Structural bias in the quality of information from death investigations for SUIDs

Abstract

Objective:

To assess Sudden Unexpected Infant Death (SUID) investigations for structural inequities by race/ethnicity and geography.

Methods:

The SUID Case Registry compiles data on death investigations. We analyzed cases from 2015 to 2018 (N=3,847) to examine likelihood of an incomplete death investigation, defined as: missing autopsy, missing scene investigation, or missing detailed information about where and how the body was found. We also analyzed which specific components of death investigations led to the greatest number of incomplete investigations.

Results:

24% of SUIDs had incomplete death investigations. Death scenes in rural places had 1.51 times the odds of incomplete death investigations (95% CI 1.19–1.92) compared to urban areas. Scene investigations led by law enforcement were more likely to result in incomplete death investigations (OR=1.49, 95% CI 1.18–1.88) compared to those led by medical examiners. American Indian/Alaska Native SUIDs were more likely than other racial groups to have an incomplete investigation (OR= 1.49, 95% CI 0.92–2.42), more likely to occur in rural places (p=0.055), and more likely to be investigated by law enforcement (p<0.001). If doll reenactments had been performed, 358 additional cases would have had complete investigations, and if SUID investigation forms had been performed, 243 additional cases would have had complete investigations. American Indian/Alaska Native SUIDs were also more likely to be missing specific components of death investigations.

Conclusions:

To produce equitable public health surveillance data used in prevention efforts, it is crucial to improve SUID investigations, especially in rural areas and among American Indian/Alaska Native babies.

Introduction

Every year in the United States about 3,500 infants die suddenly and unexpectedly without a clear cause of death,¹ a category of deaths referred to as “sudden unexpected infant deaths” (SUIDs). Rates of SUID are highest among American Indian/Alaska Native infants and second highest among Black infants² with increasing disparities compared to white infants³ that persist when examining only preterm births.⁴ In addition, rural areas have the highest infant mortality rates, and SUIDs contribute the most excess cases.⁵ SUID rates have not improved in the last twenty years,¹ so the CDC created the Sudden Unexpected Infant Death Case Registry to collect surveillance data on them.⁶ Data collection is one of the core functions of public health,⁷ one of the strategic directions in the Report of the Secretary’s Advisory Committee on Infant Mortality is to invest in improving data collection and surveillance systems.⁸

Structural racism has recently become more widely understood to be a fundamental cause⁹ of racial health inequities,¹⁰ including within data collection efforts.^11,12 Structural racism is the totality of laws, policies, practices, and norms that abet racial discrimination, often without an identifiable perpetrator.^13,14 In addition, the already higher rates of sickness and death in rural areas are heightened for racial and ethnic minority groups living in those areas.¹⁵ Public health surveillance is “the ongoing systematic collection, analysis, interpretation and dissemination of health data for the planning, implementation and evaluation of public health action.”¹⁶ If structural inequities affects our surveillance efforts, this could propel inequities throughout the planning and implementation stages of public health action.

The bedrock for SUID surveillance data is the death investigation.¹⁷ The bare minimum of a death investigation for infant deaths according to the National Association of Medical Examiners (NAME) includes 1) interviews with witnesses at the scene, 2) a detailed description of the position that the infant was found in, and 3) a thorough autopsy.¹⁷ Details from the death scene investigation are especially important for SUIDs because such deaths are often caused by suffocation, but suffocation does not have biological markers.¹⁸ Although they occasionally provide evidence for criminal investigations, the main purpose of infant death scene investigations is to find a cause of death and inform prevention efforts. Parents typically report that the scene investigation is a positive experience, despite the tragic circumstances, and want to cooperate to find the cause of the infant’s death.¹⁹ Still, it is recommended that SUID scene investigations be done by specially trained investigators¹⁷ because there are specific investigative questions to be asked and because it is important to gather complete information at the same time as being respectful to a grieving family.

