Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Power dressing and meta-analysis.
- S J Muncer (26 October 1998)
-
Poor validity?
- Ingegerd Larsson (27 October 1998)
-
Four out of five controlled trials have achieved prolonged decrease in mite allergen
- Nancy K Malone (12 November 1998)
-
The peak flow of a population is not a valid measure of asthma
- David S Morrison (12 November 1998)
-
House dust mite control measures: effectiveness depends on the severity of asthma?
- Sonja G M Cloosterman (25 November 1998)
-
House dust mite control measures in the management of atopic dermatitis
- C Gutgesell (26 April 1999)
-
Missing Data
- Ken R. Waldron (15 July 2003)
|
|
|||
|
S J Muncer, Reader School of Health, University of Teesside.
Send response to journal:
|
The recent paper by Gotzsche, Hammarquist and Burr[1] highlights a recurring problem with systematic review and meta-analysis, namely ignoring or paying lip service to the importance of power. The continued publication of articles with inadequate power in social and health sciences has been noted on many occasions[2,3,4], without any apparent effect[3]. These articles are, however, frequently combined into meta- analyses with scant regard to their power. The power of a study is the probability that it will lead to statistically significant results[2]. In a recent systematic review of the literature examining the prevention of pregnancy, for example, 9 of the 15 articles included had "low statistical power". Given that they had low statistical power, should they have been included? In the present study[1], the major results are that 41 out of 113 patients exposed to treatment interventions improved, compared with 38 out of 117 in the control groups. If we imagine this had been run as a single experimetn with this number of subjects and we carried out a chi square test on the results, they would have indeed been nonsignificant (x2=1.27 p=.161, phi 0.076). Suppose that we had originally believed that the effect size of treatment would be small, that is about 0.1[2], then to have an adequate sample with power of 0.8 to detect a significant difference, we would need 785 subjects for the 0.05 level and 1168 for the 0.01 level. To put it another way the power of this study to detect a significant difference of a small effect size is inadequate. It is also inadequate to detect a medium effect size. Given that this is the case, would it be published if it were a single study, and should it be published because it is a meta-analysis? References 1. Gotzsche PC, Hammarquist C, Burr M. House dust mite control measures in the management of asthma: meta-analysis. BMJ 1998; 317:1105-1110. 2. Cohen J. A power primer. Psych Bull 1992; 112: 155-159. 3. Sedlmeier P, Gigerenzer G. Do studies of statistical power have any effect on the power of studies? Psych Bull 1989; 105: 309-316. 4. Polit DF, Sherman RE. Statistical power in nursing research. Nurs Res 1990; 39:365-368. 5. NHS Centre for reviews and Dissemination. Effective Health Care 1997; 3(1); 1-11. |
|||
|
|
|||
|
Ingegerd Larsson, Pharmacist Medeca AB
Send response to journal:
|
Each method needs to be analysed separately in order to be able to draw the right conclusions. In this analyses 11 different prevention methods have been analysed as a whole and as if they are more or less alike. The validity is therefore poor. The maximum allergen load has been found deep inside the mattresses. Several published studies show that particle impermeable covers reduce the allergen exposure with more than 90 percent. Airborne mite allergens are only detected after vigourous disturbance. The role of chemicals, cleaning, vacuum cleaning, air cleaners and ionisers are unclear. There is a dose-response relationship between exposure and sensitization. However, some patients react to very low and others require higher doses. Besides that and the fact that some methods are found to be ineffective in reducing the allergen exposure, the allergen load vary extremely in the studies, some studies run only a couple of weeks,compliance, design and measurements differ etc. Seven completely different physical methods have been analysed. Allergen reduction were successful in nine studies. Eight used bedding covers. Twelve were unsuccessful.Six of them used aircleaning. Gillies 1987(13)(covers) is said to have no reduction though there was a significant fall both in the test(mite count fell from 40.00 to 0.85 after 12 weeks) and the control(from 21.75 to 10.33 after 6 weeks