Exploring Infant Fall Events Using Online Parenting Discussion Forums

Background: Falls are the most common mechanism of hospitalised injury in children under 12 months and commonly result in traumatic brain injuries. Epidemiological studies exploring infant falls demonstrate the burden but lack the contextual information vital to the development of preventive interventions. The objective of this study was to examine contextual information for falls involving children under 12 months. Methods: Online parenting forums provide an unobtrusive rich data source for collecting detailed information about fall events. Relevant discussions related to fall incidents were identied and downloaded using site-specic google search queries and a programming script. A qualitative descriptive approach was used to analyse the incidents and categorise contextual information into ‘pre-cursor events’ and ‘inuencing factors’ for infant falls. Results: 461 infant fall incidents were identied. Common fall mechanisms included: falls from furniture; falls when being carried or supported by someone; falls from baby products and falls on the same level. Across the spectrum of fall mechanisms, common precursor events were: infant rolling off; being alone on furniture; product misuse; caretaker falling asleep while holding the infant; and caretaker tripping/slipping while carrying the infant. Common inuencing factors included: infant’s rapid motor development; lapses in caretaker attention; and trip hazards. Conclusions: The ndings dene targets for intervention to prevent infant falls and suggest the most viable intervention approach may be to target parental behaviour change. Online forums have provided rich information critical for preventive interventions aimed at changing behaviour.


Introduction
Injury is a widespread and longstanding public health problem (1). Globally, injury is a leading cause of child death and hospitalization. (2) In Australia, children ≤ 1 year have the highest death rates due to injury among all children and an injury hospitalisation rate of 799/100,000.(3) Falls are the most common injury mechanism for this age group accounting for almost 50% of all injury hospitalisations. (3) The head is the most commonly injured body region (4) and often leads to Traumatic Brain Injuries (TBI).
(5) Similar incidence and injury patterns occur in North America and Europe. (6)(7)(8) TBI in early childhood is associated with negative behavioural and cognitive outcomes.(9) While a number of interventions are effective in minimising fall risk in older children (10,11) there is a paucity of evidence for effective countermeasures for falls in children ≤ 1 year.
Epidemiologic studies examining infant falls usually rely on administrative data or medical records. (4,12,13) These provide detail on burden and demographic risk factors but generally have limited or incomplete contextual information. This is a barrier for effective intervention development. (14) The best source of detailed contextual information about infant falls is from someone who witnessed the fall. However, one-on-one discussion and large sample sizes can be time and resource intensive. In other areas, the Internet and social media have been successfully used to collect data from people participating in online forum discussions. (15) These also provide naturalistic data as discussions occur without researcher involvement. (16) We aim to use online parenting discussion forums to unobtrusively and cost-effectively access contextual information about infant falls to identify speci c modi able factors to prevent infant falls.

Method
This was an infodemiological study (17,18) using online forum data following a qualitative descriptive approach (19) with the objective of providing comprehensive summaries of infant fall events. (20) The data source was social media forums within an online parenting website. This website was established in 1999 and is owned and operated by a large Australian media company. The website provides parenting information in the form of media articles and forums across a broad range of non-injury/prevention related child care topics. After obtaining approval from the website owner, and ethics approval from the Human Research Ethics Committee (HC180295), site speci c google search queries were chosen to identify URLs potentially containing discussions related to infant falls. These were ; 'baby fall', 'baby falling', 'baby fell', 'baby dropped' and 'baby hurt'. This approach allowed us to search all forums on this website without placing too much burden on the website. A researcher manually screened the resulting URL list and compiled a list of possible URLs containing infant fall discussions. The discussions from the selected list of web pages were downloaded using a program script written in Python, and any potentially identi able data were removed. The search was completed on 22nd of June, 2019 and included discussion threads ranging from 22nd November,2003 to 05th June 2016.
De-identi ed data were coded using QSR Nvivo12 software to identify relevant incidents. Relevant incidents were those relating to falls or near falls involving children ≤ 1 year, with age identi ed from wording in the post; speci cally mentioned age as ≤ 1 year, or posting was from a forum speci c to children aged 0-6 months or 6-12 months, or fall incident was mentioned in response to other incidents describing infant falls where age 0-12 months was mentioned, or words "newborn, tiny baby".

