From June 2015 to January 2016, patients admitted to the Hospice & Palliative Care Unit of Seoul Medical Center were primarily recruited. Admission criteria to the palliative care unit included diverse terminal stage cancer with a life expectancy of less than 6 months based on the opinion of two oncologists. Patients who were expected to die within 3 days were excluded. Except for those individuals, all patients were informed of this study by the oncologists. The oncologists received informed consent from the patients. If the patients could not communicate due to their condition, we received permission from the primary caregiver. The physiatrist then visited the patients and explained the rehabilitation approaches, including the survey, evaluation, and rehabilitation program. If the patients or caregivers did not want to get involved in this program, they were excluded.

We investigated the basic characteristics, initial pain scores using the visual analog scale, Eastern Cooperative Oncology Group (ECOG) performance scale, and the Karnofsky performance scale. ECOG and Karnofsky performance scales are widely used functional scales that describe the functional ability of cancer patients. ECOG ranges from 0 to 5, where 0 means fully active, and 5 means patient death [13]. The Karnofsky scale ranges from 0 to 100, where 0 indicates the dead and 100 indicates the normal [14]. One week after the initial survey, we followed-up with the cancer survivors. Follow-up surveys were completed in the 1st and 2nd week. After 6 months of closing this study, we investigated the survival days from the initial evaluation day using medical records.

We evaluated the patients regarding rehabilitation comprehension of their problems and functions. We made a problem list from a rehabilitation viewpoint, and for each problem, we recommended the rehabilitative approach. We evaluated their function of self-care ADL (eating, grooming, dressing, using a toilet, bathing, sphincter control, and ambulation). We also evaluation ambulation level using the Functional Ambulatory Category (FAC) [15]. FAC cannot represent wheelchair ambulators. Thus, we created and checked a 5-scale ambulation ability tool, level of ambulation. Level of ambulation consisted of five categories: (1) walking independently, (2) walking with supervision or minimal assistance, (3) walking with moderate to maximal assistance, (4) propelling the wheelchair manually, and (5) totally dependent. Level of ambulation was checked under a level surface, for 10-m ambulation with or without holding a monocane or pole.

According to the patient's problems and needs, we recommended rehabilitation approaches, such as physical therapy at the gym (GymPT), bedside physical therapy (bedsidePT), physical modalities to control pain, medication recommendations, a caregiving method change, and pain intervention. GymPT included standing training using a tilt table, parallel bar walking, stationary bicycle riding, balance, and endurance exercise, and was recommended to the patients with partial ambulation ability and stable vital signs. If the patients had unstable vital signs or were unable to ambulate to the gym, BedsidePT was recommended, which consisted of passive and active range of motion exercises. Physical modalities were recommended to those who had localized pain, and they consisted of transcutaneous electrical stimulation and heat therapy. Medication recommendations consisted of all kind of medications affecting QoL, such as pain-killers, neuropathic pain drugs, and medications aiding bowel movements or voiding. Changing methods consisted of recommendations of feeding methods, sore dressing, bladder and bowel care, and exercise methods given by caregivers. The rehabilitation intervention treatment was agreed to by the patient or the caregiver.

For the five rehabilitation approaches (GymPT, bedsidePT, physical modalities, medication, and pain intervention), we monitored the compliance of each item. We classified the compliant group as the patients who followed our rehabilitation recommendations. The others were classified as the non-compliant group.

We also received satisfaction-related questionnaires from the patients who participated in rehabilitation intervention for more than a week and agreed to respond. If the patients could not reply, the primary caregiver could answer. We use a general satisfaction score with the 11-point Likert scale (0–10). We also monitored five satisfaction items (pain relief, maintenance of ADL, improvement of fatigue, improvement of psychologic symptoms, and preference to receive any treatment) and five dissatisfaction items (no improvement in any symptom, no improvement in activity of daily living, too hard to take rehabilitation treatment, aggravation of fatigue, and aggravation of psychologic symptoms). We counted satisfaction scores (each +1 point) and dissatisfaction score (each +1 point) for each item. The total satisfaction score was calculated as the satisfaction score minus the dissatisfaction score.

Two-tailed independent t-tests were used to compare means between the compliant and non-compliant groups for each rehabilitation interventions. A Spearman's correlation was used for patient characteristics and satisfaction scores. All statistical analyses were performed using SPSS version 20.0 (IBM SPSS Inc., Armonk, NY, USA). A p<0.05 was considered statistically significant.

This study was approved by the Institutional Review Board of Seoul Medical Center (No. 2015-046).

