Scaling of the surface vasculature on the human placenta.

The networks of veins and arteries on the chorionic plate of the human placenta are analyzed in terms of Voronoi cells derived from these networks. Two groups of placentas from the United States are studied: a population cohort with no prescreening, and a cohort from newborns with an elevated risk of developing autistic spectrum disorder. Scaled distributions of the Voronoi cell areas in the two cohorts collapse onto a single distribution, indicating common mechanisms for the formation of the complete vasculatures, but which have different levels of activity in the two cohorts.

Determining source strength of semivolatile organic compounds using measured concentrations in indoor dust.

UNLABELLED: Consumer products and building materials emit a number of semivolatile organic compounds (SVOCs) in the indoor environment. Because indoor SVOCs accumulate in dust, we explore the use of dust to determine source strength and report here on analysis of dust samples collected in 30 US homes for six phthalates, four personal care product ingredients, and five flame retardants. We then use a fugacity-based indoor mass balance model to estimate the whole-house emission rates of SVOCs that would account for the measured dust concentrations. Di-2-ethylhexyl phthalate (DEHP) and di-iso-nonyl phthalate (DiNP) were the most abundant compounds in these dust samples. On the other hand, the estimated emission rate of diethyl phthalate is the largest among phthalates, although its dust concentration is over two orders of magnitude smaller than DEHP and DiNP. The magnitude of the estimated emission rate that corresponds to the measured dust concentration is found to be inversely correlated with the vapor pressure of the compound, indicating that dust concentrations alone cannot be used to determine which compounds have the greatest emission rates. The combined dust-assay modeling approach shows promise for estimating indoor emission rates for SVOCs. PRACTICAL IMPLICATIONS: The combined dust-assay modeling approach in this study can be used to predict the source strength of indoor released compounds, integrating emissions from consumer products, building materials, and other home furnishings. Our findings show that estimated emission rates are closely related to not only the level of compounds on dust, but also the vapor pressure of the compound. Thus, a fugacity-based indoor mass balance model and measured dust concentrations can be used to estimate the whole-house emission rates from all sources in actual indoor settings, when individual sources of emissions are unknown.

beta2-Adrenergic receptor gene variants and risk for autism in the AGRE cohort.

The beta2-adrenergic receptor is part of the catecholamine system, and variants at two polymorphic sites in the gene coding for the receptor (ADRB2) confer increased activity. Overstimulation of this receptor may alter brain development, and has been linked to autism in non-identical twins. The objective of this study was to determine whether alleles in ADRB2 are associated with diagnosis of autism in the Autism Genetic Resource Exchange (AGRE) population. Three hundred and thirty-one independent autism case-parent trios were included in the analysis. Subjects were genotyped at activity-related polymorphisms rs1042713 (codon 16) and rs1042714 (codon 27). Association between autism and genotypes at each polymorphic site was tested using genotype-based transmission disequilibrium tests, and effect modification by family and pregnancy characteristics was evaluated. Sensitivity to designation of the proband in each family was assessed by performing 1000 repeats of the analysis selecting affected children randomly. A statistically significant OR of 1.66 for the Glu27 homozygous genotype was observed. Increased associations with this genotype were observed among a subset of Autism Diagnostic Observation Schedule confirmed cases and a subset reporting experience of pregnancy-related stressors. In conclusion, the Glu27 allele of the ADRB2 gene may confer increased risk of autism and shows increased strength with exposure to pregnancy related stress.

Risk of colorectal cancer after breast cancer.

