- Email: [email protected]
Baseline comparisons in clinical trials
1476
the Crohn’s diease? Could some of the patients’ syndromes, in particular diarrhoea, which accounts for a high percentage of the total CDA score, have been due to other mechanisms such as bile acid induced diarrhoea or secondary to small-bowel resection? The groups were not comparable with respect to disease location. In the vivonex group 50% of patients had small-bowel disease alone and 81 % had small-bowel with or without large-bowel disease; these proportions (29% and 64%) were lower in the polymeric group. Patients with proximal disease seem to respond better to elemental diet, and this may partly explain the higher remission rate in the vivonex group. Giaffer et al do not comment upon long-term remission rates. A therapy which is so labour-intensive, both for patients and staff, should have satisfactory long-term remission rates to justify
widespread use. We have done a controlled, double-blind study with an elemental diet (’Elemental 028’; Scientific Hospital Supplies, Liverpool) and a polymevic (’Enteral 400’, Scientific Hospital Supplies) in fourteen patients with active Crohn’s disease.2 The diets were administered to inpatients by nasogastric tube for 28 days. The clinical remission rate for the polymeric group was significantly better than for the elemental group. Long-term remission rates were disappointing for both diets since only two patients in the polymeric group were well after one year. Other studies have shown that polymeric diets can improve disease activity. Supplements of ’Ensure’ improved disease activity and immune function and corrected malnutrition;3 steroid-resistant patients were able to discontinue drug therapy;4,s and patients with active disease had similar remission rates on polymeric diets as on total parenteral nutrition.6 Gastroenterology Unit, Royal Infirmary, Glasgow G31 2ER, UK
RICHARD PARK ROBIN RUSSELL
1. Teahon 2
3. 4
5.
6.
K, Bjarnason I, Levi AJ. Eleven years experience of elemental diet in active Crohn’s disease. Gut 1988; 29: A1474. Park RHR, Galloway A, Danesh BJZ, Russell RI. Double blind trial comparing elemental and polymeric diet as primary therapy for active Crohn’s disease. Gut 1988; 30: A1453-54 Harries AD, Danis V, Reatley RV, et al. Controlled trial of supplemented oral nutrition in Crohn’s disease. Lancet 1983; i: 887-90. Ginsberg AL, Albert MB. Induction of remission in Crohn’s disease with an Ensure diet: identification and exclusion of dietary substances which exacerbate disease. Gastroenterology 1988; 94: A147. Afdhal NH, Kelly J, McCormick PA, O’Donoghue DP. Remission induction in refractory Crohn’s disease using a high calorie whole diet. J Parenteral Enteral Nutr 1989; 13: 352-65. Greenberg GR, Fleming CR, Jeejeebhoy KN, Rosenberg IH, Sales D, Tremame WJ. Controlled trial of bowel rest and nutritional support in the management of Crohn’s disease. Gut 1988; 29: 1309-15.
