Risk factors of cerebral palsy.

Re: Risk factors for cerebral palsy: a case control The results of the Greek study would be more easily applied to other populations if the sample were study in Greece: Scand. J. Soc. Med:24;14–26. described in greater detail. The authors write ‘‘The protocol ... called for the identification of all children Petridou et al (1) are to be congratulated for exploring the problem of cerebral palsy in Greece. with an established diagnosis of CP born in the Greater Athens area,’’ but unless there is a surprisHowever their addition to the literature could be more valuable with the addition of some further ingly low birthrate or prevalence of cerebral palsy, their sample is likely to constitute only about one information. I would also like to clarify references to our work. fifth of the population sample. Their methods of ascertainment may bias the sample towards the more Our work (2) [ref. 21] is referenced: ‘‘Following the lead of other investigators (21) variables were distingseverely impaired. We are told that 58/103 children were functional with reference to the Oxford form uished ... on the basis of their presumed time and mode of action or time of ascertainment and (21) for describing disability. But the term functional is not defined in that form which has 4 functional the assumption being that interrelations and, therefore, confounding would be stronger among variables categories for each of the axis, trunk, lower and upper limbs. The frequency of associated non-motor within a group than among variables between groups.’’ Firstly, we categorised variables ONLY on impairment could give clues concerning the distribution of functional severity but these are not the basis of their presumed time of aetiological influence and NOT of their time of ascertainment. mentioned. Nor is any mention made of CP of post neonatal For example, a congenital malformation was not categorised as a perinatal variable (the most likely origin, ie. motor deficit recognised only following a well documented post-neonatal, early childhood time of recognition) but as an antenatal variable. Congenital malformations arise antenatally and may event with the capacity for cerebral damage (eg. cerebral infection, head trauma, anoxia such as near be a direct cause of the motor impairment (e.g. a CNS malformation), may reflect an intrauterine drowning or suffociation). Were such events sought? If they were, were these cases excluded? insult of aetiological significance for the motor impairment or may be coincidental. Secondly, I made The value of the paper would also be augmented if variables were better defined: eg. no assumptions concerning the strength of confounding and temporal proximity of events. I categ(i) Interpregnancy interval: is this the time between the termination of the pregnancy prior to the orised by presumed time of aetiological influence in order to seek time ordered aetiological pathways (3). index pregnancy and the conception of the index pregnancy? How was a subject categorised if This was accomplished by time ordered introduction of variables into the regression equation and forcing there had been no pregnancies prior to the index? (The numbers add up to 103 for cases so they retention of earlier associated variables if they were associated with later risk factors. Later steps in an have not been excluded.) Are they included in the −1 months category? Or did conception aetiological pathway will always have a stronger association with outcome than earlier ones, therefore within a month of termination of the previous pregnancy occur for nearly half of the mothers if all variables are introduced into the model simultaneously the aetiological significance of earlier steps of cases and more than one third of mothers of controls? in an aetiological pathway is unlikely to be recognised. In our study confounding was both sought (ii) The variable ‘previous lbw infant’ includes a category ‘no previous’ with 31 cases and 33 and recognised across time domains (e.g. the association with long distance postnatal transport to neocontrols. These are the same numbers as for sibship size=1. Does this mean that no first born natal intensive care was confounded by adverse perinatal events). CPs had any subsequently born sibs (up to the