Current Projects Using SEER-MHOS Data Resource

1. To estimate absolute and relative disparity in physical and mental health outcomes based on SF-36/VR-12 among patients before, during, and after their cancer diagnosis, comparing with non-cancer control groups
2. To examine overall survival by physical and mental health outcome, based on physical component summary (PCS) and mental component summary (MCS) measured by SF-36/VR-12
3. To investigate the potential effect modification of the PCS and MCS on survival by key prognostic factors (i.e., second primary lung cancer [SPLC], stage, surgery, dual eligibility to Medicare and Medicaid, etc.)

1. a) Characterize the complex MM phenotypes in older adults diagnosed with incident cancer; and b) Compare MM phenotypes between cancer patients and their non-cancer counterparts.
2. Identify phenotypes of complex MM that are associated with advanced cancer stage at diagnosis, delayed treatment initiation, and treatment.
3. Among persons receiving cancer treatment, determine the complex MM phenotypes that are associated with functional decline and poorer overall and disease-specific survival.

1. To examine clusters of social determinants of health (e.g. rurality, income) associated with health-related quality of life in cancer survivors.
2. To examine rural-urban differences in clusters of health behaviors (e.g., tobacco use, exercise, caregiving, sleep).
3. Investigate the relationship of social determinant clusters to health behavior clusters impacting health-related quality of life in cancer survivors.

1. Compare HRQOL overtime in older cancer survivors with cardiovascular diseases and older adults without cancer by race/ethnicity, and area-level characteristics (county-level SES and HSS characteristics)
2. Compare HRQOL overtime in older cancer survivors with pulmonary diseases and older adults without cancer by race/ethnicity, and area-level characteristics (county-level SES and HSS characteristics)
3. Compare HRQOL overtime in older cancer survivors with diabetes and older adults without cancer by race/ethnicity, and area-level characteristics (county-level SES and HSS characteristics)
4. Compare HRQOL overtime in older cancer survivors with musculoskeletal diseases and older adults without cancer by race/ethnicity, and area-level characteristics (county-level SES and HSS characteristics)
5. Develop a simulation model to estimate future trends (2020-2040) in HRQOL in older survivors by chronic condition clusters based on the trends observed in Research Questions 1-4.

1. Identify patterns of supportive care use (e.g., continuity, frequency, and intensity) among gyn cancer (specifically, ovarian or uterine) patients in comparison with a matched cohort of women who have not been diagnosed with cancer.
2. Identify predictors of the use of supportive care (prescription drugs) among women diagnosed with gyn cancer.
3. Identify the impact of the supportive care received on survival and HRQoL (e.g., PCS and MCS, and also VR-6D) among women diagnosed with gyn cancer.

We aim to use SEER-Medicare and SEER-MHOS data to describe stage-specific breast, prostate and colorectal cancer natural history and survival, and clarify the impact of comorbidities, medications and cancer treatment on survival outcomes and quality of life in order to best determine which groups of survivors will benefit most from lung cancer screening.

1. To elucidate pre- to post-diagnosis depression trajectories among survivors of cancer, using the ten most prevalent cancer types in the SEER-MHOS data linkage.
2. To compare predictors and survival outcomes among latent depression trajectory subtypes.
3. To compare receipt of cancer-directed surgery and timeliness of cancer-directed surgery among latent depression trajectory subtypes. Note: this aim will be restricted to solid tumor cancers.

1. explore trajectories of HRQOL among older adult survivors of breast, colorectal, lung or prostate cancer beginning pre-diagnosis and continuing to the 10-year survivorship time point.
2. identify patient demographic and clinical factors associated with different trajectories of HRQOL among older adult survivors of breast, colorectal, lung or prostate cancer within 10 years of survivorship, and to
3. identify subgroups of older adult survivors who are at the greatest risk to experience declines in HRQOL following a diagnosis of breast, colorectal, lung or prostate cancer.
4. compare demographic and clinical factors between older adults without a cancer diagnosis and older adult survivors of breast, colorectal, lung or prostate cancer.

1. Compare changes in patient reported outcomes (PROs) after a cancer-diagnosis among non-AIDS defining cancer patients living with and without HIV.
2. Estimate associations of pre-cancer diagnosis PROs with overall survival among non-AIDS defining cancer patients living with HIV.
3. Explore differences in supportive medication use to control symptoms associated with pain and emotional distress (adverse PROs) among non-AIDS defining cancer patients living with and without HIV.

