You are here

Article Text

Article menu

Download PDFPDF

Original article
Homicides of law enforcement officers responding to domestic disturbance calls
  1. Cassandra Kercher1,
  2. David I Swedler1,
  3. Keshia M Pollack1,2,
  4. Daniel W Webster3
  1. 1Department of Health Policy and Management, Johns Hopkins Center for Injury Research and Policy, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA
  2. 2Johns Hopkins Center for Occupational Safety and Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA
  3. 3Department of Health Policy and Management, Center for Gun Policy and Research, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA
  1. Correspondence to Cassandra Kercher, Department of Health Policy and Management, Johns Hopkins Center for Injury Research and Policy, Johns Hopkins Bloomberg School of Public Health, 614 North Broadway, Room 508, Baltimore, MD 21205, USA; ckercher{at}jhsph.edu

Abstract

Objective To describe the law enforcement officer (LEO), encounter, perpetrator and victim characteristics of domestic disturbance-related LEO homicides in the USA from 1996 to 2010.

Methods Narrative text analysis was conducted on the Federal Bureau of Investigation's annual report ‘Law Enforcement Officers Killed and Assaulted’. Potential cases were confirmed if the narrative included the term ‘domestic disturbance’ or a domestic disturbance situation was described.

Results 116 LEOs were killed while responding to domestic disturbance calls. Ninety-five per cent of these homicides were committed with a firearm. Sixty-seven per cent of LEOs were wearing body armour when killed; however, 52% received the fatal wound to the head/neck. Sixty-one per cent of suspects had a criminal history mentioned within the narratives and perpetrators of intimate partner violence (IPV) were more likely to be killed by LEOs than suspects involved in other forms of domestic violence. Victims of the domestic disturbance were killed in 21% of the IPV-related LEO homicide cases as opposed to only 5% of other domestic disturbance calls. A firearm was the most common weapon used in the murder of a domestic disturbance victim (86%).

Conclusions This study describes domestic disturbance-related LEO homicides. Future research in this area should further examine the dangers unique to domestic disturbance calls. A longitudinal analysis could provide greater understanding of the injury and mortality risks faced by LEOs, in order to inform homicide prevention among law enforcement.

https://doi.org/10.1136/injuryprev-2012-040723

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    According to the National Law Enforcement Officers Memorial Fund, there are currently more than 900 000 sworn state and local law enforcement officers (LEOs) serving in the USA.1 These LEOs daily place themselves in dangerous situations to ensure the safety of their communities. Of all workers, LEOs along with security guards and bartenders experienced the highest rates of non-fatal workplace violence between 2005 and 2009.2 The number of LEO fatalities has been decreasing since the 1970s.3 However, LEOs continue to experience an occupational death rate five times the national average,4 largely due to homicides by suspects being pursued.4 ,5 This may reflect a wish to evade capture, particularly among those committing more serious crimes.6 A previous study of New York City,7 an 11-year analysis of all LEO fatalities in the USA4 and an evaluation of body armour effectiveness8 indicate that LEOs are most often killed with a firearm.

    LEOs face unique challenges due to the nature of their work and response environments. The duties of LEOs range from investigating crimes to initiating traffic stops.5 ,9 LEOs often enter unstable and potentially dangerous situations, especially when responding in areas of high crime, working alone, or interacting with hostile suspects.3 They must make split-second decisions about these dynamic situations affecting the safety of everyone involved.10

    LEOs are often dispatched to calls regarding a ‘disturbance’. A disturbance call is a general term for a law enforcement response that can cover a wide range of incidents. An exploration of the 796 LEO homicides between 1996 and 2010 showed that just over 22% occurred during a disturbance call, the most common primary encounter resulting in a LEO homicide (Swedler et al, unpublished). ‘Domestic disturbance’, ‘domestic dispute’ and ‘domestic violence’ are terms used, often interchangeably, in law enforcement to describe disturbances occurring at a residence or between family members or intimate partners at any location.11 An excerpt from a narrative in the 2010 ‘Law Enforcement Officers Killed and Assaulted’ (LEOKA) report describes one such commonly encountered situation occurring within a home: “a trooper with the Pennsylvania State Police, Franklin, was ambushed in an unprovoked attack shortly before noon on January 13 while responding to a domestic disturbance involving a husband and wife at their home…” (p 26).12 Another excerpt from 2005 describes an event at a location outside the home: “Around 15:30 on August 29, a police officer with the Huntsville Police Department was fatally shot when he responded to a domestic disturbance in progress at a local restaurant” (p 1).13 Domestic disturbances are thus differentiated from other types of disturbance calls, such as out-of-control parties, through the use of the terms domestic disturbance/dispute/violence.