***This letter has been shown to Dr Giaffer and colleagues, whose reply follows.-ED. L. SIR,-Dr Park and Dr Russell raise a number of important points. It would be difficult to insist on or to justify patients being kept in hospital for a month. In our experience and that of others craving for normal food usually disappears after the first few days. The temptation to eat ordinary foods would in any case apply to each of the enteric feeds, and we are satisfied that our patients’ initial high motivation to comply was reinforced by the rapid clinical improvement. We appreciate the points on calorie and nitrogen we did not say that we used equal parts of standard and high nitrogen ’Vivonex’, which enables deliverance of 2500 Kcal with 12 g nitrogen, the same applies to ’Fortison’. No supplements were added. The poor correlation between the Crohn’s disease activity index (CDAI) and some laboratory findings is well known;2 the blood indices of activity are all non-specific and are frequently normal in active disease.3 Several of our patients with raised CDAI but normal blood tests had activity confirmed by 99’Tc-labelled neutrophil scans,4and improvement in their scan scores confirmed that shown by the CDAI. We did mention that in patients with initially abnormal tests there were highly significant improvements in serum albumin, erythrocyte sedimentation rate, and alpha-1-acid glycoprotein. We are aware of the possible spurious CDAI
content, and regret that
improvements due to diminished diarrhoea secondary to ileal resection, and excluded such patients from entry to the trial unless there was also some other clear laboratory or endoscopic evidence of inflammation. Park and Russell misquote their reference 1, which in fact showed the response to elemental diet to be independent of either disease site or severity, although the subsequent remission seemed to be for much longer in patients with only small-bowel involvement. An extensive study with a peptide-containing diet confirms this finding,’ and is in accord with our own unpublished experience of
relapse. We do not yet have data on the long-term remission in our patients, but 8 of 17 who responded were in remission on no drugs after six months. Many of these were spared surgery, and we and the patients think this well worthwhile. The relapse rate is probably lower than is seen after remission induced by prednisolone, since in several series drug withdrawal without relapse is achieved in only
20-40%. We agree that
promotion of remission by defined formula preparations may not be unique to elemental diets, and even in our investigation, 5 of 14 responded to fortison. However, of the studies Park and Russell cite in favour of non-elemental diets, reference 5 is a report of one patient, and reference 7 is not comparable with ours, as all patients continued on appreciable doses of prednisolone ranging from 15 to 45 mg per day throughout the study. The study by Parks and colleagues that they mention has been published as an abstract6 but the small number of patients included probably accounts for its unique finding that fewer than 30% (2 of 7 patients) given an elemental diet responded. This is less than published figures for spontaneous remission rates and for the response to the whole-protein preparations in our own trial. We can therefore understand their disappointing conclusion that "there is no benefit from using an elemental diet in active Crohn’s disease",6 but this is at variance with now much published experience. Even so, additional studies are needed to clarify more precisely the factors responsible for the beneficial effect of elemental and possibly other defined formula diets. M. H. GIAFFER Royal Hallamshire Hospital, C. D. HOLDSWORTH Sheffield S10 2JF, UK 1. O’Morain C, Segal AW, Levi AJ. Elemental diet as primary treatment of acute Crohn’s disease: a controlled trial. Br Med J 1984; 288: 1859-62 2. Andre C, Descos L, Landais P, Fermanian J. Assessment of appropriate laboratory measurements to supplement the Crohn’s Disease Activity Index. Gut 1981, 22: 571-74 3 Bartholomeusz FDL, Shearman DJC. Measurement of activity in Crohn’s disease. J Gastroenterol Hepatol 1989; 4: 81-94. 4. Giaffer MH, Holdsworth CD, Tindale WB, Barber DC. Correlation between a computer based method for the quantitation of Tc-99m HMPAO bowel scans and clinical and laboratory activity indices in Crohn’s disease. Gut 1990; 31: A625. 5. Malchow H, Steinhardt HJ, Lorenz-Meyer H, et al. Feasibility and effectiveness of a defined formula diet regimen in treating acuve Crohn’s disease. European co-operative Crohn’s Disease Study, III. Scand J Gastroenterol 1990; 25: 235-44. 6. Park RHR, Galloway A, Danesh BJZ, Russell RI. Double blind trial comparing elemental and polymeric diets as primary therapy for active Crohn’s disease. Gut
1989; 30: A1453-54.
Baseline
comparisons
in clinical trials
SIR,-Dr Senn (March 10, p 614) criticises some of our comments the baseline comparability of in a clinical trial. He suggests that it is not the absolute but the standardised difference (ie, difference divided by its standard error) that is important. This view seems to be based on consideration of the statistical significance of the difference between treatments, whereas our concern is with obtaining an unbiased estimate of the magnitude of the difference. Similar considerations apply to Senn’s comments on the analysis of changes from baseline values. While we agree that analysis of changes from baseline is not necessarily the best method of dealing with baseline differences, in our review we gave papers using this method the benefit of the doubt when classifying their handling of comparability.
(Jan 20,
p
149) about assessing
treatment groups
Medical Statistics Laboratory, Imperial Cancer Research Fund, London WC2A 3PX, UK
DOUGLAS G. ALTMAN
Section of Medical Statistics, Clinical Research Centre, Harrow
CAROLINE
J.
