Programmatic Impact of 5 Years of Mortality Surveillance of New York City Homeless Populations

Abstract

A homeless mortality surveillance system identifies emerging trends in the health of the homeless population and provides this information to key stakeholders in a timely and ongoing manner to effect evidence-based, programmatic change.

We describe the first 5 years of the New York City homeless mortality surveillance system and, for the first time in peer-reviewed literature, illustrate the impact of key elements of sustained surveillance (i.e., timely dissemination of aggregate mortality data and real-time sharing of information on individual homeless decedents) on the programs of New York City’s Department of Homeless Services.

These key elements had a positive impact on the department’s programs that target sleep-related infant deaths and hypothermia, drug overdose, and alcohol-related deaths among homeless persons.

Homeless individuals have a 1.5- to 11.5- times greater risk of dying relative to the general population, depending on age, gender, shelter status, and morbidity.^1–9 A point-in-time assessment of the number of homeless individuals in the United States found that nearly 650 000 individuals experienced homelessness on a single night in 2010.¹⁰ Roughly two thirds (approximately 400 000) were sheltered. The New York City (NYC) Department of Homeless Services (DHS) has estimated that on any given day at least 49 000 homeless individuals live in NYC, approximately 46 000 of whom are known to be part of the DHS shelter system.¹¹ A DHS annual outdoor population estimate survey completed in January 2012¹² estimated the number of homeless individuals who live in NYC public areas (e.g., on the street, in parks, or in the subway system) to be 3262.

Surveillance, which involves the systematic, sustained collection, analysis, and interpretation of data, can provide insight into the public health challenges faced by a population.¹³ Crucial to surveillance is the dissemination of data to stakeholders in charge of programs aimed at improving health within the population. When applied to homeless mortality, surveillance involves timely, ongoing collection of the circumstances and causes of individual homeless deaths as well as rapid dissemination of surveillance information to key stakeholders who work directly to improve the health of the homeless population. The key elements of surveillance—ongoing data collection and rapid data dissemination—allow for the identification and investigation of emerging trends in causes of death among homeless persons and, thus, facilitate the design of appropriate programmatic interventions. Using surveillance data, agencies serving homeless persons can make evidence-based changes to existing health initiatives and develop new initiatives soon after the emergence of new risks.

In 2005, after heightened media attention on homeless deaths occurring in public view during the preceding winter, the New York City Council mandated NYC’s health department, the medical examiner’s office, Human Resources Administration, and DHS to conduct surveillance of deaths of homeless individuals and report aggregate data on a regular basis.^14,15 Previously, information about the causes and circumstances of deaths in the homeless population was not systematically collected or readily available to stakeholders committed to reducing morbidity and mortality in NYC homeless populations. The limited available data were mainly retrospective.^4,6

In this essay, we describe the NYC homeless mortality surveillance system and illustrate how key elements of the system (i.e., timely dissemination of aggregate mortality data and real-time sharing of information on individual deaths among homeless persons) have had a positive impact on DHS initiatives to prevent sleep-related infant deaths and hypothermia, drug overdose, and alcohol-related deaths among homeless persons. To our knowledge, the impact of a sustained homeless mortality surveillance system on homeless mortality prevention programs has not been documented in the peer-reviewed literature.

NEW YORK CITY HOMELESS MORTALITY SURVEILLANCE

Retrospective analyses of aggregate morbidity and mortality data from a specific study period can identify health problems such as multiple comorbid conditions, substance abuse, or mental illness that result in premature death in a homeless population.^1─9 However, homeless mortality surveillance offers the advantage of ongoing, systematic, and timely data collection and dissemination that reflects the current health status of the homeless population. Ongoing surveillance can identify changing trends in illness and death among homeless populations in close to real time, allowing faster implementation of preventive interventions.⁹

Timely and sustained data collection is achieved through ongoing, scheduled reporting of cases that meet a clearly defined case definition and formal and informal information sharing between reporting agencies. DHS and the medical examiner’s office report information on the deaths of homeless individuals to the health department on a quarterly basis. The definition of a homeless individual used in surveillance is someone who at the time of death did not have a known street address of a residence at which he or she was known or reasonably believed to have resided. DHS reports the deaths of sheltered homeless individuals and any deaths they learn of among unsheltered homeless individuals, including those whom the medical examiner’s office reports to DHS.