The SUID Case Registry⁶ helps fill the knowledge gap created by the hyper-local patchwork of policies and practices that govern death investigations. These local processes make it difficult to create a comprehensive picture of how scene investigations are carried out nationwide. The CDC funds 26 states and jurisdictions to collect and record SUID data and to review each case at a multidisciplinary review committee (i.e. child protective services, pediatricians, law enforcement, medical examiners, public health practitioners). They collect data from medical records, medical examiners/coroners, autopsies, police investigations, child protective services, and birth and death certificates. Then they record the data in the National Child Fatality Review-Case Reporting System, which includes many types of child deaths. The resulting surveillance data better describe the circumstances surrounding SUIDs than what was previously available from death certificates and are used to inform prevention strategies.^3,20–24

Although the completeness of individual components of SUID scene investigations has been examined,²⁰ the overall completeness of death investigations has not been documented, including whether completeness varies by race and geography. Incomplete death investigations by race, or by characteristics patterned by race,²⁵ would indicate structural racism.²⁶ This is because practices that create or perpetuate racial inequities are racist, while practices that reduce and eliminate racial inequities are anti-racist.²⁷

In this study, we examined the completeness of death investigations for Sudden Unexpected Infant Deaths, including whether data completeness differed by race/ethnicity, geography, and the type of agency conducting the scene investigation. If our public health infrastructure systematically collects less complete data among marginalized groups, we will continue to propagate structural inequities.

Methods

Dataset: Sudden Unexpected Infant Death (SUID) Case Registry

Deaths were included in this study if they happened between 2015 and 2018, and if the state determined that they fit the CDC’s SUID definition: “the death certificate indicates the cause as unknown, undetermined, SIDS, SUID, unintentional sleep-related asphyxia/suffocation/strangulation, unspecified suffocation, cardiac or respiratory arrest without other well-defined causes, or unspecified causes with potentially contributing unsafe sleep factors” ²⁸ (N=3,847). We obtained de-identified SUID Case Registry data through a National Center for Fatality Review and Prevention agreement. In addition, we obtained a dataset of the same SUID cases from the CDC that contained the CDC team’s classification of each case²⁸, and merged the two datasets with SAS, using a unique identifier provided by both the National Center for Fatality Review and Prevention and by the CDC in their respective datasets.

Outcomes

The primary outcome of this analysis was whether each SUID had complete or incomplete death investigation as recorded in the SUID Case Registry. This measure was derived from the CDC decision-making algorithm for differentiating SUID subtypes.²⁸ Multidisciplinary review committees use the algorithm to categorize each SUID in their jurisdiction on a continuum of how likely it is that the death was caused by suffocation. The first category is for SUID cases with “no autopsy or death scene investigation”. The second category is for SUID cases with “incomplete case information”, which means they did not have detailed information about where and how the body was found, or their autopsy did not document toxicology, radiograph, and pathology. We combined these two categories into one indicating an incomplete death investigation. We collapsed the remaining categories into a second group of SUIDs that had a complete autopsy and detailed information about where and how the body was found (complete death investigation, the reference group). Although the algorithm was not designed to assess completeness of death investigations, each SUID case was categorized carefully by a team of local experts which makes it an ideal measure of completeness. A panel at the CDC categorized each SUID case in a process parallel to, but independent from, review committees in local jurisdictions. Because categorization processes vary from one jurisdiction to another, we used the CDC’s categorizations to construct our primary outcome.

The secondary outcomes were whether each component of a death investigation that should be performed¹⁷ was actually performed and shared with the SUID Case Registry team. For each specific component of the death scene investigation, we quantified the number of additional cases that would have been overall complete if that missing component had been available to the review team. This supports the primary aim because it allowed us to rank which specific components were the most important targets for improvement. The components we examined were: narrative description, scene photos, reenactment with a doll, SUID investigation form, witness interviews, and autopsy.

Exposure

The primary characteristics (exposures) we examined are the infant’s race/ethnicity, geography of the death scene (urban, suburban, rural), and the primary type of agency that performed the scene investigation (medical examiner, coroner, law enforcement).

Race/ethnicity.

SUID Case Registry grantees record infant’s race according to their death certificate. They select all applicable categories from: white, Black, Asian, American Indian, Alaskan Native, Native Hawaiian, Pacific Islander, and Unknown race; and Hispanic ethnicity. We created five racial/ethnic categories: Hispanic, non-Hispanic (NH) white, NH American Indian/Alaska Native, NH Asian American/Pacific Islander/Native Hawaiian, NH Black or African American. We created Hispanic and non-Hispanic categories because Hispanic people identify more consistently with being Hispanic than with any race.²⁹ Multi-racial infants were categorized in mutually exclusive groups in the following order: American Indian/Alaska Native, Black, Asian, white. For example, an Asian and white infant would be categorized as Asian. We chose these categories because Black/Asian and Black/white people most often choose the monoracial category of Black, and Asian-white people most often choose the monoracial category of Asian.³⁰ In order to prioritize a larger sample size among a group that is understudied due to small numbers, we categorized multi-racial American Indian/Alaska Natives as American Indian/Alaska Native even though people in these combinations sometimes choose monoracial categories of Black or white.²⁹ This means our estimates for outcomes among American Indian/Alaska Native infants may be artificially low.

