Since President John F. Kennedy established. Fatalities in the Peace Corps: A Retrospective Study, 1984 to 2003

Transcription

1 95 Fatalities in the Peace Corps: A Retrospective Study, 1984 to 2003 Nancy M. Nurthen, MPH * and Paul Jung, MD, MPH * Department of Epidemiology, George Washington University, Washington, DC, USA ; Office of Medical Services, Peace Corps, Washington, DC, USA DOI: /j x Background. To determine causes of death for Peace Corps Volunteers (PCV) between 1984 and 2003 and compare them with prior Volunteer death rates and with US death rates. Methods. We conducted a retrospective cohort study of all PCV between 1984 and 2003 and compared them to published data for prior years and against US death rates. Results. Of the 66 deaths in our study period, the major causes were unintentional injury, homicide, medical illness, and suicide. Comparisons to US mortality data controlled for age, marital status, and educational attainment found equal or lower death rates among Volunteers. When compared to previous study results from 1961 to 1983, the total number of deaths, as well as the death rate per Volunteer-year, decreased. Deaths from unintentional injury, suicide, and medical illness decreased in number and rate; only homicides increased in number during our study period, but this increase did not reach statistical significance. Conclusions. PCV are exposed to unique risks, but these risks have become significantly less fatal over the past 20 years when compared to prior Peace Corps data and matched US population data. Since President John F. Kennedy established the Peace Corps on March 1, 1961, more than 187,000 Peace Corps Volunteers (PCV) have served in 139 host countries, living and working with local communities, engaged in projects involving, among other areas, education, business development, and health. 1 The typical Volunteer serves a total of 27 consecutive months within their country of service. PCV begin service at various times in a calendar year depending on the country to which they are assigned and their job category. Living and working in developing countries enables Volunteers to make a difference in the lives of the people they encounter; however, it also may increase their risk of illness, injury, and even death. In 1985, Hargarten and Baker published a study Corresponding Author: Paul Jung, MD, MPH, Office of Medical Services, Peace Corps, Room 5102, th Street, NW, Washington, DC 20526, USA. pjung@peacecorps.gov analyzing the 185 in-service Peace Corps deaths between 1961 and 1983 and found that: 1. Unintentional injuries accounted for 128 (69.2%) of the 185 deaths. 2. Female Volunteer death rates were significantly higher than comparable US death rates. 3. Motorcycle accidents caused 12% of all Peace Corps deaths and a third of all motor vehicle related deaths. 4. Suicide emerged as a leading cause of death, accounting for 13% of all Peace Corps deaths between 1981 and These findings are consistent with other studies of overseas deaths, which show that unintentional injury, specifically motor vehicle injury, is the major cause of death of US citizens overseas 3 5 and that these death rates exceed age-specific death rates within the United States. 6, 7 Our study analyzed all the in-service deaths that occurred among PCV between 1984 and Our objective was to compare Volunteer deaths in our time frame with the previously published data, 2008 International Society of Travel Medicine, Journal of Travel Medicine, Volume 15, Issue 2, 2008,

2 96 comparing causes of death and rates of death. In addition, we compared Volunteer death rates with comparable age and gender-specific US death rates as well as nonnatural overseas deaths of US citizens. Methods Data on the deaths of PCV during service are maintained by the Peace Corps Office of Medical Services. These data are abstracted from death certificates, medical records, and administrative data using a standardized protocol. We analyzed the data on all in-service deaths between January 1, 1984, and December 31, Any data missing from the data set were retrieved from the Volunteer s medical chart, death certificate, and administrative records when available. The Peace Corps keeps data on the number of Volunteers who have served since We obtained official US mortality data from the National Center for Health Statistics Web site. We obtained data on nonnatural deaths of US citizens overseas from the State