH population and the study population for either parameter.
The incidence was calculated as follows:
Included Patients + Missed Patients (Identified Afterward)/Beds × Mean Bed Occupancy × Mean Inclusion Period.
Taking into account the duration of follow-up during the 1-year period, we identified 544 patients (474 first inclusions, 12 second inclusions, 57 exclusions, 1 at end of follow-up) plus 272 missed patients, which divided by the number of beds (2429.6) equaled 0.34 long-term care patients who met the inclusion criteria. This is equivalent to 34 patients per year in an NH with 100 beds.
After inclusion, most patients (82.9%) died within 7 days, and 92.3% died within 14 days. The median duration of survival was 3 days, with 10th and 90th percentiles of 0 (death within 24 hours) and 12 days, respectively. Fifty-seven (11.0%) of the 516 patients recovered and were excluded. Of these patients, 26 (46%) of 57 died later during our study (21.0% after the second inclusion and 24.6% between 5 and 228 days after exclusion).
Table 1 presents the demographics and disease characteristics of patients on inclusion in the study. Most of the patients were female (70.9%) and mostly widowed (59.6%), with a mean ± SD age of 83.5 years ± 8.1 years. Patients were predominantly admitted to the NH for chronic care (75.6%), and most (56.4%) stayed in a psychogeriatric ward. Of all the patients included, 33.5% had no cognitive decline, 5.2% had very mild to moderate cognitive decline, 8.1% had moderately severe cognitive decline, 27.0% had severe cognitive decline, and 26.0% had very severe cognitive decline. Possible selection bias was evaluated by testing differences between included patients and patients who were found to be incorrectly omitted from the study for 3 parameters: age, sex, and ward. No significant differences were found for sex and ward between the included and omitted patients. For age, a significant difference was found (mean age, 83.5 years for included patients vs 81.9 years for the omitted ones; P = .04). Since the mean age of both groups differed by 1.6 years, we considered this difference acceptable.
Table 2 gives the most important symptoms, signs, or problems on study inclusion, in order of magnitude, perceived by the NH physician as indicative of a limited life expectancy. In 42.6% of all cases, the problem of “(very) little/no fluid intake” was reported as a reason for the NH physician to estimate the life expectancy to be 6 weeks or less. “Generalized weakness” was reported in 31.8% of the cases, “(very) little/no nutritional intake” in 24.8%, and “respiratory problems/dyspnea” in 21.3% of the cases. The percentages of other symptoms, signs, or problems were less than 20%.
“(Very) little/no fluid intake” (17.2%) and “generalized weakness” (10.7%) were the symptoms that were most frequently reported as most important. “(Very) little/no fluid intake” (17.2%) and “(very) little/no nutritional intake” (13.6%) were most frequently reported as the second most important symptoms or problems. The third most important symptoms were “generalized weakness” (10.9%) and “(very) little/no nutritional intake” (9.9%).
As given in Table 3, the terminal phase was directly caused by a disease of the respiratory system, predominantly pneumonia, in 24.4% of the study population, “general” disorders (disorders that supersede one particular organ system) in 23.8% of the patients, and diseases of the circulatory system (such as cerebrovascular accidents and heart failure) in 14.0% of the patients. “Endocrine, nutritional, and metabolic disorders” (such as dehydration) were reported in only 9.3% of the cases. With regard to the underlying disease of the terminal phase, the 2 main categories were mental and behavioral disorders (30.2%), almost exclusively dementia, and diseases of the circulatory system (20.5%). The third category was cancer (12.0%).
Table 4 gives reported symptoms on inclusion for the 2 main underlying diseases of the terminal phase (mental and behavioral disorders and diseases of the circulatory system) and malignant neoplasms (including secondary malignant neoplasms). For 9 of the 26 symptoms, significant differences were found between the 2 main underlying diseases. Most frequently reported symptoms in the group of patients with mental disorders (mostly dementia) were problems regarding “fluid intake” (49.4%), “nutritional intake” (32.7%), “generalized weakness” (28.8%), “recurrent fever” (23.7%), “cachexia/anorexia” (17.3%), and “(worsening) pressure ulcer” (16.0%). Patients with diseases of the circulatory system more often had “respiratory problems/dyspnea” (31.1% vs 12.2%). They more often experienced “(extreme) tiredness” (10.4% vs 3.2%), severe stage of somnolence (17.0% vs 6.4%), and “other symptoms” (9.4% vs 3.2%). Included in this group was “medication not successful” (12.3% vs 3.8%), but there were fewer problems with “nutritional intake,” “recurrent fever,” “(worsening) pressure ulcer,” and ”refusal of liquid.”