A total of 62 patients were initially included during the study period. After the explanation about the rehabilitation program with the physiatrist, 45 patients (19 male and 26 female patients) agreed to undertake the rehabilitation approach. The mean age of the subjects was 64 years. The median number of survival days was 30. Demographic characteristics and functional levels are described in Table 1. The sites of the primary cancer were classified into 10 sites, and participants showed various functional status, FAC, and levels of ambulation (Fig. 1).

We investigated both subjective and objective problems. Subjective problems were complaints from patients or caregivers, which were not evaluated by any diagnostic methods. They consisted of gait-related difficulty, pain, spasticity (spastic feeling), dysphagia (feeling of swallowing difficulty), and bladder or bowel management problems. Spasticity was not evaluated objectively. Dysphagia also depended on patients' or caregivers' appeal. Bladder problems consisted of incontinence, urination frequency, urgency, dysuria, and the need for a device for voiding. Bowel problems were constipation, diarrhea, and incontinence. Thirty-two patients (71.1%) complained of gait-related difficulty (Table 2). Objective problems were problems identified or diagnosed by the physiatrist or other medical personnel. They included not-alert mental status, pressure sore, edema, sensory dysfunction, and poor medical conditions. Not alert mental state consisted of coma, semi-coma, stuporous, and delirious states. A pressure sore was defined above National Pressure Ulcer Advisory Panel stage 2. Edema included any type of limb edema. Sensory dysfunction indicated a reduced sensory function that increased the risk of injury. Poor medical conditions included any barriers of performing exercises at the gym, such as low hemoglobin/hematocrit, unstable vital signs, bleeding tendency, fever, and evidence of acute infection or inflammation. Thirty-one patients (68.9%) exhibited poor medical conditions (Table 2).

Pain was classified into cancer-related pain, musculoskeletal pain, neuropathic pain, and Complex Regional Pain Syndrome (CRPS) pain. Cancer-related pain was defined as pain near the primary cancer site, pain near the direct invasion or distant metastasis site, post-operative pain at the operation site, and radiotherapy related pain. Musculoskeletal pain included low back pain, radicular pain, frozen shoulder, shoulder impingement associated pain, joint pain irrelevant to the cancer site, osteoporosis induced compression fracture, and myofascial pain syndrome. CRPS pain included type 1 or type 2 CRPS. Musculoskeletal pain was the most frequent complaint, with 21 patients (46.7%) complaining of this type of pain. However, 11 patients (24.4%) showed cancer-related pain (Table 2).

According to each patient's problem, we recommended GymPT, bedsidePT, physical modalities, medication change, and pain interventions. GymPT included exercises at the gym, such as tilted table standing, stander, motormed, stationary bike, parallel bar gait, and any other gait training. BedsidePT consisted of active and passive range or motion exercise at the bedside. Physical modalities included heat therapy, such as a hot pack and ultrasound, transcutaneous electrical nerve stimulation or interferential current therapy, intermittent pneumatic compression, and manual lymphatic drainage. Medication change indicated a change or addition of pain medications, gastrointestinal drugs, bladder or bowel management medications, and edema control medications. Pain interventions consisted of musculoskeletal interventions including steroid injection, peripheral nerve blocks, and epidural steroid injections.

Of GymPT-recommended 22 patients, only 13 completed their rehabilitation. However, 80% patients agreed for bedsidePT. Although physical modalities were approved for 21 patients, only 12 received the treatment. No one took the recommendation for pain intervention and medication change item. Compliance of each rehabilitation recommendation is described in Fig. 2. For GymPT, a significant difference was observed in the survival days between the compliant and non-compliant groups. Compliant to GymPT patients showed higher survival (Table 3).

Only 13 patients completed the satisfaction-dissatisfaction questionnaire. They scored an overall satisfaction of 6.69 out of 10 (Table 4). Patients with a higher satisfaction score showed better ECOG and Karnofsky performance scales with significant correlations in Spearman correlation (Table 5).

Even patients in the terminal stage express strong wishes such as “I want to walk” and “I want to go to the bathroom” [3]. In our study, many patients and caregivers thought walking to the bathroom was very important. In other words, even in hospice, patients and their family wanted to experience meaningful care. Active exercises, including GymPT and bedsidePT may enhance self-esteem and psychological symptoms. Physical modalities do not include any exercise, but to undergo the modalities, patients must go to the therapy room. This may also be helpful to the patients' overall satisfaction. In our opinion, preserving ambulatory function is the largest need that is closely related to self-esteem and QoL in palliative care patients [19]. In hospice care, physiatrists must consider the various needs for ambulation.