BACKGROUND: History of breast cancer has been reported as a risk factor for colorectal cancer in women. In view of the ambiguous nature of existing evidence and the growing interest in targeted colorectal cancer prevention, we sought to quantify this risk. METHODS: We used the Surveillance Epidemiology and End Results (SEER) database to estimate risk of colorectal cancer after breast-cancer diagnosis in women with first incident breast cancer between 1974 and 1995. Observed colon and rectal cancer risk was compared with that expected in the general population. We stratified comparisons by age at breast-cancer diagnosis, stage of cancer, ethnic origin of patient, and follow-up time. FINDINGS: Overall, women with previous breast cancer were 5% less likely (95% CI 1-9) to develop colon and 13% less likely (6-19) to develop rectal cancer than women in the general population. Stratified analyses suggested that the risk reductions observed for colon and rectal cancer were most pronounced for women with breast cancer diagnosed after age 65 years, in white women, women with local stage breast cancer, and women diagnosed in the later study years (1990-94). INTERPRETATIONS: Breast cancer does not increase subsequent colorectal cancer risk, and reduced risk was seen for certain subgroups of women. Because no biologically plausible endogenous protective factor has been identified, we suggest that reduced risk could stem from an accumulation of exposures that increase breast-cancer frequency but protect against colorectal cancer.

Prenatal zidovudine use and congenital anomalies in a medicaid population.

OBJECTIVES: To examine the association of prescribed zidovudine (ZDV) during pregnancy with congenital anomalies in a population-based cohort. METHODS: Medicaid claims were used to assess prescribed ZDV and children's major congenital anomalies in 1932 liveborn deliveries from 1993 to 1996 to HIV-infected women in the state of New York (NYS), U.S.A. Prevalence of anomalies in the cohort was compared with that of a general NYS population. Within the cohort, adjusted odds of any anomaly were compared by receipt of ZDV and by trimester of first prescription. RESULTS: The adjusted prevalence of any anomaly in the study cohort was 2.76 times greater than in the general population (95% confidence interval [CI], 2.36-3. 17). Children of study women who were prescribed ZDV had increased adjusted odds of any anomaly (adjusted odds ratio [OR], 1.55; 95% CI, 1.01-2.29). Adjusted ORs (with CIs) by trimester of first prescription were 1.20 (0.58-2.51), 1.47 (0.85-2.55), and 1.84 (1. 04-3.25) for the first, second, and third trimesters, respectively. CONCLUSION: Children of HIV-infected women in this cohort had a greater prevalence of major anomalies than did the general NYS population. An increased risk of major anomalies was not evident for first trimester exposure when the association would have been most biologically plausible.

Antiretroviral use and pharmacy-based measurement of adherence in postpartum HIV-infected women.

BACKGROUND: Antiretroviral treatment for HIV-infected women is standard during pregnancy to prevent vertical transmission, but data on postpartum therapy for the mother are lacking. OBJECTIVE: The objective of this study was to examine the impact of provider and patient characteristics on receipt of antiretroviral therapy and pharmacy-based measurement of adherence by postpartum HIV-infected women. RESEARCH DESIGN: This was a retrospective cohort study. SUBJECTS: The study included 2,648 New York State Medicaid-enrolled HIV-infected women who delivered from January 1993 through October 1996 and were followed up through September 1997. MEASURES: From Medicaid claims in the first postpartum year, the study examined any prescribed antiretroviral therapy and, among women treated >2 months, adherence, defined as > or =80% days covered by prescribed therapy from first to last antiretroviral prescription. RESULTS: Antiretroviral therapy was prescribed for 681 (26%) study women. Of 292 women treated >2 months, 28% were adherent on the basis of the pharmacy-based measure. The proportion of treated women was highest in 1996 (40%), and adherence was best in 1995 (44%) when most women took monotherapy. The adjusted odds ratios (AORs) of treatment were 1.67 (95% CI, 1.24 to 2.25) for women receiving HIV-focused services and 2.71 (95% CI, 1.99 to 3.69) for women with a provider in an HIV-related specialty. The AORs of adherence were greater for women with HIV-focused services (2.13; 95% CI, 1.05 to 4.30) and for former illicit drug users versus nonusers (2.40; 95% CI, 1.05 to 5.50). CONCLUSIONS: This population-based pharmacy analysis reveals improving antiretroviral use but continuing poor pharmacy-based adherence by postpartum HIV-infected women. Receipt of HIV-focused services appears to be particularly beneficial in increasing the likelihood of treatment and adherence.

