PEDIATRICS Vol. 117 No. 5 May 2006, pp. e932-e939 (doi:10.1542/peds.2005-2078)
Distance From Home When Death Occurs: A Population-Based Study of Washington State, 1989–2002
a Pediatric Advanced Care Team, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
b Health Services Research and Development Center of Excellence, Ann Arbor Veterans Affairs Medical Center, Ann Arbor, Michigan
c Division of Biostatistics and Epidemiology, Children's Hospital of Philadelphia, Philadelphia, Pennsylvania
d Washington State Department of Health, Olympia, Washington
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
OBJECTIVE. For patients who die in hospitals, the regionalization of tertiary health care services may be increasing the home-to-hospital distance, particularly for younger patients whose care is especially regionalized and for whom access to and use of home hospice services remains limited. The objective of this study was to test the hypotheses that the distance from home at the time of death in a hospital has increased over time and is inversely related to the age of the dying patient.
METHODS. A population-based case series was conducted in Washington State of all deaths of state residents from 1989 to 2002. The main outcome measure was driving distance between home residence and location at the time of death.
RESULTS. The overall mean distance from home to the hospital where death occurred has increased by 1% annually. Children who died in hospitals were much farther from home than their adult counterparts: the mean distance was 37.4 km for neonates and 50.9 km for children who were aged 1 to 9 years, compared with 19.9 km for adults who were aged 60 to 79 years and 14.0 km for patients who were older than 79 years. Disparities of distance were even greater among patients who were at the 90th percentile for distance (85.6 km for neonates compared with 30.8 for patient who were older than 79 years).
CONCLUSIONS. The distance between home residence and the hospital where death occurs is greatest for children and has increased over time. Both of these findings have implications for the design of local and regional pediatric end-of-life supportive care services.
Key Words: end of life • palliative care • hospitalization • hospice care • geography
Despite growing awareness and acceptance of hospice and the possibility of dying at home, most Americans die in hospitals.1,2 Of the many features of dying in the United States that can make the process isolating for the patient and his or her loved ones,3,4 dislocation from home could have myriad undesirable consequences. Greater distances between the site of death and the home community can exacerbate estrangement from customary social and spiritual supports while imposing heavier burdens on family and friends regarding visitation before death and transportation of the body home after death. Patients value having loved ones at their bedside, yet greater distance may diminish this occurrence. Family and friends who do visit from farther distances incur greater travel, housing, and meal expenses; may experience greater difficulties caring for other family members who remain at home; and may encounter a greater likelihood of losing their jobs.
Despite the potential implications of increased distances between homes and the hospitals where patients die, little is known about this aspect of dying in the United States. Conceivably, home-to-hospital distance may be increasing as a result of the regionalization of specialized tertiary care, which has been especially pronounced for pediatric care. Although regionalization of tertiary medical care services can concentrate patients in high-volume centers and thereby improve outcomes for both children and adults,5–7 an increase in the distance from home would represent a potential burden or barriers imposed by regionalization.8,9 Although data specifically regarding the impact of distance from home on dying patients and family members is lacking, distance from home has been identified as a leading barrier to parents' visitation of children who are enrolled in out-of-home mental health care services,10 whereas greater distance of adult daughters and sons from their aged parents is associated with substantial financial expenditures and difficulty meeting employment obligations, contributing to significant personal stress.11
If distance from home does indeed impose these and other burdens, then both variation in home-to-hospital distances and a trend of increasing home-to-hospital distance over time would have implications for the supportive care services that hospitals should provide for dying patients and bereaved loved ones. Such services might seek to mitigate burdens (eg, by improving, the availability of short- and long-term housing near the hospital for family members, by providing financial assistance for the transport of bodies back to home towns for burials) or minimize barriers to the receipt of quality care (eg, by enhancing hospital-based spiritual care or other culturally specific services for patients and families who are far removed from their communities, or by establishing liaison services between hospitals and local hospice agencies to facilitate the transfer of end-of-life care from hospital to home).
