Adjustment of reported prevalence of respiratory symptoms for non-response in a multi-centre health survey

BACKGROUND

Estimation of non-response bias by modelling prevalence as a function of the number of mailings required to achieve a response, or of the cumulative response, has been advocated, but the models have not incorporated age and sex, differential response rates by age and sex, or season of response.

METHODS

The effect on age-sex standardized prevalence of estimating non-response bias using a variety of models was investigated using data on nine symptom and medication questions from 13,007 subjects in the three English centres of the European Community Respiratory Health Survey. Comparison was made of goodness of fit and the prediction of responses in a 25% follow-up sample with the observed values.

RESULTS

Despite low response rates in Cambridge and significant decreases in prevalence with additional mailings or increasing cumulative response in Norwich, there were only small effects on estimated age-sex standardized prevalences. No model was consistently better for any centre or question.

CONCLUSIONS

The models are useful for exploring the sensitivity of estimated prevalence to non-response bias, but should be used with caution to adjust estimates. Ideally first mailings should be staggered over the whole year so that mailing and season are not confounded, and sufficient mailings or other contacts carried out for the whole sample to ensure a high response rate.

Interpretation of measured red cell mass and plasma volume in adults: Expert Panel on Radionuclides of the International Council for Standardization in Haematology

Four European centres provided height and weight data on 202 males and 204 females undergoing red cell mass (RCM) and plasma volume (PV) measurements. For these populations, the RCM and PV predictions by the various published methods were compared. It was shown clearly that predictions based solely on body weight were inappropriate, particularly because approximately half of the male and female populations could be regarded as overweight or obese. Although there was reasonable agreement in the prediction values given by the formulae based on both height and weight, it was not possible to establish which formulae could be recommended. For that reason, the published literature containing normal RCM and PV measurements were re-examined. RCM data for 283 males and 171 females and PV data for 100 males and 67 females were included. Measurements were standardized for variables such as trapped plasma in the PCV, exclusion of buffy coat in the PCV and calculation of PV at zero time. As a result of this analysis, prediction formulae based on surface area for RCM and PV with 98/99% reference ranges have been established.

Physical fitness of 9 year olds in England: related factors

STUDY OBJECTIVE

To examine the influence of social factors, passive smoking, and other parental health related factors, as well as anthropometric and other measurements on children's cardiorespiratory fitness.

DESIGN

This was a cross sectional study.

SETTING

The analysis was based on 22 health areas in England.

PARTICIPANTS

The subjects were 299 boys and 282 girls aged 8 to 9 years. Parents did not give positive consent for 15% of the eligible sample. A further 25% of the eligible sample did not participate because the cycle-ergometer broke down, study time was insufficient, or they were excluded from the analysis because they were from ethnic minority groups or had missing data on one continuous variable.

MEASUREMENTS AND MAIN RESULTS

Cardiorespiratory fitness was determined using the cycle-ergometer test. It was measured in terms of PWC85%-that is, power output per body weight (watt/kg) assessed at 85% of maximum heart rate. The association between children's fitness and biological and social factors was analysed in two stages. Firstly, multiple logistic analysis was used to examine the factors associated with the children's ability to complete the test for at least four minutes. Secondly, multiple linear regression analysis was used to examine the independent association of the factors with PWC85%. In the logistic analysis, shorter children, children with higher blood pressure, and boys with a larger sibship size had poorer fitness. In the multiple regression analysis, only height (p < 0.001) was positively associated, and the sum of skinfold thicknesses at four sites (p = 0.001) was negatively associated with fitness in both sexes. In girls, a positive association was found with pre-exercise peak expiratory flow rate (p < 0.05), and there were negative associations with systolic blood pressure (p < 0.05) and family history of heart attack (p < 0.05). In boys an association was found with skinfold distribution and fitness (p < 0.05), so that children with relatively less body fat were fitter. Social and health behaviour factors such as father's social class, father's employment status, or parents' smoking habits were unrelated to child's fitness.