The medical examiner’s office investigates suspicious, unusual, violent, or criminal deaths and reports these deaths to the health department if the decedent was homeless.¹⁶ To determine a decedent’s homeless status, medical examiner investigators review the death scene and available information found on or near the decedent, such as identification cards and makeshift housing materials and confer with the NYC Police Department, DHS, and the decedent’s family members. When a decedent is unknown to DHS, homeless status determination is based solely on this death scene investigation, the report of a family member or friend, or both. As with all deaths in NYC, deaths of homeless individuals are certified by a medical examiner or another physician and registered by the NYC health department.

In 2007, the health department and DHS entered into a formal memorandum of understanding that allowed the health department to share medical examiner case information with DHS, which enabled the health department to identify duplicate case reports and resolve discrepancies between DHS and the medical examiner’s office regarding decedent characteristics such as age and shelter status. This review provided DHS with a more complete enumeration of deaths within the population they serve.

In NYC, homeless surveillance activities also encourage continuous information feedback loops between agencies. The process of working together to eliminate duplicates and resolve discrepancies encouraged routine communication between DHS and the medical examiner’s office. In February 2007, medical examiner investigators began contacting DHS’s agency medical director on learning of any decedent thought to be homeless to share investigation findings directly with DHS, including death scene, autopsy, and supplementary reports. DHS uses this detailed case information to evaluate the impact of its programs on the mortality risk of the homeless population and make any necessary changes to programs that could prevent similar deaths in the future. When the medical examiner’s office reports a street homeless death to DHS, DHS outreach teams determine whether the decedent was known to them. Outreach teams then canvass the area where the individual died and offer assistance to other individuals residing at the same location.

In turn, timely, ongoing reporting of cases to the health department facilitates the regular and rapid dissemination of rich surveillance reports to stakeholders. The health department publishes quarterly reports with a 3-month lag (i.e., 3 months after the end of each quarter) and issues annual reports 6 months after the end of each surveillance year. Quarterly reports contain descriptive information such as the number of homeless deaths occurring within the NYC community board district; the decedent’s age, gender, and shelter residency status (i.e., sheltered or unsheltered); the community board district of the shelter in which the decedent resided; and the location of death (i.e., in a hospital, outdoors, in a homeless shelter, or other location). Other locations include a friend or family member’s apartment, public buildings, motel or hotel rooms, building vestibules, abandoned buildings, drop-in centers, or locations within subway or train stations. Annual reports summarize the quarterly reports and also present the frequency of deaths by cause of death among all decedents and by gender and shelter residency.

Table 1 contains leading cause-of-death information for the first 5 years of surveillance, as would be found in an annual report. The categories reflect World Health Organization International Classification of Diseases, 10th revision,¹⁷ codes assigned to the cause of death provided for death registration. Each category includes multiple underlying causes of death.^18,19 The top 5 causes were ranked by frequency across the entire 5-year period. Deaths were further categorized as external on the basis of the code assigned. Natural causes are the result of disease or aging processes. External-cause deaths are the result of injury. External-cause deaths are always determined by the medical examiner’s office and include excessive exposure to natural heat or cold, results of accidents, suicide, assault, legal intervention, operations of war and their sequelae, and complications of medical and surgical care.^18,19

PROGRAMMATIC IMPACTS

Many US jurisdictions, including Seattle, Washington; San Francisco, Santa Cruz, and Los Angeles, California; Atlanta and Fulton County, Georgia; and Philadelphia, Pennsylvania, have conducted homeless mortality surveillance and published surveillance reports.^20–32 Yet, the impact of sustained homeless mortality surveillance on mortality prevention programs and homeless health policy has not been documented in the peer-reviewed literature.

Stakeholders offered continuous access to timely information about deaths among the homeless population can evaluate existing homeless mortality prevention programs and develop new programs. In NYC, DHS is committed and empowered to improve the health of the homeless population through programs and policies. Surveillance findings have demonstrated an increased risk of death from potentially preventable causes among the NYC homeless population. Next, we describe 4 DHS initiatives that were affected by or initiated on the basis of surveillance information: sleep-related infant death prevention, street outreach to prevent hypothermia deaths, drug overdose prevention, and the chronic public inebriate program to prevent alcohol-related deaths.

