Official Statistics

Infant feeding survey 2024: technical appendix

Published 4 June 2026

Applies to England

1. Introduction

Background to the survey

The infant feeding survey (IFS) asks mothers about their experiences of feeding their infant. It collects data on all aspects of infant feeding, including practices, attitudes and behaviours.

Survey design and methodology

The survey has been updated to reflect current practices in England. The Office for Health Improvement and Disparities (OHID) in the Department of Health and Social Care (DHSC) and Ipsos worked collaboratively to conduct this redesign. This included an extensive review of the questionnaires used in the 2010 survey to ensure that the content reflected current government policy and NHS guidance, and population trends.

The survey collection method was also reviewed to allow for the questionnaires to be completed either online or on paper, as in previous years.

A ‘push-to-web’ methodology was used, which means participants were sent a postal invite encouraging them to take part in the survey online. A paper questionnaire was also available, but only after 2 reminder letters had been sent. This differs from 2010, where data was largely collected from postal responses, but an online option was available.

The sample design remained broadly similar to that used in the 2010 survey. As in 2010, we invited mothers in 2024 to respond at 3 different time points, referred to as phases. These were planned to be when their baby was around 9 to 12 weeks, around 4 to 5 months and around 8 to 10 months. Due to fieldwork timings and response patterns, infant ages at each phase were:

• phase 1: when their baby was 2 to 5 months
• phase 2: when their baby was 4 to 7 months
• phase 3: when their baby was 8 to 10 months

Questionnaires were designed to reflect the ages of their infants at each phase. You can find copies of the questionnaires and other survey materials in the Box file sharing folders:

• Infant feeding survey 2024: questionnaires
• Infant feeding survey 2024: fieldwork documents  

2. Survey development

Overview of development work

The approach to updating the questionnaire and materials for IFS 2024 centred on engaging with stakeholders to understand their needs and data priorities. Stakeholders included mothers who had infants of a similar age to those in the survey. This ensured that the survey reflected the current trends and framed the questions in a sensitive manner.

This section outlines the framework that was used to develop the 3 questionnaires for IFS 2024. Overall 9 distinct engagement activities took place between April 2022 and November 2022, as set out in the section below. The pilot is covered in detail in section 4.

Scoping

Rapid evidence review

In May 2022, Ipsos conducted a rapid evidence review to explore the latest policies, guidance and developments associated with infant feeding. This assessment aimed to identify potential issues and topics that should be considered when developing the questionnaires.

For this assessment, a list of relevant documents was compiled by OHID and Ipsos’ academic partner Alison McFadden, who is Professor of Mother and Infant Public Health at the University of Dundee, and leads the Mother, Infant and Child Health Research Unit there. Professor McFadden reviewed these documents using a comprehensive analysis framework and recommended that IFS 2024:

• explore personalised care plans and information accessed during pregnancy as possible determinants of infant feeding behaviours
• collect data on breast milk at the first feed and discharge from hospital as well as support that was received
• consider including questions on mothers’ eating habits during pregnancy, particularly relating to allergenic foods
• consider collecting information on take-up of maternity income support
• include questions that capture mothers’ mental health and the relationship with infant feeding
• consider capturing details of support received by mothers as well as their experiences of feeding in public
• collect data on attitudes and behaviours to introducing solid foods

Exploratory interviews

Ipsos conducted 10 in-depth interviews with mothers who had given birth in the last 14 months to understand the opinions, attitudes and practices of mothers feeding their infant. Mothers were recruited using an expert agency that intentionally targeted mothers who are typically less represented in quantitative surveys, or those who may have poorer experiences of maternity services. Mothers were selected from face-to-face ‘on street’ methods (recruiters approached participants in their local area), as well as through local networks and panels, to capture the experiences of a wide range of mothers. Quotas were set for the:

• age of the mother
• social and economic group
• ethnicity
• geography
• age of their child
• whether this was the mother’s first child

As a result of these interviews, questions were retained from IFS 2010 on support received by mothers during pregnancy and immediately after birth. Social media was highlighted as an information source for mothers, so this response option was added to relevant questions.

Review of existing surveys

Ipsos and OHID reviewed the IFS 2010 and more recent surveys that cover a similar topic area, specifically the:

• Scottish Maternal and Infant Nutrition Survey 2017
• Maternity and Paternity Rights Survey 2018
• Care Quality Commission (CQC) Maternity Survey 2022

This provided example ‘question and answer’ option wording that could be adopted for IFS 2024. It also highlighted what information is already captured in other surveys, helping to avoid collecting duplicate data.

Questionnaire and materials development

The outcome of the scoping phase, as well as the fact that the last IFS took place more than a decade before, meant that the survey questionnaires needed updating. This phase of the development consisted of 3 distinct stages, which are outlined in the following sections.

Review of best practice for conducting surveys using smartphones

As part of the development of the 3 IFS 2024 questionnaires, Ipsos and OHID conducted a review against Ipsos’s Mobile-first best practice guide. These are best practice principles for conducting a survey on a smartphone and include the following:

• question stems (the part of the question that presents the issue being asked about) should be no more than 140 characters, including spaces
• response options should use no more than 7 options to reduce scrolling
• scaled questions should avoid ‘sliders’ as answer options and use 5-point scales where possible
• text should be in plain, direct language and not have any unnecessary words
• the question stem, instructions and response categories should be distinctly formatted and consistent
• the question should be formatted in a way that is suitable for responding online

Updates were made to all 3 questionnaires to bring them more in line with these principles.

A ‘mobile first’ approach means that the online questionnaire is designed with smartphone users in mind initially. This is because this is increasingly how participants choose to access online questionnaires. The design of the survey will then be easily transferrable for completion on other devices, such as laptops or tablets.

Ipsos developed these principles to minimise participants leaving surveys partway through completion and to increase response rates. They improve the user experience when completing a survey on a smartphone where the screen size is limited.

Review of 2010 questionnaires

The questionnaires from 2010 were initially reviewed by the Subgroup on Maternal and Child Nutrition (SMCN). This is a subgroup of the Scientific Advisory Committee on Nutrition (SACN) that advises government on child and maternal diet and nutrition.

SMCN reviewed the 3 IFS 2010 questionnaires and highlighted areas that would need to be amended to bring them in line with the latest government advice and guidance. This included:

• capturing experiences of feeding in each phase of the questionnaire
• ensuring that mixed feeding was better captured

SMCN also noted the need to capture data on:

• the use of specialised formula
• introduction of potentially allergenic foods
• family history of allergies
• types of dietary pattern followed, for example vegan diets

SMCN also suggested amendments to reduce respondent burden such as

• replacing free text boxes with answer options
• ranking questions by priority order
• suggesting specific wording changes

Project board and stakeholder workshops

A variety of stakeholders were consulted throughout the development of the survey to ensure that the questionnaires developed were fit for purpose and the data collected met their needs.

