One neurotransmitter that apparently contributes to anxiety disorders is γ-aminobutyric acid (GABA). Goddard et al. (p. 2186) report a dynamic test of GABA functioning in patients with panic disorder. At baseline, the patients’ mean GABA level in the occipital cortex was much lower than that of healthy subjects. After administration of the antianxiety agent clonazepam, the healthy subjects had a 24% decrease in GABA, but the patients showed no significant change. Their level also did not change after 4 weeks of clonazepam treatment, suggesting that low GABA is a characteristic of people with vulnerability to panic, rather than a reflection of the panic state itself.

The hippocampus is critical for memory encoding but is sensitive to chronic stress. Reduced hippocampal volume is associated with several mental illnesses, including posttraumatic stress disorder (PTSD). However, PTSD’s effects can be complicated by the trauma itself and by depression and substance use disorders, which are common in people with PTSD and are also associated with a smaller hippocampus. Winter and Irle (p. 2194) studied burn survivors, who have lower rates of depression and substance abuse. Hippocampal volume did not differ significantly between burn patients with PTSD and those without PTSD, but both groups had smaller volumes of the right hippocampus than healthy subjects. Patients with larger burns had smaller volumes, confirming the association with trauma rather than PTSD.

Most women do not become depressed during perimenopause. Attempts to predict who the unfortunate few will be have examined various factors but only at isolated time points. Schmidt et al. (p. 2238) monitored reproductive status and mood in 29 women from premenopause to amenorrhea. The 24 months surrounding the final menstrual period were associated with a greater risk of depression, but depression was not associated with any of the variables assessed, including previous depression, premenstrual syndrome, and stressful life events. Hot flushes were present in eight of the nine women who became depressed but in five cases were not temporally related. Depression was more common during late perimenopause than during early stages, indirectly supporting the role of endocrine mechanisms.

Sadness, bereavement, and clinical depression are distinct emotional experiences. Abnormalities in brain activity among depressed people may not extend to sadness and grief, and brain functioning during induced sadness reflects the subject’s memory of a sad state, not the real thing. Najib et al. (p. 2245) measured brain activity in women currently grieving the loss of a romantic relationship. Increased activity with sad thoughts of the ex-lover was generally found in posterior brain regions, and decreases were more prominent on the left and in anterior regions. In nearly all regions showing decreases with sad thoughts (grief state), decreases were greater in women with higher baseline grief ratings. The findings largely parallel previous results for sadness and depression but differ in some respects, confirming the uniqueness of the grief state.

A hospital that performs a surgical procedure frequently is likely to have a lower mortality rate for that procedure. Druss et al. (p. 2282) report a similar relationship between mental health services and mental health quality in a data set on 384 U.S. health plans covering 73 million enrollees nationwide. In 1999, plans with the fewest enrollees using outpatient mental health services, fewest enrollees hospitalized for mental illness, or fewest days of inpatient mental health treatment were more likely to have poor outcomes in terms of hospital follow-up and management of antidepressant medication. These findings suggest that for U.S. health plans, "practice" in providing mental health care is associated with better quality of care.

Estimates of the suicide rate of physicians have been higher than the national average, particularly for female physicians, but are controversial. Questions relate to methodologic quality, publication bias, and low numbers of female physicians. Schernhammer and Colditz (p. 2295) addressed these issues in a meta-analysis of 25 studies between 1960 and 2003. The suicide rate for male physicians was 1.4 times as high as for men in the general population. The rate for female physicians was 2.3 times as high as the average. Study quality did not affect these rates substantially, but publication bias had a nonsignificant effect on the rate for female physicians; i.e., small studies may have overestimated this rate. These findings confirm earlier reports of high suicide rates for physicians and the need for specialized interventions.F1

Benedict-Augustin Morel, (1809–1873), (p. 2185)F2