Fax: (919) 843-6362
Impact of diagnosis and treatment of clinically localized prostate cancer on health-related quality of life for older Americans
A population-based study
Article first published online: 27 APR 2012
Copyright © 2012 American Cancer Society
Volume 118, Issue 22, pages 5679–5687, 15 November 2012
How to Cite
Reeve, B. B., Stover, A. M., Jensen, R. E., Chen, R. C., Taylor, K. L., Clauser, S. B., Collins, S. P. and Potosky, A. L. (2012), Impact of diagnosis and treatment of clinically localized prostate cancer on health-related quality of life for older Americans. Cancer, 118: 5679–5687. doi: 10.1002/cncr.27578
- Issue published online: 30 OCT 2012
- Article first published online: 27 APR 2012
- Manuscript Accepted: 7 MAR 2012
- Manuscript Revised: 2 MAR 2012
- Manuscript Received: 28 DEC 2011
- prostate cancer;
- health-related quality of life;
- older Americans
Few studies have measured longitudinal changes in health-related quality of life (HRQOL) among patients with prostate cancer starting before their cancer diagnosis or have provided simultaneous comparisons with a matched noncancer cohort. In the current study, the authors addressed these gaps by providing unique estimates of the effects of a cancer diagnosis on HRQOL accounting for the confounding effects of ageing and comorbidity.
Data from the Surveillance, Epidemiology, and End Results registry were linked with Medicare Health Outcomes Survey (MHOS) data. Eligible patients (n = 445) were Medicare beneficiaries aged ≥65 years from 1998 to 2003 whose first prostate cancer diagnosis occurred between their baseline and follow-up MHOS. By using propensity score matching, 2225 participants without cancer were identified from the MHOS data. Analysis of covariance models were used to estimate changes in HRQOL as assessed with the Medical Outcomes Study Short Form-36 survey and the activities of daily living scale.
Before diagnosis, patients with prostate cancer reported HRQOL similar to that of men without cancer. After diagnosis, men with prostate cancer experienced significant decrements in physical, mental, and social aspects of their lives relative to controls, especially within the first 6 months after diagnosis. For men who were surveyed beyond 1 year after diagnosis, HRQOL was similar to that for controls. However, an increased risk for major depressive disorder was observed among men who received either conservative management or external beam radiation.
The current findings illustrated the time-sensitive nature of decline in HRQOL after a cancer diagnosis and enhanced understanding of the impact of prostate cancer diagnosis and treatment on physical, mental, and social well being among older men. Cancer 2012. © 2012 American Cancer Society.
In 2011, an estimated 240,890 men were diagnosed with prostate cancer, and 62% of cases were diagnosed in men aged >65 years.1 Treatments for localized prostate cancer include radical prostatectomy, external beam radiation therapy (EBRT), brachytherapy, and conservative management (androgen-deprivation therapy [ADT] alone or no treatment).2 Although treated men often experience excellent long-term survival, the treatments can have an impact on men's health-related quality of life (HRQOL).2-4
Many studies have documented the effects of prostate cancer and its treatment on HRQOL; however, those studies collected prospective HRQOL data starting after a prostate cancer diagnosis.5-9 To our knowledge, only our published study has measured the impact with a baseline HRQOL assessment before prostate cancer diagnosis to capture the true prospective impact of the disease on patients' HRQOL.10 In addition, longitudinal comparison data for men without cancer have rarely been collected,11 limiting the ability of investigators to assess the marginal effects of prostate cancer on the elderly, for whom it may be difficult to attribute longitudinal declines in HRQOL to prostate cancer while accounting for ageing and other chronic comorbid conditions.
The 2007 linkage of data from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program with the Centers for Medicare and Medicaid Services' (CMS) Medicare Health Outcomes Survey (MHOS) provides a unique opportunity to measure the burden of prostate cancer and treatment on older men's lives.12 The population-based SEER cancer registries provide detailed clinical data on incident cancers, including disease stage and initial treatment. The MHOS provides data on patients' experiences and perspectives as they relate to physical, mental, and social well being.
