0
We're unable to sign you in at this time. Please try again in a few minutes.
Retry
We were able to sign you in, but your subscription(s) could not be found. Please try again in a few minutes.
Retry
There may be a problem with your account. Please contact the AMA Service Center to resolve this issue.
Contact the AMA Service Center:
Telephone: 1 (800) 262-2350 or 1 (312) 670-7827  *   Email: subscriptions@jamanetwork.com
Error Message ......
Original Contribution |

Survival Associated With Treatment vs Observation of Localized Prostate Cancer in Elderly Men FREE

Yu-Ning Wong, MD; Nandita Mitra, PhD; Gary Hudes, MD; Russell Localio, PhD; J. Sanford Schwartz, MD; Fei Wan, MS; Chantal Montagnet, MA, MPhil; Katrina Armstrong, MD, MSCE
[+] Author Affiliations

Author Affiliations: Divisions of Population Science (Dr Wong) and Medical Science (Drs Wong and Hudes), Fox Chase Cancer Center; Department of Biostatistics and Epidemiology (Drs Mitra and Localio), Division of General Internal Medicine (Drs Schwartz and Armstrong and Mr Wan and Ms Montagnet), and Abramson Cancer Center and Leonard Davis Institute of Health Economics (Drs Schwartz and Armstrong), University of Pennsylvania, Philadelphia.

More Author Information
JAMA. 2006;296(22):2683-2693. doi:10.1001/jama.296.22.2683.
Text Size: A A A
Published online

Context Prostate-specific antigen screening has led to an increase in the diagnosis and treatment of localized prostate cancer. However, the role of active treatment of low- and intermediate-risk disease in elderly men is controversial.

Objective To estimate the association between treatment (with radiation therapy or radical prostatectomy) compared with observation and overall survival in men with low- and intermediate-risk prostate cancer.

Design and Setting Observational US cohort from Surveillance, Epidemiology, and End Results Medicare data. 

Patients At total of 44 630 men aged 65 to 80 years who were diagnosed between 1991 and 1999 with organ-confined, well- or moderately differentiated prostate cancer and who had survived more than a year past diagnosis. Patients were followed up until death or study end (December 31, 2002). Patients were classified as having received treatment (n=32 022) if they had claims for radical prostatectomy or radiation therapy during the first 6 months after diagnosis. They were classified as having received observation (n=12 608) if they did not have claims for radical prostatectomy, radiation, or hormonal therapy. Patients who received only hormonal therapy were excluded.

Main Outcome Measure Overall survival.

Results At the end of the 12-year study period, 4663 men (37%) in the observational group and 7639 men (23.8%) in the treatment group had died. The treatment group had longer 5- and 10-year survival than the observation group. After using propensity scores to adjust for potential confounders (tumor characteristics, demographics, and comorbidities), there was a statistically significant survival advantage associated with treatment (hazard ratio, 0.69; 95% confidence interval, 0.66-0.72). A benefit associated with treatment was seen in all subgroups examined, including older men (aged 75-80 years at diagnosis), black men, and men with low-risk disease.

Conclusions This study suggests a survival advantage is associated with active treatment for low- and intermediate-risk prostate cancer in elderly men aged 65 to 80 years. Because observational data cannot completely adjust for potential selection bias and confounding, these results must be validated in randomized controlled trials of alternative management strategies in elderly men with localized prostate cancer.

Figures in this Article

The widespread adoption of prostate-specific antigen (PSA) screening has led to an increasing proportion of men being diagnosed with early stage and low- or intermediate-grade prostate cancer. Cohort studies have demonstrated the indolent nature of low- and intermediate-grade disease,13 making management (observation, radiation therapy, and radical prostatectomy) of these men controversial. Population-based studies of treatment for localized prostate cancer estimate that only between 7% and 17%4 of men diagnosed with low-risk localized disease choose observation as their initial therapy.

Despite the significant burden of localized prostate cancer in the United States, there are limited randomized data to help guide treatment decisions. A recent randomized trial of radical prostatectomy vs observation in 695 Swedish men found that radical prostatectomy reduced overall mortality, disease-specific mortality, risk of metastasis, and local progression after a median of 8.2 years of follow-up.5,6 Although a subgroup analysis suggested that the effect of treatment was greater among men younger than 65 years than among men aged 65 to 75 years, the relatively small sample size and the lack of published information about the estimated effect in men older than 65 years have contributed to the uncertainty about the benefit of prostatectomy in this age group.5,6

The ongoing Veterans Affairs Prostate Cancer Intervention vs Observation (PIVOT) trial also compares radical prostatectomy and observation in 731 men and will complete follow-up in 2009. The ongoing UK Prostate Testing for Cancer and Treatment Study (Protect) will randomize men with prostate cancer to receive radiation, undergo radical prostatectomy, or remain in observation. Study accrual is estimated to be completed in 2008. The American College of Surgeons Oncology Group's Surgical Prostatectomy vs Interstitial Radiation Intervention Trial (SPIRIT) for men with clinical stage T1c or T2a tumors closed early due to poor accrual. Despite many case series suggesting long-term disease control with external beam radiation therapy or brachytherapy, there are no randomized studies comparing radiation therapy with observation.

When randomized controlled trial data are not available, observational studies can provide insight into important clinical questions. In this study, we evaluate the association of active treatment (radiation or prostatectomy) vs observation with overall survival in a large population-based sample of elderly men treated for low- or intermediate-risk localized prostate cancer.

Data Source

We used data from the linked Surveillance, Epidemiology, and End Results (SEER) Medicare database. SEER is a population-based cancer registry encompassing approximately 14% of the US population. SEER includes information on tumor histology, size, and grade.7 The National Cancer Institute has linked SEER data to Medicare claims for all individuals with Medicare coverage. Approximately 97% of individuals in SEER aged 65 years and older were successfully linked to their Medicare claims. Medicare files contain extensive medical and surgical claims data for patients covered by Medicare. The study was approved by the institutional review boards at the University of Pennsylvania and the Fox Chase Cancer Center.

Study Population

The study population included 111 640 men between 65 and 80 years of age who had an incident prostate cancer diagnosis between 1991 and 1999 in the SEER-Medicare database. Men were excluded if they were diagnosed at autopsy or death or had Medicare entitlement based on end-stage renal disease. Because Medicare does not contain complete claims information for individuals in managed care, men were excluded if they were enrolled in a managed care plan from 3 months before diagnosis to 6 months after diagnosis.