Infant Safe Sleep Promotion

Infant safe sleeping has been a long-standing priority for DHS. Established risk factors for sleep-related infant deaths include bed sharing, sleep position, and loose bedding. These and other risk factors such as low infant birth weight and gestational age, young maternal age, low maternal education, maternal smoking during pregnancy, and late or no prenatal care increase the risk of sleep-related infant deaths.^33–36 Nationally, prevention programs focus on teaching safe sleeping practices.^37–42

Before homeless mortality surveillance, DHS’s safe sleeping shelter programs focused on educating families through posters, literature, and required viewings of the video “A Life to Love”⁴³ produced by the NYC Administration for Children’s Services. The surveillance system enables DHS to continuously evaluate the safe sleeping program on the basis of the detailed circumstances surrounding each infant death and to identify any necessary changes to the program to prevent future sleep-related infant deaths. Critical to this evaluation is medical examiner case information on each infant death, which the medical examiner’s office has shared with DHS in real time since February 2007. After any infant death, DHS reviews medical examiner case information provided through the surveillance system and conducts a site visit at the affected shelter to assess sleeping conditions and protocol adherence. DHS also reviews safe sleeping principles with all parents of infants younger than 6 months in the affected shelter.

This information sharing has led to several program modifications. In April 2008, real-time information from the medical examiner’s office, supported by surveillance information, identified unsafe sleeping practices involved in infant deaths. These findings prompted several program modifications, including additional face-to-face counseling for infant caregivers at different phases of the family intake process and shelter stay, mandatory weekly room inspections for families with infants younger than 6 months, and formal documentation of noncompliance with safe sleeping protocols. Review of subsequent infant deaths in October 2009 led DHS to initiate an internal agency meeting that is convened after any infant death to review all findings and discuss any needed next steps. This meeting is conducted according to a formal process that assesses whether the safe sleeping program or its implementation could be modified to prevent a similar death.

An additional programmatic response took place in spring 2011 at the end of the fifth surveillance year. DHS, with the medical examiner’s office, the NYC Administration for Children’s Services, the health department, the New York State Center for Sudden Infant Death, and the office of NYC Deputy Mayor, coordinated a joint training on infant safety for 445 family shelter case managers. Before this training, 1 to 6 sleep-related infant deaths were reported annually. Recent surveillance information collected after the training has shown that only 1 sleep-related infant death occurred early in the sixth surveillance year and none occurred during the seventh year.

Hypothermia Prevention by Street Outreach Teams

Surveillance data confirmed the ongoing risk among homeless persons for hypothermia-related death.⁶ DHS issues a “Code Blue” alert when the National Weather Service predicts a temperature below 32° in NYC for at least 4 consecutive hours during an overnight period. During these alerts, DHS increases outreach activities and requires outreach teams to contact high-risk homeless individuals more frequently. Before homeless mortality surveillance, risk factors for hypothermia among unsheltered individuals were nonspecific and culled from the existing literature^3,4,9 and included anyone with a chronic medical condition (i.e., diabetes, heart disease, respiratory condition, severe psychiatric illness) and individuals in vehicles and other exposed areas, making it difficult for outreach teams to increase the frequency of visits given the large number of persons considered to be at high risk.

Results from the first and second surveillance years, spanning from 2005 to 2007, confirmed other health department findings⁶ that, for hypothermia deaths among the NYC homeless population, the risk factors were more specific than previously thought. During this time period, 13 homeless individuals, all of whom were unsheltered, died of hypothermia. Decedents were mostly aged 45 to 64 years (85%; n = 11) and male (85%; n = 11). In 2007, DHS refined its winter street outreach criteria using surveillance data, other health department findings,⁶ and additional information from investigations of hypothermia deaths from the medical examiner’s office. The revised criteria prioritized outreach visits for unsheltered homeless persons who were alcohol dependent, had a known history of heart disease, had a previous cold-weather injury, or were aged 45 years or older, reflecting the characteristics of the individuals who died of hypothermia in previous years. DHS instructed street outreach teams to identify individuals meeting these criteria and encourage them to be transported inside. Since 2007, DHS has used quarterly and annual surveillance reports and real-time information from medical examiner investigations to refine the outreach criteria as needed. For example, on the basis of recent surveillance findings, DHS now recognizes that Hispanic, alcohol-dependent men younger than 40 years are also at increased risk of hypothermia death. Also in 2007, the medical examiner’s office began reporting, in real time, all presumed hypothermia deaths among the homeless population to DHS. Each time a presumed hypothermia death is reported, DHS sends outreach teams to the death scene to engage and assist other individuals who may be at risk for hypothermia.

