Introduction

Patient feeding policies, following recent stroke, vary between individual hospitals and between clinicians at the same hospital. More specifically, there is often variation in the timing and method of feeding patients with stroke and this reflects the lack of reliable evidence on what is the optimal feeding strategy.

Poor nutrition is a common and under-recognised problem in patients admitted to hospital as well as in those who remain in hospital for prolonged periods (Albiin et al 1982, Sandstrom et al 1985, Cederholm & Hellstrom 1992). It is particularly frequent amongst elderly patients. It has been associated with reduced muscle strength, reduced resistance to infection and impaired wound healing (Fiatarone & Evans 1993, Potter et al 1995). Among patients with stroke, most of whom are elderly, muscle weakness and infections are very common (Davenport et al 1996). It is conceivable that malnutrition could increase the frequency of these problems and result in poorer outcomes. It is not surprising, therefore, that several workers have investigated the nutritional status of patients with recent stroke. The reported frequency of malnutrition has varied between 8% and 40% although much of this variation may be due to differences in case mix, the definitions of malnutrition, and the methods of assessment (Axelsson et al 1988, Smithard et al 1993, Unosson et al 1994, Davalos et al 1996). Furthermore, any acute illness may be responsible for a negative energy balance and greater nutritional demands and patients with stroke may be less able to meet these increased demands (Klipstein-Grobusch et al 1995). To compound the general problem of malnutrition, it has been estimated that up to 45% of hospitalised patients with stroke are unable to swallow safely, although again the reported frequency depends on the selection of cases, the timing of assessments, the sensitivity of the method used to detect swallowing problems (Gordon et al 1987, Barer 1989). Of course, even patients who are capable of swallowing liquids and food may have a poor appetite because of the effects of intercurrent illness or medication. Patients may eat more slowly because of facial weakness, lack of dentures or poor arm function. All these factors may contribute to the worsening in nutritional status which has been observed by several groups during hospital admission for stroke (Axelsson et al 1989, Smithard et al 1993, Unosson et al 1994, Davalos et al 1996). Therefore, there seems to be good evidence that a significant proportion of patients admitted to hospital with recent stroke are malnourished and that their nutritional status may further deteriorate during the admission. However, it is less clear whether this worsens patient outcomes. There is little doubt that the outcome of patients undergoing emergency surgery, and of those with other serious illness, depend on their nutritional status and nowadays careful attention is paid to their nutrition. Few studies have investigated the influence of nutrition on outcome but malnutrition has been associated with an increased risk of death after stroke (Davalos et al 1996).

If nutrition is an important determinant of outcome in the physically ill, and in particular those with stroke, the next question is whether, by improving patients’ nutrition, one might improve their outcome. There have been a large number of randomised trials, in a variety of settings, testing the effects of improving nutritional status. Most of these studies have been individually too small to demonstrate an effect but a recent systematic review of all of the available randomised trials suggests that oral or enteral (i.e. via a feeding tube) nutritional supplementation improves nutritional indicators and reduces the odds of death by 34% (95% CI 9% - 52%) (Potter et al 1996). However, this review included trials of differing methodological quality which tested various interventions in different types of patients. It also included relatively few small negative trials which may reflect a degree of publication bias and so be responsible for an over-optimistic estimate of treatment effect. None of these studies were specifically in patients with stroke and few patients with stroke were included in them. One non-randomised trial suggested that early enteral nutrition after stroke reduced length of stay in hospital but methodological limitations make this conclusion unreliable (Nyswonger & Helmchen 1992).

Even if it were shown that improving nutritional status after stroke would improve outcome, there still  remain questions about when to start any supplementary feeding regime and the best way to deliver it. This applies particularly to the important minority of patients who cannot swallow safely. Indeed, increasing emphasis has been placed on detecting those patients with swallowing difficulty so their risk of aspiration pneumonia can be reduced. This is usually done by restricting oral intake and providing fluids, and sometimes food, by alternative routes. Swallowing usually recovers over the first few days or weeks which allows patients to safely take fluids and food, if necessary with a modified consistency (Gordon et al 1987, Barer 1989). However, even during this recovery phase, patients’ fluid and food intake may be inadequate and some supplementation by an alternative route may be helpful.

