FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W1966302620> ?p ?o ?g. }

• W1966302620 endingPage "837" @default.
• W1966302620 startingPage "837" @default.
• W1966302620 abstract "Authors' replySir—Randomisation to the EUCLID trial was well underway long before either of the guidelines to which Kurt Stoschitzky refers were published, and were appropriate to practice at the time the study was designed. Ethical approval for a placebo-controlled trial was obtained from all centres, and no ethics committee raised objections to our patient inclusion criteria. At randomisation, 93% of EUCLID patients had a systolic blood pressure of less than 140 mm Hg, and 64% had a systolic pressure of less than 130 mm Hg and diastolic less than 85 mm Hg (optimum blood pressure for people with diabetes according to JNC VI guidelines1National Institutes of Health, National Heart Lung, and Blood Institute, National High Blood Pressure Education ProgrammeThe sixth report of the Joint National Committee on prevention, detection, evaluation and treatment of high blood pressure. NIH, Bethesda1997Google Scholar).Data on absolute blood pressure values were presented in the first EUCLID publication,2The EUCLID Study GroupRandomised placebo-controlled trial of lisinopril in normotensive patients with insulin-dependent diabetes and normoalbuminuria or microalbuminuria.Lancet. 1997; 349: 1787-1792Summary Full Text Full Text PDF PubMed Scopus (518) Google Scholar but it was the blood pressure difference between placebo and lisinopril groups that was included in the statistical analysis, so there was little need to represent absolute values. Interpreting clinical trials on the basis of p values alone can be misleading, and this approach has rightly been criticised.3Altman DG Bland JM Absence of evidence is not evidence of absence.BMJ. 1995; 311: 485Crossref PubMed Scopus (1156) Google Scholar The centreadjusted odds ratio for progression of retinopathy by one level in EUCLID was 0·50 (95% CI 0·28–0·92, p=0·03). In other words, lisinopril halved the progression of retinopathy. The p value of 0·03 shows this outcome could occur three times in a 100 if there was really no difference in treatment effect between placebo and lisinopril. We would therefore conclude that there is reasonable evidence for an important beneficial effect of lisinopril on retinopathy. We then added HbA1c to the model to assess its role as a confounder. The most important point when assessing the impact of a confounder is to examine the impact that adjustment has on the odds ratio, not the p value; thus a change from 0·50 to 0·55 clearly does not indicate that the treatment effect is totally accounted for by the lower HbA1c in the lisinopril group.A similar argument applies to the adjustment made for the treatment difference in blood pressure. This adjustment was made to address a mechanistic question; is the beneficial effect of lisinopril mediated wholly via its effects on systemic blood pressure, or are other mechanisms involved? The resulting odds ratio of 0·57 suggests that only part of the impact on retinopathy was mediated via changes in systemic blood pressure.Stoschitzky's assertion that we have merely shown that lower HbA1c and blood pressure slow progression of retinopathy is based on a misinterpretation of basic clinical trial statistics, and is a clear example of how important clinical findings have been ignored as a consequence of this exclusive focus on the p value.3Altman DG Bland JM Absence of evidence is not evidence of absence.BMJ. 1995; 311: 485Crossref PubMed Scopus (1156) Google Scholar Authors' reply Sir—Randomisation to the EUCLID trial was well underway long before either of the guidelines to which Kurt Stoschitzky refers were published, and were appropriate to practice at the time the study was designed. Ethical approval for a placebo-controlled trial was obtained from all centres, and no ethics committee raised objections to our patient inclusion criteria. At randomisation, 93% of EUCLID patients had a systolic blood pressure of less than 140 mm Hg, and 64% had a systolic pressure of less than 130 mm Hg and diastolic less than 85 mm Hg (optimum blood pressure for people with diabetes according to JNC VI guidelines1National Institutes of Health, National Heart Lung, and Blood Institute, National High Blood Pressure Education ProgrammeThe sixth report of the Joint National Committee on prevention, detection, evaluation and treatment of high blood pressure. NIH, Bethesda1997Google Scholar). Data on absolute blood pressure values were presented in the first EUCLID publication,2The EUCLID Study GroupRandomised placebo-controlled trial of lisinopril in normotensive patients with insulin-dependent diabetes and normoalbuminuria or microalbuminuria.Lancet. 