The goal of this analysis was to examine the associations of treatment benefit for men with low-risk tumors. Thus, we limited the study sample to men with well- or moderately differentiated T1 or T2 tumors.

Patients with T3 to T4 tumors and those with poorly differentiated or anaplastic tumors were excluded because these patients are considered at high risk for progressive disease.8 Men with metastatic disease were also excluded because they would not be candidates for primary radiotherapy or radical prostatectomy. Patients with the following characteristics were excluded from the final cohort: tumor grades of T3 or T4 (n = 9023), metastatic disease (n = 7333), unknown tumor size (n = 13 134), poorly differentiated, anaplastic, or unknown grade (n = 19 019), original or current reason for Medicare entitlement listed as disability or Medicare status code including disability (n = 3929), older than 80 years at diagnosis (n = 7047), treated with hormonal therapy alone (n = 4765), prior history of cancer (n = 666), missing socioeconomic status (n = 943), and died within a year of diagnosis (1151). The total population for analysis was 44 630.

Variable Definitions

Tumor Grade and Stage. SEER reports tumor grade as well differentiated (Gleason score, 2-4), moderately differentiated (Gleason score, 5-7), or poorly differentiated (Gleason score, 8-10), or unknown.

We used clinical extension information provided by SEER to determine tumor stage. Tumors were categorized as T2a and lower if they were described as being confined to less than one half of one lobe of the prostate. Tumors were categorized as T2b to T2c if they were described as involving greater than one half of one lobe of the prostate but were still organ-confined. We categorized 4556 patients who were described as having disease “involving one lobe of the prostate NOS [not otherwise specified]” in the T2a and lower group in our baseline analysis.

Treatment. Use of active treatment was determined by searching Medicare files for the appropriate International Classification of Diseases, Ninth Revision (ICD-9) and Healthcare Common Procedure Coding System codes during the 6 months after the date of diagnosis. These included codes for radical prostatectomy, external beam radiation, and radiation implants. (A list of all of the treatment codes, including 14 codes for radical prostatectomy, 62 codes for external beam radiation, 72 codes for brachytherapy, and 25 codes for hormonal therapy or orchiectomy is available on request from the authors). Because SEER provides the month of diagnosis only, we assumed that all patients were diagnosed on the 15th of the month and included an additional 15 days in the 6-month treatment time window. Medicare files included the inpatient claims (Part A), the carrier or physician file (Part B), and the outpatient claims file.

Patients who received hormonal therapy alone (luteinizing hormone− releasing hormone agonists or orchiectomy) were excluded from the analysis because androgen ablation monotherapy is not recommended by the National Comprehensive Cancer Network or the Hormone Therapy Study Group for nonmetastatic disease.8,9 The remaining patients were categorized as undergoing active treatment (radical prostatectomy, external beam radiation, or radiation implants) or observation.

Survival. Overall survival was defined as the interval from the date of diagnosis to the Medicare date of death. Patients alive at the end of the study period (December 30, 2002) were censored at that point and contributed the time interval from their date of diagnosis to the end of the study in the survival analysis. We excluded 1151 (645 observation, 506 treatment) patients who died within a year of diagnosis to eliminate men who were acutely ill from other causes. Although this approach may theoretically augment the survival advantage of treatment by eliminating perioperative deaths, the mortality of prostatectomy is low (ie, 30-day mortality rate is 0.66% for men aged 70-79 years).10

To measure the association between treatment and prostate cancer−specific mortality, we used cause of death provided by SEER to compare disease-specific survival among treatment groups. Patients who died from a cause other than prostate cancer (ICD-9 185.9 or ICD-10 C619) were censored at the date of death. Data on cause-specific mortality were available through the end of 2000.

Covariates.Comorbid Disease. Comorbidity has been associated with a lower probability of receiving active treatment for prostate cancer11 and a higher probability of prostatectomy-related complications.10 Comorbidities were identified by searching Medicare inpatient, outpatient, and Part B claims during the 90 days before a patient's diagnosis. Using all 3 sources captures more comorbidity data than the Charlson index, which relies on inpatient claims alone.12

Forty-four comorbidities were identified using methods described by Silber et al,13 including the following: AIDS, alcoholism, angina, aortic stenosis, arrhythmia, asthma, blood-loss anemia, collagen vascular diseases, congestive heart failure, chronic obstructive pulmonary disease, Cushing disease, depression, drug abuse, electrolyte and fluid abnormality, gastrojejunal ulcer, Graves disease, hereditary coagulopathy disorders, hypertension, hyperthyroidism, type 2 diabetes, iron-deficiency anemia, liver disease, liver dysfunction, myocardial infarction, obesity, other neurological disorders, paraplegia, peptic ulcer disease excluding gastric ulcer, peptic ulcer disease site not otherwise specified, peripheral vascular disorders, post–inflammatory pulmonary fibrosis, psychosis, pulmonary circulation disorders, pulmonic stenosis, renal dysfunction, renal failure, rheumatoid arthritis, seizures, smoking history, stroke or cerebrovascular accident, thrombocytopenia, unstable angina, valvular disorders, or weight loss.

We examined the association between comorbidity and odds of receiving treatment using each individual comorbidity as a binary variable (present or absent) and the total number of comorbidities. Both methods resulted in similar results. For clarity of presentation, we report the odds of receiving therapy based on the number of comorbidities (0, 1, 2, 3, or more) in the primary analysis. In addition, 15 “serious comorbidities” were associated with decreased probability of receiving treatment in the multivariate model and adjusted for them in the survival analyses.

Sociodemographic Characteristics. Age at diagnosis, race, marital status, and SEER site were determined from SEER data. Race was categorized as white, black, or other. Marital status was categorized as married, single, or other. Because SEER-Medicare does not collect individual patient level socioeconomic status information, we used the percentage of the 1990 census tract with a 4-year college education and the median income per census tract as proxies for an individual patient's socioeconomic status. Patients were classified as living in a rural area if they lived in a county or county-equivalent of fewer than 20 000 residents, as classified by the US Census. The remainder were classified as living in an urban area.

Statistical Analysis

Summary statistics were constructed using frequencies and proportions for categorical data elements and means and medians for continuous variables. We used propensity score methods to balance observed covariates between the treatment and observation groups. Propensity scores reflect the probability that a patient will receive therapy based on his observed covariates.14,15 We calculated propensity scores using multivariable logistic regression with active treatment as the outcome of interest. Age at diagnosis, SEER site, year of diagnosis, tumor size, tumor grade, marital status, residence in an urban setting, race, income, educational achievement, and comorbidities were the independent variables. We found a statistically significant interaction between tumor size and grade.

