Cancer in an incarcerated population

Authors

  • Paul Mathew M.D.,

    Corresponding author
    1. Department of Internal Medicine, The University of Texas Medical Branch, Galveston, Texas
    Current affiliation:
    1. Department of Genitourinary Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    • Department of Genitourinary Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Unit 427, 1515 Holcombe Boulevard, Houston, TX 77030
    Search for more papers by this author
    • Fax: (713) 745-1625

  • Linda Elting D.P.H., M.P.H.,

    1. Section of Health Services Research, Department of Biostatistics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Catherine Cooksley Dr.P.H., M.P.H.,

    1. Section of Health Services Research, Department of Biostatistics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Steven Owen Ph.D.,

    1. School of Nursing, University of Texas Medical Branch, Galveston, Texas
    Search for more papers by this author
  • James Lin M.D.

    1. Department of Internal Medicine, The University of Texas Medical Branch, Galveston, Texas
    Current affiliation:
    1. Hematology Oncology Associates, PC, Albuquerque, New Mexico
    Search for more papers by this author

  • Presented in part at the 38th annual meeting of the American Society of Clinical Oncology, Orlando, Florida, May 18–21, 2002.

Abstract

BACKGROUND

The challenge posed to prison health systems in the U.S. by an immense incarcerated population is significant. However, the patterns of presentation and associated mortality of cancer among the incarcerated population is unknown.

METHODS

An historical cohort of cancers diagnosed among inmates of the Texas Department of Criminal Justice over the course of 20 years who were followed at the University of Texas Medical Branch in Galveston, Texas was identified. There were 1807 inmates who were diagnosed with cancer. Two cohorts were chosen for comparison: a random sample of 179,757 patients from the Surveillance, Epidemiology, and End Results (SEER) registry, and an age-matched, gender-matched, race-matched SEER population comprised of 6124 patients (MSEER). Disease sites and associated mortality of the inmate cancer patients were determined and compared with SEER cohorts.

RESULTS

A marked rise in cancer diagnoses among inmates paralleled the rise in the inmate population. The leading cancers were lung carcinoma, non-Hodgkin lymphoma (NHL), and carcinomas of the oral cavity and pharynx. Among women, cervical carcinoma was the most common. Lung carcinoma, NHL, and hepatic carcinoma accounted for more cancer deaths among inmates than in the SEER cohort (P < 0.0001 for all comparisons). Lung carcinoma, hepatic carcinoma, and NHL were significantly more common in the inmate cohort than in the MSEER cohort (P < 0.001 for all comparisons). The median survival was inferior in the inmate cohort (21 mos) compared with the SEER cohort (55 mos) and the MSEER cohort (54 mos) (P < 0.0001 for both comparisons).

CONCLUSIONS

Cancers with unique epidemiology and high associated mortality have emerged among the incarcerated. This has significant implications for prison health systems. Cancer 2005. © 2005 American Cancer Society.

Unique demographics and the prevalence of specific cancer risk factors are likely to influence the presentation of cancer among the incarcerated population, which rose rapidly in the last 20 years of the Millennium to > 2 million individuals in the U.S.1 The size of the Texas inmate population increased by greater than 200% in the 1990s to > 150,000 individuals.2, 3 The demographics of the inmate population differ from those of the general population. In Texas, most inmates (92%) are men, 32% are non-Hispanic whites, 43% are non-Hispanic blacks, and 25% are Hispanics.4 In contrast, 53% of the state's general population is non-Hispanic white, 12% is non-Hispanic black, and 32% is Hispanic.5 In addition, the age distribution is different; for example, 13% of the Texas population is age 60 years or older, compared with 2% of the inmate population.

