3. Results
The research population consisted of 885 adult patients. After screening, 234 patients were randomly assigned to the three treatment groups. The most important reasons for not being eligible were not meeting the inclusion criteria (37%) or not needing ONS after nutritional assessment (21%). All allocated patients received the allocated intervention and 69% of the patients completed the study. There were no deviations from random allocation. For several reasons, 31 (13%) patients refused further use of ONS during the study; nevertheless, they were included in the final analysis. Forty-two patients were excluded from the final analysis because they were discharged from the hospital within two days of follow-up (Fig. 1). The baseline characteristics of the study groups were not different. There were no statistical significant baseline differences between treatment groups between those who completed the study and those who did not (Table 1). Median time ONS were taken was 5 days (range 1e17) (Table 1). No serious adverse events were reported; seven patients died during the study due to their primary underlying disease, but this was not related to undernutrition or the intervention.
Table 2 shows the percentages of patients who reached their treatment objective of energy and protein intake and the differences in the mean intake of ONS per day. Intervention I tended to be less effective than usual care: fewer patients reached their treatment objectives of energy and protein intake than in the usual care
group, however this result was not significant (À16.0%; 95% CI, À33.2 e 1.2, P ¼ 0.075). Also, the mean intake of ONS per day for the intervention I group tended to be lower; again this was not statistically significant. The patients in intervention group II attained their treatment objective significantly more often (23.4%; 95% CI 7.8 e 39.0, p ¼ 0.005) and also had a significantly higher mean intake than the patients in the usual care group. The number needed to treat (NNT) [12] for ONS four times a day with the prescribed medication was 4.3, meaning that four treated patients are needed to have one patient achieve the ONS treatment objective.
The risk difference between the first and second intervention group was 39.4% (95% CI 22.6e56.2). The difference in mean intake per day between the first and second intervention group was also significant (52 ml; 95% CI 23e81 ml). Correction for the 10% fewer patients with surgical conditions in Intervention 2 did not alter the results for all outcomes.
The interrater reliability of the ONS volumes as measured by the nurses and nutrition assistants was almost perfect (intraclass correlation ¼ 0.99).