[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
May 2002

An Anomaly Within the Latino Epidemiological ParadoxThe Latino Adolescent Male Mortality Peak

Author Affiliations

From the Center for the Study of Latino Health and Culture, Division of General Internal Medicine and Health Services Research, Departments of Medicine (Dr Hayes-Bautista, Messrs Hsu and Iñiguez, and Mss Hayes-Bautista and Chamberlin), and Family Medicine (Dr Solorio), UCLA School of Medicine (Dr Rico), University of California, Los Angeles.

Arch Pediatr Adolesc Med. 2002;156(5):480-484. doi:10.1001/archpedi.156.5.480

Objective  To describe the anomaly of the Latino adolescent male mortality peak in relation to the overall Latino epidemiological paradox and in relation to the need for new conceptual models describing the health of a culturally diverse population.

Design  Population-based study using California's 1989 to 1997 summary death files for death-related information and the State of California Department of Finance population estimates for population denominators for corresponding years.

Participants  California's general population for 1989 to 1997, including California's 15- to 19-year-old and 20- to 24-year-old populations. In 1997, those 2 age groups numbered 4.3 million.

Main Outcome Measures  Mortality rates for Latinos and African Americans compared with non-Hispanic whites expressed as relative risk (RR).

Results  Overall, the Latino RR of mortality follows the Latino epidemiological paradox in that it is lower (RR, <1.00) than that of non-Hispanic whites for most age groups and both sexes. The anomaly within this paradox is seen in Latino males aged 15 to 19 years (RR, 1.77; 95% confidence interval, 1.55-2.02) and 20 to 24 years (RR, 1.79; 95% confidence interval, 1.58-2.02).

Conclusions  This period of elevated mortality risk is labeled the Latino adolescent male mortality peak, and it is an anomaly within the overall Latino epidemiological paradox.

THERE WERE 4.3 million adolescents, aged 15 to 24 years, in California in 1997.1 Such a large adolescent population, larger than the entire populations of 29 states,2 affords the observation of trends that might be overlooked in smaller adolescent populations. The California adolescent population is not only large, but also experiencing demographic dynamics that make it a bellwether for the adolescent population of the United States in the mid-21st century. In 1997, its ethnic composition was as follows: non-Hispanic white, 44.8%; Latino, 34.2%; Asian/Pacific Islander, 12.1%; African American, 8.2%; and American Indian, 0.7%. It is projected that, by 2020, Latinos will comprise 47.0% of the state's adolescent population.3 With Latinos forming a large and growing percentage of the state's adolescent population, it is imperative to understand their health dynamics. This article shows how the Latino adolescent health dynamic fits into the overall Latino health profile.

Preliminary data from the 2000 census indicate that Latinos outnumber African Americans in the United States, thereby becoming the largest minority population in the country. Many models used in social policy (eg, urban underclass) have been applied to Latinos as if the behavior dynamics of all minority groups could be explained by a single model. In health care, the Race and Ethnic Health Disparities model, which repeats this tendency, has been developed during the past 30 years.

The Office of Minority Health, supporting the federal Race and Minority Health Initiative, cites "compelling evidence that race and ethnicity correlate with persistent, and often increasing, health disparities among US populations,"4 and the Office of Research on Minority Health goes on to specify that "African Americans, American Indians/Alaska Native, Asian and Pacific Islander, and Hispanic citizens suffer poorer health and higher rates of premature death than the majority population."5 In its 1997 position paper,6 the American College of Physicians provided a summary of the relationship between minority status and risk factors, such as poverty, unemployment, and minority group membership in deteriorating inner cities, concluding, "Urban health problems arise from the complex interaction of socioeconomic factors, behavior, environment, and disease that is related to race and ethnicity." In general, low income, low education, and low access to care are causes of poorer minority health.7

Although Latino populations often exhibit these classic risk factors, Latino health profiles have presented a paradox to researchers: despite high-risk factors for poor health—low income, low education, low access to care, and increased likelihood of being overweight—Latino health outcomes generally are similar to, or better than, those for whites.8,9 This observation is supported by evidence from such key health indicators as infant mortality,1012 life expectancy,9,1315 mortality from cardiovascular diseases and major types of cancer, and measures of functional health.8,9 Overall, this phenomenon is observed in California, where, in 1997, the age-adjusted mortality rate from all causes for non-Hispanic whites was 439.2, whereas the rate for Latinos was 327.4, about 25.5% lower. It seems paradoxical that a minority population with such high-risk factors manages to show such good health profiles; thus, this has been labeled the Latino epidemiological paradox (LEP).16,17

Although it has been customary to think of the Race and Ethnic Health Disparities model as applicable to all minority groups, the LEP suggests that this may not be appropriate in every case. Now that Latinos are the nation's largest minority group, it would be prudent to develop new models that can better explain the health profiles of populations that are not non-Hispanic white, covering both the Latino paradox and the African American disparities. These models have yet to be developed, but the field of adolescent health offers some insight.