The cancer risk factors that are known as highly prevalent in prison populations are smoking, drug and alcohol use, and viral infections. At least 70% of prisoners in the U.S. report a history of smoking, and > 33% of state prisoners report a history of alcohol use.2 A history of marijuana and crack cocaine use, both potential carcinogens,6, 7 is reported by > 80% of all state inmates.2

Viral infections associated with a greater risk of cancer are more prevalent in prisoners than in the general population. An epidemic of hepatitis C virus (HCV) infection in the prison population has left an estimated 30–40% of inmates infected.8 In addition, 2% of incarcerated men test positive for hepatitis B virus surface antigen.9 Human immunodeficiency virus (HIV) infection rates in the incarcerated are estimated at 6 times the national average (200 per 100,000 population).10 Although 1.7% of all Texas inmates are infected with HIV,11 it has been estimated that the seropositive rate is as high as 9% among women inmates.12 It has been estimated that 20–26% of all individuals who are living with HIV in the U.S. and 29–43% with HCV have passed through a correctional facility.13 High rates of human papillomavirus infection have been detected among women in prison,14 and the proportion of abnormal Papanicolaou smears among incarcerated women is greater than in the general population.15

The differences in demographics and the high prevalence of cancer risk factors among the incarcerated suggest that vital differences should exist with regard to the presentation and impact of cancer among prisoners compared with the general population. Such data would serve responsible policy makers in prison health systems to better understand the scope of the delivery of care challenge posed by this special population within a unique security environment. To our knowledge to date, there have been no systematic studies of the problem of cancer among prisoners. Therefore, in the current study, we examined the epidemiology of cancer in the incarcerated population of Texas and compared our findings with a general population.

MATERIALS AND METHODS

To understand the epidemiology of cancer among prisoners, we examined the demographic profile of prisoners who were diagnosed with cancer, compared the number of prison cancer diagnoses over time with the concurrent prison population, and described the distribution of organ/site-specific cancers and associated mortality by age/interval, gender, and race/ethnicity variables. We examined data stored over 2 decades (1980–1999) from the cancer registry of the University of Texas Medical Branch (Galveston, TX), which has been the tertiary referral center for the Texas Department of Criminal Justice since 1980. To establish whether differences in organ/site-specific cancers and associated mortality exist between the incarcerated and the general population, comparator populations were sought. Thus, we compared the prison profile with a sample of cancers taken from the general population obtained from the Surveillance Epidemiology, and End Results (SEER) registry,16 which contains data from the general U.S. population, and with a subset of those patients that was matched by age, gender, and race to the incarcerated population. This study was approved by the University of Texas Medical Branch Institutional Review Board under institutional guidelines for the study of the incarcerated.

The University of Texas Medical Branch Cancer Registry

From 1980 to 1995, inmates with cancer from the Texas Department of Criminal Justice were referred exclusively to the University of Texas Medical Branch. All new diagnoses of cancer were recorded in the institutional cancer registry. Between 1995–2000, 187 incarcerated patients who were diagnosed with cancer were referred to Texas Tech University Health Sciences Center (Lubbock, TX). There is no known referral bias to Texas Tech from the Texas Department of Criminal Justice; inmates were referred there for their convenience. The patients who were seen at Texas Tech were excluded from this report, because information on such patients was not available. During that same period, 751 patients were seen at the University of Texas Medical Branch. All patients were classified by self-reported race. Because ethnic status is not available through the cancer registry, ethnicity was estimated according to each patient's last name, which is customary in tumor registry practice.

The University of Texas Medical Branch cancer registry has records on cancer in inmates from 1980 onward. Data available include date of diagnosis, age at diagnosis, gender, race classification, organ or site of cancer, cancer histology, smoking history, alcohol use history, date of last follow-up, survival status at last follow-up (alive or dead), and date of death. Data regarding viral serology, recreational drug use history, ethnicity, and cause of death, as determined by autopsy, are not recorded routinely in the cancer registry.

Identification of Cases and Controls

In total, 1953 new primary cancers were identified in inmates from January 1, 1980 to December 31, 1999. Of these, 146 patients with noninvasive or skin carcinomas (111 patients with cervical intraepithelial neoplasia, 5 patients with ductal carcinoma in situ of the breast, 4 patients with basal cell carcinoma, 2 patients with Hutchinson freckles, 4 patients with squamous skin carcinoma, 4 patients with oral cavity/pharyngeal squamous carcinoma in situ, 3 patients with laryngeal squamous carcinomas in situ, 5 patients with Bowen disease, and 8 patients with vulvar intraepithelial neoplasia) were excluded, leaving a total of 1807 patients with cancer for statistical analysis. Rectal carcinomas were included with colon carcinomas. Hepatic carcinomas included tumors of the biliary system, and brain tumors included other central nervous system tumors. Second malignancies (only 2% of the total) were treated as individual cancers. All-cause mortality was reported.