observation and to 2.17 after another 6 weeks with covers). Marks 1994(14)(covers)is also said to have no reduction, though it was and a published re-analyse 1995 showed a significant correlation between allergen concentration and changes in AHR and symptom score. In Burr 1980(8)(new bedding,covers)the bedding was entirely free from mites at start - no reduction could be expected. Burr 1980(7) used washing and vacuum cleaning that later on has been shown not to be effective enough (Tovey 1997). Many patients are treated with steroids, effectively reducing the inflammation and improving the symptoms. Improvement in the symptoms would therefore not be expected from added preventive measures in a couple of weeks. In most studies medication is not monitored. In the study by Huss (25)(covers) the symptoms were improved and medication reduced. In the study by Walshaw and Evans 1986 (19)(covers) medication was reduced with 40 percent. In Carswell 1996(23)(covers) inhaled brochodilators and steroids was less in the active group than in the control group, 17% and 54% and 13% and 35% respectively. In a double-blind placebo controlled study by Halken et al(covers, JACI 1997 abs) steroids was reduced with 46 percent in the test group. |
|||
|
|
|||
|
Nancy K Malone
Send response to journal:
|
Please note: As Dr Platt-Mills's secretary, Dr Malone typed this rapid response. She does not consider herself to be an author. Dr. Gotzsche and his colleagues reported a "meta analysis" of the published controlled trials of dust mite avoidance in the treatment of asthma (1). It is well known that several of the approaches used in the past for decreasing mites in houses were not effective; e.g. vacuuming carpets or mattresses, acaricidal foams, and HEPA room air cleaners (1,2). In their paper, the authors concede that in twelve out of the 23 studies, the avoidance measures used did not decrease mite allergen. In addition, 5 studies failed to measure whether dust mite exposure changed. It is bewildering that the authors would apply, or the journal publish, an analysis including all these studies as if they were comparable to controlled trials in which a significant decrease in mite allergen was achieved. This would be equivalent to publishing an analysis of the effectiveness of inhaled steroids based on studies in which the authors concluded that the actuation failed to deliver the drug, or compliance was not assessed. Indeed, many of the older mite "avoidance" studies included in the "meta-analysis" have so little insight into the factors that influence mite allergen in a house, that it is amusing to reread them today (3,4). The authors state that the results were the same for those studies where successful reduction in exposure to mite allergen was achieved. However, this conclusion is strikingly different from that of three groups who have recently analyzed the same studies (2,5,6). Each of these reports concluded that the evidence strongly favored the use of physical avoidance measures for treating mite allergic children with asthma. There are five published controlled trials of allergen avoidance that have achieved a prolonged decrease in allergen (i.e. 6 months or more) (7,8,9,10,11). Given the conclusions of Dr. Gotzsche and his colleagues, your readers might be surprised to find that in four of those studies, the authors reported highly significant improvement in the active avoidance group (7,8,9,11). In their attempt to achieve a "meta-analysis", the authors restricted their calculations to two outcomes (symptoms and AM peak flow), ignoring the fact that these were not the primary outcomes of the successful studies. Thus, the controlled trial by Ehnert and her colleagues in Berlin (8), which produced the most convincing decrease in mite allergen and highly significant decrease in BHR, was not included in either of the comparisons in your recent paper (see Figs 2 and 3 in Ref. 1). There is consistent evidence that allergic patients who are removed from a high mite environment improve clinically, and in terms of bronchial reactivity (2). The question has always been whether it is possible to achieve the same effect under normal household conditions. Gotzche and colleagues reached a negative conclusion by including studies that had no effect, or unknown effects, on mite allergen and imposing an analysis that was simplistic and unrelated to the successful results. The correct conclusions are: first that reduction of mite allergens in a humid climate is not easy and requires understanding the factors that influence mite growth; and second, that four out of five controlled trials that have achieved prolonged decrease in mite allergen have achieved impressive clinical results. Yours sincerely, Thomas A.E. Platts-Mills, M.D., Ph.D. Martin D. Chapman, Ph.D. Head, Div of Asthma, Allergy & Immun Professor of Med. & Micro. Dir, UVA Asthma & Allergic Dis Ctr Lisa M. Wheatley, M.D. Assistant Professor of Medicine TPM/nmReferences 1. Gotzshe PC, Hammarquist C, Bur M. House dust mite control measures in the management of asthma: meta-analysis. BMJ 1998;317:1105- 1110. 