Supplementation might be achieved by intravenous feeding but in practice this is rarely used or justified in patients with stroke who generally have a gastrointestinal tract which is well able to absorb nutrients. NG tubes are often inserted to allow fluid and food to be given to patients. However, in patients who are unable to swallow, they are not always easy to insert and perhaps because they are uncomfortable they are often pulled out by patients and have to be replaced. This adds to patient distress and interrupts any feeding regime. Furthermore, NG tubes may become displaced and cause aspiration, as well as ulceration of the nostril if use is prolonged. Some workers therefore advocate the increased use of PEG (O’Mahony & McIntyre 1995) which can be performed with little or no sedation and provides an effective and quite acceptable method of enteral feeding. However, PEG is more invasive than NG and has its own complications including aspiration, peritonitis, wound infection and haemorrhage. Indeed there is a low (about 1%), but not insignificant, risk of death related to the procedure (Larson et al 1987, Miller et al 1989, Finucane et al 1991, Pender et al 1993). The published complication rates are low but may not reflect the rates in less specialised centres which do not publish their results (Wanklyn et al 1995). Two randomised comparisons of NG and PEG tube feeding have suggested that the latter provides more effective nutritional support with less interruption of feeding (Park et al 1992, Norton et al 1996). One trial was in patients with severe stroke and showed that those fed by PEG had an implausibly large (70% relative) reduction in case fatality compared with those fed via NG tube (Norton et al 1996). However, this trial only included 30 patients and little data were provided to allow any assessment of the effectiveness of randomisation. It seems most likely that some imbalance in baseline factors accounted for much of the observed difference in outcome. However, despite its limitations, this trial raises important issues about the best way to feed patients with stroke who cannot swallow safely.

Difficulties in feeding patients with stroke who cannot swallow safely mean that feeding is sometimes delayed for perhaps a week or two and only parenteral (intravenous (IV) or subcutaneous (SC)) fluids are given. During this time many patients will improve enough to be able to take at least some food and therefore avoid or reduce the need for tube feeding. On the other hand, some clinicians prefer to introduce tube feeding very soon after the stroke although many would reserve PEG feeding for those who seem likely to require prolonged tube feeding. However, as PEG feeding becomes more widely available, it is being used earlier. The pros and cons of NG and PEG feeding after stroke have recently been reviewed but with no definite conclusions (O’Mahony & McIntyre 1995).

A recent survey of clinical practice in the UK demonstrated wide variation in the timing and method of feeding in dysphagic patients with stroke which probably reflects the lack of firm evidence that any one policy is superior (Hussein et al 1995). These authors concluded that there was a need for randomised trials to establish the place of tube feeding after stroke. Against this background, we are undertaking a "family" of large randomised trials to determine the optimum feeding policies for patients with stroke.

FOOD comprises three large, simple, multicentre, randomised trials.

Trial 1 addresses the question For those who can take adequate fluids orally should we routinely supplement the normal hospital diet? This question is relevant to the majority of patients who can swallow on admission and also to those who survive to regain a safe swallow after a period of swallowing difficulty. Both groups may benefit from nutritional supplementation since even when patients can swallow they may not eat enough for a variety of reasons. We plan to randomise patients in the first month of admission between:

Normal hospital diet vs. Normal hospital diet plus oral supplements until hospital discharge.

Normal diet is that which is normally provided to patients and may be of altered consistency (e.g. for those with swallowing difficulties) or composition (e.g. for patients with special needs e.g. diabetics). Patients randomised to a normal hospital diet should not have nutritional supplements prescribed on their drug chart, although, if supplementation is the norm in a hospital, this might be continued as long as patients allocated normal hospital diet plus nutritional supplementation receive the prescribed supplement in addition to those routinely given. Oral supplements comprise 120ml of a supplement containing 1.5kcal/ml three times a day prescribed on the drug chart. We have shown in our pilot studies that this approach is practical (Reilly et al 1995), provides patients with an extra 540kcals per day and the use of drug charts allows us to monitor compliance.

