Abstract
In the 1970's and 1980's, there were large inter-village disparities in infant mortality due to congenital anomalies on Guam. A village-level analysis was conducted to determine if these disparities can be explained by behavioral (ie, median age of village females, village fertility ratio), structural (ie, population density, persons per household, single mother households per village, married females per village), and environmental (ie, living in a village where Agent Orange (AO) spraying was conducted) factors. Village-level data for live births and infant mortality due to congenital anomalies (1970–1989) was collected from Guam's Office of Vital Statistics. Data on median age of village females, village fertility ratio, population density, persons per household, single mother households, and married females were obtained from the 1980 US Census. Estimates of village-level AO use were provided through personal communications, and villages were dichotomized into AO and non-AO spray areas. Village location was classified by usual residence of the mother. Linear regression was used to determine associations between infant mortality due to congenital anomalies and the behavioral, structural, and environmental factors. The association between AO spray area and infant mortality due to congenital anomalies was statistically significant under univariable (B [95%CI] = 1.88 [0.64,3.11], P = .005) and multivariable conditions (B [95%CI] = 2.02 [0.08,3.96], P = .042). These results suggest that infants born to mothers whose usual residence was in an AO spray area on Guam are at an increased risk of mortality due to congenital anomalies. Further studies using individual-level data are needed to validate these results.
Keywords: Agent Orange, Guam, Infant, Mortality
Introduction
Guam is a United States (US) territory located in the Western Pacific Ocean. It is the southernmost island of the Mariana Archipelago and is the largest and most populous island in the Micronesian region.1 In the 1970's and 1980's, Guam experienced large disparities in infant mortality due to congenital anomalies that have gone unexplained. While some villages reported no infant deaths due to congenital anomalies, others reported cause-specific infant death rates as high as 5.62 deaths per 1,000 live births, a figure that was two times greater than the overall cause-specific death rate on Guam (2.43 deaths per 1,000 live births) and in the US (2.5 deaths per 1,000 live births).2
There are several possible behavioral, structural, and environmental explanations for these differences. At the behavioral level, mothers in high-risk villages may give birth at older ages compared to mothers in other villages. Several studies have identified advanced maternal age as a significant risk factor for genetic anomalies and stillbirth.3,4 In a review of 142 studies, advanced maternal age was associated with a 2.31–5.46 greater odds of stillbirth, and the odds of stillbirth due to congenital anomalies was found to be 7.5 (OR [95%CI] = 7.50 [3.2, 17.4]).5 Mothers in high risk villages may also have more children than mothers in low risk villages. Statistically significant associations between high fertility rates and child mortality were found in a study of 47 low and middle income countries, although the authors were unable to explain the mechanism that drives this relationship.6
Structurally, community and family dynamics may explain differences in infant mortality due to congenital anomalies. Urban environments have been associated with significant increases in congenital anomalies and infant mortality in Asia and Europe.7–10 For example, congenital anomalies decreased significantly in rural areas of Henan Province in China between 1997 and 2011 (P < .001) but increased significantly in urban locations (P = .003).9 Family dynamics may encompass both the number of individuals living in a household or single parent households. The Particulate Matter and Perinatal Events Research (PAMPER) study discovered that both infant mortality and family size decreased between 1961 and 1992,11 and in a separate study of 28,647 children conducted in Nigeria, small family size was associated with decreased under-5 mortality.12 Moreover, a study of approximately 49,000 children in Cameroon, Nigeria, and the Democratic Republic of the Congo determined that under-5 mortality was significantly, or marginally significantly, higher in children of non-widowed single mothers.13