In this study, we estimate the prospective change in HRQOL from before to after prostate cancer diagnosis among men aged ≥65 years and compare changes in their HRQOL relative to the changes in a similar cohort of men without cancer. This report differs from our earlier study using the SEER-MHOS data set, in which we compared HRQOL among 9 different cancer types.10 Here, we focus on prostate cancer-specific clinical characteristics, such as treatment type, stage of disease, and time since diagnosis, that are likely to be important influences on changes in HRQOL.
MATERIALS AND METHODS
Under the CMS and National Cancer Institute collaboration, survey data from the CMS' MHOS13 were linked in 2007 with data from National Cancer Institute's SEER cancer registries.14 During the study period from 1998 to 2003, the MHOS was administered annually to a random sample of 1000 Medicare beneficiaries from each Medicare Advantage Organization under contract with the CMS. Each participant completed a baseline survey and a 2-year follow-up survey if he remained in the same managed care plan. An overview of the SEER-MHOS data linkage is provided elsewhere.10, 12
Four MHOS cohorts were included in the study with baseline surveys for the years from 1998 to 2001 and with 2-year follow-up surveys. Response rates ranged from 64% to 72% for the baseline MHOS and from 76% to 85% for the follow-up surveys, resulting in a sample size of 11,683 MHOS respondents linked to SEER.12
Our study cohort included patients with prostate cancer whose first SEER-confirmed diagnosis occurred after their baseline MHOS and before the follow-up MHOS. We identified 467 eligible men with prostate cancer in this manner. We excluded 22 men who were diagnosed with regional or metastatic prostate cancer, resulting in a final sample of 445 men. The average time from baseline MHOS to prostate cancer diagnosis (±standard deviation) was 11.5 ± 7.1 months, and the average time from prostate cancer diagnosis to follow-up survey was 12.6 ± 7.1 months.
We selected noncancer controls (n = 16,397 men) who responded to both a baseline and follow-up MHOS and who resided in the same SEER region and participated in the same managed care plans as the cancer patients. By using propensity score matching procedures,15-17 we matched 5 controls to each cancer case to balance the proportion of patient and survey characteristics as well as pre-existing comorbid conditions between the groups.10, 18, 19 This resulted in 2225 controls matched to 445 men with prostate cancer.
The SEER-MHOS data are considered by Health Insurance Portability and Accountability Act of 1996 requirements as a limited data set, which required the investigators to sign a data use agreement before receiving the data. This exception allowed for the release of deidentified SEER-MHOS data without obtaining authorization from individual patients (see Federal Register, August 14, 2002, page 53,235).
The MHOS provides self-reported data on sociodemographics (education, age, race and ethnicity, smoking status, marital status), survey characteristics (self-report or proxy, mail or telephone-interview), HRQOL, clinical symptoms, and chronic medical conditions.20 Although the study would have been more homogeneous if we had excluded proxy reporting and men who switched assessment modes from baseline to follow-up, these data were retained in the analyses in an attempt to capture the extent of HRQOL changes, including HRQOL that may have been poor enough to require a proxy response. We controlled for these variables in the regression models.
The MHOS includes the Medical Outcomes Study Short Form-36 questionnaire (SF-36) (version 1),21 a HRQOL measure widely used in medicine.10, 22 The SF-36 includes 8 subscales: physical functioning, role-physical, bodily pain, general health, vitality, social functioning, role-emotional, and mental health. The subscales produce a physical component summary score and a mental component summary score. SF-36 scores are normalized to the general US population (mean score ± standard deviation, 50 ± 10). Higher scores indicate better HRQOL. A change in SF-36 scores of at least a half standard deviation (ie, 5 points) is considered clinically meaningful.23
The MHOS also includes a measure of activities of daily living, which assesses difficulties with getting in or out of chairs, dressing, walking, using the toilet, eating, and bathing. Each activity of daily living is reviewed individually and summed together, and higher scores reflect more limitations.