Causes of death in elderly prostate cancer patients and in a comparison nonprostate cancer cohort.

BACKGROUND: Prostate cancer tends to affect older men and to progress relatively slowly. Since the prevalence of comorbidity increases with advancing age, competing causes of death are important contributors to death rates among prostate cancer patients. Accurate determination of the underlying causes of death in older men dying with prostate cancer may thus also be more difficult. METHODS: We compared the distribution of underlying causes of death in decedents from a population-based cohort of elderly prostate cancer patients to that from a population-based comparison cohort of elderly men without prostate cancer. Among decedents from the prostate cancer patient cohort, we examined associations of patient demographics, disease stage, and initial treatment, with assignment of a prostate cancer underlying cause of death (versus any other cause) by use of multivariable logistic regression. In the subgroup of prostate cancer patient decedents having underlying causes of death other than prostate cancer, the underlying cause distribution was compared with that in nonprostate cancer cohort decedents. RESULTS: Prostate cancer was the underlying cause for 39% (95% confidence interval [CI] = 36.3-41.9) of the decedents in the prostate cancer cohort. Causes of death among prostate cancer patients not dying of prostate cancer were similar to those among the nonprostate cancer cohort decedents. However, in those who were aggressively treated, the adjusted odds of other cancer causes of death were 51% higher (odds ratio [OR] = 1.51; 95% CI = 1.08-2.10) than that in nonprostate cancer patient decedents, while in those treated with watchful waiting the adjusted odds were 34% lower (OR = 0.66; 95% CI = 0.47-0.93). CONCLUSIONS: Initial treatment may influence the underlying cause of death reported in vital statistics for prostate cancer patients.

Adherence to antiretroviral therapy by pregnant women infected with human immunodeficiency virus: a pharmacy claims-based analysis.

OBJECTIVE: To assess adherence to antiretroviral therapy with the use of Medicaid pharmacy claims data for human immunodeficiency virus (HIV)-infected pregnant women and to identify associated maternal and health care factors. METHODS: We retrospectively studied a cohort of 2714 HIV-infected women in New York State who delivered live infants from 1993-96. Among 682 women prescribed antiretroviral therapy in the last two trimesters, we studied 549 who started therapy more than 2 months before delivery. Adherence was defined as adequate if the supplied drug covered at least 80% of the days from the first prescription in the last two trimesters until delivery. Multivariable analyses were used to examine associations between maternal and health care factors and adherence. RESULTS: Only 34.2% of 549 subjects had at least 80% adherence based on pharmacy data, a rate that remained stable over time. The adjusted odds ratios (ORs) of adherence for black (OR 0.47, 95% confidence interval [CI] 0.30, 0.75) and Hispanic (OR 0.49, 95% CI 0.29, 0.82) women were nearly 50% lower than for white women. The OR of adherence was 0.32 (95% CI 0.12, 0.90) for teenagers compared with women aged 25-29 years and 0.56 (95% CI 0.34, 0.92) for women in New York City versus those residing elsewhere. Women on antiretroviral therapy before pregnancy were more likely to adhere (OR 1.55, 95% CI 1.02, 2.35). CONCLUSION: Teenagers, women of minority groups, and women living in New York City had greater risks of poor antiretroviral adherence, whereas women already prescribed antiretrovirals before pregnancy had better adherence. Our conservative pharmacy data-based measure showed that most HIV-infected women adhered poorly and adherence did not improve over the 4-year study.

Improved birth outcomes among HIV-infected women with enhanced Medicaid prenatal care.

