Objectives: To determine whether routine oral nutritional supplementation of a normal hospital diet improves outcome after stroke (Trial 1); whether early tube feeding improves the outcomes of dysphagic stroke patients (Trial 2); and if tube feeding via a percutaneous endoscopic gastrostomy (PEG) results in better outcomes than that via a nasogastric tube (NG) (Trial 3). Design: The Feed Or Ordinary Diet (FOOD) trial was a family of three pragmatic, randomised controlled trials (RCTs). They shared facilities for randomisation, data collection, follow-up and coordination. Patients could be co-enrolled in more than one of these trials. Setting: Patients were enrolled in 131 hospitals in 18 countries. Participants: A total of 5033 patients who had been admitted to hospital with a recent stroke were enrolled in the trials between November 1996 and July 2003. Interventions: In Trial 1, patients who could swallow within the first 30 days of admission were allocated to normal hospital diet versus normal hospital diet plus oral nutritional supplements (equivalent to 360 ml of 1.5 kcal/ml, 20 g of protein per day) until hospital discharge. In Trial 2, dysphagic patients enrolled within 7 days of admission were allocated to early enteral tube feeding versus avoid any enteral tube feeding for at least 1 week. In Trial 3, dysphagic patients were allocated within 30 days of admission to receive enteral tube feeding via PEG versus NG. Main outcome measures: Survival and the modified Rankin scale (MRS), a measure of functional outcome (grade 0 indicating no symptoms and grade 5 indicating severe disability, requiring help day and night). The primary outcomes were measured 6 months after enrolment, blind to treatment allocation, by the patient or their proxy completing a postal or telephone questionnaire. Results: In Trial 1, 4023 patients were enrolled by 125 hospitals in 15 countries. Only 314 (7.8%) patients were judged undernourished at baseline. Vital status and MRS at the end of the trial were known for 4012 (99.7%)and 4004 (99.5%), respectively. Of the 2007 allocated normal hospital diet, 253 (12.6%) died, 918 (45.7%) were alive with poor outcome (MRS 3-5) and 823 (41.1%) had a good outcome (MRS 0-2). Of the 2016 allocated oral supplements, 241 (12.0%) died, 953 (47.3%) were alive with poor outcome and 813 (40.4%) had a good outcome. The supplemented diet was associated with an absolute reduction in risk of death of 0.7% (95% CI -1.4 to 2.7; p = 0.5) and a 0.7% (95% CI -2.3 to 3.8, p = 0.6) increased risk of death or poor outcome. In Trial 2, a total of 859 patients were enrolled by 83 hospitals in 15 countries. MRS at the end of the trial was known for 858 (99.9%). At follow-up, of 429 allocated early tube feeding, 182 (42.4%) died, 157 (36.6%) were alive with poor outcome (MRS 4-5) and 90 (21.0%) had a good outcome (MRS 0-3). Of 430 allocated avoid tube feeding 207 (48.1%) died, 137 (31.9%) were alive with poor outcome and 85 (19.8%) had a good outcome. Early tube feeding was associated with an absolute reduction in risk of death of 5.8% (95% CI -0.8 to 12.5; p = 0.09) and a reduction in death or poor outcome of 1.2% (95% CI -4.2 to 6.6; p = 0.7). In Trial 3, 321 patients were enrolled by 47 hospitals in 11 countries. Of 162 allocated PEG, 79 (48.8%) died, 65 (40.1%) were alive with poor outcome and 18 (11.1%) had good outcome. Of 159 allocated NG, 76 (47.8%) died, 53 (33.3%) were alive with poor outcome and 30 (18.9%) had good outcome. PEG was associated with an increase in absolute risk of death of 1.0% (95% CI -10.0 to 11.9; p = 0.9) and an increased risk of death or poor outcome of 7.8% (95% CI 0.0 to 15.5; p = 0.05). Conclusions: The results of Trial 1 would be compatible with oral supplementation being associated with a 1-2% absolute benefit or harm, but do not support routine supplementation of hospital diet for unselected stroke patients who are predominantly well nourished on admission. In Trial 2, the data suggest that a policy of early tube feeding may substantially reduce the risk of dying after stroke and it is very unlikely that the alternative policy of avoiding early tube feeding would significantly improve survival. Improved survival may be at the expense of increasing the proportion surviving with poor outcome. These data might usefully inform the difficult discussions about whether or not to feed a patient with a severe stroke. In Trial 3, the data suggest that in the first 2-3 weeks after acute stroke, better functional outcomes result from feeding via NG tube than PEG tube, although there was no major difference in survival. These data do not support a policy of early initiation of PEG feeding in dysphagic stroke patients. Future research might be focused on making NG tube feeding safer and more effective, also studies need to confirm the increased risk of gastrointestinal haemorrhage associated with tube feeding and, if confirmed, establish whether any interventions might reduce this risk. Future work might also aim to establish why worse functional outcomes occurred in PEG-fed patients because patients with prolonged dysphagia or intolerance of an NG tube are inevitably fed via a PEG tube.
