Circumstances and factors of sleep-related sudden infancy deaths in Japan

Background Sudden unexpected death in infancy (SUDI) comprises both natural and unnatural causes of death. However, few epidemiological surveys have investigated SUDI in Japan. Objective This retrospective study was conducted to investigate the latest trends of circumstances and risk factors of SUDI cases in which collapse occurred during sleep. Methods Forensic pathology sections from eight universities participated in the selection of subjects from 2013 to 2018. Data obtained from the checklist form were analyzed based on information at postmortem. Results There were 259 SUDI cases consisting of 145 male infants and 114 female infants with a mean birth weight of 2888 ± 553 and 2750 ± 370 g, respectively. Deaths most frequently occurred among infants at 1 month of age (18%). According to population data as the control, the odds ratio (95% confidence interval) of mother’s age ≤19 years was 11.1 (6.9–17.7) compared with ages 30–39. The odds ratio for the fourth- and later born infants was 5.2 (3.4–7.9) compared with the frequency of first-born infants. The most frequent time of day for discovery was between 7 and 8 o’clock, and the time difference from the last seen alive was a mean of 4.1 h. Co-sleeping was recorded for 61%, and the prone position was found for 40% of cases at discovery. Mother’s smoking habit exhibited an odds ratio of 4.5 (2.9–5.8). Conclusion This study confirmed the trends that have been observed for sudden infant death syndrome; particularly, very high odds ratios were evident for teenage mothers and later birth order in comparison with those in other developed countries. Neglect was suspected in some cases of the prolonged time to discovery of unreactive infants. To our knowledge, this is the first report of an extensive survey of SUDI during sleep in Japan.


Yu Kakimoto
Response to Reviewers: The detailed review of our article is appreciated. The comments by the reviewer have been helpful in allowing us to revise the manuscript. The authors have attempted to address the questions raised as separate pages. According to the raised comments, the manuscript has been rewritten extensively to a revised version. Alterations are indicated as track changes in the revised manuscript.  Concerning the legal aspect, we have Act on the Protection of Personal Information.
But any personal information at postmortem is out of regulation of this law. The ethical committee of the Institutional Review Board for Clinical Research, Tokai University, approved this survey with a referential number of 16R-172, and permitted to gather the anonymized data set to one primary investigator without informed consent, as long as research purpose. But the committee required us to show a sample data set, and to disclose the project on the website as an optout manner (http://forensic.med.utokai.ac.jp/ethicscommittee/images/ethicscommittee_02.pdf, in Japanese). This study was also approved by the respective ethical committees of the faculties as a collaborative study. On behalf to reflect editor's concern, statements were also added into the revised manuscript. Describe where the data may be found in full sentences. If you are copying our sample text, replace any instances of XXX with the appropriate details.

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
If the data are held or will be held in a public repository, include URLs, accession numbers or DOIs. If this information will only be available after acceptance, indicate this by ticking the box below. For example: All XXX files are available from the XXX database (accession number(s) XXX, XXX.).
• If the data are all contained within the manuscript and/or Supporting Information files, enter the following: All relevant data are within the manuscript and its Supporting Information files. Tokai University. This study was also approved by the respective ethical committees of 123 the faculties as a collaborative study. All data were fully anonymized before the 124 analyzing investigators accessed them. The study protocol was disclosed to the public at 125 the website.

126
The number of 263 cases were originally registered to this project. We checked each 127 candidate case carefully at a meeting, and selected subjects in which terminal events 128 remained at speculation irrespective of the diagnosis in the death certificate. Among

159
A total of 259 cases were collected at multiple centers for the 6-year period. The  The annual frequency of SUDI during sleep was estimated to approximately 0.31 per 163 1000 LBs. However, the value could be slightly underestimated because of the 164 possibility that some cases could have been out of management by these facilities.
165 Table 1 presents the number of subjects according to sex, birth weight, gestation 166 week, maternal age, parity and maternal smoking habit along with the population data.   The mean (± S.D.) birth weight of SUDI infants was 2885 ± 556 g for male subjects 188 (n = 137) and 2763 ± 466 g for female subjects (n = 108). These birth weights were 189 lower by 191 g (6%) and 227 g (8%), respectively, than the national mean birth weights The odds ratio of incidence of infant death of mothers whose age ≤19 years was the The first responder (n = 252), who discovered the unresponsive infant, was the 224 mother in 188 cases (75%), followed by the father in 49 cases (19%), a grandmother in 225 8 cases (3%), a childminder in 2 cases (1%), and others in 6 cases (2%).