1. To describe the HRQOL change over time in elderly women. We will evaluate the secular trend of HRQOL from the time of diagnosis and will compare these trends between women with breast cancer (cases) and without breast cancer (controls).
2. To evaluate the effects of RT and comorbidities on HRQOL trajectories over time among elderly breast cancer survivors. HRQOL score will be compared among 4 groups of elderly women in the generalized linear mixed models after controlling for baseline HRQOL, age, race, BMI, socioeconomic status, cancer stage, and cancer histology. We hypothesize that women who have ≥ 2 comorbid conditions and have received RT would have the greatest HRQOL impairment (deficit) in physical functioning and mental health compared to women who had <2 comorbid conditions and did not receive RT.
3. To evaluate the effects of RT and comorbidities on the outcomes of OS/BSS among elderly breast cancer survivors. We hypothesize that women who have <2 comorbid conditions and have received radiation treatment will have best OS/BSS compared to women who had ≥ 2 comorbid conditions and did not receive RT.

1. To examine race/ethnicity (NHB/NHW/Hispanic) disparities in patient reported outcomes (PROs) in elderly breast cancer (BC) patients using the SEER-MHOS dataset.
2. To examine subtype-specific, race/ethnicity disparities in PROs in elderly BC patients using the SEER-MHOS dataset.
3. To examine stage-specific, race/ethnicity disparities in PROs in BC patients using the SEER-MHOS dataset

1. Examine the change in HRQOL of patients with hepatopancreatic cancers (pre-and post-diagnosis) and determine its predictors.
2. Compare the change in HRQOL of patients with hepatopancreatic cancers to controls without cancer.
3. Evaluate the associations between changes in HRQOL and overall survival in patients with hepatopancreatic cancer.

Among our chief aims, we plan to quantify the mortality and incidence of secondary cancer in cancer patients prescribed a statin 6 months prior to diagnosis versus those who were not. Moreover, we plan to utilize the MHOS survey to compare all breast and hepatocellular cancer patients’ BMI 6 months prior to diagnosis. This 6 month timing for statin use and self-reported BMI 6 months prior to diagnosis has been shown to eliminate bias across the board. (PMID: 27491389).

Based on the promising existing literature, our primary hypothesis states that both breast and hepatocellular cancer patients prescribed a statin will have statistically significantly lower mortality rates and incidences of secondary cancer than those not prescribed a statin. Study of the different obesity-linked cancer sites including breast and liver will be driven based on the biologic characteristics of that cancer. Specifically for our breast cancer population, we predict that adjuvant statins will be beneficial to patients with hormone receptor positive breast cancer, which is one of the most common obesity related cancers, on endocrine therapy, suggesting a synergistic benefit. No studies have explored this drug-drug interaction. Our secondary hypothesis states that patients who self-report a BMI of 30.0 or above and received a statin 6 months prior to diagnosis will have statistically significantly lower mortality rates and incidences of secondary cancer than those not prescribed a statin. Ultimately, our team will still explore other trends once the data is received.

1. To identify factors associated with HRQOL between cancer patients and matched non-cancer participants
2. To estimate absolute and relative disparity in changes in HRQOL among patients before and after their cancer diagnosis
3. To assess difference in utilization of cancer treatment among age-, stage- and histology-specific subgroups of patients, impacted by status of HRQOL
4. To examine overall survival by HRQOL status, concurrently identifying significant clinical or behavioral effect-modifiers (i.e., stage, smoking status, etc.) for the impact of HRQOL on overall survival

* All analyses will be conducted separately for BC and LC patients.

1. To examine race/ethnicity (NHB/NHW/Hispanic) disparities in patient reported outcomes (PROs) in elderly breast cancer (BC) patients using the SEER-MHOS dataset.
2. To examine subtype-specific, race/ethnicity disparities in PROs in elderly BC patients using the SEER-MHOS dataset.
3. To examine stage-specific, race/ethnicity disparities in PROs in BC patients using the SEER-MHOS dataset.

1. Multivariable general linear regression models will be used to compare mean scores on the physical component summary, the mental component summary and the 8 subscales between endometrial cancer survivors and women without cancer. Similar models will be used to evaluate associations between patient and cancer-related characteristics among endometrial cancer survivors. As sample sizes permit, we will also use general linear regression models to evaluate change in HRQOL among those with surveys completed before and after diagnosis.
2. For each ADL separately, we will use Poisson regression models with robust error variance to estimate prevalence ratios comparing the prevalence of a limitation (difficulty with activity or unable to do activity) between endometrial cancer survivors and the comparison group. We will also sum the number of ADL limitations (out of 6 possible) and compare the mean between women with and without endometrial cancer. We will then use prevalence ratios to estimate associations between patient and cancer-related characteristics and limitations in ADLs among endometrial cancer survivors.
3. Using Poisson regression models with robust error variance, we will estimate prevalence ratios comparing the prevalence (within the previous 12 months) of falls and walking or balance problems between women with and without a history of endometrial cancer. We will also estimate prevalence ratios to examine associations between patient and cancer-related characteristics and falls/walking or balance problems among endometrial cancer survivors.