Two cohorts were selected for comparison. The first cohort was a representative sample of individual patients with primary cancer among the general population, consisting of a random 10% sample of 179,757 patients with cancer from the SEER registry. This cohort was used to illustrate differences in the distribution of cancers and their associated mortality between the incarcerated and general populations. The second cohort also was selected in a random fashion from the SEER registry but differed from the first, in that it was matched with the inmate cancer cohort by age, gender, and race (black vs. nonblack). For the second sample, four patients from the SEER registry were selected for each inmate patient (when full matching was possible). This cohort was used to examine differences between the incarcerated and general populations that were not related to the different demographic profile of the populations.

Statistical Analysis

Differences between proportions were compared using a two-tailed chi-square test or Fisher exact test, as appropriate. The survival experiences of the inmate and SEER cohorts were examined according to the method of Kaplan and Meier.17 Differences among these groups were tested for statistical significance using the Breslow statistic.18 This test gives greater weight to early observations and is less sensitive to late events when few patients remain in the study. The annual number of cancer diagnoses expected in the total U.S. prison population was estimated by applying the rates observed in the Texas prison population to published reports of the U.S. prison population size,1 with direct adjustment for differences between the U.S. and Texas inmate populations by age and gender.

RESULTS

The demographics (Table 1) reflect a median age at cancer diagnosis of younger than age 50 years in the incarcerated population, which was found to be largely male (89%) and in which race-ethnic minorities accounted for greater than one-half (54%) of the patients. Among men, non-Hispanic whites, and among women, non-Hispanic blacks were the most frequently represented subgroups. Ten percent of the cancer patients (n = 186) were diagnosed in patients age 65 years or older. A history of smoking (82%) or alcohol use (67%) was common.

Table 1. Characteristics of Incarcerated Cancer Patients
CharacteristicNo. of patients (%)
MenWomenOverall
  • a

    Includes 246 whites, 2 blacks, and 9 others.

Age in yrs   
 Median494146
 Range15–8720–7215–87
Race/ethnicity   
 Non-Hispanic white759 (47.1)68 (34.7)827 (45.8)
 Non-Hispanic black620 (38.5)97 (49.5)717 (39.7)
 Hispanica226 (14.0)31 (15.8)257 (14.2)
 Other6 (0.4)0 (0.0)6 (0.3)
Total1611 (100.0)196 (100.0)1807 (100.0)

From 1980–1999, the number of individuals in Texas prisons rose steeply, with a fivefold increase reported by the end of 1999.3 The number of cancers diagnosed in Texas prisons increased correspondingly (Fig. 1), as demonstrated by the strong correlation between the number of cancers and the rising prison population (r2 = 0.9063). The number of cancers was 10-fold higher in 1999 compared with 1980, and one-half of the invasive cancers were diagnosed in the last 5 years. Extrapolating from these data to the total U.S. incarcerated population1 and adjusting for differences in the age, gender, and racial distributions of the populations, it is expected that 2813 new cancers will be diagnosed per year among the incarcerated population in the U.S.

Figure 1.

This chart compares the size of the inmate population with the number of newly diagnosed cancers.

Cancer Frequencies and Deaths in the Incarcerated Population

Lung carcinomas (443 patients) were the most common malignancy, accounting for one-fourth (24.5%) of all diagnoses (Table 2). NHL and carcinomas of the oral cavity/pharynx, colon (including rectum), prostate, and liver also were common neoplasms. Head and neck malignancies as a group (oral cavity, pharyngeal, and laryngeal carcinomas) were second only to lung carcinoma in frequency and accounted for nearly 10% of all cancer diagnoses. Cervical neoplasms originally were the second most common cancers in the inmate cancer registry, with 173 diagnoses, but 111 of those were carcinoma in situ and were excluded from the current analysis. Cervical carcinoma was the most common cancer among women inmates (39.8% of diagnoses). By contrast, female breast carcinoma was uncommon, and most diagnoses (40 of 44 breast carcinomas) occurred in women age 50 years or younger. It is interesting to note that 18 of 144 patients (12.5%) with NHL presented with a brain or central nervous system primary site. With regard to skin neoplasms, there were 29 diagnoses each of melanoma and Kaposi sarcoma.

