Search

To determine the utility (i.e., positive predictive value [PPV] and time requirement) of an adverse drug event (ADE) trigger tool in Veterans Affairs (VA) nursing facilities and to describe the most common types of potential ADEs detected with the trigger tool.Retrospective chart review.Veterans residing in three VA nursing facilities between September 29, 2010, and October 29, 2010.We used the Institute for Healthcare Improvement-endorsed nursing facility ADE trigger tool, modified to enhance its clinical relevance to detect potential ADEs. Electronic medical records were screened to identify residents with one or more abnormal laboratory values specified in the trigger tool.A potential ADE was defined as the concurrent administration of medication that could cause the abnormal laboratory value. An overall PPV, or proportion of residents with an abnormal laboratory value who had a potential ADE, and average time required to complete each trigger tool assessment, were calculated.Among 321 veterans, 50.5% (n = 162) had at least one abnormal laboratory value contained in the trigger tool. Ninety-nine potential ADEs involving 146 medications were detected in 65 veterans. The overall PPV of the ADE trigger tool was 40.1% (65/162), and the average time to complete resident assessments was 8.8 (standard deviation ± 5.7) minutes. The most common potential ADEs were acute kidney injury (n = 30 residents), hypokalemia (n = 18), hypoglycemia (n = 13), and hyperkalemia (n = 10).The modified nursing facility trigger tool was shown to be an effective and efficient method for detecting potential ADEs.

Objectives The objectives of this study were to describe changes in glyburide prescribing in cohorts that were and were not targeted by a risk reduction project, assess factors associated with glyburide discontinuation, and evaluate changes in glycated hemoglobin (ie, HbA 1c ) levels and rates of serious hypoglycemia. Methods This historical cohort study included a targeted cohort of 4368 outpatient veterans aged ≥65 years with active prescriptions for glyburide between April 1, 2007 and June 30, 2007 and serum creatinine (SCr) ≥2 mg/dL and a nontargeted cohort of 1886 outpatients meeting these same criteria between July 1, 2007 and September 3, 2007. The intervention in the risk reduction project took place on September 4, 2007 and entailed giving regional pharmacy leaders information about the increased risk of hypoglycemia with glyburide and the list of targeted patients for follow up with providers. For each patient, the study period was the time between the date they first met the eligibility criteria and March 31, 2008. All data were obtained from Veterans Affairs (VA) administrative databases. The primary outcome was the discontinuation of glyburide. Secondary outcomes were the change in HbA 1c after stopping glyburide and the rate of serious hypoglycemia after intervention. Results Incidence rate ratios (IRRs) for glyburide discontinuation in targeted versus nontargeted cohorts were statistically significantly elevated in September (IRR 2.1; 95% CI 1.7–2.5), October (IRR 1.3; 95% CI 1.1–1.6), and November 2007 (IRR 1.4; 95% CI 1.1–1.7). The intervention, black race, SCr, Charlson comorbidity score, new glyburide use, and VA region were independently associated with discontinuation. Among patients in the targeted cohort who discontinued glyburide, mean (SD) HbA 1c at baseline and after discontinuation were 7.17% (1.35%), and 7.22% (1.34%), respectively ( P = 0.36). The hypoglycemia rates/1000 person-days were 0.093 before the intervention and 0.070 afterwards ( P = 0.10). Conclusion A one-time intervention in a risk reduction project decreased glyburide use over a 3-month period in elderly outpatients with renal insufficiency without compromising glucose control.

Objective To compare the cost-effectiveness of four, six, and eight doses per month of vardenafil in the context of pharmacy benefit decision making. Methods A Markov model was used to estimate the incremental cost-effectiveness of zero, four, six, or eight doses of vardenafil per month in hypothetical cohorts of 60-year-old male veterans with erectile dysfunction. Efficacy values for vardenafil were obtained from the literature, and vardenafil costs were obtained from Veterans Affairs pharmacy data. The analysis was conducted from a third-party payer perspective with a lifetime horizon, and the effect of parameter uncertainty was explored in one-way and probabilistic sensitivity analyses. Results In the base case analysis, the cost per quality-adjusted life-year gained for four doses of vardenafil per month compared with no therapy was $576. Six doses per month compared with four cost $2585/quality-adjusted life-year gained, and eight doses per month compared with six cost $5169/quality-adjusted life-year gained. In one-way sensitivity analyses of six doses per month compared with four, variation of two parameters caused the incremental cost-effectiveness ratio to cross a willingness-to-pay threshold of $20,000: when the increased utility associated with giving two additional doses/month was less than 0.001 (baseline 0.01) and when the cost per dose increased to $15.00 (baseline $1.69). Conclusion Although four doses per month of vardenafil was the most cost-effective strategy, the use of six or eight doses per month also compares favorably with other accepted medical treatments. The results were stable across a range of inputs and help to support the current Veterans Affairs policy on the number of vardenafil doses provided per month for erectile dysfunction.

