Dr. Ismail and Colleagues Reply

To the Editor: Against the background of their own work on morphogenic variants in schizophrenia (1997), Drs. Trixler and Ti comment on our study on minor physical anomalies in 60 patients with schizophrenia, 21 normal siblings of these patients, and 75 demographically similar normal comparison subjects and raise issues concerning 1) the reliability of the Waldrop scale, 2) the relevance of items on the Waldrop scale for the development of schizophrenia, and 3) their own approach for attempting to establish the timing of insults or maldevelopment related to schizophrenia. Our study employed the Waldrop scale and 23 additional minor physical anomalies identified by means of a review of the pediatric literature. Our experience is that quite satisfactory reliability can be obtained for both the Waldrop scale and additional items when researchers train together and decide on the operational criteria used to define particular minor physical anomalies. For example, we find very high levels of agreement, not on total score, but on measurements of extent of curved fifth finger, actual head circumference, interpupil distance, and the like. Many specific minor physical anomalies occur relatively uncommonly even in select patient groups and even when percentage agreement between assessors is very high; establishing nominal reliability expressed in terms of kappa may be difficult because of limitations in the marginal frequencies within the group studied.

Results based on comparisons of patients with schizophrenia and patients with alcohol dependence (Trixler et al., 1997) may not represent an optimal strategy for determining what particular malformations and deviations are relevant for the development of schizophrenia. Dr. Trixler et al. found no significant increase in total scores in patients with schizophrenia and significant increases for only five (of 34) specific items. In contrast, our comparison with 75 demographically similar normal subjects yielded evidence of rather widespread dysmorphogenesis among patients. We also found significant increases in minor physical anomalies in total and in all six body areas studied and on 12 of 41 specific minor physical anomalies (p=0.02) in patients, including a number of items studied by Dr. Trixler et al. but not found to differentiate patients with schizophrenia from those with alcohol dependence. Regarding their relevance for schizophrenia, both items on the Waldrop scale and our new items discriminated patients from normal comparison subjects, but logistic regression analysis demonstrated that the specific minor physical anomalies making the strongest independent contribution to the discrimination of patients from normal comparison subjects were items on the Waldrop scale (curved fifth finger, epicanthus, and steepled palate). (Dr. Trixler et al. [1997, p. 693] classified these three minor physical anomalies as “phenogenetic variants” and proposed that they are developmentally identical to “normal variants.”) The items from the Waldrop scale also provided a better differentiation of normal siblings of patients and comparison subjects than did the new items.

We quite agree that a specification of the timing of the relevant insults would make a contribution to our understanding of the etiology of schizophrenia, but we raise the question of whether specification of the timing of the development of the malformation (if possible) would necessarily inform us about the timing of the insult. For example, one key minor physical anomaly associated with schizophrenia across different studies is palatal abnormality. The development of palatal morphology has a wide temporal window, which originates at 6–9 gestational weeks and continues until 16–17 weeks (1), i.e., spanning trimesters 1 and 2. Further questions could be raised about whether insults relevant for palatal morphology would actually have to occur during this very broad window or could happen even before this interval. Finally, in our effort to pinpoint exact relevant periods, remember that it is by no means established that only early prenatal factors are relevant for the later development of schizophrenia.