C
ORRESPONDENCE
Reactivity to a 35% CO
2Challenge in
Healthy First-Degree Relatives of Patients
with Panic Disorder
To the Editor:
The article by van Beek and Griez (2000) contains what appear to be some errors. On page 832, the first item in the first column in Table 4 says “VAS-A mean$25,” but presented are means6SDs. I presume this is a typesetting error of some kind. More concerning, however, are the results for the x2 tests reported by the authors. In the section VAS-A Post Score, they report ax2value of 4.46 with a p of,.05. My calculation yields ax2value of 3.61 with a “trend” level value of significance of p , .06.
The next section, CO2-Induced Panic Attacks, states that panic
attacks occurred in “twenty-one first-degree relatives (44%) and 12 control subjects (12%).” Clearly 12/50 is not 12%, so this must be wrong. Then, ax2value identical to that reported in the above paragraph (i.e., 4.46, p, .05) is reported here. But this cannot be correct. My calculation yields ax2value of 2.895 with a “trend” level of significance of p, .09.
These are two critical values upon which hinge the “take home message” from this study. It would be helpful to the field if the authors could clarify these inconsistencies in the presentation of the data.
Murray B. Stein
University of California San Diego Department of Psychiatry
8950 Villa La Jolla Drive Suite 2243
La Jolla CA 92037
PII S0006-3223(00)01054-4
Reference
van Beek N, Griez E (2000): Reactivity to a 35% CO2challenge
in healthy first-degree relatives of patients with panic disor-der. Biol Psychiatry 47:830 – 835.
Reply
To the Editor:
We thank Dr. Stein for his attentive interest and appreciate the present opportunity to correct the presentation of our data.
The abbreviated wording of Table 4 may have been mislead-ing indeed. Both the number and percentages of subjects who, after CO2intake, reported a Visual Analogue Scale (VAS) value
equal to or above 25 are shown. There were 22 out of the 50 first-degree relatives (44%) and 12 out of the 50 control subjects (24%).
Dr. Stein is right when he observes that 12/50 is not 12%. Neither does 44% fit 21/50. It is obvious here that confusing mistakes were overlooked in the final check of our report. We apologize to our readers. The sentence quoted by Dr. Stein should be: “twenty-two first degree relatives (44%) and 12 control subjects (24%).”
Accordingly, ax2calculation for panic attacks relies on the same 2 3 2 contingency table as VAS-A Post Score. We maintain that this calculation yields a result of 4.46. Although he does not mention so, Dr. Stein apparently refers to a calculation that includes Yates’ correction for continuity. Application of Yates’ correction is recommended when the frequency of any cell in a 232 contingency table is less than 10 (Ferguson 1966, 207). Otherwise, Yates correction may provide overly conserva-tive results (Conover 1974). In the present case the lowest expected frequency was 17. Consequently, there is little ground for using corrected statistics.
Finally, we would like to point out that the principal outcome measure of our study was CO2-induced increase in anxiety. This
increase was higher in the first-degree relatives of panic disorder patients than in the control subjects (Table 3). That is the finding upon which the “take home message” undoubtedly relies.
Eric Griez Nicole van Beek
Maastricht University
Department of Psychiatry & Neuropsychology Section of Clinical and Biological Psychiatry PO Box 88
6200 AB Maastricht The Netherlands
PII S0006-3223(00)01055-6
References
Conover WJ (1974): Some reasons for not using the Yates continuity correction on 232 contingency tables. J Am Stat Assoc 69(346):374 –376.
Ferguson GA (1966): Statistical Analysis in Psychology and Education. London: McGraw-Hill.