    LEOs often respond to calls concerning domestic violence and intimate partner violence (IPV), potentially unsafe situations LEOs must manage and where they may feel limited in their ability to protect victims.14 Data from the National Crime Victimization survey for 2000–2010 indicate that in the USA between 473 000 and 628 000 incidents of IPV and 166 000–294 000 incidents of violence involving other family members (ie, domestic violence) were reported to the police annually.15 In 2010 alone approximately 1300 IPV-related homicides were committed by a current intimate partner or estranged spouse.16 IPV and other forms of domestic violence may present unique challenges for LEOs responding to these situations which therefore warrant further study. For example, there are no restrictions on gun ownership within the home except for those disqualified by law. As a result, LEOs may be more likely to encounter an armed suspect when responding to an incident at a residence, which could make domestic disturbance calls more dangerous.

    There are few empirical data on the risks that LEOs face when responding to domestic disturbance calls. This presents an opportunity to examine an understudied injury burden among LEOs. The purpose of this descriptive study is to characterise the officer, encounter, perpetrator and victim characteristics specific to domestic disturbance-related LEO homicides in the USA. This research will fill an important gap by generating new data on the circumstances surrounding these encounters to help guide efforts to reduce this burden of occupational fatality among LEOs.

    Methods

    Data source

    This study is part of a larger project that aimed to describe the circumstances of all LEO homicides from 1996 to 2010. The primary source of data was the Federal Bureau of Investigation's yearly report, LEOKA. This report provides details surrounding LEO homicides in the line of duty, excluding the terrorist attacks of 11 September 2001. These data are collected as part of the Uniform Crime Reporting (UCR) process. The UCR programme in each state provides information on crimes, and certain reporting criteria must be met to ensure the data are accurate.17 A narrative is written for every line-of-duty LEO homicide (excluding ongoing investigations) and includes information on LEO demographics, and encounter and perpetrator characteristics.11 Narratives of LEO homicides from 1996–2010 were analysed after extraction from the corresponding annual LEOKA reports, which are publicly available online.18

    Narrative text analysis has been demonstrated to be useful for injury prevention research.19 This technique was used on the LEOKA reports to generate a database for analysis. Two coders entered information using an electronic template which was created based on variables of interest reported in LEOKA summary statistics (eg, suspect's weapon, encounter situations) and current literature (eg, drug/alcohol use, ambush scenarios). LEO homicides from 2003 were analysed using the coding frame to identify issues. A 10% random sample was reviewed by the primary investigator (DIS) to check the accuracy of data entry. Inconsistencies were resolved by DIS who made the final determination, and the coding frame was revised as necessary. Each remaining year was then analysed by one coder.

    Case definition and findings

    The present study examines homicides of LEOs responding to domestic disturbance calls. A total of 452 potential domestic disturbance cases were examined based on the primary and secondary encounter codes. A primary encounter is the first interaction the LEO had with the suspect or the reason for dispatch. A secondary encounter is any resulting interaction with the suspect. For example, a LEO may be dispatched to a disturbance call (primary encounter) and then ambushed by the suspect upon arrival (secondary encounter). Potential cases had a primary encounter code of Disturbance Call, Arrest Call, Ambush, Investigation, Mentally Disturbed Individual, Serving a Warrant, and Other, or a secondary encounter code of Disturbance Call. The other primary encounter codes (Crime in Progress, Handling/Transporting Inmates/Suspects, Vehicle Stop, Vehicle Pursuit/Chase) were not found to contain refer to domestic disturbances. The narrative of each potential case was then re-read by the coder (CK).