Stakeholder workshops

In June 2022, Ipsos and OHID held a stakeholder workshop with organisations, including government departments, charitable organisations and other, non-government organisations and academics. Attendees were asked to consider:

• the main areas of interest from the survey
• how the survey from 2010 was currently used
• important questions to be retained and questions that could be removed

From these discussions, there were several suggested amendments to the 2010 questionnaires as well as additional considerations for the 2024 survey. Most notably changes to:

• collect information on donor milk
• update the questionnaires to be clear on expressing milk compared with feeding from the breast
• understand more around women’s experiences of breast feeding in public and at work, given the introduction of the Equality Act in 2010
• use inclusive language (for example referring to the baby as ‘they’ rather than ‘he or she’)
• recognise that trends may be lost if questions are altered too much, but also consider how new questions might set the base for a new set of trend data
• reduce the number of open-ended response options, which add to the respondent burden and affect response rates (where appropriate, closed questions or response options were recommended)

Ipsos and OHID updated the survey questionnaires based on the feedback collected during the development process.

Project board

OHID established an IFS 2024 project board to gain expert input on the development of the questionnaires as well as provide guidance on the direction of the survey.

This project board consists of members with expertise in nutrition and infant feeding, including representatives from:

• OHID
• SACN
• NHS England
• Royal College of Paediatrics and Child Health
• University of Dundee
• Ipsos

During the development of the survey, the project board met twice. The first meeting was to agree the survey approach. In the second meeting, they reviewed the questionnaires and survey materials, with a focus on the new questions included.

The board was also consulted during 2 further stages of the project:

• between the phase 1 pilot and main survey, to gain their input on potential changes to the questionnaire - see the ‘Pilot survey’ section for further details
• before developing the survey outputs, to gain their insight on analysis and reporting

Cognitive testing

Between October and November 2022, the survey materials and a selection of questions from the 3 questionnaires were cognitively tested with mothers recruited through an expert agency in cognitive testing.

What cognitive testing is and how it was used

Cognitive testing is a method used to critically evaluate questions. It uses specialist techniques to help understand how participants process and respond to materials - in this case, the questionnaires and survey materials. For IFS 2024, the cognitive testing stage set out to test participants’ comprehension, recall, judgement and response to the selected questions.

This process seeks to expose any potential issues with the validity and measurement of the questions in the questionnaire. It can also suggest alternative wording or response categories for the items being tested.

Methodology

Between October and November 2022, a total of 28 online interviews were conducted across 3 rounds of cognitive testing. Of these, 21 interviews focused exclusively on testing a selection of questions that were new, or where significant changes had been made since 2010, across all 3 questionnaires. The remaining 7 interviews focused on the survey materials, such as the invitation letters and text messages.

During each interview, the interviewer shared their screen and showed the respondent the questions or survey materials. Participants were asked to read the content as they would normally, and talk through their initial thoughts. Interviewers then asked follow-up questions to determine their level of understanding.

Between each round of interviews OHID and Ipsos met to discuss the feedback and make any necessary changes that were then tested in subsequent rounds. Any questions that had no issues were deprioritised in the later rounds of interviews.

Changes based on feedback from cognitive testing

Changes were made to the questionnaires and survey materials to reflect the feedback received from participants. The most significant points of feedback are outlined below.

Letters

The addressee on the letters was updated to include “and baby”. Feedback from the cognitive interviews suggested that the inclusion of this wording would make mothers more likely to open the letter, as they would know it related to their baby.

Text messages

Interviewees were largely happy with the text messages. The main feedback was that they should be sent out at different points in the day to catch mothers when they might be free.

Wellbeing questions

None of the interviewees expressed a negative opinion about the personal wellbeing questions, but some felt that they could be challenging, depending on personal circumstances. These are standardised questions widely used across many national surveys, and following discussions between Ipsos and OHID, it was decided to retain them.

Food security questions

None of the interviewees expressed concerns about the food security questions being included.

Smoking and alcohol questions

The questions on drinking alcohol were only used for the phase 1 questionnaire and were revised to align with current guidance on drinking when pregnant.

The questions on smoking were included in all 3 phases and were revised to align with current guidance on smoking when pregnant and in the months after birth.

Type of diet questions

A new question about the type of diet mothers followed was added to the phase 3 questionnaire.

Parental leave question

A question about parental leave was amended to reflect current policies on Maternity pay and leave.

New questions included for IFS 2024

Once cognitive testing was complete, the questionnaires were signed off for the pilot. New questions included for the first time in the IFS 2024 are set out below.

Wellbeing

The Office for National Statistics personal wellbeing questions (see Personal well-being user guidance) were added to the phase 1, 2 and 3 questionnaires. These questions cover life satisfaction, feeling worthwhile, happiness and anxiety.

Donor milk

Two questions on the use of donor milk and its source were added to the phase 1, 2 and 3 questionnaires.

Food insecurity

Questions on food security were added to the end of the phase 1, 2 and 3 questionnaires. These questions covered concerns about food running out, food not lasting or not being able to afford a balanced meal.

Tongue tie

New questions on tongue tie covered its prevalence, associated feeding problems and experience of a tongue tie division. These questions were included in the phase 2 questionnaire only.

3. Sampling

Sample overview

The sample for IFS 2024 was drawn from NHS England’s Maternity Services Data Set (MSDS). This is a patient-level data set that captures information about activity carried out by maternity services, relating to a mother and infant or infants. It covers all activities from the point of the first booking appointment until mother and infant or infants are discharged from maternity services. The NHS Health Research Authority Confidentiality Advisory Group approved access to this sample through a section 251 application. NHS England also approved it through a Data Access Request Service application. The University of Dundee conducted an independent ethics review.

MSDS data is used for planning and research in anonymised format. The sample for the survey was taken before the data set was anonymised.

The target sample size by the end of phase 3 was to collect 5,000 responses. Ipsos used this, along with predicted response rates based on the CQC Maternity Survey (which surveys a similar population), to calculate that a starting sample size of 26,339 was required to achieve this target. It was clear from these calculations that a single month of births across England would be sufficient to sample from. For the pilot, this was births in August 2023, while for the main survey this was births in December 2023.

To increase the opportunity for subgroup analysis, the survey aimed to maximise data collection from mothers from ethnic minority groups and those living in more deprived areas. To achieve this, the sampling strategy was designed so that all mothers from the selected birth months who identified as Asian, Black, Mixed or Other ethnic groups were included in the sample.

The rest of the sample was made up of a random 50% sample of mothers who identified as White or White British, or whose ethnicity was not known (3.6% of mothers).

This approach was informed by analysis of response rates by recorded ethnicity of mothers from the CQC Maternity Survey, which identified that White or White British mothers had higher response rates than mothers from Asian or Asian British, Black, Mixed, or Other ethnic groups. Sampling was based on MSDS ethnicity data. See annex 2 for details.

The incentive strategy was designed to encourage sufficient responses from mothers living in more deprived areas. See the ‘Pilot response rates’ section for more information.

Dissent process

Ahead of the sample selection, women who were pregnant or had recently given birth were given the opportunity to opt-out of being invited to take part in the survey. Posters explaining how to do this were displayed in settings where mothers and pregnant women would see them. They were also shared on DHSC social media channels and reposted by stakeholders.

Any individual that opted out during this period was removed before sampling took place. During fieldwork, mothers that were sent invites to the survey were able to opt-out of future mailings.

Sample selection, including eligibility criteria

To be eligible for the survey, mothers had to meet the following eligibility criteria:

• gave birth to a live baby in August 2023 (pilot) or December 2023 (main survey)
• were aged 16 and over at the time of sampling
• had a home postcode that matched the baby’s postcode

Sample extraction procedure

Following approval from NHS England, MSDS data was shared with Ipsos in 2 stages to minimise the amount of personal data shared. The first stage required NHS England to share a pseudonymised data set (data that does not contain direct identifiers) with Ipsos. Ipsos then drew the sample of selected mothers, which was sent back to NHS England. In the second stage, NHS England shared the full contact details of the selected sample.

