Pneumothorax Prevalence and Mortality per Gestational Age in the Newborn

Objectives: Pneumothorax (PTX) in newborns is a life-threatening condition associated with high morbidity and mortality especially in premature infants. The frequency of PTX in neonates at different gestational ages (GA) and its impact on neonatal mortality have not been quantified. We aimed to determine: 1) the prevalence of PTX in neonates at different GA from ≤24 weeks to ≥37 weeks, 2) the impact of PTX on mortality per GA, and 3) the impact of PTX on the length of stay (LOS) per GA. Methods: The national Kids’ Inpatient Database (KID) for the years of 2006 to 2012 were used. We included all infants admitted to the hospital with a documented GA and ICD9 code of pneumothorax. Bivariate and multivariate analyses were conducted and odds ratios (OR) were calculated. Results: A total of 10 625 036 infants were included; of them 3665 infants (0.034 %) had a diagnosis of PTX, with highest prevalence at ≤24 weeks GA (0.67%), and lowest at term (0.02%). The overall mortality rate of patients with PTX was 8.8%, and greater in preterm (16.3%) vs. term infants (2.7%). The association of mortality with PTX was greatest at GA of 29−32 weeks (OR = 8.55 (95% CI: 6.56−11.13). Infants who survived until discharge had a median of 2–12 days longer length of stay depending on GA category. Conclusions: The prevalence of PTX peaks in infants <24 weeks, however its impact on mortality is greatest at 29-32 weeks. PTX is associated with longer length of stay in survivors.

Fungal Infections in Chronic Lymphocytic Leukemia before and after Treatment

Background : Patients with Chronic Lymphocytic Leukemia (CLL) are susceptible to infections due to impaired immunity, from both complications of disease and treatments. Specific treatments such as fludarabine or rituximab are incorporated in chemotherapy regimens for CLL and cause lymphocyte depletion and impaired humoral immunity, resulting in increased risk of fungal infections. The aim of this study is to explore the incidence rates of fungal infections in patients before and after treatment for CLL. Methods : We identified patients diagnosed with CLL using ICD9 code 204.1X within the Veterans Health Affairs (VHA) between 1999 and 2013, before the availability of novel agents for CLL. Pharmacy records were used to identify treatment of CLL (alemtuzumab, bendamustine, chlorambucil, fludarabine, ofatumumab, pentostatin, and rituximab) as well as anti-fungal agents (amphotericin, caspofungin,fluconazole, flucytosine, itraconazole, ketoconazole, micafungin, posaconazole, and voriconazole). Fungal infections were identified with two strategies, using ICD9 codes alone (candida 112.X; pneumocystis 136.3; coccidiomycosis 114.x; histoplasmosis 115.X; blastomycosis 116.X; aspergillus 117.3; cryptococcus 117.5; opportunistic mycoses 118.X) as well as using ICD9 codes combined with evidence of treatment with an anti-fungal (ICD9+antifungal). Number of infections and incidence rates were categorized as occurring in patients prior to first treatment (pre-treat), or after first treatment (post-treat), or in patients who had never received treatment within the VHA (no-treat). Incidence rate ratio between pre- and post-treatment period was calculated to quantify the magnitude of risk increase in patients who received CLL treatment. Results : A cohort of 20,023 patients were identified with CLL and followed for a mean of 5.33 years for a total surveillance of 106,772 person-years. Of these patients, 3,726 (18.6%) received treatment for CLL within the VHA. Using ICD9 codes alone, 970 fungal infections were identified and with ICD9+anti-fungal, 415 patients had an infection. With ICD9, the most common infection was candida with 765 infections, followed by aspergillus with 58 and pneumocystis with 39 infections. Rates of infections based on ICD9 were highest in the post-treat group with 2.98 infections per 100 person-years, followed by 1.08 infections per 100 person-years in the pre-treat group and 0.58 infections per 100 person-years in the no-treat group. The most common first treatment was rituximab alone, used in 35.9% of patients, followed by chlorambucil in 31.9%, fludarabine in 20.4%, and bendamustine in 9.9%. In comparison to pre-treat patients, post-treat patients had 2.91 times increased risk of fungal infection (95% confidence interval (CI) 2.24-3.81) determined by ICD9 code and 4.74 times increased risk (95% CI 3.09-7.57) determined by ICD9+antifungal. The rate of candida infection increased 2.9 fold (95% CI 2.24-3.81) with ICD9 and 4.7 fold (95% CI 3.09-7.57) with ICD9+antifungal while aspergillus infections increased 8.7 fold (95% CI 3.01-35.4) with ICD9 and 10.2 fold (95% CI 2.88-63.1) with ICD9+antifungal between post-treat and pre-treat groups. Conclusions: In this retrospective analysis of CLL patients, treatment for CLL significantly increased the rate of fungal infections, primarily candida and aspergillus. Further study is needed to understand the effect of modern treatments on fungal infections in CLL. Disclosures Schoen: Pharmacyclics: Research Funding. Georgantopoulos:Pharmacyclics: Research Funding. Bennett:Pharmacyclics: Research Funding.