Table 2. Distribution of Disease Sites and Disease-Related Deaths (1980–1999)
Primary siteNo. (%)Median survival in mos5 yr survival (%)
PatientsDeaths
  1. NHL: Non-Hodgkin lymphoma; HD: Hodgkin disease.

Lung443 (24.5)358 (34.2)812
NHL144 (8.0)95 (9.1)1533
Oral/pharynx119 (6.6)52 (5.0)9957
Colon111 (6.1)58 (5.5)2538
Prostate100 (5.5)33 (3.1)7958
Liver78 (4.3)60 (5.7)44
Skin73 (4.0)41 (3.9)623
Cervix62 (3.4)8 (0.8)15489
Leukemia64 (3.5)44 (4.2)2521
Larynx57 (3.2)21 (2.0)13565
Brain56 (3.1)21 (2.0)9059
HD52 (2.9)18 (1.7)> 24064
Kidney50 (2.8)30 (2.9)3341
Breast49 (2.7)15 (1.4)6964
Testis45 (2.5)4 (0.4)> 24090
Stomach39 (2.2)27 (2.6)2540
Pancreas29 (1.6)25 (2.4)40
Others236 (15.2)138 (15.7)
Total1807 (100)1048 (100)2237

In total, 1048 deaths were recorded, including 988 deaths among men (94.3%) and 60 deaths among women (5.7%). Lung carcinoma was associated with greater than one-third of all cancer deaths among inmates, followed by NHL (9.1%) and hepatic carcinoma (5.7%). Collectively, lung carcinoma and head and neck carcinomas accounted for > 40% of deaths. The 5-year survival rate estimated by the Kaplan–Meier method was best for inmates with testicular carcinoma (90%) and cervical carcinoma (89%) and was worst for inmates with hepatic carcinoma (4%) and pancreatic carcinoma (0%).

Gender and Age-Group Classification of Cancer Frequencies and Deaths

Examination of cancer frequency distribution in men by age group (Table 3) revealed that NHL and skin malignancies, which included acquired immunodeficiency syndrome (AIDS)-related Kaposi sarcoma, were more common than testicular carcinoma among men ages 20–39 years. Oral cavity/pharyngeal carcinomas and laryngeal carcinomas were among the top 5 most common cancers in the groups ages 40–59 years and ages 60–79 years, respectively. Prostate carcinomas were diagnosed less frequently than lung carcinomas, even in the group ages 60–79 years. There were 13 patients with cancer and 3 deaths reported among men younger than age 20 years and 7 patients with cancer and 5 deaths among men age 80 years and older.

Table 3. Top Five Cancers by Age Groups for Incarcerated Men
RankingDisease site: No. (%)
Ages 20–39 yrsAges 40–59 yrsAges 60–79 yrsAll ages
MenDeathsMenDeathsMenDeathsMenDeaths
  1. NHL: Non-Hodgkin lymphoma; OP: oral cavity/pharynx; HD: Hodgkin disease.

1NHL: 57 (14)Lung: 31 (15)Lung: 264 (31)Lung: 210 (39)Lung: 118 (36)Lung: 99 (42)Lung: 442 (26)Lung: 342 (35)
2Skin: 48 (12)NHL: 29 (14)OP: 78 (9)Liver: 38 (7)Prostate: 59 (18)NHL: 19 (8)NHL: 135 (8)NHL: 91 (9)
3Testis: 39 (9)Skin: 27 (13)NHL: 69 (8)OP: 38 (7)Colon: 24 (7)Colon: 15 (6)OP: 115 (7)Liver: 59 (6)
4Lung: 38 (9)Brain: 19 (9)Colon: 56 (7)NHL: 37 (7)Liver: 15 (5)Liver: 14 (6)Colon: 104 (6)Colon: 57 (6)
5HD: 35 (8)Leukemia: 19 (9)Liver: 51 (6)Colon: 29 (5)Larynx: 13 (4)OP: 10 (4)Prostate: 100 (6)OP: 52 (5)
Total4182098475443262341611988