The MHOS includes a question about urinary incontinence, which is typically impacted by treatments for prostate cancer. However, the MHOS did not include any items on 2 other key domains for this disease: sexual function and bowel function.
The MHOS includes 3 items from the Diagnostic Interview Schedule.24 Together with a question from the SF-36 mental health subscale to capture more recent depressive mood, an indicator was developed for the risk of major depressive disorder (MDD) based on the algorithm from a published study.25 This MDD indicator is more conservative than the MHOS website recommendation that a positive screen for MDD is an affirmative response to at least 1 of the 3 Diagnostic Interview Schedule questions.20
The survey included questions about the presence of multiple chronic medical conditions. We grouped these according to whether they were “pre-existing” diagnosed conditions (ie, before baseline survey) versus “newly diagnosed” conditions (ie, between baseline and follow-up surveys) (for a list of conditions, see Table 1). We summed the comorbidities and applied weights based on the association of each condition with physical and mental aspects of HRQOL. Development of the physical and mental weighted comorbidity indices were described in a previous study.10
|By Treatment Type|
|Characteristic (%)||Matched Noncancer Controls, N = 2225||All Prostate Cancer Patients, N = 445||Radical Prostatectomy, N = 72||External Beam Radiation Therapy, N = 169||Brachytherapy, N = 41||Conservative Management, N = 163|
|High school graduate or GED||25.26||25.84||23.61||27.22||17.07||27.61|
|Some college or 2-y degree||26.29||24.04||30.56||26.04||14.63||21.47|
|Asian or Pacific Islander||4.72||4.49||1.39||7.10||0||4.29|
|Marital status at baseline|
|Marital status changea||3.60||2.25||2.78||1.78||2.44||2.45|
|Assessment mode from baseline to follow-up|
|Hypertension/high blood pressure||49.48||47.64||50||47.34||58.54||44.17|
|Angina pectoris/coronary artery disease||16.18||16.63||18.06||15.98||12.20||17.79|
|Congestive heart failure||4.76||5.17||2.78||5.92||2.44||6.13|
|Myocardial infarction/heart attack||11.46||11.69||8.33||12.43||12.20||12.27|
|Other heart conditions||16.27||16.40||9.72||18.93||19.51||15.95|
|Emphysema, asthma, or COPD||13.21||13.93||6.94||13.61||19.51||15.95|
|Crohn disease, ulcerative colitis, or IBD||2.65||2.47||0||1.78||2.44||4.29|
|Arthritis of the hip or knee||24.67||25.84||20.83||27.22||31.71||25.15|
|Arthritis of the hand or wrist||22.07||22.70||23.61||23.08||29.27||20.25|
|Newly diagnosed conditions|
|Hypertension/high blood pressure||8.09||9.21||5.56||7.69||7.32||12.88|
|Angina pectoris/coronary artery disease||5.80||6.52||1.39||8.28||4.88||7.36|
|Congestive heart failure||4.22||3.60||0||4.14||2.44||4.91|
|Myocardial infarction/heart attack||4.22||4.04||1.39||4.14||2.44||5.52|
|Other heart conditions||7.55||8.09||6.94||8.28||7.32||8.59|
|Emphysema, asthma, or COPD||3.46||4.27||4.17||3.55||4.88||4.91|
|Crohn disease, ulcerative colitis, or IBD||1.39||3.15||2.78||2.96||4.88||3.07|
|Arthritis of the hip or knee||10.61||12.13||13.89||11.83||17.07||10.43|
|Arthritis of the hand or wrist||9.30||10.34||11.11||11.24||12.20||8.59|
|Time since diagnosis, mo|
|T1 or T2||12.13||16.67||6.51||12.20||15.95|
|T2 prostatic apex||15.51||9.72||17.75||14.63||15.95|
|No. of other cancer(s) before prostate cancer diagnosis|
SEER provided information regarding disease stage at diagnosis, date of diagnosis, and initial treatment received.14 Among the patients with localized disease, the clinical extent of disease was classified according to the following tumor classification used in clinical practice: T1 (clinically unapparent or incidentally detected disease), T2 (clinically or radiographically apparent disease), T2 prostatic apex (extension to prostatic apex), T1 or T2 (ie, insufficient information to classify), or unstaged based on the lack of any clinical extension information. Time from SEER-confirmed cancer diagnosis to the follow-up MHOS survey was divided into categories of 0 to 6 months, 7 to 