Article-at-a-Glance Background A mixed-methods quality improvement (QI) project for patients with pneumonia hospitalized from the emergency department (ED) was undertaken to (1) delineate the basic steps in the flow of patient care from presentation in the ED to time to first antibiotic dose (TFAD), (2) identify perceived barriers to and facilitators of reduced TFAD within these steps, (3) describe QI strategies to improve TFAD rates, and (4) identify perceived strategies for facilities to enhance performance. Methods The QI project was conducted at 10 lower- and 10 higher-performing Veterans Affairs hospitals on the basis of the proportion of patients whose TFAD was within four hours of presentation. An ED physician, an ED nurse, a radiologist, a pharmacist, and a quality manager from each site were invited to participate in a survey and focus group. Results Of the 82 survey participants, 59 (72%) perceived that ordering and performing the chest radiograph was the step most frequently resulting in TFAD delays. Medical provider assessment, chest radiograph interpretation, ordering/obtaining blood cultures, and ordering/administering initial antibiotic therapy also caused TFAD delays. The most commonly perceived barriers were patient and x-ray equipment transportation delays and communication delays between providers. The most frequently used strategies to reduce TFAD were stocking antibiotics in the ED and physician education. Focus groups emphasized the importance of multifaceted QI approaches and a top-down hospital leadership style to improve TFAD performance. Discussion TFAD relies on a series of complex, stepwise processes of care that involve numerous hospital departments and is often delayed by well-described barriers. Addressing these barriers, as well as involving facility leadership in setting institutional QI goals, could possibly improve performance on this pneumonia quality measure.

To compare users and nonusers of Veterans Affairs (VA) pharmacy services by age group.Cross-sectional.We used data on sociodemographics, health status, and medical conditions from the Medical Expenditure Panel Survey (MEPS) to compare users and nonusers of VA pharmacies for medications. Data were pooled for 2003-2005 to ensure adequate sample sizes. Student t tests were used to compare the means for each variable, and all analyses were adjusted for the complex sample design of the MEPS.Among both nonelderly (18-64 years) and elderly (or=65 years) veterans, a higher proportion who used VA pharmacy services versus those who did not use VA pharmacy services (1) were black (nonelderly: 17.7 % vs 7.4%, P.001; elderly: 9.4% vs 4.7%, P.001); (2) had no alternative insurance (nonelderly: 27.2% vs 4.8%, P.001; elderly: 36.3% vs 19.9%, P.001); (3) had lower incomes (nonelderly: 32.4% vs 11.5%, P.001; elderly: 32.4% vs 25.4%, P = .01); (4) had less than a high school education (nonelderly: 13.0% vs 6.5%, P.001; elderly: 27.5% vs 17.6%, P.001); (5) were disabled; and (6) reported poorer health. A higher percentage of nonelderly users reported a mental health condition (31.6% vs 19.4%, P.001).Veterans who use VA pharmacy services appear to be more ill than those who do not use VA pharmacy services. In addition, the VA appears to be a safety net for uninsured veterans who have mental health problems.

