|HOME||HELP||FEEDBACK||BROWSE ARTICLES||BROWSE BY AUTHOR|
clinmed/2000030003v1 (April 6, 2000)
Contact author(s) for copyright information
Dr L.Vitetta, Senior Lecturer and Co-Director of Research,
Graduate School of Integrative Medicine,
PO Box 218
Hawthorn Victoria Australia 3122.
Dr L Vitetta, Senior Research Associate,
Centre for Palliative Care, Caritas Christi Hospice.
Dr D Kenner, Palliative Care Physician,
Centre for Palliative Care, Caritas Christi Hospice and
St Vincent’s Hospital Melbourne.
The management of infections is a critical determinant of both the survival and the quality of life of patients with terminal illness1,2. It is thought that the majority of these patients will eventually succumb to infections1. The World Health Organization defines palliative care as the active care of patients whose disease is not responsive to curative treatment3. Within this framework, terminally ill patients in their final phase of care are prone to infections, which may be due to factors that have disease specific and or therapeutic induced correlates2-4. Cancer patients undergoing cytotoxic therapy often develop neutropenia and fever, and similarly, during the palliative phase of terminal illness, such patients may develop infections that have significant associations with morbidity and mortality4-6.
Principles of antibiotic therapy in neutropenic cancer patients undergoing chemotherapy have evolved over the last 30 years, largely from studies involving patients presenting with haematological malignancy7. Studies on palliative patients with solid tumors and the patterns of infection and antibiotic use are minimal8-15. In this descriptive retrospective study, we have assessed the site-specific frequencies of infection, the pattern of antibiotic use, symptom response to antibiotic therapy, and investigated the possible relationship between patient symptom profiles and subsequent infection.
This retrospective review of infections is part of a broader survey, which investigated the demography, symptom characteristics and treatment profiles of a cohort of terminally ill hospice patients during their final phase of care. At the time of this study, Caritas Christi was a 72-bed hospice with 52 dedicated palliative care beds and 20 rehabilitation beds. The hospice medical staff consisted of a full time university appointee to the professorial chair in palliative care (psychiatry and psycho-oncology), 0.8 EFT (equivalent full time) specialist palliative care physicians, two medical registrars, one of whom was a specialist physician trainee, and three junior medical officers. During the 1997-1998 period, there were 618 deaths recorded of patients who required symptom palliation for terminal illness16. Patients in this study had been admitted for palliation of symptoms for terminal malignant illness (92.2%), and non-malignant illness symptoms (7.8%) from May 1997 to October 1998. This study comprised 102 consecutive adult patients who died after admission and consisted of 50 males and 52 females.
From the medical files of the patients we retrospectively obtained information on patient demography, consisting of gender, age and patient survival details. Clinical details of disease diagnosis, symptoms on admission, co-morbid conditions, and infections present were noted, together with, a record of terminal restlessness, family support, the presence of a primary carer, family conflict and a comfortable death. The presence of infection was identified by a medical note in the file detailing the site specific diagnosis, presence of suggestive symptoms and signs, and/or a positive bacteriological culture isolate. Documentary evidence of the presence of infection in the patient record preceded all antibiotic prescriptions.
We have also investigated the frequency of de novo symptoms and catastrophic events from patients diagnosed with infection. De Novo symptoms were those, that developed during inpatient stay, not present on admission. Catastrophic events were defined as the unforeseen events that led to an untimely death during the progress phase of terminal care.
Patient files were examined concurrently by two of the authors (DK, LV). Data was evaluated from the clinical death summary, documenting major and minor palliative problems on admission, history of illness, management issues and procedures. Further, progress notes including consultation requests and replies, drug charts, and detailed nursing care plan and summary charts were reviewed. These latter summaries provided progressive information on patient behaviour and cognition, and recorded the presence of anxiety, restlessness, asthenia, dyspnea, mobility, anorexia and cachexia, nausea and vomiting, as well as the presence and nature of family support. Patients were treated according to written guidelines regarding the management of common infections in hospital patients22. Microbiological laboratory tests were readily available via St Vincent’s Hospital Melbourne, the acute teaching hospital affiliated with the University of Melbourne, and to which the hospice is associated. Antibiotic response was gauged from the nursing and medical notes, by noting documented amelioration of symptoms and completion of the antibiotic course, and/or infection related death.
As there was significant admission pressure on availability of our hospice beds and subsequent length of stay, we assessed the contribution of co-morbid conditions on overall morbidity and mortality due to the primary diagnosis, defined by the Charlson index17,18. This was based on pre-existing patient co-morbid disease states identified on admission, or those, that had recently developed and were detected during this phase of care. A value was calculated for each patient. The Charlson index is a score assigned to patients based on the presence of certain diseases with assigned values, with theoretical scores that range from 0 to 37. An adjusted score was calculated that excluded the primary diagnosis on admission, thus controlling for the primary diagnosis18.