CONCLUSION

Height and obesity are strongly associated, and systolic blood pressure to a small extent, with children's fitness, but social factors are unrelated.

Monitoring the growth of children: conclusions from a long-term study

BACKGROUND

The National Study of Health and Growth (NSHG) was set up in 1972 to monitor the growth of primary school children. Areas were selected in England and Scotland by stratified random sampling. Schools within these areas were visited annually until 1982, biennially thereafter, resulting in a mixed longitudinal design. The reasons for the original design and the study as it has operated are reviewed, with advantages and disadvantages compared to the monitoring system now to be implemented by the Department of Health.

METHODS

Description of the statistical and interpretive problems of monitoring rates of growth and a comparison of the response rates achieved in the two types of monitoring.

RESULTS

Although the design of the NSHG was selected in order to monitor rates of growth this presents statistical problems. The usable information is contained in trends in attained height rather than in rates of growth. This study has achieved an average response rate of over 95%; less than 78% can be expected from the proposed survey of households.

CONCLUSIONS

The small loss of efficiency of the mixed longitudinal design compared with repeated cross-sectional studies is more than compensated for by its high response rate and the comparability of data over time, neither of which can be guaranteed by the proposed survey.

Prevalence of asthma and asthma-like symptoms in young adults living in three east Anglian towns

BACKGROUND

The European Community respiratory health survey is examining the prevalence of asthma and risk factors for asthma.

AIM

As part of this multinational survey, a study was undertaken to determine the prevalence of asthma and asthma-like symptoms in young adults living in Cambridge, Ipswich and Norwich using a postal questionnaire.

METHOD

A previously validated symptom questionnaire was sent to 2500 men and 2500 women aged 20-44 years living in and registered with a general practitioner in each of the three towns.

RESULTS

In total, approximately 9000 adults responded. The prevalence of symptoms suggestive of asthma was found to be similar in the three towns. Of respondents, 8% reported having been woken by an attack of shortness of breath at some time in the last 12 months, higher than previously reported. Five per cent reported having had an asthma attack.

CONCLUSION

General practitioners wishing to examine asthma prevalence in their own practice population could use a similar methodology.

Response to venepuncture for monitoring in primary schools

The feasibility and acceptability of collecting blood from children by venepuncture was assessed in a sample of 593 children from seven primary schools in Canterbury. Venepuncture is necessary to obtain blood for the measurement of haemoglobin, ferritin, and cholesterol in line with Department of Health surveys in England. Return of consent forms was 87%; 75% of parents in the total sample allowed their child to be tested. Response rates differed between schools. Only 4% of eligible children refused to participate at the time of testing. In 22 (3.7%) children a blood sample could not be obtained or the volume was insufficient for analysis. There was a significant difference in the failure rate between phlebotomists. Venepuncture in the school setting was technically feasible and acceptable. The reluctance of some groups in the community to participate may bias the sample.

The European Community Respiratory Health Survey

The European Community Respiratory Health Survey (ECRHS) was planned to answer specific questions about the distribution of asthma and health care given for asthma in the European Community. Specifically, the survey is designed to estimate variations in the prevalence of asthma, asthma-like symptoms and airway responsiveness; to estimate variations in exposures to known or suspected risk factors for asthma, and assess to what extent these variations explain the variations in the prevalence of disease; and to estimate differences in the use of medication for asthma. The protocol provides specific instructions on the sampling strategy adopted by the survey teams, as well as providing instructions on the use of questionnaires, the tests for allergy, lung function measurements, tests of airway responsiveness, and blood and urine collection. The principal data collection sheets and questionnaires are provided in the appendices, together with information on coding and quality control. The protocol is published as a reference for those who wish to know more of the methods used in the study, and also to give other groups who wish to collect comparable data access to the detailed methodology.