    Each potential case was confirmed by CK if the narrative included the term ‘domestic disturbance’ or a domestic disturbance situation was described. Of the 452 potential cases, 336 were excluded, many because the disturbance call was unrelated to domestic disturbance. For example, officers may have been dispatched to a large, out-of-control party and while this incident was considered a disturbance call, no domestic disturbance actually occurred. After these exclusion criteria were applied, 116 confirmed cases where the LEO homicide occurred during a response to a domestic disturbance call (ie, where domestic violence or IPV allegedly occurred) remained for analysis.

    Analysis

    Once the domestic disturbance cases were confirmed, additional variables were captured including suspect/perpetrator demographics, criminal history, relationship to the victim in the domestic disturbance, and the outcome of the incident for the suspect. Descriptive analyses were conducted to generate data on the LEOs killed. All the narratives are de-identified and publicly available. As such, this research was classified as ‘Not Human Subjects Research’ by the Institutional Review Board of the Johns Hopkins Bloomberg School of Public Health.

    Results

    LEO characteristics

    Between 1996 and 2010, there were 116 LEO homicides that occurred while the officer was responding to a domestic disturbance call, accounting for nearly 15% of all LEO homicides during this time. When all LEO homicides in the study period (n=796) are considered, domestic disturbances were the third most common encounter resulting in LEO homicide. The average age of the LEOs killed in these incidents was 38.7 years, nearly all were male, and over 50% worked for a police agency. While 67% of LEOs were wearing body armour when killed, the fatal wound occurred in the head/neck in 52% of cases, and 34% died at the scene (table 1). None of these findings were statistically significantly different from all other LEO homicides.

    View this table:
    Table 1

    Law enforcement officer characteristics

    Encounter characteristics

    The vast majority of primary encounters between a LEO and a suspect were coded as disturbance calls (88%) compared to only 11% (n=680) of all other LEO homicides. LEOs were working alone in 28% of cases. In situations where LEOs responded with a partner or had backup present, 50% of cases resulted in more than one LEO being injured or killed, compared to only 31% for all other encounters (p<0.01). Ninety-four per cent of LEOs were killed with a firearm. These LEOs were significantly more likely to be killed with a long barrelled gun than for all other encounters (46% vs 22%, respectively; p<0.01). Almost 40% of suspects ambushed the responding LEO, and 15% barricaded themselves in their home to prevent arrest. A struggle with the suspect prior to the incident occurred in 30% of cases, and 9% of LEOs were killed by a suspect who used the LEOs service weapon (table 2).

    View this table:
    Table 2

    Encounter characteristics

    Perpetrator characteristics

    Perpetrators had an average age of 34.2 years, nearly all were male, and 27% were under the influence of drugs or alcohol at the time of the incident. Half of the perpetrators were arrested following the incident, 27% were killed by law enforcement, and 23% committed suicide. Sixty-one per cent of suspects had a criminal history mentioned. Although many reports did not document the specific extent or nature of the criminal history, 39% mentioned a history of violent crime, 15% mentioned a history of abuse, and 17% mentioned a previous weapons violation in the narrative (table 3).

    View this table:
    Table 3

    Perpetrator characteristics

    Domestic disturbance victim characteristics

    Of the 116 domestic disturbance-related LEO homicides, 46% (n=53) were committed by a suspect who was perpetrating IPV. Victims of the domestic disturbance were significantly more likely to be killed in IPV-related cases compared with other domestic disturbance calls (21% vs 5%, respectively; p<0.01). A firearm was the most common weapon used in the murder of a domestic disturbance victim: 82% of IPV-related and 100% of other domestic disturbance call victim homicides were committed with a firearm (table 4).

    View this table:
    Table 4

    Domestic disturbance victim characteristics

    Discussion

    This study of 15 years of data from the FBI's annual LEOKA report is among the first to investigate circumstances surrounding the homicides of LEOs while responding to domestic disturbances. This analysis generated information related to the suspects involved in these homicides, and detailed perpetrator and encounter characteristics. Our findings are consistent with previous research on this topic by Johnson20 who described LEO assaults and homicides while the officers were responding to domestic violence calls, and Meyer and Carroll21 who described IPV-related LEO homicides.

