Study does not rule out effect of social class

EDITOR—In a secondary analysis of the British general household survey, Cooper et al conclude that the use of health services by children and young people is equitable in terms of social class status.1 It is difficult either to agree or to disagree with them, since the results of the logistic regression analysis in table 3 do not include any details of their main social class variable (socioeconomic group of the head of family unit). We are told only that this variable was not significant and were not offered P values, odds ratios, or confidence intervals. The practice of presenting results only for variables that are statistically significant (P<0.05) is adopted throughout table 3. We are surprised that the BMJ found this acceptable, since it ignores the journal's own published statistical guidelines.2

Cooper et al found a significant association between socioeconomic group and inpatient admissions in their univariate model (table 1), and a significant association between housing tenure and inpatient admissions in their logistic regression model (table 3). One would expect there to be a strong link between socioeconomic group and housing tenure, with a higher proportion of people from lower socioeconomic groups living in local authority housing. The logistic regression model does not include both housing tenure and socioeconomic group simultaneously since these variables would provide very similar information. Cooper et al themselves claim that both socioeconomic group and housing tenure are measures of social class. Therefore, all the indications are that inpatient admissions for children are indeed related to social class, after adjustment for morbidity, with higher admission rates among the lower social classes.

Motivation for consultations may explain differential referral patterns

EDITOR—The use of health services by children and young people is an area that has been neglected. Cooper et al's study goes some way to redressing this imbalance.1 Their assertion, however, that …