Serum total cholesterol and ferritin and blood haemoglobin concentrations in primary schoolchildren

Serum total cholesterol and ferritin and haemoglobin concentrations were measured in blood samples obtained by venepuncture in 378 of 593 children aged 5-6 or 8-9 years from seven primary schools in Canterbury. This study formed part of an investigation to assess the feasibility of including a venepuncture procedure for monitoring purposes in primary schoolchildren. Although only one child had a very low haemoglobin concentration of 61 g/l, a large percentage, 25% in the rising 6 years and 7% in the rising 9 years, had concentrations between 100 and 115 g/l--that is, less than the reference fifth centile. Eight per cent had a serum ferritin concentration less than 8 micrograms/l. Cholesterol concentrations higher than 5.2 mmol/l were found in 20% of the 5-6 year olds and 23% of the 8-9 year olds, and in 19% of boys and 25% of girls. About 5% of children had cholesterol concentrations above 6 mmol/l. The number of children with anaemia, iron deficiency, and high cholesterol raises serious concerns about the nutritional and coronary heart disease risk of British children.

Estimation in community surveys of total body fat of children using bioelectrical impedance or skinfold thickness measurements

OBJECTIVE

To compare skinfold thickness measurements with bioelectrical impedance analysis (BIA) as a measure of body fat for use in a survey of children (the National Study of Health and Growth).

DESIGN

Part cross-sectional, part repeated measurement study.

SETTING

A junior school in Bath.

SUBJECTS

42 boys and 33 girls aged from 9 to 11 years.

INTERVENTIONS

Measurements of BIA, height, weight, and triceps, biceps, subscapular and suprailiac skinfolds.

RESULTS

All measurements were highly repeatable with intraclass correlation coefficients > 0.90. The level of agreement between estimates of percentage of body fat derived from prediction equations based on impedance or skinfold measurements respectively was poor: the mean difference (impedance estimate minus skinfold estimate) was 4.67% (95% range -3.47 to 12.82) for boys and 7.81% (95% range 1.27 to 14.34) for girls. The two estimates were found to correlate highly (r = 0.83 for boys and r = 0.81 for girls) because weight, used to convert estimates of fat-free mass derived from impedance to fat mass, was highly correlated with impedance and moderately highly correlated with skinfold thicknesses. The correlations of resistance (R) and (H)2/R with skinfold thicknesses were very low. There was a moderate correlation of R and H2/R with log(weight-for-height index), but lower than that of log(weight-for-height index) with each of the skinfolds.

CONCLUSIONS

As currently available equations for converting impedance-based estimates of total body water to fat mass are not fully developed for use in children of varying ages, estimates of body fat calculated from skinfold thickness measurements remain preferable in epidemiological studies of children's health and growth.

Trends in weight-for-height and triceps skinfold thickness for English and Scottish children, 1972-1982 and 1982-1990

Data from the 1972, 1982 and 1990 surveys of the National Study of Health and Growth were used to calculate changes in height, weight, triceps skinfold thickness and weight-for-height index for children aged between 4.5 and 11.99 years. There were data for 7887, 6396 and 6420 white English children in the 3 years respectively, and data for 1586, 1191 and 1317 Scottish children. Increases in all measurements were found from 1972 to 1990, except for weight-for-height in English boys, and were generally greater from 1982 to 1990 than from 1972 to 1982. Approximately a third of the increases in weight-for-height and triceps skinfold thickness from 1972 to 1990 were associated with increases in parental body mass indices and decreases in family size. No consistent differences in trends were found between social groups. Greater trends were found for girls and for Scottish children, and Scottish boys are now heavier and fatter than their English counterparts. Trends in weight-for-height and fatness may have implications for future trends in coronary heart disease, and lessen the likelihood that the targets for reductions in obesity in the White Paper Health of the Nation will be met. Preventive measures should be directed at the entire population.

