In 2022 the Academy of Breastfeeding Medicine (ABM) published Clinical Protocol #36: The Mastitis Spectrum, which aims to update clinical approaches to management of benign lactation-related breast inflammation [1]. A great deal of work is invested in the development of a Clinical Protocol. This Commentary acknowledges and respects the authors’ commitment to offering the best possible clinical care for breastfeeding pairs. However, this Commentary shares Baeza et al’s grave concerns about Clinical Protocol #36’s scientific integrity [2].

Clinical Protocol #36 introduces some concepts and recommendations that align with the analyses and the clinical guidelines for lactation-related breast inflammation published as part of the breastfeeding domain of Neuroprotective Developmental Care earlier that same year [3,4,5,6]. Specifically, elements that align include that breast inflammation:

Is a spectrum condition.

Is not helped by and is likely to be worsened by deep lump massage.

May elicit a systemic response that is not necessarily infective.

Mostly resolves with conservative care.

Is not the same as normal lactational glandular tissue which can feel ‘lumpy’ and tender.

Does not develop into infection in a period of hours.

Is not caused by retrograde spread of bacteria from a damaged nipple.

Is not caused by mammary candidiasis.

Is not helped by.

Does not require investigations of c-reactive protein or white blood cell count, since these are markers of inflammation not specific for infection.

Despite the above commonalities, I agree with Baeza et al’s concern that ABM Clinical Protocol #36 contains significant scientific flaws [2]. These flaws give rise to clinical recommendations which may be either of no benefit or which may worsen outcomes for breastfeeding pairs and their families.

Clinical protocol #36 does not conform with principles of scientific best practice

Clinical lactation support remains a research frontier [7]. Breastfeeding families and the health professionals who support them deserve clinical guidelines developed from rigorous application of implementation science [8, 9], even in the context of relative paucity of research. Best practice implementation science in health care requires:

Systematic or metanarrative review of existing and interdisciplinary research, interpreted through the lens of clinical experience, from which theoretical frames are developed;

Translation of theoretical frames and existing evidence into education programs or clinical guidelines;

Collation of iterative feedback from patients in the clinic and in pilot studies;

Improvement of education programs or clinical guidelines in response to feedback;

Layers of evaluative studies, qualitative and quantitative.

Clinical Protocol #36 fails to comply with best practice implementation science in the following ways:

1.

Inaccurate representation of existing studies.

2.

Theoretical models misrepresented as facts.

3.

Sources not reliably attributed.

4.

Strength of Recommendation Taxonomy (SORT) not reliably applied.

5.

Reliance upon single case reports.

Inaccurate representation of existing studies

To build theoretical frameworks translatable into effective recommendations, implementation science requires accurate representation and critical analysis of existing studies. Examples of studies inaccurately represented in Clinical Protocol #36 are detailed in Table 1.

Table 1 Analysis of accuracy of Clinical Protocol #36’s representation of research Theoretical models misrepresented as fact

Given that clinical lactation support is a research frontier with a relative paucity of evaluative studies to guide clinicians, clinical advice often relies upon theoretical models. Theoretical models or hypotheses need to be explicit: named, described, and debated. Naming a theoretical model and clarifying its proposed pathophysiological mechanisms is important for scientific integrity and transparency. Misrepresenting theoretical models as fact in clinical guidelines misleads clinicians.

Example A. Clinical Protocol #36 asserts that dysbiosis is one of two fundamental causes of lactation-related breast inflammation

Clinical Protocol #36 states as fact that “under physiological conditions, coagulase-negative Staphylococci and viridans Streptococci (i.e. S mitis and S salivarius) form thin biofilms that line the epithelium of the mammary ducts, allowing a normal milk flow” [1, p. 365]. The physiological, cellular, or biochemical reasons why coagulase-negative Staphylococcus and Streptococcus bacteria – and why these genera and not other micro-organisms – might form a physiologic ductal biofilm (instead of remaining planktonic) in healthy lactating women are not discussed.

Clinical Protocol #36 continues: “In the setting of dysbiosis these species proliferate and function under opportunistic circumstances whereby they are able to form thick biofilms inside the ducts, inflaming the mammary epithelium” [1, p. 365]. The pathophysiological mechanisms by which ‘dysbiosis’ develops and turns physiological ductal biofilm into pathological biofilm are also not discussed. This pathogenic microbiota hypothesis of lactation-related breast inflammation is further presented as fact in Fig. 1 and schematically illustrated in Fig. 2 of Clinical Protocol #36. The latter illustration is adapted from Figure 1 in an article by Fernandez et al, published in 2014 prior to recent advances in human microbiome and human milk microbiome science [34, 35].