Trial 2 addresses the question Does early initiation of tube feeding benefit patients?

This is relevant to 30 to 40% of stroke patients admitted to hospital who cannot safely take adequate diet and fluids orally. We plan to randomise patients within the first week of their admission between:

Immediate tube feeding vs. Delay tube feeding for at least a week and hydrate using parenteral fluids.

If randomised to immediate tube feeding, the clinician may choose the type of tube or alternatively co-enrol the patient into Trial 3 (NG vs PEG). The tube feeding should be started as soon as possible and certainly within three days of randomisation. The liquid feed would be that normally used at that institution and given in consultation with a dietitian. Patients randomised to delayed tube feeding should not have tube feeding started for at least a week and should be hydrated using parenteral fluids (IV or SC) given according to local protocols. The randomising clinician decides if and when tube feeding should start after the week has elapsed. Inevitably, some patients may be taking some oral fluids or food whilst still being fed predominantly via a tube or whilst receiving parenteral hydration. Patients do not have to remain ‘nil by mouth’.

Trial 3 addresses the question Is tube feeding via a PEG better than that via an NG tube?

This is relevant to all stroke patients who cannot safely take adequate diet or fluids orally. We plan to randomise patients within the first month of the hospital admission between:

PEG vs. NG tube feeding.

NG tubes may be wide or small bore. Percutaneous tubes may be inserted endoscopically or radiologically, into the stomach or jejunum according to local practice. The tube feeding should be started as soon as possible and certainly within three days of randomisation. The liquid feed would be that normally used at that institution and given in consultation with a dietitian.

The oral supplements (Trial 1) should normally be continued until hospital discharge. However, the responsible clinician may choose to stop supplements earlier if, for example, the patient is gaining excessive weight. Tube feeding should continue until the responsible clinician decides that the patient is taking adequate diet orally or that further tube feeding is futile. The reason for stopping the feeding regime should be recorded on the Hospital Discharge Form (Appendix C). This form should be completed on discharge from hospital, death or transfer out of the randomising centre, although the allocated feeding policy can be continued after discharge or transfer, if appropriate. Details of all types of feeding given since randomisation should be recorded on the Hospital Discharge Form (Appendix C), including those feeding regimens not randomly allocated.

If the patient has been randomised to one feeding policy but this subsequently becomes impractical or the clinician becomes certain that an alternative is better then the clinician may change the method of feeding, although our analyses will be based on intention-to-treat. Data on how often, and why, feeding policies are changed will inform our final analyses.

Any patient admitted to hospital with a stroke (excluding those with subarachnoid haemorrhage) within a week of onset, in whom the randomising clinician is substantially uncertain about the best feeding policy.

Patients can be randomised into Trial 2 (Immediate tube vs. Delay) within the first week of admission (or a stroke or recurrent stroke which occurs during hospital admission). For Trials 1 (Normal hospital diet vs. Oral supplements) and 3 (NG vs. PEG), patients can be randomised within a month (30 days) of hospital admission (or a stroke or recurrent stroke which occurs during hospital admission).

Patients who, in the opinion of the responsible clinician, are unlikely to benefit from nutritional supplementation or from PEG or NG feeding.

UK Multicentre Research Ethical Committee (MREC) approval has been granted. Each collaborating centre will need to confirm local ethics committee approval. Patients (or their carers) will be given a Patient Information Booklet (Appendix F) which describes the aims of the trial and the potential risks and benefits of a variety of feeding policies. The patients (or their carers) will be given enough time to consider the trial fully and ask any questions they may have about the implications of the trial.

Consent procedures will vary from centre to centre but they will have to be approved by the local Ethics Committee. It is generally recommended that informed consent should be obtained from the patients if they are able to understand and communicate effectively. Alternatively, a close relative may give approval/ agreement to participate in the trial. If the patient is unable to express his/her wishes and there are no close relatives, an independent clinician can be sought to provide approval/agreement.