Factors affecting peak expiratory flow variability and bronchial reactivity in a random population sample

BACKGROUND

Bronchial reactivity measurements are widely used in epidemiological studies to provide an objective marker of asthma. There are, however, several potential advantages of measuring peak expiratory flow (PEF) variability instead, particularly in large studies. PEF variability and bronchial reactivity were compared in a population sample to assess the relationships of the two measurements to factors known to be associated with airways disease, and to compare their response rates.

METHODS

Subjects aged 18-65 were randomly selected from the electoral register of an administrative area in eastern England and randomised to attend either for a bronchial challenge test measuring the provocative dose of methacholine producing a 20% fall in FEV1 (PD20), or to measure PEF at two hourly intervals during waking hours for one week. Skin tests with common allergens were performed and a smoking history obtained. PEF variability was expressed as the amplitude % mean (highest - lowest x 100/mean).

RESULTS

A total of 273 subjects (69%) collected a PEF meter but a completed record sheet was returned by only 247 (62%); this was still significantly more than the 202 subjects (54%) who attended for and successfully completed a challenge test. Amplitude % mean was higher in women than in men (9.7% v 8.5%). In multiple regression analysis amplitude % mean increased significantly with age, mean skin weal diameter, and with current smoking. The odds of having a PD20 below 24.5 mumol increased with mean skin weal diameter and were greater in current smokers. Neither age nor sex had a significant effect on bronchial reactivity but there were significant interactions between age and the effects of both smoking and atopy.

CONCLUSIONS

The higher response rate associated with the use of PEF variability measurement, and the association with factors implicated in the pathogenesis of airways disease, suggest that PEF variability would be a useful measurement to employ in epidemiological studies.

Comparison of PD20 with two alternative measures of response to histamine challenge in epidemiological studies

Bronchial responsiveness to histamine or methacholine provides a useful objective measure for epidemiological studies of airways disease, but most people in a community population do not have a 20% fall in forced expiratory volume in one second (FEV1) with the highest dose administered. Histamine challenge data were analysed to compare the repeatability. Normality and separation of symptom groups of the early dose-response slope with provocative dose producing a 20% fall in FEV1 (PD20). Tests were continued until a 20% fall in FEV1 occurred, or 4 mumol had been given. Data were available for 510 randomly selected subjects, and for an additional 283 with wheeze. A repeat test was obtained in 104 individuals. PD20 was estimated by curve fitting, with extrapolation to 8 mumol. Least-squares slope of percentage decline in FEV1 on histamine dose was calculated, using all the measured points and two-point slope as the fall from the post-saline measurement to the maximum cumulative dose divided by the maximum dose. Log transformation of PD20 and shifted reciprocal transformations of slope produced constant variance. Over all subjects the three measures had similar repeatability; in subjects with PD20 > 8 mumol the intraclass correlation for two-point slope was only 0.26, but was 0.66 for least-squares slope. Neither measure of slope was normally distributed, but the distribution of log(PD20) was consistent with a censored normal distribution. In conclusion, least-squares slope is preferable to two-point slope for epidemiological studies.(ABSTRACT TRUNCATED AT 250 WORDS)

Comparison of PD20 with two alternative measures of response to histamine challenge in epidemiological studies

Bronchial responsiveness to histamine or methacholine provides a useful objective measure for epidemiological studies of airways disease, but most people in a community population do not have a 20% fall in forced expiratory volume in one second (FEV1) with the highest dose administered. Histamine challenge data were analysed to compare the repeatability. Normality and separation of symptom groups of the early dose-response slope with provocative dose producing a 20% fall in FEV1 (PD20). Tests were continued until a 20% fall in FEV1 occurred, or 4 mumol had been given. Data were available for 510 randomly selected subjects, and for an additional 283 with wheeze. A repeat test was obtained in 104 individuals. PD20 was estimated by curve fitting, with extrapolation to 8 mumol. Least-squares slope of percentage decline in FEV1 on histamine dose was calculated, using all the measured points and two-point slope as the fall from the post-saline measurement to the maximum cumulative dose divided by the maximum dose. Log transformation of PD20 and shifted reciprocal transformations of slope produced constant variance. Over all subjects the three measures had similar repeatability; in subjects with PD20 > 8 mumol the intraclass correlation for two-point slope was only 0.26, but was 0.66 for least-squares slope. Neither measure of slope was normally distributed, but the distribution of log(PD20) was consistent with a censored normal distribution. In conclusion, least-squares slope is preferable to two-point slope for epidemiological studies.(ABSTRACT TRUNCATED AT 250 WORDS)