1. To examine the following selected self-reported comorbidity prevalence among MA beneficiaries diagnosed with cancer (top ten cancer sites) and those without any history of a cancer diagnosis: none, hypertension, angina/coronary artery disease, congestive heart failure, respiratory conditions (emphysema/asthma/chronic obstructive pulmonary disease (COPD)), osteoporosis, diabetes, and depression).
2. To utilize the MHOS-SEER-Medicare Part D linkage to evaluate the medication management of comorbidities for patients identified in Aim 1 (MA beneficiaries diagnosed with cancer and those never been diagnosed with cancer). Medication management will be assessed based on clinical guidelines for each individual comorbidity (i.e. COPD- GOLD Guidelines, JNC 8, ADA Standards of Medical Care in Diabetes).
3. To assess the relationship between comorbidity medication management and HRQOL among MA cancer patient enrollees, and the role of obesity as an interaction factor with medication management to subsequently affect HRQOL. HRQOL will be assessed using the responses to the MHOS and VR-12 Physical Component Summary (PCS) and Mental Component Summary (MCS) scores. Obesity will be calculated based on the self-reported height and weight values from the MHOS.cohorts that undergo surgery to those that do not undergo surgery.

1. To describe the HRQOL change over time in elderly women. We will evaluate the secular trend of HRQOL from the time of diagnosis and will compare these trends between women with breast cancer (cases) and without breast cancer (controls).
2. To evaluate the effects of RT and comorbidities on HRQOL trajectories over time among elderly breast cancer survivors. HRQOL score will be compared among 4 groups of elderly women in the generalized linear mixed models after controlling for baseline HRQOL, age, race, BMI, socioeconomic status, cancer stage, and cancer histology. We hypothesize that women who have ≥ 2 comorbid conditions and have received RT would have the greatest HRQOL impairment (deficit) in physical functioning and mental health compared to women who had < 2 comorbid conditions and did not receive RT.
3. To evaluate the effects of RT and comorbidities on the outcomes of OS/BSS among elderly breast cancer survivors. We hypothesize that women who have < 2 comorbid conditions and have received radiation treatment will have best OS/BSS compared to women who had ≥ 2 comorbid conditions and did not receive RT.

1. The primary objective will be to determine if patients undergoing surgical resection can return to their baseline health related quality of life following surgery. To test the null hypothesis that there is no impact on health-related quality of life, we will use a paired samples approach to compare population-level mean Physical Component Score (PCS) and Mental Component Score (MCS). In addition, at the individual level, we will determine the differential change in both PCS and MCS by stratifying patients based on the percent change in each measure.
2. Secondary objectives include estimating the impact of time from surgery on health-related quality of life measures. To achieve this, population data comparing PCS and MCS based on time the survey was completed will be used to develop a predictive model using a time-series methodology. An additional objective will be to compare health-related quality of life in study population (with cancer) to a propensity matched cohort without cancer. Matched groups will be compared on each sub-category of PCS and MCS to understand the quality of life elements most impacted by a diagnosis of surgically treatable disease. An additional propensity matched sub-group analysis is planned, comparing cohorts that undergo surgery to those that do not undergo surgery.

1. Compare older cancer patients with or without other chronic condition clusters by sociodemographic characteristics and medical conditions.
2. Compare HRQOL (PCS and MCS) by different types of cancers, number of chronic condition clusters, and time since cancer diagnosis (0-2 years, 2-5 years, 5+ years) among older cancer patients.
3. Compare HRQOL (PCS and MCS) by types of cancers, types of chronic condition clusters, and time since cancer diagnosis (0-2 years, 2-5 years, 5+ years) among older cancer patients.

1. Determine health-related quality of life before and after HPV-associated (anal, cervical, vaginal, vulvar, and oropharyngeal) cancers stratified by age and other demographic characteristics.
2. To analyze secular trends in HRQoL for HPV-associated cancers from 1998-2015.
3. To evaluate whether HRQoL differes by the site of tumor diagnosis (i.e., HPV-associated histology vs HPV-unrelated histology), stage at diagnosis, and treatment type.