Certain environmental exposures, such as herbicides, may also be associated with infant death due to congenital anomalies. In 2005 and again in 2013, the Department of Veterans Affairs (VA) concluded that herbicides, particularly Agent Orange (AO), were used on Guam from 1968 to 1970.14,15 AO is a mixture of two herbicides: 2,4,5-trichlorophenoxyacetic acid (2,4,5-T) and 2,4-dichlorophenoxyacetic acid (2,4-D),16 and has been linked to numerous health effects, primarily caused by contamination of 2,5,4-T with 2,3,7,8-tetrachlorodibenzodioxin (TCDD). TCDD has been classified as carcinogenic to humans by the International Agency for Research on Cancer and as a probable human carcinogen by the US Environmental Protection Agency (EPA).17,18 There is also evidence suggesting that AO exposure is associated with ischemic heart disease, Parkinson's disease, and respiratory cancers.19
Infant and fetal AO exposure can occur through paternal and maternal mechanisms, with recent evidence showing breast milk as an important AO vector.20 The potential effects of AO on offspring have been known since 1969 when offspring of mice exposed to contaminated 2,4,5-T were shown to be at an increased risk of developing congenital anomalies.16 A meta-analysis on human AO exposure concluded that AO significantly increased the risk of birth defects, although significant between-study heterogeneity occurred and the authors expressed concern about publication bias.21 Moreover, children of fathers who served in southeast Asia during the Vietnam War and participated in Operation Ranch Hand, the US Department of Defense's defoliation program during the war, had significantly greater risk of infant death compared to children of fathers who served in southeast Asia but did not participate in the operation.22
The reason for the disparities in infant mortality due to congenital anomalies on Guam are currently unknown. The purpose of this study is to determine if behavioral, structural, or environmental factors are associated with village-level infant mortality due to congenital anomalies.
Methods
Total births by village and infant mortality due to congenital anomalies were made available from the Office of Vital Statistics in the Guam Department of Public Health and Public Services for each year from 1970–1989. Infant mortality due to congenital anomalies was determined using ICD-8 and ICD-9 codes 740–759, and the cause specific mortality rate was defined as deaths per 1,000 live births. Village assignment for each birth and death was based on the usual residence of the mother. All villages on Guam were included in the analysis (N = 19); however, births and deaths from mothers whose usual residence was a military base or unknown were excluded because of the inability to assess potential behavioral, structural, or environmental exposure levels and their relatively brief exposure to the Guam environment.
Village-level estimates for behavioral and structural exposures were obtained from the 1980 US Census. Census data was included due to the standardized methodology and completeness of data collection. The 1980 Census was used because data collection occurred near the mid-point of the study period. Median age of village females (in years) and the village fertility ratio (ie, the number of children under 5 years old per 1,000 women 15 to 49 years old) were used to assess behavioral exposures. Population density (persons per square kilometer) was used to control for difference in community dynamics. Persons per household, single mother households (per 1,000 families), and married females (per 1,000 females 15 years old and over) were used to control for differences in family dynamics.
Although the VA has concluded that herbicides and AO were used on Guam, exact levels of AO used were unavailable, and the EPA concluded that there are no historical documents related to hazardous chemical use or disposal on Guam.23 However, a US Air Force Veteran who conducted ground-level AO spraying for vegetation control was able to provide village-level spray estimates based on the recollection of this spray routine (Foster L., Personal Communication). This information was used to dichotomize villages into AO and non-AO spray areas. The Veteran has provided documentation substantiating his claims that he conducted AO spraying on Guam. These include a notarized statement describing the Veteran's use of AO on Guam, a notarized photo identifying AO spray areas from a second Veteran who personally witnessed our Veteran spraying AO, medical records demonstrating that the Veteran has suffered from physical reactions consistent with AO exposure, including severe acne consisting of skin eruptions and scarring,24 images of the Veteran being sworn into active duty with the Air Force, and photos of polydactyly occurring in the Veteran's granddaughter. Additionally, his claims have been reported and confirmed by The Japan Times and the Daily Beast.25,26 All documents made available to the authors regarding the Veteran's AO claims have been turned over to the journal editors.
