We consider the problem of identifying a subgroup of patients who may have an enhanced treatment effect in a randomized clinical trial, and it is desirable that the subgroup be defined by a limited number of covariates. For this problem, the development of a standard, pre-determined strategy may help to avoid the well-known dangers of subgroup analysis. We present a method developed to find subgroups of enhanced treatment effect. This method, referred to as "Virtual Twins", involves predicting response probabilities for treatment and control "twins" for each subject. The difference in these probabilities is then used as the outcome in a classification or regression tree, which can potentially include any set of the covariates. We define a measure Q(Â) to be the difference between the treatment effect in estimated subgroup  and the marginal treatment effect. We present several methods developed to obtain an estimate of Q(Â), including estimation of Q(Â) using estimated probabilities in the original data, using estimated probabilities in newly simulated data, two cross-validation-based approaches and a bootstrap-based bias corrected approach. Results of a simulation study indicate that the Virtual Twins method noticeably outperforms logistic regression with forward selection when a true subgroup of enhanced treatment effect exists. Generally, large sample sizes or strong enhanced treatment effects are needed for subgroup estimation. As an illustration, we apply the proposed methods to data from a randomized clinical trial.

IMPORTANCE Zoledronic acid, a third-generation aminobisphosphonate, reduces the incidence of skeletal-related events and pain in patients with bone metastases. The optimal dosing interval for zoledronic acid is uncertain.OBJECTIVE To determine whether zoledronic acid administered every 12 weeks is noninferior to zoledronic acid administered every 4 weeks. DESIGN, SETTING, PARTICIPANTS Randomized, open-label clinical trial conducted at 269 academic and community sites in the United States. Patients (n = 1822) with metastatic breast cancer, metastatic prostate cancer, or multiple myeloma who had at least 1 site of bone involvement were enrolled between May 2009 and April 2012; follow-up concluded in April 2014.INTERVENTIONS Patients were randomized to receive zoledronic acid administered intravenously every 4 weeks (n = 911) vs every 12 weeks (n = 911) for 2 years. MAIN OUTCOMES AND MEASURESThe primary end point was the proportion of patients having at least 1 skeletal-related event (defined as clinical fracture, spinal cord compression, radiation to bone, or surgery involving bone) within 2 years after randomization and a between-group absolute difference of 7% as the noninferiority margin. Secondary end points included the proportion of patients with at least 1 skeletal-related event by disease type, pain as assessed by the Brief Pain Inventory (range, 0-10; higher scores indicate worse pain), Eastern Cooperative Oncology Group performance status (range, 0-4; higher scores indicate worse disability), incidence of osteonecrosis of the jaw, kidney dysfunction, skeletal morbidity rate (mean number of skeletal-related events per year), and, in a subset of 553 patients, suppression of bone turnover (assessed by C-terminal telopeptide levels). RESULTS Among 1822 patients who were randomized (median age, 65 years; 980 [53.8%] women; 855 with breast cancer, 689 with prostate cancer, and 278 with multiple myeloma), 795 completed the study at 2 years. A total of 260 patients (29.5%) in the zoledronic acid every 4-week dosing group and 253 patients (28.6%) in the every 12-week dosing group experienced at least 1 skeletal-related event within 2 years of randomization (risk difference of −0.3% [1-sided 95% CI, −4% to ϱ]; P < .001 for noninferiority). The proportions of skeletal-related events did not differ significantly between the every 4-week dosing group vs the every 12-week dosing group for patients with breast cancer, prostate cancer, or multiple myeloma. Pain scores, performance status scores, incidence of jaw osteonecrosis, and kidney dysfunction did not differ significantly between the treatment groups. Skeletal morbidity rates were numerically identical in both groups, but bone turnover was greater (C-terminal telopeptide levels were higher) among patients who received zoledronic acid every 12 weeks.CONCLUSIONS AND RELEVANCE Among patients with bone metastases due to breast cancer, prostate cancer, or multiple myeloma, the use of zoledronic acid every 12 weeks compared with the standard dosing interval of every 4 we...

Background Health care utilization in older adults (≥60) with acute myeloid leukemia (AML) has not been well-studied. Methods We conducted a retrospective analysis of 330 consecutive older patients diagnosed with AML between 5/1/2005 and 12/23/2011 at two hospitals in Boston to examine their health care utilization and end of life (EOL) care. Using multivariable logistic and linear regression models adjusting for covariates, we also compared health care utilization for patients undergoing intensive induction (n=197; cytarabine/ anthracyline combination) versus non-intensive chemotherapy (n=133; single-agent therapy). Results The median number of hospitalizations for the entire cohort was 4.2 (range 1–18). Patients who died spent a mean of 28.3% of their life from diagnosis in the hospital and 13.8% of their life attending outpatient clinic appointments. Although the majority (87.9%) of patients died during the 2-year follow-up period, a minority (16.2% and 23.1%) utilized palliative care or hospice services, respectively. Within 30 days of death, 84.5% of patients were hospitalized, with 61.0% dying in the hospital. Among patients who died, those treated with intensive induction (versus non-intensive therapy) spent 30% more of their life in the hospital (p < 0.0001), and were less likely to utilize hospice services (OR 0.45, P = 0.05). Conclusions These findings highlight the intensity of health care utilization of older patients with AML, regardless of treatment modality. Despite the poor prognosis, palliative care and hospice services are rarely used. Future work should study novel health-care delivery models to optimize care throughout the course of illness and at the EOL.