OBJECTIVES: This study evaluated the impact of enhanced prenatal care on the birth outcomes of HIV-infected women. METHODS: Medicaid claims files linked to vital statistics were analyzed for 1723 HIV-infected women delivering a live-born singleton from January 1993 to October 1995. Prenatal care program visits were indicated by rate codes. Logistic models controlling for demographic, substance use, and health care variables were used to assess the program's effect on preterm birth (less than 37 weeks) and low birthweight (less than 2500 g). RESULTS: Of the women included in the study, 75.3% participated in the prenatal care program. Adjusted program care odds were 0.58 (95% confidence interval [CI] = 0.42, 0.81) for preterm birth and 0.37 (95% CI = 0.24, 0.58) for low-birthweight deliveries in women without a usual source of prenatal care. Women with a usual source had lower odds of low-birthweight deliveries if they had more than 9 program visits. The effect of program participation persisted in sensitivity analyses that adjusted for an unmeasured confounder. CONCLUSIONS: A statewide prenatal care Medicaid program demonstrates significant reductions in the risk of adverse birth outcomes for HIV-infected women.

Sources of prenatal care data and their association with birth outcomes of HIV-infected women.

OBJECTIVES: Different sources of prenatal care data were used to examine the association between birth outcomes of HIV-infected women and the Adequacy of Prenatal Care Utilization (APNCU) index. METHODS: Adjusted odds ratios of birth outcomes for 1858 HIV-positive mothers were calculated for APNCU indexes on the basis of birth certificate data or 3 types of physician visits on Medicaid claims. RESULTS: Claims- and birth certificate-based APNCU indexes agreed poorly (kappa < 0.3). Only the broadest claims-based APNCU index had lower adjusted odds ratios for low birthweight (0.64; 95% confidence interval [CI] = 0.49, 0.84) and preterm birth (0.70; 95% CI = 0.54, 0.91). The birth certificate-based index had a reduced adjusted odds ratio (0.73; 95% CI = 0.56, 0.95) only for preterm birth. CONCLUSIONS: The association of birth outcomes and adequacy of prenatal care in this HIV-infected cohort differed significantly depending on the source of prenatal care data.

Effect of enhanced prenatal and HIV-focused services for pregnant women who are infected by human immunodeficiency virus on emergency department use.

OBJECTIVES: This study examines whether the receipt of enhanced prenatal or human immunodeficiency virus (HIV) medical services is associated with in-pregnancy emergency department (ED) utilization by HIV-infected women. METHODS: Medicaid and vital statistics records were linked for 1,826 women who are infected by HIV and who were delivered from 1993 to 1995 while receiving New York State Medicaid. The authors examined two types of ambulatory care--the Prenatal Care Assistance Program (PCAP) and enhanced care focused on HIV--that offer additional services in exchange for increased Medicaid reimbursement. From logistic regression models, the authors estimated adjusted associations of these types of care with ED use during pregnancy not leading directly to hospitalization. RESULTS: Fifty-three percent of pregnant women visited the ED. Women with ED use averaged 2.0 visits (SD = 1.1). After adjustment for demographic and substance use factors, enhanced care focused on HIV was not associated with any ED use (OR = 1.11, 95% CI 0.94, 1.30) or, among those using the ED at least once, with number of visits (P = 0.84). Interactions of receipt of PCAP care with the Adequacy of Prenatal Care Utilization Index (APNCU) and having a usual source of care in pregnancy improved model fit (P < 0.001 and P = 0.06, respectively). PCAP was associated with increased ED use only among women with inadequate APNCU or no usual source of prenatal care. CONCLUSION: Pregnant women infected with HIV receiving Medicaid relied heavily on ED care. Use of the ED was not associated with services focused on HIV but was positively associated with enhanced prenatal care. The association of enhanced prenatal care with greater ED use was curbed for women with more timely and adequate prenatal care visits or a usual source of prenatal care.

The efficacy of early adjuvant radiation therapy for pT3N0 prostate cancer: a matched-pair analysis.