2. Report of 3rd International Workshop, Cuenca Spain. Indoor allergens and asthma. J Allergy Clin Immunol 1997;100;S1-S24. 3. Burr MI, Dean BV, Merrett TG, Neale E, St Leger AS, Verrier-Jones ER. Effects of anti-mite measures on children with mite-sensitive asthma: a controlled trial. Thorax 1980;35:506-12. 4. Antonicelli I, Bilo MB, Pucci S, Schou C, Bonifazi F. Efficacy of an air-cleaning device equipped with a high efficiency particulate air filter in house dust mite respiratory allergy. Allergy 1991;46:594-600. 5. Guidelines for the diagnosis and management of asthma. Expert Panel Report II. Bethesda, MD: National Institutes of Health, 1997 (NIH publication no. 97-4051). 6. Colloff MJ, Ayres J, Carswell F, Howarth PH, Merrett TG, Mitchell EB, Walshaw MJ, Warner JO, Warner JA, Woodcock AA. The control of allergens of dust mites and domestic pets: a position paper. Clin Exp Allergy 1992;22:1-28. 7. Murray AB, Ferguson AC. Dust-free bedrooms in the treatment of asthmatic children with house dust or house dust mite allergy: a controlled trial. Pediatrics 1983;71:418-422. 8. Ehnert B, Lau-Schadendorf S, Weber A, Buettner P, Schou C, Wahn U. Reducing domestic exposure to dust mite allergen reduces bronchial hyperreactivity in sensitive children with asthma. J Allergy Clin Immunol 1992;90;135-8. 9. Walshaw MJ, Evans CC. Allergen avoidance in house dust mite sensitive adult asthma. QJ Med 1986;58:199-215. 10. Carswell F, Birmingham K, Oliver J, Crewes A, Weeks J. The respiratory effects of reduction of mite allergen in the bedroom of asthmatic children: a double blind controlled trial. Clin Exp Allergy 1996;26:386-96. 11. Halken S, Niklassen U, Hansen LG, Nielsen F, Host A, Osterballe O, Veggerby MC, Poulsen LK. Encasing of mattress in children with asthma and house dust mite allergy. J Allergy Clin Immunol 1997;99:S320. |
|||
|
|
|||
|
David S Morrison, Specialist Registrar in Public Health Medicine Greater Glasgow Health Board G3 8YU
Send response to journal:
|
4th November, 1998. The Editor BMJ BMA House Tavistock Square LONDON WC1H 9JR. Sir Gotzsche PC, Hammarquist C, Burr M. House dust mite control measures in the management of asthma: meta-analysis. BMJ 1998; 317: 1105-1110. The enormous heterogeneity of peak expiratory flow rates makes them very poor measures of asthma severity when studying populations. Gregg and Nunn’s results from mini-Wright peak flow meters demonstrate a wide variation between individuals - age and height explain only 30% of the variation in peak flow rates(1,2). Peak expiratory flow rate nevertheless remains a useful measure for self-monitoring of asthma. It is therefore not surprising that in using peak expiratory flow rate as one of their principle measures of asthma severity Gotzche, Hammarquist and Burr found measures to eradicate house dust mite had no statistically significant effect(3). In addition to the important distinction made by Strachan between efficacy and clinical effectiveness in efforts to reduce house dust mite(4) the measure of clinical effectiveness should be valid. I suggest that failure to distinguish between two population distributions of peak expiratory flow rate does not provide a valid measure of asthma. David S Morrison Specialist Registrar in Public Health Medicine Greater Glasgow Health Board Dalian House PO Box 15327 350 St Vincent Street GLASGOW G3 8YU. Telephone 0141 201 4900. 1 Nunn AJ, Gregg I. New regression equations for predicting peak expiratory flow in adults. BMJ 1989;298:1068-70. 2 Gregg I, Nunn AJ. Peak expiratory flow in symptomless elderly smokers and ex-smokers. BMJ 1989;298:1071-1072. 3 Gotzsche PC, Hammarquist C, Burr M. House dust mite control measures in the management of asthma: meta-analysis. BMJ 1998;317:1105-1110. 4 Strachan DP. House dust mite allergen avoidance in asthma. BMJ 1998;317:1096-1097. |
|||
|
|
|||
|
Sonja G M Cloosterman
Send response to journal:
|