Analysis
Fall mechanisms were categorised, described and mapped to ICD-10-AM codes (Table 1). Two researchers (NC and SS) independently coded the data using fall mechanism categories (Table 1) and any differences were discussed until agreement was reached. In recognition that fall circumstances are often multilayered we separated likely causative factors leading to the fall into a 'precursor event' and other 'in uencing factor'. A 'precursor event' was de ned as the event/state immediately before the fall according to the literal meaning in the discussion. An 'in uencing factor' was de ned as a factor that impacted the 'precursor event' and therefore the occurrence of the fall.
One researcher (NC) compiled a list of factors categorised as a 'precursor event' or 'in uencing factor' and a second (SS) independently re-coded the data using this list, adding new factors as necessary. The two researchers (NC and SS) compared analyses and differences were discussed until agreement was reached. To address potential coder biases and pre-conceptions, coding for each fall mechanism by each researcher was undertaken separately and consensus was reached before moving onto coding for the next mechanism.
NC (a PhD student) and SS (an undergraduate medical student) were supervised by senior co-authors experienced in injury and qualitative research methods. Figure 1 summarises the data capture process. Overall, 461 infant fall incidents were identi ed. The most common fall mechanisms were a fall from household furniture (270, 58.6%), followed by falls when being carried or supported by someone (92, 20%) and falls from baby products (55,11.9%). Other mechanisms were a fall on the same level (28, 6.1%). fall on/from stairs (6, 1.3%), falls from playmats (4, 0.9%), playground equipment (3, 0.7%) and shopping carts (3, 0.7%). "I left her in the middle of my queen bed while I did some vacuuming. As I got closer to my bedroom … I could hear her screaming like she had never screamed before. I ran into the room and she was on the oor!... "

Results
The caretaker falling asleep with the baby was another common precursor event, identi ed 18 times (6.7%).
"… I was breastfeeding him in bed and fell asleep with him on the outside. I woke up when I heard a thud and DS cry." It was clear from some discussions that the precursor event of falling asleep was often unintentional (10, 55.6%), while in others (6, 33.3%) it was intentional or intention was unclear (2, 11.1%).
For falls from change tables, a common precursor event was the caretaker reaching for something while nappy changing, identi ed 11 times within the 64 change table related incidents (17.2%).
"I was changing him on the change table and all I did was slip one hand down to put the dirty nappy in the nappy bag and ds launched himself off the table and landed on the oor…" Unexpected or rapid changes in motor development was the most common in uencing factor for furniture falls. This was identi ed 29 (10.7%) times.
"…when he had started to move -I underestimated how quick he was. I used to put him on our bed every morning while I got dressed. One day I turned my back for a second and in that time he pulled himself to the edge then did a somersault off the bed!" Lapses in caretaker attention was the next most common in uencing factor for this fall mechanism. This was identi ed 20 times (7.4%).
"… honestly it can happen in the blink of an eye. Similar to your DH I looked away from the table, and over he went. It was so quick" Falls when carried or supported by someone This was the second most common fall mechanism in the discussions (n=92). These commonly occurred when the child was carried or supported by the mother (39, 42.4%), or an unidenti ed parent (28,30.4%); when the child was carried or supported by an adult caretaker other than a parent (19,20.7%); and less commonly, when carried or supported by an older child (6, 6.5%).
"I dropped my ds he was about 10 months triped up the back step he screamed has a giant bump … " Another common precursor event for these falls was the person falling asleep while holding the infant (mentioned 15 times/ 92 incidents, 16.3%) and often involved the child's mother falling asleep while feeding (12/15 incidents, 80.0%).
"…I was totally sleep deprived. Sat down on the couch to nurse her, dozed off with her snuggled low in my arms (basically in my lap) our dog barked and I startled awake -DD rolled down my legs and into the coffee table." A tired caretaker is also a likely in uencing factor; however, this was only overtly discussed a few times (2/15). Other commonly discussed in uencing factors were inadequate holding of the child (13 times/92 incidents, 14.1%) and sudden unexpected movements of the child (9/92 incidents, 9.8%).
Improper use was the most common precursor event for these falls. Non-use or misuse of safety straps was common for many baby products (particularly for strollers/prams, baby bouncers, high chair/baby chair, child car restraints, baby capsules/carriers). This was identi ed 30 times (55%, 15 not using safety strap and 15 cases of apparent improper use of straps) "Mother of the Year here took a few months to really internalise the 'strap them in' message and DD1 bounced herself face rst out of the bouncer at about three months old" Some other critical misuses identi ed were: cot base in high position (5, 9.1%), unbalancing the stroller (3, 5.5%), not using strollers' brakes (2, 3.6%) and carrying the infant while in bouncer/portable baby bed (2, 3.6%). Falls from cots were in uenced by rapid motor development.
"…like he was balancing on the cot railing with his feet off the mattress suspended in mid air by piece of wood-… " Falls on the same level Four different types of falls on the same level were mentioned. Most common was a fall while the infant was standing (20, 71.4%). Falls while the infant was sitting, falls while the infant was crawling and other falls due to being pushed by another person each had < ve identi ed incidents.
The common in uencing factor for this fall mechanism was the child's under-developed motor skills (13, 46.4%).
"…Now that both are easily pulling themselves up against furniture to stand, they are doing it every chance they get. The only problem is once they get up they don't know how to get down or lose concentration, let go and fall...a lot of the time hitting their heads on the tiles." Fall on or from stairs Falls on or from stairs were relatively uncommon (5 incidents). Two in uencing factors for these were lapses in caretaker attention (3, 60%) and unexpected/rapid infant motor development (2, 40%).
"DS1 fell down the stairs -all 8 of them -when he was 4 months. He was lying at one end of the room, well away from the stairs. I put a book on the shelf and when I turned back he'd rolled across the room and I was just in time to see him disappear, screaming, down the stair well"