Background. Although gatifloxacin is no longer available, other fluoroquinolones may significantly interfere with glucose homeostasis. The objective of the present study was to compare the risk of severe hypo- and hyperglycemia in a cohort of patients treated with gatifloxacin, levofloxacin, ciprofloxacin, or azithromycin. Methods. This was a retrospective inception cohort study of outpatients with a new prescription for gatifloxacin, levofloxacin, ciprofloxacin, or azithromycin from 1 October 2000 through 30 September 2005 in the Veterans Affairs health care system. For patients who received one of these antibiotics, we identified outcomes of hospitalization with a primary diagnosis of hypo- or hyperglycemia. Multivariable logistic regression was used to determine the odds of hypo- and hyperglycemia with the individual fluoroquinolones versus azithromycin. Results. The crude incidence rates for severe hypo- and hyperglycemia among those who received gatifloxacin, levofloxacin, ciprofloxacin, and azithromycin were 0.35 and 0.45, 0.19 and 0.18, 0.10 and 0.12, and 0.07 and 0.10 cases per 1000 patients, respectively. Among patients with diabetes, the odds ratios for hypoglycemia compared with azithromycin were 4.3 (95% confidence interval [CI], 2.7‐6.6) for gatifloxacin, 2.1 (95% CI, 1.4‐3.3) for levofloxacin, and 1.1 (95% CI, 0.6‐2.0) for ciprofloxacin. The odds ratios for hyperglycemia were 4.5 (95% CI, 3.0‐6.9) for gatifloxacin, 1.8 (95% CI, 1.2‐2.7) for levofloxacin, and 1.0 (95% CI, 0.6‐1.8) for ciprofloxacin. Conclusions. The odds of severe hypo- and hyperglycemia were significantly greater with gatifloxacin and levofloxacin, but not ciprofloxacin, than with azithromycin. Thus, the risk of a clinically relevant dysglycemic event appears to vary among the fluoroquinolones. The fluoroquinolone antibiotics are widely used because they are generally well tolerated and have a broad spectrum of activity. However, gatifloxacin—which had been used in the treatment of a variety of infections, including those involving the skin and the respiratory and genitourinary tracts [1, 2]—was withdrawn from the market because of concerns about severe glucose disturbances. Although hypoglycemia and hyperglycemia were reported only rarely during clinical trials with gatifloxacin, severe dysglycemic events, typically in elderly patients and in patients with diabetes, were re

The objective of the study was to identify patient and provider factors associated with prescribing antibiotics for outpatients with acute respiratory tract infections of likely nonbacterial etiology.We identified outpatients who were diagnosed in the emergency department with nonspecific upper respiratory tract infections (URIs) and acute bronchitis at the VA Pittsburgh Healthcare System from June 15, 2003, to June 14, 2004, and the Philadelphia VA Medical Center from November 30, 2003, to March 31, 2004. Stepwise logistic regression was used to identify factors independently associated with antibiotic prescribing.Overall, 26% of the 667 eligible patients with URIs and/or acute bronchitis received antibiotics. Antibiotics were prescribed significantly more frequently for acute bronchitis at one site (97% vs 65%, P.001). Using multivariable analysis, the following factors were independently associated with antibiotic prescribing (odds ratio, 95% confidence interval): presence of 1 or more comorbidities (2.1, 1.2-3.5), fever (2.5, 1.4-4.4), purulent sputum (2.5, 1.5-4.4), shortness of breath (2.8, 1.4-5.4), altered breath sounds (4.6, 2.4-8.6), diagnosis of acute bronchitis (15.9, 8.0-31.8), provider ageor = 30 years (2.6, 1.1-6.3), and noninternal medicine specialty (2.7, 1.2-6.0).Antibiotic use was high and varied substantially for URIs and acute bronchitis. Specific signs and symptoms, a diagnosis of acute bronchitis, and provider age and specialty were associated with antibiotic prescribing. Interventions to decrease inappropriate prescribing should address the perceived utility of antibiotics in acute bronchitis and the accuracy of signs and symptoms in diagnosing a bacterial respiratory infection.

Objective: This paper reviews recent articles examining medication errors in the elderly. Methods: MEDLINE and International Pharmaceutical Abstracts were searched for articles published in 2006 using a combination of the terms medication errors, medication adherence, medication compliance, suboptimal prescribing , and aged . A manual search of the reference lists of the identified articles and the authors' article files, book chapters, and recent reviews was conducted to identify additional publications. Those studies that described new measures of medication errors or had a randomized controlled design, evaluated the impact of an intervention on ≥1 measure of medication errors, and involved the community-dwelling elderly were included in the review. Results: The search identified 5 studies and a new set of explicit criteria for prescribing problems in nursing homes from the Centers for Medicare and Medicaid Services (CMS). One of the studies found a new instrument, the Medication Management Instrument for Deficiencies in the Elderly, to be a reliable and valid measure of medication management in older adults. A study in the ambulatory elderly found that 13.0% reported cost-related medication nonadherence. A randomized controlled trial of a pharmacist intervention in elderly patients at high risk for coronary events found the intervention was associated with improvements in both medication adherence and systolic blood pressure control. The report from the CMS described new explicit criteria for unnecessary drug use in elderly patients in long-term care facilities, including drugs to avoid, drugs that should be limited in dose or duration, drugs to be monitored, and drug-drug interactions. A modified Delphi survey of an expert panel reached consensus on 28 drug-disease interactions in older adults. Finally, a randomized controlled trial of computerized feedback in a health maintenance organization found improvements in inappropriate prescribing of target drugs in older adults. Conclusion: Data from recently published studies may provide guidance to practitioners and help direct future research.