Death scene geography, categorized as urban, suburban, or rural according to SUID Case Registry grantees with local knowledge.

Type of agency that performed death scene investigation.

There are often several agencies present at the death scene. First, investigations were categorized as performed by a medical examiner or medical examiner investigator if either of those agencies were present. Next, investigations were classified as coroner or coroner examiner if either of those agencies were present, but if a medical examiner representative was not. Medical examiners are typically physicians trained in pathology, whereas qualifications for coroners are more variable. Only 1% of cases (n=54) were investigated by both a coroner’s office and a medical examiner’s office. Lastly, cases were classified as law enforcement if only law enforcement was present. Every case was investigated by at least one of these three agency types. If a medical examiner or coroner is present, they typically lead the scene investigation. However, if only law enforcement is present, they are tasked with collecting all relevant information, including SUID-specific data.³¹

Analysis

We examined univariate and bivariate statistics by race/ethnicity for death scene geography and agency type. We ran three logistic regression main effect models with the primary exposure variables (race, geography, agency type) and the outcome of an incomplete compared to a complete death investigation. We used directed acyclic graphs (DAGs) to help determine which variables to adjust for as potential confounders (Supplemental Figure 1).³² We adjusted all models for infant’s age in months, sex, and state where the incident occurred (all categorical variables; Table 1).

Table 1:

Descriptives of infant characteristics, overall and by whether their case had complete or incomplete death investigations in the SUID Case Registry (N=3,847)

	Overall (N=3,847)	Complete data (N=2,831)	Incomplete data (N=939)	Missing (N=77)
Outcome: Data completeness	100%	73.6%	24.4%	2.0%
Race
White (non-Hispanic)	40.7%	40.7%	40.3%	44.2%
Black (non-Hispanic)	42.1%	41.4%	43.9%	46.8%
American Indian/Alaska Native (non-Hispanic)	3.2%	3.3%	3.2%	X
Asian American (non-Hispanic)	1.6%	1.9%	1.0%	X
Hispanic	12.1%	12.6%	11.2%	X
missing	0.2%	X	X	X
Place of incident that preceded death
Urban	38.1%	39.0%	35.6%	32.5%
Suburban	32.9%	32.7%	33.1%	39.0%
Rural	22.8%	22.6%	23.3%	X
missing	6.2%	5.7%	8.0%	X
Death Investigation Agency Type
Medical Examiner	43.6%	46.5%	35.5%	40.3%
Coroner	27.0%	26.2%	29.7%	22.1%
Law enforcement	22.9%	23.5%	21.5%	16.9%
missing	6.5%	3.8%	13.5%	20.8%
Age at Death
< 1 month	11.4%	10.6%	14.0%	13.0%
1 month	19.8%	19.4%	21.4%	14.3%
2 months	19.5%	19.6%	19.3%	19.5%
3 months	16.0%	16.0%	16.3%	14.3%
4 months	11.2%	11.7%	10.1%	7.8%
5 months	7.2%	7.6%	5.8%	9.1%
6 months	5.7%	5.8%	4.8%	10.4%
7–11 months	9.2%	9.4%	8.4%	11.7%
Sex
Male	58.4%	58.7%	57.8%	53.3%
Female	41.6%	41.2%	42.2%	46.8%
missing	X	X	X	X

X indicates an estimate is suppressed because of an underlying cell size N<6

Next, we tested for an interaction between race and place of incident in predicting the death investigation completeness outcome. We produced predicted probabilities of incomplete death investigation from the logistic regression models. Our data agreement required cell sizes of n=6 or greater to be presented. Cases were excluded from analyses in which they were missing exposure or outcome data; prevalence of missingness is presented in Table 1. Overall, 2% (N=77) of cases were missing outcome data. Missingness was overrepresented among white infants and Black infants.

