Quality and accuracy of complementary feeding information on Instagram: A cross-sectional content analysis study.

In recent years, social media platforms have become dominant sources of public health information [ 1 ]. Instagram is a platform that provides a range of health-related content and includes advice on infant nutrition and feeding practices [ 2 – 4 ]. With its quick, digestible format, Instagram allows health messages to reach a wide audience; however, this open accessibility also poses challenges related to the accuracy and quality of posts due to personal biases, anecdotal experiences and commercial interests [ 5 ]. Parents and caregivers retrieve infant nutrition advice from a variety of sources such as family, friends and healthcare providers [ 6 – 8 ]. Social media is also an increasingly important resource for parents and caregivers to access infant feeding information [ 2 , 3 ]. A 2020 cross-sectional survey of parents of children aged 0–18 years in the US revealed that 96% of participating parents used social media, with 68% using these platforms for health information specifically [ 9 ]. The usage of Instagram is increasing and it is particularly prevalent among mothers and individuals under 40 [ 10 ], so can influence decision-making among young mothers in particular. Complementary feeding refers to the transition of infants from exclusive breastfeeding or formula feeding to the introduction of other foods [ 11 ]. Commencing intake of other foods is necessary to accommodate the infant’s changing requirements and physiological maturation [ 12 ]. It reduces the risk of growth faltering, malnutrition, developmental delays of feeding skills, and deficiencies of micronutrients such as iron and zinc. Complementary feeding methods can also influence lifelong flavor preferences and dietary habits that will impact long-term health [ 13 ], and is a key area of nutrition to address. The World Health Organization (WHO) provides specific recommendations on when and how to introduce complementary foods to ensure optimal growth and development. WHO recommends the introduction of complementary foods at around 6 months of age, alongside breastfeeding and a varied diet with three recommended food groups (1. animal source foods (meat, fish, eggs, dairy), 2. fruits and vegetables, and 3. pulses, nuts, and seeds) [ 12 ]. However, the extent of which publicly accessible content on Instagram aligns with these guidelines remains unknown. Previous studies have examined the quality and accuracy of general nutrition information on social media and have found the information to be inconsistent and unreliable [ 14 – 17 ]. Although data is available on general nutrition information, to date no studies have assessed the quality of complementary feeding information on social media and whether this information aligns with official recommendations. Engagement rates and use of social media for infant health advice have been investigated [ 2 , 3 , 18 – 22 ], but no studies have systematically analyzed the quality and accuracy of this content. The current study aims to evaluate the quality and accuracy of complementary feeding information disseminated on Instagram, using the PRHISM digital health tool and WHO guidelines. A secondary objective of this study was to determine whether the professional background of account holders was associated with the quality and accuracy of the information provided. The study provides important insights into the credibility of Instagram as a source of complementary feeding information.

