SKStephanie A. Kujawski

Stephanie A. Kujawski

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Stephanie A. Kujawski has a total of 120 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Estimating the Long-Term Clinical and Economic Impact of Universal Varicella Vaccination in Portugal Using a Dynamic Transmission Model

    by Colleen Burgess, Stephanie A. Kujawski, Miguel Amorim, John C. Lang

    Background Varicella vaccination is not currently included in Portugal’s childhood national immunization program. We modeled the clinical and economic impact of a two-dose universal varicella vaccination (UVV) program in Portugal. Methods A dynamic transmission model was adapted to Portugal to assess the impact of UVV over a 50-year time horizon. Two two-dose UVV strategies (first dose at 12 months, second dose at 5 years) were compared to no vaccination. Clinical outcomes included varicella and herpes zoster incidence rates, cases, deaths, quality-adjusted life years (QALYs) lost, and varicella-related health care resource use. Costs were assessed from the payer perspective (direct costs of treatment) and societal perspective (direct costs plus indirect costs of lost workdays). Results Both UVV strategies reduced total varicella cases by 79.4–85.3%, outpatient visits by 78.9–84.8%, and varicella hospitalizations by 69.3–71.3% compared to no varicella vaccination. Herpes zoster cases were reduced by 2.4–2.9%. These strategies resulted in the gain of 5977–6094 QALYs and cost savings of €12.0 million (M)–15.9M from the payer perspective and €84.2M–93.9M from the societal perspective. Conclusions A two-dose UVV program is estimated to substantially reduce the burden of varicella disease in Portugal and to be cost-saving from both the payer and societal perspectives.

  • Open Access

    Impact of Universal Varicella Vaccination in Israel: A Time-series Analysis, 2003 to 2020

    by Noga Fallach, Pieralessandro Lasalvia, Manjiri Pawaskar, Gabriel Chodick, David Greenberg, Stephanie A. Kujawski

    Background: Two-dose universal varicella vaccination (UVV) was included in Israel’s national immunization program in 2008. We aimed to evaluate the impact of the UVV program on varicella incidence and healthcare costs in Israel. Methods: A retrospective observational study was conducted using the Maccabi Healthcare Services database between 2003 and 2020. Varicella incidence and varicella-related healthcare costs were assessed in the pre- (2003–2007) and post-UVV (2010–2020) periods. We used time-series analyses with seasonal autoregressive integrated moving average models to estimate the impact of UVV on varicella incidence and regression models with gamma distributions and log-link functions to assess the impact on healthcare costs. Analyses were conducted for the overall population and vaccine-targeted (1–7 years) and nontargeted (7 years) populations. Results: Annual incidence rates per 100,000 declined from 5709.6 (pre-UVV) to 1062.9 (post-UVV) in the vaccine-targeted population and from 1269.8 to 409.6 in the overall population. Time-series analysis suggested 81.8% (95% prediction interval [PI], 67.5–87.4) and 67.1% (95% PI, 48.2–75.9) reductions in the vaccine-targeted vaccination and overall populations, respectively. The largest estimated reduction in varicella incidence was in children 1 to 2 years of age [91.2% (95% PI, 53.4–95.3)]. Mean direct medical costs decreased between pre- and post-UVV periods by 82% (95% confidence interval, 77–86) in the vaccine-targeted group and 51% (95% confidence interval, 44–58) overall. Conclusions: This real-world study demonstrates the substantial reduction in the disease and economic burden of varicella in the vaccine-targeted and overall population following the introduction of a 2-dose UVV program in Israel.


  • Open Access

    Prediction of measles cases in US counties: A machine learning approach

    by Stephanie A. Kujawski, Boshu Ru, Nelson Lee Afanador, James H. Conway, Richard Baumgartner, Manjiri Pawaskar


