Abstract
Purpose
Most patients with cancer are not screened for hepatitis B virus (HBV) infection before undergoing anticancer therapy, and optimal screening strategies are unknown. We sought to develop selective HBV screening strategies for patients who require systemic anticancer therapy.
Methods
This prospective cohort study included adults age ≥ 18 years with solid or hematologic malignancies who received systemic anticancer therapy at a comprehensive cancer center during 2013 and 2014. Patients underwent hepatitis B surface antigen, hepatitis B core antibody, and hepatitis B surface antibody testing, and completed a 19-question modified Centers for Disease Control and Prevention (CDC) HBV survey. Multivariable models that predict chronic or past HBV infection were developed and validated using bootstrapping.
Results
A total of 2,124 patients (mean age, 58 ± 13 years) completed the risk survey and HBV testing. Of these, 54% were women; 77% were non-Hispanic white, 11% Hispanic, 8% black, and 4% Asian; and 20% had a hematologic malignancy and 80% a solid tumor. Almost 12% were born outside the United States. The prevalence was 0.3% for chronic HBV infection and 6% for past HBV infection. Significant predictors of positive hepatitis B surface antigen or hepatitis B core antibody tests were as follows: men who had sex with men, black or Asian race, birthplace outside the United States, parent’s birthplace outside the United States, household exposure to HBV, age ≥ 50 years, and history of injection drug use. The area under the receiver operating characteristic curve of the model on the basis of these seven predictors was 0.79 (95% CI, 0.73 to 0.82). The modified CDC survey and brief tools with fewer than seven questions yielded similar false-negative rates (0% and 0% to 0.7%, respectively).
Conclusion
An internally validated risk tool performed as well as the modified CDC survey; however, more than 90% of patients who completed the tool would still require HBV testing. Universal HBV testing is more efficient than risk-based screening.
INTRODUCTION
After immunosuppressive therapy, patients with hepatitis B virus (HBV) infection are at risk for reactivation of the HBV infection, which can lead to a flare of hepatitis with the associated risks of liver failure and death.1-5 Identification of patients with HBV infection at the onset of anticancer therapy is crucial to prevent HBV reactivation; however, despite the existence of well-established risk factors for HBV infection6 on the basis of demographics, sexual and lifestyle behaviors, and medical conditions, risk factors are seldom assessed because of time constraints. Recommendations to identify patients with HBV infection before anticancer therapy have varied from universal serologic testing to limited testing in populations that are at risk.1-4,7 At present, there is no consensus on how to accurately and efficiently identify persons who are at risk, which has resulted in suboptimal HBV testing among patients with cancer who are anticipated to receive anticancer therapy8,9 as well as preventable adverse liver outcomes.10 We sought to develop, validate, and compare selective HBV screening strategies on the basis of predictors of HBV infection for patients who require systemic anticancer therapy.
METHODS
We conducted a prospective, observational study of adults age ≥ 18 years with cancer who presented for their first outpatient systemic anticancer therapy appointment at The University of Texas MD Anderson Cancer Center from July 2013 through December 2014. Patients who were previously found to have positive hepatitis B surface antigen (HBsAg) or total immunoglobulin hepatitis B core antibody (anti-HBc) test results more than 3 months before their first outpatient anticancer therapy appointment were excluded, as were patients who were taking any of the following anti-HBV medications: adefovir, entecavir, interferon, lamivudine, telbivudine, or tenofovir disoproxil fumarate. Because of the high volume of eligible patients and logistical challenges of approaching patients receiving cancer therapy in three centers in two buildings, after the first 3 months of our study, we used a simple random sampling algorithm to identify 80% of eligible patients who received cancer therapy during an 11-hour period each weekday. We assessed the age, gender, race, and ethnicity of eligible patients who did not enroll. All patients provided written informed consent. This study was approved by the institutional review board of The University of Texas MD Anderson Cancer Center before data collection.