Validation of a food frequency questionnaire for assessing dietary intake in a study of coronary heart disease risk factors in children

A food frequency questionnaire (FFQ) was designed and validated for use in an epidemiological survey measuring coronary heart disease risk factors in British children. The questionnaire asked about the frequency of children's consumption of 35 food items over the previous month and was completed by parents/guardians. Food items represented a range of food groups from which children may eat, with emphasis on foods with a high fat and fibre content. The questionnaire was validated against 14 daily recalls of consumption, using the same food list as the FFQ. The sample consisted of 272 children, aged 5-11 years, The response rate for the FFQ was 92% and for the recall, 82% in the first week and 66% in the second week. The level of agreement between the FFQ and the recall was measured by calculating (McGinnis JM & Nestle M, 1989, Am. J. Clin. Nutr. 49, 23-28) the median difference between results from the two methods and (Willett WC (ed.), 1990, Nutritional epidemiology, ch. 15; Oxford University Press) the percentage of children classified by FFQ to within +/- 1 day per week of the recall. Median differences between individual items on the questionnaire and the recall were < or = 0.5 days for 91% of items and equal to 1 day for the remainder. The percentage agreement to within +/- 1 day per week between frequencies reported in the two methods ranged from 99.8% for lamb to 46.8% for low-fibre cereal. Better agreement was found for food items representative of fat intake than those of fibre intake.(ABSTRACT TRUNCATED AT 250 WORDS)

Effects of prematurity and intrauterine growth on respiratory health and lung function in childhood

OBJECTIVE

To determine whether birth weight and gestational age are associated with respiratory illness and lung function in children aged 5-11 years.

DESIGN

Cross sectional analysis of parent reported birth weight, gestational age, and respiratory symptoms; parental smoking and social conditions; forced vital capacity (FVC), forced expiratory volume in one second (FEV1), forced expiratory rates between 25% and 75% and 75% and 85% (FEF25-75 and FEF75-85), and height.

SETTING

Primary schools in England and Scotland in 1990.

SUBJECTS

5573 children aged 5-11 (63.3% of eligible children) had respiratory symptoms analysed and 2036 children (67.1% of eligible children) had lung function measured.

MAIN OUTCOME MEASURES

Symptoms of asthma, bronchitis, occasional and frequent wheeze, cough first thing in the morning, and cough at any other time and lung function.

RESULTS

Birth weight adjusted for gestational age was significantly associated with all lung function measurements, except FEF25-75. The association remained for FVC (b = 0.475, 95% confidence interval 0.181 to 0.769) and FEV1 (b = 0.502, 0.204 to 0.800) after adjustment for gestational age, parental smoking, and social factors. FEF75-85 was the only lung function related to gestational age. Respiratory symptoms, especially wheeze most days (adjusted odds ratio 0.9, 0.84 to 0.97) were significantly associated with prematurity. Every extra week of gestation reduced the risk of severe wheeze by about 10%.

CONCLUSIONS

Lung function is affected mainly by intrauterine environment while respiratory illness, especially wheezing, in childhood is related to prematurity.

Lung function, respiratory illness, and passive smoking in British primary school children

BACKGROUND

Many studies have reported a significant association between parents' smoking and reduced lung function in their children, but often the association has been found to be significant only in relation to maternal smoking. There have been few epidemiological studies on this topic in Britain.