Lastly, we estimated 1) the risk of having an incomplete death investigation by whether each component of the scene investigation was done 2) a corresponding risk difference for each component and 3) the number of SUID cases with incomplete death investigations that were attributable to having each incomplete component. This number is a derivative of ‘attributable risk in the exposed’, calculated by multiplying the component-specific risk difference by the total number of cases without that component (‘exposed’) cases. The resulting estimate is the so-called “number of excess cases”,³³ or in this analysis, the number of cases with incomplete death investigations that could have had complete death investigations if that component of the scene investigation had been completed and shared with the SUID Case Registry team, scaled to the 2015 through 2018 U.S. SUID registry data. We also assessed the percentage of whether each component was done by infant race.

Results

Of the 3,847 infant deaths in the SUID Case Registry from 2015 to 2018, 24% of cases had incomplete death investigations (Table 1). Medical examiner representatives were present at 47% of SUID scenes, followed by coroner representatives (27%), and law enforcement (23%). Death scenes were most often in urban areas (41%) followed by suburban (35%) and rural (24%) areas.

The death scene geography and the type of investigating agency both differed by infant race (p<.001; Table 2). Non-Hispanic (NH) American Indian/Alaska Native infant deaths were the most likely to be investigated only by law enforcement (41%) compared to Hispanic (29%), NH white (27%), NH Black (20%), and NH Asian American (19%) infants. NH American Indian/Alaska Native infant deaths were also most likely to occur in rural places (49%) compared to NH white (40%), Hispanic (18%), NH Black (11%), and NH Asian American (10%) infant deaths.

Table 2:

Sudden Unexplained Infant Deaths in the US: Distribution of death scene and primary investigating agency, by race/ethnicity in the SUID Case Registry Dataset, 2015–2018. (N=3847)

	Overall	Non-Hispanic								Hispanic (N=466)		Overall race/ ethnicity χ2 test
		White (N=1564)		Black (N=1621)		Native American (N=124)		Asian American (N=63)		Percent	χ2 p-value compared to white
		percent		percent	χ2 p-value compared to white	percent	χ2 p-value compared to white	percent	χ2 p-value compared to white
Primary Investigating agency
Medical Examiner	47%	40%	ref	51%	<0.001	51%	0.017	53%	0.046	52%	<0.001	p<.001
Coroner	29%	34%	ref	29%	0.003	8%	<0.001	29%	0.438	19%	<0.001
Law Enforcement	24%	27%	ref	20%	<0.001	41%	0.001	19%	0.164	29%	0.422
Total	100%	100%		100%		100%		100%		100%
Place of death
Urban	41%	21%	ref	59%	<0.001	32%	0.004	37%	0.002	43%	<0.001	p<.001
Suburban	35%	39%	ref	30%	<0.001	19%	<0.001	54%	0.020	39%	0.821
Rural	24%	40%	ref	11%	<0.001	49%	0.055	10%	<0.001	18%	<0.001
Total	100%	100%		100%		100%		100%

Figure notes: All estimates supported by an underlying cell size of N<6

“ref” denotes the reference group

In the covariate-adjusted regression models (Figure 1, Supplemental Table 1), NH American Indian/Alaska Native deaths were more likely to have incomplete death investigations (OR=1.49, 95% CI: 0.92–2.42) and NH Asian Americans were less likely to have incomplete death investigations (OR=0.54, 95% CI: 0.25–1.15) compared to information collected on NH white infants, although both estimates were imprecise, likely due to small sample sizes. Black infants had complete death investigations at similar rates to white infants (OR=1.00, 95% CI: 0.82–1.23).

Figure 1:

Main effect logistic regression odds ratios of having incomplete death investigations, compared to having complete death investigations by race: SUID Case Registry 2015-2018 (N=3847)

Figure notes

Models adjusted for state, age, sex, death scene geography

Error bars display 95% confidence intervals

Death scene geography predicted whether an infant had incomplete death investigations (Figure 2, Supplemental Table 1). Death scenes in rural places had 1.51 times the odds of having incomplete death investigations (95% CI 1.19–1.92) and death scenes in suburban places had 1.23 times the odds of producing incomplete death investigations (95% CI 1.00–1.51) compared to urban areas.

