Malnutrition is highly prevalent among older adults. Adequate treatment is crucial to maintain health, improve functional status and independency. For severely malnourished patient, tube feeding is often the most effective treatment, but it also implies a risk of developing refeeding syndrome (RFS). RFS is described as fluid and electrolyte shifts particular hypophosphatemia when recommencing nutrition, potentially causing fatal complications. In 2006 National Institute for Health and Care Excellence (NICE) developed guidelines regarding 1) detecting patients at risk for RFS and 2) a treatment plan for refeeding severely malnourished patients. These guidelines have never been validated. Hence, there is a need to explore the use of different refeeding regimens in the treatment of older malnourished patients, in regards to assess safety, as well as the impact of the treatment, on functional status, independency, quality of life and RFS. This will be the first randomized controlled trial (RCT) that challenges the existing NICE guidelines. A blinded RCT study, with the aim of tube feeding patients at risk for developing RFS, and assigned either to an intervention group or to a control group, comparing the different refeeding regimens. This study will help develop safe and validated refeeding protocols for severely malnourished older patients.

Before 2006, the recommended initial refeeding rate was around 20-25kcal/kg/day for severely malnourished patients. But due to the risk of RFS, the National Institute of Health and Clinical Excellence (NICE) in 2006 published new guidelines, recommending a very cautious refeeding level of only 5-10 kcal/kg/day. However, the supporting literature underpinning the evidence that results in the above recommendations is weak, and is mostly based upon cohort studies, case series and consensus expert opinions. RFS was first identified and reported on in the late forties when starved concentration camp inmates and prisoners of war developed cardiac arrest, neuralgic symptoms and/or peripheral edema during recommencement of feeding. Similar symptoms were noticed again during the 1970's and 1980's, when parenteral nutrition was used widely, and carbohydrate provision was not restricted.

Patients defined as at risk for RFS are those who are malnourished and have recently lost weight. In addition, patients with, eating disorders, uncontrolled diabetes, cancer, malabsorption syndromes, alcoholics, surgical patients, elderly, and patients who are fasting or have undergone bariatric surgery, are high risk groups. It is well known that low levels of electrolytes are fatal, and may cause cardiac-and lung problems, affect the gastrointestinal-and the immune system, and increase the occurrence of sepsis and infections. Even severely malnourished patients have normal electrolyte levels upon admission, but when refeeding, a rapid drop in electrolytes, especially phosphate, may occur, usually within 2-5 days. Electrolytes like potassium and sodium are measured routinely, but magnesium and phosphate are not included in routine laboratory tests, and therefore often overlooked. Studies have found that patients screened as "not at risk" also develop RFS confirming the need for monitoring phosphate levels on a broader group than just patients at risk. Two main challenges regarding RFS are the absence of a clear definition and the uncertainties regarding the optimal refeeding rate. However, new onset hypophosphatemia after refeeding is a good predictor of RFS and will be used in this study when defining RFS.

Refeeding syndrome in Anorexia Nervosa:

Most studies on RFS conducted on patients with eating disorders, but few are Randomized Controlled Trials (RCT). In later years there has been a paradigm shift in regards to refeeding patients with anorexia nervosa (AN). For years, a low calorie approach has been recommended, but these recommendations has led to poor weight gain and prolonged hospital stay. Garber et al. did a systematic review on refeeding patients with AN, and concluded that higher calorie approaches on AN patients in hospital is feasible, when close medical monitoring is provided. Studies on patients with eating disorders indicate that the severity of malnutrition rather than the total energy intake predict the development of RFS. For instance a calorie level at both 125 and 1900 kcal a day led to hypophosphatemia. The great difference in recommendation for initial refeeding rates in patients with eating disorders, illustrates the lack of studies and international consensus on the topic.

Older adults and refeeding syndrome:

Overall only a few studies are done on tube feeding (TF) in older adults with the aim of improving nutritional status. Regarding RFS and older adults, there are only six clinical trials, but none of them RCT's. These trials are mostly looking at occurrence of RFS, and they all use different cut off levels when defining hypophosphatemia (HP). In addition, they all used different outcomes measures; either mortality rates, delirium or muscle strength, which in the end makes it difficult to compare the studies.

When refeeding malnourished patients the effect on physical function is often the first sign of improvement. Hand grip strength (HGS) is a relevant marker of physical function, and has been shown to be a legitimate indicator of nutritional status. In some populations HGS may even predict outcomes better than traditional nutritional markers such as weight to height ratios, weight loss or limb circumference . Reduced HGS is associated with increased LoS), increased risk of complication, premature mortality and earlier onset of dependency. Recently a study has shown that lower HGS at discharge is associated with higher 30-day readmission rate, and might be a useful indicator to detect patient at risk of readmission, and the need for individualize discharge planning including nutrition care.

Although it is reported that refeeding too cautiously can place patients at even further risk of malnutrition, postpone recovery and lead to unnecessary loss of physical function but NICE has stated that guidelines cannot be changed unless more high-quality RCTs are performed. This study will provide new knowledge regarding how to refeed severely malnourished elderly patients. By providing them with 40% more calories and protein the first 7 days of hospitalization one can expect faster improvements in regards to physical function, independency, reduced readmission-and mortality rates. In addition this study provides new knowledge regarding the necessity to monitor phosphate levels in patients not identified as at risk of developing RFS.