METHODS

Spirometry, in 2756 children aged 6.50-11.99 years, was carried out in a representative sample of English children, an inner city and ethnic minority sample, and a Scottish sample. Forced vital capacity (FVC), forced expiratory volume in one second (FEV1), and forced expiratory flow rates of 25-75% and 75-85% (FEF25-75 and FEF75-85) were measured and standardised scores obtained separately for the English representative sample, the Scottish sample and subgroups in the inner city sample, white and Afro-Caribbean children and those originating from the Indian subcontinent. Multiple regression analyses were used to assess associations of FVC, FEV1, FEF25-75 and FEF75-85 with the passive smoking and respiratory illness, with adjustment for a large number of potential confounders. Passive smoking was defined in terms of reported number of cigarettes smoked at home by each parent. The respiratory symptoms and illnesses assessed were wheeze, asthma and bronchitis attacks, cough in the morning, and cough at any other time as reported by parents.

RESULTS

Maternal smoking, but not paternal smoking, was associated with reduced FEF25-75 and FEF75-85 in boys. No association was found between passive smoking and lung function in girls, but in an analysis including both sexes the interaction of sex and parental smoking on lung function was not significant. With few exceptions, FEV1, FEF25-75 and FEF75-85 were reduced in children with wheeze and asthma attacks.

CONCLUSIONS

The effect of passive smoking may depend on the close contact of a parent with a susceptible child as only maternal smoking in boys was significantly associated with impaired lung function. However, this explanation remains unsubstantiated. A parent's report of wheeze and asthma attacks in the child is reflected in reduced lung function.

Portable liquid oxygen and exercise ability in severe respiratory disability

BACKGROUND

The development of portable liquid oxygen systems, capable of delivering high flow rate oxygen for long periods, justifies reassessment of the value of supplemental oxygen to aid exercise tolerance in patients with chronic respiratory insufficiency. The type of exercise test and the low oxygen flow rates previously used may account for the variable and often poor responses to supplemental oxygen reported in earlier studies.

METHODS

The walking tolerance of 30 patients with severe respiratory disability was measured while they were breathing air and increasing doses of supplemental oxygen (2, 4, 6 1/min) by using both the standard six minute walking test and an endurance walking test. To assess the initial learning effect and repeatability of the walking tests, three six minute walks and three endurance walks were performed on day 1 and a single walk of each type on days 2, 3, and 14. In addition, oxygen dosing studies were performed on days 2 and 3 after the initial baseline walking tests. Each dosing study comprised four endurance walking tests or four six minute walking tests with patients breathing either air at a flow rate of 4 1/min from a portable cylinder or supplemental oxygen at a flow rate of 2, 4 or 6 1/min from a portable liquid oxygen supply. The order of the tests was randomised. Walking distance with each flow rate of oxygen was compared with walking distance with patients carrying cylinder air and for the initial unburdened walks. Breathlessness was assessed by visual analogue scoring on completion of each walk.

RESULTS

Exercise ability and breathlessness were significantly improved with supplemental oxygen and this benefit outweighed the reduction in performance resulting from carrying the portable device. Supplemental oxygen at flow rates of 2, 4, and 6 1/min increased mean endurance walking distances by 37.9%, 67.7% and 85.0% and six minute walking distances by 19.2%, 34.5%, and 36.3% by comparison with distances when the patient was carrying air with a flow rate of 4 1/min. The additional work of carrying the portable gas supply reduced endurance walking distance by 22.2% and six minute walking distance by 14.1% by comparison with a baseline unburdened walk. Comparison of supplemental oxygen at 2, 4, and 6 1/min with the baseline unburdened performance showed increased endurance walking distances of 7.3%, 30.4%, and 43.9% and six minute walking distances of 2.3%, 15.5%, and 17.0%. Walking distance was increased by more than 50% by comparison with an unburdened walk in seven patients with the endurance walking test but in only three patients with the six minute walking test. The benefit was similar in patients with obstructive and with interstitial lung disease. Individual responses were variable and only desaturation during the baseline walk in patients with obstructive lung disease had any predictive value for benefit with oxygen.