All variables included in the analysis were examined for symmetry and skewness prior to analysis. No transformations were needed. Villages were classified based on whether the village infant mortality due to congenital anomalies rate was above (high risk) or below (low risk) the overall rate for the island. Chi-square analysis, Fisher's exact test, and independent sample t-tests were used to identify significant differences in the included factors between high and low risk villages. Univariable linear regression models were created to determine unconditional associations between infant mortality due to congenital anomalies and the selected factors. AO spray area was dummy coded and non-AO spray areas were used as the reference group. The remaining independent variables and village infant mortality due to congenital anomalies were treated as continuous variables. R2 values were used to estimate the percent variance in infant death due to congenital anomalies explained. A multivariable linear regression model was created to determine conditional significant associations. Multicollinearity was assessed using variance inflation factors (VIF). Variables with a VIF greater than 10 were examined and removed if significantly correlated with other variables in the model. The analysis was conducted using IBM SPSS Statistics for Windows, Version 21.0 (Armonk, NY, USA), and statistical significant was set at 0.05 a priori.
Results
Eleven villages on Guam were classified as high risk villages and eight were classified as low risk villages (Figure 1). From 1970–1989, there were 121 infant deaths due to congenital anomalies among 49,841 live births on Guam, resulting in a cause-specific death rate of 2.43 per 1,000 live births (Table 1). Congenital anomalies included congenital heart disease, anencephaly, and diaphragmatic hernia. Within high risk villages, the cause-specific death rate was 2.96 per 1,000 live births while the rate was 1.31 per 1,000 live births among low risk villages. The difference between these values was statistically significant (χ2(1) = 11.628, P < .001).
Figure 1.
Villages of Guam Above (High Risk) and Below (Low Risk) Guam's Overall Rate of Infant Mortality due to Congenital Anomalies
Table 1.
Infant Mortality due to Congenital Anomalies (1970–1989) by Village*
| Village | Infant Deaths | Live Births | Infant Mortality Rate (/1,000 Live Births) |
| High Risk | |||
| Asan | 5 | 889 | 5.62 |
| Piti | 5 | 948 | 5.27 |
| Chalan-Pago-Ordot | 8 | 1,905 | 4.20 |
| Agat | 10 | 2,999 | 3.33 |
| Yona | 8 | 2,412 | 3.32 |
| Inarajan | 4 | 1,308 | 3.06 |
| Agana | 3 | 1,106 | 2.71 |
| Santa Rita | 4 | 1,516 | 2.64 |
| Dededo | 29 | 11,257 | 2.58 |
| Tamuning | 17 | 6,635 | 2.56 |
| Yigo | 7 | 2,797 | 2.50 |
| Sub-Total | 100 | 33,772 | 2.96 |
| Low Risk | |||
| Merizo | 2 | 1,022 | 1.96 |
| Barrigada | 6 | 3,620 | 1.66 |
| Mongmong-Toto-Maite | 5 | 3,036 | 1.65 |
| Agana Heights | 2 | 1,413 | 1.42 |
| Mangilao | 4 | 3,278 | 1.22 |
| Sinajana | 2 | 1,926 | 1.04 |
| Talofofo | 0 | 1,271 | 0.0 |
| Umatac | 0 | 503 | 0.0 |
| Sub-Total | 21 | 16,069 | 1.31 |
| Total | 121 | 49,841 | 2.43 |
Village of mother's usual residence
Overall mean median age of village females was 21.7 years (SD = 2.27) and the mean fertility ratio was 468.9 children under 5 years old per 1,000 women 15 to 49 years old (Table 2). There were approximately 367 persons per square kilometer and 4.24 persons per household. There were 90.8 single mother households per 1,000 families, and there were 384.0 females married per 1,000 females 15 years old and over. Twelve villages were classified as AO spray areas. There was a statistically significant association between village type and AO spray area (P = .006). No other statistically significant differences based on village type were detected (P's = .065–.921).
Table 2.