Our investigation revealed that musculoskeletal pain was the most frequently-mentioned complaint. For breast cancer patients, diverse musculoskeletal disorders can provoke shoulder problems [20,21]. Many cancer patients suffer from cancer-related pain and other musculoskeletal pain [22]. This is also applicable to advanced cancer patients receiving palliative care. Proper management of musculoskeletal pain like physical modalities, painkillers, injection therapy, and exercises could reduce pain and opioid use to enhance physical activity and QoL. Among the musculoskeletal pain, frozen shoulders, low back pain, and radicular pain are the most common areas that patients mention. This pain is not fatal, but hinders activity and ADL. It may also aggravate psychological symptoms. Injection therapies such as intraarticular steroid or epidural steroid injections can reduce these kinds of pain immediately, and can help to reduce activity limitation. We suggest that active pain intervention can mitigate these losses.

Frozen shoulders might be due to immobilization and surgery [20,21]. Neuropathic pain might be related to chemotherapy [20,23]. There was one case of complex regional pain syndrome that was related to brain metastasis. We also observed one case of spastic pain due to spinal cord compression. This disease entity is related to rehabilitation. A rehabilitation approach could afford a proper diagnosis and management for these individuals. Considering the diverse cause of pain and activity limitations of advanced cancer patients, from an early stage of treatment till their death, a rehabilitative approach can help the patients maintain a higher QoL [17].

For the rehabilitation recommendation, patients with better performance scales showed greater compliance. In GymPT, more compliant patients had longer survival days. The rehabilitative approach consisted of GymPT, bedsidePT, physical modalities, recommendation of pain interventions, and medications. All patients who were recommended for pain interventions and medication change refused the modifications. Almost all patients and caregivers thought they were terminal and did not want any invasiveness. For a similar reason, some patients and care-givers thought that exercises were meaningless. We recommended to some patients changing medications, musculoskeletal pain interventions and different methods of caregiving at bedside, to the patients, but no one participated. Some patients felt that exercise was too hard. We, along with our physical therapists, had little experience with terminal cancer patients. To improve compliance, research regarding the proper intensity of rehabilitation would be necessary. In the satisfaction-related survey, many patients expressed being satisfied with the receipt of any treatment. Hospice care does not equate to doing nothing. We could not perform a study of effectiveness, but patients with better performance showed compliance and satisfaction. It may mean that the rehabilitation approach in patients with better performance could increase their effectiveness. Even though these patients suffer from an incurable disease, early application of comprehensive rehabilitation may be needed.

This study has some limitations. First, this study included patients with various types of cancers. Each cancer may have various characters and burdens to the patients that are diverse. Furthermore, we did not investigate whether cancers were locally advanced or distantly metastatic. Locally advanced cancer may show different features compared with distant metastasis. However, many participants were transferred from other hospitals, little medical information was available on their condition. The patients only wanted to receive palliative care and did not want to undergo more diagnostic tests. A study of a single type of solid cancer will elicit the effectiveness of rehabilitation more precisely. Second, we did not investigate the effectiveness of comprehensive rehabilitation. All the patients in the palliative care unit were near death. As time went by, all patients experienced a functional decline and QoL aggravation. This resulted in poor compliance. Thus, general evaluation to check the effectiveness of rehabilitation was inadequate. We could not find the right tool to monitor the effectiveness. Only the compliance and satisfaction could have been evaluated, and we could not have investigated factors affecting compliance. Furthermore, the satisfaction-related survey was our innovative idea without validation data. We could not find an adequate tool to evaluate the compliance and satisfaction of cancer patients in the terminal stage. For a similar reason, we could not assign a control group. To prove the effectiveness of any intervention, a randomized controlled study is the most powerful. However, for these patients, a controlled study was difficult due to the physical status and some existing ethical issues. Lastly, the follow-up time was short. In our institution, hospitalization in the hospice palliative care unit was limited to 6 weeks (Nowadays, the limitation is 2 months due to the change in insurance policy). The median number of survival days was 30, and many patients did not complete the 3-week follow-up. Only 11 patients could answer the satisfaction-related questionnaire. In Korea, many hospice palliative care centers receive patients in the terminal stages of cancer. If a patient with advanced cancer is in the terminal stage, early palliative care and rehabilitation could be helpful to preserve QoL and independent ADL until the end of life. Thus, the development of early palliative care and an early rehabilitation program is necessary. In addition, early palliative rehabilitation research may solve the problems associated with the randomized controlled study. This was a preliminary study and the first study on palliative rehabilitation. A multicenter research study of a larger size and longer duration is needed.

In conclusion, advanced cancer patients hospitalized in hospice palliative care unit face several issues, and comprehensive rehabilitation approaches help to resolve these issues. Individuals with better performance showed more compliance and satisfaction. To improve the palliative rehabilitation program, a larger population is needed.