Association of changes in functional status with radiation for prostate cancer in older veterans.

148 Background: Although radiation therapy (RT) for prostate cancer is generally well tolerated, frail older adults with less functional reserve may experience more toxicity and loss of independence. Our objective was to determine trajectories of activities of daily living (ADL) for prostate cancer patients admitted to a nursing home for RT. Methods: We used the Veterans Affairs (VA) Minimum Data Set (MDS) to identify men age ≥65 with an ICD9 diagnosis of prostate cancer living in a VA nursing facility (CLC) 1/2005-12/2015, an MDS evaluation reporting RT and no ICD9 code for bone metastasis. Functional status was assessed using MDS-ADL score (range 0-28, higher scores = greater disability). A piecewise linear mixed effects model (nodes at months 1 and 3) modeled the relationship between baseline characteristics and MDS-ADL score. Results: 645 patients were identified, of whom 585 (90.7%) had not resided in a CLC prior to RT. Median age 74 (range 65-94), median baseline PSA 5.33 ng/mL (IQR 1-14.57), and median Charlson Comorbidity Index (CCI) 5 (30.5% CCl ≥8). Baseline median MDS-ADL score was 1 (range 0-28). Patients with CCI 2-3 did not have appreciable change in functional status in 6 months following start of RT, while patients with CCI 4-7 had an increase months 1-3, followed by improvement (Table). Compared to patients with CCI 2-3, those with CCl ≥8 had an increase in MDS-ADL score of 1.8 points/month months 1-3 after starting RT (p = 0.008), and increase in MDS-ADL score of 3 points/month from 3 months onward (p < 0.001). Older age and higher CCI were associated with increase in MDS-ADL score (p < 0.05). Conclusions: In a cohort of elderly prostate cancer patients with significant comorbidity, RT led to different functional trajectories depending on comorbidity burden. While patients with moderate comorbidity had an initial decline in functional status followed by improvement, patients with high comorbidity had continued functional decline. [Table: see text]

Anatomically Associated Venous Thromboembolism in Patients With Peripherally Inserted Central Catheters

Background: Peripherally inserted central catheters (PICCs) are ubiquitous in modern hospitals, but are associated with venous thromboembolism (VTE), which includes deep vein thrombosis (DVT) and/or pulmonary embolism (PE). We retrospectively examined this association in hospitalized patients, highlighting anatomically associated VTEs (those with DVT in the PICC extremity). Methods: Charts with an International Classification of Diseases, Ninth Revision (ICD9) code for VTE were collected from a discharge database of PICC-managed patients at a tertiary hospital. A sample (52.3%) of the VTE charts was manually reviewed to verify PICC-associated VTE (unverified charts were excluded), and determine such data as the extremity in which each DVT was diagnosed (using ultrasound reports). VTE rates were calculated using an uncorrected method (from charts with VTE ICD9 code) and a corrected method (from charts with manually verified PICC-associated VTE). Results: Our uncorrected VTE rate was 3.9% (P &lt; .0001), whereas the corrected rate was 1.5%. Among 125 charts with manually verified PICC-associated VTE, 69 (60.5%) out of 114 patients with a DVT had their DVT occur in the PICC extremity, yielding an anatomically associated VTE rate of 0.84%. The most common reason for a chart being excluded (60.2%) was a VTE occurring before PICC placement. Conclusions: We found clinically significant rates of PICC-associated VTE. The majority of patients' DVT occurred in the same extremity as their PICC, lending further evidence that PICCs are an independent risk factor for VTE and require judicious use. There was also a discrepancy in VTE rate derived from ICD codes alone vs. manual chart review.