The survey materials were mailed out by post, and text message reminders were sent for the phase 1 questionnaire. Ipsos activated the online survey and sent respondents their log-in details by post and text message reminders.

A recontact question was included at the end of the phase 1 questionnaire. Once fieldwork for phase 1 ended, Ipsos processed responses to create a file of respondents who consented to being sent questionnaires at phases 2 and 3. This included a flag to highlight who should be sent a multiple births questionnaire at phase 2 (see the ‘Main survey’ section for further details).

Sample refreshes

Ahead of each mailing, efforts were made to remove mothers that should no longer be contacted. These included mothers:

• who asked to be removed from the sample
• experiencing loss or bereavement
• that had died

For each refresh, NHS England ran the exclusion checks and supplied a file to Ipsos of ID numbers for mothers who should be contacted. Ipsos’s sampling team created a file of ID numbers to exclude from the next mailing by comparing the previous list with the most recent NHS England list.

Sample checks were conducted at each stage to ensure the accuracy, reliability and validity of the sample data.

Pilot starting sample sizes

The starting sample size for the pilot was 2,631 mothers. This was designed to be about 10% of the overall sample. The pilot sample was selected from births in August 2023. The starting sample sizes for each pilot phase are shown in table 3.1 and the profile of each phase in tables 3.2 to 3.4.

The phase 1 starting sample is the sample of mothers selected from MSDS data. The phase 2 starting sample is mothers who responded at phase 1 who agreed to be recontacted for phases 2 and 3. The phase 3 starting sample is mothers who responded at phase 2. The phase 1 ethnicity sample profile reflects the sampling strategy which oversampled mothers from ethnic minority groups to achieve sufficient responses for subgroup analysis by ethnicity.

Table 3.1: pilot starting sample sizes

Phase	Starting sample size
Phase 1	2,631
Phase 2 (see note)	690
Phase 3	455

Note: 690 of the mothers who responded at phase 1 agreed to be recontacted.

Table 3.2: pilot starting sample profile per phase: ethnicity

Ethnicity	Phase 1	Phase 2	Phase 3
White (or ethnicity not known or stated)	62.0%	66.2%	70.5%
Asian or Asian British	19.8%	15.7%	12.3%
Black or Black British	8.8%	10.0%	9.2%
Mixed	3.5%	3.6%	3.3%
Other ethnic group	5.9%	4.5%	4.6%

Table 3.3: pilot starting sample profile per phase: Index of Multiple Deprivation quintiles

Index of Multiple Deprivation (IMD) is the official measure of relative deprivation in England, and broadly defines deprivation based on an individual’s living conditions. IFS 2024 uses IMD quintiles where IMD 1 is the most deprived and IMD 5 is the least deprived.

IMD quintile	Phase 1	Phase 2	Phase 3
IMD 1 (most deprived)	27.0%	23.5%	23.1%
IMD 2	22.7%	17.8%	15.6%
IMD 3	18.9%	20.0%	20.0%
IMD 4	16.8%	19.3%	18.9%
IMD 5 (least deprived)	14.6%	19.4%	22.4%
Blank	0.1%	0.0%	0.0%

Table 3.4: pilot starting sample profile per phase: mother’s age

Mother’s age	Phase 1	Phase 2	Phase 3
16 to 19 years	2.5%	0.1%	0.0%
20 to 24 years	11.5%	7.3%	5.5%
25 to 29 years	26.3%	22.8%	22.5%
30 to 34 years	34.3%	38.2%	38.8%
35 years and over	25.3%	31.6%	33.3%

Main survey starting sample sizes

A total of 23,708 letters were posted out in phase 1 of the main survey. The sample was selected from births in December 2023. The starting sample size at each phase is shown in table 3.5. Tables 3.6 to 3.8 provide details of the sample profile at each phase.

The phase 1 starting sample is the sample of mothers selected from MSDS data. The phase 2 starting sample is mothers who responded at phase 1 who agreed to be recontacted for phases 2 and 3. The phase 3 starting sample is mothers who responded at phase 2.

The phase 1 ethnicity sample profile reflects the sampling strategy which oversampled mothers from ethnic minority groups to achieve sufficient responses for subgroup analysis by ethnicity.

Table 3.5: main survey starting sample size

Phase	Starting sample size
Phase 1	23,708
Phase 2 (see note)	7,580
Phase 3	4,609

Note: 7,580 of the mothers who responded at phase 1 agreed to be recontacted.

Table 3.6: main survey starting sample profile: ethnicity

Ethnicity	Phase 1	Phase 2	Phase 3
White (or ethnicity not known or stated)	59.4%	66.2%	70.4%
Asian or Asian British	21.8%	15.3%	12.8%
Black or Black British	9.2%	10.5%	9.5%
Mixed	3.8%	3.4%	3.1%
Other ethnic groups	5.9%	4.6%	4.1%

Table 3.7: main survey starting sample profile: IMD quintile

IMD quintile	Phase 1	Phase 2	Phase 3
IMD 1 (most deprived)	29.2%	26.4%	24.4%
IMD 2	22.0%	24.4%	25.0%
IMD 3	18.2%	16.6%	16.4%
IMD 4	16.0%	16.0%	16.7%
IMD 5 (least deprived)	14.6%	16.5%	17.6%
Blank	0.1%	0.0%	0.0%

Table 3.8: main survey starting sample profile: mother’s age

Mother’s age	Phase 1	Phase 2	Phase 3
16 to 19 years	2.1%	0.9%	0.7%
20 to 24 years	11.8%	7.3%	5.7%
25 to 29 years	27.0%	24.3%	22.6%
30 to 34 years	34.4%	38.8%	40.4%
35 years and over	24.7%	28.7%	30.6%

4. Pilot survey

Overview of pilot survey methodology

A pilot study was conducted between October 2023 and June 2024, using a sample of births from August 2023. This pilot ensured that the questionnaires and methodology were tested in full, ahead of the main survey.

The survey used a sequential push-to-web methodology. Mothers were invited to take part in the survey with 3 invitation letters sent by post. Each letter was followed by a text message reminder if a mobile number was available for the mother. All mailings included information about how to access the online survey as well as a universal link that took participants to a sign-in page. A paper questionnaire was included in the final mailing only.

At the end of the questionnaire for phase 1, mothers were asked if they were happy to be recontacted for phases 2 and 3 of the survey. Those that agreed were then invited to take part in phase 2. The phase 3 survey was sent to those that completed phase 2.

The main aims of the pilot were to:

• test the proposed contact strategy
• test the questionnaires for each of the 3 survey phases in full
• provide an indication of the expected response rate at each phase
• test the proposed incentive strategy and trial different options (see the ‘Incentives’ section for more information)

Pilot contact strategy and fieldwork dates

The same contact strategy was used for each of the 3 phases as outlined in table 4.1.

Contact started with an initial invitation letter posted to all survey participants on the first day of each phase. Once the letters had arrived in the post, a text message reminder was sent to all those in the sample with a valid mobile number.

