Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Low allergy diet and hypertension prevention
- Ellen C G Grant (1 May 2004)
-
At last somrthing to save time
- Rupert A Gude (3 May 2004)
-
Anatomy of a trial: useful or useless?
- Francesco P Cappuccio (6 May 2004)
-
Re: Anatomy of a trial: useful or useless?
- Paul S Little, Jane Barnett, Jo Kelly, Martina Dorward, Barrie Margetts (21 June 2004)
|
|
|||
|
Ellen C G Grant, physician and medical gynaecologist 20 Coombe Ridings, Kingston-upon-Thames, KT2 7JU, Surrey, UK
Send response to journal:
|
Editor- Neither the low salt diet nor the extra fruit and vegetable diet used by Paul Little and colleagues reduced blood pressure in hypertensive patients over a 6 months’ period.1 In contrast, an individually tailored low allergy diet is effective in a week or two, as I described in the Lancet in 1979.2 Avoiding foods causing a pulse change or a blood pressure response, when re-introduced after a short avoidance period, was effective in preventing further attacks in 85% of 60 migraine patients. 15 out of 60 had labile hypertension and they became normotensive. Over the next 25 years I have continued to find this individualised diet prevents troublesome headaches, migraine and labile hypertension. The British Society for Allergy, Environmental and Nutritional Medicine recommends exclusion diets for a range of illnesses. Wheat, orange, eggs, coffee,(or chocolate for children),milk and corn were the commonest “hidden or masked” food allergens which caused adverse reactions in mine and in another nine studies.3 Avoiding wheat consumption alone may normalise blood pressure for some patients. 1 Little P, Kelly J, Barnett J, et al. Randomised controlled factorial trial of dietary advice for patients with a single high blood pressure reading in primary care. BMJ 2004; 328: 1054. 2 Grant ECG. Food allergy and migraine. Lancet 1979; 2: 358-359. 3 Anthony H, Birtwhistle S, Eaton K, Maberly J. Food allergy and intolerance – investigation and management. In Environmental Medicine in Clinical Practice.BSAENM Publications, Southampton, 1997. Competing interests: None declared |
|||
|
|
|||
|
Rupert A Gude, General Practitioner Abbey Surgery, Tavistock, Devon PL19 8BU
Send response to journal:
|
Dear Sir, If I read the tables correctly on bmj.com, the contol group had a blood pressure reduction of mean 10.3 / 5.1 at 6 months. So there is no advantage in spending time encouraging and educating newly diagnosed hypertensives. Just the diagnosis and brief advice and the people of Southampton sort out on their own their new problem with the lowest urinary NA:K ratio and subsequent blood pressure reduction of all the groups. I suppose it is better than being told you can add low sodium salt to your food at the table rather than being warned off all added salt. Yours sincerely Rupert Gude Competing interests: None declared |
|||
|
|
|||
|
Francesco P Cappuccio, Professor of Clinical Epidemiology & Primary Care Department of Community Health Sciences, St George's Hospital Medical School, London SW17 0RE
Send response to journal:
|
EDITOR - The paper by Little et al.(1) whilst addressing an important topic, stimulates several points of discussion. A number of methodological issues need clarification. The study was carried out in nurse run hypertension clinics in six general practices. This is a model of delivery of care widely used in the primary care setting. Unfortunately, there is lack of robust evidence from randomised controlled trials of the effectiveness of nurse-led hypertension management in primary care (2). Furthermore, patients were individually randomised to one of eight arms of the trial. There is no mention of the possible cluster effect due to the contribution of different practices, nor do authors clarify whether randomisation was balanced within practice. It is also likely that neither nurses assessing outcomes nor patients were blinded to the intervention arm. Patients were enrolled on a single blood pressure reading in the practice, assessment at one month was through blood pressure measured three times, yet the final assessment consisted of 14 home measurements of blood pressure carried out in the patient’s home. The entry assessment does not follow British Hypertension Society guidelines (nor does that at six months) that indicate that at least two measurements (1-2 min apart) should be taken on each occasion (3). Also appropriate cuff sizes should be used. It is likely that the majority of patients enrolled would have had normal