Other fall mechanisms
Other mechanisms identi ed from the discussions included falls from playground equipment or playmats and falls from shopping carts (10 incidents). Precursor events relating to falls from shopping carts was non-use of straps (3, 30%).
"…didn't bother to buckle him in. I was squatting down looking at something when I heard a horrible splat sound, he had fallen face rst onto the cement oor …" There were no detailed discussions to identify causal factors for falls from playground equipment and playmats. Also, there were no discussions of falls between levels e.g. from windows Table 2 summarises pre cursor events and in uencing factors for different fall mechanisms.

Discussion
Using a novel qualitative infodemiological approach, we identi ed contexts requiring intervention to prevent the majority of falls in children ≤ 1 year. These are leaving children alone on furniture; misuse of change tables and baby products such as strollers, baby carriers and baby chairs; slips and trips and falling asleep while holding an infant. Furthermore, the richness of our dataset allowed us to link speci c in uencing factors to speci c precursor events for these fall types to identify modi able factors to prevent the fall. These include awareness of unexpected or rapid changes in infant motor development; lapses in caretaker attention; importance of adequately holding the infant; and reducing hazards in the home environment.
Our ndings align well with previously reported studies using administrative datasets and medical record reviews. (4,7,21,22) While some factors we have identi ed have been noted previously, (8,(23)(24)(25) this is the rst to provide this level of detail and identify targets for intervention across the spectrum of fall mechanisms among infants of this age.
Contextual information like that identi ed in this work also provides evidence for and adds to studies that have previously suggested using age appropriate injury prevention education for caregivers and home safety assessment programs. (7,23) However currently there is no evidence for any effective interventions speci cally targeting falls in this age group. (26) Given the magnitude and potential impact of this problem (5,8) there is an urgent need to ll this gap and identify effective targeted interventions. The outcomes of this current work identify modi able factors to be targeted in these interventions.
While our ndings demonstrate that no single intervention would prevent all falls, there is a common need for parent/caretaker behavioural change across many of the fall mechanisms. It therefore appears that a behaviour change or active approach rather than purely structural change (passive approach)(27) may be effective to prevent infant falls. For example, having the safety harness in a change table is structural whereas the parent using it appropriately is behavioural. But behaviour change is complex and educational interventions alone usually do not enact behaviours. (28) Behaviour change intervention are more likely to be successful when based on behaviour theory. (29) One challenge to developing effective behaviour change interventions is that they require detailed understanding of the problem and target behaviours. (30) This study lls this gap.
In this study we did not attempt to examine data by age of infant but it is clear from our earlier work(4) that risk of fall by different mechanisms changes as children move through development stages in the rst year of life. Different behaviour change intervention is likely needed at different times through this year and this needs to align with the infant's developmental stage.(6) For example it is possible the risk of a mother falling asleep while feeding is higher in early infancy and the risk of rolling off the bed is higher when the infant is gaining motor skills. Intervening at a single time point may also not be as effective as a targeted strategy to deliver behavioural change interventions at different time points over time.