    When domestic disturbance-related cases are compared with all other LEO homicides, several findings stand out. When a LEO was working with a partner or had backup, the likelihood of more than one LEO being injured or killed was significantly greater for domestic disturbance cases. Additionally, LEOs responding to domestic disturbance calls were significantly more likely to be shot with a long barrelled weapon, which is consistent with a high proportion these of perpetrators ambushing LEOs. These findings suggest that domestic disturbance calls are often dangerous encounters for LEOs whether working alone or with backup and that further research is warranted to design policies to prevent LEO homicide.

    The domestic disturbance calls in this study involved a wide range of participants and relationships. However, nearly half of the LEO homicides occurred while the officer was responding to an IPV call. A larger proportion of suspects in IPV-related calls were killed by LEOs than in other domestic disturbance-related calls. Additionally, victims of IPV-related calls were also killed in 21% of LEO homicides compared to only 5% of other domestic disturbance-related calls. While other types of encounters, such as traffic stops and robberies, are becoming increasingly dangerous for LEOs, our findings suggest that IPV perpetrators and the circumstances in which LEOs respond may be more dangerous than encounters with other types of offenders. There are laws that allow, or mandate, the removal of firearms from subjects of a protection order and 24 states had such laws in 2004.22 Firearm restrictions for domestic violence restraining order defendants have been shown to decrease domestic homicides.23 ,24 However, victims of IPV may not receive the full benefit of these laws due to poor implementation.25 Adequate enforcement of removal orders for firearms could increase the safety of those experiencing IPV and/or domestic violence, and the LEOs responding to these calls.

    A slightly higher proportion of IPV perpetrators committed suicide after killing a LEO compared to suspects involved in other domestic disturbance calls. Barber et al26 also found that perpetrators of intimate partner homicide were more likely to commit suicide than those killing a non-intimate partner. Consistent with prior studies, this finding could reflect mental health issues, as discussed by Thomas et al who found a higher history of mental health issues among those killing intimate partners versus others.27 The LEOKA narratives did not include information on whether or not perpetrators had mental health issues, and so this could not be confirmed in this analysis. Future research in this area should explore the underlying influence of mental health on these incidents and utilise additional sources of data that capture the presence and type of mental health conditions.

    Overall, 102 (95%) of these LEO homicides and 12 (85%) of the domestic disturbance victim homicides were committed with a firearm. Wright and Wintemute found a higher likelihood of felonies and violent crimes among those with a previous criminal history who bought guns legally.28 This was found in our study as well, 20 perpetrators (17%) were described as having previously committed a weapons violation. While the specific crime previously committed was not detailed, all 20 of these LEO homicides were committed with a firearm. This finding illustrates the need for more effective regulations and enforcement to prevent dangerous individuals from accessing firearms. Additionally, it may be beneficial for LEOs responding to domestic disturbance calls to be told upon dispatch if potential suspects have registered firearms or a history of gun crime; research has shown LEOs are safer when they receive more information from dispatch prior to entering a scene.29 Widespread implementation of such a system would make responding to these scenes safer for LEOs.

    Body armour is becoming standard issue to protect LEOs from injury and death. As of 2010, nearly 75% of law enforcement agencies require the use of body armour and it has proved effective at reducing LEO fatalities; a LEO wearing body armour is 3.4 times more likely to survive a gunshot to the torso than a LEO without body armour.8 The use of body armour does increase survival, but risk is influenced by the type of firearm used and the distance of the shot,30 and body armour will not prevent a fatal shot to other parts of the body, as shown by our finding that 67% of LEOs were wearing body armour when killed. Of those LEOs wearing body armour, 73% received the fatal wound in a location not covered by their armour. More research is needed on ways to extend body armour to provide greater coverage and protection for LEOs.

    Although this research provided insights into the circumstances surrounding the killing of LEOs responding to domestic disturbance calls, there are limitations to this study based on the nature of the data source. Not every LEO homicide is included in the LEOKA narratives as some are excluded due to ongoing investigations at the time of report publication. During our study period 18 homicides mentioned in the LEOKA reports did not have narratives. Therefore, although this analysis included all available cases, the results may have been influenced by missing data. In addition, since LEO homicides are rare, this study was based on a small sample of LEO fatalities and may not be generalisable to all law enforcement.

