Advanced cancer patients have a poor quality of life and a high rate of unplanned hospital admissions. Palliative care has been shown to improve patient outcomes such as quality of life and symptom burden. Our long-term aim is to study the effect of Supportive and Palliative Care Review Kit (SPaRK) - a novel model of early palliative care in the acute hospital inpatient setting. To this end, we propose a pilot study that will provide the necessary information for the planning of a subsequent larger scale Phase III trial of SPaRK.

The specific aims of this pilot study are: 1) to estimate the recruitment rate and ability of advanced cancer patients to complete the Functional Assessment of Cancer Therapy - General (FACT-G) questionnaire, which measures health-related quality of life (QOL) in the four domains of physical well-being, social/family well-being, emotional well-being, and functional well-being, 2) to obtain a tentative estimate of the effect size of the SPaRK model of care on change in patient QOL over 6 days, 3) to explore the responsiveness to change of FACT-G over 3 days and 4) to explore views of healthcare professionals and stakeholders on SPaRK as a model of integrated palliative care and oncology care service delivery.

In the short term, the results from this study will be used to plan a larger scale Phase III study of SPaRK. In the long term, a resource-efficient and effective model of care needs to be developed for the aging population and rising palliative care needs of patients. If successful, this pilot study will lead on to the testing of SPaRK, which does not rely wholly on specialist palliative care manpower, but involves both specialists and non-specialists working together to provide palliative care, and hence can be feasibly scaled up across institutions and extended to non-cancer patients.

Specific Aims ==========

This is a pilot study of the Supportive and Palliative care Review Kit (SPaRK) model of care, which is integrated palliative care and oncology care for advanced cancer patients, and will be focused on improving outcomes in lung cancer patients in the acute hospital inpatient setting. The current usual care will be provided in the first 3-month period from 1 October to 31 December 2015, and the novel SPaRK model of care will be piloted in the latter 3-month period from 1 January to 31 March 2016. The overall aim of this pilot study is to facilitate the planning of an upcoming phase III trial.

Specific aim 1: To estimate the recruitment rate and data completion rate of the FACT-G instrument We will estimate the recruitment rate and data completion rate for FACT-G, which will demonstrate the feasibility and help in the planning of the future Phase III trial. The measurement properties of FACT-G will be detailed later in the 'Methods' section of this proposal. The FACT-G questionnaire will be administered at up to four time points - day 1, 4 and 7 of hospital admission and day of discharge.

Specific aim 2: To obtain tentative estimate the effect size of SPaRK on change in FACT-G from day 1 to day 7 of hospital admission The control group will be hospital admissions that are managed under the current model of care. The intervention group will be hospital admissions that are managed under the new SPaRK model of care. We hypothesize that the SPaRK model of care will result in a greater improvement in FACT-G score from day 1 to day 7 of hospital admission.

Specific aim 3: To explore the responsiveness to change of FACT-G from day 1 to day 4 of hospital admission Within the three-day period from day 1 to day 4 of hospital admission, we expect the acute problems that prompted the admission to be addressed. The patient's 'phase of illness' will be recorded by the primary medical oncology physician on day 1 and day 4 as part of the study protocol. For patients who change from an 'unstable phase' on day 1 to 'stable phase' on day 4, we will estimate the responsiveness to change of FACT-G score in terms of association with this change. Even though FACT-G uses a recall period of seven days, we hypothesize that FACT-G score will nonetheless increase among patients who change from an 'unstable phase' to 'stable phase' over the shorter time period of three days from day 1 to day 4 of hospital admission while there is no change in FACT-G score among patients who remained in the 'unstable phase. This would be useful for measuring outcomes in patients who are discharged before day 7 of hospital admission.

Specific Aim 4: To explore the views of healthcare professionals regarding SPaRK as a model of integrated palliative care and oncology care service delivery Semi-structured interviews will be conducted among a purposive sample of healthcare professionals, management and key decision makers involved in the pilot study. We will explore their views on how integrated palliative care and oncology care can be delivered in the acute hospital inpatient setting, and whether SPaRK is an acceptable model for providing this care.

Methods

=======

Study setting This study will be conducted in Singapore General Hospital (SGH), which is a 1,597-bedded acute hospital with more than 78,000 admissions per year. About 5,000 hospital admissions per year are under DMO, out of which about 70% are unplanned admissions for advanced cancer patients, with a mean length of hospital stay of 7.2 days. These inpatients are cared for by one of four Division of Medical Oncology (DMO) teams broadly organized by tumor site -gastrointestinal (GI), lung, breast and lymphoma.

In the current model of service, if the inpatient DMO team decides that a patient would benefit from palliative care input, they would request for a review from the Division of Palliative Medicine (DPM) team. The DPM team will provide palliative care (PC) by reviewing the patients and making suggestions for patient management, but the patient remains under the care of DMO.

