Archives

  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • data include demographic clinical and mortality data along w

    2020-08-14

     data include demographic, clinical, and mortality data along with each individual’s Medicare claims data.
    To obtain the study sample, we included patients diagnosed be-tween 2007 and 2012 with a single, stage I-IV cancer of the kidney, identified using the International Classification of Diseases for Oncology, Third Edition code C649. We limited the sample to patients age 66 and older at diagnosis who were continuously enrolled in fee-for-service Medicare (both Parts A and B) from the time of diagnosis until death or last follow-up. Additionally, we restricted the sample to patients who received surgery as the initial treatment (defined as nephrectomy) within 6 months of diagnosis and excluded deaths on the day of surgical discharge since we would be unable to observe their subsequent rate of ED visits. Surgery was identified as inpatient or outpatient nephrectomy with an ICD-9 procedure code of: 55.3,
    Our primary goal was to understand the relationship between 3 primary outcomes—hospital-level risk-adjusted ED visits, risk-adjusted mortality rates, and risk-adjusted costs. We identified an ED visit from an inpatient claim with a positive value for the emergency room charge23 or an outpatient claim with a revenue center code of 0450-0459 or 0981.24 Any inpatient claim that occurred within 1 day of discharge of an outpatient claim was assumed to represent the same ED visit.
    For each hospital we calculated a ratio of 30-day and 365-day total actual ED visits divided by risk-adjusted predicted ED visits. The ratio measure was similar to the type of measures often used in the literature to examine how factors such as volume affect hospital outcomes.25,26 To estimate predicted ED visits, we used logistic regression to model the occurrence of an ED visit within 30 and 365 days of surgery. The former was meant to capture initial complications while the latter was meant to capture some of the longer term follow-up often incorporated into quality measures. The logistic models included patient stage at diagnosis (stages I, II, III, IV), age at diagnosis, sex, Ferrostatin-1 (White, Black, Asian, or Other), rurality (big metro, metro, urban, less urban/rural), marital status (married, nonmarried), histology group (clear cell carcinoma, renal cell carcinoma, other), Charlson comorbidity index (CCI) score,22 median household income in patient’s zip code, and receipt of systemic targeted therapy before and after surgery but before eval-uation point (ie, 30 or 365 days post surgery). Based on the model predictions, the ratio of actual ED visits divided by predicted ED visits was calculated. Values greater than 1 represented hospitals with higher-than-predicted ED visit rates.
    We used the same approach to estimate 30 and 365-day risk-adjusted mortality. However, our sample contained only 60 deaths within 30 days, so 30-day mortality was excluded. For costs, we estimated 6- and 12-month total costs for patients and total costs less the cost of any ED visits using generalized linear models with a gamma distribution and log link function and the same set of control variables used in the ED rate models. The 2 cost variables were meant to capture whether costs were driven by the ED visit itself or were indicative of higher overall costs.
    Clinical Genitourinary Cancer June 2019 - e651
    Table 1 Patient Characteristics by Disease Stage at Diagnosis
    Risk-Adjusted ED Visits
    Stage
    Stage
    Table1Continued
    StagesIandIIcombinedduetoreportingrequirementrelatedtosmallsamplesize.
    [CharacteristicStageI(N3791) CharlsonComorbidityScore Mean(SD)1.3(1.52) Median1.0 Histology Clear-cellcarcinoma2182(57.6%) Renal-cellcarcinoma664(17.5%) Otherhistology945(24.9%) SystemicTargetedTherapyBeforeSurgery Received39(0.9%) SystemicTargetedTherapyAfterSurgery(Within365Days) Received29(0.8%)
    a
    a
     Joel E. Segel et al
    Finally, we examined the associations between a hospital’s 30-day and 365-day risk-adjusted ED visit measure and its risk-adjusted mortality measures and its cost measures. To estimate the associa-tions we used simple linear regression models restricted to hospitals with more than 10 patients during the course of the study.
    Results
    The data contained 6078 kidney cancer patients in 667 hospitals life history met all inclusion and exclusion criteria. As shown in Table 1, patient characteristics tended to vary by stage. Patients with more advanced stage of kidney cancer tended to be male, more likely White, and married. As shown in Table 2, ED visits, mortality, and costs all increased with stage. The percentages of patients with an ED visit within 30 days ranged from 12.4% for stage II disease to 23.8% for stage IV disease, where 53.1% of all ED visits within 30 days ultimately led to an inpatient admission. In addition, 365-day percentages ranged from 34.8% with stage II disease to 67.2% with stage IV disease where 58.2% of all ED visits within 365 days led to an inpatient admission.