Monitoring breastfeeding indicators in high‐income countries: Levels, trends and challenges

Abstract Monitoring indicators of breastfeeding practices is important to protect and evaluate the progress of breastfeeding promotion efforts. However, high‐income countries lack standardized methodology to monitor their indicators. We aimed to update and summarize nationally representative annual estimates of breastfeeding indicators in high‐income countries and to describe methodological issues pertaining to the data sources used. A review was conducted through population‐based surveys with nationally representative samples or health reports from nationally representative administrative data of electronic surveys or medical records. Methodological aspects and rates of all breastfeeding indicators available were summarized by country. The median and annual growth of breastfeeding in percentage points within countries with time‐series data were estimated. Data from 51 out of 82 high‐income countries were identified. The data were obtained through surveys (n = 32) or administrative data (n = 19). Seventy‐one percent of countries have updated their indicators since 2015. Ever breastfed was the indicator most frequently reported (n = 46), with a median of 91%. By 6 months of age, the median equals 18% for exclusive and 45% for any breastfeeding. At 12 months, the median of continued breastfeeding decreased to 29%. The annual growth rate for ever breastfed, exclusive and any breastfeeding at 6 months and continued at 12 months varied from 1.5 to −2.0, 3.5 to −3.1, 5.0 to −1.0 and 5.0 to −1.9, respectively, with positive changes for most countries. Stronger interventions are needed to promote breastfeeding in high‐income countries as a whole, and investments are required to monitor trends with standardized methodologies.


| INTRODUCTION
Monitoring indicators of breastfeeding practices is important to protect and evaluate the progress of breastfeeding promotion efforts.
Despite the prevalence of breastfeeding initiation being over 80% in most high-income countries, a drastic drop in breastfeeding rates is observed within the first 6 months of life, especially in the case of exclusive breastfeeding . Longer breastfeeding reduces the infant risk of infectious morbidity and mortality, dental malocclusions and probably the risk of obesity and type 2 diabetes, and it increases the child's intelligence (Victora et al., 2015), an effect that persists until adult life. For mothers, breastfeeding may prevent breast and ovarian cancer and may prevent the risk of diabetes . The Lancet Breastfeeding Series in 2016 highlighted the many benefits of breastfeeding to mothers and children, both in poor and in rich countries . Nevertheless, only a small proportion of all children receive any breast milk at 12 months of age (Sarki, Parlesak, & Robertson, 2019;Victora et al., 2016).
Indeed, there has been growing attention around the failure to protect, promote and support breastfeeding in high-income settings (Bagci Bosi, Eriksen, Sobko, Wijnhoven, & Breda, 2016;Mirkovic, Perrine, Scanlon, & Grummer-Strawn, 2014). Mothers and children in these countries are placed in a vulnerable position as a result of aggressive marketing by the infant formula industry. The violation of the International Code of Marketing of Breast Milk Substitutes is frequently reported (Baker et al., 2016;Grummer-Strawn, Holliday, Jungo, & Rollins, 2019;Theurich et al., 2019). Few hospitals adopt baby-friendly practices aimed at promoting the successful initiation of breastfeeding (Theurich et al., 2019). Additionally, there is wide variability among high-income countries regarding the implementation of recognized interventions to support breastfeeding after hospital discharge, such as counselling by a nurse, trained lactation counsellor, post-discharge telephone calls and home visits (Rollins et al., 2016;Skouteris et al., 2014). There are also marked differences in maternal schooling levels and family income in breastfeeding indicators within countries of the World Health Organization (WHO) European Region (Baerug et al., 2017;Hughes, 2015;Sarki et al., 2019) and in North America (Hughes, 2015). Another critical issue is paid parental leave, which also differs widely among high-income countries. The length and level of maternal and parental leaves are important determinants to support mothers and to prolong breastfeeding rates (Theurich et al., 2019).
However, few high-income countries monitor their indicators according to the current WHO guidelines (Sarki et al., 2019;Theurich et al., 2019;Victora et al., 2016). The lack of standardized surveys and inconsistencies in data collection among these countries prevent WHO breastfeeding recommendations from being monitored. This may also delay the timely implementation of appropriate policy actions to protect and promote breastfeeding (Sarki et al., 2019;Theurich et al., 2019;Victora et al., 2016). We aimed to update and summarize nationally representative annual estimates of breastfeeding indicators in high-income countries and to describe methodological issues pertaining to the data sources used, providing a critical appraisal of available data.