PURPOSE: This study examines the effect of adjuvant radiation therapy (RT) on outcome in patients with pT3N0 prostate cancer and makes comparisons to a matched control group. METHODS AND MATERIALS: At our center, 149 patients undergoing radical prostatectomy were found to have pT3N0 prostate cancer, had an undetectable postoperative prostate-specific antigen (PSA) level, and had no immediate hormonal therapy. Fifty-two patients received adjuvant RT within 3 to 6 months of surgery. Ninety-seven underwent radical prostatectomy alone and were observed until PSA failure. From these two cohorts, we matched patients 1:1 according to preoperative PSA (<10 ng/ml vs. >10 ng/ml), Gleason score (<7 vs. > or =7), seminal vesicle invasion, and surgical margin status. Seventy-two patients (36 pairs) were included in the analysis. Median follow-up time was 41 months. We calculated a matched-pairs risk ratio for cumulative risk of PSA relapse (a rise above 0.2 ng/ml). RESULTS: After controlling for the prognostic factors by matching, there was an 88% reduction (95% confidence interval [CI]: 78-93%) in the risk of PSA relapse associated with adjuvant RT. The 5-year freedom from PSA relapse rate was 89% (95% CI: 76-100%) for patients receiving adjuvant RT as compared to 55% (95% CI: 34-79%) for those undergoing radical prostatectomy alone. CONCLUSIONS: These data suggest that adjuvant RT for pT3N0 prostate cancer may significantly reduce the risk of PSA failure as compared to radical prostatectomy alone. Its effect on clinical outcome awaits further follow-up.

Risk for colorectal cancer after gynecologic cancer.

BACKGROUND: Studies have suggested that women with previous diagnoses of gynecologic cancer (cervical, endometrial, or ovarian) have an increased risk for colorectal cancer. OBJECTIVE: To quantify risk for colorectal cancer after gynecologic cancer, both overall and for subgroups defined by age at diagnosis, cancer stage at diagnosis, ethnicity, and duration of follow-up. DESIGN: Retrospective cohort analysis of the Surveillance, Epidemiology, and End Results (SEER) program database from 1974 through 1995. SETTING: U.S. cancer registry. PATIENTS: 21,222 patients with cervical cancer, 51,680 patients with endometrial cancer, and 28,832 patients with ovarian cancer. MEASUREMENTS: Standardized incidence ratios (SIRs) were calculated for each gynecologic cancer site and for subgroups to represent the relative risk for colorectal cancer in women with previously diagnosed gynecologic cancer compared with women without gynecologic cancer. Poisson regression methods adjusting simultaneously for all study variables were used to estimate relative risks for colorectal cancer across subgroups with each gynecologic cancer. RESULTS: Overall, risk for colorectal cancer was elevated among women with previous ovarian cancer (SIR, 1.36 [95% CI, 1.21 to 1.53]). Risk was greatest in women who received a diagnosis before 50 years of age (SIR, 3.67 [CI, 2.74 to 4.80]) but was also elevated in women who received a diagnosis between 50 and 64 years of age (SIR, 1.52 [CI, 1.25 to 1.83]). The risk for colorectal cancer after endometrial cancer was also elevated substantially if endometrial cancer was diagnosed before the age of 50 (SIR, 3.39 [CI, 2.73 to 4.17]). No apparent risk elevation was associated with previous cervical cancer. CONCLUSIONS: Previous endometrial or ovarian cancer, particularly when diagnosed at an early age, increases subsequent risk for colorectal cancer. Greater emphasis on colorectal cancer screening in these populations may be necessary.

Translating clinical trial results into practice: the effect of an AIDS clinical trial on prescribed antiretroviral therapy for HIV-infected pregnant women.