We therefore sought to measure for all deaths that occurred in Washington State from 1989 to 2002 the distance from home to the site where death occurred (including deaths that occurred at home, in nursing facilities, in hospitals, and in emergency departments) and then to analyze specifically how the home-to-hospital distance varied across age groups. We specifically hypothesized that the home-to-hospital distance (1) has increased over time, (2) has been greatest for the youngest patients and declined across age groups (as a result of the combined effects of greater regionalization of pediatric care and greater motivation to travel farther for tertiary care in younger patients), and (3) has been greater in larger hospitals that serve as regionalized care centers.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
After obtaining institutional review board approval, we conducted a retrospective, population-level case series study of all deaths of Washington state residents that occurred in Washington State during the years 1989–2002.
Data Regarding Decedents' Characteristics
We used death certificate information that is held by the Washington State Department of Heath. The data files specified the street address of decedents, as well as the location of death as either home or a specific facility. For deaths that occurred at home or in nursing facilities, the distance from home was defined as 0. For all deaths that occurred in hospitals or emergency departments, we geocoded 93% of home addresses and 100% of facility addresses to street-level accuracy using geocoding supplied by the Washington State Department of Health, derived from tax parcels and census TIGER line files. Home addresses that were not geocoded through this process were geocoded manually using TransCAD geographic information system software (Caliper Corp, Newton, MA). With these combined techniques, >95% of the death records were geocoded to street-level accuracy.
We next used TransCAD to calculate driving distances between decedents' home residences and the hospitals where death occurred. We did not consider either alternative modes of transportation (eg, air flight, ferry crossings), which would have reduced the travel distance for some decedents, or possible road closures as a result of inclement weather or the use of public transportation, which would have increased some travel distances. For 0.12% of records for which the route could not be determined, we calculated the great circle distance between home residence and hospital. For the 5% of cases of decedent hospital death that could not be geocoded, we imputed the driving distance on the basis of a multivariate technique (implemented using the SAS-based IVEware program; SAS Institute, Cary, NC) that fit a sequence of regression models, drew values from the corresponding predictive distributions, and calculated the mean imputed value.12 We used data regarding decedents' home residence ZIP codes to classify each decedent into an ordinal scale of 10 rural-urban commuting area codes, which classifies settlements on the basis of the community's size and the direction of primary commuting flow.13,14
Hospital Characteristics
For each hospital, we used data from the American Hospital Association to determine the number of available hospital beds in 1996. We used data from the Washington State Department of Health to classify the hospital as rural or urban.
Statistical Analysis
We conducted our analyses using SAS 9.1 (SAS Institute) and Stata 9.0. (Stata Corp, College Station, TX) software. After univariate and bivariate analyses, we tested our 3 hypotheses regarding increased distance from home to the hospital where death occurred (1) over time, (2) for younger compared with older patients, and (3) among larger compared with smaller hospitals by fitting a linear regression model with the log of distance as the response variable and (1) the year (centered), (2) the patient age (in years), or (3) the number of hospital beds as the predictor. We tested our hypotheses using the log-transformed distance (with the level of statistical significance set a priori at 
= .05) but present our findings regarding mean distances based on linear models that regressed on the untransformed driving distance. The lines graphed in Figs 3 and 4 were fitted on the mean values for age (Fig 3) and year (Fig 4) using fractional polynomial models, with the corresponding 95% confidence interval for Fig 3.
|
|
Because the impact of home-to-hospital distance on a patient and his or her family members may increase as distance increases, the mean and the median of the distance are not necessarily the primary parameters of interest when comparisons are made, because cases that are farther on the distribution of distances may be of proportionally greater concern. We therefore also examined the 75th and 90th percentiles of driving distances among all cases. The technique of quantile regression permits estimates of effect size to be made at specific percentiles of the distribution of a response variable15–17; we used quantile regression models that were composed of the same covariates that were used in the linear regression models to calculate adjusted parameter estimates of distance at the median, 75th, and 90th percentiles of driving distances with adjustment in these models for year and decedents' age.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Site of Death
Between 1989 and 2002 (Table 1), a total of 511791 Washington State residents died in hospitals (37.1%), in nursing facilities (31.3%), at home (28.8%), or in emergency departments (2.8%). During this time interval, the proportion of hospital deaths declined from 44.2% in 1989 to 35.4% in 2002, whereas the proportion of deaths that occurred at home rose from 24.1% in 1989 to 31.1% in 2002 (Fig 1, top).