1. We will evaluate the association between the number of patients and the type of targeted therapy, according to patient-level demographic characteristics, neighborhood-level demographic characteristics, cancer characteristics, clinical characteristics, and treatment characteristics. We will use the Chi-square test for categorical characteristics variables and t-test/ANOVA for continuous characteristics variables.
2. We will use a generalized linear model (GLM) to assess the association of the different types of targeted therapy with HRQoL, controlling for covariates. Additionally, we will conduct subgroup analyses to assess the association of the different types of targeted therapy with HRQoL according to comorbidity status.
3. We will use an item response theory (IRT) model to estimate the utility score based on HRQoL scores according to the use of the different types of targeted therapy.

1. We aim to extend the DAFI for use as a prognostic indicator in gynecologic oncology patients with the goal of personalizing treatment options.

1. The aim of this project is to investigate the association between complex multimorbidity and cancer outcomes in older adults newly diagnosed with cancer, including cancer stage, time to treatment initiation, cancer treatment modalities, and mortality. In patients with survey data before and after cancer diagnosis, we will examine changes in quality of life measures, accounting for varying levels of multimorbidity before cancer diagnosis, cancer stage, and treatment modalities.

1. Our primary endpoint will be improved quality of life in cancer patients with secondary endpoints of improved overall survival, improved disease-free survival, and patterns of morbidity and mortality.

1. Determine the prevalence of depression among older cancer survivors before and after cancer diagnosis. Further examine the difference in prevalence in the time periods before cancer diagnosis with the post cancer diagnosis period.
2. Assess the level of daily physical activity measured through patient reported activities including lifting or carrying groceries, climbing flights of stairs, bending, kneeling, sitting, using toilet, eating, walking and bathing among older cancer survivors in the past 12-month period before and after cancer diagnosis. Further examine the difference in levels of physical activity in the time periods before cancer diagnosis with the post cancer diagnosis period.
3. Examine health-related quality of life measured using VR-12 instrument among older cancer survivors in the past 24-month period before and after cancer diagnosis. In addition, inspect any difference in measured HRQOL values in the time spans before and after cancer diagnosis.

1. To prospectively identify changes in HRQol after diagnosis of HCC in comparison with a group of adults without HCC.

1. Determine whether HRQOL will be lower for SEER-MHOS participants in rural areas than in urban areas.
2. Determine whether HRQOL will be lower for SEER-MHOS cancer patients in rural areas than cancer patients in urban areas.
3. Determine whether the association between cancer status and HRQOL will vary by urbanicity, such that the negative relationship between cancer status and HRQOL will be larger in rural areas than in urban areas.
4. Determine whether the associations among cancer status, urbanicity, and HRQOL will vary by race/ethnicity, such that racial/ethnic minority cancer patients living in rural areas will have the lowest HRQOL.

Using data from cohorts of older adults a decade apart, Cohort 5 (2002-2004) and Cohort 15 (2012-2014), we will explore the following specific aims:

1. To compare HRQOL at baseline between propensity-score matched cancer survivors and non-cancer participants within Cohorts 5 and 15.
2. To compare HRQOL at baseline and 2-year follow-up between cancer survivors from Cohort 5 and Cohort 15.
3. To explore relationships between patient demographic and clinical variables and HRQOL in cancer survivors from Cohort 5 and Cohort 15.
4. To explore whether significant correlates of HRQOL at baseline serve as predictors of HRQOL at 2-year follow-up in Cohorts 5 and 15.

1. We plan to assess prevalence of vitality/fatigue post-diagnosis and predictors of vitality/fatigue change from pre- to post-diagnosis among breast, prostate and colorectal cancer survivors in SEER-Medicare Health Outcomes Survey (MHOS).
2. As a secondary outcome we will look at prevalence and predictors of changes in physical and mental summary scores for quality of life from pre- to post-diagnosis.

