Breastfeeding difficulties in the first 6 weeks postpartum among mothers with chronic conditions: a latent class analysis

The advantages of breastfeeding over formula feeding for maternal and child health are well established, including lowered risk of infection, obesity, and asthma in children and reduced risk of cardiovascular morbidity and breast and ovarian cancers for mothers [1‐5]. Breastfeeding is recommended as the primary source of infant nutrition until 6 months when complementary foods are introduced, and sustained for longer–up to 2 years and beyond–according to maternal preference [6]. Yet a substantial proportion of women discontinue breastfeeding earlier than is recommended or planned [7, 8]. Difficulties with the mechanics of breastfeeding and physiology of lactation are the most commonly cited reasons for early cessation [8‐10]. Women who have breastfeeding difficulties often report intense feelings of inadequacy, failure, and powerlessness and face higher risk for postpartum depression [11‐14], particularly in light of inadequate lactation support [15]. Public health emphasis on breastfeeding promotion should therefore be matched with high-quality care to prevent and address breastfeeding difficulties [16], taking into account that difficulties frequently cluster together and exacerbate one another [17].

Increasingly, evidence has shown that mothers with pre-existing physical health conditions are at high risk of early breastfeeding cessation compared to the general maternal population [7, 18‐20]. Evidence on the clusters of breastfeeding difficulties in women with chronic conditions that may underpin this disparity is scant [21]. Moreover, our understanding of whether the type of chronic condition, each with its own set of pathological and clinical features and management, impacts lactation through breastfeeding difficulties is limited. For example, Berg et al. reported that mothers with type 1 diabetes were more likely to report low milk supply than mothers without diabetes at 2 months postpartum. Hormonal aberrations from diabetes are thought to reduce milk production; [22] this may extend to other endocrine conditions such as thyroid disorders [23, 24], though few epidemiologic studies have investigated this. Qualitative studies on mothers with musculoskeletal conditions have detailed distinct challenges with breastfeeding positioning and latch due to pain and mobility limitations [25‐29]. We therefore sought to explore clusters of breastfeeding difficulties experienced by mothers with a wide range of chronic conditions up to 6 weeks postpartum, and to examine potential associations between chronic condition types and breastfeeding difficulty clusters.

Of the 405 participants enrolled, 371 completed the 32-week pregnancy questionnaire (91.6% response rate) and 360 completed the 6-week postpartum questionnaire (88.9% response rate). From the 360 women who responded to both questionnaires, we excluded 11 women who did not initiate breastfeeding and 1 woman who did not provide complete data on breastfeeding difficulties, resulting in a sample size of 348 for this analysis (Fig. 1).

Table 1 describes characteristics of the MaCI Study sample used for this analysis. Participants were predominantly White, nulliparous or primiparous, reported a household income above the LICO for their area of residence, and held a post-secondary degree. Mean maternal age was 31.7 years and mean pre-pregnancy BMI was 27.4 kg/m², which falls in the overweight range. Sample demographic characteristics were comparable to that of the baseline MaCI cohort and Alberta maternal population in recent years; however, the MaCI sample slightly under-represented mothers who were younger than 24 years of age, self-identified as BIPOC, did not hold a post-secondary degree, or were multiparous (Supplementary Table 1).

Within the sample, prenatal mental health symptoms were slightly elevated; on average, women reported a score of 9.3 on the EPDS (a cut-off of 10 indicates elevated depressive symptoms) and 38.6 on the STAI (a cut-off of 40 indicates elevated anxiety symptoms). Nearly two thirds of women delivered vaginally and one third experienced obstetrical complications. Endocrine conditions were the most prevalent condition type, reported by 36.2% of women, followed by musculoskeletal (24.4%), gastrointestinal (21.8%), and neurological (13.2%) and respiratory (13.8%) conditions. Maternal fatigue (68.4%), leaking (44.0%) and engorged (35.9%) breasts, sore nipples (43.1%), low milk supply (32.5%), and latch problems (30.7%) were the most commonly reported breastfeeding difficulties rated as moderate to severe in the sample (Supplementary Table 2).

Model fit indices indicated that the 3- or 4-cluster model fit the data best, as evidenced by the lower values for the AIC, BIC, and SABIC and larger values for the log likelihood (Supplementary Table 3). Additional consideration of model parsimony and clinical utility of the clusters led us to select the 3-cluster model as the final model.