Descriptive Statistics of Independent Variablesa
| Variable | Overall | High Risk Villagesb | Low Risk Villagesb | Te | Pf |
| Median Age of Village Femalesc | 21.7 (2.27) | 22.1 (2.32) | 21.0 (2.18) | 1.05 | .307 |
| Fertility Ratioc | 468.9 (81.0) | 470.4 (101.7) | 466.9 (45.7) | 0.10 | .921 |
| Population Densityc | 366.5 (401.5) | 246.7 (245.7) | 531.3 (524.2) | 1.43 | .187 |
| Persons per Householdc | 4.24 (0.67) | 4.11 (0.69) | 4.42 (0.64) | 1.01 | .326 |
| Single Mother Householdsc | 90.8 (31.6) | 81.9 (29.8) | 103.0 (31.7) | 1.49 | .155 |
| Females Marriedc | 384.0 (51.5) | 402.5 (50.3) | 358.6 (43.9) | 1.98 | .065 |
| AO Spray Villagesd | 12 (63.2) | 10 (90.9) | 2 (25.0) | .006 |
Mean (SD);
Data from the Office of Vital Statistics in the Guam Department of Public Health and Public Services;
Data from the 1980 US Census;
Fisher's exact test was used, no t-statistic was produced;
the T statistic is calculated by dividing the difference between groups by the standard error of the total sample;
P < .05 is considered statistically significant. AO, agent orange.
Univariable linear regression determined that the association between AO spray area and infant mortality due to congenital anomalies was statistically significant (B [95%CI] = 1.88 [0.64, 3.11], P = .005) (Table 3). AO spray area explained approximately 38% of the variance (R2 = 0.38). There were no other statistically significant associations with infant mortality due to congenital anomalies (P's = .111–.400).
Table 3.
Associations between Village-Level Variables and Infant Death due to Congenital Anomalies, Univariable Linear Regression
| Variable | B | 95% CI | Ta | Pb | R2 |
| AO Spray Area | 1.88 | 0.64, 3.11 | 3.21 | .005 | 0.38 |
| Median Age of Village Females | 0.20 | −0.13, 0.53 | 1.30 | .213 | 0.09 |
| Fertility Ratio | −0.004 | −0.01, 0.01 | 0.86 | .400 | 0.04 |
| Population Density | −0.001 | −0.003, 0.001 | 1.14 | .270 | 0.02 |
| Persons per Household | −0.73 | −1.83, 0.37 | 1.41 | .178 | 0.05 |
| Single Mother Households | −0.01 | −0.04, 0.01 | 1.03 | .317 | 0.06 |
| Married Females | 0.01 | −0.003, 0.03 | 1.68 | .111 | 0.14 |
the T statistic is calculated by dividing B by the standard error of B;
P < .05 is considered statistically significant. AO, agent orange; CI, confidence interval.
In the multivariable linear regression model, median age of village females, persons per household, and married females exhibited high multicollinearity (VIF's = 12.17–25.83). Therefore, three multivariable models were created, each using only one of the correlated variables plus three additional covariates (fertility ratio, population density, and single mother households). Under multivariable conditions that included median age of village females (Model 1), the association between AO spray area and infant mortality due to congenital anomalies was statistically significant (B [95% CI] = 2.02 [0.08, 3.96], P = .042), and the model explained approximately 51% of the variance (R2 = 0.51) (Table 4). No other covariates were statistically significant (P's = .244–.687). Similar statistically significant coefficients for AO spray area were produced when median age of village females was replaced by persons per household (Model 2) (B [95% CI] = 1.71 [0.04, 3.39], P = .046) and married females (Model 3) (B [95% CI] = 1.82 [0.08, 3.55], P = .042).
Table 4.
