Demographic characteristics

Demographic variables included age, sex, educational attainment, race/ethnicity, marital status, and income. Total individual and family annual incomes were reported for all respondents, with missing data imputed using logical editing and weighted, sequential hot decks, as well as top coding to preserve confidentiality. Sources of income included the following: interest, business, dividends, refunds, retirement, alimony, sales, trust, social security, unemployment, workers’ compensation, veterans’ income, cash, child support, public assistance, and any other form. Family income as a percentage of the poverty line was categorized as either low income (1=199% of the federal poverty level or less) or middle and high income (0=200% of the federal poverty level or higher).

Continue Reading

Insurance coverage

CSAQ respondents indicated if they were covered by insurance that paid for all or part of their medical care, tests, or cancer treatment at the time of diagnosis. These responses were dichotomized as either not insured (1= no) or insured (0= yes).

Cancer characteristics

Cancer characteristic variables included the type of cancer and current treatment status. Current treatment status was defined as those who were actively undergoing cancer treatment at the time of CSAQ administration (1=yes, 0=no). The three most common cancer diagnoses along with the respondents’ current treatment status were also extracted from the CSAQ.

Related Articles

Employment status

Employment status was based on a single item and was coded to reflect unemployed (1=yes, 0=no). Those who were employed at the time of survey or had a job to return to were coded as 0 (ie, not unemployed); those who were employed during the reference period but not employed at the time of survey or who were not employed with no job to return to at the time of the survey were coded as 1 (ie, unemployed). This allowed logistic regressions to estimate the odds of cancer survivors being unemployed at the time of the survey.

Cancer-related psychological job distress

Two items surveyed respondents’ anxiety related to the cancer–work interface. First, respondents were asked if they were worried about being forced to retire or quit early due to cancer-related effects on their health. Next, respondents were asked if they were worried about fulfilling home or job responsibilities if their cancer returned or got worse. Both items were coded dichotomously (1=yes, 0=no).

Cancer-related interference with job tasks

Two items assessed how a cancer diagnosis, its treatment, or the effects of the treatment interfered with job tasks. The first asked if cancer affected the respondent’s ability to perform any physical tasks required at work, whereas the second probed if cancer interfered with any mental tasks required at work. Responses were also coded dichotomously (1=yes, 0=no).

Employment characteristics

Employment characteristics referred to the job that respondents had since their cancer diagnosis and were described in two categories as follows: work leave and workplace supports. Two items assessed respondent’s leave from work due to their cancer diagnosis. Respondents indicated if they took extended paid time off from work (no, yes), and subsequently if they took extended unpaid time off from work (no, yes); to aid in interpretation of odds of unemployment, these items were recoded (1=no, 0=yes). The following three questions identified workplace supports: 1) changing to a flexible schedule; 2) to a less demanding job; or 3) from a full to part time job; these were coded dichotomously as either did not change (0=no) or changed (1=yes).

Statistical analyses

All analyses accounted for weighting using complex sample statistical design. Independent samples Student’s t-tests compared employed vs unemployed cancer survivors for age, whereas differences in sex, income level, health insurance status, major cancer diagnoses, current cancer treatment status, mental and physical work interference, paid and unpaid time off, and changes in work conditions were compared by employment status using Chi-square statistics. Logistic regression analyses were used to calculate the odds of being unemployed in the 5 years since the respondents’ cancer diagnosis using age, sex, income, and health insurance status as control variables. Cancer-related psychological job distress, cancer interference with job tasks (mental and physical), leave, and changes in schedules and job status were assessed in a series of stepwise logistic regression models. All analyses were conducted using SPSS software, version 24 (IBM Corporation, Armonk, NY, USA). Findings were considered statistically significant using two-tailed P-values of <0.05.