12 months, 13 to 18 months, and ≥19 months to allow for comparison of HRQOL in patients who were surveyed at different points in their postdiagnosis trajectory. We categorized treatments into radical prostatectomy (also including those who received adjuvant radiation therapy), EBRT, brachytherapy, or conservative management (defined as the absence of documented curative-intent surgery or radiation treatments). Because neither SEER nor Medicare includes data on the use of outpatient-based hormone therapy or ADT, it is likely that a sizable proportion of men in our conservative management group did receive such therapy. SEER contains nearly complete data on surgery, but there has been recent evidence of a slight under-ascertainment of radiotherapy.26, 27
A mixed-effects analysis of covariance model was used to estimate changes in HRQOL scores adjusting for baseline scores. The model adjusted for clustering of respondents enrolled in the same Medicare Advantage Organization and for the 5 controls matched to each cancer patient.28 The model included the patient characteristics, comorbidities, and survey characteristics listed in Table 1 as covariates. Models were tested to examine for differences in HRQOL by treatment type and time since diagnosis, both as main effects and as interactions. Models that included the variable “time since diagnosis” alone combined prostate cancer patients across all treatments; thus, significant findings could be weighted by treatment groups that were over-represented in the sample (EBRT and conservative management) (see Table 1). The SAS statistical software package (version 9.2; SAS Institute Inc., Cary, NC) was used for all analyses.
Characteristics of the men with prostate cancer (n = 445; categorized by initial treatment) and their matched controls (n = 2225) are described in Table 1. No statistically significant differences were observed on demographic characteristics, survey characteristics, or pre-existing health conditions after propensity score matching. For newly diagnosed health conditions (not included in the propensity score model), men with prostate cancer had a higher percentage of gastrointestinal disorders than men without cancer (3.19% vs 1.39%, respectively; P < .01).
Changes in Physical Health
At baseline, the mean SF-36 physical component summary score (±standard deviation) was 45.6 ± 11.2 for the matched noncancer controls and 45.1 ± 10.9 for the men with prostate cancer before diagnosis. These scores were similar by treatment group and were consistent with the age-adjusted US general population 1998 SF-36 norms.29, 30
In general, there were few statistically significant differences between the treatment groups and the control group in changes over time for the overall physical component summary score or for the SF-36 subdomains of physical function, role-physical, or bodily pain. The exception was that, for both the EBRT group and the conservative management group, a statistically significant decline was observed in general health (adjusted mean change, −3.9 and −3.6, respectively; P < .01) relative to the matched controls (mean change, −1.7).
Next, we examined the changes in patients' HRQOL relative to controls according to the time from diagnosis to the follow-up survey (Fig. 1). Men who were surveyed within 6 months of diagnosis had statistically worse overall physical component summary scores (mean change, −4.9 vs −2.1 in controls; P < .01), role-physical scores (−7.6 vs −3.0; P < .01), and general health scores (−5.0 vs −1.7; P < .01). Patients with prostate cancer who were surveyed from 7 to 12 months after diagnosis also had significantly decreased role-physical scores (−6.3; P = .046) and general health scores (−3.5; P = .03) relative to controls. For physical component summary, role-physical, and general health scores, we did not observe an interaction between treatment group and time since diagnosis. When both variables were included in the model, time since diagnosis, but not treatment, retained a significant association with decreased physical health.