EDITOR-The meta-analysis of Grtzsche1 concluded that house dust mite control measures are not clinically effective in house dust mite allergic asthmatics. Strachan2 indicated that this was probably a consequence of the fact that several control measures used in the included studies, did not result in a (relevant) reduction of house dust mite allergens. Improvements in clinical condition are consequently not to be expected. Some studies in the meta-analysis did find clinical effects while others did not1. This might, however, not be a result of whether or not effective allergen reduction occurred, but more a result of measuring different groups of asthmatic subjects, who are in a different stage of their asthmatic disease. We have reason to believe that early treatment of very mild asthma might have more impact than treatment of moderate to severe asthma. In our department we performed two studies. First we investigated the (clinical) effects of a combined allergen avoidance strategy (house dust mite (HDM) impermeable covers for mattresses and bedding and Acarosan® for living room and bedroom floors) in a group of subjects, allergic to HDM, without a diagnosis of asthma yet but with some early signs of asthma3. Secondly, we investigated this same combined strategy in a group of patients, with a confirmed diagnosis of moderate asthma. In the first study we found that peak flow parameters and asthma symptom scores were stabilised during the follow-up period (figure 1a), suggesting that a delay in the onset of asthma may occur in some allergic subjects. In the second study, no clinical effects could be observed in diagnosed allergic subjects with moderate asthma (figure 1b). This negative result was not a consequence of a failure in allergen avoidance. Allergen concentrations were reduced, especially on mattresses (ten-fold, p=0.0001)4. The question is why no positive clinical effects were found in the group of allergic subjects with already established asthma, but were found in an allergic group that had not developed asthma yet? We hypothesise that allergen avoidance as early preventive measure has more impact than as treatment of moderate asthma. When asthma is already established, it may be that very small amounts of allergens are sufficient to trigger a deterioration in asthma. Furthermore, allergen reduction need some time to reverse the already developed process of inflammation and consequently it needs some time to establish effects. In allergic patients without asthma yet, a reduction in allergen load might prevent further development of asthma, as it is probably possible to prevent or to slow down the process of inflammation in this early stage of the disease. The general observation that avoidance measures are more effective in children than in adults, does support this vision5. This implies that allergen avoidance measures have to be applied in an early stage of the disease (secondary or even primary prevention6) in order to be clinically effective. Summarised, we believe that the conclusion of Grtzsche et al. is covering only one aspect. It seems that HDM control measures might be effective in patients in an early stage of their asthmatic disease, and might therefore recommended as an early intervention. More research has to be performed to investigate this hypothesis. Sonja G.M. Cloosterman, MSc1 Constant P. van Schayck, MSc, PhD1/2 1 Dept. of General Practice and Social Medicine, University of Nijmegen 2 Dept. of General Practice, University of Maastricht The Netherlands Corresponding address: Mrs. S.G.M. Cloosterman, MSc Dept. of General Practice and Social Medicine, 229 University of Nijmegen P.O.box: 9101 6500 HB Nijmegen Tel.: +31 24-3616968/3613315 Fax.: +31 24-3617084 email: S.Cloosterman@hsv.kun.nl References 1. Gotzsche PC, Hammarquist C, Burr M. House dust mite control measures in the management of asthma: meta-analysis. Br Med J 1998;317:1105-10. 2. Strachan DP. House dust mite allergen avoidance in asthma: Benefits unproved but not yet excluded. Br Med J 1998;317:1096-7. 3. Cloosterman SGM, Hofland ID, Lukassen HGM, Wieringa MH, Folgering HTM, van der Heide S, Brunekreef B, Schayck CPv. House dust mite avoidance measures improve peak flow and symptoms in patients with allergy but without asthma: A possible delay in the manifestation of clinical asthma? J Allergy Clin Immunol 1997;100:313-9. 4. Cloosterman S, Hofland I, van der Heide S, Brunekreef B, Eslhout Fv, Folgering H, Schayck Cv. Effects of house dust mite impermeable covers and an acaricide on Der p 1 Levels. Eur Respir J 1998;12:69S(Abstract) 5. Ehnert B, Lau-Schadendorf S, Weber A, Buettner P, Schou C, Wahn U. Reducing domestic exposure to dust mite allergen reduces bronchial hyperreactivity in sensitive children with asthma. J Allergy Clin Immunol 1992;90:135-8. 6. Schönberger HJAM, Schayck CPv. Prevention of asthma in genetically predisposed children in primary care-from clinical efficacy to a feasible programme. Clin Exp Allergy 1998;28:1325-31 |