Propensity scores were then used to group patients into quintiles according to the probability of receiving active treatment based on each patient's baseline known characteristics. This procedure has been estimated to remove more than 90% of the bias resulting from each of the covariates.16 We used t tests, χ2 tests, and generalized linear models to determine whether the covariates were balanced within strata and found that number of comorbidities, tumor grade, and tumor size were still not balanced.

Cox proportional hazard models were used to compare survival in patients who received active treatment compared with those who were managed with observation. We examined the association of treatment and propensity score quintile adjusting for comorbidities, tumor grade, and tumor size, which were not balanced even after adjustment for propensity scores. In addition, we examined the association of treatment and survival in the entire cohort adjusting for propensity score as a continuous variable and comorbidities, tumor grade, and tumor size as categorical variables. We also measured the association between treatment vs observation and outcomes in the full cohort adjusting for grade, size, and comorbidities stratified by quintiles.

To examine the relationship between age and outcomes associated with treatment, we calculated the propensity score as described above excluding age. We then measured the interaction of treatment and age as a categorical variable (65-67, 68-70, 71-73, 74-77, and 78-80 years) in their association with survival, stratifying by propensity score quintile.

To measure the association of treatment and outcomes in specific populations, we created cohorts of patients who were diagnosed before and after 1994, black, elderly (age 75 to 80 years at diagnosis), and low risk (≤T2a, well-differentiated). We then calculated propensity scores using the methods described above and used Cox proportional hazard methods to measure the association between treatment and overall survival adjusting for propensity score as a continuous variable. Using the same methods, we examined the treatment within individual SEER sites to determine whether geographic location influenced the overall point estimates.

We also compared the association between radiation therapy and survival by creating a cohort of patients who were treated exclusively with radiation. We compared this outcome to the observation cohort by calculating propensity scores and using the Cox proportional hazard methods adjusting for propensity scores. We also compared patients who received radical prostatectomy with those who were observed using the same methods. We also examined outcomes associated with treatment in the “healthiest” patients (no comorbidities for 6 months prior to diagnosis) to determine whether the benefit associated with treatment was because sicker patients were in the observation group and did not receive treatment.

Finally, we performed a sensitivity analysis to measure the potential influence an unmeasured covariate might have on the hazard ratio (HR) estimates of the association between treatment and survival.17,18 Propensity scores calculations and survival analyses were performed using STATA 8.0 (STATA Inc, College Station, Tex). Sensitivity analyses were conducted using Microsoft Excel. Statistical significance was set at .05, and all tests were 2-tailed.

Patient Baseline Characteristics and Predictors of Treatment

The final group for analysis consisted of 44 630 patients, of whom 12 608 (28.25%) were in the observation group and 32 022 (71.75%) were in the treatment group. The baseline characteristics of the 2 groups and the results of multivariate analyses of predictors associated with receiving treatment with radical prostatectomy or radiation therapy are shown in Table 1. Patients who were younger, white, married; had higher income or educational status; or lived in an urban area were more likely to receive active treatment. In addition, patients with larger tumors (T2b-T2c) or moderately differentiated tumors were also more likely to receive active treatment. Patients with 3 or more comorbidities were less likely to receive treatment. The odds of receiving active treatment also varied by SEER site and year of diagnosis.

Table Graphic Jump LocationTable 1. Patient Baseline Characteristics and Predictors of Receiving Treatment

Dividing the cohort into quintiles based on propensity scores balanced all of the predictors of active treatment within each quintile, with the exception of tumor grade and size and number of comorbidities. The distributions of race, marital status, grade, size, number of comorbidities, place of residence, year of diagnosis, and SEER site by quintile propensity score are shown in Table 2. Survival analyses were conducted within quintile adjusting for comorbidities and tumor size and grade.

Table Graphic Jump LocationTable 2. Distribution of Covariates by Propensity Score Quintiles*
Survival Analysis

Overall Survival and Prostate Cancer−Specific Mortality. At the end of the 144-month (12-year) long study period, 12 302 patients (27.6%) had died. Thirty-seven percent of the 4663 patients in the observation group and 23.8% of the 7639 patients in the treatment group died (P value <.001). The unadjusted Kaplan-Meier survival curves for the entire cohort are shown in the Figure. The 5- and 10-year survival probabilities for the entire cohort and within quintiles are also shown in Table 3. Using Cox proportional hazard models, active treatment was associated with a significant improvement in survival in the overall cohort and within quintiles of propensity score, adjusting for tumor grade and size (Table 4). The estimated association between treatment and survival was similar (0.71; 95% confidence interval [CI], 0.69-0.74) if the model also adjusted for the 15 individual comorbidities that were associated with survival in multivariate analysis.

Figure. Kaplan Meier Survival Curves for Full Cohort
Graphic Jump Location

Patients who survived less than 12 months were excluded from the analysis.

Table Graphic Jump LocationTable 3. Patient Distribution and 5- and 10-Year Overall Survival by Treatment and Propensity Score Strata
Table Graphic Jump LocationTable 4. Association Between Active Treatment and Overall Mortality

Nine hundred twenty-six deaths were attributed to prostate cancer in SEER, including 314 in the observation group and 612 in the treatment group. In our analysis, patients who were treated also had a reduced risk of death from prostate cancer when adjusting for propensity score alone (HR, 0.67; 95% CI, 0.58-0.77). In addition, among the patients whose deaths were attributed to prostate cancer, the median overall survival in the treatment group was 55 months compared with 47 months in the observation group (log-rank P<.001).

Subgroup Analyses and Sensitivity Analyses. The association between treatment and reduced mortality risk was evident within all prespecified subgroups examined (Table 5). In addition, treatment was associated with reduced risk of death in patients who received either radiation therapy or radical prostatectomy, suggesting that the benefit of treatment was not confined to an individual treatment modality. Although treatment was associated with reduced mortality risk for each of the age categories examined, the relationship was not as strong in the older age groups as in the younger age groups (P = .03 for interaction; HR, 0.67 in ages 65-67; HR, 0.61 in ages 68-70; HR, 0.70 in ages 71-73; HR, 0.71 in ages 74-77; and HR, 0.74 in ages 78-80 years). The findings were similar across SEER sites, suggesting that the overall association between treatment and reduced mortality was not dominated by practice variations in any particular region of the country (Table 6).