Study design This is a cluster-controlled study with 1 control cluster and 1 intervention cluster, with patient admission episodes as units of inference. The control cluster will comprise patient admission episodes occurring during the 3-month control period (1 October to 31 December 2015) when the current model of care is in place; the intervention cluster will comprise patient admission episodes during the subsequent 3-month intervention period (1 January to 31 March 2016) when the SPaRK model of care is in place.

Study patients This pilot study will only be conducted in the DMO lung team as their cohort of patients has the highest percentage of our target population - advanced cancer patients (86.0%) with unplanned hospital admissions (84.6%). Furthermore, the patients under the DMO lung team also have the highest unplanned readmission rate (11.8% within 7 days, 35.3% within 30 days), which is a surrogate marker for unresolved palliative care problems.

All advanced cancer patients who have an unplanned admission under the lung team during a 6-month period from 1 October 2015 to 31 March 2016 will be included. In the first 3-month period from 1 October to 31 December 2015, current usual care will be provided and baseline data collected. The novel SPaRK model of care will be piloted in the latter 3-month period from 1 January 2016 to 31 March 2016.

Inclusion criteria

• 21 years old and above
• Has an unplanned admission to SGH under the care of the inpatient DMO Lung team
• Has a cancer diagnosis of stage 4 solid tumor
• Has capacity to give written informed consent to participate in the study (for patients who complete the patient questionnaires)

Exclusion criteria - Unable to complete self-administration of the FACT-G questionnaire

The SPaRK intervention The SPaRK intervention will be piloted in the DMO lung team for latter 3-month period (1 Jan to 31 Mar 2016).

All advanced cancer patients who have an unplanned admission in the SPaRK model of care will receive palliative care, which will be provided by the DMO team in conjunction with a palliative care advanced practice nurse (APN) and a part-time specialist palliative care (PC) physician who will be integrated members of the DMO team (the DPM/DMO SPaRK team. It is envisaged that PC physician will spend an average of 2 hours per weekday discussing cases with the APN, reviewing selected cases in person and participating in a weekly team meeting with the medical oncology team where the inpatient list is discussed. In the current model of care, patients referred to DPM for PC are also discussed in a weekly team meeting to establish their plan of care.

The components of palliative care (PC) that is provided in the control group by the DPM team and in the intervention group by the DPM/DMO SPaRK team will include assessments of physical symptoms, functional status, psychosocial issues and spiritual issues, followed by the formulation of a problem list and establishment of a care plan.

If any patient has complex palliative care needs that cannot be met by the SPaRK team, then either the DMO consultant, palliative care APN or palliative care physician will trigger a formal referral to the DPM team for review.

Data collection and instruments Patients will complete the FACT-G questionnaire (Available English, Chinese [simplified], Malay and Tamil validated translations will be used). The doctor in charge will complete the 'phase of illness' and Eastern Cooperative Oncology Group (ECOG) score.

We will be using FACT-G to measure the primary outcome of patient's health-related quality of life (QOL) as it is an established instrument that has been well-validated both internationally and locally for patients with many kinds of cancer. It has a low level of noise, as evidenced by reliability measures consistently shown to be between 0.8 and 0.95, which is suitable for comparison between groups. 21, 22, 24 As compared to some popular alternatives in cancer QoL research, FACT-G has been shown to be more sensitive in detecting differences between groups. It covers four QOL domains: physical well-being (7 items), social/family well-being (7 items), emotional well-being (6 items) and functional well-being (7 items), and each item has response choices in a 5-point Likert-type scale. We will adopt the official scoring method, which is part of the validated instrument that has been demonstrated to have good measurement properties as aforementioned. Response scores on negatively phrased questions are reversed so that higher scores always mean better QOL. The domain score will be obtained by summing individual item scores within the domain, subject to imputation for item non-response by the "half-rule", i.e. missing item score replaced by average of other scores in the subscale if more than 50% of the items in the same subscale were answered. It has been shown that the half-rule is very robust and that very few Singaporean cancer patients who participated in FACT-G assessment had item non-response not imputable by the half-rule. The total FACT-G score is the sum of the four domain scores.

The FACT-G questionnaire will be administered at four time points: T1 - day 1 of hospital admission; T2 - day 4 of hospital admission (if patient has not been discharged); T3 - day 7 of hospital admission (if patient has not been discharged); T4 - day of discharge. The standard administration method of FACT-G is self-report. Hence self-administration is the chosen mode of questionnaire completion. As there is some evidence that FACT-G is not equivalent between self-administration and interviewer-administration, we will not be using interviewer-administration in this study.

At all time points, the patient's 'phase of illness' will also be recorded. The standard definitions of the five phases are:

Phase 1 - Stable: patient problems and symptoms are adequately controlled. Phase 2 - Unstable: an urgent change in the plan of care or emergency treatment is required because the patient experiences a new problem that was not anticipated in the existing plan, or care and/or the patient experiences a rapid increase in the severity of a current problem, and/or family/carers circumstances change suddenly impacting on patient care.

Phase 3 - Deteriorating: the care plan is addressing anticipated needs but requires periodic review because there is a gradual worsening of an existing problem or there is a new but anticipated problem.