  • Open Access

    Healthcare resource utilization and costs associated with hepatitis A in the United States: a retrospective database analysis

    by Salome Samant, Edith Chen, Cristina Carias, Stephanie A. Kujawski

    Aim. To investigate hepatitis A-related healthcare resource use and costs in the US.Methods. The Merative Marketscan Commercial Claims and Encounters database was retrospectively analyzed for hepatitis A-related inpatient, outpatient, and emergency department (ED) claims from January 01, 2012 to December 31, 2018. We calculated the hepatitis A incidence proportion per 100,000 enrollees, healthcare resource utilization, and costs (in 2020 USD). Results were stratified by age, gender, and select comorbidities.Results. The overall hepatitis A incidence proportion was 6.1 per 100,000 enrollees. Among individuals with ≥1 hepatitis A-related claim, the majority (92.6%) had ≥1 outpatient visit related to hepatitis A; 9.1% were hospitalized, and 4.2% had ≥1 ED visit. The mean (standard deviation [SD]) length of hospital stay was 5.2 (8.1) days; the mean (SD) number of outpatient and ED visits were 1.3 (1.3) and 1.1 (0.6), respectively. The incidence proportion per 100,000 was higher among adults than children (7.5 vs. 1.5), individuals with HIV than those without (126.7 vs. 5.9), and individuals with chronic liver disease than those without (143.6 vs. 3.8). The total mean (SD)/median (IQR) per-patient cost for hepatitis A-related care was $2,520 ($10,899)/$156 ($74-$529) with the mean cost of hospitalization was 18.7 times higher than that of outpatient care ($17,373 vs. $928).Limitations. The study data included only a commercially insured population and may not be representative of all individuals.Conclusions. In conclusion, hepatitis A is associated with a substantial economic burden among privately insured individuals in the US.


  • Open Access

    Clinical and Economic Burden of Antibiotic Use Among Pediatric Patients With Varicella Infection in the Outpatient Setting: A Retrospective Cohort Analysis of Real-world Data in France

    by Stephanie A. Kujawski, Caroline S. Casey, Hervé Haas, Amisha Patel, Christina Msc Diomatari, Tim Bsc Holbrook, Manjiri Pawaskar

    Background: Varicella infects 90% of children before age 9. Though varicella is self-limiting, its complications may require antibiotics, though how antibiotics are utilized for varicella in France is not well known. This study assessed antibiotic use and costs associated with varicella and its complications in pediatric patients managed in the outpatient setting in France. Methods: A retrospective cohort study using the Cegedim Strategic Data-Longitudinal Patient Database, an electronic medical record database from general practitioners and office-based specialists in France, was conducted. Children <18 years old diagnosed with varicella between January 2014 and December 2018 with 3-month follow-up available were included. We used descriptive analysis to assess varicella-related complications, medication use, healthcare resource utilization and costs. Results: Overall, 48,027 patients were diagnosed with varicella; 15.3% (n = 7369) had ≥1 varicella-related complication. Antibiotics were prescribed in up to 25.1% (n = 12,045/48,027) of cases with greater use in patients with complications (68.1%, n = 5018/7369) compared with those without (17.3%, n = 7027/40,658). Mean medication and outpatient varicella-related costs were €32.82 per patient with medications costing a mean of €5.84 per patient; antibiotics contributed ~23% to total costs annually. Conclusion: This study showed high antibiotic use for the management of varicella and its complications. A universal varicella vaccination program could be considered to alleviate complications and associated costs in France.


  • Open Access

    Etiology of Acute Lower Respiratory Illness Hospitalizations Among Infants in 4 Countries

    by John Kubale, Stephanie Kujawski, Irena Chen, Zhenke Wu, Ilham Abu Khader, Iris Hasibra, Brett Whitaker, Lionel Gresh, Artan Simaku, Eric A F Simões, et al.

    Background: Recent studies explored which pathogens drive the global burden of pneumonia hospitalizations among young children. However, the etiology of broader acute lower respiratory tract infections (ALRIs) remains unclear. Methods: Using a multicountry study (Albania, Jordan, Nicaragua, and the Philippines) of hospitalized infants and non-ill community controls between 2015 and 2017, we assessed the prevalence and severity of viral infections and coinfections. We also estimated the proportion of ALRI hospitalizations caused by 21 respiratory pathogens identified via multiplex real-time reverse transcription polymerase chain reaction with bayesian nested partially latent class models. Results: An overall 3632 hospitalized infants and 1068 non-ill community controls participated in the study and had specimens tested. Among hospitalized infants, 1743 (48.0%) met the ALRI case definition for the etiology analysis. After accounting for the prevalence in non-ill controls, respiratory syncytial virus (RSV) was responsible for the largest proportion of ALRI hospitalizations, although the magnitude varied across sites—ranging from 65.2% (95% credible interval, 46.3%–79.6%) in Albania to 34.9% (95% credible interval, 20.0%–49.0%) in the Philippines. While the fraction of ALRI hospitalizations caused by RSV decreased as age increased, it remained the greatest driver. After RSV, rhinovirus/enterovirus (range, 13.4%–27.1%) and human metapneumovirus (range, 6.3%–12.0%) were the next-highest contributors to ALRI hospitalizations. Conclusions: We observed substantial numbers of ALRI hospitalizations, with RSV as the largest source, particularly in infants aged <3 months. This underscores the potential for vaccines and long-lasting monoclonal antibodies on the horizon to reduce the burden of ALRI in infants worldwide.