Mean (sd) years for patient ages and median days of survival with interquartile ranges (25%-75%) were calculated when necessary. Confidence intervals (95%) were calculated for diagnoses on admission. Mann-Whitney U-test was used to compare groups with and without an outcome of interest, for ordinal (eg. prominent symptoms) or continuous (eg Charlson co-morbid scores) data that were not normally distributed25.
Patients in this cohort consisted of 102 terminally ill adults admitted to Caritas Christi for palliative care (Table 1). Thirty-seven patients (36.3%) of the 102 patients reviewed were diagnosed with infection, and in 13 of these patients, a positive culture isolate was obtained. In 35 patients, each episode of infection resulted in either a single or multiple antibiotic prescription. In 2 patients with infections on admission, no antibiotics were prescribed, as survival was limited to that day. Median patient survival in the total patient cohort was 12 days (IQR: 4-29). In patients with infection, the median survival was 22 days (IQR: 18-33) (P= .013). Infections were diagnosed after a median of 8.5 days. Infections were diagnosed within 72 hours from the time of admission in 25/37 (67.7%) patients. In 8/25 (32%) of these patients the infection had been identified prior to admission.
Two-thirds of the patients who developed an infection, irrespective of its locus, had 2 or more co-morbid conditions diagnosed on admission. Twelve patients had none (7) or one (5) associated co-morbid condition. Of all the major co-morbid condition categories reported on admission, the most commonly reported from patients with infections, had metabolic [17/38 (44.7%)], cardiovascular [4/35 (11.4%)] and pulmonary [7/17 (41.2%)] correlates. The most common co-morbid conditions recorded in this group of patients with infections, were non-insulin dependent diabetes mellitus [NIDDM] (8/12), chronic obstructive airways disease and or asthma (9/17), and a number of cardiovascular conditions (12/25). Of the 5 patients with skin infections, all were diagnosed with diabetes mellitus. The mean (sd) adjusted Charlson co-morbid score for patients with infections was higher [5.0 (2.9)] than for patients without infections [4.6 (3.3)]. The trend however was not significant (P.05) (Table 5). An increase in survival was associated with the development of infections (P = .013).
The most frequent site of infection was the urinary tract (41.0%), and the distribution of infections by site is presented in Table 3. In 8 patients infections were suspected on clinical grounds. The sites of these infections were the eyes (4), lungs (4) and the urinary tract (5). Thirteen individual specimen samples obtained from 13 patients yielded 20 positive bacterial cultures. These samples comprised 12 mid-stream urines, 3 blood cultures, 4 skin swabs and 1 sputum sample. Escherichia coli was the predominant organism cultured (36.8%). Other organisms cultured included Enterobacter fecalis (15.8%) Staphylococcus aureus (10.5%), Pseudomonas aeruginosa (10.5%) Streptococcus bovis (5.3%) Enterobacter aerogenes (5.3%), Enterococcus species (5.3%), Streptococcus pyogenes group A (5.3%) and Proteus mirabilis (5.3%) (Table 4). Urinary tract infections were associated with 5 patients with indwelling urinary catheters.
From the medical, nursing and drug chart notes, symptom amelioration and antibiotic course completion was observed in a minimum of 15 (40%) patients with infections. However, incomplete medical and nursing notes and death of patients during antibiotic therapy precluded any further estimate of antibiotic efficacy in this series of patients. The antibiotics prescribed that were associated with a positive symptom response included first line cotrimoxazole/trimethoprim, which had an overall success rate for symptom relief of 40%. The success to failure rate observed in these patients was: urinary tract infections [4/10 (40%)], respiratory tract infections [2/4 (50%)] and subcutaneous skin infection [0/1 (0%)]. First line use of cephalexin for urinary tract infections had a success to failure rate of 50% (2/4) (Table 3).
De novo symptomatology was recorded in only 42 (41.2%) patients26. Infections comprised approximately 20% of the de novo symptoms recorded. These were diagnosed in 10 patients and consisted of septicemia in 5, urinary tract infections in 4, and an abscess of the forearm in 1 other. Catastrophic events that lead to sudden death were observed in 11 patients, of which 10 patients had been diagnosed with malignant diseases26. Infections comprised approximately 27% of these events. These were observed in 2 patients with septicemia and a further 2 patients with pneumonia.