|
|
Most neonates who died did so in hospitals, with the percentage averaging 92.9% during the study period (Fig 1, bottom). The percentage of infant deaths that occurred in hospitals increased from 44.2% in 1989 to 49.7% in 2002 (P < .01 for the trend over all years), whereas the percentage of hospital deaths declined for decedents who were aged 20 to 79 years from 43.5% in 1989 to 34.8% in 2002 (P < .01). For children who were aged 1 to 19 years, the percentage who died in the hospital averaged 53.3% during the study interval with no significant trend.
Distance From Home
The mean distance from home at the time of death diminished only by 1.3 km, from 8.9 km in 1989 to 7.6 km in 2002 (P < .001 for the trend over time; Fig 2, top). This decrease was moderated in part by a countervailing 1.3-km increase in the mean distance from home among patients who died in hospitals, rising from 18.9 km in 1989 to 20.2 km in 2002 (P < .001 for the trend over time; Fig 2, bottom).
|
Age and Distance From Home
Among all deaths, the distance from home declined substantially with age. Among deaths that occurred in hospitals, the mean distance increased during infancy and early childhood, then declined steadily from adolescence onward (Fig 3). This pattern appeared at the mean, median, 75th, and 90th percentiles of the distribution of distance from home (Table 2).
|
Scrutinizing the age-based variation in the change in distance from home over time for all deaths, the distance from home increased for deaths of infants who were aged between 1 and 11 months (P = .02). Among deaths that occurred in hospitals, distance from home likely increased for decedents who were aged 1 month through 19 years (P = .1 for each age group, which is above the preset level of statistical significance) and increased a smaller, absolute amount but to a statically significant degree for older decedents (P < .01 for each of the older age groups; Fig 4).
Adjusting for decedents' ages, the mean distance from home to hospital increased from 1989 to 2001 by 0.23 km each year (Table 3). Of note, patients whose distances from home were greater than the overall mean distance experienced larger annual increases (Table 3), such that decedents at the overall 90th percentile of distance from home experienced an increase of 0.5 km each year.
|
Hospital Characteristics and Distance From Home
Large hospitals with 350 or more available beds had substantially greater distances across their hospital-to-home distance distributions (Table 4). Patients who died in urban hospitals were moderately farther (3-km difference of mean distance) from home than patients who died in rural hospitals.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The provision of quality palliative, end-of-life, and bereavement care to dying children and their families confronts various challenges that distinguish pediatric from adult care.4 This study greatly clarifies 1 of these challenges. Proximity to home is regarded as an important aspect of the experience of dying.3,4,18 Investigating the experience of Washington State decedents during a 13-year interval, we document that patients who die in hospitals are increasingly distant from home and that the distance is greatest for pediatric decedents. This unique population-level perspective, made possible through the use of geographic information systems and the resulting precise measurements of home-to-hospital distances, provides support for our hypotheses that (1) home-to-hospital distance has increased over time, (2) children die farther from home than adults, and (3) larger hospitals care for dying patients who are farther from home.
The results of this study should be evaluated with 2 caveats in mind. First, because we analyzed the distance from home at the time of death only for Washington State residents, excluding decedents from other states or countries, the data presented in this article underestimate the true home-to-hospital distances, but the omission of out-of-state decedents is unlikely to bias our main findings regarding the 3 hypotheses. Second, whether these findings hold true in settings outside Washington State or during other time periods remains to be evaluated. Each of this study's 3 main findings suggests a set of mechanisms that warrant additional investigation:
Temporal Trend
We postulated that the home-to-hospital distance was increasing over time as a result of the increasing regionalization of tertiary medical care. Although our results support this observation, the annual increase in distance alternatively could reflect migration over time of Washington State residents away from the most densely urban settings, where the majority of hospitals are located. This hypothesis is supported by the increasing degree of rurality observed among decedents over time (with a small but statistically significant 0.24 increase in the 10-point urban-to-rural code of decedents' home residences noted during the 14-year study interval; these data are not shown elsewhere).
