Postpartum-specific anxiety as a predictor of infant-feeding outcomes and perceptions of infant-feeding behaviours: new evidence for childbearing specific measures of mood

Postpartum anxiety (PPA) remains among the most under-studied, under-diagnosed, and under-treated complications of childbirth (Smith and Kipnis 2012). Moreover, PPA has been associated with a variety of suboptimal outcomes in the infant (Lonstein 2007; Glasheen et al. 2010). Research into PPA currently utilises a range of general self-report measures which are seldom validated for use postpartum. This may lead to erroneous data, inaccurate interpretation, and incomparable results across studies (Meades and Ayers 2011).

These oversights have been addressed in the pregnancy anxiety literature (Van den Bergh 1990; Levin 1991; Wadwha et al. 1993; Huizink et al. 2002) where a distinct presentation from general anxiety and depression (Huizink et al. 2004) has been revealed. As a result, a number of self-report measures have been developed (e.g. Pregnancy Related Anxiety Questionnaire (PRAQ; Van Den Bergh 1990), and the PRAQ-R (Huizink et al. 2004)). Studies using these scales consistently find that pregnancy-specific anxiety is a more efficacious predictor of perinatal outcomes than other general forms of stress, anxiety, and depression (Guardino and Schetter 2014). This is evident across studies of pre-term birth (Dunkel Schetter 2011), cognitive and motor performance (Davis and Sandman 2010), attention regulation (Huizink et al. 2002), temperament (Davis et al. 2004), and infant-feeding (Fairlee et al. 2009).

An equivalent measure of postpartum-specific anxiety has recently been developed and validated (Fallon et al. 2016b). The Postpartum-specific Anxiety Scale (PSAS) is a 51-item scale that taps into four domains of anxiety which are specific to the postpartum period. The PSAS proved acceptable to postpartum women and demonstrated high validity and reliability in initial psychometric work. As with all novel measures, validation is an iterative process and the predictive utility of the PSAS has not yet been examined in comparison to its pregnancy-specific counterparts.

One fundamental infant health outcome lies in the nourishment of the infant. Appropriate infant-feeding (i.e. responsive maternal feeding, exclusive breastfeeding to 6 months of age) confers significant health benefits. A recent systematic review provides evidence that women with PPA are less likely to breastfeed exclusively, and more likely to terminate breastfeeding earlier (Fallon et al. 2016a). Furthermore, mothers who report anxiety are at risk of non-responsive feeding behaviours (Hurley et al. 2008) characterised by impaired feeding interactions, insensitivity to infant cues of hunger and satiety, and lack of uptake to current feeding recommendations (Birch and Fisher 1995; Hughes et al. 2005).

The aims of this paper are twofold. Firstly, the predictive validity of the PSAS will be examined within the context of infant-feeding using a short-term prospective design. Second, it will be examined whether the PSAS may be more efficacious at predicting infant-feeding and perceptions of infant-feeding behaviours than the more commonly used general measures. It is hypothesised that after controlling for the effects of general anxiety and depression, postpartum-specific anxiety will have a significant, independent effect on infant-feeding outcomes and perceptions of infant-feeding behaviours.

Results demonstrate that higher levels of postpartum-specific anxiety are associated with lower odds of breastfeeding exclusively, and breastfeeding in any quantity in the first 6 months postpartum. These findings provide evidence for the predictive validity of the PSAS as they are comparable with a body of literature which finds that PPA is inversely associated with exclusive breastfeeding (Virden 1988; Groër 2005; Clifford 2006; Britton 2007; Zanardo et al. 2009; Adedinsewo et al. 2014) and breastfeeding in any quantity (Adedinsewo et al. 2014; Britton 2007; Brown and Arnott 2014; Courtois et al. 2014; Paul et al. 2013). Biological theories of anxiety and lactation posit that PPA may negatively influence breastfeeding through physiological stress responses and subsequent hormone imbalance (Lonstein 2007). In particular, anxiety is associated with lower oxytocin and prolactin which may inhibit the milk ejection reflex and subsequent breast milk production (Chen et al. 1998; Dewey 2001; Stuebe et al. 2012). It is theorised that postpartum-specific anxieties negatively affect breastfeeding practices via the same physiological mechanisms.