226
Co-sleeping was recorded for 143 cases (61%) among a total of 230 available cases.
227 Table 3 presents the child sleeping position when the collapse was discovered. The 228 prone position in late SUDI infants accounted for 40% of cases, and there were 19% of 229 cases with the prone position even in 0 to 2-month-old subjects who cannot roll over.  the present study was more similar to that reported in Taiwan.

262
Traditional bedding of cotton mat, known as futon, on the floor is common in Japan.

263
Therefore, it is more appropriate to use the term co-sleeping (sharing a sleeping surface) 264 than bed-sharing. Such co-sleeping is a common style of sleep. Tokutake et al. [17] 265 reported that 84% of mothers practice co-sleeping, of whom half also practice 266 breastfeeding. The father was found to be the first responder in up to 20% of cases, and 267 in most of these cases the father also co-slept and discovered the infant death upon

272
It is a traditional practice in Japan to lay infants in the supine position. However,

273
40% of infants were found in the prone position, of which frequency was higher than 274 that reported in an earlier study. [17] Li et al. [30] reported that 60% of SIDS infants 275 were found in the prone position in the United States. It is possible that turning over by 276 infants during sleep is a causal factor. However, approximately 28% of 0 to 2-month-old 277 infants who were unable to turn over were found in the prone and side positions. They 278 might have been placed in the prone position or been breastfed during co-sleeping, but 279 the original position was not recorded sufficiently at DSI.

280
A striking finding was the prolonged time to discovery. In nine cases, it spent more 281 than 10 hours to find the unreactive infants. No apparent infanticide was involved in all 282 subjects, however neglect by parents was suspected in a couple of cases from 283 circumstances at DSI. We think that the time difference should be an important indicator 284 to suspect careless infant rearing.

285
A relationship between the occurrence of SIDS and the smoking habit of parents has 286 been found in Japan.
[31] In the present investigation, the incidence of pregnant 287 mother's smoking among SUDI cases was 34%. This incidence in the general female 288 population was reported as 11%, which also resulted in the high odds ratio of 4 information. After a new law related to child health was enacted in 2018, child death 304 reviews will be introduced to the society in the near future. These reviews in 305 combination with multiple agencies will be helpful in investigating the sleeping 306 environments of infants in detail.

307
In conclusion, we conducted an effective epidemiological analysis of sleep-related 308 SUDI using the checklist form. This approach has revealed the present critical features 309 prevailing in the country. This report displayed the latest trends of SUDI in Japan.       information at postmortem.

148
The annual frequency of sleep-related SUDI was estimated to approximately 0.3 per 149 1000 births. However, the value could be slightly underestimated because of the 150 possibility that some cases could have been out of management by these facilities.
151 Table 1 presents the number of subjects according to sex, birth weight, gestation 152 week, maternal age, parity and maternal smoking habit along with the population data.
153 Table 2 summarizes the odds ratios (95% CI) and p values in terms of the related factors.

154
The factors were also analyzed for two groups, i.e., the early group in which death 155 occurred within 2 months of age (n = 98) and the late group in which death occurred 156 after 3 months (n = 161).  The mean (± S.D.) birth weight of SUDI infants was 2885 ± 556 g for male subjects 176 and 2763 ± 466 g for female subjects. These birth weights were lower by 191 g (6%) 177 and 227 g (8%), respectively, than the national mean birth weights of 3076 g for males 178 and 2990 g for females recorded in 2017. The percentage of low birth weight infants 179 was significantly higher than the control group in both sexes. For low birth weight 180 infants, the highest odds ratio of 2.9 was observed in the late male group.

181
Infants of premature birth were found to be 2.6 times more likely to die from SUDI 182 than those of mature birth. The late group also showed higher odds ratio than the early 183 group.