The first letter reminder was sent out a week later followed by a text message reminder. The second letter reminder, which included a copy of the paper questionnaire, was then sent out 2 weeks later. This was followed closely by a reminder text message. A further week later, a final text message reminder was sent.

Table 4.1: contact strategy for each pilot phase

Contact type	Day
Invitation letter arrives	Day 1
Text message 1	Day 2
Reminder letter 1 arrives	Day 7
Text message 2	Day 8
Reminder letter 2 to include a paper questionnaire arrives	Day 14
Text message 3	Day 15
Text message 4	Day 21

At each mailing, every effort was made to remove those that had already taken part in the survey.

Fieldwork for the pilot took place in 3 phases between October 2023 and June 2024. 

Table 4.2: fieldwork dates for each pilot phase 

Phase	Start date	End date
Phase 1	27 October 2023	4 December 2023
Phase 2	19 December 2023	26 February 2024
Phase 3	17 April 2024	16 June 2024

Incentives

Incentive strategy

A priority for the infant feeding survey is to capture the experiences of mothers from a variety of backgrounds. To achieve this, Ipsos and OHID employed an incentive strategy with the aim of encouraging a response from mothers from groups that are known to be less likely to complete a survey.

The pilot provided an opportunity to test the use of incentives to ensure the chosen strategy would be effective at encouraging responses. A mixture of unconditional and conditional incentives were tested:

• an unconditional incentive is included in the initial invitation to take part in the survey, meaning that the respondent does not have to complete the survey to receive the incentive
• a conditional incentive is only sent when the respondent has completed the survey

For the IFS 2024 pilot, £5 gift vouchers were offered for completing each questionnaire.

Incentive strategies tested

For the phase 1 pilot, an unconditional incentive (£5 voucher) was given to sampled mothers living in the most deprived areas. This strategy was used to see if the incentive would boost response rates among the most deprived 20% of the sample.

The phase 1 pilot showed that incentives were effective in boosting the response rate. For phase 2, alternative options were tested to be able to offer an incentive to more respondents. The phase 2 starting sample was split into the following 4 groups:

• no incentive: those that fell into IMD quintiles 3 to 5 and the less deprived half of IMD quintile 2 did not receive an incentive
• unconditional incentive - IMD quintile 1: those in the most deprived 20% were split randomly with half receiving an unconditional incentive
• conditional incentive - IMD quintile 1: the remaining half of those in the most deprived 20% received a conditional incentive, where they had to complete the survey to receive the £5 voucher
• conditional incentive - IMD quintile 2: those in the more deprived half of IMD quintile 2 received a conditional incentive, where they had to complete the survey to receive the £5 voucher

Once again, it was found that the incentives were having a positive effect on the response rate and the difference in response between groups receiving unconditional and conditional incentives was minimal (see the ‘Pilot response rates’ section for details).

As a result, the incentive strategy was revised for phase 3 of the pilot and for the main survey. Conditional incentives were offered at phases 2 and 3, at which point respondents were already engaged with the survey having agreed to be recontacted.

So, for phase 3 of the pilot, the sample was split into the following 2 groups:

• no incentive: those that fell into IMD quintiles 3 to 5 did not receive an incentive
• conditional incentive: those in IMD quintiles 1 and 2 received a conditional incentive

Pilot response rates

Response rates for each phase of the pilot are shown in tables 4.3 and 4.4.

Table 4.3: overall pilot response rates per phase

Phase	Unadjusted response rate
Phase 1	33.8%
Phase 2	65.9%
Phase 3	75.8%

Table 4.4: incentive response rates per pilot phase

Pilot phase and incentive	Response rate
Phase 1 overall	33.8%
Phase 1 unconditional incentive	33.4%
Phase 1 no incentive	33.9%
Phase 2 overall	65.9%
Phase 2 unconditional incentive	65.1%
Phase 2 conditional incentive	68.3%
Phase 2 no incentive	65.6%
Phase 3 overall	75.8%
Phase 3 conditional incentive	71.8%
Phase 3 no incentive	77.6%

Changes made as a result of the pilot

During the pilot, Ipsos and OHID met after each phase to review the performance of the pilot and discuss whether the survey was working as intended. This included reviewing the:

• response at each question
• contact strategy
• drop-out rates
• response rates per mailing

Ipsos and OHID also used these meetings to review the answers given by mothers to:

• open-ended questions that allowed a free text response
• ‘other (please specify)’ response options that gave a free text box in addition to formulated answer codes

The 2010 survey had used several open-ended questions to capture the variety of breastfeeding experiences. However, as part of the mobile first principles, the number of open-ended questions included in IFS 2024 was reduced to lessen the burden on respondents. Responses given in 2010 to open-ended questions were used to develop answer option lists for 2024. The pilot provided the opportunity for these lists to be sense checked to ensure that the codes were appropriate.

Following the pilot, OHID and Ipsos agreed amendments to the questionnaires, survey materials and contact strategy. These were minor changes, with the most notable outlined in tables 4.5 and 4.6.

Table 4.5: pilot questionnaire changes

Phase	Changes made
Phase 1	Additional response code at question 62 (pain relief during labour) (see note)
Phase 1	Instructions for how to answer questions on infant’s weight
Phase 2	New questions on infant’s experience of tongue-tie
Phase 2	Additional response code at question 41 (reason for giving drinks) (see note)
Phases 2 and 3	Additional question on food security created for phases 2 and 3 of the main survey
All 3 phases	Additional code created to capture whether more support from a community midwifery team would have helped mothers breastfeed for longer

Note: you can find information about questions 41 and 62 in the Box file sharing folder ‘Infant feeding survey 2024: questionnaires’.

Table 4.6: pilot material or method changes

Material or method	Changes made
Letters	Additional sentence added to letters to emphasise the importance of hitting the submit button at the end of the online survey.
Letters	NHS logo added to help legitimise the survey and encourage respondents to open each mailing.
Text messages	Line added to encourage mothers to make sure they submit their responses if they have completed the survey.
Text messages	Language reviewed to ensure that it was encouraging for mothers.
Incentives	Revised incentive strategy so that an unconditional incentive (£5 gift voucher) would be provided to those in IMD quintiles 1 and 2 (most deprived) at phase 1. For phases 2 and 3, the same group would be offered a conditional incentive, meaning they had to complete the survey to be able to claim the £5 voucher.

5. Main survey

Overview of main survey methodology

As the methodology proved successful in the pilot (see the ‘Pilot survey’ section for more information), the main survey also used a sequential push-to-web methodology. Mothers were invited to take part in the survey with 3 invitation letters sent by post.

Each letter was followed by a text message reminder, if a mobile number was available for the individual. The third letter was followed up by 2 text message reminders sent a week apart. All mailings included information about how to access the online survey. A paper questionnaire was included in the final mailing only.

Main survey contact strategy and fieldwork dates

Similar to the pilot, the main survey used the same contact strategy for each phase.

Table 5.1: contact strategy for all phases

Contact type	Day
Invitation letter arrives	Day 1
Text message 1	Day 2
Reminder letter 1 arrives	Day 7
Text message 2	Day 8
Reminder letter 2 to include a paper questionnaire arrives	Day 14
Text message 3	Day 15
Text message 4	Day 21

Table 5.2: main survey fieldwork dates

Fieldwork for the main survey took place in 3 phases between February and October 2024. 