Patients in the final phase of care were poly-symptomatic. When the patients were divided between those with and without infections no significant differences were observed in adjusted Charlson co-morbidity scores, severe pain, dyspnea, immobility, depression and/or anxiety, confusion, weakness, loss of appetite and nausea and/or vomiting (Table 5). Further there was no significant difference between the sub-groups in terms of primary carers, family support, family conflict or a comfortable death. The significant differences were observed in median survival and major psychological distress (P<.05) (Table 5).
Although the rate of infections in this retrospective cohort was not as high as that in previously published reports10, our results confirm that terminally ill patients are prone to secondary infections during their final phase of care. In this study we observed that multiple factors were responsible for the increased susceptibility to infection. These included asthenia, impaired cognition, immobility, and the use of various foreign bodies (eg urinary catheters) in the care of these terminally ill patients. Fifteen patients (40%) recovered from their diagnosed infections, and infection was not the direct cause of death in this sub-group.
Data from both acute and long term care settings (including nursing homes), have consistently indicated that urinary tract infections are those most frequently encountered, and are responsible for some 35% to 45% of all hospital acquired infections19-21. Pneumonias account for approximately 20% of all nosocomial infections and specific surgical site infections for around 24%. In this study cohort, urinary tract infections were the most common and respiratory infections the second most common. Septicaemia was diagnosed in only 5 patients, and in 4 of these cases, another source of infection was identified, and was presumably the primary septic focus. Nevertheless, this figure may be an underestimate, as generalized sepsis may be difficult to diagnose and monitor in terminally ill patients, and blood cultures are not routinely requested in this care setting. The pattern of bacterial pathogens was similar to other studies whether originating from a palliative care unit or hospital based studies. Escherichia coli was the frequent bacterial species isolated, and strains were predominantly ampicillin / amoxycillin and gentamicin sensitive. Moreover, the trends of the distribution of pathogens by site was also similar to other studies previously reported22, 24. Vancomycin resistant enterococcus species (VRE) was isolated from one patient. There were no isolates of methicillin resistant staphylococcus aureus (MRSA) in this patient cohort.
This study demonstrated that approximately one third of terminally ill hospice patients develop infections during their final phase of care. Urinary and respiratory sepsis predominated, and a thorough examination of the patient files showed that an antibiotic response was obtained in at least 40% of patients. There are no comparable published data on antibiotic response rates in hospice patients. Given the inherent limitations of antibiotic therapy in a terminally ill patient population with advanced disease and prevalent co-morbidity, the apparent response rate in this study was perhaps not unexpected. It is difficult to give conclusive recommendations regarding guidelines for antibiotic therapy in hospice patients based on this study, with the small numbers involved and its retrospective nature. However, our results suggest that, for first-line therapy in cases of urinary sepsis, trimethoprim, cephalexin and augmentin forte are effective agents. Cotrimoxazole and augmentin forte appeared to be effective agents for respiratory infections.
The decision whether to treat an infection or not in the palliative care setting may be complex, and requires an individualised clinical approach. Despite the fact that a minimum of 40% of patients prescribed antibiotics appeared to respond clinically in this study, the subsequent course in these patients lead to a comfortable death in almost 90% of those treated. Although there are obvious limitations inherent in a retrospective study of this nature, the criteria used for defining a response to antibiotic therapy are highly dependent on the presence of adequate medical documentation of symptomatic improvement in the patient file. We found this particularly difficult in this retrospective study, given also that the patients had multiple symptoms. However, significant symptom amelioration documentation and drug charts review that showed that at least 15 patients had successfully completed an antibiotic course and that death in these patients had not been due to bacterial infections, were strong indicators for positive antibiotic responses.
Patients with documented infection had a longer median survival in our study. The most plausible interpretation of this data is that the probability of infection increases with duration of survival, given the similar prevalence of cachexia and immobility. Similarly, the increased psychological distress observed in these patients with infection may be the result of a more prolonged terminal phase, rather than an association with infection per se. In this case, psychotropic drug therapy and psycho-social interventions by pastoral care and other workers may have had a major impact on the outcome of a comfortable death26,27. Nevertheless, we cannot exclude the possibility that antibiotic therapy also contributed to this outcome.
We noted with interest that a third of the patients who experienced a catastrophic event that led to an unexpected death, did so through infections that comprised septicemia in two patients and pneumonia in a further two patients. All of these patients were immobile and on oxygen therapy26, a future area of research may investigate the effect of immobility, oxygen therapy and the supine position as risks for such infections. Nevertheless, we conclude that antibiotic therapy may be beneficial in palliative symptom control in terminally ill patients. Further studies carefully documenting both patient symptoms and the individualised approach to decision making regarding the use or withholding of antibiotic therapy in palliative patients are recommended.
|HOME||HELP||FEEDBACK||BROWSE ARTICLES||BROWSE BY AUTHOR|