Abstract 251: Long-term Outcomes and Implantable Cardioverter Defibrillator Use in Peripartum Cardiomyopathy

Background: Many studies of outcomes in peripartum cardiomyopathy (PPCM) are limited by short-term follow-up of 6-12 months. Long-term mortality rates and outcomes are largely unknown. Prior research has also suggested under-utilization of implantable-cardioverter defibrillators (ICD) in this population. Methods: Patients with PPCM at a tertiary care center were retrospectively identified using ICD9 code 674.5x from 2000-2011, and each chart was manually reviewed. Clinical, demographic, and echocardiographic data were reviewed, with follow-up through November 2016. Recovery was defined as ejection fraction (EF) >= 55%. Results: Of 60 patients, 35 were white (58.3%) and 16 were black (26.7%). Mean age at diagnosis was 29.6 years (range 18-44 years) and average follow-up was 5.8 years. Mean EF at time of diagnosis was 20.2 +/- 4%, at 6-months follow-up it was 39.6 +/- 4%, and at final follow-up was 38.7 +/- 3%. Recovery (EF>=55%) occurred in 38%. Of those who did not recover (62%), the final mean EF was 25%. ICD was placed in 19 patients (51%). Only one patient with an ICD later recovered at one year follow-up. LVAD or transplant occurred in 8 patients, and death occurred in 7 patients with a mean survival of 4.4 +/- 1.4 years. Five patients experienced subsequent deterioration in their EF (with a range of 10-35%) in the absence of a subsequent pregnancy, with a mean decrease in EF of 23.2%. Three of these patients had achieved full recovery (EF>=55%) prior to subsequent decline. Conclusions: Patients with PPCM who have initial improvement in EF or even full recovery, can later suffer deterioration, even in the absence of subsequent pregnancy. This finding warrants continued surveillance and raises the question whether lifelong medical therapy may be indicated despite recovery. This study also shows a higher rate of ICD utilization than prior reports.

Abstract 309: Trends in LVAD Use and Outcomes Among Patients Admitted with Acute Myocardial Infarction

Background: Mortality rates among patients with acute myocardial infarction (AMI) complicated by cardiogenic shock remain high. Implantation of left ventricular assist devices (LVAD) has become increasingly available since the approval of continuous flow devices in 2008 and in severe cases, may be used to prolong survival post AMI. Little is known about how the frequency of LVAD implantation and subsequent outcomes in AMI patients have changed over time. Methods: We used the National Inpatient Sample, a 20% stratified sample of all hospital discharges that uses scaled weights to approximate national estimates. We identified all patients with AMI (ICD9 code 410.1x) and LVAD implantation (ICD9 code 37.66) from 2006 through 2012.The primary outcome was in-hospital mortality. Baseline characteristics were compared over time using the chi-square test for categorical variables. Univariate logistic regression was used to examine the association between baseline characteristics and risk of mortality after LVAD. Results: The number of LVADs implanted for any indication increased from 713 to 2,960 during the study period whereas LVAD use among AMI patients remained stable (Figure). AMI patients who received an LVAD were predominately male and white and the average age was 56.3 years. The number of AMI patients receiving ECMO, Impella, or other short-term mechanical support devices as a bridge to LVAD increased over time whereas IABP use remained stable. Among patient and hospital factors studies, non-white race and later year of implantation were associated with lower mortality after LVAD. Use of other mechanical support devices was associated with higher mortality (OR 2.7, p=0.029). Post-LVAD mortality rates were higher for AMI compared to non-AMI patients but decreased for all patients over time: 57.1% to 21.2% for AMI patients (p <.0001) and 36.8% to 12.8% for patients without AMI, p < .0001). Conclusion: Among patients with AMI, LVAD use remains low and has not increased as has LVAD use for other indications. Although LVAD use in this population was initially associated with higher in-hospital mortality, our analysis suggests a narrowing of this gap. Future studies are needed to determine how long-term survival is affected and which patients are appropriate candidates for LVAD implantation after AMI.