1. An ANOVA will be calculated to test hypothesis one, to determine if patients who undergo prostate brachytherapy will exhibit the highest rates of depression compared to the other two treatment groups with patients undergoing external beam radiation exhibiting the lowest scores. The independent variable will be prostate cancer treatment, and the dependent variable will positive screening of depression. My goal is to see whether various cancer treatment methods are more likely to put individuals at risk of depression.
2. Second, an ANOVA will be calculated to test hypothesis two, to determine if patients who undergo prostate brachytherapy will exhibit the lowest scores of HRQoL compared to the three treatment groups with patients undergoing external beam radiation exhibiting the highest scores of HRQoL. The independent variable will be prostate cancer treatment, and the dependent variable will be the HRQoL rating.
3. Third, a multiple regression will be calculated to test hypothesis three, to determine if older men have a decreased rate of depression compared to younger men and if African American men are at a greater risk of depression compared to men of other races. Age and race will be entered as independent variables, with the sample racial category being dummycoded in separate bi-level variables, using the methods outlined by Cohen, Cohen, West and Aiken (2003).
4. Lastly, a mediation analysis will be calculated to test hypothesis four, to determine if the relationship between the prostate cancer treatment and depression is mediated by HRQoL. To conduct these analyses, guidelines provided by Frazier, Tix, and Barron (2004) will be used. First, the predictor (prostate cancer treatment) must predict the presumed outcome variable (depression). Second, the predictor variable (prostate cancer treatment) should have a statically significant relationship with the presume mediator (HRQoL). Third, the presumed mediator (HRQoL) must predict the outcome variable (depression). Finally, the analyses must demonstrate the relationship between the predictor (prostate cancer treatment) and outcome (depression) is nonsignificant/equivalent to zero when the mediator is added to the model (Frazier et al.).

1. To describe the HRQOL distributions in SEER-MHOS patients diagnosed with the cancer of interest. All subscales scores and summary scores (based on SF-36 or VR-12) will be summarized, respectively. A health utility score (SF-6D/VR-6D) as an overall measure of health will also be calculated. The findings will be stratified by cancer stages if sample sizes permit, and survey time since diagnosis (every 6 months). Same analysis will be conducted in MHOS participants without cancer. The findings from the cancer free group will serve as a benchmark for the cancer group.
2. To explore the associations between HRQOL and patient characteristics, including demographic factors (e.g., age, gender, race/ethnicity, marital status, household income, education level, geographic location), life style factors (e.g., smoking, BMI), and medical history (e.g., self-reported comorbidities, cancer stage and treatment modalities).

1. Develop a patient-reported geriatric assessment tool, mapping onto existing CGA domains (Table 1.), and to examine the prognostic value of that tool on survival of older cancer patients. In creating this tool, we plan to develop the tool in a sample of older adults without cancer and validate the tool in a sample of older cancer survivors.
2. Determine the relationship between self-reported geriatric assessment (individual domains and overall score) and overall survival (time to event) among older adults without a history of cancer.
3. Determine the relationship between self-reported geriatric assessment (individual domains and overall score) and overall survival (time to event) among older adults with cancer.

1. Examine disparity trends in physical and mental health status as a measure of health-related quality of life (HRQoL) among Medicare Advantage beneficiaries with and without a cancer history, by race/ethnicity.
2. Examine disparity trends in physical and mental health status as measure of health-related quality of life (HRQoL) among Medicare Advantage beneficiaries with and without a cancer history, by highest level of education attained.
3. Assess if disparity of HRQoL by race/ethnicity widens over the study time period (1998-2012).
4. Assess if disparity of HRQoL by educational attainment widens over the study time period (1998-2012).

1. Compare the receipt of physical activity conversations with healthcare providers between older cancer survivors and their peers without a history of cancer.
2. Categorize older cancer survivors into age groups (young-old (65-69 years), older-old (70-80 years), and oldest-old (80+ years)), and compare frequencies of physical activity discussions between these groups.
3. Assess the interactions between chronological age and self-reported physical and mental health (PCS and MCS) and their impact on receipt of physical activity discussions with healthcare providers.
4. Assess the interactions between cancer history and self-reported physical and mental health (PCS and MCS) and their impact on receipt of physical activity discussions with healthcare providers.

1. Will involve descriptive analysis of patterns of pain and depression and changes in these symptoms between the first and second interview (for those with more than one interview), comparing beneficiaries with and without cancer.
2. Analyses will begin with a description of patterns of medication therapy specific to reported pain and/or depression.
3. Will examine whether medication therapy is associated with reduction or elimination of symptoms of pain and depression from the first to the second MHOS survey.
4. Will examine whether symptoms of pain or depression are associated with adherence (duration and adherence, measured by medication possession ratio (MPR) to oral therapy, particularly hormone therapy for women diagnosed with breast cancer and who have initiated therapy.

1. In the population of primary brain tumors, we would like to examine patient functional status and cognition and correlation with treatment received, including surgery, radiation, and chemotherapy.
2. For prostate cancer survivors, we intend to evaluate the incidence of, and modifying factors for, long-term outcomes after treatment of prostate cancer, with a focus on late gastrointestinal (GI), genitourinary (GU), and sexual toxicities.
3. For survivors of early stage lung cancer, we would like to evaluate quality of life measures as well as respiratory status and chest pain after primary treatment.