185
Maternal age and birth order 186 The odds ratio of incidence of infant death of mothers whose age ≤19 years was the 187 highest at 11.1 compared with mothers aged 30-39 years, and that for mothers aged 188 20-29 years was 2.1, which showed significant differences. Furthermore, the child of a The first responder (n = 252), who discovered the unresponsive infant, was the 215 mother in 188 cases (75%), followed by the father in 49 cases (19%), a grandmother in 216 8 cases (3%), a childminder in 2 cases (1%), and others in 6 cases (2%).

217
Co-sleeping was recorded for 143 cases (61%) among a total of 230 available cases.
218 Table 3 presents the child sleeping position when the collapse was discovered. The 219 prone position in late SUDI infants accounted for 40% of cases, and there were 19% of 220 cases with the prone position in the early group.  the present study was more similar to that reported in Taiwan.

247
Traditional bedding of cotton mat, known as futon, on the floor is common in Japan.

248
Therefore, it is more appropriate to use the term co-sleeping (sharing a sleeping surface) 249 than bed-sharing. Co-sleeping is a common style of sleep, and Tokutake et al. [17] 250 reported that 84% of mothers practice co-sleeping, of whom half also practice 251 breastfeeding. The father was found to be the first responder in up to 20% of cases, and 252 in most of these cases the father also co-slept and discovered the infant death upon It is a traditional practice in Japan to lay infants in the supine position. However,

258
40% of infants were found in the prone position, of which frequency was higher than 259 that reported in an earlier study. [17] Li et al. [30] reported that 60% of SIDS infants 260 were found in the prone position in the United States. It is possible that turning over by 261 infants during sleep is a causal factor. However, approximately 28% of infants in the 262 early group who were unable to turn over were found in the prone and side positions.

263
They might have been placed in the prone position or been breastfed during co-sleeping, 264

Commented [O53]:
We thought that the difference should be caused by that a particularly higher risk was evident among teenage mothers than that found in an earlier study [25], and that more first-born infants were dead during 0-2 months of age.    Answer; In the initial attempt, we selected the five-year-period from 2013 to 2017. However, the number of gathered cases was around 200, which seemed insufficient for statistical analysis. Then, we extended the period one more year to 2018 because we thought that more than 250 cases should be desirable for statistically significant analysis. A phrase was added into the sentence of Methods.
3 Method lone 90-96 and 120-129. Explain the selection and classification methods of the study subjects precisely. Not clear even how many SUDI were registered totally.
When and who decided? How to get a final agreement among investigators? Any audit by the third party? This is the most important point of this study. Answer; As the reviewer pointed out, the explanation was short in the manuscript. Answer; The area to which the eight joined faculties covered metropolitan cities like Fukuoka and Kyoto to rural side like small islands and isolated mountain side. The distribution was biased around the country, we thought that the data should represent the whole nation.
Concerning the limitation of regional bias, a sentence is added into line 110 of the track changed text.
7 Results Table 2 line 162-163. Were these odds ratios adjusted against the background distributions of the control? Answer; No, the odds ratios were not adjusted. The reviewer should imply the Pearson's χ 2 test, which is calculated based on deviation of each fraction. The mean value and deviation are available for the whole national population data. But it was impossible to obtain the mean and deviation for grouped fractions separately such as < 2500 g and > 2500 g.
Further, the χ 2 value of exact method is usually compatible with that in the Pearson's test.
8 Results the same as above. Explain why the early group consists of 0-2 months of age. Answer; As the reviewer pointed out, it should be unreasonable to separate the subjects into two groups of 0-2 month-old infants and others. The authors quitted the grouping. Tables 1   and 2  Answer; The authors added major findings of this study into the first paragraph of the Discussion (line 235 of the track changed text). Please also see lines 246-251 in the revised version.
10 Discussion. Better to make a paragraph of limitations of the study with potential bias or imprecision. Answer; As the reviewer pointed out, a paragraph including the statement concerning the limitation is inserted into the beginning of the Discussion, which is the same way as indicated to the last comment.
11 Discussion line 250-252. Be careful about this interpretation because of the lack of control subjects. Answer; This sentence is not our interpretation. This is the results that Tokutake et al. investigated [17]. To circumvent the misleading, the sentence is changed a bit.
Minor comments 1 Discussion lines 279 and 286. The authors said the DSI was not reliable while the checklist was an official, very confusing description. Answer; The statement was inappropriate. The checklist form is widely utilized among health care stuffs of pediatricians and pathologists. The form is also distributed to police officers in some degree, but it is not mandatory for them to fill in. The authors discarded the word of 'official' at two sites. Please see lines 70 and 286 in the original version.
2 Better to show a flow diagram of subjects. Answer; The number of originally registered cases was 263. As mentioned earlier, only 4 cases were eliminated from the reasons of not found in sleep environment (n = 3) and fully explained clinical cause of heart disease (n = 1) among them. As mentioned earlier, the statement was inserted into the revised text. No flow diagram has been constructed because of the small number of selection.
Reviewer #2:: The detailed review of our article is appreciated. According to the raised comments, the manuscript has been rewritten to a revised version. Alterations are indicated as track changes in the revised manuscript. The answers to the raised comments are as follows, Line 61. it might be easier to make a international comparison by giving rates instead of numbers per year, e.g.if rates of SIDS are given as 0.15 per 1000 livebirths Answer; As the reviewer advised, the number has changed to the annual rate per 1000 livebirths.
Line 73. Good that Japan has a SIDS investigation form -is it used for by the death scene investigation team or forensic pathologists? The authors might want to mention that it is similar to CDC SUDI investigation form since it is in Japanese.