Data, collected at baseline, will include centre identifiers, patient identifiers and information which will provide a prediction of outcome (e.g. is the patient able to lift both arms off the bed?). Data required to allow minimisation (see below) and to calculate the delay from stroke onset to hospital admission and randomisation will also be collected. A swallowing assessment, carried out by the randomising clinician or a member of his/her team, will determine which trial(s) the patient can be entered into. The swallowing assessment should be performed in line with local guidelines, but as a minimum should comprise a bedside assessment.

Co-enrolment - Randomising a patient into more than one of these three

Of course, patients cannot be randomised twice in the same trial during the course of their hospital admission. To enter a patient into another of the three trials later in the admission, the randomising clinician would simply complete another randomisation form and telephone the randomisation service again.

At discharge, transfer from the randomising centre or death, the randomising clinician, or the trial support staff in that centre will review the case notes and drug charts, complete a Hospital Discharge Form and return it to the FOOD Trial Co-ordinating Centre in Edinburgh. Clinicians who routinely transfer their patients with stroke to another ward/unit/hospital very soon after admission will need to reach an agreement with that ward/unit/hospital such that the allocated feeding regimen will be continued and that the Hospital Discharge Form will be completed on discharge, death or transfer from that ward/unit/hospital.

These data should be available from the medical or nursing notes.

The primary outcome for Trial 1 will be the proportion of patients who are surviving free of dependency (defined as a Modified Rankin <3) six months after first randomisation.

The primary outcome for Trials 2 and 3 will be the proportion of patients surviving free of severe disability (defined as a Modified Rankin <4) six months after first randomisation.

Trial 1 (Normal hospital diet vs. Oral Supplements).

We plan to randomise at least 6000 patients divided equally between the two groups which will provide us with at least 80% power to detect an increase in the proportion of patients surviving free of dependency (Modified Rankin <3) from 52% to 56% when the null hypothesis is rejected at p-values of 0.05 and below (i.e. a =0.05, b =0.2).

Trial 2 (Immediate tube vs. Delay).

We plan to randomise at least 2000 patients divided equally between the two groups which will provide us with at least 80% power to detect an increase in the proportion of patients surviving free of severe disability (Modified Rankin <4) from 30% to 36% when the null hypothesis is rejected at p-values of 0.05 and below (i.e. a =0.05, b =0.2).

Trial 3 (NG vs. PEG).

We plan to randomise at least 1000 patients divided equally between the two groups which will provide us with at least 80% power to detect an increase in the proportion of patients surviving free of severe disability (Modified Rankin <4) from 30% to 39% when the null hypothesis is rejected at p-values of 0.05 and below (i.e. a =0.05, b =0.2).

We plan to continue to randomise patients into each of the three trials until we have achieved these minimum sample sizes in all three trials. Thus it is likely that we will exceed these sample size estimations in two of the trials to allow us to detect more modest treatment effects. Our Data Monitoring Committee may advise us to stop or prolong randomisation in any one of the three trials depending on the results of their confidential interim analyses.

FOOD Trial Co-ordinating Centre Personnel

Principal Investigator: Dr. Martin Dennis

Trial Co-ordinator: Gina Cranswick

Trial Statistician: Dave Signorini

Trial Programmer: Vera Soosay

Steering Committee

The trial will be managed and co-ordinated by a combined scientific and administrative Steering Committee. The scientific advisory group, with a particular interest in nutritional problems, will comprise: Campbell Chalmers, Martin Dennis (Chair), John Forbes, Subrata Ghosh, Peter Langhorne, Carole Ann McAteer, Jean McIntyre, Paul O’Neill, Jan Potter and Margaret Roberts.

The administrative, data management and trial development group will comprise: Gina Cranswick, Martin Dennis, Barbara Farrell, Anne Leigh Brown, Dave Signorini, Vera Soosay and Charles Warlow (Chair).