Main survey phase	Start date	Close date
Main survey phase 1	23 February 2024	8 April 2024
Main survey phase 2	24 April 2024	10 June 2024
Main survey phase 3	13 August 2024	10 October 2024

Incentives used in the main survey

As detailed in section 4, incentives were used to help encourage survey participation by people from groups that are known to be less likely to respond to a survey. It was clear from the pilot that incentives were beneficial to encourage responses.

For the main survey, the following incentive strategy was used.

Phase 1

No incentive for mothers that fall into IMD quintiles 3 to 5.

An unconditional incentive (£5 voucher) for mothers that fall into IMD quintiles 1 and 2.

Phase 2

No incentive for mothers that fall into IMD quintiles 3 to 5.

A conditional incentive (£5 voucher) for mothers that fall into IMD quintiles 1 and 2 upon completion of the questionnaire.

Phase 3

No incentive for mothers that fall into IMD quintiles 3 to 5.

A conditional incentive (£5 voucher) for mothers that fall into IMD quintiles 1 and 2 upon completion of the questionnaire.

Mailing and text message process

Letter and questionnaires

Ipsos sent the final name and address sample to the printing house using a secure file transfer protocol (SFTP) with high-level encryption. All letters and questionnaires were digitally printed as required, ahead of each postal mailing. Business return envelopes and outer envelopes were also printed in advance. These envelopes included the NHS logo to help encourage respondents to open the letters.

The letters were personalised with name, address and the individual’s unique survey code, which also appears on the questionnaire. Letters were handed to Royal Mail for delivery.

Text messages

The final mobile number sample was sent to the text message provider using SFTP with high-level encryption. All text messages were personalised with a short URL that was unique to each participant, allowing them direct access to the online survey.

To manage the volume of text messages that were issued into the mobile network at any time, an automated system scheduled a set number of messages in batches, every 15 minutes over a 3-hour slot.

Reminders

Reminders (letters and text messages) were sent over a period of 21 days following the initial invitation (see table 5.1). Ahead of each mailing, efforts were made to ensure that reminders were not sent to mothers who met the following criteria at the point of the relevant deadline:

• had taken part in the survey online (or returned a paper questionnaire ahead of text message 3 or text message 4)
• had taken part in the survey through the helpline (see the ‘Online completion’ section for more information)
• had opted out of taking part in the survey
• mothers recorded as no longer eligible on the NHS England database

Online completion

Each mother in the sample was assigned a unique survey code. This was printed on the letter and on the front page of the paper questionnaire. This allowed them to access the online survey by using the shortened URL available in the letter. To complete the survey online, participants needed to either enter their unique survey code on a sign-in page or click on the unique URL in the text message reminder.

The wording and design of questions in the online survey were identical to those on the paper questionnaire. To ensure comparability between the online survey and paper questionnaire, participants were able to skip questions in the online survey. However, a prompt asked them if they were sure they wanted to skip each question, to encourage completion.

To improve accessibility, the online survey included:

• large answer option buttons
• keyboard navigation
• labelled ‘Next’ and ‘Back’ buttons
• compatibility with screen reader software on both desktop and mobile devices

Only one online response per participant was accepted. If participants tried to complete the survey more than once online, a message appeared letting them know they had already completed it. If they did not complete the survey in one sitting, their unique survey code returned them to where they had left off.

Telephone completion

Participants were able to complete the survey questionnaires on the telephone in the 5 languages offered (see the ‘Survey completion by language’ section) by calling the freephone helpline. However, no participants completed the survey by telephone.

Multiple birth questionnaire

IFS 2024 included a short additional questionnaire for mothers of multiple births.

Mothers of multiple births were identified at phase 1 and were subsequently asked to complete additional questions at phase 2. Completion was possible using the online survey (in a section at the end of the questionnaire only visible to mothers of multiple births) or as an additional paper questionnaire.

The questionnaire gathered data on how these mothers feed all their infants (including experience of solid foods) and the support they may have received. Unless directed otherwise, mothers were asked to complete the questions about their infant or infants other than the first-born twin, triplet or quadruplet (who were the focus of the main questionnaire).

You can find the full questionnaire in the Box file sharing folder ‘Infant feeding survey 2024: questionnaires’.

Main survey response rates

Response rates for each phase of the main survey are shown in table 5.4.

Table 5.4: overall main survey response rates per phase

Phase	Unadjusted response rate
Phase 1	39.1%
Phase 2	60.8%
Phase 3	69.4%

6. Response rates and survey completion

The pilot and main survey data sets were combined for reporting to use all the available data and provide larger sample sizes for analysis at subgroup level. The figures in this section reflect combined pilot and main survey responses.

Overall response rates by phase

The overall response rate for phase 1 of IFS 2024 was 38.6% based on 26,339 invites sent out and 10,168 valid returns.

The overall response rate for phase 2 was 61.2% based on 8,270 invites sent out (those at phase 1 who agreed to be recontacted) and 5,064 valid returns.

The overall response rate for phase 3 was 70.0% based on 5,064 invites sent out and 3,542 valid returns.

Table 6.1: overall response rate

Phase	Unadjusted response rate
Phase 1	38.6%
Phase 2	61.2%
Phase 3	70.0%

Mothers were only contacted in later phases of the survey if they had responded to the previous phase. This means the effect of non-response on the response rate at each stage is cumulative. So, the response rate at phase 3 of the survey based on the initial sample of mothers was 13.4%.

Response by group

Tables 6.2 to 6.4 provide response rates by ethnicity, IMD quintile and mother’s age.

See annex 2 for the demographic profile of mothers at each phase of the survey.

Table 6.2: response rate per phase by ethnicity

Ethnicity	Phase 1	Phase 2	Phase 3
Asian or Asian British	31.1%	50.9%	60.4%
Black or Black British	45.7%	55.4%	64.0%
Mixed	35.0%	56.3%	66.9%
White (or ethnicity not known or stated) (see note)	41.1%	65.1%	72.6%
Other ethnic group	32.1%	56.2%	69.8%

Note: at phase 1, 3.6% of the sample did not have a recorded ethnicity.

Table 6.3: response rate per phase by IMD

IMD quintile	Phase 1	Phase 2	Phase 3
IMD 1 (most deprived)	35.8%	56.9%	64.1%
IMD 2	40.7%	61.9%	71.4%
IMD 3	36.4%	60.4%	71.7%
IMD 4	39.4%	63.3%	71.3%
IMD 5 (least deprived)	42.9%	65.9%	72.9%
No IMD group (see note)	28.6%	33.3%	100.0%

Note: at phase 1, less than 0.1% of the sample did not have an IMD quintile. This was because their postcode did not have a corresponding IMD score, which can happen with new postcodes.

Table 6.4: response rate per phase by mother’s age

Mother’s age	Phase 1	Phase 2	Phase 3
16 to 19 years (see note)	19.6%	44.4%	37.5%
20 to 24 years	26.2%	48.1%	58.5%
25 to 29 years	35.3%	57.1%	64.3%
30 to 34 years	42.9%	63.6%	73.6%
35 years and over	43.7%	65.2%	72.1%

Note: mothers aged 15 and under were not eligible to take part in the survey.