Participants completed a survey of HBV risk factors from a publicly available Centers for Disease Control and Prevention (CDC) hepatitis risk assessment11 that was consistent with national recommendations.1,6 The CDC hepatitis risk assessment included 18 questions about age and gender; place of birth; blood transfusion or organ transplantation before 1992 or clotting factor disorder; history of chronic liver disease, HIV or AIDS, or diabetes; sexual behaviors; injection drug use; household exposure to HBV infection; and other factors. We added a question about race and ethnicity for a total of 19 questions in our survey, hereafter referred to as the modified CDC survey. Patients self-reported their type of cancer.
Participants had the following HBV serologic tests performed within 3 months before or 2 months after study enrollment: HBsAg, total anti-HBc, and hepatitis B surface antibody (anti-HBs quantitative assay). Positive HBsAg and anti-HBc test results indicated chronic HBV infection. Negative HBsAg and positive anti-HBc test results, regardless of anti-HBs result, indicated past HBV infection. Positive anti-HBs test results indicated immunity from past HBV infection (in patients with positive anti-HBc results) or previous immunization (in patients with negative anti-HBc results). Per study protocol, patients with chronic or past HBV infection and hematologic malignancies were started on antiviral prophylaxis. Per usual care, patients with positive HBsAg or anti-HBc test results had HBV DNA tests performed using the COBAS AmpliPrep/COBAS TaqMan HBV test (version 2.0; Roche Molecular Systems, Pleasanton, CA).
Fisher’s exact and χ2 tests12 were used to compare patient characteristics by HBV infection status in univariable analyses. The initial multivariable logistic regression model was obtained by including an initial set of candidate predictor variables with P values < .20 from the univariable analysis. Clinical predictors of chronic or past HBV infection were chosen using stepwise backward elimination in which the P value required for a variable to enter the model was .10, and the P value for a variable to be retained in the model was .05. The goodness of fit of a selected model was assessed using the Hosmer-Lemeshow χ2 test. The discriminative capacity of the model was evaluated via receiver operating characteristic curve analysis and measured by the area under the receiver operating characteristic curve (AUC). Only patients who completed all questions were included in the modeling—complete-case analysis. To obtain reliable parameter estimates, we recoded gender into three groups as follows: female, men having sex with men, and men not having sex with men.
To address the issue of overfitting from using the same data set for fitting and predicting, we used bootstrapping,13 a data-based simulation method for statistical inference, with 500 bootstrap samples with replacement to obtain a bias-corrected estimate of the true AUC along with CIs.14 We also obtained the top five models by running stepwise logistic regressions independently on each bootstrap sample. From these top five models, we selected the model that we considered to be best on the basis of the highest AUC results of bootstrapping and clinical relevance. Screening strategies that had at least one affirmative answer to one of the predictors in the selected strategy or a subset of these predictors were compared with strategies that had at least one affirmative answer to an item on the modified CDC survey in terms of sensitivity and false-negative rate (FNR). To reduce the risk of HBV reactivation, we selected strategies with subsets of predictors with a low FNR.
RESULTS
A total of 4,131 patients were eligible. With the implementation of random sampling after month 3, we approached a total of 3,534 patients and enrolled 2,206 (62.4% of those approached and 53.4% of eligible patients). Of those enrolled (Fig 1), 67 patients did not complete serologic HBV testing, six did not complete the modified CDC survey, and nine withdrew, which left 2,124 patients in our final cohort. There was no difference in age (P = .57) or gender (P = .48) between our study cohort and nonenrollees; however, the proportions of black and Hispanic patients were lower in our study cohort—7.5% and 10.3%, respectively—than among nonenrollees—12.2% and 14.6%, respectively—and the proportions of Asian and non-Hispanic white patients were higher in our study cohort—3.2% and 77.1%, respectively—than among nonenrollees—0.5% and 63%, respectively (overall P < .001).
Fig 1.
