The participants, recruited by advertisements in two daily newspapers, were obese and asthmatic. Inclusion criteria were ability to cope with the study protocol, body mass index 30-42, age 18-60 years, previously diagnosed asthma with a spontaneous diurnal variation or a bronchodilator response of 15% or more, and being a non-smoker or having stopped smoking for two years or more before age 50. Exclusion criteria were pregnancy, history of bulimia or anorexia, unstable angina or arrhythmia, untreated thyroid disease, symptomatic liver or gall bladder disorder, any other severe disease, insulin treatment, systemic steroid treatment, or history of food allergy or of intolerance to any component of the very low energy dietary preparation that would be used in the study—such as soya, fish, chocolate, or lactose. Participants with a history of adverse reactions to peas, beans, or peanuts were excluded because of possible cross reactions to soya protein.3
Telephone interviews were held with 202 potential participants; 133 were excluded (see fig A, a flow chart showing exclusions, on the BMJ 's website), and the remaining 69 were invited for further evaluation. These 69 potential participants underwent a personal interview and a clinical check up, and their inhalation technique was checked. Antiasthmatic medication was not changed if the stability of the asthma was clinically acceptable. In six cases medication had to be intensified. During the baseline period and the study period, medication was changed only during episodes of exacerbation of the asthma. One participant with an uncertain history of allergy to peas and beans was excluded because a test dose of the dietary preparation caused immediate allergic symptoms.
After a two to three week run-in period with lung function measurements and laboratory tests, 38 participants fulfilled the inclusion criteria (see BMJ 's website for flow chart showing exclusions). After two weeks of baseline measurements, the participants were randomised to a treatment group (19) and a control group (19). Randomisation was by “shuffling cards,” with the help of someone not involved in the study. Clinical data are presented in table A on the BMJ 's website.
The treatment group took part in a weight reduction programme that included 12 group sessions and lasted 14 weeks, including eight weeks—“the dieting period”—in which participants took a very low energy dietary preparation (Nutrilett (Nycomed Pharma, Oslo)). The daily dose gave 1760 kJ of energy and contained daily allowances of all essential nutrients, as described elsewhere.4
The principles and behaviour methods of this weight reduction programme are also described elsewhere.5,6
The control group had sessions at the same intervals as the treatment group; each session lasted half an hour, during which time themes chosen by the group were discussed freely. These themes were discussed with the treatment group at later sessions; by the end of the first year each group had had the same amount of education about asthma and allergy.
All participants received normal medical care throughout the study. Using a mini-Wright peak flow meter and taking the best of three consecutive blows into account, the participants measured their daily morning and evening pre-bronchodilator and postbronchodilator peak expiratory flow (PEF) during the dieting period, and thereafter during the two weeks before each group meeting. The mean of the previous two weeks' premedication PEF values served for calculation. Forced vital capacity (FVC) and forced expiratory volume in one second (FEV1
) were measured by means of a mini-Wright spirometer at baseline, at the end of the dieting period, at the end of the 14 week weight reduction programme, at six months, and at one year. During the peak flow follow up periods, asthma symptoms were recorded on a 100 mm visual analogue scale (where 0 mm represented “best possible” and 100 mm “worst possible”) and rescue medication was recorded as the number of daily doses of an inhaled bronchodilator. Serum cortisol and diurnal urine cortisol excretion concentrations and blood and urine concentrations of sodium, potassium, calcium, magnesium, triglycerides, and cholesterol were measured at baseline, at the end of the dieting period, and after 14 weeks. Health status was investigated four times during the first year, using the St George's respiratory questionnaire, which is divided into three parts: symptoms (distress caused by specific respiratory symptoms); activity (physical activities that cause or are limited by breathlessness); and impact (social and psychological effects of the disease). The total score is derived from all items and is expressed as a percentage of the maximum possible. A decrease in the score indicated an improvement in health status. A translated Finnish version of this questionnaire has been used previously.7
The data from the questionnaires on health status were compiled with the help of software designed by P W Jones (St George's Hospital, London). All 38 participants were followed for one year. Two participants found the consistency or taste of the dietary preparation intolerable but followed a low energy diet; one participant started smoking during the study. These three participants were retained in the study.
Determination of sample size was based on the change in daily PEF (l/min). It was originally assumed that PEF would increase by 50 l/min in the treatment group and by 20 l/min in the control group. It was assumed that a difference in change of 30 litres or more between the groups would be clinically significant.8
To demonstrate this, with a standard deviation of 25 l/min, α error of 5%, and a power of 90%, 15 participants per group were required to complete the study.9
However, owing to chance, the baseline difference between the groups was 36 litres (treatment group, 423 litres; control group, 387 litres)—that is, greater than the expected treatment difference. Also owing to chance, the proportion of men was higher in the treatment group, which can explain the baseline differences expressed in litres. To control for the possible differences in sex, age, and height distributions, the analysis was based on percentage of predicted values.10,11
The groups were similar for these baseline values and for all other participant characteristics except rhinitis (see table A on website). The mean numbers of PEF recordings were 13.9 (range 12-14), 13.6 (7-14), 13.7 (11-14), 13.3 (5-14) and 13.7 (8-14) for the five two-week periods. The mean of all usable PEF recordings was used as the mean value for the period.
The data were analysed by means of StatView 512+TM (Brainpower) for Apple Macintosh and SPSS 9.0 for Windows. The changes from baseline were calculated for lung function, and the t test for independent samples was used to compare the treatment group with the control group. The 95% confidence intervals were calculated for change within a treatment group and for difference between groups. Changes in symptoms and use of rescue sympathomimetics were analysed by the non-parametric Mann-Whitney U test. Changes in health status subscores and in laboratory variables were analysed by the t test for independent samples. The χ2 test was used for binary variables.
Secondary analyses were performed to control for the multiplicity of data. Analysis of variance for repeated measurements was performed to compare the changes from baseline in the lung function variables in the two study groups. We studied (a) the difference between study groups; (b) the time effect—that is, change during the follow up period; and (c) interaction between group and time effect. Repeated measurements of symptoms (dyspnoea, cough) and use of sympathomimetics during the follow up were reduced to the mean of all measurements, and the overall changes from baseline were calculated. The Mann-Whitney U test was used to compare the treatment group with controls.