|||
|
|
|||
|
C Gutgesell
Send response to journal:
|
EDITOR-Gotzsche et al1 conclude that avoidance of indoor allergens by physical and chemical measures is not effective as treatment for atopic asthma. We investigated the effect of allergen avoidance strategies (encasing, acaricides) in patients with atopic dermatitis and also obtained some negative results: A placebo-controlled allergen elimination study was performed in 20 adult patients with moderate to severe atopic dermatitis; the study period was 1 year. All patients were sensitised to house dust mite (RAST-class ³ 3) and were exposed to a significant amount of the mite allergen Der p 1 on their mattress (³ 2 µg/g). Disease severity was evaluated every 2 months using an established score (SCORAD) and the serum concentration of eosinophil cationic protein (ECP), a disease activity marker, was monitored by a commercial assay. In addition, the use of topical steroids was quantified and the patients noticed their subjective impression of pruritus and sleeplessness on a visual scale weekly. Der p 1 levels dropped in the verum group to almost undetectable concentrations. However, although in both groups (placebo and verum) objective parameters (SCORAD and ECP) improved temporarily, no statistical difference between placebo and verum group was observed. But there was a marked difference between both groups with respect to the subjective parameters: After 4 months of allergen exclusion, the verum group constantly reported to suffer significantly less from nocturnal pruritus and sleeplessness than the placebo group. We conclude that allergen avoidance leads to a better quality of life since sleep is improved. But reducing indoor allergen exposure is not sufficient to improve the overall disease activity. The reason for this may be an ongoing exposure to other allergens including outdoor allergens and irritants. Our results disagree with those of another placebo-controlled trial on dust mite avoidance measures in atopic dermatitis published recently2: Performing a 6 months- trial, these authors had observed a significant improvement of the clinical score in the verum group. The difference may be explained by a partly different study population (inclusion of children in the latter study). Allergen avoidance may be more effective in children than in adults and this should be investigated in controlled trials. Taken together, our results are in accordance with D. Strachan3 and we conclude: House dust mite allergen avoidance in atopic dermatitis: ''Benefits unproved but not yet excluded". C. Gutgesell S. Heise S. Seubert A. Seubert Ch. Neumann Department of Dermatology, University of Göttingen, Germany 1 Gotzsche PC, Hammarquist C, Burr M. House dust mite control measures in the management of asthma: meta-analysis. BMJ 1998 Oct 24;317(7166):1105-10. 2 Tan BB, Weald D, Strickland I, Friedmann PS. Double-blind controlled trial of effect of housedust-mite allergen avoidance on atopic dermatitis. Lancet 1996 Jan 6;347(8993):15-8 3 Strachan DP. House dust mite allergen avoidance in asthma. Benefits unproved but not yet excluded. BMJ 1998 Oct 24;317(7166):1096-7. |
|||
|
|
|||
|
Ken R. Waldron, Instructor, Research and Statistics Boucher Institute of Naturopathic Medicine, New Westminster, BC V3M 5Y6
Send response to journal:
|
Careful readers of this review might have noticed an inconsistency between the list of studies selected for inclusion, and those actually used for the meta-analysis. The review purports to include 23 studies (counting one three-arm trial as two studies). However, examination of Figures 1 through 3 reveals that outcomes for the following 6 studies were not included in the meta-analysis:
Ehnert et al (1992) Van der Heide et al (1997) Sette et al (1994) Gillies et al (1987) Verrall et al (1988) Ehnert et al (1992) The explanation, from one of the authors (Peter C. Gøtzsche, private communication), is that "some trials did not provide data that could be used for meta-analysis." Competing interests: None declared |
|||