Causes of death among cancer patients

Abstract

Background: The purpose of our study was to characterize the causes of death among cancer patients as a function of objectives: (i) calendar year, (ii) patient age, and (iii) time after diagnosis.

Patients and methods: US death certificate data in Surveillance, Epidemiology, and End Results Stat 8.2.1 were used to categorize cancer patient death as being due to index-cancer, nonindex-cancer, and noncancer cause from 1973 to 2012. In addition, data were characterized with standardized mortality ratios (SMRs), which provide the relative risk of death compared with all persons.

Results: The greatest relative decrease in index-cancer death (generally from > 60% to < 30%) was among those with cancers of the testis, kidney, bladder, endometrium, breast, cervix, prostate, ovary, anus, colorectum, melanoma, and lymphoma. Index-cancer deaths were stable (typically >40%) among patients with cancers of the liver, pancreas, esophagus, and lung, and brain. Noncancer causes of death were highest in patients with cancers of the colorectum, bladder, kidney, endometrium, breast, prostate, testis; >40% of deaths from heart disease. The highest SMRs were from nonbacterial infections, particularly among <50-year olds (e.g. SMR >1,000 for lymphomas, P < 0.001). The highest SMRs were typically within the first year after cancer diagnosis (SMRs 10-10,000, P < 0.001). Prostate cancer patients had increasing SMRs from Alzheimer's disease, as did testicular patients from suicide.

Conclusion: The risk of death from index- and nonindex-cancers varies widely among primary sites. Risk of noncancer deaths now surpasses that of cancer deaths, particularly for young patients in the year after diagnosis.

Keywords: SEER; United States; cancer; clinical trials; comorbidity; competing events; epidemiology; heart disease; mortality; non-cancer death; second cancer.

Figure 1.

Plots of patient death versus attained calendar year (from 1973 to 2012), for various cancers. Death was characterized as due to “index-cancer,” (black lines; i.e. the cancer originally diagnosed in the patient), “nonindex-cancer” (gray lines; i.e. a second cancer diagnosed in the patient, and not a metastasis from the original cancer), and “noncancer death” (green lines; i.e. death from any medical cause that does not include cancer). Attained calendar year refers to the year in which the death occurred, among all cancers diagnosed since 1973. Cancers subsites are organized by color: light tan = head and neck, skin; genitourinary = yellow; male specific = light blue; female specific = pink; gastrointestinal = dark tan; gray = hematologic.

Figure 2.

(A) A plot of death of the proportion of all cancer patients experiencing death due to index-cancer (black), nonindex-cancer (gray), and noncancer cause (gray) among all patients, illustrates that noncancer causes of death have increased while index-cancer causes of death have decreased since the 1970s. Plots of death due to index-cancer (B), nonindex-cancer (C), and noncancer cause (D), among all patients experiencing a death of that cause, stratified by various cancers.

Figure 3.

SMRs for the leading causes of noncancer death were characterized after diagnosis, binned by patient age. An SMR above 1 represents a higher relative risk of death for a particular cause, when compared with the noncancer population. Black lines represent SMRs from all sites; purple from lung and bronchus cancer; blue from prostate cancer; pink from breast cancer; and brown from colorectum. Select cancers with outlier SMRs (green, red, and yellow lines) are also plotted.

Figure 4.

SMRs for the leading causes of noncancer death were characterized after diagnosis, binned by time periods. An SMR above 1 represents a higher relative risk of death for a particular cause, when compared with the noncancer population. Black lines represent SMRs from all sites; purple from lung and bronchus cancer; blue from prostate cancer; pink from breast cancer; and brown from colorectum. Select cancers with outlier SMRs (green, red, and yellow lines) are also plotted.