Article Text
Abstract
Objectives The Palliative Performance Scale (PPS) was introduced across NHS Grampian. Our aim was to determine how practical and useful the PPS was for clinicians looking after palliative patients in a variety of settings.
Methods A prospective audit approach was used in primary, secondary and nursing home care settings who. Demographic and assessment data were gathered for 3 months; feedback was gathered at the end of the data collection phase. Patient follow-up status was determined at 12 months.
Results Fifteen clinical sites participated and feedback was obtained from all clinical areas (n=30). Most respondents found the PPS easy to use and that it helped recognise disease progression in cancer patients, but not in patients with dementia/frailty. Assessment data were gathered on 666 patients. Sixty per cent had a malignant diagnosis and 62.5% of the sample died within 12 months. Lower PPS scores at initial assessment indicated poorer prognosis. Median survival figures differed from previously published data. Falling PPS scores increased the risk of death compared with patients whose PPS scores remained static or improved.
Conclusion Clinicians found the PPS to be a quick, useful way of assessing and reviewing functional changes in palliative patients. However, it may not identify the subtle changes in individuals with advanced dementia. The survival figures confirm that caution is needed in generalising survival data across different settings and populations. Further work is needed to examine changing functional status in patients with non-malignant diseases or dementia/frailty.
Statistics from Altmetric.com
Introduction
Palliative care is part of the daily work for almost all healthcare professionals. Patients are likely to spend 90% of their final year of life in a primary care setting,1 and at least 12% of hospital inpatients are in their last 3 months of life.2 Functional decline is a feature of many palliative conditions, but the rate of decline may vary according to the particular diagnosis3 4 and, therefore, it can be difficult to recognise when patients are entering the palliative or terminal phase of their illness. These uncertainties about prognosis can impede appropriate advance care planning.5 6
It has been recognised that there is a need for more work on prognostic indicators that can inform end-of-life care planning.7 A number of functional assessment tools that link changes in functional status with survival have been developed and validated. The Scottish Government published ‘Living and Dying Well: a national action plan for palliative and end of life care’ in 2008. Action Point 2 states that all palliative patients should be assessed and reviewed in all care settings using recognised tools currently available.8 One such tool is the Palliative Performance Scale (PPS). The PPS was developed from the Karnofsky Performance Scale in Canada in 1996 as a way of providing a framework to assess the ‘progressive decline in palliative patients’.9 The PPS is a measure of performance, where an assessment is made of a person's ability to ambulate, undertake self-care, eat and drink independently, level of consciousness and presence of disease, based on which a score ranging from 0 to 100 is allocated. The assessment takes less than 1 min to carry out and the score correlates with survival time.10 The lower the score, the shorter the survival time.
There are reports of the use of the PPS in palliative care settings—palliative care units, hospices or hospital-based palliative consultation services—with the majority of patients having a cancer diagnosis.11,–,19
A meta-analysis exploring the results of studies undertaken in Canada and the USA reported that although the PPS is a strong predictor of survival from initial assessment following referral to specialist palliative care services, caution is needed in applying these findings to other care settings.11 To our knowledge, there are no published data looking at the use of the PPS in hospital and community settings in the UK.
The NHS Grampian Palliative Strategy Group was allocated the task of rolling out the Action Plan across the Grampian region. However, we felt it would be beneficial to understand more about the PPS and its use in practice, prior to full rollout across the region.
This paper describes the introduction of the PPS across a range of primary and secondary care clinical settings in the northeast of Scotland and feedback from the clinicians using it. This is a pragmatic study describing the ‘real-life’ utility of the PPS. It is not a controlled clinical trial.
The aim was to investigate how practical and useful the PPS is for clinicians looking after palliative patients in primary and secondary care settings across northeast Scotland. The specific objectives were:
▶ To gather feedback from clinicians about the ease of use of the PPS in the assessment of palliative patients.
▶ To gather feedback from clinicians about the helpfulness of the PPS in the identification of palliative nursing needs.
▶ To assess the usefulness of PPS as a predictor of mortality.
Method
Approach
A prospective audit approach was used. Clinicians in primary, secondary and nursing home settings who care for palliative patients were eligible to participate. The sample included three general practices (primary care), a specialist palliative care unit, hospital specialist palliative care team and Community Macmillan Nurse referrals, an acute oncology ward, two long-stay dementia wards, three medical wards for the care of older people and three care homes. The settings were chosen to reflect the range of palliative care work undertaken, both in primary and secondary care. Patients were selected for inclusion if the clinician looking after them believed they were ‘palliative’. We deliberately did not enforce more specific inclusion or exclusion criteria as we wanted to assess how the PPS would fit with ‘normal’ clinical practice.
