Incidence and Risk Factors of Refeeding Syndrome

Consecutive patients referred to the Department of Otorhinolaryngology and Audiology on suspicion of cancer during the study period from February to August 2011. Patients were asked for participation if they met the following inclusion criteria: age ≥ 18 years, suspicion or diagnosis with HNC, and first time referral. Patients were excluded if not able to understand and speak Danish, having severe dementia and/or renal impairment (women: p-creatinine > 180 μmol/l, men: p-creatinine > 200 μmol/l).

Every participant was observed for eight days after admission or referral to the clinic. Blood samples and interviews were performed at baseline and at day 2, 4 and 7.

Potential factors defining high risk were analyzed by comparing those patients developing RFS or RFF with those that did not. Baseline information consisted of data from the interview and supplemental information in medical records. As a part of the interview, the participant was asked about dietary history in order to assess daily habitual energy end protein intake. Hand grip strength (HGS) was measured, and the participant performed a maximal step test on a 20 cm high bench. The nutritional status was evaluated using the NRS-2002 screening tool (13) and refeeding risk according to the NICE-guidelines (14). Intake of energy, protein, and fluids was registered and compared to the estimated requirements. Standardized questionnaires were used for the interviews, and blood samples were analyzed as part of daily routine in in-patients, while outpatients had their blood samples collected by their general practitioner or by accredited hospital laboratories close to their homes. For participants with a body mass index (BMI) < 30 kg/m2, energy requirement was calculated using Harris \& Benedict's equations for men and women respectively (15). A physical activity factor of 1.1 for bedridden participants and 1.3 for non-bedridden participants was used. Protein requirement was calculated as 18 E-% of total energy requirement. For participants with a BMI ≥ 30 kg/m2 the following equations were used: Energy requirement (bedridden) = 85 kJ/kg body weight/day, energy requirement (non-bedridden) = 100kJ/kg body weight/day, protein requirement (bedridden) = 0.9 g/kg body weight/day, protein requirement (non-bedridden) = 1.1 g/kg body weight/day (16). Energy and protein intake was estimated from the 24-hour dietary recall by using the computer program "Dankost 3000" (Dankost 3000, version 2.5, Danish Catering Center, Herlev, Denmark).

For the step test, an audio file was used to dictate the rhythm, similar to the audio file used in The Danish Step Test (17). HGS was measured on the right hand with a hydraulic Jamar® hand dynamometer (scale 0-200 lbs). Three measurements were conducted with a break of 15 seconds between each measurement, and the mean value was used for analysis. Head and neck pain were measured twice by visual analogue scale (VAS) (18) with added marks for consistent pain and for pain during eating.

All statistical analyses were performed using STATA/IC 10.0 for Windows and Microsoft Excel 2003. The Mann-Whitney rank sum test was used for group comparisons, and dose-response relationships were tested by the Spearman's rank correlation test. Potential risk factors were analyzed by the Chi2-test with Yates correction or the Fisher's exact test as appropriate.

The primary outcomes were changes in p-phosphate and the development of RFS-symptoms after start of nutritional intake. A discriminative value for a clinically relevant hypophosphatemia or decline in p-phosphate was estimated from the coherence between the development of clinical symptoms and the lowest p-phosphate as well as the change in p-phosphate from baseline to the study day with the lowest p-phosphate value.

In case of significant differences between groups, odds ratios (OR) were calculated and Spearman's rank correlation test was used to test for correlation between the potential risk factor and change in p-phosphate, and correlations were tested for linearity by linear regression analysis. Sensitivity, specificity, and predictive values were calculated for NRS-2002 total score ≥ 3, NRS-2002 A-score = 1 and NICE guidelines, respectively.

The maximal increase in energy intake to be allowed without causing RFS was analyzed by correlating changes in p-phosphate and changes in intake in percentage of calculated energy requirements.