The Data Monitoring Committee comprises: Professor C Bulpitt (London), Professor A Grant (Aberdeen, Chair), Professor G Murray (Edinburgh) and Dr P Sandercock (Edinburgh).

During the period of recruitment into the trial, interim analyses of the proportion of patients surviving free of dependency/severe disability as well as data available on other major outcome events will be supplied, in strictest confidence, to the chairman of the Data Monitoring Committee, along with any other analyses that the Committee may request. In the light of these analyses, the Data Monitoring Committee will advise the chairman of the Steering Committee if, in their view, the randomised comparisons have provided both (i) ‘proof beyond reasonable doubt’ that for all, or some, the intervention is clearly indicated or clearly contra-indicated and (ii) evidence that might reasonably be expected to materially influence patient management in normal practice. Appropriate criteria of proof beyond reasonable doubt cannot be specified precisely, but some members of the committee have expressed sympathy with the view that a difference of at least 3 standard deviations in an interim analysis of a major outcome event may be needed to justify halting, or modifying, such a study prematurely. If this criterion were to be adopted, it would have the practical advantage that the exact number of interim analyses would be of little importance, and so no fixed schedule is proposed. The Steering Committee can decide whether to modify intake to the trial (or seek extra data). Unless this happens, however, the Steering Committee, the collaborators and central administrative staff will remain ignorant of the interim results.

All publications relating to the main trial will be published in the name of the International Stroke Trials (IST) Collaboration - FOOD.

Abstracts relating to the main study will be submitted as the International Stroke Trials (IST) Collaboration - FOOD along with the presenter’s name.

Papers and abstracts relating to ‘Add-on’ studies will be in the name of those collaborators who took part or the group’s name, but recognise the input of the entire Collaboration by putting ‘part’, ‘member’ or ‘on behalf of’ the International Stroke Trials (IST) Collaboration - FOOD.

Anyone wishing to use the data generated from this trial for higher degrees, PhDs etc. must first seek the permission of the Steering Committee. All papers must be approved by the Steering Committee prior to submission for publication. Anyone wishing to use the data in this way, will be asked to sign a confidentiality agreement which will prevent them from publishing the data until the results of the main trial have been published.

No group of collaborators should publish the results of any sub-study which splits patients by treatment allocation without the agreement of the Steering Committee, on behalf of the other members of the Collaboration. Studies which report any of the process or outcome data collected as part of the main study must acknowledge the collaboration as an author e.g. Smith on behalf of the International Stroke Trials (IST) Collaboration - FOOD.

Albiin, N., Asplund, K. and Bjermer, L. (1982) Nutritional status of medical patients on emergency admission to hospital. Acta Med Scand 212:151-156

Axelsson, K., Asplund, K., Norberg, A., Alafuzoff, I. (1988) Nutritional status in patients with acute stroke. Acta med Scand 224:217-224

Axelsson, K., Asplund, K., Norberg, A., Eriksson, S. (1989) Eating problems and nutritional status during hospital stay of patients with severe stroke. Journal of American Dietary Association. 8:1092-6

Barer, D.H. (1989) The natural history and functional consequences of dysphagia after hemispheric stroke. Journal of Neurology, Neurosurgery and Psychiatry 52:236-241

Carver, A.D. (1996) Can nurses identify nutritionally depleted elderly patients? Abstract presented at Annual British Dietetic Association Conference, Stratford Upon Avon. Publication pending.

Cederholm, T. and Hellstrom, K. (1992) Nutritional status in recently hospitalized and free-living elderly subjects. Gerontology 38:105-110

Davalos, A., Ricart, W., Gonzalez-Huix, F., Soler, S., Marrugat, J., Molins, A., Suner, R,, and Genis, D. (1996) Effect of malnutrition after acute stroke on clinical outcome. Stroke 27: 1028-1032

Davenport, R.J., Dennis, M.S., Wellwood, I. and Warlow, C.P. (1996) Complications following acute stroke. Stroke 27: 415-420

Fiatarone, M.A., Evans, W.J. (1993) The etiology and reversibility of muscle dysfunction in the aged. The Journal of Gerontology Vol. 48 (Special issue):77-83