| METHODS
Our analyses were based on nationally representative breastfeeding indicators from countries classified as high-income economies. The potential list of countries was obtained from the World Bank database that classifies global economies based on levels of gross national domestic product per capita (World Bank, 2019

Key messages
• Current breastfeeding practices in most high-income countries fall well short of international recommendations of exclusive breastfeeding under 6 months and continued breastfeeding until 2 years.
• Thirty-six out of 51 high-income countries have updated their indicators since 2015, and eight have collected breastfeeding indicators for more than two decades.
• National censuses of children estimated from maternity hospitals or follow-up in primary care services have replaced national health and child surveys in high-income countries.
• Lack of standardized methodologies and definitions affect the comparability of breastfeeding indicators of high-income countries, especially related to the variability of time frame to characterize exclusive breastfeeding under 6 months.
in the field from several countries. From these contacts, we requested information about recent and upcoming surveys, as well as on the existence of other possible data sources with national representative samples (e.g., health reports of administrative data).
We selected documents with sufficiently detailed methodological descriptions. We only incorporated breastfeeding rates published in country card reports and websites from international agencies and civil society organizations after accessing the original complete references. The documents were obtained in a diversity of national languages and, if necessary, we used the Google Translate website. In some cases, a brief report and tables in English were obtained when contacting national researchers by e-mail.

| Data extraction
We extracted all available annual data points into an Excel spreadsheet. Countries were characterized according to the type of institution that produced the data, data source type (survey or electronic data), year, follow-up, target population and response rate (in case of electronic data, percent of health units that submitted data reports).
We carefully reviewed the surveys or health statistical report methodology to identify changes in study design or time frame estimation over time that could affect the retrospective comparison of annual data. For countries with more than one data source, we prioritized those with the highest national representativeness and those provided by governmental institutions.

| Breastfeeding indicators
Several national surveys and health reports did not estimate breastfeeding indicators according to the internationally standardized definitions (WHO, 2008(WHO, , 2018. To allow comparison, we grouped the available indicators into specific breastfeeding variables as follows: i. Early initiation of breastfeeding within the first hour of birth, using the international recommendation. ii. Ever breastfed: infants reported having been breastfed, even if for a short period, including breastfeeding at the hospital, or any breastfeeding up to 2 weeks after birth. If needed, we estimated 'ever breastfed' by the complement of the percent of children never breastfed. Five countries (Antigua and Barbuda, Aruba, Israel, Japan and Latvia) that did not have data on ever breastfeeding but reported on the percent of children receiving any breast milk at 1, 2 or 3 months; these countries were analysed separately.
iii. Exclusive breastfeeding was extracted according to the variety of time frames presented in national surveys and reports: 1 week, 2 weeks, 1 month, 2 months, 3 months, 4 weeks, 5 months, and 6 months, 0-3 months and 0-5 months.
vi. Continued breastfeeding at 2 years, including estimations at 20 to 23 months, 21 to 23.9 months and 24 months.
A detailed summary of the WHO indicators and breastfeeding rates grouped is presented in Table 1.

| Statistical analysis
Descriptive analyses were performed with Stata 16.0 and R software, version 4.0.2. We presented a methodological description of the data extracted (i.e., source of data, annual updates, population surveyed and response rate) and the number of indicators by country and data source. Bar charts were prepared with the last updated breastfeeding rates per indicator. For countries with two estimates for the same indicator over time, we used the Wilcoxon signed-rank test to assess whether there was a significant change over time in the breastfeeding indicator when all countries with two data points were considered.
Annual growth analyses in percentage points (pp) were carried out by calculating the difference between the last and first annual rates divided by the difference in years between rates. Analyses were restricted to countries with annual rates estimated in a similar time frame and without differences in methodology that restricted comparisons between years. Graphs were displayed for each indicator to allow visual comparisons across countries.

| Ethical considerations
The datasets used are publicly available with no identifiable information. Ethical clearance for conducting the surveys were the responsibility of the institutions conducting the survey.

