Analysis of Outcome of Patients with Acute Myeloid Leukemia According to Socio-Economic Status

Background: Many factors have been found to impact the treatment outcomes of acute myeloid leukemia (AML) patients including age of patient at diagnosis, race, cytogenetic risk grouping, and white blood cell count. Although these clinical and demographic features might be easy to measure and identify, models using socioeconomic factors predicting treatment outcomes of AML have not been as rigorously explored. This study aims to investigate how some socioeconomic factors may impact the treatment outcomes of patients with newly diagnosed AML.

Methods: We retrospectively analyzed patients from January 2000 to June 2012 diagnosed with AML over 18 years of age, who were treated at the University of Oklahoma with induction chemotherapy. 215 AML patients were identified. The AML patient, or next of kin, was contacted and asked to complete a telephone-based questionnaire with 11 questions. A total of 181 patients or next of kin responded to questions regarding marital status, dependency on another, difficulty with transportation, renting hotel/motel for treatment, car ownership, and whether the patient's life was stressful. Simple descriptive statistics were created for all covariates [n (%)] overall. Kaplan-Meier survival curves and log-rank tests of homogeneity identified which covariates were associated with survival (α=0.25) and would be included in a multivariable Cox proportional hazards regression model. In this model, all two-way interactions were explored and backward selection was used to identify the variables included in the final model. If the exclusion of a variable changed the estimate of a significant covariate by ≥ 20%, it was deemed a confounder and retained in the model. A similar approach was used to assess complete remission (CR); however, binary logistic regression models were used to examine the univariate association of each covariate with CR to identify associated variables to include in the multivariable model (α=0.25).

Results: The majority of patients were males (61%), married (45%), and achieved CR (64%). The median age was 52 years. Most patients reported not being dependent on another (65%), not having served in the military (94%), and not having difficulty with travel (73%), although most reported owning a car (70%). After adjusting for age and risk grouping, military status is marginally associated with overall survival (OS) (p-value =0.0548). The hazard of death for those in the military is 2.36 times higher for patients with AML relative to those who have never been part of the military (95% CI: 0.98, 5.69). Marital status is not associated with OS. There is a significant association between difficulty with transportation (p-value=0.0440) and CR. Patients who reported difficulty with transportation have 0.35 times the odds of CR (or 65% lower) relative to those who do not report this difficulty (95% CI: 0.13, 0.97). Marital status is marginally significant (p-value=0.0698) when comparing married vs widowed patients. Married patients have 1.35 times the odds of remission relative to those patients who are widowed (95% CI: 0.32, 5.68).

Conclusion: There is evidence that those who served in the military have lower OS relative to those who have not. We also find that patients who reported having difficulty with transportation have lower rates of CR. Further exploration of how these variables might be associated with OS and CR is warranted given evidence from this study.