Department Web site. 8 US mortality data are reported per population unit. To compare Peace Corps mortality rates with US data, we used the total number of Volunteers who served at any time in the Peace Corps over the period of analysis. We calculated standardized mortality ratios (SMRs) of observed deaths divided by expected deaths within a population to compare the number of deaths of Volunteers and of other comparison groups over the 20-year study period. To standardize our mortality ratios, we divided US mortality data into age categories beginning with a lower age limit of 18 years because all PCV must be 18 years of age or older to be eligible for service. We calculated a specific SMR among 20- to 39-year-olds for the 5-year interval of 1998 to 2002 because US data for these years were reported in 10-year age intervals (eg, 20 29, 30 39, etc.), which allowed for robust comparisons with Volunteer data because the majority of Volunteers fall into the 20- to 39-year age category. US mortality data by marital status and education in the years 1998 to 2002 were categorized in age ranges different than those for overall mortality data (ie, 15 24, 25 34, etc.), so we compared SMRs between Volunteers and US statistics for the 25- to 34-year-old age group as this age category included the largest number of Volunteers. Not all Volunteers complete the usual 27-month tour, and a few extend service past their standard 27-month tour of duty. To account for this variation in the denominator of our analysis, we calculated the number of PCV-years by determining the actual Nurthen and Jung number of days served by all Volunteers between January 1, 1984, and December 31, 2003, and dividing by 365. We calculated incidence rates, incidence rate ratios (IRRs), and confidence intervals with Volunteer-years using OpenEpi. 9 Last, we calculated the distribution of time to death from start of in service to determine a Volunteer s average time to death. Results There were 66 deaths of PCV during service between January 1, 1984, and December 31, A total of 71,198 PCV (55.8% female) served during that time frame, and they contributed 121,564 Volunteer-years to our analysis. As a historical comparison, there were 105,539 Volunteers (38.4% female) between the creation of the Peace Corps on March 1, 1961, and December 31, 1983, who contributed 167,639 Volunteeryears. There were 185 deaths in that prior time period for a total of 251 Volunteer in-service deaths between 1961 and The calculated death rate for our time frame is 54.3 deaths per 100,000 Volunteer-years compared to per 100,000 Volunteer-years in the previous study. A comparison between the two study periods shows a statistically significant difference in the incidence rates, with an IRR of (95% CI = ). Causes of Death Figure 1 presents the causes of death among Volunteers between 1984 and Unintentional injuries accounted for the largest number of deaths at 45 (68.2%), followed by homicide, death due to medical illness, and suicide. Table 1 shows all the categories for causes of death, including subcategories of unintentional injuries. Among unintentional Figure 1 Causes of death among Peace Corps Volunteers, 1984 to 2003.

3 A Retrospective Study, 1984 to Table 1 Causes of death among volunteers by age group, 1984 to 2003 Age group Total, n (%) Total unintentional injuries (67) Total motor vehicle accidents (33) Motorcycle 1 1 2/22 (9) Auto /22 (50) Truck 1 1 2/22 (9) Public transport /22 (32) Drowning (9) Plane (6) Bicycle accident (3) Game park accident 1 1 (2) Carbon monoxide poisoning 1 1 (2) (cardiac arrest) Accidental fall (5) Accidental asphyxiation (3) Bee sting 1 1 (2) Accidental gun shot 1 1 (2) Cranial fractures 1 1 (2) Alcohol/drug-related 1 1 (2) Total homicides (17) Total suicide 1 1 (2) Total illnesses (14) Heart disease (8) Cancer (3) Cerebral malaria 1 1 (2) Sepsis 1 1 (2) All totals (%) 22 (33%) 28 (42%) 5 (7.6%) 3 (4.5%) 4 (6.1%) 4 (6.1%) 0 (0.0%) 66 (100) PCV-years by age group 48,338 55,468 5,935 4,353 4,851 2, ,564 Death rate per 100,000 PCV-years PCV = Peace Corps Volunteers. injuries, motor vehicle crashes accounted for 22 of the 45 deaths (48.9%), most frequently in automobiles (11; 24.4%) but also in buses, trucks, taxis, and minibus taxis. For the 11 motor vehicle crashes involving automobiles, only 1 death record confirmed seat belt use; in 5 instances, the death record noted that a seat belt was not used; and in the remaining 