Each area was visited by the Senior Lecturer in Palliative Care and a short tutorial was given to explain the project and how to use the PPS. Clinicians were asked to record an initial PPS score for their palliative patients, then either regularly or whenever they felt there was a change in the patient's condition for 3 months. Each clinical area identified a key contact (nurse or doctor) who forwarded assessment data as well as limited demographic data, which was collated by one of the investigating team (LM).
Clinician feedback
After data collection was completed, feedback was sought from the key contact in each pilot site. The key contact was also invited to seek feedback from other members of their clinical team who had used the PPS. The location and designation of the respondent was recorded. Three questions were included in the feedback form with ‘yes’, ‘no’, ‘unsure’ responses and space for additional comments. The questions were:
▶ 1. Has the PPS been easy to use?
▶ 2. Do you think the PPS has been useful in helping to recognise disease progression in your palliative patients?
▶ 3. Is the PPS helpful in identifying palliative nursing needs?
The feedback data were collated according to the comments made about each of the questions and grouped into recurring themes.
Statistical analysis
Audit data collected included PPS assessment scores, patient location, diagnostic group,4 gender and date of death or status 12 months after the initial PPS assessment. Each participant had at least one PPS assessment with some as many as 13. The patients were grouped into three diagnostic categories (malignant, non-malignant and dementia/frailty) as these groupings are widely used to distinguish different disease trajectories, notably within the Gold Standard Framework.20
SPSS (V.18) was used to analyse the data. To investigate whether initial PPS predicted mortality, Cox regression for time from the first PPS assessment to death (censored at 1-year follow-up) was carried out. Initially unadjusted models were fitted followed by an adjusted model (for gender and location). For the purposes of modelling, the PPS was recalculated as a score out of 10, with 10 representing death, and included in the model as a continuous variable to allow the interpretation of the HR to be with respect to a 10% decrease in the PPS score (a worsening). Separate models were then fitted for each of the three diagnostic groups.
To compare mortality at 12 months between those whose PPS scores had remained unchanged/showed improvement and those whose PPS scores worsened from the initial to the second PPS assessment, χ2 tests were used. Separate logistic models for each diagnosis for 12-month mortality from the first PPS assessment were fitted with change in PPS as the explanatory variable (no change/improvement vs worsening). To account for the differences in timing between assessments, these models were adjusted for time between the first and second assessments (≤14 days, 15–28 days, >28 days) for the malignant group and (≤28 days, >28 days) for the non-malignant and dementia/frail groups (due to smaller numbers).
Ethical approval
Confirmation was obtained from the Regional Ethics Committee that full ethical approval was not required as this was a clinical audit. The audit was registered with NHS Grampian's clinical effectiveness department.
All data received from the clinical settings were stored on an encrypted, password protected computer database in a locked office and were anonymised.
Results
Fifteen clinical sites participated in this audit. We first present the feedback from clinicians on using the PPS, followed by analysis of the PPS to predict 12-month mortality.
Feedback from clinicians
Completed forms were received from all (15/15) key contacts at the clinical sites. Forms were also received from other clinicians who had used the PPS at the clinical sites. In total, 30 forms were returned. Nineteen forms were returned from secondary care locations, 10 were from primary care and one person did not identify his/her location.
The forms were completed by 23 nurses, 3 doctors, 3 nursing home managers and 1 person who did not identify his/her role.
Responses to the questions about how easy the assessment tool was to use, whether it helped in the recognition of disease progression and the identification of nursing needs are shown in table 1. Most respondents found the PPS easy to use and that it helped to recognise disease progression in cancer patients, but they were less likely to report that it helped to recognise palliative nursing needs.
Ease of use of the PPS assessment tool
A number of respondents commented that the PPS was user friendly, easy to explain and understand as well as being quick to use. There were a number of comments about the difficulty in placing a patient on a specific percentage band in the PPS. For example, a nurse working in a setting for the care of older people noted ‘sometimes difficult fitting patients in 50–60% category’, and this view was also shared by a specialist nurse working in secondary care who said ‘it was quick and straightforward, although some debate about which percentage to use’.
It was noted by staff working in both the care home sector and a ward for the care of older people that the assessment tools did not increase workload, with another person saying that it was easier to use than another integrated care pathway. One respondent regarded the use of the tools as adding to the bureaucracy of the admission procedure, but another felt that it provided ‘evidence’ to back up their own feelings and predictions.
PPS helping to recognise disease progression
The comments reflected different views about the helpfulness of the assessment tool in the recognition of disease progression. For some respondents, the tool did not help to recognise specific disease progression. For example, a nursing home manager observed that ‘dementia's slow decline doesn't fit particularly well’. Another respondent who looks after people with dementia felt that the scores did not represent the actual clinical picture and that deterioration was difficult to see. For patients with a non-malignant condition such as chronic obstructive pulmonary disease, it was observed that the results fluctuated according to the patient's condition and the level of exacerbation.