In general, adolescence is a period of increased mortality. National statistics show that adolescents aged 15 to 19 years have higher mortality rates than younger children.18 As Latinos are becoming the largest portion of the adolescent population in California, their patterns rapidly are becoming the norm for adolescent behavior in general. We wanted to determine if Latino adolescent mortality followed the LEP model or presented some other profile. This article considers Latino adolescent mortality (15-19 years and 20-24 years) as part of Latino mortality during a lifespan (birth to ≥85 years). We found that Latina (female) adolescents in these age groups exhibit the general tendency of the LEP, showing no elevated mortality risk during the late adolescent (15-19 years) and early adult (20-24 years) periods. Because female adolescents conform to the LEP, no further analysis of their mortality rates is presented herein. By contrast, males show a definite countertrend: in the late adolescent and early adult periods, their mortality is nearly twice as high as that of non-Hispanic whites, returning later in life to the levels expected by the LEP. Although mortality rates for the general Latino population conform to the LEP pattern, male adolescents are an interesting anomaly in the paradox.


The primary data sources for this analysis were California's 1997 summary death files19 for death-related information and the State of California Department of Finance population estimates1 for population denominators. Both sets of files include age, sex, and ethnicity identifiers, with Latinos identified separately from non-Hispanic whites, African Americans, and Asian/Pacific Islanders. The deaths and population estimates were aggregated into 5-year age groups (0-4 years, 5-9 years, 10-14 years, and so on) to compute age-specific death rates for males and females separately.

Because the Latino population in California is so large, there were sufficient events in the adolescent 5-year age groupings to allow estimation of significance. During 1997, in the male 15- to 19-year age group, there were 387 non-Hispanic white deaths, 518 Latino deaths, and 140 African American deaths. In the male 20- to 24-year age group, there were 456, 635, and 235 deaths, respectively. The relative risk (RR) for Latinos and African Americans for each age group was computed for 1997 deaths, using non-Hispanic whites as the comparison base, and 95% confidence intervals (CIs) were calculated.20

There were too few deaths in most of these specific age groups to analyze specific causes of death for 1997 alone with reasonable estimates of significance. To examine specific causes of death for each race/ethnic and age group, data were aggregated from the summary death files for the 9-year period of 1989 through 1997 to calculate an annualized rate for a particular cause of death, using the midpoint 1993 population figures for the denominator (because race/ethnic- and age-specific denominators were not available for 1989, 1991, or 1992), then the RR and 95% CI was generated. To track long-term trends for specific causes of death by race/ethnicity, the age groups 10 through 14 years, 15 through 19 years, and 20 through 24 years were combined for each year. Although this aggregation loses the age specificity, it allows a 9-year window to observe trends during that period.


The pattern of Latino male mortality is seen in Figure 1. Before the age of 15 years, the number of deaths among Latino boys is slightly lower than among non-Hispanic whites, but this finding was not statistically significant. In the 15- to 19-year age group, however, the RR for Latinos is nearly twice as high as that for non-Hispanic whites (RR, 1.77; 95% CI, 1.55-2.02). This is the front side of the Latino adolescent male mortality peak (LAMMP). The mortality peak continues into the next age group (20-24 years) in which Latino RR remains nearly twice as high (RR, 1.79; 95% CI, 1.58-2.02). At ages 25 to 29 years, the RR plunges, so that Latino mortality rates are virtually identical to non-Hispanic white rates (RR, 1.12; 95% CI, 1.00-1.26). By this age, the LAMMP has passed. In the age groups 30 to 34 years and 35 to 39 years, the RR for Latino men is slightly lower than for non-Hispanic white men, but this finding is not statistically significant. From the age of 40 years to 90 years or older, the RR for Latino men is significantly lower than for non-Hispanic white men. In contrast, in all age groups until 90 years or older, the RR for African Americans is significantly higher than for non-Hispanic whites.

Figure 1.
Relative risk of Latino and African American mortality to white mortality by age group, 1997. Bars indicate 95% confidence intervals.

