sion, which would be expected to be a moderator. The inconsistencies are not owing to how antenatal depression was measured, since all but one of the studies relied on symptom scales rather than on diagnoses.
Another plausible explanation for the inconsistencies is comorbid alcohol or substance use during pregnancy. Researchers have found high rates of antenatal depression co-occurring with cigarette smoking, alcohol consumption, and abuse of such substances as cocaine, often in combination with each other (e.g., Amaro and Zuckerman, 1990, 1991; Amaro, Zuckerman, and Cabral, 1989; Zuckerman, Amaro, and Beardslee, 1987; Zuckerman et al., 1989). For example, depression is especially prevalent (35–56 percent) in samples of drug-dependent pregnant women (Burns et al., 1985; Fitzsimons et al., 2007; Regan et al., 1982).
Several researchers have found that infants of depressed mothers, relative to controls, have more difficult temperament. Whiffen and Gotlib (1989) found that infants of depressed mothers were perceived as more difficult to care for and more bothersome. A large study in Japan found direct effects of maternal depression on the infant temperament constructs of “frustration tolerance” and “fear of strangers and strange situations” (Sugawara et al., 1999). A meta-analysis found a significant, moderate correlation between postpartum depression and infant temperament, with a 95 percent confidence interval that ranged from 0.26 to 0.37 (Beck, 1996). Although relying on depressed mothers as reporters of their children’s temperament has raised concerns (Forman et al., 2003), researchers have found these significant associations even with lab-based measures of temperament (Goldsmith and Rothbart, 1994) and with self-report measures such as those developed by Rothbart (Garstein and Rothbart, 2003), which minimize maternal report bias by asking situation-specific questions and taking advantage of the extent and breadth of experience that mothers have with their infants.
Among the potential moderators or correlated risk factors that have been studied, co-occurring anxiety in mothers has been found to play a role in associations between maternal depression and infant temperament, but the precise role is not clear. It may be that the anxiety, known to be highly correlated with depression, matters more. In one study, maternal trait anxiety predicted difficult infant temperament independent of antenatal and postnatal depression scores (Austin et al., 2005). Another study found that both anxiety and depression in mothers matter. One study, which measured temperament not only with maternal reports but also with observations, found that mild parental dysphoria and mild parental anxiety were associ-