Figure 2 depicts the prevalence of each moderate-to-severe breastfeeding difficulty (compared to none-to-mild) for each of the three clusters. Cluster 1 was labelled “physiologically expected,” characterized by modest prevalence of leaking breasts (49.3%), engorgement (34.9%), and maternal fatigue (55.8%) and low prevalence of the remaining difficulties (not exceeding 20%). Cluster 2 was labelled “low milk production,” characterized by high prevalence of low milk supply (97.4%) and maternal fatigue (79.3%) and modest prevalence of concerns about infant weight gain (50.0%). Cluster 3 was labelled “ineffective latch,” characterized by high prevalence of sore nipples (75.9%), latch problems (63.2%), and maternal fatigue (82.5%) and modest prevalence of the remaining difficulties (35–55%). Most participants were assigned to the physiologically expected cluster (51.1%), followed by the ineffective latch (33.5%) and low milk production (15.4%) clusters. The distribution of breastfeeding difficulty clusters differed by method of breast milk feeding between birth and 6 weeks postpartum (Supplementary Table 4). The physiologically expected cluster was most prevalent in participants who fed only from the breast (69.3%); the low milk production cluster was most prevalent in participants who fed only expressed breast milk (38.1%); and the physiologically expected cluster was most prevalent (45.6%), followed by the ineffective latch cluster (37.7%), in participants who fed both at the breast and expressed milk.

Table 2 displays the results of multinomial logistic regression estimating the association between chronic condition types and breastfeeding difficulty clusters, using the physiologically expected cluster as the reference outcome group. Endocrine (adjusted RRR 2.34, 95% CI 1.10–5.00), cardiovascular (adjusted RRR 2.75, 95% CI 1.01–7.81), and gastrointestinal (adjusted RRR 2.51, 95% CI 1.11–5.69) conditions were associated with higher risk of belonging to the “low milk production” cluster. Gastrointestinal conditions were associated with higher risk of belonging to the “ineffective latch” cluster (adjusted RRR 2.44, 95% CI 1.25–4.77).

In this community-based cohort study of 348 postpartum mothers with chronic conditions, we identified three clusters of breastfeeding difficulties in the first 6 weeks postpartum. Half of mothers (49.3%) experienced physiologically expected difficulties of fatigue, leaking breasts, and engorgement. One third (33.5%) experienced ineffective latch, where latch problems co-occurred with several other difficulties including nipple pain. One in six (15.4%) experienced low milk production, where low milk supply co-occurred with concerns about infant weight gain.

Existing research has rarely accounted for the interrelatedness of breastfeeding difficulties, often studying them individually or as a composite [32, 44]. Exploratory factor analysis has been used to group similar types of breastfeeding difficulties [9, 14], but in the context of women’s underlying reasons for breastfeeding cessation. This approach overlooks women who are still breastfeeding despite potential difficulties or whose difficulties have resolved. Building on existing work, we used LCA to identify that most mothers with chronic conditions experienced co-occurring fatigue and breast overfullness in the first 6 weeks postpartum, which corresponds to the physiologically expected trajectory of breastfeeding establishment. Milk supply is upregulated to meet rapid increases in infant intake requirements, which mothers may experience as breast engorgement and leaking, and consolidated sleep is interrupted by frequent feeds and newborn care [45].

Sore nipples are experienced by up to 80% of breastfeeding mothers [44, 46]; while etiology is multifactorial, the most frequent cause is improper infant latch at the breast [47]. We identified an ineffective latch cluster which captured this causal link as well as co-occurring challenges with infant sleepiness, fussiness, and weight gain concerns, difficulty positioning the infant at the breast, and breast engorgement. Improper latch can lead to suboptimal draining of milk from the breasts during feeds, which initially presents as engorgement but will downregulate milk supply when sustained over time [48]. Moreover, limited transfer of milk to the infant can result in increased hunger cues and slower weight gain [48]. Our analysis of chronic condition types revealed that gastrointestinal conditions were associated with the ineffective latch cluster. Physical sensations of infant oral grasping and suckling on nipple tissue may be more painful for these women given that hypersensitivity to pain is a feature of several gastrointestinal conditions [49, 50]. Pain and discomfort in the abdominal area may also interfere with comfortable positioning of the infant to avoid cross-body breastfeeding positions.