Recent surveillance information has suggested that DHS’s rigorous efforts to help at-risk homeless individuals come indoors may be effective. In the initial 5 surveillance years, 3 to 7 hypothermia deaths were reported among homeless individuals each year. During the sixth year of surveillance, no hypothermia deaths among homeless individuals were reported for the first time since surveillance began.

Drug Overdose Prevention

First-year surveillance findings showed that drug overdose was the leading cause of death among single adults living in shelters, accounting for 32% (n = 25) of these deaths. In response to the magnitude of this risk, which confirmed results from a previous health department study of sheltered homeless adults,⁶ DHS and the health department began development of a harm reduction protocol in 2006 to be implemented in homeless shelters. The protocol included training single adult shelter and medical staff and DHS police to treat opioid overdoses with intranasal naloxone and thus prevent overdose deaths. For the next 3 years, DHS worked with multiple stakeholders including provider agencies, employee unions, and DHS administration to finalize the protocol.

Subsequent surveillance data showed that drug overdose remained the leading cause of death in this subpopulation, consistent with published findings for other homeless populations (Table 1).^{3,4,6–8,44–46} These data motivated DHS leadership to remove all obstacles and pave the way for the January 2009 pilot of the harm reduction protocol at select shelters. Subsequently, DHS and the health department went on to train more than 300 shelter staff and DHS police to carry out the protocol in homeless shelters. After completing training, DHS identified a protocol implementation challenge: some trainees were uncomfortable with performing resuscitation and administering medication, including naloxone, to reverse drug overdose. DHS police, however, embraced both the training and the implementation. Currently, 300 DHS police are trained, and each class of graduating officers receives certification before deployment. Three shelter providers, accounting for 23% (15 of 65) of single adult shelters in NYC, as well as 2 of 4 providers serving street homeless individuals, have also achieved certification as New York State overdose responders.

TABLE 1—

Leading Causes of Death Among Homeless Decedents: New York City, July 1, 2005–June 30, 2010

Cause of Deatha	Sheltered,b % (No.)	Unsheltered, % (No.)	Total, % (No.)
All-cause deaths
Heart disease	23 (86)	23 (105)	23 (191)
Drug overdose	26 (98)	18 (81)	21 (179)
Accidents except drug poisoning	7 (28)	20 (93)	14 (121)
Alcohol abuse	1 (4)	10 (47)	6 (51)
Assault/homicide	6 (21)	3 (15)	4 (36)
All other causes	38 (143)	26 (120)	31 (263)
Total	100 (380)	100 (461)	100 (841)
External causes (injury related)
Poisoning by psychoactive substance	53 (81)	31 (66)	40 (147)
Assault/homicide	14 (21)	7 (15)	10 (36)
Poisoning by noxious substance	6 (9)	10 (21)	8 (30)
Cold exposure/hypothermia	0 (0)	13 (27)	7 (27)
Suicide	7 (10)	6 (12)	6 (22)
Undetermined intent	9 (14)	12 (26)	11 (40)
All other external causes	12 (19)	21 (45)	17 (64)
Total	100 (154)	100 (212)	100 (366)

Note. Percentage totals may not equal 100 as a result of all percentages being rounded to the nearest whole number.

^aInternational Classification of Diseases, 10th revision,¹⁷ codes used for each cause-of-death category are as follows: heart disease I00-I09, I11, I13, I20-I51; drug overdose F11-F16, F18-F19, X40-X42, X44; accidents except drug poisoning V01-X39, X43, X45-X49, X50-X59, Y85-Y86; alcohol abuse F10; assault/homicide X85-Y09, Y87.1; poisoning by psychoactive substance X40-X42, X44; poisoning by noxious substance X43, X45-X49; cold exposure/hypothermia X31; suicide X60-X61, X63-X84, Y87.0; and undetermined intent Y10-Y34, Y87.2, Y89.9.

^bHomeless decedents were categorized as being sheltered if they lived in a shelter at the time of death or if they had stayed in a shelter for at least 1 night in the past 30 days. Most (69%) of these deaths occurred outside of a shelter.

Harm reduction is also carried out by the health department, which provides medical care and discharge case management for NYC inmates. The health department concurrently expanded its efforts to prevent overdose among persons being released from NYC jails, a group that overlaps substantially with the sheltered homeless population. In the fifth and sixth years of surveillance (July 1, 2009–June 30, 2011), 13 and 12 overdose deaths were reported among homeless individuals, respectively, compared with 20 to 23 in prior years, which suggests these joint programmatic efforts have been successful.