Associations between Village-Level Variables and Infant Death due to Congenital Anomalies, Multivariable Linear Regression
| Variable | B | 95% CI | Ta | Pb |
| Model 1c | ||||
| AO Spray Area | 2.02 | 0.08, 3.96 | 2.25 | .042 |
| Median Age of Village Females | −0.16 | −0.82, 0.50 | 0.53 | .607 |
| Fertility Ratio | −0.008 | −0.02, 0.01 | 1.22 | .244 |
| Population Density | −0.001 | −0.003, 0.002 | 0.78 | .450 |
| Single Mother Households | −0.005 | −0.03, 0.02 | 0.41 | .687 |
| Model 2d | ||||
| AO Spray Area | 1.71 | 0.04, 3.39 | 2.21 | .046 |
| Persons per Household | 0.48 | −2.08, 2.17 | 0.05 | .962 |
| Fertility Ratio | −0.005 | −0.02, 0.01 | 0.82 | .427 |
| Population Density | −0.001 | −0.004, 0.001 | 1.10 | .290 |
| Single Mother Households | −0.003 | −0.04, 0.03 | 0.19 | .852 |
| Model 3e | ||||
| AO Spray Area | 1.82 | 0.08, 3.55 | 2.26 | .042 |
| Married Females | −0.003 | −0.03, 0.02 | 0.26 | .798 |
| Fertility Ratio | −0.01 | −0.02, 0.004 | 1.26 | .230 |
| Population Density | −0.001 | −0.003, 0.001 | 1.25 | .232 |
| Single Mother Households | −0.01 | −0.05, 0.03 | 0.33 | .745 |
the T statistic is calculated by dividing B by the standard error of B;
P < .05 is considered statistically significant;
R2 = 0.51;
R2 = 0.50;
R2 = 0.50. AO, agent orange; CI, confidence interval
Discussion
Infants of mothers whose usual residence was in AO spray areas had an increased risk of mortality due to congenital anomalies compared to infants of mothers whose usual residence was in non-AO spray areas. AO spray area was the only statistically significant predictor of infant mortality due to congenital anomalies under univariable and multivariable conditions. This increased risk is unlikely to be due to overall greater rates of infant mortality due to congenital anomalies on Guam. Indeed, the mean US infant mortality rate due to congenital anomalies between 1970 and 1989 was 2.5 deaths per 1,000 live births, slightly greater than that of Guam found here.2
Significant associations between AO and infant mortality are consistent with previously published literature. In a meta-analysis of 9 peer-reviewed and 13 unpublished studies, parental exposure to AO resulted in a 95% increased risk of birth defects (RR 95% CI) = 1.95 (1.59, 2.39).21 The risk of birth defects was dose dependent, with studies in Vietnamese populations producing greater relative risk estimates (RR (95% CI) = 3.00 (2.19, 4.12) than studies in non-Vietnamese populations (RR (95% CI) = 1.29 (1.04, 1.59). A later meta-analysis of 4 peer-reviewed and 3 unpublished studies determined that the pooled relative risk for spina bifida, a birth defect resulting in the incomplete closer of the spinal column around the spinal cord, after paternal AO exposure was 2.02 (95% CI = 1.48, 2.74).27 Moreover, paternal AO exposure has been previously linked with increased risk of infant death.22
There are two possible explanations for why all other included independent variables were non-significant. First, there may be a high degree of homogeneity within Guamanian communities and in individual behavior patterns. Indeed, at only 571 square kilometers, Guam is the smallest US territory,28 and in 1980, the island's population was only 106,000 individuals.29 Second, non-significant associations may have occurred due to inadequate sample size. Because the analysis was conducted at the village-level, the maximum sample size available was 19, regardless of how many births and infant deaths had actually occurred. This value may have been insufficient to detect true associations between non-significant variables and infant mortality due to congenital anomalies.
There are several limitations to this study. Exact levels of AO spraying that occurred on Guam during the Vietnam War could not be specified and official records likely do not exist.23 The study relied primarily on the recollection of one individual who claimed to be in charge of AO spraying. Although his claims have been verified by others, caution is required because he may harbor significant biases due to suffering from conditions associated with AO exposure. Furthermore, confirmatory laboratory testing to determine whether any individual included in the study was exposed to AO could not be performed, although environmental AO may pose a threat for years after initial spraying due to its long half-life, that is typically 11–15 years in humans and has been observed to be as long as 20 years.30 In the environment, AO's half-life is heavily dependent on soil type and depth of penetration. When exposed to sun, primarily on leaf and soil surfaces, TCDD will break down in 1 to 3 years.31 When under the soil surface or deep within the sediment of rivers and other bodies of water, TCDD's half-life can be more than 100 years.
