Infant feeding guidelines
Indicators must be broadly consistent with standards, recommendations and best-practice. Although most allergists and immunologists in Australia recommend the introduction of solid foods between four and six months and not beyond six months [
6,
7], the recommended duration of exclusive breastfeeding is six months [
8,
9]. The current National Health and Medical Research Council (NHMRC) Infant Feeding Guidelines in Australia related to breastfeeding are 'exclusive breastfeeding for the first six months' and to 'continue breastfeeding for at least 12 months' [
9]. Also, current NHMRC guidelines indicate that breastfeeding should be complemented with appropriate, hygienically prepared food from
around six months of age. This 'looseness' in the wording for the timing of introduction of solid foods caters for some infants being developmentally ready for solids prior to six months. However this creates some problems as it is juxtaposed with the recommendation for exclusive breastfeeding to six months.
Measurement of breastfeeding
In Australia [
1] the current recommendation for determining rates of exclusive breastfeeding, hence the introduction of substances other than breast milk, is by current status methods, often 24-hour recall, as recommended by the World Health Organization (WHO) [
3]. The advantages of the 24-hour recall method are that it requires a small number of questions in order to report against many indicators important for policy and practice, the analysis and interpretation of the data is straightforward, and recall error is significantly reduced. Consequently, the method can be used to gain reliable estimates of trends in breastfeeding practices. In Australia it has been used in state-level surveys in New South Wales (NSW) and Queensland (QLD) - together with longer-term recall questions - and it has been used for data collection at the area health service level (South Eastern Sydney Illawarra Health), and state-wide in the Personal Health Record in NSW.
The other main method to measure and report on breastfeeding practices involves questions relating to prior practices involving longer-term (longer than seven days) recall, often since birth, in cross-sectional population surveys or generally over several months in cohort studies.
Issues with longer-term recall
The longer-term recall method commonly employs survival analysis techniques to determine duration and intensity of breastfeeding across the population. There are a number of limitations or issues associated with this method. First, there are problems of maternal recall. Second, there are issues with determining boundary points or cut-offs and use of prepositions. Third, questions can relate to the first or regular giving of something other than breast milk to determine cessation of exclusive breastfeeding, an issue also observed in the suggested 'second pass' questions associated with 24-hour recall indicated above. Fourth, there is no standard method of converting from weeks to months.
Maternal recall
The issue of maternal recall has been discussed extensively [
1,
10‐
16,
19]. This problem can be overcome by use of a cohort, prospective study design to reduce the period of recall, although this is often deemed not practical, affordable, and/or necessary at the population level [
1,
14].
Boundary points, cut-offs and prepositions
The issue of boundary points has been less discussed. While the current status method clearly enables the reporting of breastfeeding practices at a particular age, in surveys containing questions related to longer-term recalled practice, e.g. since birth, the cut-offs or boundary points become less clear.
Survival analysis techniques including the Life-table and Kaplan-Meier methods have been used to determine breastfeeding rates in Australia and internationally and both techniques are equally applicable in most instances. Garden et al indicate that the Life Table method was adopted in their analyses because the duration of breastfeeding and age of the child are recorded as an interval (in months) rather than an exact date [
20].
In survival analysis, prevalence rates of breastfeeding are usually sourced from survey questions asking age at which breastfeeding stopped, and an infant is assumed to have been breastfed for all of the time up until the age of cessation, although it is not clear if they are also considered to be breastfed at the month of cessation. For example, "
it is assumed that someone who stopped breastfeeding at a particular age was breastfeeding for all the time up until the age of cessation (for example a child who stopped being breastfed at 4 months is assumed to have been breastfed at ages 1, 2 and 3 months)"[
1,
20] (p. 54; p. 29). Presumably the infant is assumed to also have been breastfed at less than one month.
Exclusive breastfeeding ceases once something other than breast milk is introduced into the infant diet. The implication of this is that if a mother reports that she introduced solids to her infant when he/she was six months old, and had given the infant nothing other than breast milk until that time, then the infant was assumed to have been exclusively breastfed AT each month of age TO 6 months, i.e. AT < 1 month, AT 1 month, AT 2 months, AT 3 months, AT 4 months, and AT 5 months, but not AT 6 months. Similarly, if solid foods were introduced when the infant was five months old, then the infant was exclusively breastfed FOR 5 months, TO 5 months and AT each month of age UP TO 5 months but not AT 5 months.
Uncertainty remains however, where the boundary points for various breastfeeding practices are being set within the survival analyses, and/or if this tallies with the actual breastfeeding practice. How the raw data are 'used' within the survival analysis, and how the survival curves are interpreted - specifically how percentages are 'read off' from the curves, particularly from the different survival analysis methods - is often not evident. For example, are mothers who stop breastfeeding during the infant's first month included in the 'at less than one month' percentage, are they 'ignored', or are they included in the 'at one month' percentage?