Study flowchart. anti-HBc, hepatitis B core antibody; anti-HBs, hepatitis B surface antibody; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; (*) 80% random sampling implemented after month 3.
The mean (± standard deviation) age of the 2,124 patients in our final study cohort was 58 years (± 13 years), and 54% of the patients were women (Table 1). Approximately 77% of the patients were non-Hispanic white, 11% Hispanic, 8% black, and 4% Asian. Of the 2,124 patients, 424 (20%) had a hematologic malignancy, 23 (1%) had hepatocellular cancer (HCC), and 1,677 (79%) had a non-HCC solid tumor.
Table 1.
HBV Status According to Cancer Type and Responses to the Modified 19-Question Centers for Disease Control and Prevention HBV Survey
The final study cohort included 135 patients with chronic (seven patients; 0.3%) or past (128 patients; 6%) HBV infection (Fig 1). Of the 128 patients with past HBV infection, 98 (77%) had evidence of resolved HBV infection (positive anti-HBc and anti-HBs test results), 27 (21%) had occult HBV infection (positive anti-HBc and negative anti-HBs test results), and three patients had indeterminate anti-HBs results that indicated a detectable antibody level that was below the threshold of quantification. Three hundred forty patients (17%) had only positive anti-HBs test results, which indicated previous immunization. At the onset of anticancer therapy, median HBV DNA level for patients with chronic HBV infection was 594.5 IU/mL (range, 26 to 3180 IU/mL), and none of the 104 patients with past HBV infection who underwent HBV DNA testing had detectable HBV DNA in serum. Among 1,677 patients with non-HCC solid tumors, four (0.2%) had chronic HBV infection and 99 (5.9%) had past HBV infection. Among the 23 patients with HCC, one had chronic HBV infection and one had past HBV infection. Among the 424 patients with hematologic malignancies, two (0.5%) had chronic HBV infection and 28 (6.6%) had past HBV infection.
We examined HBV infection status according to patient characteristics as assessed in the modified CDC survey (Table 1). Rates of chronic or past HBV infection in various subgroups were as follows: 31% in Asian patients and nearly 16% in black patients versus 4.6% in white patients; 7.5% in men versus 5.4% in women; nearly 18% in patients who were born outside the United States and approximately 15% in patients with a parent who was born outside the United States versus less than 5% in patients who were born in the United States and patients with a parent who was born in the United States; almost 16% in patients with a household exposure to HBV infection versus approximately 6% in patients without such a household exposure; and more than 42% in men having sex with men and nearly 28% in patients who ever used injected drugs versus 6% in patients without these behaviors (6% for each group). Mean age was higher among patients with HBV infection than among those without HBV infection (61 ± 11.5 years v 58 ± 13.3 years).
The initial multivariable model that predicted chronic or past HBV infection (Table 2) included seven significant predictors: age 50 to < 65 years or age ≥ 65 years, men having sex with men, black or Asian race, birthplace outside the United States, parent’s birthplace outside the United States, household exposure to HBV infection, and history of injection drug use. The bias-corrected AUC for this initial seven-question predictive model was 0.79 (95% CI, 0.73 to 0.82). Using the same bootstrap samples, we identified the five most frequently chosen models from bootstrapping methods (Table 3). Hosmer-Lemeshow goodness-of-fit tests indicated that all five models fitted the data well (P > .15). From these, we selected model C, which had seven questions, including patient’s birthplace and parent’s birthplace.
Table 2.
Initial Multivariable Model Predicting Chronic or Past HBV Infection
Table 3.