Table Graphic Jump LocationTable 5. Association Between Active Treatment and Overall Mortality Among Subgroups
Table Graphic Jump LocationTable 6. Association Between Active Treatment and Overall Mortality by SEER Site Adjusting for Propensity Score Only

An observed association between treatment and outcome may reflect the effects of unknown or unmeasured confounders and the results of sensitivity analyses to estimate the potential effects of such an unmeasured confounder, such as poor functional status, on the estimated HR for treatment vs observation are shown in Table 7. These sensitivity analyses are based on the estimated HR for treatment (0.69, 95% CI, 0.66-0.72), propensity score, comorbidity, tumor grade, and tumor size. The strength of the association between treatment and mortality (measured by the HR) is affected by (1) the prevalence of the unmeasured confounder in the observation group, (2) the prevalence of the unmeasured confounder in the treated group, and (3) the hazard associated with the unmeasured confounder, which was assumed to be the same in the treated and untreated groups.

Table Graphic Jump LocationTable 7. Sensitivity Analysis Estimating the Effect of an Unmeasured Confounder on the Hazard Ratio of Death

In these sensitivity analyses, we used prevalence rates of overall disability,19 Alzheimer dementia,20 and osteoporotic fractures21 for examples of prevalence rates of an unmeasured confounder in the treated group (10%-30%). We used the increased risk of death associated with these conditions as examples of the hazard associated with unmeasured confounder (1.25-3.3). We then varied the prevalence of the unmeasured confounder in the observation group to determine the extent to which its distribution under these conditions would need to be imbalanced to influence the statistical significance of the results (ie, the upper bound of the 95% CI crosses 1.0). For instance, (as shown in rows 4-6 of Table 7) for an unmeasured confounder with a prevalence of 10% in the treated group and with an associated 2-fold increased risk of death (HR 2.0), the prevalence of the unmeasured confounder would need to be 60% in the untreated group to influence the estimated HR (for death in the treatment group) enough to make the results nonsignificant.

Although prostate cancer is the most common cancer affecting US men, management of localized prostate cancer remains controversial. SEER data suggest that up to 15% of PSA-detected prostate cancer in white men and up to 37% in black men might never have presented clinically within the patient's lifetime.22 Cohort studies have shown that low-grade disease may have a long natural history, even if conservatively managed.13 Furthermore, large randomized clinical trials of treatment of localized prostate cancer have proven difficult to conduct, adding to the uncertainty about the role of active treatment for men with low- and intermediate-risk disease. The only published randomized trial of men in this risk group included fewer than 700 patients, was conducted in Scandinavia, and left considerable uncertainty about the effect of treatment on men older than 65 years.5

In our observational analysis, using SEER data, we found that men with early stage and low- or intermediate-risk prostate cancer who underwent active treatment with either radical prostatectomy or radiation therapy within 6 months after diagnosis were 30% less likely to die during the subsequent 12 years of follow-up than men who did not undergo active treatment within 6 months after diagnosis. To address the issue of confounding, we combined observed potential confounders into 1 composite variable, the propensity score, and demonstrated that this approach balanced important measured covariates between groups except comorbidity, tumor size, and tumor grade, which we included as separate covariates. The association between active treatment and mortality was consistent across propensity score quintiles and within clinical subgroups. These results extend the currently available literature about the effectiveness of treatment for localized prostate cancer. The effect size we observed was similar to that of the only published randomized trial of radical prostatectomy (relative risk, 0.74; 95% CI, 0.56-0.99).5 However, this trial also reported that the effect of surgery was smaller among men older than 65 years (P = .08 for interaction), although the point estimate in this subgroup was not reported.5 Our finding is also supported by nonrandomized studies that have reported prolonged disease control with radiation therapy.23,24

Because our study used an observational cohort rather than a randomized controlled trial design, these findings must be interpreted within the context of the limitations of observational data. In particular, because patients were not randomized, treatment and observation groups may differ in measured and unmeasured ways that are associated with differences in survival. Of greatest concern, men who are offered treatment (or specific types of treatment) or those who choose treatment may be “healthier” than men who are not offered or do not choose treatment, raising the possibility that the apparent benefit of treatment arises from the selection of healthier men rather than a direct effect of treatment on prostate cancer mortality.

This possibility certainly must be considered when interpreting the study findings, but several factors reduce the likelihood that the observed benefit of treatment is solely due to selection bias. First, Medicare data provide extensive information on comorbidities and the benefit of treatment persisted irrespective of the method of adjusting for these comorbidities.

Second, the observed treatment survival advantage remained when we examined the “healthiest” patients (ie, those without any claims for comorbid conditions within 90 days of date of diagnosis) or excluded patients who died within 1, 2, and 3 years of diagnosis, respectively.

Third, the vast majority of localized prostate cancer is diagnosed through PSA screening, and men who undergo PSA screening already have been selected based on functional status and estimated life expectancy.25 Thus, although SEER-Medicare data do not provide a direct measure of functional status, we believe that this cohort is likely to represent a relatively healthy subset of elderly men, making functional status potentially less important as a confounder.

Fourth, the association between treatment and reduced mortality risk was strongest among younger patients and weakest among older patients. This relationship also was seen in the Swedish randomized controlled trial5 and argues against a pure effect of selection bias because greater selection pressure might be expected to target surgery for the healthiest individuals among older rather than among younger men.

And fifth, our sensitivity analyses demonstrated that our main result is fairly robust and only can be explained by an uneven distribution of an unknown cofounder that had a high influence on mortality (Table 7). For example, using a baseline prevalence rate of 10% in the untreated group and an increased risk of death of 2.0, an unmeasured confounder would need to have a prevalence of 60% (6 times the baseline prevalence) to eliminate a statistically significant association of treatment and reduced mortality. Although we believe that these factors support the validity of these results, the possibility of significant selection bias remains and must be considered when applying these findings to clinical decisions. These issues can only be addressed definitively by a rigorously designed and conducted randomized controlled trial.