CONCLUSION

As there was no clear relation between response to oxygen therapy and the patients' characteristics, assessment for supplemental oxygen therapy will depend on exercise testing. It is suggested that portable oxygen should be considered only if a patient shows a 50% improvement in exercise ability with high flow rate oxygen (4-6 1/min) by comparison with an unburdened walk.

Height and age adjustment for cross sectional studies of lung function in children aged 6-11 years

BACKGROUND

No standard exists for the adjustment of lung function for height and age in children. Multiple regression should not be used on untransformed data because, for example, forced expiratory volume (FEV1), though normally distributed for height, age, and sex, has increasing standard deviation. A solution to the conflict is proposed.

METHODS

Spirometry on representative samples of children aged 6.5 to 11.99 years in primary schools in England. After exclusion of children who did not provide two repeatable blows 910 white English boys and 722 girls had data on FEV1 and height. Means and standard deviations of FEV1 divided by height were plotted to determine whether logarithmic transformation of FEV1 was appropriate. Multiple regression was used to give predicted FEV1 for height and age on the transformed scale; back transformation gave predicted values in litres. Other lung function measures were analysed, and data on inner city children, children from ethnic minority groups, and Scottish children were described.

RESULTS

After logarithmic (ln) transformation of FEV1 standard deviation was constant. The ratios of actual and predicted values of FEV1 were normally distributed in boys and girls. From the means and standard deviations of these distributions, and the predicted values, centiles and standard deviation scores can be calculated.

CONCLUSION

The method described is valid because the assumption of stable variance for multiple regression was satisfied on the log scale and the variation of ratios of actual to predicted values on the original scale was well described by a normal distribution. The adoption of the method will lead to uniformity and greater ease of comparison of research findings.

Weight-for-height in children aged 4-12 years. A new index compared to the normalized body mass index

Data from the 1990 survey of the National Study of Health and Growth, comprising 3357 white English boys and 3050 white English girls, were used to construct and evaluate a new index of weight-for-height. Prediction of triceps + subscapular skinfold thickness, using half of the data, led to the index (weight -9)/height3.7, where weight was in kilograms, height in metres. Using the second half of the data the index was shown to have almost as good a correlation with normalized (triceps + subscapular skinfold) as the body mass index (BMI) in children aged 4-12 years. The new weight-for-height index had stable variance from age 4 to 12 years after simple log transformation, unlike the BMI which required transformation via three age-related parameters prior to analysis or centile calculation. Neither index was a good proxy for skinfold measurements for comparison of ethnic groups. From data from inner city areas surveyed in 1989 children of Indian subcontinent origin had substantially lower weight-for-height, but only Gujarati children were thinner than white children, whereas Afro-Caribbean children were slightly heavier for their height, but thinner. Centiles of weight-for-height, and of BMI by age, for ages 4-12 years, were calculated for white English boys and white English girls.

Trends in body mass index in young adults in England and Scotland from 1973 to 1988

STUDY OBJECTIVE

The aim was to determine whether the mean body mass index of young adults in England and Scotland had increased during the years 1973 to 1988.

DESIGN

The study was an analysis of reported heights and weights for parents of children participating in a mixed longitudinal study of children's growth.

SETTING

20 study areas (16 in England and four in Scotland) were selected by stratified random sampling.

SUBJECTS

Subjects were parents of 5229 children who were new entrants to the study in the years 1973-6 and 1982-8. After excluding cases containing missing values on continuous variables, data for 4568 (87%) women (mean age 30 years) and 4029 (77%) men (mean age approximately 32 years) were analyzed.