BACKGROUND: The success of Pediatric AIDS Clinical Trials Group (PACTG) Protocol 076 in preventing vertical HIV transmission prompted intensive efforts to inform lay-persons and professionals about the trial's results. OBJECTIVE: To explore community responsiveness to these efforts by assessing temporal, maternal, and health care factors associated with prescribed antiretroviral therapy before and after PACTG Protocol 076. DESIGN: Retrospective cohort study. SETTING: New York State Medicaid program. PATIENTS: 2607 HIV-infected women who delivered a living child between January 1993 and September 1996. MEASUREMENTS: Adjusted odds of being prescribed antiretroviral treatment in the second or third trimester for women who delivered in period 1 (during the trial [January 1993 to February 1994]), period 2 (after the trial's end and announcement of the results to publication of the results [March 1994 to November 1994]), and period 3 (after publication of the trial results [December 1994 to September 1996]). RESULTS: The adjusted odds of being prescribed antiretroviral therapy increased 21% per month in period 2 and decreased to 3% per month in period 3. In all time periods, the adjusted odds of being prescribed antiretroviral therapy were at least 60% greater (P < 0.05) for women who were treated at an institution that performed HIV clinical trials, received HIV-focused ambulatory care, or had adequate prenatal care visits. After the trial, women receiving methadone treatment had at least twofold (95% CI, 1.5- to 4.3-fold) greater adjusted odds of being prescribed antiretroviral therapy than women who did not take any illicit drugs. Latin-American women, older women, and women born in the United States had greater adjusted odds (P < 0.05) of being prescribed treatment in period 3. CONCLUSION: Community practice responded rapidly to efforts to disseminate the results of PACTG Protocol 076; however, the absolute increase in prescribed therapy was greatest for women who had adequate prenatal visits or were receiving HIV-focused care, care at a site that performed clinical trials, or methadone therapy.

Predictors of compliance with recommended cervical cancer screening schedule: a population-based study.

BACKGROUND: The prevalence of routine cervical cancer screening and compliance with screening schedules are low compared to the Year 2000 objectives. Identifying predictors of routine screening and screening schedule compliance will help achieve these objectives. METHODS: We analyzed data from probability samples of 1,609 Missouri women responding to both the 1994 Behavioral Risk Factor Surveillance System (BRFSS) and the Missouri Enhanced Survey (ES). We generated prevalence odds ratios to identify predictors of non-compliance to cervical cancer screening guidelines. Also, among a sample of women reporting a reason for last Pap test, we estimated the relative odds of a screening v. diagnostic exam. RESULTS: In the combined probability sample, compliance with screening schedule was likely among women younger than 50 years of age and women who had either a recent mammography or a clinical breast exam. Being African-American, not experiencing a cost barrier when seeking medical care, having at least a high-school education and health coverage were each associated with an increased compliance with a screening schedule in the combined probability sample. Among women in the combined probability sample, whites, those who experienced no cost barrier to seeking medical care, the non-obese, and those who had a recent mammography were each more likely to have had a screening as opposed to a diagnostic exam. DISCUSSION: Cancer control and cardiovascular (CVD) prevention programs should consider jointly targeting those at high risk for cervical cancer and CVD because of aging and associated high-risk behavior such as non-compliance with cervical cancer screening, smoking, and obesity. Also, further research is needed to examine whether the increased compliance with cervical cancer screening guidelines among African American women may be in part due to higher occurrence of diagnostic Pap smears.

Predictors of Medicare costs in elderly beneficiaries with breast, colorectal, lung, or prostate cancer.

