• 2022-08
  • 2022-07
  • 2022-05
  • 2022-04
  • 2021-03
  • 2020-08
  • 2020-07
  • 2020-03
  • 2019-11
  • 2019-10
  • 2019-09
  • 2019-08
  • 2019-07
  • sensitivity analyses using a period of days prior to


     sensitivity analyses using a Lactacystin of 14 days prior to treatment start date to distinguish staging MRI from MRI studies performed in the context of treatment planning. The primary study outcome was initial management for PCa, ie, observation vs definitive therapy. Observation was defined as the absence of definitive cancer-directed therapy within 12 months of diagnosis.15,16 Consistent with prior work, we identi-
    fied treatment based on Medicare claims for radical prostatec-tomy (RP), radiation therapy, androgen deprivation therapy, or ablative therapy. The date of treatment was assigned using inpa-tient Medicare Provider Analysis and Review files for admis-sions, outpatient claims, and Medicare Part D files. We compiled relevant sociodemographic, clinical and healthcare-related char-acteristics, including race, age at diagnosis, year of diagnosis, marital status, comorbidity (measured with Elixhauser score, using Medicare claims in the 12 months prior to PCa diagnosis), SEER region, metropolitan or rural status based on rural-urban continuum codes specified in SEER in 2013, state buy-in and/or low income subsidy for insurance, median household income at the zip code level, and urologist density divided into tertiles (based on 2011 hospital referral region capacity measures, Dart-mouth Atlas).17,18 The clinical factors considered included clini-cal stage (T1 vs T2), prostate-specific antigen level, and Gleason score.
    Statistical Analysis
    The primary study objective was to examine the potential associ-ation between prostate MRI and observation among men newly diagnosed with low-risk PCa. We described patient characteris-tics using frequency tables, means, standard deviations, and chi-square tests as appropriate. To identify factors associated with use of prostate MRI in the study cohort, we fit multivariable logistic regression models examining clinical, sociodemographic, and healthcare-related factors. We used propensity score match-ing to control for confounding associated with the preferential use of MRI.19 Specifically, we computed propensity scores using logistic regression models including all available clinical, socio-demographic, provider, facility, and cancer-related variables. We used 1:4 greedy matching to pair patients with a similar propen-sity for exposure (ie, receipt of MRI) within a specific limited range or caliper.20 We confirmed directional selection the group of patients who underwent MRI and the comparison group of patients who did not receive MRI were balanced across multiple covariates based on standardized differences less than 0.1. Using propensity weighted data, we then constructed conditional logistic regres-sion models to examine the associations between prostate MRI and type of initial treatment (observation vs definitive treat-ment) for low-risk PCa.
    Figure 1. Percentage of patient received prostate MRI and definitive therapy among 8144 low-risk prostate cancer patients by year of diagnosis, 2010−2013.
    clinical, sociodemographic, and treatment related characteristics are detailed in Table 1.
    We identified significant associations of patient and provider-level factors with the use of prostate MRI surrounding the diag-nosis of PCa. In multivariable logistic regression analysis, non-white race, Lactacystin age older than 75 years, residence outside of the northeast, higher urologist density, were associated with lower likelihood of receiving prostate MRI. Diagnosis in later years, highest zip code-level median household income, and clinical tumor stage T2 were associated with higher likelihood of receiv-ing prostate MRI (Table 2).
    We first examined the association between prostate MRI and initial management for PCa using multivariable logistic regres-sion. Patients who received prostate MRI were significantly more likely to receive observation (OR = 2.01, 95% CI: 1.66-2.44). In addition, non-white race, older age, being unmarried, low clinical stage, and residence in a region with higher urologist density were significantly associated with the likelihood of receiving observation vs definitive treatment (Table 3).