DORÉ
the Crohn’s diease? Could some of the patients’ syndromes, in particular diarrhoea, which accounts for a high percentage of the total CDA score, have been due to other mechanisms such as bile acid induced diarrhoea or secondary to small-bowel resection? The groups were not comparable with respect to disease location. In the vivonex group 50% of patients had small-bowel disease alone and 81 % had small-bowel with or without large-bowel disease; these proportions (29% and 64%) were lower in the polymeric group. Patients with proximal disease seem to respond better to elemental diet, and this may partly explain the higher remission rate in the vivonex group. Giaffer et al do not comment upon long-term remission rates. A therapy which is so labour-intensive, both for patients and staff, should have satisfactory long-term remission rates to justify
widespread use. We have done a controlled, double-blind study with an elemental diet (’Elemental 028’; Scientific Hospital Supplies, Liverpool) and a polymevic (’Enteral 400’, Scientific Hospital Supplies) in fourteen patients with active Crohn’s disease.2 The diets were administered to inpatients by nasogastric tube for 28 days. The clinical remission rate for the polymeric group was significantly better than for the elemental group. Long-term remission rates were disappointing for both diets since only two patients in the polymeric group were well after one year. Other studies have shown that polymeric diets can improve disease activity. Supplements of ’Ensure’ improved disease activity and immune function and corrected malnutrition;3 steroid-resistant patients were able to discontinue drug therapy;4,s and patients with active disease had similar remission rates on polymeric diets as on total parenteral nutrition.6 Gastroenterology Unit, Royal Infirmary, Glasgow G31 2ER, UK
RICHARD PARK ROBIN RUSSELL
1. Teahon 2
3. 4
5.
6.
K, Bjarnason I, Levi AJ. Eleven years experience of elemental diet in active Crohn’s disease. Gut 1988; 29: A1474. Park RHR, Galloway A, Danesh BJZ, Russell RI. Double blind trial comparing elemental and polymeric diet as primary therapy for active Crohn’s disease. Gut 1988; 30: A1453-54 Harries AD, Danis V, Reatley RV, et al. Controlled trial of supplemented oral nutrition in Crohn’s disease. Lancet 1983; i: 887-90. Ginsberg AL, Albert MB. Induction of remission in Crohn’s disease with an Ensure diet: identification and exclusion of dietary substances which exacerbate disease. Gastroenterology 1988; 94: A147. Afdhal NH, Kelly J, McCormick PA, O’Donoghue DP. Remission induction in refractory Crohn’s disease using a high calorie whole diet. J Parenteral Enteral Nutr 1989; 13: 352-65. Greenberg GR, Fleming CR, Jeejeebhoy KN, Rosenberg IH, Sales D, Tremame WJ. Controlled trial of bowel rest and nutritional support in the management of Crohn’s disease. Gut 1988; 29: 1309-15.