The patient's probability of developing adverse effects and his willingness to tolerate them should be an important part of the patient's decision to pursue active treatment. Treatment with radiation and surgery can be associated with urinary, bowel, and sexual dysfunction symptoms, which may affect quality of life but are not likely to affect overall survival. Patients with conservatively managed prostate cancer may develop obstructive urinary symptoms due to local disease progression or systemic symptoms due to metastatic disease. In the Scandinavian study, men who underwent radical prostatectomy had increased rates of urinary leakage and erectile dysfunction but lower rates of obstructing voiding symptoms.26 The National Cancer Institute–sponsored Patterns of Care Study found that men who underwent active treatment had higher rates of treatment related adverse effects but were more satisfied with their overall treatment decision.27 Thus, patients should be informed of the risks and benefits of treatment options, including the effects of treatment on quality of life and overall survival before deciding upon a treatment strategy.

Our study has several other limitations. Men with localized prostate cancer are likely to have participated in PSA screening, and these men may be healthier, more motivated, and have greater access to the health care system than the general population. Therefore, our results may not be generalizable to the unscreened US population. For example, we do not have direct measures of comorbidities that may be common in the unscreened US population. Condiditions common in geratric patients, such as Alzheimer or vascular dementia, fraility, Parkinson disease, may make patients poor candidates for aggressive treatment for early stage prostate cancer. We also did not evaluate the role of delayed therapy for patients in the observation group. However, if patients received treatment more than 6 months past diagnosis, this would improve the overall survival in the observation group and bias the results toward the null. We also do not have data about the use of oral antiandrogen agents or palliative chemotherapy. However, their effect on the overall survival of men with localized prostate cancer is likely to have been small. Because managed care provides incomplete claims data to Medicare, our cohort included only patients with fee-for-service coverage. If there are differences in the quality of prostate cancer treatment between managed care and fee-for-service health care centers, the association between treatment and survival among men in managed care may differ from that in our cohort.

Studies of surgery vs other treatment may result in stage and grade migration when the final surgical pathology reveals a higher tumor grade or stage than expected from clinical staging. This can result in an apparent benefit of surgery, simply from reclassifying some patients into a higher stage or grade group. This bias may be further exacerbated if surgeons abort a prostectomy procedure if positive nodes are discovered intraoperatively and these patients are classified as not having undergone surgery.28 However, nomograms that incorporate PSA level, clinical stage, and Gleason score are now widely used to predict pathologic stage. It is recommended that patients who have a high probability of having lymph node–positive disease undergo pelvic imaging and fine-needle aspiration sampling of suspicious appearing nodes, reducing the number of patients with node-positive disease undergoing radical prostatectomy.29,30 In addition, intraoperative frozen section of pelvic lymph nodes is no longer routinely performed in the United States, making it unlikely that patients in our study cohort would have had their procedure terminated due to lymph node involvement found intraoperatively.

We attempted to minimize bias from possible stage migration by using clinical extension classification in the analyses. This classification relies solely on tumor size and would exclude information on nodal metastases that may have been found intraoperatively. Although this classification only became available in SEER in 1994, we found no difference in the results for patients diagnosed prior to 1994 and those diagnosed after 1994. Similarly, grade migration has become unlikely since the practice of routinely collecting 10 to 12 biopsies has resulted in a high correlation between presurgical and postsurgical tumor grade. It is generally accepted that Gleason scoring methods have changed over time, resulting in an “upgrading” of tumors diagnosed in the latter years of our cohort compared with those diagnosed in the earlier years. Although this Gleason score upward migration may result in improved grade-specific outcomes over time, the association of therapy with reduced mortality was apparent in the subgroups diagnosed both before and after 1994. Upgrading would have resulted in a cohort of patients with a better prognosis, yet the association with treatment was still statistically significant.

Despite these limitations, our analysis has several strengths. By using SEER-Medicare data, we were able to examine the outcomes of treatment in the Medicare population (≥65 years), a group that is traditionally under-represented in clinical trials. Although 70% of incident cases of prostate cancer occur in men aged 65 years or older, all of the large randomized clinical trials have an age cutoff for enrollment (75 years for PIVOT31 and the Scandinavian study, 69 years for ProtecT),32 limiting their generalizablity to the Medicare population.

We used propensity scores in our analysis, which is a powerful method of adjusting for confounding by multiple covariates. One limitation of traditional covariate-based regression models is that it is difficult to determine if there is adequate overlap across all covariates between the observation and treatment groups. In traditional multivariable models, the treatment effects may be attributed to covariates that are unevenly distributed between groups. However, grouping patients into quintiles by propensity scores allows evaluation of whether adequate overlap exists between groups to balance potential confounding covariates.18

Our study has a relatively long follow-up, which is essential for studies of prostate cancer given the long natural history of this disease. For example, in the Scandinavian study, the effect of surgery on overall survival was not evident in the analysis conducted at 6.2 years of follow-up, but it was found at the analysis reported at 8.2 years.5 This long follow-up of our cohort also allowed us to detect differences in overall survival as our primary end point rather than necessitating reliance on intermediate markers such as biochemical recurrence, which may be less strongly correlated with clinical status or survival.33,34 We were able to examine both overall survival and disease-specific survival, demonstrating a similar magnitude of effect for both measures. Furthermore, among men with prostate cancer−specific mortality, patients who initially underwent treatment had a longer median survival than those who were observed. The median survival of patients whose deaths were attributed to prostate cancer was significantly shorter in both groups than in the overall cohort. Whether this outcome reflects inaccuracies in death certificate reporting35 or the natural history of the disease is uncertain.

In summary, even though prostate cancer commonly is considered an indolent disease, this observational study suggests a reduced risk of mortality associated with active treatment for low- and intermediate-risk prostate cancer in the elderly Medicare population examined. Because observational data can never be free of concerns about selection bias and confounding, these results must be validated by rigorous randomized controlled trials of elderly men with localized prostate cancer before the findings can be used to inform treatment decisions.

Corresponding Author: Yu-Ning Wong, MD, Divisions of Population Science and Medical Science, Fox Chase Cancer Center, 333 Cottman Ave, Philadelphia, PA 19111 (Y_Wong@fccc.edu).

Author Contributions: Dr Wong had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Wong, Mitra, Hudes, Schwartz, Armstong.

Acquisition of data: Montagnet, Armstong.

Analysis and interpretation of data: Wong, Mitra, Hudes, Localio, Wan, Armstong.

Drafting of the manuscript: Wong, Mitra, Localio.

Critical revision of the manuscript for important intellectual content: Mitra, Hudes, Localio, Schwartz, Wan, Montagnet, Armstong.

Statistical analysis: Wong, Mitra, Localio, Wan, Armstong.

Obtained funding: Armstrong.

Administrative, technical, or material support: Wan, Montagnet, Armstong.