Top Five Models Predicting Chronic or Past HBV Infection Using Bootstrapping Methods of Validation
We next compared the performance of our 19-question modified CDC survey, the 18-question CDC hepatitis risk assessment, model C (Table 3), and models that contained seven, six, and five questions on the basis of the questions in model C predicting chronic or past HBV infection (Table 4). FNR for the modified CDC survey was 0%—in other words, all patients with chronic or past HBV infection would have been identified if we had serologically tested all patients with at least one affirmative answer to any of the survey’s 19 questions. FNR for the CDC hepatitis risk assessment was 3%; use of this model to screen our cohort would have caused us to miss four patients with HBV infection, all of whom had past HBV infection. Model C, which included male gender by history of sex with men, had an FNR of 1.5% and would have resulted in our missing two patients with past HBV infection. We then explored a seven-question model that was identical to model C, except that the question about sexual behaviors was omitted, and found that the FNR was 0%. A briefer six-question HBV screening model performed as well as the modified CDC survey, with an FNR of 0%. Omission of a potentially embarrassing question about injection drug use yielded our five-question HBV screening model, which had an FNR of 0.7% and would have missed one patient with past HBV infection (Table 4).
Table 4.
Comparison of Selective HBV Screening Strategies
The proportion of patients who responded “Yes” to at least one survey question was 91% (1,932 of 2,124 patients) for the CDC hepatitis risk assessment, 92% (1,956 of 2,124 patients) for the modified CDC survey, 90% (1,908 of 2,124 patients) for our five-question survey, and 90% (1,910 of 2,124 patients) for our six-question survey. In other words, with the use of any of these surveys, approximately 90% of patients would require HBV serologic testing to achieve an FNR < 1%.
DISCUSSION
Identification of patients with HBV infection is the first step in preventing HBV reactivation as a result of anticancer therapies. In this study, we found that HBV risk surveys with five or six questions performed as well as a 19-question modified CDC survey. Our brief surveys demonstrated excellent sensitivity and low FNRs, which indicated that they could be practicable tools with which to identify patients with cancer with HBV infection. To our knowledge, our study is the first to develop an HBV screening tool on the basis of HBV risk factors in a population of patients with cancer. Our study is also unique in employing bootstrapping to validate an initial predictive model and offering selective screening strategies for consideration. Only one previous study identified predictors of HBV infection—male gender and birthplace in an HBV-endemic country—and this was in individuals without cancer.15
Currently, there are no clear guidelines for HBV serologic testing in patients with cancer. Several approaches have been recommended—testing all patients, testing only those with risk factors for HBV infection, or testing only patients who would be treated with anticancer therapies associated with a high risk of HBV reactivation (eg, anti-CD20 monoclonal antibodies or stem cell transplantation).1,3-5 At present, HBV testing among patients with cancer remains suboptimal, and preventable adverse clinical outcomes continue to occur.
Each of the HBV identification methods that have been proposed to date has benefits and limitations. Universal HBV testing would identify all patients who are at risk for reactivation and would obviate questioning about HBV risk factors; however, universal serologic testing may not be cost effective in countries with a low prevalence of HBV infection, such as in the United States,16 and would have limited benefit among patients with solid tumors, whose risk of reactivation from cancer treatment might be low.17,18 HBV screening can be based on the therapy-related risk of HBV reactivation; however, many cancer therapies can cause HBV reactivation, and we do not have data by which to discern the risks of HBV reactivation associated with all cancers therapies. Furthermore, most patients receive a combination of anticancer drugs, and when one regimen fails, other regimens are used. Thus, screening only patients who will receive therapies associated with a high risk of HBV reactivation, such as anti-CD20 therapy, might expose some patients to the risk of HBV reactivation.
In this study, we developed HBV screening tools using data from patients who were treated at our tertiary comprehensive cancer center. Most patients at our center have commercial insurance or Medicare, and our previous studies have found a low prevalence of chronic HBV infection in our patients,8,19 which makes our institution an ideal setting in which to develop brief HBV screening tools.