To examine factors associated with fluoroquinolone prescribing among adults receiving antibiotics for acute respiratory tract infections (ARIs) in emergency departments.Cross-sectional.We analyzed data from 8 Department of Veterans Affairs medical centers and 7 nonfederal US hospitals. At each hospital, we randomly sampled 200 ARI visits with International Classification of Diseases, Ninth Revision discharge diagnoses for nonspecific upper respiratory infections, acute bronchitis, pharyngitis, sinusitis, and pneumonia between November 1, 2003, and February 29, 2004. Patient and provider factors associated with each visit were extracted from medical records. System characteristics were obtained by surveying pharmacy directors. Multivariable logistic regression was used to evaluate independent predictors of fluoroquinolone prescribing.Fluoroquinolones accounted for 14% of these prescriptions. At hospitals with at least 1 unrestricted fluoroquinolone on formulary (n = 12), the average fluoroquinolone prescription rate was 17%, compared with a 6% prescription rate at hospitals where fluoroquinolone access was restricted by the hospital formulary (n = 3) (P.0001). Factors associated with increased fluoroquinolone prescription rates were hospital admission (odds ratio [OR] = 1.8; 95% confidence interval [CI] = 1.1, 3.1) and the diagnoses of acute bronchitis (OR = 2.3; 95% CI = 1.3, 4.2), acute exacerbations of chronic bronchitis (OR = 2.6; 95% CI = 1.2, 5.6), and pneumonia (OR = 6.4; 95% CI = 3.3, 12.4). Restricted hospital status was associated with decreased fluoroquinolones accounted for 14% of the antibiotic prescriptions.Hospital formulary policies represent a potentially important target for influencing outpatient drug prescribing in emergency departments.

The Outpatient Bleeding Risk Index (BRI) prospectively classified patients who were at high, intermediate, or low risk for warfarin-related major bleeding. However, there are only 2 published validation studies of the index and neither included veterans.To determine the accuracy of the BRI in patients attending a Veterans Affairs (VA) anticoagulation clinic and to specifically evaluate the accuracy of the BRI in patients with atrial fibrillation.Retrospective cohort study.Using the BRI, all patients managed by the Anticoagulation Clinic between January 1, 2001 and December 31, 2002 were classified as high, intermediate, or low risk for major bleeding. Bleeds were identified via quality-assurance reports. Poisson regression was used to determine whether there was an association between the index and the development of bleeding.The rate of major bleeding was 10.6%, 2.5%, and 0.8% per patient-year of warfarin in the high-, intermediate-, and low-risk groups, respectively. Patients in the high-risk category had 14 times the rate of major bleeding of those in the low-risk group (incidence rate ratio (IRR) 14; 95% confidence interval (CI), 1.9 to 104.7). The rate of major bleeding was significantly different between the high- and intermediate-risk categories (P.001). Among those with atrial fibrillation, patients in the high-risk category had 6 times the major bleeding rate of those in the intermediate- and low-risk groups combined (IRR=6; 95% CI, 2.4 to 15.3).The BRI discriminates between high- and intermediate-risk patients in a VA anticoagulation clinic, including those with atrial fibrillation.