5, seat belt use was indicated as unknown. There were only two motorcycle deaths in our study period, which accounted for only 3% of all deaths and less than a tenth of motor vehicle related deaths. Drowning accounted for 6 of the 45 (13.3%) unintentional injury deaths. Four of the death records noted that the Volunteer knew how to swim; ability to swim was indicated as unknown in the remaining records. Homicides accounted for 11 of the 66 (16.7%) deaths. Of the 11 homicides, 6 were women and 5 were men, but the homicide death rate was slightly higher among males than females (9.10 and 9.01 per 100,000 PCV-years, respectively), but this difference was not statistically significant (IRR = 0.828; 95% CI = ). Six of the homicides occurred in Africa, one in Eastern Europe, two in Asia, and two in Latin America. Six of the 11 homicide records mentioned robbery as the possible motive. Nine Volunteers died from medical illness. Heart disease accounted for five of the nine medical illness deaths, two were a result of cancer, and one each was from cerebral malaria and sepsis. There was only one suicide between 1984 and 2003; a 23-year-old male Volunteer with no history of mental illness in his medical records hanged himself in 2003 after serving 23 months in the Peace Corps. Analysis by Gender, Age, and Ethnicity Although males contributed only 45% of the Volunteer-years in our analysis (54,949.1 male Volunteer-years vs 66,613.1 female Volunteer-years), they accounted for 35 (53%) of the 66 deaths. Male Volunteers were 1.2 times more likely to die during service than female Volunteers, but this increased likelihood is not statistically significant (IRR = 1.369; 95% CI = ).

4 98 The number of Volunteer deaths was highest in the 25- to 34-year age category, constituting 28 (42.4%) of the 66 deaths between 1984 and When calculating death rates per Volunteer-year, however, rates were highest in the 65 to 74 category (185 deaths per 100,000 Volunteer-years). Table 1 also shows causes of death by age category. Motor vehicle crashes were the leading cause of death in both the 18- to 24- and the 25- to 34-year groups, while medical illness, specifically heart disease, and unintentional injuries tied as the leading cause in the 35- to 44-year groups and was the leading cause of death in the 65- to 74-year groups. Homicide was the leading cause of death in both the 45- to 54- and the 55- to 64-year groups. Medical illness was the leading cause of death in the 65 to 74 age category. Because applicants to the Peace Corps are not required to indicate their ethnicity, we could not obtain accurate figures for ethnic makeup of the Volunteer population. However, based on medical record and death certificate abstractions, we noted that 56 of the 66 (85%) deaths were among white Volunteers, 4 (6%) were among Hispanic Volunteers, and 2 (3%) deaths each were among Asian and black Volunteers. The remaining two deaths were among Volunteers for whom no ethnicity was noted on the death certificate. Time Distribution of Fatalities With respect to the time distribution of deaths during Peace Corps service, the mean and median time to death were 14 months from the beginning of service, and the standard deviation was 8 months. For unintentional injury deaths, the mean and median were also 14 months, with a standard deviation of 8 months. Homicide deaths occurred with a mean of 15 months, a median of 13 months, and a standard deviation of 8 months. Comparison With Peace Corps Deaths, 1961 to 1983 As stated previously, the overall number of deaths declined significantly in the two study periods (Table 2). Although the number of deaths from unintentional injuries declined significantly from 128 to 45 (IRR = 0.485; 95% CI = ), these deaths accounted for nearly the same proportion of all deaths in the current study as in the first (68.2% vs 69.2%, respectively). Motorcycle accidents were responsible for 22 deaths (11.9%) in the previous Peace Corps study and only 2 (3.0%) deaths in our study. This is a statistically significant decrease (IRR = 0.126; 95% CI = ). Nurthen and Jung The number of deaths due to illness has significantly declined from 40 (21%) in the first study period to 9 (14%) in our study period (IRR = 0.310; 95% CI = ). Homicide ranked second as a cause of 11 deaths in our study, an increase from 7 in the original study. However, the increase in numbers of homicides did not reach statistical significance (IRR = 0.878; 95% CI = ). Although the previous study warned of suicide as an emerging cause of death among Volunteers, only one Volunteer suicide occurred between 1984 and 2003, which is a significant reduction from the 10 suicides recorded in the prior study period (IRR = 0.137; 95% CI = ). Comparison With US Death Rates The overall mortality ratio comparing Volunteer deaths to deaths in the United States between 1984 and 2003 was 0.40, indicating far fewer observed deaths in the Peace Corps than expected when compared to US death rates over our 20-year study period. However, because the majority of US deaths occur in those aged 55 years and older, while the majority of Peace Corps deaths occurred in those aged 55 years and younger [the 20- to 39-year age group accounted for 50 (75.8%) of the 66 Peace Corps deaths], we calculated a specific SMR for the 20- to 39-year-old age group for the years 1998 to For these years, there were fewer deaths than expected for each year and significantly fewer than expected from 1999 to For males in the 20- to 39-year-old age group, there were fewer observed deaths than expected for males in each year (there were no male deaths in 2002) and significantly fewer in 1999 and 2002 ( Table 3 ). For females in the 20- to 39-year-old age group, as with males, there were fewer observed deaths among females in each year, but no year reached a statistically significant difference ( Table 4 ). Because 93% of Volunteers are single, never married, we compared their SMRs with their US counterparts and found significantly fewer deaths than expected for each year from 1998 to Because most Volunteers have completed college degrees, when comparing those in the United States with more than a high school education to all Volunteers, the three deaths in 1998 approximated the expected number of deaths (2.72), but from 1999 to 2002, there were fewer Volunteer deaths than expected. No difference in death rates in the educational attainment comparison reached statistical significance.

5 A Retrospective Study, 1984 to Table 2 Comparison of fatalities between study periods Cause of death No. (%) Rate * No. (%) Rate * Incidence rate ratio 95% CI All unintentional injury 128 (69) (68) Homicide 7 (4) (17) Medical illness 40 (22) (14) Suicide 10 (5) (2) Total 185 (100) (100) * Rate per 100,000 Volunteer-years. The majority of unintentional deaths were among Volunteers between the ages of 18 to 34 years. This correlates with US data, where unintentional injuries were the leading cause of death among 18- to 34-year-olds for each year from to Peace Corps homicide rates during our study period were less than or equal to reported rates worldwide. Data from the United Nations indicate homicide rates of 25 per 100,000 inhabitants of Latin America and the Caribbean, 17 to 20 per 100,000 inhabitants of Africa, 8 per 100,000 in Eastern Europe, and 3 to 4 per 100,000 in Southeast Asia and the Pacific between 1986 and By comparison, the rates found among Volunteers in our study show rates of 2.8 per 100,000 Volunteers in Latin America, 8.4 per 100,000 Volunteers in Africa, 1.4 per 100,000 in Eastern Europe, and 2.8 per 100,000 in Asia. The homicide rate among Volunteers is also consistent with US homicide data, which showed a 25% increase in the US age-adjusted homicide rate between 1985 and In addition, homicide was the second leading cause of death among 15- to 24- year-olds in the United States for each year from to Comparison With US Resident Overseas Deaths There were 1,799 nonnatural, overseas deaths of US citizens reported to the US State Department between October 1, 2002, and December 31, We were not able to find a reliable estimate of a denominator for these data, so we could only make a descriptive comparison. Of the 1,799 nonnatural deaths, 1,128 (63%) were due to unintentional injuries, and 569 of these unintentional injury deaths (50%) were due to motor vehicle crashes. Homicide ranked second in nonnatural causes of death, accounting for 407 of the 1,799 (23%) injury deaths. Suicide ranked third among nonnatural causes of death among US citizens overseas, accounting for 204 of the 1,799 (11%) injury deaths. If we remove illness-related deaths among PCV, the percentage of deaths from unintentional injury is 79% (with motor vehicle related deaths at 49% of all unintentional injury deaths). Deaths from homicide account for 19% of nonillness deaths, and deaths from suicide are 