For our six-question HBV screening tool, with questions about age, gender, race and ethnicity, birthplace, household exposure to HBV infection, and injection drug use, the sensitivity was 100%, and the FNR was 0%. Eliminating the question about injection drug use, which is potentially embarrassing and might not be answered honestly, resulted in a modest decrease in sensitivity, to 99.3%, and a modest increase in FNR, to 0.7%. This five-question model is brief and effective; thus, it may be suitable for geographic areas of low HBV infection prevalence. Moreover, this model could be incorporated into electronic medical records to increase HBV testing rates.20,21 In contrast, in areas where HBV infection is endemic, screening tools are likely not necessary and universal HBV testing—resources permitting—may be advisable.
An important finding of our study was that, regardless of the survey—CDC hepatitis risk assessment, modified CDC survey, or five- or six-question survey—approximately 90% of patients at our institution would answer affirmatively to at least one of the survey questions and would therefore require HBV serologic testing. Thus, even though we provided a brief, internally validated risk tool, our findings indicate that universal HBV screening is more efficient and practical. In populations in which the prevalence of risk factors is lower, the selective risk tool could be of benefit in that it could decrease the need for HBV testing to a lower proportion (< 90%) of patients; however, this hypothesis needs to be validated.
Our study has several limitations. It was conducted at a single academic cancer center, and the number of patients with HBV infection was small. We used bootstrapping, an internal validation approach, to validate our findings; thus, our models should be independently validated in large, multicenter, prospective studies. Owing to the low prevalence of chronic HBV infection in the United States, we were unable to develop separate models for chronic HBV infection and past HBV infection. Having separate models is important as the risk of HBV reactivation is higher for patients with chronic HBV infection than it is for patients with past HBV infection. We excluded patients who were previously tested or known to have HBV infection, which likely reduced the prevalence of HBV infection in our cohort. We included patients with HCC in our risk assessment, though all patients with HCC should be tested for HBV as part of standard medical practice. We did not ask participants about their HBV vaccination history. Because our study was conducted in the United States, our results are unlikely to be applicable to high-prevalence settings outside the United States. Our findings, though limited to patients with cancer who were awaiting anticancer therapy, may have relevance for the broader population of patients receiving immunosuppressive therapy for GI, rheumatologic, or dermatologic diseases.
In summary, our selective screening models on the basis of HBV risk factors and demographics performed as well as the 19-question modified CDC survey. A five-question HBV screening tool, with questions about age, gender, race and ethnicity, birthplace, and household exposure to HBV infection, may be administered by medical assistants or self-administered by patients to identify patients who are at risk for HBV reactivation from anticancer therapy; however, our data indicated that the majority of our study population needed HBV testing despite the risk assessment tool if we aim for an FNR of zero, which reinforces the need for universal HBV screening. More efficient screening tools could be developed but would sacrifice sensitivity, leading to missed patients with HBV infection who are at risk for adverse liver outcomes and death as a result of reactivation. Future studies are needed to assess the cost effectiveness of universal testing compared with selective screening across populations who receive low-to-intermediate intensity anticancer regimens and with varying prevalence of HBV infection and range of FNR.
ACKNOWLEDGMENT
We thank Cynthia Jorgensen at the Centers for Disease Control and Prevention (CDC) for providing technical support for the use of the CDC hepatitis risk assessment. At The University of Texas Anderson Cancer Center, we thank Rhodrick Haralson, Sheila Khalili-Ahmadi, and Diane Stryk for assistance with patient enrollment; Stephanie Deming (Department of Scientific Publications) for editorial assistance; Laurissa Gann for assistance with references; Reeni Luke and Sanjivkumar Dave for assistance with institutional databases; and Jean Caputo and Calvin Harris for support with electronic survey implementation. We are grateful to the study patients who generously gave their time and effort to this project.
Footnotes
Supported by US National Cancer Institute Grant Nos. K07-CA132955 and R21-CA167202 (both to J.P.H.). M.E.S.-A. received Grant No. K24-AR053593 from the US National Institute for Musculoskeletal and Skin Disorders during the conduct of this study. The University of Texas MD Anderson Cancer Center is supported by US National Cancer Institute Grant No. P30-CA016672.
Presented at the Society of General Internal Medicine Annual Meeting, Washington, DC, April 19, 2017.