The current study also observed that postpartum-specific anxiety was significantly associated with negative perceptions of infant-feeding behaviours including a lower enjoyment of food, and greater food responsiveness and satiety responsiveness in the infant. The latter two findings may initially appear counterintuitive; food responsiveness is characterised by excessive hungriness, unnecessary and frequent demands for milk, and an inability to recognise satiety cues. Conversely, satiety responsiveness is characterised by under-consumption of milk during feeds and an oversensitivity to cues of satiety (Llewellyn et al. 2011). However, all three findings map onto previous work which finds high levels of anxiety negatively affect maternal perceptions of infant-feeding behaviour and impede maternal ability to interpret infant cues of hunger and satiety (Farrow and Blissett 2005; Hellin and Waller 1992; Hurley et al. 2008; Richter and Reck 2013; Savage et al. 2008). This study adds to this by demonstrating that postpartum-specific anxiety can negatively affect maternal perceptions across a number of feeding domains simultaneously and regardless of polarity. Distorted perceptions are a fundamental characteristic of unresponsive feeding practices which are linked to a range of adverse health outcomes (Cullen et al. 2000; Farrow and Blissett 2008; Hurley et al. 2011). Despite these findings, postpartum-specific anxiety was not associated with slowness in eating or general appetite. Given that this was the case for all of the measures of mood in these models it may be that these domains of feeding behaviour are too diffuse or that they do not elicit the same emotional response from mothers.

As hypothesised, postpartum-specific anxiety was a stronger predictor of infant-feeding outcomes and behaviours than general anxiety (state and trait) and depression. The PSAS was the only significant measure of mood across all of the feeding outcomes and behaviours apart from slowness in eating and general appetite analyses (in which none of the measures of mood were significant). Furthermore, the PSAS was a significant predictor after controlling for general anxiety and depression indicating that postpartum-specific anxiety elicits a unique effect upon infant-feeding outcomes and behaviours. This is a novel finding in the postpartum anxiety literature and resonates with a body of work which finds pregnancy-specific anxiety is a more potent predictor across a range of infant health and behaviour outcomes (Davis and Sandman 2010; Dunkel Schetter 2011; Guardino and Schetter 2014; Huizink et al. 2002, 2003). Theories of pregnancy-specific anxiety posit that it is a distinct construct which is rooted in the emotional and physical context of a specific pregnancy (Huizink et al. 2004). It has been suggested that pregnancy-specific anxiety may differ in its predictive power because it is more proximally linked to physiological stress responses than general measures of mood (Guardino and Schetter 2014). Furthermore, studies attempting to distinguish between general and pregnancy-specific anxiety typically report moderate correlations, suggesting that there is overlap but there is also an inimitable construct (Rini et al. 1999; Green et al. 2003; Huizink et al. 2004). These findings extend the applicability of this theory to the postpartum period. It is proposed that postpartum-specific anxiety is a distinct construct which is embedded in the emotional and physical context of the months following childbirth with a new infant. The study provides new evidence for childbearing specific measures of mood in the postnatal period and calls for an increased uptake in the use of these measures when attempting to predict childbearing related outcomes. Future research should aim to replicate these findings across other indices of maternal and infant health and behaviour in the postpartum period, particularly those with previously inconsistent results using general measures.

One strength of this study is its simultaneous consideration of infant-feeding outcomes and behaviours which provides a more comprehensive overview of the relationship between PPA and infant-feeding than other work (Fallon et al. 2016a). Furthermore, the analysis distinguished between exclusive and any breastfeeding which is consistent with current breastfeeding recommendations (McAndrew et al. 2012) and more detailed than previous research (Dusdieker et al. 1985; Hellin and Waller 1992; Mezzacappa and Katkin 2002; Cooke et al. 2007; O’Brien et al. 2008; Courtois et al. 2014). However, some limitations must also be acknowledged. Although the study controlled for a range of established confounders, the short-term prospective design precludes causality. Future research should aim to replicate the findings prospectively over a longer follow-up period. An online convenience sample was used which was adequately powered for the analyses conducted but lacked sampling control. As such, the samples were predominately married, primiparous, housewives which limit the generalizability of findings to other populations. Finally, although the PSAS was a highly significant predictor in the models discussed, the variance explained in the outcome variables was low which indicates that there is a reliable, albeit small, relationship between variables. Infant-feeding practices are complex and multifaceted, with many cultural (Scott et al. 2015), social (Hauff 2014), physical (Arbour and Kessler 2013), and emotional (O’Brien et al. 2008) factors affecting behaviours and outcomes. Given the current lack of uptake to infant-feeding recommendations (McAndrew et al. 2012), identification of any factor that consistently impacts upon feeding practices is important.

The domains of anxiety encompassed in the PSAS may all be potentially modifiable through support, education, and treatment. Replication of these findings in relation to infant-feeding and other fundamental maternal and infant health outcomes will provide an evidence base to inform interventions aimed at reducing postpartum-specific anxiety. Interventions designed to alter feeding perceptions in anxious postpartum populations may also increase the likelihood of positive feeding interactions, reduce the onset of feeding difficulties, and alleviate the emotional consequences brought about by them.