An increase in drug overdose deaths among unsheltered homeless individuals was also recently detected in the surveillance data. As a result, in 2011 and 2012, DHS implemented opiate overdose reversal training for more than 200 street outreach team members and drop-in center and safe haven staff. Safe havens are a transitional housing alternative to shelters. They are smaller than traditional shelters, laid out in private or semiprivate rooms, have few rules, and have no curfew.

Between January 2009 and November 2011, the first 3 years after the initial training of single adult shelter staff, medical staff, and DHS police, 3 overdose reversal attempts were made, 2 by DHS police and 1 by a medical provider. After the 2011–2012 trainings, 8 overdose reversals occurred within a 6-month period, most performed by DHS police.

Chronic Public Inebriate Program

DHS continues to develop new programs in response to surveillance information. Alcohol abuse is a leading cause of death among the unsheltered homeless population, accounting for 10% (n = 47) of deaths in this subpopulation over the initial 5 surveillance years (Table 1). Unsheltered individuals are defined as “chronically street homeless” by DHS if they have lived on the street for at least 9 of any of the previous 24 months. Recently, DHS has partnered with NYC hospital emergency departments to reduce mortality among persons who are chronically street homeless. In a pilot of the chronic public inebriate program, a joint initiative of Bellevue Hospital Center, DHS, and the Goddard Riverside Community Center, Bellevue identified the most frequent emergency department users who were thought to be street homeless and had been diagnosed with at least 1 alcohol-related disorder during an emergency department visit. With the patient’s consent, the hospital and DHS’s outreach teams provided case management and helped place the individual in a stabilization bed or safe haven. The ultimate goal is permanent housing placement, thus improving individuals’ health status and decreasing their risk of death. Preliminary data from the pilot showed a 38% and 35% reduction in hospital emergency department visits and in-patient days, respectively, as well as a reduction in associated costs, for individuals enrolled in the program.⁴⁷ The majority of program participants (79%; 19 of 24) are in transitional or permanent housing.⁴⁷ DHS has expanded this program to 2 additional NYC hospitals.

Surveillance data continue to highlight additional targets for homeless mortality reduction efforts, including accidents among street homeless individuals and heart disease, which has become the leading cause of death among the homeless population as injury-related deaths have decreased. Ongoing analysis of current and future surveillance data may suggest approaches to further reduce morbidity and mortality in the homeless population.

CONCLUSIONS

Surveillance systems collect and disseminate information that informs and helps in targeting interventions. Surveillance is only successful if this information is provided to key stakeholders and decision makers who are motivated to take action and have the resources required to address the needs of the population.⁴⁸ In NYC, a homeless mortality surveillance system provides timely and sustained information about homeless deaths to DHS, the NYC agency specifically tasked with and committed to serving the needs of the homeless population. Surveillance has facilitated information exchange, such as sharing of the medical examiner’s office death scene investigation reports with DHS. Surveillance data allow DHS to monitor emerging trends regarding the causes and conditions of death among the homeless population. DHS uses this information to concentrate resources on developing or modifying policies and programs with the greatest potential to prevent mortality among homeless persons. The DHS accomplishments outlined here and others have led to broad support of continued homeless mortality surveillance from DHS, the health department, and city council.

Policymakers in jurisdictions with large homeless populations should consider implementing similar surveillance systems to improve their local homeless mortality prevention programs. Homeless mortality surveillance could be scaled up or down, depending on the size of the jurisdiction and available resources; the number and type of variables collected can be tailored to diverse homeless populations. In NYC, surveillance uses existing health department, DHS, and medical examiner information. To initiate surveillance, jurisdictions with these data sources already in place would only need 1 part-time analyst to collect the data from the 2 data sources, conduct the matching process, and analyze the data for the reports. For maximum benefit, we recommend that jurisdictions encourage formal stakeholder involvement in the collection and analysis of the data and implement a memorandum of understanding to allow for easy data sharing between agencies when surveillance is initiated.

With active participation across agencies and the resources to prevent deaths among the homeless population, homeless mortality surveillance can drive changes in established policies and practices, spur development of new programs, improve timeliness of communication and information sharing between involved agencies, and allow for dynamic, real-time reaction to individual deaths and mortality trends among homeless populations.