1. To describe demographics of elderly patients diagnosed with blood cancers between 2000-2013.
2. To identify trends in health outcomes and quality of life among elderly patients with blood cancers between 2000-2014, stratified by disease (leukemia, lymphoma, myeloma). This will include 2 years of follow up for all patients diagnosed through 2012. We will compare these results to survey respondents without a cancer diagnosis.
3. To identify the proportion of patients who have a change in their quality of life outcomes from diagnosis to 24 months after diagnosis.

1. To identify demographic, clinical, and functional factors associated with HRQOL among older women after diagnosis with gynecological cancer.
2. To evaluate factors associated with depressive symptoms among older women after diagnosis with gynecological cancer.
3. To determine the impact of diagnosis with gynecological cancer on HRQOL in older women.
4. To evaluate the effect of HRQOL on survival for older women with gynecological cancer.

1. We hypothesize that early stage lung cancer treated with radiation yields better quality of life than if it is treated with surgery.

1. To characterize functional impairments, multimorbidity, and health-related quality of life in older survivors of lung/bronchus and GI (esophagus, stomach, Small Intestine, colorectal, liver & bile duct, gallbladder & other biliary, pancreas, other digestive system) cancers who were surgically treated;
2. To evaluate oncologic outcomes (overall survival, cancer-specific survival) of older cancer survivors who were surgically treated for lung/bronchus and GI cancers;
3. To characterize the association of functional impairment and comorbidities on oncologic outcomes;
4. To evaluate the effect of race/ethnicity and socioeconomic status on functional impairment, multimorbidity, HRQOL, and oncologic outcomes.

1. Using SEER-MHOS, we seek to investigate self-reported quality of life in patients with renal cell carcinoma selecting surgical versus non-surgical management, and analyze its predictive ability on cancer-specific and overall survival.
2. Furthermore, we intend to compare and validate the results observed in the renal cell carcinoma population to patients with other genitourinary malignancies, like blagder, prostate, testis, etc.
3. Our hypothesis is that differences are likely to exist in the self-reported quality of life between patients who choose surgery versus non-surgical management and that self-reported quality of life questionnaires are a better measure of life expectancy/risk stratification than other metrics like the Charlson Comorbidity Index.

1. Assess the health-related quality of life (HRQOL) of underserved racial/ethnic populations that have been diagnosed with breast, cervical, colorectal, and prostate cancer.

1. “The goal of this study is to examine the patterns of aggressive cancer care in different patient strata by life expectancy. For many cancers, such as prostate cancer, aggressive treatment may involve surgery and/or multi-modality therapy. However, for older patients and those with comorbidities, often less aggressive treatment is pursued.

   Using patient age and existing comorbidities prior to cancer diagnosis (which are available in the SEER-MHOS dataset (the investigators will categorize patients into different life expectancy groups. Then in each group, patterns of care in terms of treatments received will be examined. While patterns of care studies are common in the cancer literature, how patient life expectancy affects aggressiveness of cancer treatment is not a well-studied subject. The latter is the focus of this proposed study.”

1. To investigate the feasibility of a study using the Surveillance, Epidemiology, and End Results- Medicare Health Outcomes Survey (SEER-MHOS) linked database to measure health-related quality of life in cancer survivors treated with an operation.
2. To characterize the physical and mental health of patients diagnosed with cancer using a large population-based database.
3. To compare self-reported physical and mental health between cancer survivors treated with and without surgical resection after controlling for age, race, socioeconomic status, co-morbidities, cancer histology, and cancer stage.
4. To compare self-reported physical and mental health between cancer and noncancer patients after controlling for age, race, co-morbidities, and socioeconomic status.
5. To examine disparities in cancer outcomes and its underlying attributes.

1. To describe racial differences in self-reported HRQOL in women after gynecologic cancer diagnosis compared to propensity-matched non-cancer controls.
2. To describe racial differences in self-reported HRQOL in gynecologic cancer patients compared to a cohort of breast cancer patients.
3. To examine factors associated with poor HRQOL in women diagnosed with gynecologic cancer.

1. Measure the calibration and discrimination of the original PROMPT in the sample of SEER-linked cancer patients for the top four cancers (breast, prostate, lung, and colorectal).
2. Using the SEER variables in addition to the MHOS data, develop a cancer-specific model to predict six month mortality in the four most prevalent cancers, first for combined cancer and then for each of the four cancers separately.