Finucane, P., Aslan, S.M. and Duncan, D. (1991) Percutaneous endoscopic gastrostomy in elderly patients. Postgrad Med Journal 67:371-373

Gordon, C., Langton Hewer, R, and Wade, D.T. (1987) Dysphagia in acute stroke. British Medical Journal 295:411-414

Hussain, A., Rodgers, H., Barer, D. (1995) Management of dysphagia in stroke patients: A postal survey. British Geriatrics Society 24 (Suppl 1): P5

Klipstein-Grobusch, K., Reilly, J.J., Potter, J., Edwards, C.A., Roberts, M.A. (1995) Energy intake and expenditure in elderly patients admitted to hospital with acute illness. British Journal of Nutrition 73:323-334

Larson, D.E., Burton, D.D., Schroeder, K.W., (1987) Percutaneous endoscopic gastrostomy. Gastroenterology 93:48-52

Miller, R.E., Castlemain, B., Lacqua, F.J., Kotler, D.P. (1989) Percutaneous endoscopic gastrostomy. Surgical Endoscopy 3:186-90

Norton, B., Homer-Ward M., Donnelly, M.T., Long, R.G., Holmes, G.K.T. (1996) A randomised prospective comparison of percutaneous endoscopic gastrostomy and nasogastric tube feeding after acute dyshagic stroke. British Medical Journal 312:13-16

Nyswonger, G.D., and Helmchen, R.H. (1992) Early enteral nutrition and length of stay in stroke patients. J Neurosci Nurs 24:220-223

O’Mahony, D., McIntyre, A.S. (1995) Artificial feeding for elderly patients after stroke. Age & Ageing 24:533-535 Park, R.H., Allison, M.C., Lang, J., Spence, E., Morris, A.J., Danesh, B.J., Russell, R.I., Mills, P.R. (1992) Randomised comparison of percutaneous endoscopic gastrostomy and nasogastric tube feeding in patients with persisting neurological dysphagia. British Medical Journal 304 (689):1406-9

Pender, S.M., Courtney, M.G., Rajan, E., McAdam, B. and Fielding, J.F. (1993) Percutaneous endoscopic gastrostomy -results of an Irish single unit series. Ir J Med Sci 162:452-455

Peto, R., Pike, M.C. et al Design and analysis of randomised clinical trials requiring prolonged observation of each patient Part 1: Introduction and design. Br J Cancer 1976; 34: 585-612, and Part II: Analysis. Br J Cancer 1977; 35: 1-39

Pocock, S.J., (1983) Clinical Trials: A practical approach, John Wiley and Sons, Chichester

Potter, J., Langhorne, P. (1996) Nutritional supplementation in the elderly: A statistical overview. Age & Ageing 25 (Suppl 1): 42

Potter, J., Klipstein, K., Reilly, J.J., Roberts, M. (1995) The nutritional status and clinical course of acute admissions to a geriatric unit. Age & Ageing 24:131-136

Reilly, J.J., Mackintosh, M., Potter, J., and Roberts, M.A. (1995) An evalution of the feasibility of sip-feed supplementation in undernourished, acutely sick, elderly patients. Proc Nutr Soc 54:135A

Sandstrom, B., Alhaug, J., Einarsdottir, K., Simpura, E-M. and Isaksson, B. (1985) Nutritional status, energy and protein intake in general medical patients in three Nordic hospitals. Hum Nutr: Appl Nutr 39A:87-95

Smithard, D.G., Renwick, D. and O’Neill, P.A. (1993) Change in nutritional status following acute stroke. (Abstract) Age & Ageing 22 (supp no 3):11

Unosson, M., Ek, A.C., Bjurulf, P., von Schenck, H. and Larsson, J. (1994) Feeding dependence and nutritional status after acute stroke. Stroke 25:366-371

Wanklyn, P., Niall, C., Belfield, P. (1995) Outcome in patients who require a gastrostomy after stroke. Age & Ageing 24: 510-514