    Additionally, the LEOKA reports are not generated for research: they contain administrative data collected by the FBI for the purposes of understanding crimes, guiding officer training, and providing better operational and strategic responses to given criminal situations.17 As such, the narratives demonstrated inconsistent reporting of information on perpetrators and domestic disturbance victims. Although the LEOKA report has very strict procedures for submitting fatality statistics, it does not have standardised requirements for homicide narratives. This results in potentially inconsistent provision of information. For example, 35% of the narratives did not discuss the perpetrator's criminal history. As a result, our ability to uniformly describe additional context is limited. There is also some indication that the LEOKA reports undercount the number of LEO fatalities compared to other commonly used databases (Tiesman et al, unpublished data) and thus the data presented here may not reflect the true magnitude of LEO fatalities resulting from domestic disturbance calls.

    This descriptive study has provided new information on officer and encounter circumstances, and perpetrator and victim characteristics of domestic disturbance-related LEO homicides in the USA. These data may help guide future epidemiological analysis, as well as inform fatality prevention efforts. Future research in this area should include longitudinal analyses of LEO safety, including both fatal and non-fatal assaults, to examine trends over time and provide greater understanding of the factors related to injury and death among LEOs.

    What is already known on the subject

    • Law enforcement officers experience an occupational death rate five times the national average.

    • Law enforcement is one of the top three occupations with the highest rates of non-fatal workplace violence.

    What this study adds

    • New understanding of the circumstances of law enforcement homicides that occur while law enforcement officers (LEOs) are responding to domestic disturbance calls.

    • Nearly half of the 116 LEO homicides that occurred while LEOs were responding to domestic disturbance calls were specific to intimate partner violence (IPV).

    • Victims of IPV-related calls were also killed in 21% of LEO homicides compared to only 4% of other domestic disturbance-related calls.

    Acknowledgments

    We would like to thank Ms Molly Simmons for her work on data entry.