However, a specialist palliative care nurse working in primary care with patients diagnosed with cancer felt that it did focus on disease progression, especially after a significant event such as a hospital admission. This view was supported by staff in the oncology ward who found the change of status useful to follow if the patient had been re-admitted. One respondent said ‘it is good to see at a glance time scale and rate of decline’ while another noted ‘useful, especially when the patient is re-admitted as you can see if they've deteriorated and at what speed’.
For a nurse working in the care of older people, the tool was useful for patients who had been admitted for more than 2 weeks. One respondent noted that the tools were useful, but in combination with patient observation and staff feedback.
PPS helping to recognise palliative nursing need
As in the previous question, respondents had different views about the usefulness of the PPS in the recognition of palliative nursing needs. From a positive perspective, a number of respondents noted that it was useful to see the rate of decline at a glance and that this prompted a review of care. These respondents worked in palliative care settings.
By contrast, for clinicians working with individuals with dementia, the PPS did not identify a need for change in nursing due to the slow deterioration. One person noted ‘underlying conditions may have required them to need assistance for a couple of years with washing and dressing’. Another clinician working in primary care felt that there was an instinctive knowledge of patient deterioration as the patient was known for such a long time. There was also a concern raised about the potential variability in assessment ability by different clinicians. A general practitioner felt that the doctors and nurses gave different scores. A final comment mentioned the need for consistency in the use of any assessment tool.
Survival data
The patients were grouped according to three diagnostic trajectories: malignant, non-malignant and dementia/frailty.4 Patient location was also recorded (specialist palliative care unit, primary care, secondary care, nursing home). Table 2 details the patient characteristics. Overall, 62.5% of the patients died within 12 months, with the malignant group showing the highest mortality (74.6%). A greater proportion of men died within 12 months (66.5%) compared with women (59.5%). Those being cared for in a specialist palliative care unit showed the highest 12-month mortality (94.3%) compared with those in primary care (57.9%), secondary care (62.9%) and nursing home (31.7%).
Most (93.5%) patients whose initial PPS score was 10 (N=31) died during the 12-month follow-up, decreasing to 64.9% for those with initial PPS=50 (N=134), with only 22.7% of those with initial PPS=100 (n=22) dying during the 12-month follow-up. Median time to death for those with initial PPS=10 was 3 days (IQR=1–11 days) while for those with initial PPS=30 this was 35.5 (17.25–123.5) days and for those whose initial PPS=50 this was 52 (18–141) days. For those who scored 100 at the first assessment (N=22), median (IQR) days to death was 138 (115.5–335.5).
A diagnosis of malignancy, male gender and being an inpatient in a specialist palliative care unit were all associated with an increased hazard of death (table 3).
The cumulative survival for the adjusted Cox model split by diagnosis is shown in figure 1.
Analysing the diagnosis groups separately showed that a reduction of 10% in the PPS at first assessment increased the hazard of death in each of the diagnosis groups (table 4).
Analysis of rate of change in PPS
Overall, 422/666 (63%) patients had at least a second PPS. Of these, 319 (76%) had no change or a better PPS score at the second assessment, while 103 (24%) had a worsening of PPS compared with the first assessment. Within the no-change/improvement group, 49.8% had died by the 12-month follow-up compared with 81.6% from the worsened PPS group (p<0.001).
Breaking this down into separate diagnoses and using logistic regression for 12-month mortality with change in the PPS as the explanatory variable (adjusting for time between first and second assessment) showed that a worsening of PPS was indicative of increased mortality in all three diagnosis groups, but not statistically significant for the non-malignant group (table 5).
Discussion
This study aimed to prospectively evaluate the introduction and use of the PPS across primary, secondary and tertiary care settings in Grampian to assist advance care planning for palliative patients as has been recommended by ‘Living and Dying Well’.8
The PPS was developed in Canada to assess and document physical change in palliative patients and to facilitate communication between health professionals.9 Subsequent publications have suggested that the PPS (in particular the initial PPS score at first assessment) could be used as a prognostic predictor.10,–,12 15 16 21 However, the literature available examines the use of the PPS in a limited number of populations in Canada9 11 13 15 16 18 22 and the USA.10 12 21 Moreover, all these studies examined patients referred for hospice care or specialist palliative care consultation.