However, the latest human milk microbiome research renders the pathogenic microbiota hypothesis of lactation-related breast inflammation outdated [4]. ‘Eubiosis’ needs to be defined before ‘dysbiosis’ can be described. Taxonomic categorisation of ‘eubiosis’ in human microbiomes is increasingly considered unachievable and less relevant; the research focus has shifted to microbiome functionality, which emerges from complex interactions between various elements [36,37,38,39,40,41,42,43]. The human milk microbiome, leucocytes, epithelial cells, oligosaccharides, exosomes and metabolome are each complex systems interacting within the complex adaptive system of the mammary immune system [4]. Composition of the human milk microbiome is extremely variable both within the one woman over time, and between lactating women.

Kvist et al. found no correlation between scores for erythema, breast tension, pain or total severity of symptoms and the type of bacteria in breast milk during episodes of breast inflammation [44]. High counts of Staphylococcus aureus and decreased microbial diversity associated with breast inflammation are most plausibly explained as secondary to wound-healing inflammatory responses of the mammary immune system, rather than as causes of breast inflammation [4]. That is, the dramatic influx of leucocytes (with their powerful bactericidal properties) into alveoli and lactiferous duct lumens as a result of an inflammatory cascade is likely to decrease counts of more susceptible bacteria and increase counts of more adaptive bacteria like S. aureus, without typically tipping the patient into an infective process requiring antibiotic treatment [4].

Clinical Protocol #36 implicitly acknowledges the implausibility of the pathogenic biofilm hypothesis by classifying postpartum engorgement as “a distinct clinical entity related to interstitial edema and hyperemia” [1, p. 362]. This separate classification seems to acknowledge that engorgement is unlikely to result from generalised narrowing of lactiferous ducts by whole-of-breast pathological biofilm. Yet Clinical Protocol #36 doesn’t offer pathophysiological mechanisms to explain the interstitial fluid and hyperaemia of engorgement.

Applying the Neuroprotective Developmental Care classification system which arises from the mechanobiological model of lactation-related breast inflammation, engorgement belongs on the spectrum of breast inflammation, subject to the same aetiological model as other clinical presentations of benign lactation-related breast inflammation [4, 5].

Example B. Clinical Protocol #36 asserts that ‘hyperlactation’ is the second fundamental cause of breast inflammation

Clinical Protocol #36 states as fact that “ductal lumens can be narrowed by edema and hyperemia associated with hyperlactation”, not only by pathological biofilm formation [1, p. 361]. This implicitly acknowledges the mechanical effects of raised stromal pressure on lactiferous ducts, which are then compressed, as detailed earlier in the mechanobiological model of breast inflammation [4]. The causative role of ‘hyperlactation’ is further presented as fact in Fig. 1 of Clinical Protocol #36 [1].

Clinical Protocol #36 states Clinical Protocol #32 on ‘hyperlactation’ may be considered as an adjunct. Yet there is no workable definition of ‘hyperlactation’, since the term requires comparison with a state of ‘normal’ milk production. Normal breast milk volumes are highly variable between individuals, ranging from 478 to 1356 mls over a 24 h period in women who are exclusively and successfully breastfeeding [45]. Also, a woman successfully exclusively breastfeeding twins, generating milk volumes of two litres over a 24 h period, is not in a state of ‘hyperlactation’. The term ‘production mismatch’ more accurately identifies the contextual nature of milk production.

Clinical Protocol #36 doesn’t offer explicit pathophysiological mechanisms by which ‘hyperlactation’ causes stromal oedema and hyperaemia. Subsequently, proponents have argued that ‘hyperlactation’ causes lactose to leak through intra-lactocyte tight junctions to inhibit milk secretion and to penetrate the stromal space, increasing interstitial fluid volume [46, 47]. Problems such mechanism of penetration of the basement membrane are not addressed. The studies cited to support this ‘hyperlactation’/lactose hypothesis of breast inflammation more plausibly corroborate the mechanobiological model of breast inflammation [4].

Infant lactose overload or maternal recurrent breast inflammation as a result of production mismatch, with production exceeding the infant’s caloric needs, may occur more often in sociocultural contexts where mechanical milk removal appears to occur more commonly, for example, in the United States [48, 49].

Sources not reliably attributed

Clinical Protocol #36 does not comply with scientific standards for image description and source attribution, which may mislead clinicians. For example, scientific guidelines require authors to specify preparation, type of equipment used, and resolutions of an image at acquisition and downstream after processing [50].

Details such as source, anatomic site specifics, preparation, type of organism, and magnification for the three images in Fig. 4 entitled “Human milk microbiota”, “Healthy mammary gland” and “Mastitis” are missing. Two images from Fig. 4 are repeated in Fig. 17, again with explanatory details missing, other than the labels “Mammary duct – NO mastitis” and “Mammary duct - Mastitis S. epidermidis biofilm” [1]. In personal communication, Professor Juan Rodriguez (22 September 2021) explains that the mastitis image is of tissue biopsied from an area of mastitis in a lactating human breast. The extracted tissue had fixative applied before being photographed under electron microscopy at 5000x magnification.