***This letter has been shown to Dr Giaffer and colleagues, whose reply follows.-ED. L. SIR,-Dr Park and Dr Russell raise a number of important points. It would be difficult to insist on or to justify patients being kept in hospital for a month. In our experience and that of others craving for normal food usually disappears after the first few days. The temptation to eat ordinary foods would in any case apply to each of the enteric feeds, and we are satisfied that our patients’ initial high motivation to comply was reinforced by the rapid clinical improvement. We appreciate the points on calorie and nitrogen we did not say that we used equal parts of standard and high nitrogen ’Vivonex’, which enables deliverance of 2500 Kcal with 12 g nitrogen, the same applies to ’Fortison’. No supplements were added. The poor correlation between the Crohn’s disease activity index (CDAI) and some laboratory findings is well known;2 the blood indices of activity are all non-specific and are frequently normal in active disease.3 Several of our patients with raised CDAI but normal blood tests had activity confirmed by 99’Tc-labelled neutrophil scans,4and improvement in their scan scores confirmed that shown by the CDAI. We did mention that in patients with initially abnormal tests there were highly significant improvements in serum albumin, erythrocyte sedimentation rate, and alpha-1-acid glycoprotein. We are aware of the possible spurious CDAI
content, and regret that
improvements due to diminished diarrhoea secondary to ileal resection, and excluded such patients from entry to the trial unless there was also some other clear laboratory or endoscopic evidence of inflammation. Park and Russell misquote their reference 1, which in fact showed the response to elemental diet to be independent of either disease site or severity, although the subsequent remission seemed to be for much longer in patients with only small-bowel involvement. An extensive study with a peptide-containing diet confirms this finding,’ and is in accord with our own unpublished experience of
relapse. We do not yet have data on the long-term remission in our patients, but 8 of 17 who responded were in remission on no drugs after six months. Many of these were spared surgery, and we and the patients think this well worthwhile. The relapse rate is probably lower than is seen after remission induced by prednisolone, since in several series drug withdrawal without relapse is achieved in only
20-40%. We agree that
promotion of remission by defined formula preparations may not be unique to elemental diets, and even in our investigation, 5 of 14 responded to fortison. However, of the studies Park and Russell cite in favour of non-elemental diets, reference 5 is a report of one patient, and reference 7 is not comparable with ours, as all patients continued on appreciable doses of prednisolone ranging from 15 to 45 mg per day throughout the study. The study by Parks and colleagues that they mention has been published as an abstract6 but the small number of patients included probably accounts for its unique finding that fewer than 30% (2 of 7 patients) given an elemental diet responded. This is less than published figures for spontaneous remission rates and for the response to the whole-protein preparations in our own trial. We can therefore understand their disappointing conclusion that "there is no benefit from using an elemental diet in active Crohn’s disease",6 but this is at variance with now much published experience. Even so, additional studies are needed to clarify more precisely the factors responsible for the beneficial effect of elemental and possibly other defined formula diets. M. H. GIAFFER Royal Hallamshire Hospital, C. D. HOLDSWORTH Sheffield S10 2JF, UK 1. O’Morain C, Segal AW, Levi AJ. Elemental diet as primary treatment of acute Crohn’s disease: a controlled trial. Br Med J 1984; 288: 1859-62 2. Andre C, Descos L, Landais P, Fermanian J. Assessment of appropriate laboratory measurements to supplement the Crohn’s Disease Activity Index. Gut 1981, 22: 571-74 3 Bartholomeusz FDL, Shearman DJC. Measurement of activity in Crohn’s disease. J Gastroenterol Hepatol 1989; 4: 81-94. 4. Giaffer MH, Holdsworth CD, Tindale WB, Barber DC. Correlation between a computer based method for the quantitation of Tc-99m HMPAO bowel scans and clinical and laboratory activity indices in Crohn’s disease. Gut 1990; 31: A625. 5. Malchow H, Steinhardt HJ, Lorenz-Meyer H, et al. Feasibility and effectiveness of a defined formula diet regimen in treating acuve Crohn’s disease. European co-operative Crohn’s Disease Study, III. Scand J Gastroenterol 1990; 25: 235-44. 6. Park RHR, Galloway A, Danesh BJZ, Russell RI. Double blind trial comparing elemental and polymeric diets as primary therapy for active Crohn’s disease. Gut
1989; 30: A1453-54.
Baseline
comparisons
in clinical trials
SIR,-Dr Senn (March 10, p 614) criticises some of our comments the baseline comparability of in a clinical trial. He suggests that it is not the absolute but the standardised difference (ie, difference divided by its standard error) that is important. This view seems to be based on consideration of the statistical significance of the difference between treatments, whereas our concern is with obtaining an unbiased estimate of the magnitude of the difference. Similar considerations apply to Senn’s comments on the analysis of changes from baseline values. While we agree that analysis of changes from baseline is not necessarily the best method of dealing with baseline differences, in our review we gave papers using this method the benefit of the doubt when classifying their handling of comparability.
(Jan 20,
p
149) about assessing
treatment groups
Medical Statistics Laboratory, Imperial Cancer Research Fund, London WC2A 3PX, UK
DOUGLAS G. ALTMAN
Section of Medical Statistics, Clinical Research Centre, Harrow
CAROLINE
J.
DORÉ