The feedback from the pilot sites suggests that clinicians find the PPS convenient, easy to use and useful to establish a baseline from which to monitor overall deterioration or stability in condition as well as a sense of the timescale of a patient's deterioration. This may help with advance care planning, especially if patients are being admitted to/discharged from hospital as it shows the rate and degree of deterioration following a significant clinical event. This supports the original aim of the PPS.9 However, the individuals caring for people with advanced dementia found it less useful as the rate of deterioration was usually too slow to be picked up by the PPS within the timeframe of the study.
It is interesting to note that the median survival times for patients in our sample tended to be longer than those previously published. For example, the median survival for a patient with PPS=30 in our sample is 32 days (95% CI 14.5–75) compared with just 5 days (95% CI 5–5) reported in a large (n=6066) Canadian specialist palliative care service study.16 Much of this difference is accounted for by patients with non-malignant disease and dementia/frailty in our sample having statistically longer survival trajectories than cancer patients. This would appear to support the use of the three-trajectory model4 introduced into UK clinical practice by the Gold Standards Framework20 and contrasts with previously published findings suggesting that diagnosis may not be an important variable22 or that patients with non-malignant disease may have shorter survival rates.10 23 Notably, in our patient population, relatively few patients with non-malignant disease or dementia/frailty had low initial PPS scores. It has been recognised that the PPS is most accurate at predicting survival at low PPS scores and that there is greater subjectivity in grading patients with higher scores.13 It is possible that the relatively high proportion of patients with high PPS scores in our study reflects current ‘best practice’ in the UK.20 This contrasts with the much shorter estimated prognoses required for access to hospice programmes in other healthcare systems.24 Our findings confirm the need to be wary of generalising survival data from one patient population/healthcare setting to another.11 13 15 16
The data from patients who had a repeat PPS score suggest that repeat scoring may help refine prognostic estimates. This is a provisional finding only, as these repeat scores were not done in a standardised way. Some clinicians repeated scores regularly, some only if there had been a clinical change. The time from initial score and repeat score also varied considerably between patients. Despite this, we saw a clear survival reduction in patients with any reduction in PPS score versus patients with stable or increasing scores. While most studies report on initial PPS scores only, repeat PPS scores have been shown to help predict patients who could be discharged.21 This suggests that further research is needed to more rigorously examine the impact of changing functional status on prognosis. This approach may be most valuable in non-malignant disease and frailty/dementia where the usual disease trajectory does not follow the more predictable accelerating decline of cancer.25
A number of the participating sites continue to use the PPS following the completion of the audit. Community hospital nurses have found that the PPS provides a useful communication aid, promoting anticipatory care, specifically used when a patient's PPS is ≤20. At this point, the nurses contact the covering medical team to get a ‘Just in Case’ prescription to allow rapid symptom control at the end of life. While the nurses have often been aware of the deterioration of their patient, the PPS score ‘backs up’ their clinical assessment and gives them more confidence when communicating with their medical colleagues and patients' families.
The specialist palliative care unit has integrated the PPS into the daily assessment of inpatients. The PPS scores are used to guide care planning and are used in handovers between day and night staff.
The study has several limitations. There was no inter-rater reliability testing in any of the participating sites. There was also no attempt to standardise the timing or frequency of repeat PPS scoring. Furthermore, the clinicians themselves determined who was ‘palliative’. We did not specify any inclusion or exclusion criteria. Although this might be a weakness from a methodological perspective, it does reflect the real world of clinical practice.
The assessment data were collected for a period of 3 months, which might be too short. This was a pragmatic decision due to the time pressure associated with the implementation of a national action plan. Our dataset was not exhaustive, again, to allow for easy data collection in busy clinical environments. As a result, our statistical analysis is necessarily limited, for example, for the survival analysis we controlled for gender and location only. Many other confounding variables may have been present and this limits the generalisability of our results. This emphasises the need for care when trying to extrapolate prognostication data from one area to another, which may have different patient populations, healthcare systems and even definitions of palliative care.
Conclusion
This project was undertaken as part of the rollout of ‘Living and Dying Well: a national action plan for palliative and end of life care’ in Scotland. An audit approach was used to determine if the PPS was practical, useful and accurate. Fifteen pilot sites volunteered to use the PPS over a period of 3 months when they were reviewing their palliative patients. Overall, the PPS was found to be a useful and easy to use adjunct to advance care planning. Any tool that can help in the recognition of deterioration has to be succinct and easy to use, in order for it to be integrated into standard care.
However, the prognostic element of the PPS was different to the published literature, especially in palliative patients with non-malignant disease and dementia/frailty.
This project builds on the Gold Standards Framework, Scotland,20 which promotes the early recognition and review of palliative patients to help in advance care planning with patients, families and other members of the multi-professional team.
Acknowledgments
The authors acknowledge the advice of Professor A W Chambers on an earlier draft of the manuscript.
References
Footnotes
-
Funding Friends of Roxburghe House charity.
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.