Study supervision: Mitra, Hudes, Localio, Schwartz, Armstong.

Financial Disclosures: Dr Schwartz reports that he serves on a health outcomes advisory board for Sanofi Aventis and has applied for grant funding from Sanofi Aventis; otherwise no other financial disclosures were reported.

Funding/Support: Dr Wong was supported by grant R25 CA057708 from the National Institutes of Health when this research was conducted and is supported in part by a Clinical Research Curriculum Award (K30 RR022255) and by a University of Pennsylvania School of Medicine Scholarship. This research was sponsored by the Center for Population Health and Health Disparities at the University of Pennsylvania under Public Health Services grant P50-CA105641. This study used the linked SEER-Medicare database.

Role of the Sponsors: The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.

Disclaimer: The interpretation and reporting of these data are the sole responsibility of the authors.

Acknowledgment: We appreciate the assistance from Dr Robert Uzzo, Fox Chase Cancer Center for his helpful comments on an earlier draft of the manuscript, for which he received no compensation. The authors acknowledge the efforts of the Applied Research Program, NCI; the Office of Research, Development and Information, CMS; Information Management Services (IMS) Inc; and the SEER Program tumor registries in the creation of the SEER-Medicare database.

This article was corrected on 12/12/06, pror to publication of the correction in print.

Albertsen PC, Hanley JA, Fine J. 20-Year outcomes following conservative management of clinically localized prostate cancer.  JAMA. 2005;293:2095-2101
PubMed   |  Link to Article
Albertsen PC, Hanley JA, Gleason DF, Barry MJ. Competing risk analysis of men aged 55 to 74 years at diagnosis managed conservatively for clinically localized prostate cancer.  JAMA. 1998;280:975-980
PubMed   |  Link to Article
Chodak GW, Thisted RA, Gerber GS.  et al.  Results of conservative management of clinically localized prostate cancer.  N Engl J Med. 1994;330:242-248
PubMed   |  Link to Article
Harlan SR, Cooperberg MR, Elkin EP.  et al.  Time trends and characteristics of men choosing watchful waiting for initial treatment of localized prostate cancer: results from CaPSURE.  J Urol. 2003;170:1804-1807
PubMed   |  Link to Article
Bill-Axelson A, Holmberg L, Ruutu M.  et al.  Radical prostatectomy versus watchful waiting in early prostate cancer.  N Engl J Med. 2005;352:1977-1984
PubMed   |  Link to Article
Holmberg L, Bill-Axelson A, Helgesen F.  et al.  A randomized trial comparing radical prostatectomy with watchful waiting in early prostate cancer.  N Engl J Med. 2002;347:781-789
PubMed   |  Link to Article
Potosky AL, Harlan LC, Stanford JL.  et al.  Prostate cancer practice patterns and quality of life: the Prostate Cancer Outcomes Study.  J Natl Cancer Inst. 1999;91:1719-1724
PubMed   |  Link to Article
 Prostate Cancer Clinical Practice Guidelines in Oncology. Version 2.2005. Jenkintown, Pa: National Comprehensive Cancer Network Inc; 2006. http://www.nccn.org. Accessed October 1, 2006
Chodak GW, Keane T, Klotz L. Critical evaluation of hormonal therapy for carcinoma of the prostate.  Urology. 2002;60:201-208
PubMed   |  Link to Article
Alibhai SM, Leach M, Tomlinson G.  et al.  30-day mortality and major complications after radical prostatectomy: influence of age and comorbidity.  J Natl Cancer Inst. 2005;97:1525-1532
PubMed   |  Link to Article
Houterman S, Janssen-Heijnen ML, Hendrikx AJ, van den Berg HA, Coebergh JW. Impact of comorbidity on treatment and prognosis of prostate cancer patients: a population-based study.  Crit Rev Oncol Hematol. 2006;58:60-67
PubMed   |  Link to Article
Klabunde CN, Potosky AL, Legler JM, Warren JL. Development of a comorbidity index using physician claims data.  J Clin Epidemiol. 2000;53:1258-1267
PubMed   |  Link to Article
Silber JH, Rosenbaum PR, Trudeau ME.  et al.  Multivariate matching and bias reduction in the surgical outcomes study.  Med Care. 2001;39:1048-1064
PubMed   |  Link to Article
Rosenbaum PRRD. Reducing bias in observational studies using subclassification on the propensity score.  J Am Stat Assoc. 1984;79:516-524
Link to Article
Rubin DB. Estimating causal effects from large data sets using propensity scores.  Ann Intern Med. 1997;127:757-763
PubMed   |  Link to Article
Cochran WG. The effectiveness of adjustment by subclassification in removing bias in observational studies.  Biometrics. 1968;24:295-313
PubMed   |  Link to Article
Lin DY, Psaty BM, Kronmal RA. Assessing the sensitivity of regression results to unmeasured confounders in observational studies.  Biometrics. 1998;54:948-963
PubMed   |  Link to Article
Mitra N, Heitjan DF. Sensitivity of the hazard ratio to non-ignorable treatment assignment in an observational study. University of Pennsylvania Biostatistics Working Papers Series; 2005. Paper 5. http://biostats.bepress.com/cgi/viewcontent.cgi?article=1005&context=upennbiostat. Accessed November 2, 2006
Spillman BC. Changes in elderly disability rates and the implications for health care utilization and cost.  Milbank Q. 2004;82:157-194
PubMed   |  Link to Article
Evans DA, Funkenstein HH, Albert MS.  et al.  Prevalence of Alzheimer's disease in a community population of older persons: higher than previously reported.  JAMA. 1989;262:2551-2556
PubMed   |  Link to Article
Chang KP, Center JR, Nguyen TV, Eisman JA. Incidence of hip and other osteoporotic fractures in elderly men and women: Dubbo Osteoporosis Epidemiology Study.  J Bone Miner Res. 2004;19:532-536
PubMed   |  Link to Article
Etzioni R, Penson DF, Legler JM.  et al.  Overdiagnosis due to prostate-specific antigen screening: lessons from US Prostate Cancer Incidence Trends.  J Natl Cancer Inst. 2002;94:981-990
PubMed   |  Link to Article
Shipley WU, Thames HD, Sandler HM.  et al.  Radiation therapy for clinically localized prostate cancer: a multi-institutional pooled analysis.  JAMA. 1999;281:1598-1604
PubMed   |  Link to Article
Potters L, Morgenstern C, Calugaru E.  et al.  12-year outcomes following permanent prostate brachytherapy in patients with clinically localized prostate cancer.  J Urol. 2005;173:1562-1566
PubMed   |  Link to Article
Hoffman RM, Stone SN, Espey D, Potosky AL. Differences between men with screening-detected versus clinically diagnosed prostate cancers in the USA.  BMC Cancer. 2005;5:27
PubMed   |  Link to Article
Steineck G, Helgesen F, Adolfsson J.  et al.  Quality of life after radical prostatectomy or watchful waiting.  N Engl J Med. 2002;347:790-796
PubMed   |  Link to Article
Hoffman RM, Hunt WC, Gilliland FD, Stephenson RA, Potosky AL. Patient satisfaction with treatment decisions for clinically localized prostate carcinoma. Results from the Prostate Cancer Outcomes Study.  Cancer. 2003;97:1653-1662
PubMed   |  Link to Article
Lu-Yao GL, Yao SL. Population-based study of long-term survival in patients with clinically localised prostate cancer.  Lancet. 1997;349:906-910
PubMed   |  Link to Article
Beissner RSSJ, Speights VO, Coffield KS, Spiekerman AM, Riggs M. Frozen section diagnosis of metastatic prostate adenocarcinoma in pelvic lymphadenectomy compared with nomogram prediction of metastasis.  Urology. 2002;59:721-725
PubMed   |  Link to Article
Briganti A, Chun FK, Salonia A.  et al.  Validation of a nomogram predicting the probability of lymph node invasion among patients undergoing radical prostatectomy and an extended pelvic lymphadenectomy.  Eur Urol. 2006;175:721-725
Wilt TJ, Brawer MK. The Prostate Cancer Intervention vs Observation Trial: a randomized trial comparing radical prostatectomy vs expectant management for the treatment of clinically localized prostate cancer.  J Urol. 1994;152:1910-1914
PubMed
Donovan J, Hamdy F, Neal D.  et al.  ProtecT Study Group. Prostate Testing for Cancer and Treatment (ProtecT) feasibility study.  Health Technol Assess. 2003;7:1-88
PubMed
Kim-Sing C, Pickles T. Intervention after PSA failure: examination of intervention time and subsequent outcomes from a prospective patient database.  Int J Radiat Oncol Biol Phys. 2004;60:463-469
PubMed   |  Link to Article
Pound CR, Partin AW, Eisenberger MA, Chan DW, Pearson JD, Walsh PC. Natural history of progression after PSA elevation following radical prostatectomy.  JAMA. 1999;281:1591-1597
PubMed   |  Link to Article
Smith Sehdev AES, Hutchins GM. Problems with proper completion and accuracy of the cause-of-death statement.  Arch Intern Med. 2001;161:277-284
PubMed   |  Link to Article