The issue of 'boundary points' was raised in the UK Infant Feeding Survey 2005: "
when deriving certain parameters such as time of cessation of exclusive breastfeeding, then the boundary point needs to be defined. For example, if a mother responded that they first introduced formula at six weeks, should they be counted as being exclusive or not exclusive at six weeks? It was decided that in such a situation the baby would be counted as being exclusively breastfed UP TO six weeks, but not at the six week point itself. This same principle was applied for all different ages." [
21] (p. 41). However, despite raising this issue, the indicator was then reported as 'the proportion of all babies who are being exclusively breastfed AT specific ages' (also page 41). Hence the interpretation must be that the infant in their given example was not included in the rate of exclusive breastfeeding at six weeks. However, in accordance with the uncertain interpretation of output from survival analysis, it is not clear whether this is, in fact, the case.
The uncertainty around how data are analysed and interpreted has led to the mixed reporting of breastfeeding practices, exclusive breastfeeding in particular. Often the two prepositions, 'at' and 'to' (as well as 'for' and 'until', largely equivalent to 'to') are used interchangeably when describing rates of exclusive breastfeeding. Most commonly, the indicator 'exclusive breastfeeding AT six months' is used to describe data which may or may not accurately portray exclusive breastfeeding TO six months. 'Exclusive breastfeeding AT six months' is a redundant indicator, as solid foods are recommended to be introduced at six months of age. Not surprisingly therefore, many studies have determined exclusive breastfeeding at six months of age to be non-existent. For example, in the UK Infant Feeding Survey 2005 [
21] (p. 23) exclusive breastfeeding was reported as: "
at six months the prevalence of exclusive breastfeeding was negligible in all [UK] countries (<1%)". Other examples include "
at six months, none of the women was exclusively breastfeeding"[
22] and "
rates of exclusive breastfeeding amongst infants aged six months or younger were 39%"[
23]. Where reported exclusive breastfeeding rates at six months are high, e.g. "
exclusive breastfeeding rates at six months after birth (14.4%) were lower than desirable" [
24] (p. 122), it seems likely that these data more accurately reflect exclusive breastfeeding TO six months. Such data could also be reported as 'at five months', if the survey and analysis method permits.
'First' or 'regular' introduction
As in the suggested options for follow-on questions when using 24-hour recall methods (see above), questions for longer-term recall differ in whether they ask about the
first or
regular introduction of substances other than breast milk. All of the population-level studies in the United States have used the
first introduction of something other than breast milk [
5]. In Australia research studies have often used
first[
25], whereas state-level and national-level surveys have tended to use
regular or
first regular. In NSW and QLD
regular has been defined as 'once per day' and in national surveys
regular has been left open to respondent interpretation. No studies have compared the response differences, if any, to these two types of questions.
It is well-established that infants who receive infant formula in hospital have much poorer breastfeeding outcomes [
26‐
28]; much less is known about the clinical relevance of small amounts of formula supplementation in hospital [
29] or later in the infant period among otherwise exclusively breastfed babies. There are known effects of formula-feeding on the digestive tract [
10,
30], but the impact of these effects on health outcomes has not been specifically examined. The problems of giving water to newborns have been summarised [
31].
Infants may be given formula in hospital and then breastfed exclusively post-discharge [
32]. In the United States, rates of exclusive breastfeeding for the first few days of life (up to four days) were about 20% lower when measured through maternity database collections compared to longer-term recall of 19-35 months through the National Immunization Survey [
29]. These data suggest that many infants were given formula in hospital, but this isn't remembered or known by mothers via longer-term recall. Another possibility is that mothers might consider small amounts of early formula as not important and therefore not report it [
29]. Flaherman et al summarise the literature regarding maternal recall of infant feeding practices and conclude that there is no available literature supporting the accurate recall of early formula use beyond six months postpartum [
29]. They suggest that an appropriate survey question could be '
how much formula did your child receive (if any) during newborn hospitalization?'.
Accurate data on the 'proportion of infants exclusively breastfed since birth', in the strictest, exact sense, are important for research relating breastfeeding practices to specific health outcomes, particularly in relation to mother-to-child-transmission of Human Immunodeficiency Virus [
11,
12,
15]. Overestimation of rates of exclusive breastfeeding due to not precisely measuring it in its strictest, exact sense, since birth, leads to an underestimation of the health benefits of this breastfeeding practice. However, whether it is important to measure exclusive breastfeeding in its exact sense, since birth, in monitoring surveys at the population level is arguable.