Village-level data on some potential confounders were identified, although the list is not exhaustive. Data on race, ethnicity, socioeconomic factors, infectious disease rates, maternal nutritional status, and other environmental factors were not included. It is possible that survival could be a function of socioeconomic status because of restricted healthcare access. Healthcare access for many Pacific Islanders has often been limited, with many needing to travel to Hawai‘i or the US mainland to see a specialist. Only recently has healthcare access been expanded in the Pacific Islands, primarily through telemedicine programs.32 Additionally, the reliance on vital statistics data may diminish study validity because such databases may be incomplete and data quality may be poor.33
The study presented here is the first to identify associations between AO and infant mortality in civilian populations outside of Southeast Asia, but it is important to stress that the ecological design of the study makes causal inferences of the study results impossible. The results, however, do propose intriguing hypotheses regarding the impact of AO on Guam, and there are several follow-up studies that are needed to confirm the association between AO and infant mortality due to congenital anomalies. Case-control and retrospective cohort studies need to be conducted to determine if the association remains when individual data is used. These studies should use multiple control groups, including populations from non-AO spray areas on Guam, other Pacific islands where AO was never used or stored, and unexposed areas of the contiguous US. Because of AO's potentially long half-life, environmental studies are also required to determine if AO has persisted in the Guamanian soil, and if present, studies of the current population on Guam area needed to determine if AO continues to be a public health threat.
In conclusion, the results suggest that infants born to mothers who resided in AO spray areas were at an increased risk of infant mortality due to congenital anomalies. This is the first study to examine the link between AO and infant mortality on Guam. Further studies using individual-level data are required to validate the claims of the Veteran and confirm a link between AO and infant mortality on Guam. Additional studies should also determine if excess risk of infant mortality on Guam exists decades after AO spraying ceased.
Acknowledgement
The assistance of Carolyn R. Garrido, Registrar, Office of Vital Statistics, Guam Department of Public Health and Social Services and her staff is gratefully acknowledged.
Conflict of Interest
None of the authors identify a conflict of interest.
References
- 1.Cox LM, Dorn EJ, McIntosh KC, Cook MG, White AH. The Island of Guam. Washington, D.C.: Government Printing Office; 1926. [Google Scholar]
- 2.Rosano A, Botto LD, Botting B, Mastroiacovo P. Infant mortality and congential anomalies from 1950 to 1994: an international perspective. J Epidemiol Community Health. 2000;54:660–666. doi: 10.1136/jech.54.9.660. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Fretts R. Stillbirth epidemiology, risk factors, and opportunities for stillbirth prevention. Clin Obstet Gynecol. 2010;53(3):588–596. doi: 10.1097/GRF.0b013e3181eb63fc. [DOI] [PubMed] [Google Scholar]
- 4.Hassold TJ, Jacobs PA. Trisomy in man. Annu Rev Genet. 1984;18:69–97. doi: 10.1146/annurev.ge.18.120184.000441. [DOI] [PubMed] [Google Scholar]
- 5.Aminu M, Unkels R, Mdegela M, Utz B, Adaji S, van den Broek N. Causes of and factors associated with stillbirth in low- and middle-income countries: a systematic literature review. BJOG. 2014;121(Suppl 4):141–153. doi: 10.1111/1471-0528.12995. [DOI] [PubMed] [Google Scholar]
- 6.Kozuki N, Sonneveldt E, Walker N. Residual confounding explains the association between high parity and child mortality. BMC Public Health. 2013;13(Suppl 3):S5. doi: 10.1186/1471-2458-13-S3-S5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Alonso V, Fuster V, Luna F. Causes of neonatal mortality in Spain (1975–98): influence of sex, rural-urban residence and age at death. J Biosoc Sci. 2006;38(4):537–551. doi: 10.1017/S0021932005026957. [DOI] [PubMed] [Google Scholar]