The prostatectomy group experienced a statistically significant improvement relative to controls (P = .03) in overall activities of daily living from baseline (change, 0.69 vs 0.61 on a scale from 0 to 6) to follow-up (mean, 0.38 vs 0.88). The prostatectomy group reported improvements after diagnosis in bathing, dressing, eating, and using the toilet; whereas the controls reported greater limitations for all activities of daily living.
Urinary incontinence was a significant problem reported by all patients with prostate cancer regardless of treatment (Table 2). Each treatment group experienced more urinary incontinence than controls at the follow-up survey, and the prostatectomy group reported the largest increase in incontinence.
|Treatment Type||No. of Patients||Baseline/Before Cancer Diagnosis % Yes||Two Years Later/After Cancer Diagnosis % Yes||Logistic Regression OR (95% CI)a|
|External beam radiation therapy||169||20.12||39.64||2.89 (2.00-4.17)|
|Conservative management||163||29.45||38.65||1.92 (1.31-2.83)|
Changes in Mental Health
At baseline, the mean SF-36 mental component summary score (±standard deviation) was 53.5 ± 10.3 for the matched noncancer controls and 53.4 ± 10.1 for the patients with prostate cancer. These scores were similar by treatment group and were consistent with age-adjusted general US population 1998 SF-36 norms.29, 30
Compared with the matched noncancer controls, patients who received EBRT reported significantly greater declines in mental component summary scores (adjusted mean changes, −1.1 and −3.8, respectively; adjusted P < .01), role emotional scores (−2.3 and −5.9, respectively; P < .01), vitality scores (−1.4 and −3.2, respectively; P = .01), and social function scores (−1.4 and −4.2, respectively; P < .01).
Men who were surveyed 7 to 12 months after diagnosis had greater declines in mental component summary scores (Fig. 1) (mean change, −3.7 vs −1.1 for controls; P = .02) and vitality scores (−3.7 vs −1.4 for controls; P < .01) than controls. Social function scores also were affected by prostate cancer. Among men who were surveyed within 6 months of diagnosis, the adjusted mean change in social function was −3.6 compared with −1.4 for controls (P = .03). Among men who were surveyed 7 to 12 months from diagnosis, the mean change in social function was −4.8 compared with −1.4 for controls (P < .01).
Men in the conservative management group were at the greatest risk for MDD, with 13% and 23% classified as at risk for MDD at the baseline and follow-up surveys, respectively. This differed significantly (odds ratio [OR], 1.9; 95% confidence interval [CI], 1.2-3.0) from the risk among matched controls (11% and 14% at the baseline and follow-up surveys, respectively). Men with prostate cancer (all treatments combined) who were surveyed >19 months after diagnosis had a greater change in the risk of MDD, with 13% and 24% at risk for MDD at the baseline and follow-up surveys, respectively (OR, 2.2; 95% CI, 1.3-3.8). Men who received EBRT and were surveyed >19 months after diagnosis were at an increased of MDD compared with controls (OR, 3.1; 95% CI, 1.3-7.1). In addition, relative to controls, men in the conservative management group who were surveyed from 13 to 18 months after diagnosis had the greatest increased risk of MDD (OR, 3.4; 95% CI, 1.5-7.4), and this increased MDD risk for the conservative management group relative to controls continued for men who were surveyed after 19 months (OR, 2.5; 95% CI, 1.0-6.0).