Figures

Figure. Kaplan Meier Survival Curves for Full Cohort
Graphic Jump Location

Patients who survived less than 12 months were excluded from the analysis.

Tables

Table Graphic Jump LocationTable 1. Patient Baseline Characteristics and Predictors of Receiving Treatment
Table Graphic Jump LocationTable 2. Distribution of Covariates by Propensity Score Quintiles*
Table Graphic Jump LocationTable 3. Patient Distribution and 5- and 10-Year Overall Survival by Treatment and Propensity Score Strata
Table Graphic Jump LocationTable 4. Association Between Active Treatment and Overall Mortality
Table Graphic Jump LocationTable 5. Association Between Active Treatment and Overall Mortality Among Subgroups
Table Graphic Jump LocationTable 6. Association Between Active Treatment and Overall Mortality by SEER Site Adjusting for Propensity Score Only
Table Graphic Jump LocationTable 7. Sensitivity Analysis Estimating the Effect of an Unmeasured Confounder on the Hazard Ratio of Death

References

Albertsen PC, Hanley JA, Fine J. 20-Year outcomes following conservative management of clinically localized prostate cancer.  JAMA. 2005;293:2095-2101
PubMed   |  Link to Article
Albertsen PC, Hanley JA, Gleason DF, Barry MJ. Competing risk analysis of men aged 55 to 74 years at diagnosis managed conservatively for clinically localized prostate cancer.  JAMA. 1998;280:975-980
PubMed   |  Link to Article
Chodak GW, Thisted RA, Gerber GS.  et al.  Results of conservative management of clinically localized prostate cancer.  N Engl J Med. 1994;330:242-248
PubMed   |  Link to Article
Harlan SR, Cooperberg MR, Elkin EP.  et al.  Time trends and characteristics of men choosing watchful waiting for initial treatment of localized prostate cancer: results from CaPSURE.  J Urol. 2003;170:1804-1807
PubMed   |  Link to Article
Bill-Axelson A, Holmberg L, Ruutu M.  et al.  Radical prostatectomy versus watchful waiting in early prostate cancer.  N Engl J Med. 2005;352:1977-1984
PubMed   |  Link to Article
Holmberg L, Bill-Axelson A, Helgesen F.  et al.  A randomized trial comparing radical prostatectomy with watchful waiting in early prostate cancer.  N Engl J Med. 2002;347:781-789
PubMed   |  Link to Article
Potosky AL, Harlan LC, Stanford JL.  et al.  Prostate cancer practice patterns and quality of life: the Prostate Cancer Outcomes Study.  J Natl Cancer Inst. 1999;91:1719-1724
PubMed   |  Link to Article
 Prostate Cancer Clinical Practice Guidelines in Oncology. Version 2.2005. Jenkintown, Pa: National Comprehensive Cancer Network Inc; 2006. http://www.nccn.org. Accessed October 1, 2006
Chodak GW, Keane T, Klotz L. Critical evaluation of hormonal therapy for carcinoma of the prostate.  Urology. 2002;60:201-208
PubMed   |  Link to Article
Alibhai SM, Leach M, Tomlinson G.  et al.  30-day mortality and major complications after radical prostatectomy: influence of age and comorbidity.  J Natl Cancer Inst. 2005;97:1525-1532
PubMed   |  Link to Article
Houterman S, Janssen-Heijnen ML, Hendrikx AJ, van den Berg HA, Coebergh JW. Impact of comorbidity on treatment and prognosis of prostate cancer patients: a population-based study.  Crit Rev Oncol Hematol. 2006;58:60-67
PubMed   |  Link to Article
Klabunde CN, Potosky AL, Legler JM, Warren JL. Development of a comorbidity index using physician claims data.  J Clin Epidemiol. 2000;53:1258-1267
PubMed   |  Link to Article
Silber JH, Rosenbaum PR, Trudeau ME.  et al.  Multivariate matching and bias reduction in the surgical outcomes study.  Med Care. 2001;39:1048-1064
PubMed   |  Link to Article
Rosenbaum PRRD. Reducing bias in observational studies using subclassification on the propensity score.  J Am Stat Assoc. 1984;79:516-524
Link to Article
Rubin DB. Estimating causal effects from large data sets using propensity scores.  Ann Intern Med. 1997;127:757-763
PubMed   |  Link to Article
Cochran WG. The effectiveness of adjustment by subclassification in removing bias in observational studies.  Biometrics. 1968;24:295-313
PubMed   |  Link to Article
Lin DY, Psaty BM, Kronmal RA. Assessing the sensitivity of regression results to unmeasured confounders in observational studies.  Biometrics. 1998;54:948-963
PubMed   |  Link to Article
Mitra N, Heitjan DF. Sensitivity of the hazard ratio to non-ignorable treatment assignment in an observational study. University of Pennsylvania Biostatistics Working Papers Series; 2005. Paper 5. http://biostats.bepress.com/cgi/viewcontent.cgi?article=1005&context=upennbiostat. Accessed November 2, 2006
Spillman BC. Changes in elderly disability rates and the implications for health care utilization and cost.  Milbank Q. 2004;82:157-194
PubMed   |  Link to Article
Evans DA, Funkenstein HH, Albert MS.  et al.  Prevalence of Alzheimer's disease in a community population of older persons: higher than previously reported.  JAMA. 1989;262:2551-2556