Abstract 152: Daily Dietary Sodium Intake and Advice on Dietary Sodium among Heart Failure Patients in the Community

Objective: Little is known about the sodium intake of heart failure (HF) patients in the community. We used data from the Measuring Disparities in Chronic Conditions Study to examine the receipt of advice on dietary sodium and its relationship with sodium intake among adults in a large urban county. Method: We recruited adults in King County, WA using a home-address based sample as well as sampling from health facilities using medical record billing codes for key cardiovascular conditions. Survey questions addressed past medical history as well as receipt of advice and behavior related to dietary sodium. Diet history was assessed using the National Cancer Institute Diet History Questionnaire II (DHQ) for intake in the past year. We also compared responses and estimated sodium intake for those who did and did not report a diagnosis of HF. As a sensitivity analysis, we repeated our analysis using the subset of individuals who had a medical record with an ICD9 code for HF in the previous two years. Result: Our results are based on 3357 respondents from the address-based sample and 3477 from the medical record-based sample. The DHQ was completed by 940 and 1291 of these respondents, respectively. HF was reported by 290 respondents (median age 66 years, 57% male, 6.6% black race, 85% from medical record sample, 43% with a DHQ). Among those reporting a diagnosis of HF, 61% (112 of 183) reported being advised to decrease dietary sodium and 37% (67 of 183) reported being told to buy food items labeled as low salt. These rates were 30% (755 of 2515) and 13% (337 of 2515) respectively among individuals without reported HF. Estimated daily dietary sodium intake over the prior year was lower among those with HF completing a DHQ if they reported receiving counseling (mean 2.5 vs 2.8 gm if advised to cut down on salt and 2.3 vs 2.8 gm if advised to buy low salt labeled products). However, these differences were not statistically significant. Only 65% of respondents with HF (183 of 282) said they buy low salt labeled foods and 26% (61 of 290) reported eating processed, fast, or canned food at least 5 times in the prior month. Overall mean estimated daily dietary sodium intake over the prior year was 2.7 gm for respondents both with and without HF. Results were similar for those with an ICD9 code for HF. Conclusion: Our findings revealed that a large percentage of HF patients report no receipt of medical advice to reduce dietary sodium or purchase foods labeled as low salt. Moreover, the dietary intake of salt was similar for HF patients and the general population. These results point to a need for dietary interventions in HF that are both widely adopted and can actually change behavior.

Provider mix and costs associated with treatment of patients with bone metastases secondary to prostate cancer.

e15175 Background: Treatment of patients with bone metastases secondary to prostate cancer can involve several provider types and combinations of chemotherapy, surgery, radiation, and pharmaceutical treatment. This study evaluated the combinations of provider types and associated treatment patterns for a cohort of patients with bone metastases secondary to prostate cancer. Methods: Continuously enrolled patients older than 20 years of age in the MarketScan database between January 2004 and December 2010 with evidence of bone metastases (ICD9 code 198.5 or treatment with zolderonic acid, pamidronate, or demosumab) were included. Inpatient and outpatient medical claims data were used to define provider combinations. Treatment patterns were determined from prescription fill/refill claims and procedure codes from inpatient and outpatient medical claims. Results: A total of 4,493 patients had evidence of bone metastases. A radiologist was involved in care for a vast majority (n=4,054, 90%). Less than half of the population, (n=1,751, 39%) had an oncologist actively involved in care. Most patients (n=2633, 59%) had both an urologist and a radiologist involved in their care. The most common combinations of providers were urologist and radiologist (n=998, 22%); urologist, radiologist, and surgeon (n=951, 21%), and urologist, radiologist, and oncologist (n=781, 17%). About 15% (n=684) of patients had a surgeon, urologist, oncologist, and radiologist involved in their care. Only approximately half (n=2,274, 51%) of the population had evidence of receiving radiation therapy, suggesting that the radiologist plays a diagnostic role for many patients. A vast majority of patient were prescribed hormone therapy (89%) and 76% were prescribed steroid agents (mostly glucocorticoids). Less than half of the population (n=1,838, 41%) received surgery related to their prostate cancer. Conclusions: There is significant variation in combinations of provider types and associated treatment patterns for patients who have bone metastases secondary to prostate cancer. Follow-up studies should examine optimal conditions for different provider mixes and treatment patterns.