| RESULTS
We obtained data from 51 out of 82 high-income countries and territories. The time period covered by the data points ranged from 1986 to 2019. National breastfeeding indicators were estimated through breastfeeding surveys (n = 4), infant surveys (n = 13), infant immunization surveys (n = 5), national health surveys (n = 10), national health statistical reports summarized from maternity hospitals (n = 3) and primary health facilities (n = 16). Thus, data were gathered from 32 national surveys and 19 electronic medical record systems (Table 2, Appendix B).
Seventy-one percent of the countries had updated their indicators since 2015, and 16% have collected breastfeeding indicators for more than two decades. Updates were conducted annually (n = 22), every three (n = 2) or five to nine (n = 6) years. The other 21 countries had no regular updates, or these had been discontinued (Table 2).
Breastfeeding rates were estimated from mothers of children of different age groups: newborns followed up to 8 weeks after birth (n = 5), children up to 12 to 15 months (n = 17), children under 24 months (n = 9) or children under 3 to under 6 years (n = 11). In eight surveys, the population sampled was women who had a child in the prior 3 to 5 years ( Table 2).
The final response rate was ≥80% in 16 surveys and statistical reports, in contrast with seven with rates below 50% and 13 other sources that did not report this information (Table 2).
Malta and Latvia had data in previous years but discontinued in the updated ones.

| Ever breastfeeding
Forty-six countries presented data on the prevalence of ever breastfed children, with a median of 91%. Twenty-three countries reported a prevalence higher than 90%, 13 countries reported a prevalence between 80% and 90%, and 10 reported a prevalence between 60% and 79%. The lowest prevalence levels were reported in Scotland (65%), Wales (62%), Northern Ireland (60%) and the Republic of Ireland (60%) (Figure 1, Table S1).
The annual growth rate for ever breastfed children varied from 1.5 pp to −2.0 per year, with positive changes in half of the countries with data. The highest growth rates were found in Puerto Rico  Table S2). There was no significant trend (P = 0.273) in this indicator over time, with median values of 90% in the earlier and 91% in the most recent estimate (data not shown in tables).

| Exclusive breastfeeding under 6 months
Forty countries presented at least one indicator of exclusive breastfeeding at a given age under 6 months. Data availability according to time frame varied widely among countries. Of the 20 countries with data at the age of 3 months, the highest rates were reported by Uruguay (72%), Iceland (69%) and Norway (68%), and the lowest were Czech Republic (21%), Northern Ireland (19%) and Aruba (12%) ( Table 3). The annual growth was estimated for 16 countries with changes varying from 6.5 pp to −3.0 pp. The highest growth rates were observed for Puerto Rico (6.5 pp), Guam (3.0 pp) and Uruguay (2.7 pp), and the lowest for the Virgin Island   Rate refers to the last study follow-up; more than one rate corresponds to different phases within the study. For electronic data, rate was based on the number of health units that had submitted data or the percent of unknown/missing data reported. (−3.0 pp), Czech Republic (−1.9 pp) and Estonia (−0.3 pp) (Table S2).
Eight out of 14 countries presented multiple data to estimate the annual growth, with changes varying from 3.0 pp to −0.5 pp. The highest annual growth rates were reported by Finland (3.0 pp), Uruguay (2.9 pp), Portugal (0.9 pp) and Aruba (0.9 pp) (Table S2).

| Exclusive breastfeeding at 6 months
Thirty countries presented data on exclusive breastfeeding at 6 months with levels varying from 0.1% to 57%, with a median of   Table S1).
Sixteen out of 21 countries presented an annual increase, ranging from 5.5 pp to 0.1 pp, and Puerto Rico (5.0 pp), Estonia  Table S2). In the group of countries with two measurements over time, the median value increased significantly from 33% to 45% (P = 0.0003 in the Wilcoxon test).
Annual changes in continued breastfeeding at 12 months the median value increased significantly from 24% to 32% (P = 0.002 in the Wilcoxon test) (data not shown in tables).

| Annual changes within countries with conflicting trends
The annual changes in breastfeeding rates were not consistent within some countries, in which an observed increase in one indicator con-  Table S2).