Programs have targeted individual patient and physician behaviors to reduce the use of antibiotics for upper respiratory infections (URIs), but such efforts have had limited success to date.The aim of this study was to measure the extent of variation in antibiotic prescribing patterns at the hospital-facility level to determine whether organizational factors may be associated with patterns of antibiotic prescribing.This was a cross-sectional study using linked pharmacy and encounter data to measure hospital-level variation in patterns of antibiotic prescribing at US Department of Veterans Affairs (VA) medical centers between October 1, 2000, and September 30, 2001. The main outcome measure was the proportion of visits for URIs or acute bronchitis with an antibiotic dispensed within 1 day before to 3 days after the encounter, restricted to primary-care and emergency/urgent care clinics at VA medical centers withor =100 annual visits for URIs.A median of 523 visits for URIs occurred across 108 medical centers. The median proportion of visits with an antibiotic dispensed was 52% (range, 14%-88%). Hospitals in the South had increased odds of prescribing antibiotics for veterans with URIs compared with hospitals in the Northeast (odds ratio, 1.8 [95% CI, 1.2-2.5]). Among facilities with200,000 visits per year, an increase in the percentage of unscheduled outpatient visits increased the odds of prescribing antibiotics for veterans with URIs (odds ratio per 10% increase, 1.3 [95% CI, 1.1-1.5]).Our results suggest variation in antibiotic prescribing for URIs at the hospital-facility level within the VA health care system. Organizational factors, such as time pressure, may be important targets for future interventions designed to reduce inappropriate antibiotic use in ambulatory care settings.

The possible reactivities of commonly used antibiotics of fungal, nonfungal, and nonmicrobial or synthetic sources with the Platelia Aspergillus galactomannan assay were assessed. For drugs that tested positive, the minimal concentration of the antibiotic in serum that yielded a positive test (index, >0.5) was determined. At undiluted concentrations, piperacillin and multiple lots of piperacillin-tazobactam tested positive, whereas amoxicillin, ampicillin-sulbactam, nafcillin, cefazolin, ceftazidime, erythromycin, gentamicin, and levofloxacin tested negative. All three lots of piperacillin-tazobactam and all bags within each lot tested positive, with a mean index value of 5.168. At achievable concentrations in serum, however, only one of three lots of piperacillin-tazobactam yielded a positive test. Concentrations of 75, 150, and 300 μg/ml of serum tested positive with the Platelia Aspergillus enzyme immunoassay, whereas lower concentrations, mimicking the trough levels, tested negative. Thus, while achievable serum piperacillin-tazobactam concentrations may potentially result in a positive test for galactomannan, the timing of the collection of serum samples from patients may influence the test results, with reactivity being less likely in samples collected at trough levels or prior to the administration of a dose of the antibiotic.

The detection of adverse drug reactions (ADRs) by a traditional passive reporting system and by a method involving patient and provider interviews was studied. The study sample consisted of randomly selected outpatients seen by their primary care provider during scheduled appointments in January and February 2001 at a Veterans Affairs medical center. After ambulatory care clinic sessions, patients and providers were asked (by telephone and in person, respectively) to identify potential ADRs. Also obtained were demographic data, information about drug regimens, and the severity and management of each ADR. A standardized ADR-assessment tool was used to determine the severity of each reported reaction and its causal relationship with the medication. A total of 198 patients were included. Of these, 51 (26%) had one or more ADRs. The patient and provider interviews identified a total of 83 ADRs, compared with 1 ADR identified by the passive reporting system. When providers were made aware of the ADRs they had not identified, changes were made to the patient's medication regimen in 34% of cases. The risk of an ADR was not associated with age, number of medications, or provider type. Direct patient and provider interviews yielded a significantly higher rate of ADR detection in an ambulatory care setting than did a passive ADR-reporting system.

OBJECTIVE: To report three cases of a suspected interaction between warfarin and fluvastatin. CASE SUMMARIES: Three patients receiving stable warfarin dosages with therapeutic international normalized ratios (INRs) exhibited increased INRs when fluvastatin was added to their maintenance regimens. While none of the patients experienced a bleeding episode, they did require a reduction in their weekly warfarin dosage to achieve an appropriate level of anticoagulation. DISCUSSION: Reports of an interaction between warfarin and lovastatin have been described previously; however, to our knowledge, this is the first published report of a possible interaction between warfarin and fluvastatin. These cases were chosen because other factors that could potentially increase the INR were ruled out as significant contributors. CONCLUSIONS: The exact mechanism for the potential interaction between warfarin and fluvastatin is unknown. Until more is known, it is advisable to monitor patients more frequently when fluvastatin is initiated, discontinued, or adjusted in patients taking warfarin.