Abstract 3889: Diagnosis of Transient Ischemic Attack Varies by Neurologist

Introduction: The short duration of symptoms and multitude of non-vascular mimics make diagnosis of transient ischemic attack (TIA) subjective and challenging. Physician practice may influence whether a patient receives the diagnosis of TIA. Pathophysiology is the same for TIA and ischemic stroke (IS). Therefore, the proportion of TIA versus IS diagnoses should be the same across demographic and clinical strata and theoretically should not vary between physicians. Hypothesis: The TIA-IS ratio, defined as the proportion of TIA among all TIA and IS cases, varies between medicine and neurology services and between board-certified vascular and other neurologists. Methods: TIA and IS cases were identified using primary ICD9 discharge codes at Montefiore Medical Center from January 2009 through June 2011. Each TIA diagnosis was confirmed by chart review, excluding cases with acute IS on brain imaging, duration >24 hours or a clearly non-vascular syndrome. The TIA-IS ratio was assessed based on the primary ICD9 code and again on the diagnosis by chart review for discharges from the neurology versus the medicine service and also individually for each board-certified neurologist with more than 25 IS or TIA discharges during the study period. Mantel-Hantsel chi-square was used for univariate statistical comparison between groups. Results: There were 659 cases of TIA and 1927 cases of IS by primary ICD9 discharge code, for an overall TIA-IS ratio of 25.5 %. The ratio was higher for discharges from medicine (28.6%) than neurology services (23.8%, p=0.009). The ratio was higher for 12 non-vascular neurologists (27.0%, range 17.9-44.4%, 904 cases), than 5 vascular neurologists (20.0%, range 18.7-23.0, 715 cases, p=0.001). Chart review reduced the overall TIA-IS ratio to 16.3% and the difference between medicine and neurology discharges became non-significant but non-vascular neurologists still had a higher TIA-IS ratio (20.7%) than vascular neurologists (14.4%, p=0.001). Conclusions: The TIA-IS ratio is different on medicine and neurology services and varies widely among board-certified neurologists, with vascular neurologists less likely to assign the diagnosis of TIA than other neurologists even after excluding obvious non-TIA cases. The ABCD2 score, assessing patients’ symptoms and vascular risk factors, is promoted as a criterion for admission; however, these data suggest that physician training and perspective affect the diagnosis, and thus assessment of future stroke risk, as well.

Gender Disparities in Procedure Use for Acute Myocardial Infarction in the United States, 1995-1997

P03 Gender disparities in procedure use for acute myocardial infarction (AMI) have been well documented in selected populations in the 1980s and early 90s. However, little is known about recent trends in disparities in the general population. Therefore we conducted a series of cross-sectional analyses of data from the Nationwide Inpatient Sample for years 1995 through 1997 to compare rates of catheterization, angioplasty and coronary artery bypass grafting (CABG) performed prior to discharge for acute myocardial infarction in men vs. women. The NIS includes data (including demographics, diagnoses and procedures) on all discharges from over 900 representative civilian hospitals in 22 states. We identified 425,236 discharges with AMI (ICD9 code 410) as the first listed diagnosis during 1995-97. The cohort was 39% female (mean age 71.7) and 61% male (mean age 64.4). From 1995 to 97, catheterization rates increased in women (38.8% to 41.8%)and men (49.7% to 52.4%) as did rates of angioplasty (women 17.1% to 19.9%, men 23.0% to 26.6%) and CABG (women 8.3% to 8.7%, men 11.9% to 12.4%) After adjusting for age, race, and comorbidities, women were less than men to undergo catheterization in 1995 (odds ratio [OR] 0.85, 95% confidence interval [CI] 0.83-0.87), 1996 (OR 0.85, 95%CI 0.83-0.87), and 1997 (OR 0.87 95%CI 0.85-0.87). This gender disparity was even greater among adults aged 65 and older (adjusted OR 0.70, 95%CI 0.68-0.72 in 1995 and 96; and 0.71, 95%CI 0.69-0.72 in 1997). Women were also less likely to undergo CABG (adjusted OR 0.72, 95%CI 0.70-0.74 in 1995, 96 and 97). Among patients who underwent catheterization, women remained less likely to go on to CABG than men (adjusted OR 0.77, 95%CI 0.75-0.79). In contrast,women were only slightly less likely to undergo angioplasty (adjusted OR 0.93, 95%CI 0.91-0.94) and, once catheterized, were actually more likely to go on to angioplasty (adjusted OR 1.03, 95%CI 1.01-1.05). While procedure use for AMI is rising for men and women, these recent nationwide data suggest that women remain much less likely to undergo catheterization or CABG, but about equally likely to undergo angioplasty.