blood pressure and, therefore, would have been less likely to respond to any non-pharmacological treatment. Indeed, the blood pressure response to any therapy is greater the higher the blood pressure. Also, power calculations are based on standard deviations (of differences?) based on a study of elderly hypertensives patients, not of younger normotensive individuals (4). The study may have been grossly underpowered to show any effect. For instance, as indicated in the BHS guidelines (3), a reduction of sodium intake to <100 mmol/day (or to <6g of salt) would be expected to reduce systolic blood pressure by 2-8 mmHg. In the present trial Low Sodium Salt achieved a reduction in sodium-to-potassium ratio of 0.32 [0.08 to 0.56], likely to be due to an increase in potassium rather than a reduction in sodium. This is fully comparable to the changes seen by Geleijnse et al (4) in older patients with a higher blood pressure. Nevertheless, a 1.3 mmHg lower systolic blood pressure at one month and 1.4 mmHg at six months is observed, though not statistically significant. This is exactly the effect on blood pressure you would expect from that change in sodium-to- potassium ratio in younger normotensive individuals (5). The study was grossly underpowered to detect such a difference. Finally the claim that the Low Sodium Salt may have adverse effects in the short term is rather strong. Indeed, Little and coll. only refer to a modest rise in the anxiety score at one month (not sustained at six months) difficult to interpret since the trial was not double-blind. In conclusion, the main outcome of the trial could have been predicted by a scrupulous examination of the study design. Regrettably, the study was funded and carried out and it represents a missed opportunity to provide valid and generalisable useful information. References 1. Little P, Kelly J, Barnett J, Dorward M, Margetts B, Warm D. Randomised controlled factorial trial of dietary advice for patients with a single high blood pressure reading in primary care. BMJ 2004;328:1054-7 2. Oakeshott P, Kerry S, Austin A, Cappuccio FP. Is there a role for nurse-led blood pressure management in primary care? Family Practice 2003;20:469-73 3. Williams B, Poulter NR, Brown MJ, Davis M, McInnes GT, Potter JF, Sever PS, Thom SMcG. Guidelines for management of hypertension: report of the fourth working party of the British Hypertension Society, 2004 – BHS IV. J Hum Hypert 2004;18:139-85 4. Geleijnse JM, Witteman JCM, Bak AAA, den Breeijen JH, Grobbee DE. Reduction in blood pressure with a low sodium high potassium high magnesium salt in older subjects with mild to moderate hypertension. BMJ 1994;309:436-40 5. Cappuccio FP. Salt and blood pressure. Issues for population-based prevention and public health strategies. Public Health Medicine 2000;2(2):57-61 Competing interests: None declared |
|||
|
|
|||
|
Paul S Little, Professor of Primary Care Research Aldermoor Health Centre, University of Southampton, SO16 5 ST, Jane Barnett, Jo Kelly, Martina Dorward, Barrie Margetts
Send response to journal:
|
Anatomy of a trial - perhaps useful after all? Professor Cappucio raises some important points – that we have overplayed the possibility of adverse effects, and suggests that the reasons no effect on blood pressure was found with the low sodium salt – for example when comparing our results with the previous Dutch study using a low sodium salt1 - were numerous design issues. Regarding the adverse effects we were quite careful to say that this was a secondary finding and should be treated with caution, but nevertheless should not be dismissed. Taking each of the other points in turn: 1) Using 8 groups. Although patient were randomised to 8 groups, the design was a factorial design. In a factorial design unless a significant interactions are shown between interventions (and none were) then the estimates for each intervention use the whole sample. Thus for the low sodium salt at one month the estimates are comparing those who had low salt (n=125) compared to those who did not have the salt (n=144), controlling for any possible effects of co-interventions. This allows precise adjusted estimates of the effect of each intervention. 