Answer;
The checklist form is widely utilized among medical stuffs including pediatricians and pathologists. Items in the form is similar to the CDC one. The sentence is changed a bit.
Line 79. It would be interesting as an introduction to mention incidence of SUDI in Japan (0.4(?) per 1000 livebirths from Taylor et al's study) and any findings of sleep related SUDI from previous studies such as ref 17 or from using the DSI form, since that is the authors' major focus, rather than a mention of vaccination related SUDI Answer; As the reviewer found, the incidence of 'SUDI' in Japan is around 0.4 per 1000 livebirths, which is shown in the Tayler's article [7]. But what the authors would like to say in this sentence was the number of 'SIDS', not 'SUDI'. In the 1990s, the diagnosis of 'SIDS' was popular, but the diagnosis is considerably avoided among pathologists these days.
Anyhow, the aim of our study was not clearly described in the last paragraph of Introduction. One sentence is added into the paragraph on line 82 of the track changed text). Further, the maternity passbook that was used in the last study was very useful for the present study as well, particularly for the maternal smoking habit. The term of 'maternity passbook' is added into the last sentence as well (line 84).
Line 88. The authors' case definition included infants 0-365 days. According to most definitions, unexpected deaths in infants under 7 days of age are excluded from the SUDI category, and instead have been termed "sudden unexpected early neonatal death (SUEND)". It would be interesting to know how many infants in the study were actually below 7 days of age to see if it would impact the findings related to early or late SUDI. Answer; In this revision, we provided the whole data set that was used in this series of studies. The earliest death was a 9-day-old infant. A phrase of 'age less than 1 week' is added into line 89 of the track changed text.
Line 134 -alignment of words (for editor) Answer; An equal line-width makes the prolonged arrangement. Sorry for no other expression.
Line 157, do the authors mean the later group is in where death occurred at or after 3 months of age .. . What about the infants of 2 months plus to 3 months of age if the later group is AFTER 3 months of age and Early group is within 2 months of age (<= 2 months old). Do the authors mean "below 3 months of age" ?
From the existing literature, the age groups are usually neonatal SUDI (i.e. below 28 days , approximately 1 month of age) and after the neonatal period. Is it an empirical decision or to take into account the preterm babies ? Answer; As the reviewer pointed out, it should be unreasonable to separate the subjects into two groups of 0-2 and 3-11-month-old infants. The authors agree with spoiling away the grouping. Tables 1 and 2 have been reconstructed completely. For Table 3, we demonstrated the data within 2-month-old infants because some of SUDI cases found in prone position prior to rolling over, which should be a good finding, we thought.
Line 159 Table 1. "SUDI" is used in the text but "SUID" is used in the Table, suggest to use one term only for consistency. Suggest to add 'livebirths' to become "Approx annual (incidence) rate per 1000 livebirths(LB) " Answer; We appreciate finding such careless spell miss. The term has been corrected at a total of three sites. Please see Table 1.
Line 179 -Is it the annual incidence rate of SUDI in low birth weight infants being significantly higher rather than the "percentage", the latter which is not shown in table 1 Answer; The term of 'percentage' is corrected to 'incidence', as indicated by the reviewer.
Line 223 Table 3 -typo Supine Answer; The spell miss is corrected.

