br Statistical Analysis br After using census tract data to

Statistical Analysis
After using census tract data to estimate the household income for each individual patient, the median household income for the entire study population was calculated. Patients were then catego-rized into low- or high-income groups relative to the median household income for the entire study population.
Univariate comparisons of clinicopathologic outcomes between the low- and high-income groups were then performed using c2 analysis to test for statistical significance. Outcomes of interest included: preoperative prostate-specific antigen (PSA) level, patho-logical Gleason score, pathologic T stage, extraprostatic extension (EPE), seminal vesicle invasion (SVI), surgical margin status, and rates of UNC1999 node involvement (LNI). Rates of adverse pathology (pathological stage T3, pathological Gleason score 4þ3, or positive lymph nodes) were then assessed. Postoperative disease risk was further characterized by calculating the postsurgical Cancer of the Prostate Risk Assessment (CAPRA-S) score for each patient, as previously described.11 We then compared the distribution of CAPRA-S risk scores and rates of adverse pathology with patients stratified into income quartiles.
To analyze income as a continuous variable, linear and logistic regression analyses were conducted to test for associations between income and CAPRA-S score or the presence of adverse pathology, respectively. Each model included income (as a continuous variable) and the institution where the patient was treated as covariates. Additional measures of disease risk were not included as covariates in these models, because many of these measures are used to define the outcome of interest (CAPRA-S score or adverse pathology). Odds ratios for these models were calculated on the basis of $10,000 incremental increases in income. All statistical analyses were per-formed using STATA software (version 14; StataCorp LLC), with P values < .05 meeting statistical significance.
Results
Of the 392 patients who met initial inclusion criteria, 45 patients were excluded because of incomplete demographic or pathologic data, leaving 347 patients available for analysis (institution 1: 127 [36.6%]; institution 2: 220 [63.4%]). Table 1 shows patient demographic and disease characteristics (including CAPRA-S score and rates of adverse pathology), in the entire cohort and with patients stratified according to institution. Median age and PSA level at diagnosis were 60 (range, 40-80) and 5.9 (range, 0.9-86.7), respectively. Using the CAPRA-S scoring system, 163 (47%) patients met criteria for intermediate or high risk of disease recurrence, whereas 178 (51%) were classified as low
2 - Clinical Genitourinary Cancer Month 2019
Table 1 Patient Demographic and Disease Characteristics
Institution 1,
Institution 2,
Characteristic n (%)
Age
Data missing
P
PSA at
Diagnosis
P
Pathologic
Gleason Score
Pathologic
T-Stage
P
Surgical
Margins
P
Extraprostatic
Extension
P
Seminal Vesicle
Invasion
P
 Samuel A. Weprin et al
Table 1 Continued
Institution 1,
Institution 2,
Characteristic n (%)
Lymph Node
Metastases
Data missing
P
CAPRA-S Score
CAPRA-S
Category
Low risk
Intermediate
risk
High risk
Data missing
P
Adverse
Pathologya
No
Yes
Data missing
P
a ¼ postsurgical Cancer of the Prostate Risk Assessment.
Abbreviation: CAPRA-S
Adverse pathology: pathological stage pT3, Gleason score 4þ3, or positive lymph nodes.
risk. Approximately one-third (35.7%) of the study population fulfilled our definition of adverse pathology after RP.
The median household income for the entire study population was $37,954 (range, $11,621-$200,001). Patients were stratified into low- or high-income categories on the basis of their individual household income relative to this median. For reference, the United States federal poverty line for a family of 4 in 2015 was $24,250.
Univariate comparisons of disease characteristics and assessments of disease risk between the low- and high-income groups are shown
Clinical Genitourinary Cancer Month 2019 - 3
Poverty and Adverse Prostate Cancer Pathology
Table 2 Comparison of Disease-Specific Parameters Accord-ing to Income Level
Low
High P
Parameter
Income Income
PSA
Median (range)
Pathologic
Gleason Score
Pathologic
T-Stage
Surgical