Relative risk of Latino and African American mortality to white mortality by age group, 1997. Bars indicate 95% confidence intervals.

African American male mortality seems to follow the Race and Ethnic Health Disparities model. Latino male mortality generally follows a different model, the LEP pattern, but with an important anomaly within the paradox: the LAMMP. Further analysis of the LAMMP for the years 1989 through 1997 revealed that the major causes of deaths were homicides and motor vehicle crashes (MVCs), as given in Table 1. The LAMMP was not observed in suicide rates.

Comparison of Annualized Mortality Rates for Male Latinos and Whites in California, 1989-1997 (per 100 000) Population* 
Comparison of Annualized Mortality Rates for Male Latinos and Whites in California, 1989-1997 (per 100 000) Population*
Comparison of Annualized Mortality Rates for Male Latinos and Whites in California, 1989-1997 (per 100 000) Population*

Latinos consistently had higher rates for homicide than non-Hispanic whites (in ages 10-14 years, RR, 3.95; 95% CI, 3.04-5.13; in ages 15-19 years, RR, 7.71; 95% CI, 6.96-8.54; in ages 20-24 years, RR, 5.34; 95% CI, 4.90-5.82; and in ages 25-29 years, RR, 3.58; 95% CI, 3.28-3.89). The rates for deaths due to MVCs were closer to non-Hispanic white rates, with the Latino rate actually slightly lower in the 10- to 14-year age group (RR, 0.80; 95% CI, 0.66-0.96) and slightly higher in the rest (in ages 15-19 years, RR, 1.06; 95% CI, 0.98-1.15; in ages 20-24 years, RR, 1.20; 95% CI, 1.13-1.28; and in ages 25-29 years, RR, 1.20; 95% CI, 1.11-1.29). Latinos had significantly lower suicide rates than non-Hispanic whites for all age groups.

There have been secular trends in mortality for homicide and MVCs that need to be used as context for the 1997 figures. The rate of male homicides seems to have peaked in the mid-1990s, with a gradual decrease since then (Figure 2). This was most noted in African American males, with a rise from 1989 to 1993, then a marked decrease, so that in 1997 African American male rates were much lower than at the beginning of the decade. Latino rates rose slowly from 1989 to 1992, reached a plateau until 1995, then decreased, approaching the 1989 level. Rates for both African American and Latino adolescents are far greater than those for non-Hispanic whites; at one point (1995), Latinos were more than 9 times as likely to die from homicide, and at another (1993), African Americans were 18 times as likely to die as non-Hispanic whites.

Figure 2.
Male adolescent (age 15-24 years) mortality rates for homicide by ethnicity and year.

Male adolescent (age 15-24 years) mortality rates for homicide by ethnicity and year.

For MVCs, there has been a general decline over time for all races/ethnicities (Figure 3). The sharpest decrease occurred before 1992, with some leveling off and then a less striking decline afterward. Latino rates were the highest overall, about 30% higher than white rates in some years. More recently, African American deaths due to MVCs have dipped below those of whites.

Figure 3.
Male adolescent (age 15-24 years) mortality rates for motor vehicle crashes by ethnicity and year.

Male adolescent (age 15-24 years) mortality rates for motor vehicle crashes by ethnicity and year.

For suicides, the general pattern over time is less clear (Figure 4). Although there has been a slight decline in suicides over the years for non-Hispanic white adolescents, these adolescents are consistently at highest risk overall. Latino suicides seem to be increasing somewhat. African American rates have fluctuated from rising above the white rate to decreasing below the Latino rate, with such variation possibly a result of extremely small numbers of cases (eg, 32 in 1996).

Figure 4.
Male adolescent (age 15-24 years) mortality rates for suicide by ethnicity and year.

Male adolescent (age 15-24 years) mortality rates for suicide by ethnicity and year.

The limitations of the data are those inherent in using any large, public administrative data set. A recent National Center for Health Statistics report,21 providing a summary of current knowledge on the quality and reliability of death rates by race and Latino origin in the official mortality statistics of the United States, concluded that for the non-Hispanic white and African American populations, published death rates were overstated in official publications by an estimated 1.0% and 5.0%, respectively, principally because of undercounts of these population groups in the census. Death rates for Latinos were underestimated by approximately 2.0%. In addition, although there has been nearly 20 years' experience with Latino ethnic coding, there still is the possibility of miscoding. The population estimates also are subject to error, although by using official State of California Department of Finance estimates, this analysis will share the same error as any other analysis using these population estimates. The strength of using these data is that they are comprehensive, covering the entire universe, and hence not subject to the selection bias inherent in small area studies.