Survey representativeness

The sampling strategy and use of incentives were designed to encourage response from mothers in groups that are typically underrepresented. The aim was to maximise data collection from mothers from ethnic minority groups, and those in more deprived areas, to increase the opportunity for subgroup analysis. Annex 2 shows the demographic profile of mothers at each phase of the survey compared with the profile of mothers from the sampled birth months (August and December 2023) and from births in the 2023 to 2024 year by IMD, ethnicity, maternal age and region. This shows the effect of the sample design, and the impact of the response rates achieved.

Mothers from ethnic minority backgrounds were overrepresented in the survey starting sample, and among phase 1 respondents due to the sample design. By phase 3 the profile of mothers responding was generally in line with mothers in the sampled birth months, although Asian mothers were slightly underrepresented and Black mothers were slightly overrepresented.

Mothers living in the most deprived areas (IMD 1) were slightly overrepresented in the starting sample, and among phase 1 respondents. By phase 3, mothers living in the most deprived areas were slightly underrepresented compared with mothers in the sampled birth months.

There were some differences by maternal age. Younger mothers were less likely, and older mothers were more likely, to respond to the survey at each phase. By phase 3, mothers aged 24 years and under made up 5% of respondents compared with 14% of mothers in the sampled birth months. Mothers aged 35 and over were overrepresented among phase 3 respondents (32%) compared with mothers in the sampled birth months (25%).

The survey data collected was weighted for analysis to correct for the sampling design and to reduce the impact of non-response bias (differences between groups in likelihood to respond to the survey), to ensure that the results are representative of mothers overall for ethnicity, deprivation and maternal age.

There may still be bias for other reasons. See the ‘Weighting’ section for more information, including how infant feeding status was used. Also, see the ‘Survey limitations’ section.

Multiple birth response rate

IFS 2024 included additional questions for mothers of multiple births, who were identified at phase 1 of the survey.

A questionnaire was sent out at phase 2 to 121 mothers of multiple births who had agreed to be recontacted. Of those, 59 were returned providing a response rate of 48.8%.

Infant age

The aim was to ask mothers to complete surveys when their infant was around:

• 9 to 12 weeks old (phase 1)
• 4 to 6 months old (phase 2)
• 8 to 10 months old (phase 3)

Table 6.5 shows the age range of infants whose mothers completed the survey at each phase.

As a result of the sample being taken from a month of births, and the length of the fieldwork periods, the infant age range at each phase was broader than expected.

There is also some overlap in the ages at phase 1 and phase 2. This was because the gap between fieldwork periods for the pilot and main surveys was narrow, and some mothers took longer to respond.

Table 6.5: infant age by survey phase

Phase	Age range (weeks)	Age range (months) (see note)	Mean age	Median age
Phase 1	7 to 20 weeks	2 to 5 months	2.6 months	2.7 months
Phase 2	15 to 30 weeks	4 to 7 months	4.4 months	4.7 months
Phase 3	33 to 44 weeks	8 to 10 months	8.4 months	8.5 months

Note: this is based on one month being 4.35 weeks. Figures are rounded up.

Online and paper responses

Table 6.6 shows the proportion of online and paper responses at each phase. More than 90% of surveys were completed online. Of these, 96% were completed on a smartphone.

Table 6.6: response rate by online and paper completion

Response type	Phase 1	Phase 2	Phase 3
Online	90.8%	93.8%	92.1%
Paper	9.2%	6.2%	7.9%

Survey completion by language

To ensure the survey was as accessible as possible, participants were offered the option of completing the survey in 5 different languages, which were:

• English
• Arabic
• Portuguese
• Spanish
• Polish

These languages were chosen based on the top 5 languages that the CQC Maternity Survey 2022 was completed in. Participants were informed of the languages offered in the invitation letter and were able to select the preferred language on the sign-in page to the online survey.

Table 6.7 provides details of the number of mothers that completed the online survey in each language at each phase.

Table 6.7: number of mothers that answered the online survey in each language per phase

Language	Phase 1	Phase 2	Phase 3
Arabic	74	29	12
English	9,031	4,671	3,213
Spanish	34	12	9
Polish	44	17	11
Portuguese	50	23	17

Free text responses to open ended questions were translated into English before analysis.

7. Data processing

Paper questionnaire processing

Paper questionnaires were returned in supplied freepost business reply envelopes (second class) to the scanning house. Once received, envelopes were machine opened, and questionnaires collated, guillotined and prepared for scanning.

Paper questionnaires were then processed using unique survey codes allocated to each case in the sample. All marks on the forms were recognised at this stage, regardless of whether they followed the questionnaire instructions.

At phase 2, where the multiple births questionnaire was provided, 5 mothers responded to that separate questionnaire and not to the main phase 2 questionnaire. These responses were not included.

Data checks and edits

All completed online responses where the respondent had clicked on the final ‘submit’ button were eligible for inclusion. All paper questionnaires received with identifiable unique survey codes were also eligible for inclusion.

The response rates were based on all completed, valid questionnaires returned and all questionnaires sent. They were not adjusted to exclude questionnaires that did not reach the sampled individual, for example where envelopes had been returned undelivered.

The following questionnaires were excluded from the data:

• all questionnaires where there was only data for the first page of the paper questionnaire
• all questionnaires where only the demographic questions had been completed (questionnaires where the front page and the demographic questions had been completed were not excluded)
• duplicates were removed - the most complete version of the survey had priority (see the ‘Duplicates’ section for more information)

Inclusion of pilot data

The questionnaires, survey materials and methodology performed well in the pilot surveys, and only minimal changes were made between the pilot and main survey at each phase.

Analysis to look at the differences in results between the pilot and main survey across a range of survey estimates and demographic profiles found that there would be no effect on the overall results from including the pilot data. So, the pilot and main survey data sets were combined for reporting. This allowed us to use all the available data and provide large enough sample sizes for analysis at subgroup level.

Responses from the pilot equate to 9% of the total number of returned questionnaires at phase 1, 9% at phase 2 and 10% at phase 3.

Duplicates

Questionnaire data was combined from online and scanned data sources. Where duplicates existed, the response used is the case that is the most complete (that is, with the fewest unanswered questions). If there was no difference in completeness, online data had precedence.

Free text coding

At each phase, survey questionnaires contained several opportunities for respondents to write in their responses, through fully open-ended questions or ‘other (please specify)’ answer options.

These responses were reviewed by Ipsos researchers and grouped based on themes appearing in the responses. Where a respondent’s answer matched an answer option that appeared in the answer response option list, it was ‘back coded’ to that specific answer option.

At the end of each questionnaire, mothers could add their final thoughts on their experience of feeding their infant. This open text box allowed respondents to comment on any aspect of feeding not covered in the questionnaire. These comments were treated in the same way as other free text boxes, by grouping similar comments into themes.

These responses were included in the reporting where relevant.

Editing the data

Completed paper questionnaires sometimes contained completion errors. For example, a respondent:

• ticked more than one box when only one response was required
• answered a question that was not relevant to them
• missed questions out altogether

So, it was necessary to undertake a certain amount of editing to ensure the data was logical. For example, a response was excluded if:

• a respondent ticked more than one box where only one answer was required
• a respondent selected 2 conflicting answers on a question that allowed multiple responses
• all boxes were left blank
• a respondent did not answer a filter question - all subsequent questions relating to that filter question were also excluded

Response errors

In addition to the editing rules set out above, Ipsos conducted further checks to ensure that outlying or contradicting responses were removed from the data. This largely affects the paper responses, as conditions were established on the online survey to avoid this happening.