MEASUREMENTS AND MAIN RESULTS

Changes in weight for height over time were determined using body mass index as dependent variable, adjusting for age, social class, family size, and study area. In women, mean body mass index showed an annual increase of 0.10 (95% CI 0.03 to 0.17%) per year of study. Body mass index also increased with increasing age and family size and was greatest for women with husbands in manual occupations. The secular trend in body mass index in women was not explained by changes in the distribution of these variables. The proportion of women with BMI greater than 25 kg/m2 increased over the study period. In men the secular trend in body mass index was not quite significant in this age group [annual increase 0.05% (-0.01 to 0.12%)].

CONCLUSIONS

There has been an increase in the body mass index of young women in England and Scotland over the years 1973 to 1988. This increase was not explained by changes in the age, parity, social class of the subjects sampled. Evidence of a trend in men was not found.

A new method for calculation of height centiles for preadolescent children

A new method of calculation of cross-sectional attained height centiles is proposed. This exploits the variance-stabilizing transformation, which has the form log(a+b height) up to age 11 years. Residuals of transformed height about the expected value for age are antilogged to give a single Gaussian distribution from which centiles can be calculated. In contrast to previous methods the standard errors of the centiles can be estimated without imprecise adjustment for smoothing. The parameters a and b of the transformation do not need precise estimation and the method can be used for population subgroups from which large samples cannot be obtained.

Comparison of bronchial reactivity and peak expiratory flow variability measurements for epidemiologic studies

Inclusion of a standardized measurement of airway function is important in epidemiologic studies of asthma to facilitate comparison between different studies. Bronchial reactivity is widely used in such studies, but measurement of peak expiratory flow (PEF) variability has a number of potential advantages. We compared PEF variability with methacholine challenge tests in a community population sample. Subjects selected at random (n = 95) and on the basis of having experienced wheeze in the last 12 months (n = 130) performed a challenge test with methacholine to a maximum dose of 12.25 mumol and made serial PEF recordings every 2 h for a week. PEF variability was expressed as mean daily maximum amplitude as a percentage of the mean (amplitude % mean). Increased bronchial reactivity and PEF variability were arbitrarily defined as values above the 10th or below the 90th percentiles in the random sample. A measurement of amplitude % mean was available from all 225 subjects, whereas only 115 (51%) had a measurable PD20 methacholine. PD20 measurements correlated weakly but significantly with amplitude % mean (r = -0.44, p less than 0.001). Increased values of both bronchial reactivity and PEF variability were related to the presence of respiratory symptoms in the week before testing. Asthma was more strongly related to increased bronchial reactivity than to PEF variability. Both measurements showed a strong association with atopy and the intraclass correlation coefficients (ratio of between-subject to total variance) were similar for both.(ABSTRACT TRUNCATED AT 250 WORDS)

Interregional variations in measures of health from the Health and Lifestyle Survey and their relation with indicators of health care need in England

STUDY OBJECTIVE

The aim was to assess the extent to which a range of routinely available need indicators which have been suggested for use in NHS spatial resource allocation formulas were associated geographically in England with the different dimensions of population health status collected in the 1985/86 Health and Lifestyle Survey (HLS).

DESIGN

Regional health authorities were ranked according to each of the HLS health variables which varied significantly between authorities. The HLS health variables were regressed on a selection from the range of routinely available morbidity and socioeconomic indicators available from the 1981 census. The potential need indicators were also regressed on the health variables.

SETTING

The analyses were undertaken at individual level and at regional health authority level in England.

SUBJECTS

The study comprised the English component of the HLS random sample representative of the population in private households in Great Britain.

MAIN RESULTS

The different HLS health variables did not yield consistent regional health authority rankings. Among the variables, forced expiratory volume in one second (FEV1) and self assessed health appeared to be associated with most of the other health and need variables except longstanding illness. Longstanding illness was not strongly associated with any of the other HLS health variables but appeared to show some association with three deprivation indices constructed from the 1981 Census.

CONCLUSIONS

There may be a case for including a measure of chronic ill health in the new NHS system of capitated finance in addition to the all cause standardised mortality ratio which is used currently as a measure of need for health care.