Finding Staphylococcus epidermis biofilm in lactiferous ducts after biopsy and fixative does not corroborate the hypothesis that lactiferous ducts are lined with physiological biofilm in vivo, nor that pathological Staphylococcus epidermis biofilm in lactiferous ducts causes breast inflammation. Moreover, a 2022 Australian nested case-controlled study examined the breast milk of 20 women with mastitis and 16 women without mastitis, and did not find any clear association between Staphylococcus epidermidis and mastitis [51].

Strength of recommendation taxonomy (SORT) not reliably applied

Clinical Protocol #36 provides an analysis of existing research using the 2004 Strength of Recommendation Taxonomy (SORT) [52]. SORT recommendations are intended to be based “on a body of evidence (typically more than one study)” [48, p. 549], compiled from comprehensive review of all existing evaluations of that specific intervention [52].

It is not clear that comprehensive review of all existing evaluations has been undertaken in the preparation of Clinical Protocol #36, since relevant studies are omitted (Table 1). Also, studies which are highly heterogenous and which apply a wide range of measures have been grouped together to create recommendations. Clinical Protocol #36 acknowledges that many of its recommendations derive from Level C evidence which includes consensus, usual practice and/or opinion. However, high levels of subjectivity in Clinical Protocol #36’s application of SORT mean that its rating of evidence is unhelpful for clinicians.

In one example, Clinical Protocol #36 recommends use of probiotics for lactation-related breast inflammation on the basis of SORT Level of Evidence 1–2 and Strength of recommendation B. To reach this SORT recommendation, the authors group together and analyse a scoping review by Barker et al. 2020, a Cochrane review by Crepinsek et al. 2020, a narrative analysis by Amir et al. 2016, and a narrative review of bovine and human milk microbiomes by Oikonomou et al. 2020 [29, 39, 53, 54]. Barker et al. does not support Clinical Protocol #36’s SORT recommendation for reasons detailed in Table 1 [49]. Crepinsek et al. investigated the use of probiotics as prevention of mastitis after childbirth (finding very low certainty of evidence) not as intervention for breast inflammation. Amir et al. note that probiotics are vigorously marketed for treatment of mastitis despite lack of reliable evidence demonstrating efficacy. Oikonomou et al. briefly analyse both bovine and human evidence under a subtitle “Can we manipulate the milk microbiota in order to improve mammary gland or offspring health?”, citing two human studies and drawing no conclusions [39, p. 10].

In a second example, Clinical Protocol #36 recommends avoidance of nipple shields on the basis of SORT Level of evidence 3. Strength of recommendation C. To reach this conclusion, ABM Clinical Protocol #36 considers just one 2010 study [19], and doesn’t include more recent studies and a systematic review on this topic, further discussed in Table 1.

Reliance upon single case reports

Case reports, such as in Figs. 10 and 11, are unable to address multiple potential confounding factors. For this reason, they may mislead clinicians and are best avoided in clinical guidelines [1].

Clinical Protocol #36 makes recommendations which may risk worsened outcomes for breastfeeding pairs and their families

In summary, Clinical Protocol #36 recommends:

Interventions which have been investigated but lack convincing evidence of efficacy;

Interventions which have been subject to very few evaluations and which also lack pathophysiological rationale; and

Avoidance of an intervention which has demonstrated positive effects in management of breastfeeding problems.

As a result, some or many of Clinical Protocol #3’s recommendations may have no benefit or may even worsen outcomes for breastfeeding pairs and their families. Baeza et al. draw a similar conclusion [2].

Clinical protocol #36 recommends unnecessary interventions which increase risk of unintended outcomes

Recommendations by Clinical Protocol #36 which lack an evidence-base or credible theoretical frame, such as ‘lymphatic drainage’, therapeutic ultrasound, and lecithin, are discussed in Table 1. These recommendations may appear benign. However, unnecessary interventions increase risk of unintended outcomes, exacerbate patient anxiety and disempowerment, increase financial burden, and may be accessible only by affluent patients within advanced economies [55,56,57,58,59].

Clinical Protocol #36 recommends to not increase or to reduce milk removal when the breast becomes inflamed, which may risk worsened outcomes for breastfeeding pairs

Clinical Protocol #36 states that “overfeeding from the affected breast …. is a major risk factor for worsening tissue edema and inflammation” [1, p. 367]. By foregrounding the theoretical model of ‘hyperlactation’ as a key aetiological factor for breast inflammation without clear criteria for diagnosing ‘hyperlactation’, Clinical Protocol #36 derives the recommendation that milk removal should not be increased or be reduced when breast inflammation presents. This recommendation occurs in six places.

1.