Figure 2:

Main effect logistic regression odds ratios of having incomplete death investigations, compared to having complete death investigations by death scene geography, and investigation agency: SUID Case Registry 2015–2018 (N=3847)

Figure notes

Models adjusted for state, age, sex, race/ethnicity

Error bars display 95% confidence intervals

Type of death investigation agency was important as well (Figure 2, Supplemental Table 1), with law enforcement-led investigations more likely to result in incomplete death investigations (OR=1.49, 95% CI 1.18–1.88) compared to medical examiner-led investigations. There was an interaction between place of incident and infant’s race/ethnicity (Supplemental Figure 2). NH American Indian/Alaska Native infants in rural places were twice as likely to have incomplete death investigations compared to NH white infants in urban areas.

Among all the specific components we examined that make up the death investigation, the component that was complete most often was the autopsy (96%), narrative description (89%) and witness interviews (81%). The component completed least often was the reenactment with a doll (44%). American Indian/Alaska Native infants had higher rates than any other race of missing a narrative description (p=.03), a doll reenactment (p<.001), a SUID investigative form (p=.004), an autopsy (p=0.19), and witness interviews (p=.35) (Supplemental Table 2).

Our last analyses examined the relative importance of each component of death investigations for resulting in an overall incomplete death investigation (Table 3). For instance, 93% of investigations without an autopsy had incomplete death investigations, compared to 24% with an autopsy. However, because autopsies were completed for 96% of cases, there is not much room for improvement (i.e., lack of autopsy resulted in only 41 excess cases). To determine which of the components resulted in the greatest number of cases with incomplete death investigations, we estimated the number of additional cases (scaled to the 2015 through 2018 SUID Case Registry) that could have had complete death investigations, if a specific component had been completed (Table 3, last column): a measure of excess cases, to inform prevention and intervention efforts. Completing doll reenactments would result in the most (n=358) additional cases with complete death investigations and completing a SUID investigation form would result in 243 additional cases with complete death investigations. Therefore, these specific missing components are prime targets for improving the completeness of death investigations.

Table 3:

Measures of excess risk of incomplete death investigations on the infant death, by specific components of the death scene investigation among all SUID cases in the U.S. SUID Case Registry, 2015–2018 (N=3847)

	Completeness of component (%)	Completeness of Death Investigations (N)		# of Cases Total (N)	Incomplete Death Investigation % (Risk)	Risk Difference	Excess Cases (N)
		Incomplete (N=939)	Complete (N=2831)
Narrative description	89%
No (exposed)		203	155	358	57%	35%	126
Yes (unexposed)		736	2676	3412	22%
Scene photos	72%
No		394	592	986	40%	20%	201
Yes		545	2239	2784	20%
Reenactment with doll	44%
No		676	1386	2062	33%	17%	358
Yes		263	1445	1708	15%
SUID investigation form	75%
No		408	484	892	46%	27%	243
Yes		531	2347	2878	18%
Witness interviews	81%
No		263	398	661	40%	4%	26
Yes		676	2433	3109	22%
Autopsy Completed	96%
No		55	4	59	93%	69%	41
Yes		884	2827	3711	24%

Table Notes

(Number of cases total without component completed) * (risk difference) = Excess cases

“Excess cases” are the number of additional cases that would have had complete death investigations if component of investigation were done

X indicates the estimate was suppressed because of a cell size smaller than n=6

Fillable Excel worksheet with built-in equations available in supplemental materials

Discussion

Of all the cases in the SUID Case Registry between 2015 and 2018, 24% had incomplete death investigations. Death investigations were less likely to be complete 1) when investigations are conducted by law enforcement 2) for Native American/Alaska Native infants, and 3) in rural areas. The doll reenactment is the component of the death investigation with the most potential to improve death investigations.

Investigations led by law enforcement were less likely to be complete, and one way that structural racism is operating is that Native American infant deaths were more likely to be investigated by law enforcement. While law enforcement in some states receive training on best practices in infant death scene investigations, it is not a core part of their job. First responders³⁴ and police¹⁹ prefer that trained medical professionals lead infant death scene investigations. Indeed, law enforcement officials report that they are inclined to treat parents as potential suspects.³⁵ Death investigations performed only by police were upsetting to parents, and information was less complete.¹⁹ First responders also feel less confident conducting investigations for sudden unexpected infant and child deaths compared to conducting investigations for events that occur more frequently.³⁴ Therefore these findings suggest that medical examiners or coroners be involved in as many infant death cases as possible, to improve the completeness of this important surveillance data. One possible barrier is that medical examiners and coroners are allocated by population size, which in rural areas can result in long distances to travel to death scenes. Because death investigation systems vary widely by state, the solutions to standardizing SUID investigations are likely also state-specific.