Following screening, a total of 302 posts met the criteria for inclusion in the analysis ( Fig 1 ). Inter-coder reliability tests revealed a Krippendorff’s alpha value of 0.799, indicating a high level of agreement between coders and acceptable reliability for content analysis [ 38 ]. Over a third of the posts (n = 103, 34.1%) included information on when to introduce complementary feeding, while most of the posts (n = 268, 88.7%) mentioned types of complementary foods to introduce. The professional status of the majority of account holders (83.4%) could not be confirmed from the posts alone, however a small proportion identified as dietitians or doctors. The ‘Other’ subgroup (n = 3) was included in descriptive statistics; however, due to its small sample size, it was not included in subgroup statistical analyses. The ‘Other’ subgroup included n = 2 nutritionists and n = 1 parenting coach. a Posts not in English included Arabic, German, Portuguese, and Spanish b Posts excluded for not relating to complementary feeding included content on maternal nutrition, feeding beyond 23 months of age, or other infant care topics not specific to feeding c A sample of posts (n = 69) included recommendations for both timing of introduction and types of foods, and were accounted for in both categories During analysis, one of the 11 PRHISM principles was excluded from scoring (Principle 8 [Privacy]) due to the complete lack of applicable content across all posts. Quality scores were therefore calculated based on the remaining 10 principles. The majority of posts were classified as mediocre, with few achieving a ‘Good’ quality rating. No posts were identified as ‘Excellent’ quality ( Table 2 ). A Fisher’s Exact test revealed a statistically significant association between account type and PRHISM category χ 2 (6, N = 302) = 33.42, p  < 0.001. Dietitians were identified as the primary contributors to high-quality content, with the highest proportion of ‘Good’ posts. Other account types, particularly unverified accounts, presented less in the ‘Good’ category. *Account type based on self-identification of credentials When separated into the individual principles, out of a total score of 4, the lowest mean scores presented for complementary information, attribution, and associated risks and benefits. Three principles had a mean score of two or more (Action-orientated, Accessibility, Images) ( Table 3 ). *Principle 1 and 2 were excluded as they related to account bios which were not evaluated The overwhelming majority of posts recommended introducing complementary foods at 6 months of age, aligning with WHO guidelines ( Table 4 ). All of the posts from dietitians recommended introduction at 6 months of age. Deviations from the WHO guideline came primarily from unverified accounts and one doctor account ( Table 5 ) A Chi-Square test of independence revealed a statistically significant association between account type and the month recommended for introducing complementary foods (χ 2 (4, N = 103) = 17.84, p  = 0.001). Dietitians had the highest accuracy for timing of introduction relative to other account types. Over half of posts with information on foods to include mentioned all 3 food groups recommended by WHO. Most of the remaining posts mentioned 2 of the 3 groups ( Table 4 ). There was a significant association between account type and recommended food groups to introduce (χ 2 (6, N = 268) = 22.25, p  = 0.001). Dietitians most frequently mentioned all three WHO-recommended food groups ( Table 5 ). The most common food group to recommend was fruit and vegetables (96.6%), followed by animal products (85.4%), and then other plant sources (63.8%). The most frequently recommended individual foods included bananas (45.5%), avocados (42.1%), and apples (18.0%) for fruits; broccoli (37.3%), sweet potatoes (32.8%), and carrots (29.5%) for vegetables; eggs (51.5%), yoghurt (34.3%), and chicken (30.6%) for animal products; and nuts (34.7%), peas (28.4%), and seeds (27.6%) for other plant sources. Advice to avoid certain foods appeared infrequently across all account types, with avoidance of high-sugar foods, high-salt foods, and sugar-sweetened beverages mentioned in approximately 20% of all posts ( Table 4 ).

Complementary feeding information on Instagram is of variable accuracy and overall mediocre quality, and has notable deficiencies in referencing, recognition of personalized healthcare needs, and acknowledgement of associated risks and benefits. Improving content quality can be supported through public guidance on identifying credible sources, increased engagement from healthcare professionals, and platform-level screening for inaccurate information. This study contributes to the limited body of research exploring the role of social media as a resource for complementary feeding information. As social media increasingly shapes parental decision-making, ensuring the dissemination of evidence-based and trustworthy advice is essential to support optimal complementary feeding and improve infant health outcomes.