Under the heading Key Information: Pathophysiology of Mastitis Spectrum Conditions: “Reducing milk removal may transiently increase pain and erythema …; however, it ultimately prevents future episodes” [1, p. 361].

2.

Under the subheading Ductal narrowing (e.g., “plugging”): “Patients may feel relief of a “plug” with breastfeeding because this decreases alveolar distension. However, repeated feeding in an attempt to relieve the “plug” will suppress FIL, increase milk production, and ultimately exacerbate inflammation and ductal narrowing” [1, p. 363].

3.

Figure 10: “Patient with early inflammatory mastitis … was treated with ice, ibuprofen, acetaminophen, and feeding first off the left, less congested breast first to avoid overstimulation of the affected right breast” [1, p. 365].

4.

Figure 11: “Bacterial mastitis that progressed from early inflammation in the inner quadrant to all quadrants being affected. This patient also pumped and continually fed the infant …. in an attempt to prevent ‘milk stasis’. This approach resulted in worsened ductal inflammation and bacterial overgrowth as well as milk obstruction” [1, p. 366].

5.

Under the heading “Spectrum-wide recommendations c. Feed the infant on demand, and do not aim to “empty” the breasts”: “In some instances, in which the retroareolar region is so edematous and inflamed that no milk is expressible by infant breastfeeding or hand expression, the mother should not continue to attempt feeding from the affected breast during the acute phase” [1, p. 367].

6.

Figure 19: “Ice and decreased removal of breast milk reduce ductal narrowing and breast swelling” [1, p. 370].

The belief that an infant can overfeed from or overstimulate the breast contradicts the evolutionary biology model which underpins the Neuroprotective Developmental Care concept of frequent and flexible breastfeeds, necessary for adequate milk production and infant weight gain [5]. From the mechanobiological model’s perspective, frequent milk removal is integral to the downregulation of breast inflammation, because the ductal dilations of milk ejection reduce intra-alveolar pressures and counter stromal pressures [5]. A true case of production mismatch, with supply exceeding the infant’s caloric needs, is identified at presentation but addressed once breast inflammation has resolved. No matter how swollen and inflamed the areolae are, clinical support continues to facilitate positional stability of the infant and milk removal.

Breast inflammation is already associated with subsequent low milk production [60]. Clinical Protocol #36’s key recommendation of not increasing or reducing milk removal has two effects predicted by the mechanobiological model, both of which risk worsened inflammatory cascades:

1.

Perpetuation of excessive intra-luminal pressures;

2.

Limited episodes of ductal dilations.

Worsened inflammatory cascades are predicted to worsen the clinical presentation of breast inflammation, and to result in a greater decrease in milk production post-resolution.

Clinical Protocol #36 introduces terms or diagnoses which are poorly defined and potentially confusing or misleading

New diagnoses for clinical presentations in breastfeeding pairs should be introduced with great caution. Overdiagnosis and overtreatment are escalating international trends, including in breastfeeding women and their babies, driving unnecessary costs for families and health systems and risking unintended outcomes [55,56,57,58,59]. For further discussion of unnecessary or poorly defined terms or diagnoses in Clinical Protocol #36, see Table 2.

Table 2 Unnecessary and poorly defined diagnoses in Clinical Protocol #36 may increase the risk of antibiotic overtreatment Example A. Erroneous use of the diagnosis ‘lymphedema’

Clinical Protocol #36 confuses the temporary increase in breast stromal interstitial fluid associated with inflammation with the medical condition of lymphoedema. The diagnosis of lymphoedema is only relevant to the lactating breast in the exceptional case of a genuine primary or secondary lymphoedema co-morbidity.

Secondary or acquired lymphoedema is a chronic and progressive disease. It occurs subsequent to destruction of normal lymphatic vasculature by systemic disease, trauma, or surgery. Secondary lymphoedema often results in fibrosis [64]. Although the most common cause of secondary or acquired lymphoedema world-wide is filariasis, in advanced economies the most common cause is surgical excision or irradiation of lymph nodes due to breast cancer treatment, predominantly affecting the upper limbs and occasionally the breast. The phenotypes of primary lymphoedema are rare, mostly genetic, and also often progressively fibrotic [65].

Clinical Protocol #36 states in Fig. 5: “Day 5 postpartum breast engorgement showing edematous nipple areolar complex and dependent lymphedema with overlying erythema” [1, p. 363]. This patient may have some increased interstitial fluid and stromal tension, but her nipple areolar complex swelling is most likely from high intraluminal pressures in both her alveoli and lactiferous ducts [4]. Lactiferous ducts become dilated and tense when the volume of milk produced exceeds milk removed. Most glandular tissue is subareolar, in a 3 cm radius from the base of nipple [33].

Similarly, in Fig. 21 entitled “Technique of lymphatic drainage”, Clinical Protocol #36 states that