Hypothesis: The investigators hypothesize that providing a higher initial refeeding rate, reaching estimated calorie and protein needs faster, is as safe as following NICE guidelines. Furthermore, the investigators hypothesize that the intervention group will improve HGS, and achieve a better long-term effects compared to the control group.

Study aims are to:

1. Investigate if a high initial refeeding rate is safe and leads to improvement of the patient's physical function and independency compare to a slower refeeding rate?
2. Evaluate the long-term effects three months after discharge between two different refeeding regimens regarding physical function, independency, quality of life, mental test score, readmission and mortality rates

Feasibility of the study:

Regarding the feasibility of the project: it will be carried out on all medical and surgical wards at Diakonhjemmet hospital. Refeeding of patients using tube feeding is part of routine care. The study is already approved by the Regional Ethic Committee (REC) and the head of the departments at Diakonhjemmet hospital have been informed and approved the research project.

Refeeding guidelines forms are premade based on kg and type of intervention (fast or slow) and will be inserted as phrases in the electronic medical journal.

Regarding recruitment of patients, it is estimated to include 1.5 patients a week (63 patients in a year based on 42 recruiting weeks in a year). The estimation is based on a point prevalence of malnutrition performed in 2016, where 12 patients out of the 72 screened patients 65 years and older (16.7%), had BMI <18.5. The calculation was based on a 10 % agreement of participation. To be even more cautious the investigators estimates a recruitment of 126 patients ,within 3 years, based on patients accessibility of 4000 new admissions a year.

During the study period there will be regular meetings with the supervisors and the project coordinator, located at the same ward.

A research assistant (RA) will perform the physical test and questionnaires, in hospital and at 3 months

Publications:

1. Will a higher initial refeeding rate, meeting estimated calories and protein needs faster, in severely malnourished older patients affect patient's handgrip strength?
2. Effects of two different refeeding regimes in severely malnourished older patients admitted to hospital in regards to physical function, cognition, quality of life, activities of daily living, and readmission and mortality rates.
3. The difference in occurrence of electrolyte disturbances comparing two refeeding regimes in severely malnourished older patients admitted to hospital.

Statistics Statistics The sample size estimation is performed using HGS as the primary endpoint. A clinically important difference was defined as a mean difference in HGS of 1 kg between baseline and follow up between the two treatment groups. Then, if the true mean difference in the intervention and control group in HGS between baseline and follow up is 1 kg with an estimated standard deviation of 1.8, the investigator will need to study a minimum of 52 patients in each treatment group, a total of 104 patients, considering a type I error of 5% and a power of 80%. Due to possible drop-outs the sample size will be increased with 20 %, which gives a total of 126 patients i.e. 63 patients in each group.

Statistical analysis will be performed using SPSS version 20. Data will be presented as the mean and standard deviation (SD) for continuous variables and as percentages for categorical variables. Differences between the intervention group and control group will be assessed by t-tests for continuous data, with non-parametric test for ordinal data and with the chi-squared test for categorical data. The main statistical assessment of primary and secondary endpoint with repeated measurements will be mixed models for repeated measurements.

The data will be analysed following the intention to treat principles and all the randomized patients will be included in the analyses. LoS will be calculated as the difference between the day of admission and the day of hospital discharge.

User involvement The project is presented to and discussed with the User Committee (UC) at Diakonhjemmet Hospital. This study will recruit a very heterogenic patient group, which makes it difficult to find suitable users. In that matter, the UC has been chosen to represent the patients. UC has approved the project as they considers it to be clinically relevant and benefit for the users, and further supports an application. UCs representatives made important contributions regarding the target group of the project. The UC have raised concerns regarding patients with dementia making sure that their integrity is cared for throughout the study. The UC also emphasized patients degree of dementia, and the importance to reflect critically and continually on the participative process. It is important that they understand the nature of the research and his /her participation. In addition that patients with dementia, shows the ability to consider alternatives, including the option not to participate in the study. Another concern from the UC is the older adults without dementia that might feel obligated to sign the informed consent and the need for the researcher to clearly state that this is voluntary.

UC has also provided input to the consent form to ensure clear and user-friendly language.

This project will be presented to the UC at their meetings in 2016 and will be regularly informed of the progress and given the opportunity to provide input and advice.

The PhD candidate will attend a course in "User involvement in research and service design" to strengthen her expertise and ensure the safeguarding of the user's role and participation in the research project (25th of May 2016).

Inclusion Criteria:

• Malnourished: According to NICE screening tool:
• BMI <16kg/m2, or weight loss, or little nutritional intake, Or: BMI< 18.5kg/m2, and; weight loss, or little nutritional intake, or alcohol, drug abuse, chemotherapy, antacids diuretics.

Exclusion Criteria:

• Severe dementia (palliative stage).
• Severe kidney disease, or congestive heart failure NYHA class III or IV(medical doctor to decide if patient is eligible for inclusion)
• Already on artificial nutrition
• Terminal stage of life, defined as receiving only palliative medical treatment
• Readmissions (patients will not be included twice)
• Intensive care patients.

Patients and care provider(nurse) will be masked(not informed), Patients will be screened (height and weight) and handgrip strength will be performed by the investigator. Patients will sign a written consent, and then randomized to either the control or an intervention group. Patients will be tested on day 4, 7 and at 3 months by a research assistant/outcome assessor and is blinded treatment allocations. Blood tests are drawn daily. A nurse will administered the tubefeeding according to written orders in the medical journal; nurses are not informed about treatment allocation or the outcome measures. Physicians are informed about the study, but will not know what treatment patients will receive.