Deaths among infants (1–11 months), unlike other age groups, experienced a statistically significant increase in the overall distance from home for all deaths. One possible explanation stems from the reduction in sudden infant death syndrome as a cause of mortality during the 1990s, because many of these deaths occurred at home.19
Age Effect
Three mutually compatible and potentially complementary mechanisms may underlie the substantial age-related differences in home-to-hospital distances. We had hypothesized (first) that regionalization of hospital care was considerably more pronounced for pediatric than adult care, but the continued diminishment of home-to-hospital distances over the adult age range suggests (second) that as patients age, they undergo a gradual process whereby they or the surrogates who make decisions for them are increasingly less willing to travel a given distance to obtain hospital care or are more likely to return home for hospice care than are younger patients. This process could be based on the adaptation and development over the life span of patient values and preferences, or the process could be driven (third) by increasing ease of access to community-based palliative and hospice services as people age.
Hospital Characteristics
The largest hospitals in this study were substantially more distant from dying patients' homes than were smaller hospitals, likely as a result of these hospitals' role as regional referral centers that care for particularly complicated and sickly patients over a larger catchment area. Left unresolved by this study but warranting additional investigation is whether the greater average home-to-hospital distance exhibited by the largest hospitals is attributable to patients' (or family members') electing to travel greater distances to obtain care at these institutions or to patterns of patient triage and transfer from an emergency department or a hospital to another, more distant hospital (with less influence exerted by the patient or the family).
The findings of this study should motivate hospitals and the health care system to begin to investigate the consequences of these greater home-to-hospital distances and respond with enhanced or targeted psychosocial support for families away from home. Some hospitals, especially those that provide care for children and those with more hospital beds, need to pay proportionally more attention to the burdens or barriers that are imposed by increased distances from home on dying patients and their families and caregivers. After the consequences of farther home-to-hospital distances are delineated better, various services should be tested rigorously to determine whether they ameliorate distance-related difficulties. Such services might include (1) improving the quality, availability, and affordability of transportation, housing, and meals near the hospital; (2) facilitating, if desired by the patient or the family, more timely and increased access to spiritual care services; (3) delivering culturally sensitive supportive care services for people of diverse backgrounds; (4) creating hospital-based hospice liaisons who could meet patients and families in the hospital so that the option of dying at home can be explored more thoroughly; and (5) expanding community-based supportive care resources so that palliative, end-of-life, and bereavement services are located wherever dying people and their families live.
|
ACKNOWLEDGMENTS |
|---|
Support for this study was provided in part by the Life Sciences Values and Society Program at the University of Michigan. Dr Feudtner was supported by grant KO8 HS00002 from the Agency for Health Care Research and Quality. Dr Silveira was supported by The Robert Wood Johnson Generalist Scholars Program and a Career Development Award from the US Department of Veterans Affairs.
We thank the Washington State Department of Health, especially Craig Erickson of the Department of Information Services for geocoding the death certificates; Ann Lima, Phyllis Reed, and John Whitbeck, PhD, Center of Health Statistics, for providing high-quality data and excellent documentation; and Richard Ordos, Hospital and Patient Data System Services, for hospital facility information. Caliper Corp (Newton, MA) greatly facilitated the use of their TransCAD program for the geographic information systems analysis. We are grateful for the critique offered by David A. Asch, MD.
|
FOOTNOTES |
|---|
Accepted Oct 24, 2005.
Address correspondence to Chris Feudtner, MD, PhD, MPH, General Pediatrics, 3535 Market St, Room 1523, Children's Hospital of Philadelphia–North, 34th and Civic Center Boulevard, Philadelphia, PA 19104. E-mail: feudtner{at}email.chop.edu
Drs Feudtner, Silveira, and Hoskins jointly conceived and oversaw the conduct of the study. Dr Hoskins obtained the data and conducted the geographic information systems calculations of distance. Dr Feudtner and Ms Shabbout performed the analyses. For ensuring confidentiality of information, as specified in the study's institutional review board agreements, all geocoding and calculation of distances were performed within the Washington State Department of Health by Dr Hoskins. Dr Feudtner had full access to the calculated distances and the de-identified death certificate data used in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr Feudtner drafted the manuscript, and Drs Silveira, Hoskins, and Ms Shabbout assisted in revision for intellectual content. Drs Feudtner and Silveira obtained funding for the conduct of the study. All authors approved this version of the manuscript for publication.