American Indian/Alaska Native infants were more likely to have incomplete death investigations. This is consistent with our hypothesis that the quality of the death investigation would be patterned by race/ethnicity, which indicates that structural racism may be operating within this data collection. Strikingly, American Indian/Alaska Native infant deaths were both 1) more often investigated in rural areas and 2) more often investigated by law enforcement, which were each associated with having incomplete death investigations in our study. This is particularly concerning because American Indian/Alaska Native infants also experience the highest rates of SUIDs,² which means the public health system collects the least complete information in the population with the highest mortality burden. We may be missing vital information on how to prevent SUIDs among American Indian/Alaska Native infants.

There were also structural inequities by geography, with death scenes in rural areas more likely to be incomplete. On top of that, we documented an interaction between geography and race, where American Indian/Alaska Native infants in suburban and rural areas were more likely to have incomplete death investigations compared to other groups (Supplemental Figure 2). This means that the structural inequities by race and geography exacerbate each other to result in incomplete death investigations for marginalized communities.

This study found that doll reenactments and use of SUID-specific investigative forms were the most important components that if left incomplete, generated the largest numbers of incomplete death investigations. These two components were also missing at the highest rates for American Indian/Alaska Native infants, indicating a specific pathway through which structural inequity may be operating. Both components are unique to infant death scene investigations and perhaps why they are more likely to be forgotten. It is also possible that investigators avoid performing doll reenactments because parents find them unpleasant.³⁶ Agencies that conduct death scene investigations should add these two items – a doll and a SUID investigative protocol form – to their required list of things to bring to an infant death investigation. In addition, death scene investigators should be specifically trained and prepared to walk a family through these procedures with sensitivity.

SUID Registry grantees and others who use the CDC’s SUID classification can replicate Table 3 analyses to identify the largest contributors of incomplete death investigations in their local systems. Using a SUID investigative protocol is recommended both by the National Association of Medical Examiners, the National Sheriff’s Association,³⁷ and the FBI³¹, and an exemplar is available from the CDC as a fillable, printable form. Both the CDC and the National Center have resources to train scene investigators to perform SUID investigations. One complementary solution to improving data from SUID investigations could be to recruit health professionals such as nurses to perform the aspects of a scene investigation that involve interacting with families, which has been done successfully in England.¹⁹ This could be especially useful in rural areas where medical examiners’ offices are in short supply.

Limitations

The data in the SUID Case Registry represent the 26 states and jurisdictions that participate in it and is not generalizable to the entire country. We were not able to identify individual states, which could have allowed us to assess which states have high-performing death investigation systems.

Several of our primary variables were measured with error. States are instructed to record the race from the infant’s death certificate, but the infant’s birth certificate tends to be more accurate because it is reported by parents, rather than by death investigators.³⁸ In addition, we would have preferred to use an objective measure of geography derived from the death scene address, but the geographic data are masked for confidentiality. We relied on local data abstractors’ determination of urban, suburban, or rural. We hope that the abstractors’ local knowledge leads to meaningful categorizations, but it is possible that aspects of the scene investigation could influence their determination of geography and introduce bias into that measure.

We excluded cases from analyses for which they were missing exposure or outcome data. Missing data were rare for race (0.2%), age and sex (suppressed due to small cell sizes), and slightly higher for place of incident (6.2%) and death investigation agency type (6.5%).

Implications for Policy & Practice.

• It is important to assess public health infrastructure, such as surveillance data and death investigations, for structural inequities among underserved and marginalized groups of people.

• All SUID investigations should include a doll reenactment and a SUID-specific protocol such as CDC’s SUID investigative reporting form (SUIDIRF).

• States and localities who collect SUID Case Registry data should replicate the Table 3 analyses to determine which components of SUID investigations have the highest potential for improvement.

• Whenever possible, SUIDs should be investigated by a coroner’s or medical examiner’s office rather than law enforcement.

• Public health agencies should seek out the expertise of Tribal Nations and American Indian/Alaska Native organizations to get a better understanding of barriers and needs regarding SUID death investigations.

• States should explore how to support local agencies in improving SUID investigation in rural areas.

• Additional federal, state, and local funding for SUID investigations and surveillance data is warranted to equitably and comprehensively collect data used in prevention efforts.

Abbreviations:

CDC

Centers for Disease Control and Prevention

CI

confidence interval

OR

odds ratio

SUID

sudden unexpected infant death