In this prospective, population-based study, we evaluated the burden of prostate cancer and treatment on the lives of American men aged ≥65 years. Although patients with prostate cancer who were surveyed before diagnosis were similar in health status to matched controls, notable HRQOL decrements after diagnosis occurred in men with cancer, especially for those who received EBRT or conservative management. HRQOL decrements were greatest for those patients who were surveyed within 6 months of diagnosis, when the men probably were receiving active treatment. For men who were surveyed >1 year after diagnosis, HRQOL was similar to that of controls across most HRQOL domains. Our study is among the very few to examine HRQOL changes before and after a prostate cancer diagnosis and to use a matched control group of men without cancer. These findings illustrate the time-sensitive nature of variations in physical, mental, and social well being for men after a cancer diagnosis.
Men who were diagnosed with prostate cancer within 6 months of the survey reported significantly larger decrements in overall physical health relative to controls and experienced more role limitations because of their poorer general health than matched controls. Each of these changes in physical health were at least a half standard deviation or greater on the SF-36, representing a clinically meaningful change.23 Patients with prostate cancer who were surveyed 12 months after diagnosis had decrements in physical health that were comparable to those in similarly aged men without cancer. This trend has been observed in other prospective studies, in which men often returned to baseline physical health levels (defined in those studies as after cancer diagnosis but before treatment).7, 9
When examining men by treatment type, we observed that those who received EBRT or conservative management reported worse general health than controls. Because we did not have information on ADT, it is likely that a greater percentage of these patients received ADT for months to years after diagnosis compared with men in the surgery group.31, 32 It is known that ADT use causes fatigue, muscle weakness, and weight gain, which may cause a decline in perceived general health.33 In men who are receiving radiation therapy, the late morbidity from treatment with older EBRT techniques may partly explain this decline.34, 35 Recent technical improvements, such as intensity-modulated radiation therapy and image-guided radiation therapy, may reduce such effects36 but were not widely disseminated until after the collection of these data.37, 38
The rate of incontinence was significantly greater for men with prostate cancer than for those without cancer; especially among men who underwent radical prostatectomy. However, even men in the conservative management group experienced a statistically significant increase in urinary incontinence at 9.2%. The rates we observed were similar to those reported in the literature.5-7, 9
An unexpected finding in this study was that men who underwent prostatectomy reported improved activities of daily living relative to matched controls. Further investigation is needed to replicate this finding and to explore the potential reasons for these improvements.
Differences in mental health changes also were detected between the prostate cancer patients and controls. Men who received EBRT reported statistically significant declines relative to controls in overall mental health, limitations in activities because of emotional problems, declines in social functioning, and increased fatigue. In addition, an unexpected result was that men who were surveyed >19 months after their cancer diagnosis were approximately 2.2 times more likely to have an elevated risk of MDD than men without cancer, especially in the conservative management and EBRT groups. Other studies reported similar evidence of increased depression in older prostate cancer survivors.39-42 Although receipt of ADT may be a contributing factor in these treatment groups,43 the impact of prostate cancer and age-associated physical limitations also may affect depressive mood.41
Limited social functioning was reported for men with prostate cancer who were surveyed within 12 months after diagnosis; however, men who were surveyed >12 months after diagnosis reported social functioning similar to that reported by noncancer controls, suggesting some adaptation or return to baseline across all treatment groups.
The current study has some limitations. The data set was limited to beneficiaries in Medicare managed care plans. Past studies suggest that enrollees in Medicare Advantage Organizations have better health status than fee-for-service beneficiaries,44, 45 whereas other studies have indicated that health status is similar for both groups.46 In addition, the current data did not capture men who disenrolled from a managed care plan or died before responding to the follow-up survey.47 We lack sufficient data to estimate the effect of response bias; however, changes in HRQOL scores probably will be worse when considering men who were close to end of life.
Another possible limitation is that the study included men who required a proxy to report their HRQOL on the MHOS. The concern here is that responses from proxies may be influenced by their own feelings about and experiences of caring for the cancer patient.48 However, proxies provide critical information for men who are too ill to self-report.48 Removing proxy data may bias the HRQOL results to reflect better HRQOL outcomes from cancer than what are actually experienced. Proxy reporting was controlled in the regression models and was associated with poorer HRQOL.