- 8.Huo K, Zhao Y, Feng H, Yao M, Sävman K, Wang X, Zhu C. Mortality rates of children aged under five in Henan province, China, 2004–2008. Paediatr Perinat Epidemiol. 2010;24(4):343–348. doi: 10.1111/j.1365-3016.2010.01126.x. [DOI] [PubMed] [Google Scholar]
- 9.Xia L, Sun L, Wang X, Yao M, Xu F, Cheng G, Wang X, Zhu C. Changes in the Incidence of Congenital Anomalies in Henan Province, China, from 1997 to 2011. PLoS One. 2015;10(7):e0131874. doi: 10.1371/journal.pone.0131874. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Tromp M, Eskes M, Reitsma JB, Erwich JJ, Brouwers HA, Rijninks-van Driel GC, Bonsel GJ, Ravelli AC. Regional perinatal mortality differences in the Netherlands; care is the question. BMC Public Health. 2009;9:102. doi: 10.1186/1471-2458-9-102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Glinianaia SV, Rankin J, Pless-Mulloli T, Pearce MS, Charlton M, Parker L. Temporal changes in key maternal and fetal factors affecting birth outcomes: a 32-year population-based study in an industrial city. BMC Pregnancy Childbirth. 2008;8:39. doi: 10.1186/1471-2393-8-39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kayode GA, Adekanmbi VT, Uthman OA. Risk factors and a predictive model for under-five mortality in Nigeria: evidence from Nigeria demographic and health survey. BMC Pregnancy Childbirth. 2012;12:10. doi: 10.1186/1471-2393-12-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ntoimo LF, Odimegwu CO. Health effects of single motherhood on children in sub-Saharan Africa: a cross-sectional study. BMC Public Health. 2014;14:1145. doi: 10.1186/1471-2458-14-1145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Department of Veterans Affairs Regional Office, author. Citation Nr: 0527748, Hearing, October 13, 2005 (Docket No. 02-11 819) Boston, MA, USA: 2005. [August 6, 2015]. Available at: http://www.va.gov/vetapp05/Files4/0527748.txt. [Google Scholar]
- 15.Department of Veterans Affairs Regional Office, author. Citation Nr: 1328764, Hearing, September 9, 2013 (Docket No. 11-19 894) Honolulu, HI, USA: [August 6, 2015]. Available at: http://www.va.gov/vetapp13/Files3/1328764.txt. [Google Scholar]
- 16.Buckingham WA. Operation Ranch Hand: The Air Force and herbicides in Southeast Asia 1961–1971. Washington, D.C.: Office of Air Force History, United States Air Force; 1982. [March 18, 2015]. Available at: http://www.afhso.af.mil/shared/media/document/AFD-100928-054.pdf. [PubMed] [Google Scholar]
- 17.Baan R, Grosse Y, Straif K, et al. A review of human carcinogens—Part F: chemical agents and related occupations. Lancet Oncol. 2009;10(12):1143–1144. doi: 10.1016/s1470-2045(09)70358-4. [DOI] [PubMed] [Google Scholar]
- 18.Technology Transfer Network— Air Toxics Web Site, author. 2,3,7,8-Tetrachlorodibenzo-p-Dioxin (2,3,7,8,-TCDD) Washington, D.C.: United States Environmental Protection Agency; 2000. [March 19, 2015]. Available at: http://www.epa.gov/ttn/atw/hlthef/dioxin.html. [Google Scholar]
- 19.Committee to Review the Health Effects in Vietnam Veterans of Exposure to Herbicides, author. Veterans and Agent Orange. Washington, D.C.: Institute of Medicine; 2008. [Google Scholar]
- 20.Hue NT, Nam VD, Thuong NV, Huyen NT, Phuong NT, Hung NX, Tuan NH, Son LK, Minh NH. Determination of PCDD/Fs in breast milk of women living in the vicinities of Da Nang Agent Orange hot spot (Vietnam) and estimation of the infant's daily intake. Sci Total Environ. 2014;491–492:212–218. doi: 10.1016/j.scitotenv.2014.02.054. [DOI] [PubMed] [Google Scholar]
- 21.Ngo AD, Taylor R, Roberts CL, Nguyen TV. Association between Agent Orange and birth defects: systematic review and meta-analysis. International Journal of Epidemiology. 2006;35(5):1220–1230. doi: 10.1093/ije/dyl038. [DOI] [PubMed] [Google Scholar]
- 22.Michalek JE, Rahe AJ, Boyle CA. Paternal dioxin, preterm birth, intrauterine growth retardation, and infant death. Epidemiology. 1998;9(2):161–167. [PubMed] [Google Scholar]
- 23.United States Environmental Protection Agency, author. EPA Superfund Record of Decision. Andersen Air Force Base. 2003. [July 9, 2015]. EPA ID: GU6571999519. Available at: http://www.epa.gov/superfund/sites/rods/fulltext/r0904002.pdf.