    References

    1. ↵
      National Law Enforcement Officers Memorial Fund. Law enforcement facts. Washington, DC, 2012.
    2. ↵
      1. Harrell E
      . Workplace violence, 1993–2009. In: Statistics BoJ, ed. Violence in the workplace series. Bureau of Justice Statistics, Washington, DC, 2011:1–18.
    3. ↵
      1. Fridell L,
      2. Faggiani D,
      3. Taylor B,
      4. et al
      . The impact of agency context, policies, and practices on violence against police. J Crim Justice 2009;37:542–52.
      OpenUrlCrossRefWeb of Science
    4. ↵
      1. Tiesman HM,
      2. Hendricks SA,
      3. Bell JL,
      4. et al
      . Eleven years of occupational mortality in law enforcement: the census of fatal occupational injuries, 1992–2002. Am J Ind Med 2010;53:940–9.
      OpenUrlPubMed
    5. ↵
      1. Clarke C,
      2. Zak MJ
      . Fatalities in law enforcement officers and firefighters, 1992–97. Compensation and Working Conditions 1999; Summer: 3–7.
    6. ↵
      1. Kaminski RJ,
      2. Marvell TB
      . A comparison of changes in police and general homicides: 1930–1998. Criminology 2002;40:171.
      OpenUrlCrossRefWeb of Science
    7. ↵
      1. Kyriacou DN,
      2. Monkkonen EH,
      3. Peek-Asa C,
      4. et al
      . Police deaths in New York and London during the twentieth century. Inj Prev 2006;12:219–24.
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. LaTourrette T
      . The life-saving effectiveness of body armor for police officers. J Occup Environ Hyg 2010;7:557–62.
      OpenUrlCrossRefPubMed
    9. ↵
      1. Waters JA,
      2. Ussery W
      . Police stress: history, contributing factors, symptoms, and interventions. Policing 2007;30:169–88.
      OpenUrl
    10. ↵
      1. McCarty WP,
      2. Zhoa J,
      3. Garland BE
      . Occupational stress and burnout between male and female police officers. Policing 2007;30:672–91.
      OpenUrl
    11. ↵
      FBI. Law enforcement officers killed and assaulted. Washington, DC: United States Department of Justice, 2004.
    12. ↵
      FBI. Law enforcement officers killed and assaulted. Washington, DC: United States Department of Justice, 2010.
    13. ↵
      FBI. Law enforcement officers killed and assaulted. Washington, DC: United States Department of Justice, 2005.
    14. ↵
      1. Horwitz S,
      2. Mitchell D,
      3. LaRussa-Trott M,
      4. et al
      . An inside view of police officers’ experience with domestic violence. J Fam Violence 2011;26:617–25.
      OpenUrlCrossRefWeb of Science
    15. ↵
      BJS. NCVS Victimization Analysis Tool Report: number of violent victimizations by victim-offender relationship and reporting to the police, 1993–2010. Washington, DC: United States Department of Justice, 2012.
    16. ↵
      FBI. Crime in the United States, 2010. Washington, DC: United States Department of Justice, 2011.
    17. ↵
      FBI. Uniform crime reporting handbook. Washington, DC: United States Department of Justice, 2004.
    18. ↵
      FBI. Uniform crime reports. Washington, DC: United Stated Department of Justice, 2012.
    19. ↵
      1. McKenzie K,
      2. Scott DA,
      3. Campbell MA,
      4. et al
      . The use of narrative text for injury surveillance research: a systematic review. Accid Anal Prev 2010;42:354–63.
      OpenUrlCrossRefPubMed
    20. ↵
      1. Johnson RR
      . Officer firearms assaults at domestic violence calls: a descriptive analysis. Police J 2008;81:25–45.
      OpenUrlCrossRef
    21. ↵
      1. Meyer S,
      2. Carroll RH
      . When officers die: understanding deadly domestic violence calls for service. Police Chief 2011;78:24–7.
      OpenUrl
    22. ↵
      1. Frattaroli S,
      2. Vernick JS
      . Separating batterers and guns. Eval Rev 2006;30:296–312.
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Vigdor ER,
      2. Mercy JA
      . Do laws restricting access to firearms by domestic violence offenders prevent intimate partner homicide? Eval Rev 2006;30:313–46.
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. Zeoli AM,
      2. Webster DW
      . Effects of domestic violence policies, alcohol taxes and police staffing levels on intimate partner homicide in large US cities. Inj Prev 2010;16:90–5.
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Webster DW,
      2. Frattaroli S,
      3. Vernick JS,
      4. et al
      . Women with protective orders report failure to remove firearms from their abusive partners: results from an exploratory study. J Womens Health (15409996) 2010;19:93–8.
      OpenUrlCrossRefWeb of Science
    26. ↵
      1. Barber CW,
      2. Azrael D,
      3. Hemenway D,
      4. et al
      . Suicides and suicide attempts following homicide. Homicide Stud 2008;12:285–97.
      OpenUrlCrossRefWeb of Science
    27. ↵
      1. Thomas KA,
      2. Dichter ME,
      3. Matejkowski J
      . Intimate versus nonintimate partner murder. Homicide Stud 2011;15:291–311.
      OpenUrlCrossRefWeb of Science
    28. ↵
      1. Wright MA,
      2. Wintemute GJ
      . Felonious or violent criminal activity that prohibits gun ownership among prior purchasers of handguns: incidence and risk factors. J Trauma 2010;69:948–55.
      OpenUrlCrossRefPubMed
    29. ↵
      1. Fridell LA,
      2. Binder A
      . Police officer decisionmaking in potentially violent confrontations. J Crim Justice 1992;20:385–99.
      OpenUrlCrossRefWeb of Science
    30. ↵
      NIJ. Ballistic resistance of body armor: NIJ Standard-0101.06. NIJ Standards. Washington, DC: United States Department of Justice, 2008.

    Footnotes

    •  Contributors All authors made substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data, participated in drafting the article or revising it critically for important intellectual content, and gave their final approval of the version to be published.

    • Funding This research was supported in part by funding from the NIOSH Education and Research Center for Occupational Safety and Health at the Johns Hopkins Bloomberg School of Public Health (#T42OH008428). This research was also supported in part by funding from a pre-doctoral traineeship in Interdisciplinary Research Training on Violence through a National Institute of Child Health and Human Development grant (T32-HDO64428).

    • Competing interests None.

    • Ethics approval The Johns Hopkins Bloomberg School of Public Health Institutional Review Board approved this study.

    • Provenance and peer review Not commissioned; externally peer reviewed.