Objective: To report on a patient with recurrent methicillin-resistant Staphylococcus aureus (MRSA) osteomyelitis and bacteremia successfully treated with combination antibiotic therapy. Case Summary: Two sets of blood cultures from a 55-year-old man with fever, malaise, and low back pain grew MRSA. Radiologic studies of the spine showed bony changes consistent with osteomyelitis. Soon after completing 6 weeks of vancomycin, the patient experienced a recurrence of back pain. Laboratory values included an increase in the sedimentation rate to 53 mm/h and positive blood cultures for MRSA. Vancomycin, gentamicin, and rifampin were administered for 8 weeks. Serum inhibitory and bactericidal titers were more than 1:1024 for both the peak and trough concentrations. Radiologic studies of the spine showed healing osteomyelitis. Two years after completion of antibiotic therapy, the infection has not recurred. Discussion: Antibiotic therapy alone was attempted because the patient was considered a risky surgical candidate. Serum inhibitory and bactericidal titers documented the high in vivo activity of the vancomycin, gentamicin, and rifampin combination. Initiation of vancomycin therapy led to disappearance of the fever and back pain. Cure was documented by sustained normalization of the erythrocyte sedimentation rate and radiologic evidence of healing. Conclusions: Combination antibiotic therapy with vancomycin, rifampin, and low-dose gentamicin (1 mg/kg q12h) may be useful for deep-seated tissue infection caused by MRSA.

The Veterans Health Administration (VHA) runs the largest integrated healthcare system in the nation. Formulary management within VHA primarily involves 3 national groups: the Medical Advisory Panel, the Veterans Integrated Service Network Formulary Leaders, and the Pharmacy Benefits Management Strategic Healthcare Group. Together, these groups manage the VHA national drug formulary with a goal of providing a comprehensive, safe, and cost-effective pharmacy benefit for veterans. Traditionally, VHA has relied on cost-minimization analyses in formulary decisions. More recently, VHA has emphasized the use of cost-effectiveness data, especially for newer, costly drugs. In addition to including this data in drug monographs, the VHA has begun requiring formal cost-effectiveness analysis from manufacturers of selected pharmaceuticals. VHA has also requested that clinically relevant information such as quality of life plus mortality benefit be made available from industry so that internal cost analyses can be performed. It is hoped that by setting the expectation that cost-effectiveness will be formally considered in all VHA formulary decisions, the pharmaceutical industry and others will be stimulated to collect and report data that enables these analyses. We believe that if other organizations also place an emphasis on economic evaluations, industry and the public will be more accepting of decisions that incorporate cost considerations.

Background Given the paucity of information on dose intensity, the objective of this study is to describe the use of adjuvant chemotherapy for stage III colon cancer, focusing on relative dose intensity (RDI), overall survival (OS) and disease-free survival (DFS). Methods Retrospective cohort of 367 patients diagnosed with stage III colon cancer in 2003–2008 and treated at 19 VA medical centers. Kaplan-Meier curves summarize 5-year OS and 3-year DFS by chemotherapy regimen and RDI, and multivariable Cox proportional hazards regression was used to model these associations. Results 5-fluorouracil/leucovorin (FU/LV) was the most commonly initiated regimen in 2003 (94.4%) and 2004 (62.7%); in 2005–2008, a majority of patients (60%-74%) was started on an oxaliplatin-based regimen. Median RDI was 82.3%. Receipt of >70% RDI was associated with better 5-year OS (p 70% RDI at the first year (HR = 0.58; 95% CI = 0.37-0.89) and married status (HR = 0.66; 95% CI = 0.45-0.97) were associated with significantly decreased risk of death, while age ≥75 (versus 55–64) (HR = 2.06; 95% CI = 1.25-3.40), Charlson Comorbidity Index (HR = 1.17; 95% CI = 1.06-1.30), T4 tumor status (versus T1/T2) (HR = 5.88; 95% CI = 2.69-12.9), N2 node status (HR = 1.68; 95% CI = 1.12-2.50) and bowel obstruction (HR = 2.32, 95% CI = 1.36-3.95) were associated with significantly increased risk. Similar associations were observed for DFS. Conclusion Patients with stage III colon cancer who received >70% RDI had improved 5-year OS. The association between RDI and survival needs to be examined in studies of adjuvant chemotherapy for colon cancer outside of the VA. Electronic supplementary material The online version of this article (doi:10.1186/s12885-015-1038-y) contains supplementary material, which is available to authorized users.