1. Compare 8 domains of HRQL among four groups of women age 65 and older: Group 1 (BrCa+HF): women with a history of breast cancer and HF; Group 2 (BrCa): women with a history of breast cancer but no history of HF; Group 3 (HF): women with HF but no history of cancer; Group 4 (NoCaHF): healthy women with no history of cancer or heart failure. To control for demographic and clinical variables, women in groups 2, 3, and 4 will be identified based on propensity score matching with women in group 1.25
2. Among the four groups, compare how each of the 8 domains of HRQL changes over time, from baseline to two years later.
3. For all women in the sample, determine the relationship between baseline HRQL and diagnosis status (breast cancer and HF diagnoses) and HRQL at two-year follow-up in each of the 8 domains.

1. To examine the changes in HRQOL, prevalence in limitations in ADLs, problems with balance or walking, fall risk management, and falls after the cancer diagnosis in older cancer survivors.
2. To evaluate the relationship between cancer-related clinical variables, HRQOL, ADLs, balance or walking, and falls in cancer survivors after their cancer diagnosis.
3. To identify among variables at baseline, including cancer-related clinical variables, HRQOL, limitations in ADLs, problems with balance or walking, fall risk management, and the previous history of falls, the unique factors predictive of falls in older cancer survivors.
4. Develop evidence-based clinical algorithms to optimize the outcomes including problems with balance or walking as well as falls in older cancer survivors.

1. Identify comorbidity clusters of multi-morbidities pre- and post- cancer diagnosis across cancer types.
2. Determine the symptom burden associated with different multi-morbidity clusters among cancer patients.
3. Determine the functional limitations associated with different multi-morbidity clusters among cancer patients.

1. Patients completing a MHOS survey after initiation of treatment for HNC will have inferior HRQOL and ability to perform ADLs compared to patients completing a MHOS survey before initiation of treatment for HNC (see part 1 below).
2. Differences in HRQOL and ability to perform ADLs in patients completing a MHOS survey before versus after initiation of treatment will be affected by tumor-, treatment and patient-specific variables, including specific tumor subsite (oral cavity, nasopharynx, oropharynx & tonsil, larynx & hypopharynx, and salivary gland) and the interval of time between treatment initiation and survey completion (see part 1 below).
3. Patients completing a MHOS survey after initiation of treatment for thyroid cancer will have inferior HRQOL and ability to perform ADLs compared to patients completing a MHOS survey before initiation of treatment for thyroid cancer (see part 2 below).
4. Differences in HRQOL and ability to perform ADLs before versus after initiation of treatment in patients with laryngeal or hypopharyngeal cancer will be affected by initial treatment modality (see part 3 below).
5. Differences in HRQOL and ability to perform ADLs before versus after initiation of treatment in patients with oropharyngeal or tonsil cancer will be affected by initial treatment modality (see part 4 below).

1. Develop an ontology-guided ML method that allows for analyzing heterogeneous healthcare datasets.
2. Test the ontology-guided ML method to analyze the SEER-MHOS dataset.
3. Investigate the associations between the quality of life (QoL), survival, and other characteristics in patients with cancer.

1. Compare HRQoL and health utility changes among selected cancer patients across disease stages and type of cancer from before to after cancer diagnosis with matched non-cancer controls.
2. Assess HRQoL and health utility changes among selected cancer patients across initial course of treatment from before to after cancer diagnosis with matched non-cancer controls.
3. Determine HRQoL and health utility changes among selected cancer patients across race/ethnicity from before to after cancer diagnosis with matched non-cancer controls.

1. Examine changes in HRQoL from before to after cancer among Medicare Beneficiaries stratified by treatment type and burden of comorbid health conditions.
2. Compare the impact of cancer, treatment type, and comorbid conditions on Medicare beneficiaries' lives relative to matched controls without cancer.
3. Conduct in-depth analyses by cancer treatment type to examine clinical and sociodemographic factors that may be associated with HRQoL declines for Medicare beneficiaries with cancer.

1. To determine the effect of different treatment approaches for common urologic malignancies (prostate, kidney and bladder) on patient reported HRQoL.
2. To evaluate the change in patient reported HRQoL compared to healthy controls following diagnosis and treatment of bladder, kidney and prostate cancer.