Among men, lung carcinoma was the most common cancer related to death among all age groups including, the group ages 20–39 years. The high mortality associated with skin cancers among men in the group ages 20–39 years was likely because of AIDS-related complications in patients with Kaposi sarcoma. Hepatic, colon, and oral cavity/pharyngeal carcinomas were significant causes of death in the groups ages 40–59 years and ages 60–79 years. Among women (Table 4), cervical carcinoma matched breast carcinoma in cancer deaths among women age younger than 40 years, whereas lung carcinoma dominated cancer deaths among women older than age 40 years.

Table 4. Top Three Cancers by Age Group for Incarcerated Women
RankingDisease site: No. (%)
Age < 40 yrsAge > 40 yrsAll ages
WomenDeathsWomenDeathsWomenDeaths
1Cervix: 41 (47)Cervix: 6 (33)Breast: 29 (27)Lung: 14 (31)Cervix; 62 (32)Lung: 15 (25)
2Breast: 15 (17)Breast: 5 (28)Cervix: 21 (19)Breast: 8 (18)Breast: 44 (22)Breast: 13 (21)
3Endocrine: 6 (7)Lung: 2 (11)Lung: 18 (17)Cervix; 4 (9)Lung: 21 (11)Cervix: 10 (16)
Total88171084319660

Race-Ethnic Classification of Cancer Frequencies and Deaths

Lung carcinoma remained the leading cancer in all race-ethnic categories in both cancer frequencies and cancer deaths (Table 5). However, any apparent differences observed between categories were not tested for statistical significance.

Table 5. Distribution of Disease Sites and Mortality Related to Race and Ethnicity among Incarcerated Cancer Patients
Disease siteNo. (%)
Non-Hispanic white (n = 827)Hispanic (n = 257)Non-Hispanic black (n = 717)
PatientsDeathsPatientsDeathsPatientsDeaths
  1. NHL: non-Hodgkin lymphoma; OP; oral cavity/pharynx; HD: Hodgkin disease.

  2. a Six other incarcerated patients (four deaths) outside this race-ethnic classification were excluded.

Lung226 (27)185 (38)26 (10)22 (17)191 (27)151 (36)
NHL50 (6)35 (7)24 (9)12 (9)69 (10)47 (11)
OP65 (8)26 (5)10 (4)5 (3)42 (6)20 (5)
Colon52 (6)28 (6)20 (8)9 (7)39 (5)21 (5)
Prostate34 (4)12 (2)11 (4)4 (3)55 (8)17 (4)
Cervix23 (3)7 (1)12 (5)0 (—)27 (4)1 (< 1)
Liver33 (4)24 (5)24 (9)19 (15)21 (3)17 (4)
Skin46 (6)29 (6)5 (2)3 (2)22 (3)9 (2)
Brain30 (4)14 (3)10 (4)3 (2)16 (2)4 (1)
Leukemia28 (3)15 (3)14 (5)9 (7)22 (3)20 (5)
Larynx32 (4)11 (2)5 (2)0 (—)20 (3)10 (2)
Breast16 (2)6 (1)5 (2)1 (< 1)28 (4)8 (2)
HD22 (3)7 (1)10 (4)4 (3)19 (3)7 (2)
Kidney14 (2)11 (2)21 (9)13 (11)14 (2)5 (1)
Testis24 (3)1 (< 1)12 (5)2 (2)9 (1)1 (< 1)
Stomach13 (2)9 (2)6 (2)4 (3)20 (3)14 (3)
Pancreas8 (1)7 (1)5 (2)3 (2)16 (2)15 (4)
Others111 (13)65 (13)37 (14)17 (13)87 (12)55 (13)
Total827 (100)492 (100)257 (100)130 (100)717 (100)422 (100)

Trends over 20 Years

For lung carcinoma, the proportion of patients (25%) and cancer deaths (35%) did not change appreciably over the past 20 years, whereas the proportion of deaths from NHL rose from 5% to 10% from the first 5-year period to the last 5-year period, respectively. Hepatic carcinomas became the third most common cause of cancer-related death in the second decade.