Phase 4 - Terminal: death likely within days. Phase 5 - Bereavement: patient has died.

The 'phase of illness' will be used to determine responsiveness to change of FACT-G, specifically for patients who change from an 'unstable phase' to 'stable phase'.

The following demographic data and background information will also be collected: age, gender, ethnicity, primary cancer diagnosis, stage of cancer and Eastern Cooperative Oncology Group (ECOG) performance status score. The ECOG score is a set of standard criteria for measuring how the disease impacts a patient's daily living abilities and describes a patient's level of functioning in terms of their ability to care for themself, daily activity, and physical ability (walking, working, etc.).

The following data for each patient will be obtained from the computer system:

• Hospital length of stay
• Discharge destination (e.g. home, nursing home, inpatient hospice)
• Referral to DPM during the admission and date of referral to DPM
• The presence of a plan for follow up by community palliative care services upon discharge from hospital (including outpatient palliative care clinic, home hospice, day hospice and inpatient hospice)
• The occurrence of another unplanned hospital admission within 7 days and 30 days after being discharged from hospital

These outcomes reflect the usage of acute hospital and community palliative care resources and will be used to decide if any of these outcomes would be suitable for further study in the future Phase III trial.

Data analysis and sample size planning Recruitment rate and data completion rate for FACT-G It is estimated that there will be 95 eligible patient admissions per month (based on data from May 2013-June 2014), and thus approximately 280 patient admissions in the 3-month control period and 280 patient admissions in the 3-month intervention period, totally 560 eligible patient admissions. We assume that, on average, one patient may have about 1.5 admissions during the study period, and that the intra-class correlation coefficient between outcomes measured at two admissions is about 0.5. This implies a design effect of about 1.25 and this is taken into account in the sample size planning to allow for clustering of data within the patient.

This study will update the percentage of eligible patient admissions, recruitment rate and data completion rate for the measured outcomes. A total of 560 eligible admissions in six months (effective sample size 450 after discounting for design effect) will allow us to more accurately estimate of the consent cum data completion rate with 95% confidence interval of approximately +/- 4%.

Effect size of SPaRK on change in FACT-G from day 1 to day 7 of hospital admission (T1 to T3) Assuming about 25% consent cum data completion rate for FACT-G at T1 (day 1 of hospital admission) and T3 (day 7 of hospital admission), we would have approximately 75 patient admissions each in the control group and intervention group (total 150 patient admissions) with complete FACT-G at both time points. This sample size would provide about 80% power at a 20% 1-sided Type 1 error rate to detect an effect size of 0.3 (PASS 13 software).

Responsiveness to change of FACT-G from day 1 to day 4 of hospital admission (T1 to T2) Sensitivity to change tests the ability of an instrument to detect important changes over time in the concept being measured. For patients who change from the 'unstable phase' at T1 (day 1 of hospital admission) to the 'stable phase' at T2 (day 4 of hospital admission), we will estimate the sensitivity to change of FACT-G scores between T1 and T2. We hypothesize that FACT-G scores will improve for patients whose phase of illness changes from 'unstable phase' at T1 to 'stable phase' at T2 while there will be no significant change in FACT-G score among patients who remain in the 'unstable phase' in the same period.

Views of healthcare professionals regarding SPaRK as a model of integrated palliative care and oncology care service delivery

Healthcare professionals involved in this pilot study will be invited to participate in one-to-one interviews. Purposive sampling will be used. Heads of department from the DMO, DPM and the nursing division will also be included. The format of the interviews will be semi-structured, where participants will be asked to discuss their responses to the following questions:

i) How do you think integrated palliative care and oncology care should be delivered in the acute hospital inpatient setting? ii) What do you think of the SPaRK model as a possible way of delivering integrated palliative care and oncology care to advanced cancer patients? iii) How can the SPaRK model be modified to improve its effectiveness in providing integrated palliative care and oncology care? iv) What do you perceive to be the role of the palliative care nurse and palliative care doctor in the oncology team? v) Do you think the SPaRK model should be implemented and why?

The interviews will be audio-taped and transcribed verbatim. Written notes will also be taken during the interview. Interview transcripts will be analyzed using framework analysis.

Conclusion

========= This is a pilot study of the SPaRK - a novel model of integrated palliative care and oncology care for advanced cancer patients who have an unplanned hospital admission. The overall aim of this pilot study is to establish the feasibility of a larger-scale Phase III trial of SPaRK. If successful, this pilot study will help design and plan the subsequent phase III trial, which will test a potentially viable as well as effective way of providing early palliative care for advanced cancer patients in the acute hospital inpatient setting.

Inclusion Criteria:

• Has an unplanned admission to SGH under the care of the inpatient DMO Lung team
• Has a cancer diagnosis of stage 4 solid tumor
• Has capacity to give written informed consent to participate in the study (for patients who complete the patient questionnaires)

Exclusion Criteria:

• Unable to complete self-administration of the FACT-G questionnaire