The frequencies of reported symptoms for patients with cancer show a different pattern when compared with the 2 main underlying diseases (Table 4). Significant differences between patients with cancer and those with mental and behavioral disorders and diseases of the circulatory system were found for 11 and 12 (of 26) symptoms, respectively.
Compared with patients with mental and behavioral disorders, those with cancer more often experienced “generalized weakness” (46.7% vs 28.8%), “other symptoms” (20.0% vs 3.2%), “(extreme) tiredness” (26.7% vs 3.2%), “loss of appetite” (10.0% vs 1.3%), “vomiting” (11.7% vs 1.3%), “nausea” (13.3% vs 1.3%), and “feeling sick” (10.0% vs 0.6%). They less frequently had “(very) little/no fluid intake” (21.7% vs 49.4%), “recurrent fever” (10% vs 23.7%), “(worsening) pressure ulcer” (0.0% vs 16.0%), or “subcoma” (0.0% vs 10.3%).
When compared with patients who had diseases of the circulatory system, patients with cancer more often experienced “generalized weakness” (46.7% vs 28.3%), “cachexia/anorexia” (26.7% vs 10.4%), “(extreme) tiredness” (26.7% vs 10.4%), “vomiting” (11.7% vs 0.9%), “nausea” (13.3% vs 0.9%), and “feeling sick” (10.0% vs 1.9%). They less frequently had “(very) little/no fluid intake” (21.7% vs 45.3%), “respiratory problems/dyspnea” (13.3% vs 31.1%), “difficulty swallowing” (5.0% vs 17.9%), “subcoma” (0.0% vs 9.4%), “severe stage of somnolence” (3.3% vs 17.0%), and “medication not successful” (1.7% vs 12.3%).
To our knowledge, this is the first nationwide study to provide insight into the symptoms, direct causes, and underlying diseases of the terminal disease phase in NH patients. Furthermore, this study estimated the incidence of NH patients who entered the terminal phase. The strengths of this study are the large-scale design, the representativeness of the NHs, and the prospective character, which enhance the validity and reliability of the findings.
However, the study has some limitations. The underlying disease of the terminal phase was assessed with a question analogue to the cause-of-death question on Dutch death certificates.13 The NH physicians who participated in the study might have had difficulties in reporting the underlying disease of the terminal phase and therefore may have incorrectly registered the cause of death on the death certificates.18 Problems occur mainly with regard to the sequence of the primary and secondary causes of death, and the primary cause (underlying disease) cannot always be determined.
There was also missing information about a large number of terminally ill patients. These patients should have been included. Although we did not find substantial differences between the study sample and the omitted group in the demographic variables (age, sex, and ward), we cannot, in case physicians selectively missed patients, exclude a possible bias on the outcome variables of this study. Also, no separate incidence of patients (eg, with and without cognitive impairment) who would enter a terminal disease phase could be calculated.
In addition, although sudden and unexpected death was an exclusion criterion, different physicians may differently read the signs in the phase before death and may miss signs of nearing death. It is possible that physicians missed the signs of nearing death and mistakenly described patients as having died suddenly. However, we believe that physicians did not deliberately misclassify patients. Furthermore, individual interpretation was possible regarding the data on having a limited life expectancy of 6 weeks or less; however, this factor was decided on the basis of clinical daily practice.
The 4 most frequently reported symptoms, signs, or problems in the terminal disease phase of NH patients, which physicians identified as related to terminal illness, were problems with fluid and nutritional intake, generalized weakness, and respiratory problems or dyspnea. The patients were frequently in a state of somnolence and experienced recurrent fever. These symptoms alone might be highly predictive of approaching death and could be an indication for the physician and all others involved that the patient is dying. However, this must be confirmed in future research on predictive models. Since the identified period of terminal illness is limited, the question is whether palliative care is sufficiently addressed.