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
- Weitzen S, Teno JM, Fennell M, Mor V. Factors associated with site of death: a national study of where people die. Med Care. 2003;41 :323 –335[CrossRef][Web of Science][Medline]
- Hansen SM, Tolle SW, Martin DP. Factors associated with lower rates of in-hospital death. J Palliat Med. 2002;5 :677 –685[CrossRef][Medline]
- Field MJ, Cassel CK, Institute of Medicine (U.S.). Committee on Care at the End of Life. Approaching Death: Improving Care at the End of Life. Washington, DC: National Academy Press; 1997
- Field MJ, Behrman RE, Institute of Medicine (U.S.). Committee on Palliative and End-of-Life Care for Children and Their Families. When Children Die: Improving Palliative and End-of-Life Care for Children and Their Families. Washington, DC: National Academy Press; 2003
- Birkmeyer JD, Lucas FL, Wennberg DE. Potential benefits of regionalizing major surgery in Medicare patients. Eff Clin Pract. 1999;2 :277 –283[Medline]
- Birkmeyer JD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. Surgeon volume and operative mortality in the United States.
N Engl J Med. 2003;349
:2117
–2127
[Abstract/Free Full Text] - Tilford JM, Simpson PM, Green JW, Lensing S, Fiser DH. Volume-outcome relationships in pediatric intensive care units. Pediatrics. 2000;106 (pt 1):289–294
- Finlayson SR, Birkmeyer JD, Tosteson AN, Nease RF Jr. Patient preferences for location of care: implications for regionalization. Med Care. 1999;37 :204 –209[CrossRef][Web of Science][Medline]
- Birkmeyer JD, Siewers AE, Marth NJ, Goodman DC. Regionalization of high-risk surgery and implications for patient travel times.
JAMA. 2003;290
:2703
–2708
[Abstract/Free Full Text] - Kruzich JM, Jivanjee P, Robinson A, Friesen BJ. Family caregivers' perceptions of barriers to and supports of participation in their children's out-of-home treatment.
Psychiatr Serv. 2003;54
:1513
–1518
[Abstract/Free Full Text] - Miles Away: The MetLife Study of Long-Distance Caregiving. Bethesda, MD: National Alliance for Caregiving; 2004. Available at: www.caregiving.org/pubs/data.htm. Accessed October 19, 2005
- Raghunathan TE, Lepkowski JM, Van Hoewyk J, Solenberger P. A multivariate technique for multiply imputing missing values using a sequence of regression models. Surv Methodol. 2001;27 :85 –95
- Morrill R, Cromartie J, Hart G. Metropolitan, urban, and rural commuting areas: toward a better depiction of the United States settlement system. Urban Geography. 1999;20 :727 –748[Web of Science]
- Economic Research Service, United States Department of Agriculture. Measuring rurality: rural-urban commuting area codes. Available at: www.ers.usda.gov/Briefing/Rurality/RuralUrbanCommutingAreas/. Accessed October 19, 2005
- Koenker R, Bassett GJ. Regression quantiles. Econometrica. 1978;46 :33 –50[CrossRef]
- Gould W. Quantile regression with bootstrapped standard errors. Stata Tech Bull. 1992;9 :19 –21
- Koenker R, Xiao Z. Inference on the quantile regression process. Econometrica. 2002;70 :1583 –1612[CrossRef]
- Lunney JR, National Cancer Policy Board (U.S.), National Research Council (U.S.). Division on Earth and Life Studies. Describing Death in America: What We Need to Know. Washington, DC: National Academies Press; 2003
- Feudtner C, Christakis DA, Connell FA. Pediatric deaths attributable to complex chronic conditions: a population-based study of Washington State, 1980–1997. Pediatrics. 2000;106(pt 2) :205 –209
PEDIATRICS (ISSN 1098-4275). ©2006 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