2%. This shows that PCV have a higher proportion of deaths from unintentional injury and lower proportions from homicide and suicide. The proportion of unintentional injury deaths related to motor vehicles remained relatively similar. Discussion The results of our study show that there were relatively few deaths among PCV between 1984 and 2003 when compared to the previous study period of 1961 to Unintentional injury remained the leading cause of death at nearly 70%. Notable reductions in causes of death include those from motorcycle accidents, suicide, and medical illness. Although the number of homicides increased from Table 3 Male SMR for 20- to 39-year-old age groups, 1998 to 2002 Year Male volunteers, age US death rate, age (per 100,000) Expected volunteer deaths Observed volunteer deaths SMR Chi-square p -Value , , , , ,

6 100 Nurthen and Jung Table 4 Female SMR for 20- to 39-year-old age groups, 1998 to 2002 Year Female volunteers, age US death rate, age (per 100,000) Expected volunteer deaths Observed volunteer deaths SMR Chi-square p -Value , , , , , to 11, and the proportion of deaths due to homicide also increased, the increase was not statistically significant. One possible explanation of the significant reduction in motorcycle-related deaths could be related to a change in Peace Corps policy after the first study was published, Peace Corps substantially restricted motorcycle use by Volunteers, required mandatory motorcycle training, and enforced a strict helmet policy for those eligible for motorcycle use. 19 These policies likely contributed to the significant decline in motorcycle-related deaths. The decrease in illness-related deaths may be due to a combination of factors, including improvements in the Peace Corps health care system, careful screening of applicants health conditions, and general improvements in health care worldwide. Compared to US mortality data, overall death rates for PCV are lower than or equal to their agespecific counterparts in the United States. In addition, the age-specific causes of death among PCV are similar to those in the United States. Comparisons of Peace Corps deaths with nonnatural deaths of US citizens living or traveling abroad show a higher proportion of deaths from unintentional injury but lower proportions of homicide and suicide in the Peace Corps population. There are significant limitations to this study. Comparisons of PCV with US counterparts, even those of similar age or educational level or marital status, are limited by differences in the two populations. In general, PCV are in better health than the general US population due to their younger age and the medical standards required for clearance to serve in the Peace Corps. Other factors include tailored training for Volunteers in avoiding hazardous situations and self-monitoring their own health conditions and better access to health care services and providers who are trained and experienced in conditions specific to Volunteers. In addition, there may be an inherent difference between those who pursue and complete Peace Corps service and those who do not, and this difference may result in behavior patterns or other factors that may affect exposure to mortality risk. We are also limited in assessing the effect of ethnicity as a factor in these differences because reporting of ethnic status is optional and not required for Volunteers. PCV undergo extensive health and safety training before their service, including training on motor vehicle safety. This may explain the reduction in the total number of deaths due to motor vehicle injuries, even though the proportion of deaths from this cause remained essentially the same from the previous study period. As noted earlier, only one death record of a motor vehicle crash indicated that a seat belt was used. The potential for misclassification of cause of death on death certificates is well documented. 