BACKGROUND: Determining the apportionment of costs of cancer care and identifying factors that predict costs are important for planning ethical resource allocation for cancer care, especially in markets where managed care has grown. DESIGN: This study linked tumor registry data with Medicare administrative claims to determine the costs of care for breast, colorectal, lung and prostate cancers during the initial year subsequent to diagnosis, and to develop models to identify factors predicting costs. SUBJECTS: Patients with a diagnosis of breast (n = 1,952), colorectal (n = 2,563), lung (n = 3,331) or prostate cancer (n = 3,179) diagnosed from 1985 through 1988. RESULTS: The average costs during the initial treatment period were $12,141 (s.d. = $10,434) for breast cancer, $24,910 (s.d. = $14,870) for colorectal cancer, $21,351 (s.d. = $14,813) for lung cancer, and $14,361 (s.d. = $11,216) for prostate cancer. Using least squares regression analysis, factors significantly associated with cost included comorbidity, hospital length of stay, type of therapy, and ZIP level income for all four cancer sites. Access to health care resources was variably associated with costs of care. Total R2 ranged from 38% (prostate) to 49% (breast). The prediction error for the regression models ranged from < 1% to 4%, by cancer site. CONCLUSIONS: Linking administrative claims with state tumor registry data can accurately predict costs of cancer care during the first year subsequent to diagnosis for cancer patients. Regression models using both data sources may be useful to health plans and providers and in determining appropriate prospective reimbursement for cancer, particularly with increasing HMO penetration and decreased ability to capture complete and accurate utilization and cost data on this population.

Breast cancer mortality in black and in white women: a historical perspective by menopausal status.

To examine racial/ethnic differences in breast cancer mortality over time by menopausal status, data from published U.S. Vital Statistics tables (1950-1992) and the Surveillance, Epidemiology, and End Results Program of the National Cancer Institute (1973-1991) were used to calculate age-adjusted breast cancer mortality and incidence rates. Overall, breast cancer mortality rates for white women were relatively stable from 1950 to 1992. In contrast, breast cancer mortality rates for black women increased during this period. Among premenopausal women there was no difference in breast cancer mortality between black and white women from 1950 to about 1975. However, after 1975, mortality rates in black premenopausal women increased, whereas those in white women decreased. Among postmenopausal women, breast cancer mortality was substantially lower in blacks than in whites in 1950. Between 1950 and 1992, rates in blacks increased and eventually exceeded rates in whites, which remained stable during this period. This excess in breast cancer mortality in black women is not explained by changes in breast cancer incidence rates. There is an unexplained epidemic of breast cancer mortality in black women that appears to differ somewhat by menopausal status. Reasons for temporal increases in breast cancer mortality seen only among black women need to be identified, as do reasons for the heterogeneity of trends by menopausal status.

Does comorbid disease interact with cancer? An epidemiologic analysis of mortality in a cohort of elderly breast cancer patients.

BACKGROUND: Although widely believed that co-occurring chronic diseases in elderly persons do not act independently in causing death, there has been little empirical research assessing prognostic interrelationships between comorbidities. METHODS: Nonconcurrent prospective follow-up of 3,549 Virginia-resident elderly women diagnosed with a first breast cancer and 2,114 elderly women with no breast cancer history admitted to Virginia hospitals with principal diagnoses of genital prolapse during 1986-1988 was conducted through linkage of cancer registry and Medicare administrative records. Aggregate comorbidity was measured from Medicare claims via the Charlson comorbidity index (CCI). Mortality rates and relative risks were estimated for the breast cancer and non-breast-cancer groups stratified by the presence and level of comorbidity. Proportional hazards models were used to estimate Rothman's synergy index (S) measure of additive interaction. RESULTS: Over full follow-up, the excess mortality rate for women with breast cancer and other comorbidity was 17% greater than expected under the null hypothesis that risks were additive and independent (S = 1.17, p = .12). Stratified analyses revealed a pattern of S estimates across cancer stage subgroups that was biologically sensible, but this pattern was not supported by strong statistical evidence. CONCLUSIONS: This study provides the first empirical estimates of statistical interaction between breast cancer and other chronic comorbidity. S index values tended to be small, but these small effects would translate into substantial numbers of deaths attributable to interaction between cancer and comorbidity. Interactions between breast cancer and comorbid disease should be explored further in large studies that can estimate these effects with increased precision.