In addition, our cohort was diagnosed and treated between 1998 and 2003; thus, we were unable to include some newer treatments (eg, robotic prostatectomy and intensity-modulated radiation therapy). Furthermore, we were unable to include hormone therapy, which reportedly is associated with various physical and mental effects and, thus, potentially confounds some of our HRQOL findings.8, 49, 50
A final limitation is that the MHOS did not include cancer-specific, functional measures. Many studies have demonstrated that treatments for prostate cancer can result in significant problems. Radical prostatectomy also has been associated with increased sexual dysfunction,2, 7-9, 51-53 and both EBRT and brachytherapy have been associated with urinary irritative-obstructive symptoms and bowel dysfunction.2, 7-9, 51-54
Despite these limitations, the current SEER-MHOS prostate cancer study provides important information for researchers, clinicians, patients, and health plans. Understanding the HRQOL effects from prostate cancer and its treatments may allow health care providers to better anticipate and, thus, more effectively manage the physical and psychosocial sequelae of prostate cancer.10 This study demonstrates that the diagnosis of prostate cancer has a significant impact on health status, especially for patients who receive EBRT or conservative management. These results provide useful benchmarks of the net impact of prostate cancer on HRQOL in the presence of ageing, comorbid health conditions, and sociodemographic factors. Finally, our estimates also may be informative for incorporating health status into cost-effectiveness analyses of prevention and early detection interventions. Future research will attempt to replicate these findings when new SEER-MHOS cohorts become available. The results from such studies will enable health care payers and clinicians to actively monitor how improvements in the treatment of prostate cancer may enhance men's lives.
Dr. Reeve's work was supported under a National Cancer Institute contract to the University of North Carolina at Chapel Hill (HHSN261201000642P).
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 1American Cancer Society. Cancer Facts & Figures 2011. Atlanta, GA: American Cancer Society; 2011.
- 14SEER Cancer Statistics Review, 1975-2004 (based on the November 2006 SEER data submission). Bethesda, MD: National Cancer Institute; 2007. Available at: http://seer.cancer.gov/csr/1975_2004/. Accessed October 13, 2011., , , et al.
- 16Constructing a control-group using multivariate matched sampling methods that incorporate the propensity score. Am Stat. 1985; 39: 33-38., .
- 19GMATCH SAS Macro: computerized matching of cases to controls using the greedy matching algorithm with a fixed number of controls per case. Controls may be matched to cases using 1 or more factors (X's). Rochester, MN: Mayo Clinic; 2003. Available at: http://mayoresearch.mayo.edu/mayo/research/biostat/sasmacros.cfm. Accessed October 14, 2011., .
- 20Centers for Medicare and Medicaid Services. Medicare Health Outcomes Survey: report on the health status of managed care smokers and nonsmokers cohort III baseline and follow up. Final report prepared by Health Services Advisory Group [monograph online]. Available at: http://www.hosonline.org/surveys/hos/download/HOS% 20Smoking%20Report%202005.pdf. Accessed October 14, 2011.
- 22SF-36 Health Survey and “SF” Bibliography. 3rd ed. Lincoln, RI: QualityMetric Inc.; 2002., , .
- 27Studying radiation therapy using SEER-Medicare-linked data. Med Care. 2002; 40: IV-49-IV-54., , , , , .
- 29SF-36 Physical and Mental Health Summary Scales: A User's Manual. Boston, MA: The Health Institute; 1994., , .
- 30SF-36 Physical and Mental Health Summary Scales: A Manual for Users of Version 1. 2nd ed. Lincoln, RI: QualityMetric Inc.; 2001., .
- 33Impact of androgen deprivation on physical well-being in patients with prostate cancer: analysis from the CaPSURE (Cancer of the Prostate Strategic Urologic Research Endeavor) registry. Cancer. 2011; 117: 4406-4413., , , , , .