- 24.Patterson AT, Kaffenberger BH, Keller RA, Elston DM. Skin diseases associated with Agent Orange and other organochlorine exposures. J Am Acad Dermatol. 2015 doi: 10.1016/j.jaad.2015.05.006. S0190-9622(15)01644-8. [DOI] [PubMed] [Google Scholar]
- 25.Mitchell J. Poisons in the Pacific: Guam, Okinawa and Agent Orange. The Japan Times. 2012. Aug 7, [June 23, 2015]. Available at: http://www.japantimes.co.jp/community/2012/08/07/issues/poisons-in-the-pacific-guam-okinawa-and-agent-orange/#.VehLDZcsAcN.
- 26.Dimond D. Were Vets Who Served in Guam Exposed to Agent Orange and Denied Benefits? The Daily Beast. 2013. Sep 25, [June 23, 2015]. Available at: http://www.thedailybeast.com/the-hero-project/articles/2013/09/25/were-vets-who-served-in-guam-exposed-to-agent-orange-and-denied-benefits.html.
- 27.Ngo AD, Taylor R, Roberts CL. Paternal exposure to Agent Orange and spina bifida: a meta-analysis. Eur J Epidemiol. 2010;25(1):37–44. doi: 10.1007/s10654-009-9401-4. [DOI] [PubMed] [Google Scholar]
- 28.US Census Bureau, author. State Area Measurements and Internal Point Coordinates. Washington, DC: US Department of Commerce; 2012. [August 15, 2015]. Available at: https://www.census.gov/geo/reference/state-area.html. [Google Scholar]
- 29.US Census Burear, author. 1980 Census of the Population, Volume 1 Characteristics of the Population, Chapter C/D: Detailed social and economic characteristics, Part 54: Guam. Washington, DC: US Department of Commerce; 1984. [Google Scholar]
- 30.Armitage JM, Ginevan ME, Hewitt A, Ross JH, Watkins DK, Solomon KR. Environmental fate and dietary exposures of humans to TCDD as a result of the spraying of Agent Orange in upland forests of Vietnam. Sci Total Environ. 2015;506–507:621–630. doi: 10.1016/j.scitotenv.2014.11.026. [DOI] [PubMed] [Google Scholar]
- 31.Brauer J. War and nature: The environmental consequences of war in a globalized world. Rowman & Littlefield. 2009 [Google Scholar]
- 32.Person DA. The pacific island health care project. Front Public Health. 2014;2:175. doi: 10.3389/fpubh.2014.00175. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Carter KL, Rao C, Lopez AD, Taylor R. Mortality and cause-of-death reporting and analysis systems in seven Pacific Island countries. BMC Public Health. 2012;12:436. doi: 10.1186/1471-2458-12-436. [DOI] [PMC free article] [PubMed] [Google Scholar]