| DISCUSSION
We described the current patterns and annual growth in breastfeeding rates in high-income national economies, with emphasis on the strengths and weaknesses of available data. Using different sources of data, we were able to obtain at least one breastfeeding indicator for 51 countries, with recent estimates and time trends for most of these countries. Nevertheless, the lack of standardized methodologies and definitions affected the comparability of breastfeeding indicators, especially related to the variability of the time frame used to characterize exclusive breastfeeding under 6 months. The present analyses provide the basis for future studies comparing national recommendations and actual practices.
Our data confirm that most children (median of 91%) in highincome countries are ever breastfed, but there are marked declines in the first months of life. By 6 months of age, the median prevalence equals 18% for exclusive and 45% for any breastfeeding. The number of countries that monitor breastfeeding indicators from 6 months onwards decreases by one third. The lack of data suggests that continued practice of breastfeeding after 6 months does not seem to be strongly encouraged in high-income countries, as these data are being monitored by a few countries. It may also reflect the fact that longer durations of breastfeeding are not sufficiently valued in several highincome countries.

Most countries have adopted the WHO and United Nations
Children's Fund (UNICEF) recommendation of exclusive breastfeeding for 6 months, have initiatives to support breastfeeding and attempt to monitor trends. Nevertheless, our review observed some contradictions regarding recommendations on the duration of exclusive breastfeeding. The European Society for Paediatric Gastroenterology, Hepatology, and Nutrition Committee on Nutrition supports the recommendation of exclusive breastfeeding for the first 6 months, yet advises that complementary feeding should not be introduced before 4 months nor be delayed beyond 26 weeks (6 months) (Fewtrell et al., 2017). Consequently, in most European countries, solid food is ventions (Rollins et al., 2016). Initiatives to support breastfeeding must consider its multifactorial determinants, which include individual factors, social attitudes and values, workplace protection, maternity leave legislation and enforcement, control of advertising by formula companies and health care services that support breastfeeding.
Programmes and interventions must target both antenatal and post-natal periods, be extended to mothers, fathers and family members and be combined with the relevant policies and regulations mentioned above (Rollins et al., 2016). Assessment of breastfeeding programmes and monitoring standardized indicators every 5 years are also essen-  (2019) in their attempt to use electronic data from Croatia (Milos et al., 2019). In their case, because breastfeeding information was not collected at similar ages for all children, the database only allowed for the estimation of indicators in broad age categories, and in addition, the percentage of unreported data fluctuated in specific years and imposed difficulties in drawing time trends.
Some difficulties were faced during our search process. Some reports are published in their native language, which make them difficult to identify using our search strategies, as well as hindering their use by the global scientific community. Although we have followed a broad protocol for searching the grey literature, in some countries, this failed to produce any information, but existing data reports were identified after contacting institutions and researchers through e-mail. Additionally, some reports lacked details on important methodological aspects, such as description of breastfeeding definitions, target population and sample response rate. In three cases, raw data were obtained from national sources, and we were able to calculate the indicators. While a number of countries stand out for their high quality, methodologically sound reports (Australia, Luxembourg, Norway, Finland, Northern Ireland and Oman), the available reports for other countries (such as Antigua and Barbuda, Aruba, Saudi Arabia, Portugal and Spain) were not very informative.
The strengths of our review include the ability to obtain updated information on breastfeeding indicators for a substantial number of countries without standardized surveys due to our comprehensive search of different data types. In addition, we were able to estimate time trends for most indicators. The main challenge of the current review was the lack of regular surveys in most countries and difficulties in identifying existing data sources in some countries. For example, we were unable to find any data for 31 high-income countries and territories. In addition, breastfeeding data were derived from a variety of sources, and the age of measurement often failed to agree with the international definitions, which limited between-country comparisons. Lack of uniformity was particularly common for exclusive breastfeeding indicators. As a consequence, it was not possible to stratify our analyses based on methodology (study design, population and age definitions), source of data (electronic versus survey) and final response rates.

| CONCLUSIONS
Our review shows that current breastfeeding practices in most highincome countries fall well short of international recommendations of exclusive breastfeeding for 6 months and continued breastfeeding until 2 years. Our analyses of the median values of countries with available data show that although nine out of 10 children start to be breastfed, only about half are still breastfed (and one quarter exclusively breastfed) at 6 months, and one third remain on the breast at 12 months. On the positive side, there were significant increases in all indicators under study, except for ever breastfeeding. Stronger interventions are needed to promote breastfeeding in high-income countries as a whole, and investments are required to monitor trends with standardized methodologies.

ACKNOWLEDGMENTS
The authors gratefully acknowledge the researchers and institutions listed as follows: Australian Bureau of Statistics, Australia; Phillip