20 However, the data used for our study were obtained from death certificates as well as from medical records and, when available, written accounts of each Volunteer s death. Thus, misclassification of the causes of death in our study is likely minimized. The average time to death from injury is at a point immediately prior to the completion of their first year of service (the 15-month mark, after 3 months of training and 12 months of service). However, given the standard deviation of 8 months, it is difficult to identify any particular time during a Volunteer s service when fatalities may be more likely, thereby limiting the ability to introduce any possible intervention to coincide with a high-risk period. The Peace Corps is a unique organization, and its work exposes Volunteers to exceptional circumstances and environments that require special training and support. However, the safety considerations for Volunteers do extend to other populations such as expatriates, leisure travelers, and overseas employees of the US government, private corporations, and humanitarian relief organizations. As the Peace Corps progresses through the 21st century, it will likely encounter challenges as it accomplishes its distinctive mission. Our hope is that this study will assist

7 A Retrospective Study, 1984 to 2003 the Peace Corps in continuing to preserve the health and safety of its Volunteers as well as others who choose to travel abroad. Declaration of Interests N. M. N. was a graduate student at George Washington University and P. J. was Chief of Epidemiology for the Peace Corps at the time of this study. The authors state that they have no conflicts of interest. References 1. Fast Facts. What is the Peace Corps? Available at : = learn. whatispc.fastfacts. ( Accessed 2007 Jan 30 ) 2. Hargarten SW, Baker SP. Fatalities in the Peace Corps. A retrospective study: 1962 through JAMA 1985 ; 254 : McInnes RJ, Williamson LM, Morrison A. Unintentional injury during foreign travel: a review. J Travel Med 2002 ; 9 : Baker TD, Hargarten SW, Guptill KS. The uncounted dead American civilians dying overseas. Public Health Rep 1992 ; 107 : Guptill KS, Hargarten SW, Baker TD. American travel deaths in Mexico. Causes and prevention strategies. West J Med 1991 ; 154 : Frame JD, Lange WR, Frankenfield DL. Mortality trends of American missionaries in Africa, Am J Trop Med Hyg 1992 ; 46 : Hargarten SW, Baker TD, Guptill K. Overseas fatalities of United States citizen travelers: an analysis of deaths related to international travel. Ann Emerg Med 1991 ; 20 : U.S. citizen deaths from non-natural causes by foreign country. Reporting period: October 01, 2002 to December 31, Available at : gov/family/family_issues/death/death_594.html. ( Accessed 2005 Feb 23 ) 9. Dean AG, Sullivan KM, Soe MM. OpenEpi: Open source epidemiologic statistics for public health. Updated June 7, Available at : OpenEpi.com. ( Accessed 2007 Jun 17 ) Peace Corps Office of Medical Services. The 2003 Annual Report of volunteer health. The health of the volunteer. Volume 9, 2004 : Hoyert, DL, Kochanek, KD, Murphy, SL. Deaths: Final data for National Vital Statistics Report, Volume 47, Number 19. Hyattsville, MD : National Center for Health Statistics, Available at : 19.pdf. (Accessed 2007 Jan 30) 12. Murphy, SL. Deaths: Final data for National Vital Statistics Report, Volume 48, Number 11. Hyattsville, MD : National Center for Health Statistics, Available at : nvsr/nvsr48/nvs48_11.pdf. (Accessed 2007 Jan 30) 13. Anderson, RN. Deaths: Leading causes for National Vital Statistics Report, Volume 49, Number 11. Hyattsville, MD : National Center for Health Statistics, Available at : nvsr/nvsr49/nvsr49_11.pdf. (Accessed 2007 Jan 30) 14. Anderson, RN. Deaths: Leading causes for National Vital Statistics Report, Volume 50, Number 16. Hyattsville, MD : National Center for Health Statistics, Available at : nvsr/nvsr50/nvsr50_16.pdf. (Accessed 2007 Jan 30) 15. Anderson, RN, Smith, BL. Deaths: Leading causes for National Vital Statistics Report, Volume 52, Number 9. Hyattsville, MD : National Center for Health Statistics, Available at : cdc.gov/nchs/data/nvsr/nvsr52/nvsr52_09.pdf. (Accessed 2007 Jan 30) 16. Anderson, RN, Smith, BL. Deaths: Leading causes for National Vital Statistics Report, Volume 53, Number 17. Hyattsville, MD : National Center for Health Statistics, Available at : cdc.gov/nchs/data/nvsr/nvsr53/nvsr53_17. (Accessed 2007 Jan 30) 17. Shaw M, van Dijk J, Rhomberg W. Determining trends in global crime and justice: an overview of results from the United Nations surveys of crime trends and operations of criminal justice systems. Forum on Crime and Society 2003 ; 3 : US Preventive Task Force. Guide to clinical preventive services. 2nd Ed. Baltimore, MD: Williams & Wilkins, Hynes, NA. Healthwise: a newsletter for Peace Corps Medical Officers worldwide ; IV: Messite J, Stellman SD. Accuracy of death certificate completion: the need for formalized physician training. JAMA ; 275 :