If a respondent entered an age for their baby (for example, an age when they first received infant formula) that was higher than their current age, their response was removed.

If a respondent entered an age for their baby that was older than they could have been for that phase, limitations were put in place.

Weighting

Weights were generated to correct for the sampling design and to reduce the impact of non-response bias. Different weights were created for each phase of the data, to account for differential likelihood of response by phase. Weights were calculated and applied for the pilot and main surveys combined.

Phase 1 weighting

The phase 1 weights were designed to match the population of mothers who gave birth in the specified sampling period by maternal age, maternal ethnicity, IMD quintile and region. This accounted for the differential likelihood of selection, based on the oversampling of mother from ethnic minority groups, as well as the differential non-response, where some groups were more likely to take part than others. The weight was then scaled to the number of responses received, so the total unweighted base and total weighted base match.

Phase 2 and phase 3 weighting

The phase 2 and phase 3 weights were calculated based on a non-response multivariate regression model, which calculated probability of response by maternal age, maternal ethnicity, IMD quintile, region, reported feeding status at 6 weeks (whether mothers were giving breast milk only, infant formula only, or breast milk and infant formula), multiple births, and maternal education, weighted by each participant’s phase 1 weight. The output variable for each model was the likelihood of responding to each phase, compared with the responding profile at phase 1.

The weights were calculated as the inverse probability of likelihood of responding (that is, 1 divided by the likelihood of responding). This means that those who were less likely to respond, according to the model, received a larger weight, and those who the model identified as more likely to respond received a smaller weight. To avoid extreme weights, the weights were capped for the largest 1% of values. The weights were then combined with the phase 1 weight for each participant, rescaled to the number of responses received at each phase.

Weighting efficiency

Weighting efficiency is a statistical metric that measures the amount of data distortion required to weight a sample. A higher percentage (closer to 100%) means less manipulation and more reliable statistical results.

The design effects, effective base sizes and weighting efficiencies for the pilot and main surveys combined are set out in table 7.1. The design effect shows the increase in variance resulting from the weighting approach. The weighting efficiency indicates how much the data has been manipulated and the resulting effective base size.

Table 7.1: weighting efficiency

Phase	Design effect	Unweighted base	Effective base size	Weighting efficiency
Phase 1 weight	1.09	10,168	9,334	91.8%
Phase 2 weight	1.25	5,064	4,065	80.3%
Phase 3 weight	1.42	3,543	2,489	70.3%

8. Analysis and reporting

Analysis

Analyses were undertaken by Ipsos and OHID using data tables to identify significant differences between groups. Additional derived variable analysis was performed in SPSS (statistical analysis software).

Weighted percentage estimates are presented in the survey report and data tables, showing how respondents answered each question.

Typically, all response options are included in the calculation of a question result. However, for some questions, certain response options are excluded from the result calculation where appropriate (for example, ‘Don’t know or can’t remember’).

In some instances, summary results are presented. These are a single statistic that provides a quick way of viewing the result for a question. This is usually an aggregation of 2 individual responses (for example, ‘%Yes’ is a combination of ‘%Yes, vitamin D only’ and ‘%Yes, vitamin D and other vitamins’).

Specific differences between subgroups mentioned in the report text are statistically significant at the 95% confidence level. This means we are 95% confident that there is a true difference.

Data tables showing the weighted percentages for each survey question are published alongside the report. These tables also show results by subgroup and show statistically significant differences between groups.

No statistical significance testing was carried out on differences between responses at different phases of the survey.

Subgroup reporting

Data is reported by the following subgroups.

Age of mother

Mothers were asked their year of birth at phase 1 of the survey and the results are grouped into 5 age bands:

• 16 to 19 years
• 20 to 24 years
• 25 to 29 years
• 30 to 34 years
• 35 years and over

Ethnicity

Ethnicity is based on self-reported ethnicity from participants at phase 1 and categorised into 5 overarching groups. See annex 2 for full details.

Deprivation

Deprivation is based on the IMD quintile (English indices of deprivation 2019) of the address where the invitation letter was sent. Reporting focuses on the observed differences for mothers and infants living in the most and least deprived areas.

First child

At phase 1 of the survey, mothers were asked if this was their first baby. Those that selected ‘yes’ have been reported as first-time mothers, whereas those that selected ‘no’ have been reported as mothers that have other children.

Bases

The base is used in tables and figures to show how many people answered each question. Not all questions are asked of, or answered by, everyone. For example, some questions only apply to respondents depending on their answers to a previous question.

Where results are presented in tables or figures within the report, unweighted bases are shown. Data tables show weighted and unweighted bases.

Where base sizes are small (under 50 responses), specifically at phase 3 of the survey, results should be treated with caution. Results for some groups may not be included in the report due to small numbers. For example, these groups include mothers aged 16 to 19 years and mothers from Mixed or Multiple ethnic groups.

The effective base is the unweighted base adjusted to take into account the design effect resulting from the weighting. This is used for statistical significance testing because it is designed to reduce the likelihood that weighting adjustments produce ‘false positive’ significant results.

The effective base can also show whether weighting is inflating the answers from a particular group by a large factor. It is calculated by dividing the squared sum of weights for all the respondents in the weighting matrix table by the sum of the squared weights.

9. Comparing IFS 2024 with previous surveys

The IFS 2024 has undergone significant changes compared with the 2010 survey. Changes to the content of the questionnaires have been outlined in earlier sections (see ‘Finalising questionnaire and materials’ in section 2 and ‘Changes made as a result of the pilot’ in section 4). Changes to sampling, methodology and weighting are summarised below.

Sampling changes

For IFS 2024, sample data was drawn from MSDS. This is a different approach to the last IFS in 2010 when the sample was drawn from birth registration records (MSDS data was not available for sampling at this time). This change in sample source has implications for the age of infants surveyed. In 2024, their ages were slightly older in the first phase of the survey compared with 2010, as NHS trusts have an 8-week period to submit their birth data.

Methodological changes

The 2010 infant feeding survey was mainly a paper survey, with the option to complete it online. For IFS 2024, the ‘push-to-web methodology’ encouraged mothers to complete the survey online, and a paper survey was available only after 2 reminders had been sent.

Weighting approach

The weights for IFS 2024 were designed to be as consistent as possible with the previous IFS, while reflecting the changes to the survey design and to statistical learning since the last iteration of the survey in 2010.

The main differences between the 2010 and 2024 approaches to weighting were as follows.

In 2024, we oversampled mothers from ethnic minority groups, so the weights needed to adjust for the effect of this on an individual’s probability of selection.

In the 2010 survey, the oldest age group was ‘35 years and over’, and the youngest age groups were ‘under 20’ and ‘20 to 24 years’. In 2024, due to the changing age profile of mothers, it was decided for weighting purposes to split the oldest age group into ‘35 to 39 years’ and ‘40 years and over’. For the phase 2 and 3 weights, the youngest 2 groups were combined into an ‘under 25’ group (it was still possible to differentiate these groups for the phase 1 weight).