Clinical trial information: NCT01970254.
See accompanying Editorial on page 935
AUTHOR CONTRIBUTIONS
Conception and design: Jessica P. Hwang, Anna S. Lok, Maria E. Suarez-Almazor
Financial support: Jessica P. Hwang
Administrative support: Jessica T. Foreman
Collection and assembly of data: Jessica P. Hwang, Andrea Barbo, Jessica T. Foreman
Data analysis and interpretation: Jessica P. Hwang, Anna S. Lok, Michael J. Fisch, Scott B. Cantor, Andrea Barbo, Heather Y. Lin, John M. Vierling, Harrys A. Torres, Bruno P. Granwehr, Ethan Miller, Cathy Eng, George R. Simon, Sairah Ahmed, Alessandra Ferrajoli, Jorge Romaguera, Maria E. Suarez-Almazor
Manuscript writing: All authors
Final approval of manuscript: All authors
Accountable for all aspects of the work: All authors
AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
Models to Predict Hepatitis B Virus Infection Among Patients With Cancer Undergoing Systemic Anticancer Therapy: A Prospective Cohort Study
The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/site/ifc.
Jessica P. Hwang
Research Funding: Gilead Sciences, Merck & Co.
Anna S. Lok
Research Funding: Bristol-Myers Squibb, Gilead Sciences
Michael J. Fisch
Stock or Other Ownership: Anthem
Other Relationship: AIM Specialty Health
Scott B. Cantor
Research Funding: Hitachi, Intuitive Surgical
Andrea Barbo
No relationship to disclose
Heather Y. Lin
No relationship to disclose
Jessica T. Foreman
No relationship to disclose
John M. Vierling
Consulting or Advisory Role: Bristol-Myers Squibb, Gilead Sciences
Research Funding: Bristol-Myers Squibb (Inst), Gilead Sciences (Inst)
Travel, Accommodations, Expenses: Gilead Sciences
Harrys A. Torres
Consulting or Advisory Role: Gilead Sciences
Research Funding: Gilead Sciences, Merck, Vertex
Bruno P. Granwehr
Research Funding: Merck
Ethan Miller
Stock or Other Ownership: Amgen
Cathy Eng
Honoraria: Roche, Eli Lilly, Bayer
Consulting or Advisory Role: Genentech, Bayer Schering Pharma, Sirtex Medical
Research Funding: Keryx, Advaxis, Genentech
Travel, Accommodations, Expenses: Genentech, Bayer
George R. Simon
No relationship to disclose
Sairah Ahmed
Honoraria: Seattle Genetics
Alessandra Ferrajoli
No relationship to disclose
Jorge Romaguera
No relationship to disclose
Maria E. Suarez-Almazor
Consulting or Advisory Role: Bristol-Myers Squibb, Endo Pharmaceuticals, Pfizer, Eli Lilly
Research Funding: Pfizer
REFERENCES
- 1.Weinbaum CM, Williams I, Mast EE, et al. :Recommendations for identification and public health management of persons with chronic hepatitis B virus infection. MMWR Recomm Rep 57:1-20, 2008 [PubMed] [Google Scholar]
- 2.American Association for the Study of Liver Diseases : AASLD practice guideline update: Chronic hepatitis B: Update 2009. www.aasld.org/practiceguidelines/Documents/Bookmarked%20Practice%20Guidelines/Chronic_Hep_B_Update_2009%208_24_2009.pdf
- 3.Reddy KR, Beavers KL, Hammond SP, et al. : American Gastroenterological Association Institute guideline on the prevention and treatment of hepatitis B virus reactivation during immunosuppressive drug therapy. Gastroenterology 148:215-219, quiz e16-e17, 2015. [Erratum: Gastroenterology 148: 455, 2015] [DOI] [PubMed] [Google Scholar]
- 4.Hwang JP, Somerfield MR, Alston-Johnson DE, et al. : Hepatitis B virus screening for patients with cancer before therapy: American Society of Clinical Oncology provisional clinical opinion update. J Clin Oncol 33:2212-2220, 2015 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Baden LR, Swaminathan S, Angarone M, et al. : Prevention and treatment of cancer-related infections, version 2.2016, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 14:882-913, 2016 [DOI] [PubMed] [Google Scholar]