Conversion from weeks to months
Infant feeding recommendations are indicated in months and most often breastfeeding practices are reported in months, yet many data collections allow mothers to respond in days and/or weeks, particularly during the early months, with some studies only recording breastfeeding in weeks. There is no standard way of converting between weeks and months. Often the number of weeks in a year is divided by the number of months so that 1 month is assumed equal to 4.33 weeks, 2 months assumed equal to 8.66 weeks, etc. In other instances 8 weeks is considered to be equal to 2 months and 24 weeks is considered to be equal to 6 months, for example. In a study by Scott et al responses were allowed only in weeks [
33]; they observed a peak at 16 weeks around the introduction of solid foods, which suggests that perhaps mothers interpret 16 weeks as 4 months. How mothers consider infant age when they report on infant feeding practices has not been explored.
Definitions and other wording issues
Calls for consistency in breastfeeding definitions began in the late 1980s [
34] and have occurred more recently [
10,
35]. However, I contend that the main issue now is not one of definition of breastfeeding practices, at least for exclusive breastfeeding. The WHO definition of exclusive breastfeeding is generally accepted as a reasonable description of what is allowed to be received by an infant, other than breast milk, i.e. drops or syrups consisting of vitamins, mineral supplements or medicines [
2,
3], and can be successfully applied to all survey methods and research studies. Rather, there are two issues that emerge when considering the reporting of exclusive breastfeeding: (i) the period over which exclusive breastfeeding is determined, and (ii) the text description of how exclusive breastfeeding is determined.
The period over which exclusive breastfeeding is recalled and measured varies between survey methods. In the current status method it is usually 24 hours, 48 hours or 7-days, whereas in longer-term recall or prospective studies it is usually 'since birth'; the latter regardless of whether
first or
regular introduction signifies the deviation from exclusivity. Therefore an indicator using current status or short-term recall methods might be more accurately described in a more cumbersome manner involving the time period of recall. This was recommended by Binns et al, who suggested that data collected using the 24-hour recall method be referred to as the '
24-h full breastfeeding rate at "X" months' [
10] (p. 179). However, reporting those exclusively breastfed in the previous 24 hours as a fully breastfed rate only takes account of potential misclassification due to water or watery drinks on other days - infant formula and/or solid foods can also be given intermittently. Exclusive breastfeeding, as opposed to full breastfeeding, according to the WHO definitions, is easily measureable and reportable using the 24-hour recall method, hence the indicator in this instance would most appropriately be the '
24-h exclusive breastfeeding rate at "X" months'.
Often the WHO definition is accepted as 'true', but then compromises are made in terms of the survey questions, disabling accurate reporting against the definition despite the indicator 'exclusive breastfeeding' being reported. For example, Lande et al state in the body text of their paper that: "
Exclusive breastfeeding at 0.5-5.5 mo included infants given water (use of water was not covered by the retrospective questions). Breast milk was their only energy source and therefore we preferred to use the category exclusive breastfeeding rather than predominant breastfeeding." [
36] (p. 154). The inclusion or exclusion of the giving of water to infants in the monitoring questions used to report against exclusive breastfeeding has been recently commented on in studies in the US [
5,
29]. Similarly, as indicated by Binns et al [
10], the NSW Population Health Survey reports exclusive breastfeeding as the indicator, but in the body text describes the indicator as only taking into account water and fruit juice, not other watery drinks [
20,
37]; as well as allowing the infant to receive substances other than breast milk irregularly. In comparison, Kristiansen et al, for example, clearly indicate that infants were classified as exclusively breastfed at a particular age if they had received only breast milk and had not been introduced to any additional food or drink, not even water, but could have received vitamin-mineral supplements [
38].
Perhaps what is needed is a 'prohibition' on the reporting of exclusive breastfeeding unless the data collection methods allow the accurate reporting against the WHO definition or the indicator stipulates the deviation from the WHO definition, as well as including the time period over which it is measured.
Another 'wording' issue relates to bottle feeding. In 2008 the WHO revised their optional indicators to include bottle feeding as the '
proportion of children 0-23.9 months of age who are fed with a bottle' (during the previous day) [
3]. This indicator includes the giving of expressed breast milk in a bottle and was in recognition of the potential interference of bottle feeding with optimal breastfeeding practices, as well as the association with increased incidence of diarrhoeal disease. Thulier comments on the potential differential effect of bottle feeding
versus breast feeding of breast milk on some health outcomes, notably obesity; however despite purporting to, her proposed definitions and resultant groups for infant feeding do not specifically incorporate this distinction [
35]. Indeed, Thulier's suggested alternative definitions for infant breast feeding offer considerably less consistency than the WHO definitions and only relate to a largely unmeasurable intensity of breastfeeding, particularly in relation to population monitoring, but probably even for the stated purpose of research.
Finally, although of less importance than some of the other issues raised, there is a lack of consistency around survey questions and indicators in relation to whether they are phrased as a maternal behaviour or as an infant behaviour [
5]. Different terms include breastfed (breast fed, breast-fed), breast milk (breastmilk) fed, expressed breast milk fed, breastfeeding (breast-feeding, breast feeding), breast milk feeding, and expressed breast milk feeding.