Statistical analysis :
Please excuse me if my understanding of the analysis is incorrect ... years) ] divided by 6 = 0.008 and so forth for the other variables. It may also not be necessary to have an annual incidence but just an overall incidence over the 6 years since the annual rate is very low Answer; The number was further divided by 0.14, of which number was from the rate of covered population in this study. 145 *1000 / no. of LB (in the 6 years) / 0.14 = 0.34 2. Line 190 and 191 -I am not sure if the number of early SUDI in one age group compared to early SUDI in the rest of the population can be used in chi squared calculation against total SUDI population-should it instead be compared with non event population in the same age group ? Answer; The authors agreed with the comment by the reviewer. The distribution of occurrence is not even in the other generations, meaning that the comparison was very rough. As mentioned earlier, we quitted grouping of 0-2 and 3-11 month-old infants. We would like to delete this sentence.
Line 255. The effects of co-sleeping could not be evaluated but perhaps more details if available could be described under findings? Was there any related to wedging, inadvertent suffocation especially in co-sleeping cases ? Or to mention Line 280 ..."not well trained in documentation". Answer; Situations of co-sleeping varied among cases such as with mother, with parents and with brothers and sisters, and in the same mattress and the separate mattresses, etc. Actually, two fathers confessed overlay during drunken sleeping. However, the detailed information was not obtained for all. It was difficult to evaluate the effects of co-sleeping further. In the next time, we would like to do another prospective study to evaluate effects of co-sleeping. The sentence has been changed a bit on line 256.
Discussion -Some comment re missing data would be helpful , 14 cases missing in gender , 25 cases in gestation and more than half cases had missing data on maternal smoking -how it may affect the reliability of the data esp with regards to smoking , although this is a known risk factor in sleep related SUDI. Sleeping position was missing in 35 cases. These are understandable given the retrospective nature of the study -so a comment may be useful to the reader Answer; Because this was a retrospective study, it was impossible to get all data without any lucks.
A statement concerning this limitation was added into the revised text. The sentence was inserted into the first sentence of the first paragraph of Discussion (line 235 in the track changed text). Please also see lines 244-249 in the revised version.
Line 238 -Could the authors comment if the peak age of death at one month is related to the age group being studied to be 0-12 months of age as compared to other SUDI studies where the study group is from 2 to 12 months ? Answer; Good point! Other SUDI studies [4,24] showed the peak age to 2 months. As far as we read the articles, no apparent differences were found. However, we thought that the difference in peak age might be caused by that a particularly higher risk was evident among teenage mothers, and that more first-born infants were dead during 0-2 months of age. This message is replaced on lines 239-241.
Prevention issues -of the risk groups are not discussed much . Any comments on how the babies of later birth order might be at risk ? Answer; The authors are sorry for no idea about prevention issues to infants in later birth order, because we are not clinicians who care a number of alive patients. But according to the advice from the reviewer, we added a general sentence concerning prevention into line 278 of the track changed text.
Is there a concern about neglect if the duration of an infant seen alive is more than 5-6 hours for those found in the morning, and more than 2-3 hours for those found in the afternoon (since the parents or childminder would be awake then)? Is there a likely delay in reporting due to possible infanticide/ negligence in those reporting that last seen alive was 8 hours ? Or is it a non carer who was reporting his duration? Lack of this data could be a discussion point in your study re improvement in investigation by the social workers or police. Issues related to DSI itself could be a discussion point on quality improvement in