In the 2010 survey, the National Statistics Socio-economic classification was collected at phase 1, and then used in the weights at phase 2 and phase 3. However, since it was decided not to collect this data in the 2024 survey, it was also decided to use ‘highest qualification’ as collected at phase 1 of the survey (in combination with IMD of the address, which was already a factor in the weighting) in the weighting, as well as for analysis. It was also agreed that some other variables used in the weighting for the 2010 survey (such as drinking and smoking behaviour) were controlled for by other weighting variables, and were asked in a way that made their use for weighting more complex. So, these were not included in the 2024 weights.

In the 2010 model, the phase 2 and phase 3 weights were designed using a chi-squared automatic interaction detection (CHAID) model to adjust for intersectional non-response. Based on advances in statistical approaches to weighting, particularly related to the instability of CHAID models compared with regression models, it was agreed that the 2024 survey would use a regression model for these weights.

As the survey in 2010 was UK-wide, multiple weights were produced to allow for analysis for the 4 countries separately, England and Wales in combination, and for the UK as a whole. As the 2024 survey was England-only, only one set of these weights was required for each phase.

Comparisons with previous infant feeding surveys

For England-only estimates, differences in weighting between surveys should not affect comparability. However, the results from IFS 2024 are not directly comparable with previous surveys due to:

• differences in the survey design
• a different approach to sampling
• changes to the questionnaires

So comparisons have not been made in the report with previous reports.

10. Survey limitations

The figures in this publication come from a survey, which gathers information from a sample rather than from the whole population. Results from sample surveys are always estimates, not precise figures.

Behaviours are self-reported by mothers at specific time points in their baby’s life. Actions were taken to minimise the effect of social desirability (a response bias where participants over-report ‘good’ behaviours and under-report ‘bad’ ones to conform to social norms). For instance, making clear that results were reported anonymously. Actions were also taken to increase accuracy, for example by specifying time frames. However, there is likely to be some difference between self-reported and actual behaviours.

Respondents were not always consistent when responding to similar types of questions. For example, in the first phase of the survey, there was more than one question about alcohol intake before and during pregnancy, and after birth. Responses to these questions did not always match.

The sampling design aimed to achieve a representative sample of mothers responding to the survey. The oversampling of mothers from ethnic minority groups and use of incentives generally achieved this and provided sufficient responses for subgroup analysis based on maternal age, ethnicity and IMD. Weighting was applied to correct for the sampling design and to reduce the effect of non-response bias, although there may still be bias for other reasons that it was not possible to account for.

The sampling design and weighting approach did not take feeding method into account for phase 1 of the survey, as there is no reliable way of achieving this. This means that there may be response bias, for example mothers who breastfeed may be more likely to respond to the survey at phase 1. Other published estimates suggest lower rates of breastfeeding than reported in this survey, although differences in data collection methods and definitions mean that figures cannot be directly compared. For example, Breastfeeding at 6 to 8 weeks after birth: annual data April 2023 to March 2024 reports the prevalence of breastfeeding in England at 52.7%, compared with the prevalence of breastfeeding at 6 weeks in this report at 71%.

Mothers who breastfeed were more likely to respond to phases 2 and 3 of the survey. The weighting applied to phase 2 and 3 data takes feeding status at 6 weeks as reported at phase 1 into account. However, there may still be bias where the results reported are based on responses from phases 2 and 3 of the survey. This should be taken into account when considering the survey findings.

Annex 1: population and sample profiles

Table A1: profile of population, starting and responding samples by IMD quintile, ethnicity, age and region

Subgroup	All births in England 2023 to 2024	Aug 2023 and Dec 2023 births	IFS 2024 starting sample	Phase 1 respondents	Phase 1 agreed to recontact	Phase 2 respondents	Phase 3 respondents
Total (number)	524,740	71,047	26,339	10,168	8,270	5,064	3,542
IMD 1 (most deprived)	26%	26%	29%	27%	26%	24%	22%
IMD 2	22%	22%	22%	23%	24%	24%	25%
IMD 3	19%	19%	18%	17%	17%	17%	17%
IMD 4	17%	17%	16%	16%	16%	17%	17%
IMD 5 (least deprived)	15%	15%	15%	16%	17%	18%	19%
Asian or Asian British	15%	15%	22%	17%	15%	13%	11%
Black or Black British	6%	6%	9%	11%	10%	9%	9%
Mixed	3%	3%	4%	3%	3%	3%	3%
White	68%	69%	57%	60%	63%	67%	70%
Other ethnic groups	4%	4%	6%	5%	5%	4%	4%
Not known or not stated	4%	4%	3%	3%	3%	3%	3%
16 to 19 years	3% (see note)	2% (see note)	2%	1%	1%	1%	0%
20 to 24 years	13%	12%	12%	8%	7%	6%	5%
25 to 29 years	28%	27%	27%	25%	24%	23%	21%
30 to 34 years	34%	34%	34%	38%	39%	40%	42%
35 to 39 years	18%	20%	20%	23%	24%	26%	27%
40 years and over	4%	5%	5%	5%	5%	5%	5%
East Midlands	Not available	8%	8%	8%	8%	8%	8%
East of England	Not available	12%	11%	12%	12%	12%	12%
London	Not available	14%	16%	13%	13%	12%	12%
North East	Not available	5%	5%	5%	6%	5%	5%
North West	Not available	15%	15%	14%	15%	15%	15%
South East	Not available	16%	15%	16%	17%	17%	17%
South West	Not available	8%	7%	8%	9%	10%	10%
West Midlands	Not available	12%	12%	12%	11%	11%	10%
Yorkshire and the Humber	Not available	11%	11%	11%	11%	11%	11%

Note: MSDS age group was ‘under 20’.

Annex 2: ethnicity categories

Ethnicity categories for sampling based on MSDS variables

The ethnicity categories for sampling based on MSDS variables are set out in the groups below.

Asian or Asian British

Categories in the Asian or Asian British group include:

• Indian
• Pakistani
• Bangladeshi
• any other Asian background

Black or Black British

Categories in the Black or Black British group include:

• Caribbean
• African
• any other Black background

Mixed

Categories in the Mixed group include:

• White and Black Caribbean
• White and Black African
• White and Asian

White

Categories in the White group include:

• British
• Irish
• any other White background

Other ethnic groups

Categories in the ‘Other ethnic groups’ group include:

• Chinese
• any other ethnic group

Not known or not stated

Categories in the ‘Not known or not stated’ group include:

• not stated
• not known

Ethnicity categories for reporting based on participant response at phase 1

The ethnicity categories for reporting based on participant response at phase 1 are as follows.

Asian or Asian British

Categories in the ‘Asian or Asian British’ group include:

• Bangladeshi
• Chinese
• Indian
• Pakistani
• any other Asian background

Black or African or Caribbean or Black British

Categories in the ‘Black or African or Caribbean or Black British’ group include:

• African
• Caribbean
• any other Black African or Caribbean background

Mixed or multiple ethnic groups

Categories in the ‘Mixed or multiple ethnic groups’ group include:

• White and Asian
• White and Black African
• White and Black Caribbean
• any other mixed background

White

Categories in the White group include:

• English, Welsh, Scottish, Northern Irish or British
• Irish
• Gypsy or Irish Traveller or Roma
• any other White background

Other ethnic group

Categories in the ‘Other ethnic group’ group include:

• Arab
• any other ethnic background
• would prefer not to say