1997; 349: 1787-1792Summary Full Text Full Text PDF PubMed Scopus (518) Google Scholar but it was the blood pressure difference between placebo and lisinopril groups that was included in the statistical analysis, so there was little need to represent absolute values. Interpreting clinical trials on the basis of p values alone can be misleading, and this approach has rightly been criticised.3Altman DG Bland JM Absence of evidence is not evidence of absence.BMJ. 1995; 311: 485Crossref PubMed Scopus (1156) Google Scholar The centreadjusted odds ratio for progression of retinopathy by one level in EUCLID was 0·50 (95% CI 0·28–0·92, p=0·03). In other words, lisinopril halved the progression of retinopathy. The p value of 0·03 shows this outcome could occur three times in a 100 if there was really no difference in treatment effect between placebo and lisinopril. We would therefore conclude that there is reasonable evidence for an important beneficial effect of lisinopril on retinopathy. We then added HbA1c to the model to assess its role as a confounder. The most important point when assessing the impact of a confounder is to examine the impact that adjustment has on the odds ratio, not the p value; thus a change from 0·50 to 0·55 clearly does not indicate that the treatment effect is totally accounted for by the lower HbA1c in the lisinopril group. A similar argument applies to the adjustment made for the treatment difference in blood pressure. This adjustment was made to address a mechanistic question; is the beneficial effect of lisinopril mediated wholly via its effects on systemic blood pressure, or are other mechanisms involved? The resulting odds ratio of 0·57 suggests that only part of the impact on retinopathy was mediated via changes in systemic blood pressure. Stoschitzky's assertion that we have merely shown that lower HbA1c and blood pressure slow progression of retinopathy is based on a misinterpretation of basic clinical trial statistics, and is a clear example of how important clinical findings have been ignored as a consequence of this exclusive focus on the p value.3Altman DG Bland JM Absence of evidence is not evidence of absence.BMJ. 1995; 311: 485Crossref PubMed Scopus (1156) Google Scholar Angiotensin II, VEGF, and diabetic retinopathyNish Chaturvedi and colleagues (Jan 3, p 28)1 claim that their trial was done in normotensive individuals. However, they included patients with a systolic blood pressure up to 155 mmHg, but this limit is higher than that of the guidelines issued by WHO/International Society of Hypertension2 and the JNC VI3 which define normotension as a systolic blood pressure of less than 140 mm Hg (<120/<80 mm Hg being the optimum3). Is this upper limit acceptable, especially since Sawicki4 has drawn attention to the same mistake in the EUCLID study?5 Full-Text PDF" @default.
• W1966302620 created "2016-06-24" @default.
• W1966302620 creator A5089503693 @default.
• W1966302620 creator A5091135967 @default.
• W1966302620 date "1998-03-01" @default.
• W1966302620 modified "2023-09-25" @default.
• W1966302620 title "Angiotensin II, VEGF, and diabetic retinopathy" @default.
• W1966302620 cites W1525161423 @default.
• W1966302620 cites W2101337813 @default.
• W1966302620 cites W2414167269 @default.
• W1966302620 doi "https://doi.org/10.1016/s0140-6736(05)78973-x" @default.
• W1966302620 hasPublicationYear "1998" @default.
• W1966302620 type Work @default.
• W1966302620 sameAs 1966302620 @default.
• W1966302620 citedByCount "1" @default.
• W1966302620 crossrefType "journal-article" @default.
• W1966302620 hasAuthorship W1966302620A5089503693 @default.
• W1966302620 hasAuthorship W1966302620A5091135967 @default.
• W1966302620 hasConcept C118487528 @default.
• W1966302620 hasConcept C126322002 @default.
• W1966302620 hasConcept C134018914 @default.
• W1966302620 hasConcept C167734588 @default.
• W1966302620 hasConcept C2776004912 @default.
• W1966302620 hasConcept C2778313320 @default.
• W1966302620 hasConcept C2779829184 @default.
• W1966302620 hasConcept C2908929049 @default.
• W1966302620 hasConcept C555293320 @default.
• W1966302620 hasConcept C71924100 @default.
• W1966302620 hasConcept C84393581 @default.
• W1966302620 hasConceptScore W1966302620C118487528 @default.
• W1966302620 hasConceptScore W1966302620C126322002 @default.
• W1966302620 hasConceptScore W1966302620C134018914 @default.
• W1966302620 hasConceptScore W1966302620C167734588 @default.
• W1966302620 hasConceptScore W1966302620C2776004912 @default.
• W1966302620 hasConceptScore W1966302620C2778313320 @default.
• W1966302620 hasConceptScore W1966302620C2779829184 @default.
• W1966302620 hasConceptScore W1966302620C2908929049 @default.
• W1966302620 hasConceptScore W1966302620C555293320 @default.
• W1966302620 hasConceptScore W1966302620C71924100 @default.
• W1966302620 hasConceptScore W1966302620C84393581 @default.
• W1966302620 hasIssue "9105" @default.
• W1966302620 hasLocation W19663026201 @default.
• W1966302620 hasOpenAccess W1966302620 @default.
• W1966302620 hasPrimaryLocation W19663026201 @default.
• W1966302620 hasRelatedWork W2042789158 @default.
• W1966302620 hasRelatedWork W2365043821 @default.
• W1966302620 hasRelatedWork W2377184508 @default.
• W1966302620 hasRelatedWork W2409877198 @default.
• W1966302620 hasRelatedWork W2566684050 @default.
• W1966302620 hasRelatedWork W2994330360 @default.
• W1966302620 hasRelatedWork W3080076760 @default.
• W1966302620 hasRelatedWork W3203345339 @default.
• W1966302620 hasRelatedWork W4229050006 @default.
• W1966302620 hasRelatedWork W864080607 @default.
• W1966302620 hasVolume "351" @default.
• W1966302620 isParatext "false" @default.
• W1966302620 isRetracted "false" @default.
• W1966302620 magId "1966302620" @default.
• W1966302620 workType "article" @default.