Direct causes of the terminal phase, according to NH physicians, were diseases of the respiratory system, mostly pneumonia, and disorders of a general nature, such as cachexia, malaise, coma, fever, and septicemia. Diseases of the circulatory system were less frequently reported by NH physicians. Hanson et al6 found that NH deaths are preceded by a slow trajectory punctuated by acute and reversible illnesses, such as pneumonia, sepsis, or dehydration. The results of the present study are in line with this observation. Health care professionals are therefore challenged with providing care for patients with diseases that have trajectories of slow decline with periodic crises and a less well-defined terminal phase.19
Diseases of the circulatory system and mental or behavioral disorders dominate as the underlying diseases in the terminal phase of life of NH patients. However, regardless of the underlying disease, symptoms of “(very) little/no fluid intake,” “generalized weakness,” “somnolence,” and “cachexia/anorexia” were common in both groups. For other symptoms, different patterns could be determined between these 2 main groups. For patients with mental disorders, mainly dementia, the beginning of the terminal phase was marked with problems of nutritional intake, and they experienced recurrent fever. This group was most likely to refuse liquid and to develop a (worsening) pressure ulcer. In these NH patients, it appeared that low food and fluid intake, together with episodes of fever, led to a deterioration in their overall condition, which may have caused (worsening of) pressure ulcers. In patients with disorders of the circulatory system, mainly caused by poor heart function, the presence of respiratory problems or dyspnea can be expected. Extreme tiredness and severe stage of somnolence can be the consequences of a circulatory disease. A low level of oxygen in the blood, due to insufficient heart function, causes tiredness, and the mental state of patients in the severe stage of somnolence can probably be explained by the high prevalence of cerebrovascular accidents (9.3%) in the study population. In this group of patients, medication was significantly less successful than in the group of patients with mental disorders.
The findings of the present study reflect a (mainly) noncancer population. Only 12% of the patients had cancer as the underlying disease. Those patients showed a different pattern of symptom prevalence with regard to the noncancer population. These characteristics are in line with the findings of Addington-Hall et al,20 who reported a difference in the prevalence of symptoms between patients with and without cancer.
It is estimated that in an NH with 100 beds, 34 patients will enter the terminal phase of life each year. In other words, for every 3 beds, an NH can expect 1 patient per year to reach a terminal disease phase that requires palliative care. This information is useful for policy making and assessment of the required terminal care. Although the estimated incidence of the terminal phase is high, the duration of this phase, as noticed by the physicians, is most often limited to a couple of days.
The study demonstrates that most of the patients were considered to be terminally ill only when death was actually already very near. After inclusion most patients died within 7 days, and after 14 days almost all patients were dead. This finding highlights the late identification of terminally ill patients by NH physicians. On the other hand, 1 of 10 patients recovered. Nevertheless, almost half of these patients died later during the inclusion period. Even the advantage that the physicians are employed by the NH and therefore know their patients very well is no guarantee of an accurate prediction of life expectancy. The difficulty of defining patients without cancer as terminally ill or predicting their life expectancy has been recognized in several studies.21- 23 An estimation of the possible survival period is important for the patients, their families, and the NH physicians and nurses and also has implications for the planning of appropriate medical care. Therefore, improving prognostication of life expectancy for patients without cancer is needed.
In conclusion, it is perhaps even more difficult to provide appropriate palliative care for elderly patients with chronic diseases than for patients with cancer. The course of chronic diseases is more difficult to predict because it shows a more gradual decline when compared with patients dying of cancer.24
As our study shows, when patients were identified as entering a terminal disease phase, death was only a few days off. This indicates that the traditional focus of palliative terminal care on patients with cancer needs to be extended and adopted to the specific needs of elderly patients without cancer. This public health challenge was recently adopted by the World Health Organization.1
Our study also highlights how ineffective NH physicians are at determining prognosis with regard to life expectancy and poses the question of whether this limited time is adequate for the implementation of palliative terminal care, if this has not already been provided. High-quality symptom control and organization of palliative care in NHs is needed owing to the apparent speed of deterioration and therefore a steep increase in symptoms. The findings of this study stress the difficulty of estimating incidence of palliative care needs of the terminally ill patients in NHs. Further research to estimate these needs is desirable.
Correspondence: Hella E. Brandt, MA, Department of Nursing Home Medicine, Institute for Research in Extramural Medicine, VU University Medical Center, Van der Boechorststraat 7, 1081 BT Amsterdam, the Netherlands (firstname.lastname@example.org).
Accepted for Publication: July 20, 2004.
Acknowledgment: We thank all participating nursing homes for the data collection for this study. We also thank the Ministry of Health, Welfare and Sports, The Hague, the Netherlands, for project funding.
Financial Disclosure: None.