Clearly, further study of the LAMMP phenomenon is needed to understand the causes of this striking anomaly in the LEP. Certain risk factors for adolescent mortality already have been identified for the general population, including structural components such as ease of access to firearms, alcohol use, and education levels of both adolescents and their parents, and psychosocial ones such as exposure to violence or substance abuse.2230 Counterbalancing these are certain general protective factors, such as parental presence, positive role models, and involvement in prosocial activities.3032 The role of both sets of factors, risk and protective, in specifically adolescent male Latino culture requires study and analysis. Furthermore, the LEP in general has been attributed to "cultural protective factors,"3335 but these have yet to be defined. Research aimed at identifying and analyzing these factors not only would add to the further understanding of this Latino norm, but also might provide a basis of comparison for identifying how and why adolescent Latino males in California depart from it, only to return to the norm in early adulthood. The patterns of other Latino adolescent populations (eg, Puerto Rican, Cuban) in other states (eg, New York, Texas, Florida) need to be calculated and compared for similarities and differences.

Accepted for publication January 17, 2002.

What This Study Adds

Even though Latino populations often exhibit high-risk factors for mortality (eg, low income, low education, low access to care), they also usually exhibit the LEP (eg, lower age-adjusted mortality). In the general US population, adolescence is seen as a period of increased mortality. This study analyzes, for the first time to our knowledge, the phenomenon of Latino adolescent mortality within the context of the overall LEP. By analyzing the sexes separately, it is seen that the LEP holds for Latina (female) adolescents. In contrast, Latino male adolescents show anomalously higher mortality among the age groups of 15 to 19 years and 20 to 24 years, returning to the LEP pattern after age 25 years. Specific suggestions for further research are provided.

Corresponding author and reprints: David E. Hayes-Bautista, PhD, Center for the Study of Latino Health and Culture, 924 Westwood Blvd, Suite 730, Los Angeles, CA 90024 (e-mail: dhayesb@ucla.edu).