Distribution of Cancers in Incarcerated and SEER Populations

Significant differences in the proportional distribution of cancers were found between the incarcerated population and the general population from the random SEER cohort. Lung carcinoma was twice as common in the prison population compared with the SEER cohort (25% vs. 13.8%, respectively; P < 0.001). Also significantly more common among inmates were NHL (8% vs. 3.4% SEER; P < 0.001), oral cavity and pharyngeal carcinomas (6.6% vs. 2.3% SEER; P < 0.0001), and cervical carcinomas (31.6% women vs. 12.2% SEER women; P < 0.0001). The following site-specific, cancer-related deaths were significantly higher among inmates: lung carcinoma (34.2% vs. 21.9% SEER; P < 0.0001), NHL (9.2% vs. 3.8% SEER; P < 0.0001), hepatic carcinoma (5.3% vs. 1.3% SEER; P < 0.0001), oral cavity and pharyngeal carcinomas (4.9% vs. 2.6% SEER; P < 0.0001), and Kaposi sarcoma (3.4% vs. 0.9% SEER; P < 0.0001).

Because the age, gender, and racial demographics of the cohort differed significantly from those of the SEER cohort, an age-matched, gender-matched, and race-matched sample was selected randomly from the SEER population (the MSEER cohort) to determine whether the observed differences were because of demographic variations alone. The MSEER cohort (n = 6124 patients) included 89% male patients and 38% black patients, with a median age of 48 years (n = 6124 patients). The matched inmate cohort (n = 1538 patients) included 89% male patients and 39% black patients, with a median age of 47 years. After controlling for demographic differences, we found that the proportion of patients with lung carcinoma remained greater in the inmate population than in the MSEER population (26% vs. 16%, respectively; P < 0.001), which also was true for patients with NHL (9% vs. 6%, respectively; P < 0.0001) and patients with hepatic carcinoma (4% vs. 1%, respectively; P < 0.001). Conversely, patients with colon carcinoma (6% vs. 10%, respectively; P < 0.001), melanoma (1% vs. 5%, respectively; P < 0.001), and prostate carcinoma (5% vs. 11%, respectively; P < 0.001) were less common proportionately in the inmate population compared with the MSEER population.

Survival Rates in Incarcerated and SEER Populations

The median survival of inmates was 21 months, and 37% of inmate patients survived for 5 years. The survival of patients in both the SEER cohort and the MSEER cohort were significantly superior to the survival of the inmate cohort (SEER, 55 mos; MSEER, 54 mos; P < 0.0001 for both comparisons). The 5-year survival rate was 49% for the SEER cohort and 47% for the MSEER cohort (P < 0.0001 for both comparisons) (Fig. 2). The median survival in the MSEER cohort was superior to that of the inmate cohort both for women (100 mos; MSEER, > 200 months; P = 0.009) and for men (18 mos; MSEER, 42 months; P < 0.0001). The median survival for female inmates was 8.4 years, with a 5-year survival rate of 67%; whereas, for male inmates, the median survival was 1.5 years, with a 5-year survival rate of 33% (P < 0.00001) (Fig. 3). The median survival was 2.7 years for Hispanics, 1.9 years for non-Hispanic whites, and 1.6 years for non-Hispanic blacks. Only the difference between blacks and Hispanics approached statistical significance (P = 0.046).

Figure 2.

This chart illustrates overall survival of the inmate cohort and the Surveillance, Epidemiology, and End Results (SEER) cohort. MSEER: the SEER population of age-matched, gender-matched, and race-matched patients (matched to the inmate cohort).

Figure 3.

This chart illustrates gender-specific overall survival outcomes of the inmate cohort and the Surveillance, Epidemiology, and End Results (SEER) cohort.

