some biochemical correlates of psychological states—they are a prerequisite for any hormonal
byproduct hypothesis. Interestingly, changes in progesterone, estrogen, prolactin, and cortisol
levels appear to have surprisingly little to do with PPD (Harris, 1994; O’Hara, 1995; Owens,
Smith, Brinsmead & al, 1987; Smith, Owens, Brinsmead & al, 1987). Additionally, fathers—
who are not undergoing dramatic hormonal changes—experience PPD at 50-100% the rate of
mothers (e.g., O’Hara, 1985; Richman, Raskin & Gaines, 1991).
The relationship between postpartum thyroid dysfunction and PPD may support a
“hormonal” etiology for a small number of cases of PPD. The results of Pop, et al. (1991), as
well as the work of Harris et al. (1992) and Pederson et al. (1993) clearly demonstrate that
thyroid dysfunction is associated with a small but significant fraction of PPD4. Nevertheless, in
light of the strong associations of PPD with social support and infant variables, and its frequent
occurrence in fathers, a strictly hormonal etiology for most cases of PPD is untenable.
Other correlates and non-correlates
Most studies find a strong correlation between either a previous history of emotional
problems, depression, or depression during pregnancy, and PPD (Atkinson & Rickel, 1984;
Cutrona & Troutman, 1986; Gotlib et al., 1991; Graff et al., 1991; Logsdon et al., 1994;
O’Hara et al., 1984; O’Hara et al., 1983; Whiffen, 1988; Whiffen & Gotlib, 1993).
Unfortunately, the defection hypothesis does not clearly illuminate this important aspect of
PPD.
Most demographic variables are not reliably associated with PPD. These include age
(Campbell & Cohn, 1991; Gotlib, Whiffen, Mount, Milne & et al., 1989; Gotlib et al., 1991),
marital status (Gotlib et al., 1991), occupational level (Campbell & Cohn, 1991), work status
(Campbell & Cohn, 1991; Gotlib et al., 1989; Gotlib et al., 1991), the sex of the infant
(Campbell & Cohn, 1991), the number of children in the home (Gotlib, Whiffen et al. 1989;
Gotlib, Whiffen et al. 1991; cf. O’Hara 1986), and education (Gotlib et al. 1989; Gotlib et al.
1991; cf. Campbell and Cohn 1991; O’Hara 1986). The expected correlations between PPD
and age will be discussed below. No strong correlation is expected with marital status as this
merely represents the formal status of a mother’s relationship and not the presence or absence
of an investing mate. While the occupational level, work status, and education of the mother
might be expected to predict availability of resources, and thus PPD, the relationship of these
variables to resources is ambiguous at best. If the mother has a low occupational or educational
level, but her husband has a high occupational level, then the mother’s access to resources will
be sufficient. Two incomes from low occupational levels might also be sufficient. As noted
above, financially impoverished populations do appear to experience higher levels of PPD, and
maternal unemployment (as opposed to housewife status) and head-of-household
unemployment have both been found to be risk factors for PPD.
Number of children in the home and sex of the infant are both potentially evolutionarily
significant variables, but the data reported in existing studies do not allow tests of hypotheses
regarding differential investment based on sex or number of existing offspring. The
(unreported) age distribution of the existing children would be critical for evaluating the costs
and benefits they represent to the mother. In ancestral environments, infants represented
significantly higher costs and lower benefits (due to high rates of infant mortality) than did
older children. Children of different sexes can also have different relative values to the mother
when resource availability varies (Trivers and Willard, 1973), and thus, in conjunction, are
possible predictors of PPD, but this hypothesis cannot be assessed with existing data.
4Though here, too, there may be a connection with fetal strategies and lack of paternal support. See Haig (1993).
17