Falls among infants on the same level were discussed relatively less commonly by parents in the forums than other falls, yet it is likely that these occur very commonly. As noted by Adolph and Berger (31) falling is a common by-product of children learning to walk with children at this stage of development falling within the vicinity of 17 times an hour and 100 times a day. As we previously observed (4) these types of falls very rarely occur among infants hospitalised from a fall (< 2% of all patients) and therefore this lack of severity might underpin the lack of discussion in the forums. As these falls occur while children are developing an important motor skill, it would not be appropriate to try to prevent the activities leading to these. Instead, injury risk might best be reduced by attention to the environment in which children are placed during this stage of development.
Another aspect warranting further environmental examination is the adequacy of both design and instructions of common baby products used by parents of infants. Improper use was the most common pre-cursor event for falls involving child products. In other areas, e.g. child car seats, it is becoming increasingly clear that correct use requires attention to how usage information is communicated, and the inherent design of the product in addition to the general behaviour of the user. (32) Extension of this approach to all baby products may be useful.

Limitations
As this is a qualitative study based on ad hoc reporting of fall types, the frequency of different fall types reported might not re ect true frequencies. However common fall mechanisms aligned with other epidemiological and medical record reviews. (4,7,22) Data used were from a convenience sample of online forum discussion participants and sample characteristics are unknown. Therefore, this may introduce some unknown bias and ndings may not be generalizable to the whole population. Another limitation was use of a single search engine and different search engines may provide different results sets. Also, this kind of study collects data from a static point in time and precludes active engagement with caregivers and ability to clarify or to get additional details from parents compared to other qualitative approaches. However, the unobtrusive nature of this data collection method may be a strength as it provides data extracted from naturalistic parental discussion.

Conclusion
This study used infant fall incidents from online parenting forums to identify precursor events and in uencing factors leading to different fall types among infants ≤ 1 year. This information is paramount to development of preventive interventions, particularly given the ndings suggest interventions need to target parental behaviour.

Declarations
Ethics approval and consent to participate Ethics approval from the Human Research Ethics Committee (HC180295), consent to participate was not applicable since the data was from a public online forum.

Consent for publication
Each of the authors con rms that this manuscript has not been previously published and is not currently under consideration by any other journal. Additionally, all of the authors have approved the contents of this paper and have agreed to the BMC Public Health Journal policies.

Availability of data and materials
The data that support the ndings of this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical restrictions.

Competing interests
The named authors have no con ict of interest, nancial or otherwise.

Funding
This research received no speci c grant from any funding agency in the public, commercial or not-forpro t sectors.

Authors' contributions
Nipuna Cooray conceptualized and designed the study, did the site-speci c searches, developed the web scripting script, downloaded and cleaned the data, carried out the analysis, drafted the initial manuscript, reviewed and revised the manuscript.
Si Sun analysed the data, drafted the initial manuscript, reviewed and revised the manuscript. Susan Adams, Lisa Keay and Natasha Nassar conceptualized and designed the study, coordinated and supervised data collection, and critically reviewed the manuscript for important intellectual content. Julie Brown conceptualized and designed the study, coordinated and supervised data collection and data analysis, drafted the initial manuscript, reviewed and revised the manuscript and critically reviewed the manuscript for important intellectual content. All authors approved the nal manuscript as submitted and agree to be accountable for all aspects of the work.