PubMed   |  Link to Article
Chang KP, Center JR, Nguyen TV, Eisman JA. Incidence of hip and other osteoporotic fractures in elderly men and women: Dubbo Osteoporosis Epidemiology Study.  J Bone Miner Res. 2004;19:532-536
PubMed   |  Link to Article
Etzioni R, Penson DF, Legler JM.  et al.  Overdiagnosis due to prostate-specific antigen screening: lessons from US Prostate Cancer Incidence Trends.  J Natl Cancer Inst. 2002;94:981-990
PubMed   |  Link to Article
Shipley WU, Thames HD, Sandler HM.  et al.  Radiation therapy for clinically localized prostate cancer: a multi-institutional pooled analysis.  JAMA. 1999;281:1598-1604
PubMed   |  Link to Article
Potters L, Morgenstern C, Calugaru E.  et al.  12-year outcomes following permanent prostate brachytherapy in patients with clinically localized prostate cancer.  J Urol. 2005;173:1562-1566
PubMed   |  Link to Article
Hoffman RM, Stone SN, Espey D, Potosky AL. Differences between men with screening-detected versus clinically diagnosed prostate cancers in the USA.  BMC Cancer. 2005;5:27
PubMed   |  Link to Article
Steineck G, Helgesen F, Adolfsson J.  et al.  Quality of life after radical prostatectomy or watchful waiting.  N Engl J Med. 2002;347:790-796
PubMed   |  Link to Article
Hoffman RM, Hunt WC, Gilliland FD, Stephenson RA, Potosky AL. Patient satisfaction with treatment decisions for clinically localized prostate carcinoma. Results from the Prostate Cancer Outcomes Study.  Cancer. 2003;97:1653-1662
PubMed   |  Link to Article
Lu-Yao GL, Yao SL. Population-based study of long-term survival in patients with clinically localised prostate cancer.  Lancet. 1997;349:906-910
PubMed   |  Link to Article
Beissner RSSJ, Speights VO, Coffield KS, Spiekerman AM, Riggs M. Frozen section diagnosis of metastatic prostate adenocarcinoma in pelvic lymphadenectomy compared with nomogram prediction of metastasis.  Urology. 2002;59:721-725
PubMed   |  Link to Article
Briganti A, Chun FK, Salonia A.  et al.  Validation of a nomogram predicting the probability of lymph node invasion among patients undergoing radical prostatectomy and an extended pelvic lymphadenectomy.  Eur Urol. 2006;175:721-725
Wilt TJ, Brawer MK. The Prostate Cancer Intervention vs Observation Trial: a randomized trial comparing radical prostatectomy vs expectant management for the treatment of clinically localized prostate cancer.  J Urol. 1994;152:1910-1914
PubMed
Donovan J, Hamdy F, Neal D.  et al.  ProtecT Study Group. Prostate Testing for Cancer and Treatment (ProtecT) feasibility study.  Health Technol Assess. 2003;7:1-88
PubMed
Kim-Sing C, Pickles T. Intervention after PSA failure: examination of intervention time and subsequent outcomes from a prospective patient database.  Int J Radiat Oncol Biol Phys. 2004;60:463-469
PubMed   |  Link to Article
Pound CR, Partin AW, Eisenberger MA, Chan DW, Pearson JD, Walsh PC. Natural history of progression after PSA elevation following radical prostatectomy.  JAMA. 1999;281:1591-1597
PubMed   |  Link to Article
Smith Sehdev AES, Hutchins GM. Problems with proper completion and accuracy of the cause-of-death statement.  Arch Intern Med. 2001;161:277-284
PubMed   |  Link to Article

Letters

CME
Meets CME requirements for:
Browse CME for all U.S. States
Accreditation Information
The American Medical Association is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The AMA designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 CreditTM per course. Physicians should claim only the credit commensurate with the extent of their participation in the activity. Physicians who complete the CME course and score at least 80% correct on the quiz are eligible for AMA PRA Category 1 CreditTM.
Note: You must get at least of the answers correct to pass this quiz.
You have not filled in all the answers to complete this quiz
The following questions were not answered:
Sorry, you have unsuccessfully completed this CME quiz with a score of
The following questions were not answered correctly:
Commitment to Change (optional):
Indicate what change(s) you will implement in your practice, if any, based on this CME course.
Your quiz results:
The filled radio buttons indicate your responses. The preferred responses are highlighted
For CME Course: A Proposed Model for Initial Assessment and Management of Acute Heart Failure Syndromes
Indicate what changes(s) you will implement in your practice, if any, based on this CME course.

Multimedia

Some tools below are only available to our subscribers or users with an online account.

Web of Science® Times Cited: 144

Related Content

Customize your page view by dragging & repositioning the boxes below.

See Also...
Articles Related By Topic
Related Collections
PubMed Articles