This was a cross-sectional content analysis study of Instagram posts related to complementary feeding, guided by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist [ 23 ]. This study was exempt from formal ethics review due to the public nature of the data. To maintain ethical standards, identifiable details of Instagram accounts were excluded from analysis and reporting. To be included, posts had to include information on when to introduce complementary feeding and/or the types of foods to introduce or avoid. The content must have been published within the last five years and written in English. The posts could not contain video or video-related content. Posts that did not meet the inclusion criteria were excluded from analysis. A new Instagram account was created on a desktop’s incognito mode to avoid algorithmic bias [ 15 , 16 ]. Hashtags related to complementary feeding were collected through an exploratory search using the search bar for the terms ‘complementary feeding’, ‘starting solids’, and ‘weaning’. Posts generated from these searches were reviewed, and any hashtags that appeared in the captions were documented. For each hashtag, the number of associated posts was recorded and used to rank them in order of popularity. Hashtag collection continued until no newly identified hashtags had a higher post count than those already listed [ 24 ]. The 13 most frequently used hashtags (each with >200,000 uses), such as #babyledweaning, #firstfoods, and #startingsolids, were selected to identify complementary feeding-related posts (full list of hashtags used available in S1 Table ). The top 50 posts for each hashtag were collected over a 5-day period between 9 th and 13 th of March 2025. This resulted in an initial dataset of 650 posts, consistent with that of similar social media content analysis studies [ 25 – 27 ]. Posts were manually screened for eligibility and duplicates based on matching captions and dates of publication. Content analysis was conducted on the images and captions of eligible posts. As has been done in previous studies [ 28 – 30 ], the content analysis focused solely on the information contained within the posts themselves, without examining account profiles or bios. However, any profession stated within the post content was recorded. The recommended age for introducing complementary foods and the types of foods to introduce or to avoid were documented as applicable ( Table 1 ). Any food mentioned that could be categorized into three WHO-recommended food groups (1. animal-source foods (meat, fish, eggs, dairy), 2. fruits and vegetables, and 3. pulses, nuts, and seeds) were recorded. Within each food group category, individual food items were documented; for instance, banana was placed under ‘fruit’ and yoghurt under ‘animal-source foods.’ Any commentary on food categories to avoid when complementary feeding was documented, including foods high in salt and sugar, and sugar sweetened beverages. * WHO= World Health Organization The professional background of account holders (dietitian, doctor, other, or unable to confirm) was noted. In this context, ‘account holders’ refers to the individuals who posted complementary feeding content on Instagram. The terms ‘dietitian’ and ‘doctor’ were recorded as those who self-identified as dietitians or medical doctors; however, professional credentials could not be independently verified due to the limitations of confirming qualifications on social media [ 31 ]. ‘Unable to confirm’ referred to cases where the account holder’s professional background could not be identified from the post. Accounts that did not fall into these categories were classified as ‘other’, such as individuals in non-healthcare professions. Quality was assessed using a modified version of the Principles for Health Information on Social Media (PRHISM) tool, which was a structured framework designed to evaluate the quality of public health information on social media [ 32 ]. Unlike traditional website-focused tools, PRHISM addresses the unique characteristics of social media content [ 33 , 34 ]. Developed via a modified Delphi process [ 35 ], the tool outlines 13 quality principles relating to aspects such as clarity, accessibility, and the provision of evidence-based information. For this study, Principle 1 (Authorship) and Principle 2 (Authoritativeness) were excluded, as they pertain to information in account bios, which were not evaluated. The current study focused solely on assessing the quality of content within the posts themselves. The remaining 11 principles were assessed: Action-Oriented Content, Financial Disclosure, Attribution, Balance & Justifiability, Risks & Benefits, Privacy, Complementary Information, Referrals & Support, Readability & Comprehensibility, Accessibility, and Use of Images . A comprehensive breakdown of the PRHISM tool and its principles are described elsewhere ( S2 Table ) [ 15 , 32 ]. Each principle was scored on a 5-point Likert scale, based on how well the content aligned with the principle. The overall PRHISM score was calculated as a percentage using total score/possible score x 100. Scores range from 0 to 100, with higher scores indicating higher content quality. Quality ratings were categorized into four levels: scores from 0 to 25 as poor, 26–50 as mediocre, 51–75 as good, and 76–100 as excellent [ 15 ]. Accuracy was evaluated based on alignment with WHO-recommended food groups ( Table 1 ). Yes/no labels were input based on whether a food group was mentioned and/or the timing mentioned was correct or not. Total PRHISM scores were used to identify the overall quality category. A structured coding framework, with a set of coding rules was created to improve consistency and minimize conflicts [ 36 ]. To assess inter-coder reliability, a sample of every 10 th post (10% of posts) was independently coded by a second researcher as has been utilized in other content analysis studies [ 37 ]. Krippendorff’s alpha was used to determine the level of agreement between coders. Its values range from 0 to 1, with 1 indicating perfect reliability and 0 suggesting the absence of reliability [ 38 ]. Disagreements were resolved through discussion until consensus was reached. Based on the patterns of disagreement and the consensus, the remaining 90% of posts were updated to reflect the new understanding through the consensus building process. The dataset was exported from Google sheets to Microsoft Excel for preliminary data cleaning, before being imported into IBM SPSS Statistics (Version 29.0) (SPSS, IBM, USA) for statistical analysis. Within SPSS, a combination of descriptive and inferential statistical analyses was completed. Descriptive statistics, such as frequencies and percentages, were used to describe categorical data, such as PRHISM scores and alignment with WHO accuracy criteria. Mean ± standard deviations [SDs] were reported for individual PRHISM principles, to provide insight into content strengths and weaknesses. Chi-Square tests of independence and Fisher’s Exact Tests were conducted to assess associations, depending on expected cell sizes. Statistical significance was defined as p value <0.05.

(TIF) (TIF)