DISCUSSION

Our data illustrate that the epidemiology of cancer in prison differs from that of cancers in the general U.S. population and that, overall, these cancers are associated with a high mortality. We estimate that nearly 3000 inmates in the U.S. are diagnosed with cancer annually. The numbers of inmates diagnosed with cancer rose proportionately as the prison population increased rapidly in the latter part of the last century, and males from race-ethnic minorities with a median age younger than 50 years formed a dominant subgroup. The vast majority of these patients had a history of smoking and/or alcohol use.

The pattern presented by the cancers among the incarcerated is influenced by the unique cancer risk factors and demographics prevalent in the prison population. The finding that lung carcinoma, NHL, and hepatic carcinoma were more common proportionately in the prison population compared with the demographically matched general population is consistent with the greater incidence reported among inmates of the known risk factors of smoking, HIV, HCV, and alcohol, respectively. It is interesting to note that primary brain lymphoma was more common in the inmate population (12.5%) than in the general population (1%)—all of the patients with primary brain lymphoma in this survey were HIV positive. An illustration of the demographic impact on the presentation of cancers was that > 90% of female breast carcinomas occurred in inmates younger than age 50 years, and 10% of breast carcinomas were found in males. Prostate and colon carcinomas were less common, perhaps as a result of an ascertainment bias, because cancer screening was restricted to Papanicolaou smears and mammograms among women.

We also speculate that the inferior survival outcomes among inmates compared with the general population resulted from the greater frequency of lethal neoplasms, such as lung carcinomas, NHL, and hepatic carcinomas. Variations in the distribution of lethal malignancies among gender and racial-ethnic groups of the incarcerated also may explain the various survival outcomes among these subgroups.

The current study data do not represent fully the entire incarcerated population of the U.S.: Data from federal prisoners are not included, and it is known that federal inmates have a different demographic and behavioral cancer risk profile, with a lower rate of drug and alcohol use compared with state inmates.2 Important variations in the demographic compositions of prison cohorts are likely to exist from state to state, and these differences may influence the epidemiology of cancer. For example, Hispanics in the inmate population had superior overall survival outcomes compared with non-Hispanic blacks. In general, Hispanics have lower incidence and mortality rates from cancer compared with the non-Hispanic black and white populations.19 Access to tertiary care institutions with expertise in cancer care also is likely to vary, and treatment outcomes may differ accordingly. Linkages between data on incarcerated populations and state-based registries may offer more accurate data on the trends and impact of cancer among the prison population and patterns of care.

The rapid increase in cancer diagnoses in prison with the associated high mortality suggests that the overall burden of cancer care for the incarcerated is growing. Implicit in these findings are the rising therapeutic, economic, psychosocial, and ethical challenges for healthcare delivery in a unique security environment, all of which define the nature of prison oncology. With the incarcerated population in the U.S. at its greatest numbers ever and the lengths of prison stay increasing significantly,2 the problems of prison oncology are unlikely to diminish in the near future. Careful consideration may be given to the design of prison oncology programs to improve the detection and treatment of cancer and associated supportive care. However, it is premature to make screening recommendations for prison populations that differ from those for the general population based on these data.

Management challenges often are exacerbated in the prison setting. One example is the provision of effective cancer pain control with opioids in a population in which it is known that drug abuse and misuse are common.20 Another challenge is that the diagnosis of a terminal illness does not guarantee parole; therefore, more patients are dying in prison from cancer than were dying previously. Prison hospices that are available in Texas offer supervised environments in which effective palliative care may be delivered. However, prison hospices are not available widely in the U.S., and palliative-care guidelines for healthcare practitioners in a prison system are needed. Healthcare resources are limited in the prison system; expensive but potentially life-saving procedures, such as bone marrow transplantation for patients with NHL or leukemia, are not covered routinely, even when they are indicated clinically. To understand the full scope of the challenges and opportunities of prison oncology, comprehensive nationwide studies will be required.

Acknowledgements

The authors thank the following for their assistance: Cynthia C. Dunn, Sonia Barre, Kleanthe Caruso, Michael Megna, and John Pulvino, P.A., at The University of Texas Medical Branch, Galveston, Texas.

Ancillary