2) Outcome assessment bias? In an open pragmatic trial patients and nurses cannot be blinded to their intervention, but nurses measured blood pressure using semi-automated blood pressure monitors according to a structured schedule. Furthermore any possible measurement bias is likely to favour the interventions, but we found no evidence of this. 3) Placebo effect? In an open trial there is clearly the possibility of a placebo effect favouring the active interventions. However we have shown in previous trials that the estimates using a structured approach - where each group intervention is supported by structured advice - can abolish the placebo effect and provide estimates very similar to blinded trials2;3. In this study there was no evidence of a placebo effect favouring the interventions. 4) We used the wrong sample? We apologise for the ambiguity regarding ‘a single blood pressure reading’. GPs and nurses were asked to take 2 or three readings according to BHS guidelines on a single occasion (i.e. this was not a single reading, but on a single occasion), and including using the appropriate cuff sizes, when referring patient to the study. This is the population where GPs and nurses are advised to give non- pharmacological advice while performing watchful waiting before definitive diagnosis of hypertension. We argue this is exactly the time when the effect of something like a low sodium salt should be assessed in practice. 5) A sample who did not have hypertension? After the blood pressure reading on the first occasion, and after a few weeks waiting to enter the study our baseline mean blood pressure (i.e. the second reading the patient had) was 153/93 mm Hg - not dissimilar to the Dutch study which investigated a low sodium salt (158/91 mm Hg). If we confine our estimates to those with DBP>90 on this second reading (n=171) the estimates of the effect of low salt on DBP are very similar -1.16 mm Hg (95% CI -3.5 to 1.18) 6) The sample was too young, and so we were underpowered? This was the sample that presented is likely to be more representative of a general practice population than just selecting an elderly population. We also found no interaction of effect with age (the estimate of interaction term was -0.5mm Hg i.e. a lower effect in older patients). If we confined the estimates of effect to those over age 60 this creates a subsample with mean age 68, older than Dutch study, n=94 , and a baseline BP 160 /92 mm Hg (i.e. a similar sample size, with similar power, and a higher baseline BP than Dutch study). In this sub-sample the estimates of effect of advice to use lowsalt on BP are also very similar to our main study findings - + 0.001 mm Hg (-3.17 to 3.17). 7) We have provided misleading estimates due to clustering, i.e. we have made errors estimating the standard errors (SEs) in the study? As expected, allowing for clustering by practice or nurse (i.e. estimating robust SEs allowing for clustering) results in almost identical SEs (in fact slightly lower). This is not surprising since this was an individually randomised trial not a cluster randomised trial, and furthermore the nurses used highly structured proformas. Thus none of Professor Capuccio’s potential explanations for our results are borne out in the data. There are two likely reasons for the lack of effect: a) Low salt is not likely to be very effective for individuals, and the previous Dutch study provides an overestimate of effectiveness b) In a pragmatic setting the control group are given basic advice, are not constrained to keep their diets constant, and furthermore know that they are not getting low salt – and thus may be more motivated to change their diets according to basic advice. Indeed there was evidence in our trial that the control group did make significant changes in their diet. In this context the additional effect of a low sodium salt may be lower than expected even among the elderly and those with more definite hypertension. For evidence to be relevant to practice we need both efficacy trials which constrain patients more, but also pragmatic open trials designed to look at the effect of advice in everyday practice where patient behaviour is rather more realistic. Yours Sincerely Paul Little
Reference List 1. Geleijnse JM, Witteman JCM, Bak AAA, den Breeijen JH, Grobee DE. Reduction in blood pressure with a low sodium high potassium high magnesium salt in older subjects with mild to moderate hypertension. B.M.J. 1994;309:436-40. 2. Little PS, Williamson I, Warner G, Gould C, Kinmonth AL, Gantley M. An open randomised trial of prescribing strategies for sore throat. B.M.J. 1997;314:722-7. 3. Little P, Gould C, Williamson I, Moore M, Warner G, Dunleavey J. A pragmatic randomised controlled trial of two prescribing strategies for acute otitis media. BMJ 2001;322:336-42. Competing interests: None declared |
|||