- 6.LeFevre ML, U.S. Preventive Services Task Force : Screening for hepatitis B virus infection in nonpregnant adolescents and adults: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med 161:58-66, 2014 [DOI] [PubMed] [Google Scholar]
- 7.National Comprehensive Cancer Network : Prevention and treatment of cancer-related infections (version 2.2014). http://www.nccn.org/professionals/physician_gls/pdf/infections.pdf
- 8.Hwang JP, Fisch MJ, Zhang H, et al. : Low rates of hepatitis B virus screening at the onset of chemotherapy. J Oncol Pract 8:e32-e39, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Day FL, Link E, Thursky K, et al. : Current hepatitis B screening practices and clinical experience of reactivation in patients undergoing chemotherapy for solid tumors: a nationwide survey of medical oncologists. J Oncol Pract 7:141-147, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hwang J, Suarez-Almazor ME, Cantor SB, et al. : Impact of the timing of hepatitis B virus identification and anti-hepatitis B virus therapy initiation on the risk of adverse liver outcomes in patients receiving cancer therapy. Cancer 123:3367-3376, 2017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Centers for Disease Control and Prevention : Hepatitis risk assessment. https://www.cdc.gov/hepatitis/riskassessment/
- 12.Woolson RF, Clarke WR. Statistical Methods for the Analysis of Biomedical Data. New York, John Wiley and Sons, 2002. 10.1002/9781118033050 [DOI] [Google Scholar]
- 13.Gonen M. Analyzing Receiver Operating Characteristic Curves Using SAS. Cary, NC, SAS Press, 2007 [Google Scholar]
- 14.Cameron AC, Trivedi PK. Microeconomics Using Stata. Stata Press, College Station, TX: 2009 [Google Scholar]
- 15.Spenatto N, Boulinguez S, Mularczyk M, et al. : Hepatitis B screening: who to target? A French sexually transmitted infection clinic experience. J Hepatol 58:690-697, 2013 [DOI] [PubMed] [Google Scholar]
- 16.Zurawska U, Hicks LK, Woo G, et al. : Hepatitis B virus screening before chemotherapy for lymphoma: A cost-effectiveness analysis. J Clin Oncol 30:3167-3173, 2012 [DOI] [PubMed] [Google Scholar]
- 17.Voican CS, Mir O, Loulergue P, et al. : Hepatitis B virus reactivation in patients with solid tumors receiving systemic anticancer treatment. Ann Oncol 27:2172-2184, 2016 [DOI] [PubMed] [Google Scholar]
- 18.Paul S, Saxena A, Terrin N, et al. : Hepatitis B virus reactivation and prophylaxis during solid tumor chemotherapy: A systematic review and meta-analysis. Ann Intern Med 164:30-40, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hwang JP, Fisch MJ, Lok AS, et al. : Trends in hepatitis B virus screening at the onset of chemotherapy in a large US cancer center. BMC Cancer 13:534, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Hsu PI, Lai KH, Cheng JS, et al. : Prevention of acute exacerbation of chronic hepatitis B infection in cancer patients receiving chemotherapy in a hepatitis B virus endemic area. Hepatology 62:387-396, 2015 [DOI] [PubMed] [Google Scholar]
- 21.Sanagawa A, Kuroda J, Shiota A, et al. : Outcomes of the implementation of the computer-assisted HBView system for the prevention of hepatitis B virus reactivation in chemotherapy patients: A retrospective analysis. J Pharm Health Care Sci 1:29, 2015 [DOI] [PMC free article] [PubMed] [Google Scholar]