1. To determine the effect of different treatment approaches for common urologic malignancies (prostate, kidney and non-muscle invasive bladder) on patient reported HRQoL.
2. To evaluate the change in patient reported HRQoL compared to healthy controls following diagnosis and treatment of non-muscle invasive bladder, kidney, and prostate cancer.
3. To assess the change in patient reported HRQoL of muscle invasive bladder cancer patients undergoing radical cystectomy with urinary diversion (ileal conduit) compared to colo-rectal cancer patients undergoing extirpative surgery with colostomy or ileostomy.

1. To analyze secular trends in post-diagnosis HRQOL among OWBC from 1998-2013.
2. To analyze secular trends in changes in HRQOL during the pre- to post-diagnosis period from 1998-2013 among OWBC and to compare them to secular trends in changes in HRQOL from baseline to follow-up survey among older women without cancer.
3. To analyze secular trends in differences in HRQOL between OWBC and matched, non-cancer controls from 1998-2013.

1. To describe the patient factors associated with prostate cancer treatment allocation (aggressive vs. conservative therapies) and to identify factors associated with patients dying of prostate cancer versus other causes of death.
2. To identify predictors (patient factors, quality of life indicators, and information routinely collected in clinical practice) of 5- and 10-year other cause mortality among men with prostate cancer.
3. To develop a risk calculator from risk prediction models using a subset of variables routinely available in clinical practice to predict 10-year risk of other cause mortality.
4. Using the prediction scores derived from Specific Aim 2, to estimate the cost of evidenced-based, appropriately aggressive, care delivery vs. actual care received as reported in SEER, among men with prostate cancer, stratified across levels of other cause mortality risk.

1. Retrospective data analyses to determine the associations of race, socioeconomic status (SES), comorbidities, etc. with multiple myeloma (MM) survival, MM treatment, healthcare utilization etc.

1. The LCPM can accurately represent the natural history, treatment, and outcomes of lung cancer patients with chronic lung, heart, and renal disease as well as by overall burden of comorbidities.
2. Stage I NSCLC patients with severe comorbidities will benefit from less aggressive forms of therapy.
3. Stage II and IIIA NSCLC patients with moderate to severe chronic lung, heart, or renal disease or with high burden of comorbidities will not benefit from adjuvant chemotherapy.
4. Resection will lead to better outcomes than chemoradiotherapy in limited stage SCLC patients with severe renal disease or high burden of comorbidities.

1. Describe differences in prescription drug use patterns and differences in recent history of falls among older adults with cancer and between older adults with cancer and those without a history of cancer.
2. Characterize the associations between number of concurrent prescription medications and specific classes of prescription medication and recent history of falls among older adults with prevalent cancers (prostate, breast, colorectal, lung, ovary, uterus, cervix, bladder, melanoma, and head/neck), controlling for comorbidities, socioeconomic status, age at time of survey, race/ethnicity, limitations in ADL or IADL, cancer stage, time since diagnosis, and health behavior-related variables.
3. Determine which specific medication classes predict a history of falls in in older adults with prevalent cancers (prostate, breast, colorectal, lung, ovary, uterus, cervix, bladder, melanoma, and head/neck), controlling for comorbidities, socioeconomic status, age at time of survey, race/ethnicity, limitations in ADL or IADL, and health behavior-related variables.
4. Assess the extent to which balance problems, urinary incontinence, lower extremity weakness, and/or sensory impairment mediate the relationship between number and/or class of concurrent prescription medications and falls in older adult cancer survivors.

1. To describe the HRQOL change over time in elderly women. We will evaluate the secular trend of HRQOL from the time of diagnosis and will compare these trends between women with breast cancer (cases) and without breast cancer (controls).
2. To evaluate the effects of RT and comorbidities on HRQOL trajectories over time among elderly breast cancer survivors. HRQOL score will be compared among 4 groups of elderly women in the generalized linear mixed models after controlling for baseline HRQOL, age, race, BMI, socioeconomic status, cancer stage, and cancer histology. We hypothesize that women who have ≥ 2 comorbid conditions and have received RT would have the greatest HRQOL impairment (deficit) in physical functioning and mental health compared to women who had <2 comorbid conditions and did not receive RT.
3. To evaluate the effects of RT and comorbidities on the outcomes of OS/BSS among elderly breast cancer survivors. We hypothesize that women who have < 2 comorbid conditions and have received radiation treatment will have best OS/BSS compared to women who had ≥ 2 comorbid conditions and did not receive.

1. Calculate the health utility of patients before and after cancer treatment, by converting the SF-36 data to a single health utility.
2. Compare the impact of different cancer treatments on changes to health utility.
3. To incorporate the impact of cancer treatment and its changes to a patient's measured health utility on modifying the measurement of comparative effectiveness that has been measured without incorporating health utility.