State of California Web site, Department of Finance, Demographic Research Unit, Race/Ethnic Population With Age and Sex Detail, 1970-2040 Sacramento December1998;Available athttp://www.dof.ca.gov/HTML/DEMOGRAP/Race.htmAccessed January 2000
US Census Bureau Web site, United States Department of Commerce.  Population Estimates for States by Race and Hispanic Origin July1 1999;Available athttp://www.census.gov/population/estimates/state/srh/srh99.txtAccessed October 5, 1996
State of California Web site, Department of Finance, County Population Projections With Age, Sex and Race/Ethnic Detail Sacramento December1998;Available athttp://www.dof.ca.gov/HTML/DEMOGRAP/Proj_age.htmAccessed December 17, 2000
US Department of Health and Human Services Web site, Office of Minority Health, Eliminating Racial and Ethnic Disparities in Health. 1998;Available athttp://raceandhealth.hhs.gov/sidebars/sbinitover.htmAccessed August 17, 2000
New NIH Minority Health Initiative Web site. NIH Office of Research on Minority Health., New Web Site Provides Information About the NIH Minority Health Initiative.  April14 2000;Available athttp://www.nih.gov/news/pr/apr2000/od-14.htmAccessed August 10, 2000
American College of Physicians, Inner-city health care [position paper].  Ann Intern Med. 1997;126485- 490Crossref
US Department of Health and Human Services, Health, United States, 1998: With Socioeconomic Status and Health Chartbook.  Hyattsville, Md Centers for Disease Control and Prevention, National Center for Health Statistics1998;DHHS publication (PHS) 98-1232
Hayes-Bautista  DEHsu  PIñiguez  DHayes-Bautista  MStein  RM Health and medicine in the Southwest: a window to medicine in the US in 2050.  Med Am. 2000;14- 12
Hayes-Bautista  DEBaezconde-Garbanati  LSchink  WOHayes-Bautista  M Latino health in California, 1985-1990: implications for family practice.  Fam Med. 1994;26556- 562
Fuentes-Afflick  ELurie  P Low birthweight and Latino ethnicity: examining the epidemiologic paradox.  Arch Pediatr Adolesc Med. 1997;151665- 674Crossref
Fuentes-Afflick  EHessol  NAPerez-Stable  EJ Testing the epidemiologic paradox of low birth weight in Latinos.  Arch Pediatr Adolesc Med. 1999;153147- 153Crossref
Sorlie  PDBacklund  EJohnson  NJRoger  E Mortality by Hispanic status in the United States.  JAMA. 1993;2702464- 2468Crossref
Liao  YCooper  RSCao  G  et al.  Mortality patterns among adult Hispanics: findings from the NHIS, 1986 to 1990.  Am J Public Health. 1998;88227- 232Crossref
Abraido-Lanza  AFDohrenwend  BPNg-Mak  DSTurner  JB The Latino mortality paradox: a test of the "salmon bias" and healthy migrant hypotheses.  Am J Public Health. 1999;891543- 1548Crossref
Los Angeles County Public Health Programs and Services, Vital Statistics of Los Angeles County, 1986.  Los Angeles, Calif Public Health Programs and Services, Data Collection and Analysis Division1989;
Markides  KSCoreil  J The health of Hispanics in the southwestern United States: an epidemiological paradox.  Public Health Rep. 1986;101253- 265
English  PBKharrazi  MGuendelman  S Pregnancy outcomes and risk factors in Mexican-Americans: the effect of language used and mother's birthplace.  Ethn Dis. 1997;7229- 240
Centers for Disease Control, Youth Risk Behavior Surveillance: United States, 1997.  MMWR Morb Mortal Wkly Rep. 1997;47 ((No. SS-3)) 2
State of California, Department of Health Services, ZIP code summary death files, 1989-1997.  Sacramento Vital Statistics Section2000;Diskettes ZIPDTH89.ARJ-ZIPDTH97.ARJ
Bland  JMAltman  DG Statistic notes: the odds ratio.  BMJ. 2000;3201468Crossref
Rosenberg  HMMaurer  JDSorlie  PD  et al.  Quality of death rates by race and Hispanic origin: a summary of current research, 1999.  Vital Health Stat 2. 1999; ((128)) 1- 13
Heide  KM Juvenile homicides in America: how can we stop the killing?  Behav Sci Law. 1997;15203- 220Crossref
Muscat  JE Characteristics of childhood homicide in Ohio 1974-84.  Am J Public Health. 1988;78822- 824Crossref
Singh  GKStella  MY Trends and differentials in adolescent and young adult mortality in the United States, 1950 through 1993.  Am J Public Health. 1996;86560- 564Crossref
Centers for Disease Control and Prevention Web site, 1997 Youth Risk Behavior Surveillance System (YRBSS) Summary.  Last updated December23 1998;Available athttp://www.cdc.gov/nccdphp/dash/yrbs/natsum97/suun97.htmAccessed November 28, 2000
Mayhew  DRDonelson  ACBeirness  DJSimpson  HM Youth, alcohol and relative risk of crash involvement.  Accid Anal Prev. 1986;18273- 287Crossref
Scribner  RAMacKinnon  DPDwyer  JH The risk of assaultive violence and alcohol availability in Los Angeles County.  Am J Public Health. 1995;85335- 340Crossref
Fingerhut  LAIngram  DDFeldman  JJ Homicide rates among US teenagers and young adults: differences by mechanism, level of urbanization, race, and sex, 1987 through 1995.  JAMA. 1998;280423- 427Crossref
Sanders-Phillips  K The ecology of urban violence: its relationship to health promotion behaviors in low-income black and Latino communities.  Am J Health Promot. 1996;10308- 317Crossref
Costa  FMJessor  RTurbin  MS Transition into adolescent problem drinking: the role of psychosocial risk and protective factors.  J Stud Alcohol. 1999;60480- 490
Raine  TRJenkins  RAarons  SJ  et al.  Sociodemographic correlates of virginity in seventh-grade black and Latino students.  J Adolesc Health. 1999;24304- 312Crossref
Taylor-Seehafer  MRew  L Risky sexual behavior among adolescent women.  J Soc Pediatr Nurs. 2000;515- 25Crossref
Black  SMarkides  KS Acculturation and alcohol consumption in Puerto Rican, Cuban-American and Mexican-American women in the United States.  Am J Public Health. 1993;83890- 893Crossref
Marin  GMarin  BVOOtero-Sabogal  RSabogal  FPerez-Stable  E The role of acculturation in the attitudes, norms and expectations of Hispanic smokers.  J Cross-Cult Psychol. 1989;20399- 415Crossref
Marin  BVOFlores  E Acculturation, sexual behavior and alcohol use among Latinas.  Int J Addict. 1994;291101- 1114