Perceived low milk supply in the absence of self-reported latch issues suggests that factors intrinsic to maternal physiology may be interfering with lactation [51]. This was captured in the low milk production cluster, wherein low milk supply was nearly universal and occurred in relative isolation. In our study, endocrine, cardiovascular, and gastrointestinal conditions were associated with the low milk production cluster. Endocrine conditions are an established risk factor for impaired lactation [51], through mechanisms related to insulin resistance, breast hypoplasia, and reduced prolactin responsiveness [52]. Previous studies have found that low milk supply was associated with diabetes in pregnancy [53], and that fewer women with polycystic ovary syndrome (PCOS) breastfed to 6 months relative to women without PCOS (44.3% vs. 54.2%, respectively) [54]. Our findings related to cardiovascular and gastrointestinal conditions are novel and merit further investigation. Lactation interfaces with several body systems, stimulating changes in hormone activity, gastrointestinal blood flow, metabolic rate, and cardiac output [55, 56]. Yet the influence of underlying maternal disease pathophysiology on lactation physiology has received minimal attention to date. Alternatively, given that we could not verify self-reported low milk supply with objective measurements of milk volume, our findings may reflect a hypervigilance among mothers with these conditions towards milk adequacy as opposed to differences in actual milk production.

Strengths of this study include the prospective design, community-based sampling of diverse chronic conditions, and graded measurement of breastfeeding difficulties in all women who initiated breastfeeding regardless of duration. However, some limitations should be considered. Validity of self-reported chronic conditions is imperfect relative to clinical exams or medical records; however, a quarter of the sample were recruited directly from obstetric clinics specializing in chronic medical disorders or using mailed letters sent to women with diagnostic codes for chronic conditions and pregnancy. Of the remaining participants recruited through social media or other methods, over 80% reported receiving prenatal care from an obstetrician and/or specialist physician. It is therefore likely that most women recruited into the sample have a true clinical diagnosis. Chronic condition types were based on the ICD system, but often involved grouping heterogeneous conditions. For example, endocrine conditions included type 1 diabetes, an autoimmune disease, as well as PCOS, a metabolic gynecologic disorder. Additional research on individual conditions is needed to verify the associations we observed and investigate distinct underlying mechanisms. Our use of complete case analysis for handling missing covariate data (< 3% of observations) may have slightly reduced the precision of our estimates [57]. Finally, compared to the maternal population in Alberta, the MaCI sample slightly under-represented mothers who were younger than 24 years of age, who self-identified as BIPOC race/ethnicity, or who did not hold a post-secondary degree. Caution is needed when generalizing findings from the MaCI Study to these underrepresented groups.

Our findings have important clinical implications. Women should be counselled prenatally about the potential breastfeeding difficulties they may experience related to their chronic condition. Given that our data suggests half of mothers with chronic conditions who choose to breastfeed will experience low milk production or ineffective latch, health care providers should closely monitor breastfeeding experiences for these clusters of breastfeeding difficulties and promptly offer evidence-based interventions when they arise [17, 45]. For example, mothers reporting low milk supply in the absence of latch issues should be evaluated for potential underlying physiologic or psychosocial contributors and counselled on appropriate feeding frequency, infant weight gain, and wet diapers, as well as supported to feed or express more frequently or initiate galactagogues as indicated to increase supply [17, 45]. More broadly, future research that employs latent class analysis of breastfeeding difficulties would be valuable to ascertain whether the clusters we identified in mothers with chronic conditions are similarly observed in the general maternal population and to compare the distributions of each.

In summary, we identified three clusters of breastfeeding difficulties in mothers with chronic conditions, corresponding to physiologically expected lactation changes, ineffective latch, and low milk production. Approximately half of mothers with chronic conditions belonged to either the ineffective latch or low milk production cluster. Relative to the physiologically expected cluster, endocrine, cardiovascular, and gastrointestinal conditions were associated with the low milk production cluster and gastrointestinal conditions were additionally associated with the ineffective latch cluster. These findings can aid with differentiating and treating breastfeeding difficulties in clinical practice and suggest that the influence of disease pathophysiology should be considered when providing early postpartum breastfeeding support.