  • Open Access

    The Clinical and Economic Burden of Antibiotic Use in Pediatric Patients With Varicella Infection: A Retrospective Cohort Analysis of Real-World Data in England

    by Stephanie A Kujawski, Victoria Banks, Caroline S Casey, Simon B Drysdale, Amisha Patel, Olivia Massey, Tim Holbrook, Manjiri Pawaskar

    Background: Varicella is a highly infectious disease, particularly affecting children, that can lead to complications requiring antibiotics or hospitalization. Antibiotic use for varicella management is poorly documented. This study assessed antibiotic use for varicella and its complications in a pediatric population in England. Methods: Data were drawn from medical records in the Clinical Practice Research Datalink and Hospital Episode Statistics data sets. The study included patients <18 years old with varicella diagnosed during 2014–2018 and 3-month follow-up available. We determined varicella-related complications, medication use, healthcare resource utilization, and costs from diagnosis until 3 months after diagnosis. Results: We identified 114 578 children with a primary varicella diagnosis. Of these, 7.7% (n = 8814) had a varicella-related complication, the most common being ear, nose, and throat related (37.1% [n = 3271]). In all, 25.9% (n = 29 706 of 114 578) were prescribed antibiotics. A higher proportion of patients with complications than without complications were prescribed antibiotics (64.3% [n = 5668 of 8814] vs 22.7% [n = 24 038 of 105 764]). Mean annualized varicella-related costs were £2 231 481 for the study cohort. Overall, antibiotic prescriptions cost approximately £262 007. Conclusions: This study highlights high antibiotic use and healthcare resource utilization associated with varicella management, particularly in patients with complications. A national varicella vaccination program in England may reduce varicella burden and related complications, medication use, and costs.


  • Open Access

    Predicting Measles Outbreaks in the United States: Evaluation of Machine Learning Approaches

    by Boshu Ru, Stephanie Kujawski, Nelson Lee Afanador, Richard Baumgartner, Manjiri Pawaskar, Amar Das

    Background: Measles, a highly contagious viral infection, is resurging in the United States, driven by international importation and declining domestic vaccination coverage. Despite this resurgence, measles outbreaks are still rare events that are difficult to predict. Improved methods to predict outbreaks at the county level would facilitate the optimal allocation of public health resources. Objective: We aimed to validate and compare extreme gradient boosting (XGBoost) and logistic regression, 2 supervised learning approaches, to predict the US counties most likely to experience measles cases. We also aimed to assess the performance of hybrid versions of these models that incorporated additional predictors generated by 2 clustering algorithms, hierarchical density-based spatial clustering of applications with noise (HDBSCAN) and unsupervised random forest (uRF). Methods: We constructed a supervised machine learning model based on XGBoost and unsupervised models based on HDBSCAN and uRF. The unsupervised models were used to investigate clustering patterns among counties with measles outbreaks; these clustering data were also incorporated into hybrid XGBoost models as additional input variables. The machine learning models were then compared to logistic regression models with and without input from the unsupervised models. Results: Both HDBSCAN and uRF identified clusters that included a high percentage of counties with measles outbreaks. XGBoost and XGBoost hybrid models outperformed logistic regression and logistic regression hybrid models, with the area under the receiver operating curve values of 0.920-0.926 versus 0.900-0.908, the area under the precision-recall curve values of 0.522-0.532 versus 0.485-0.513, and F2 scores of 0.595-0.601 versus 0.385-0.426. Logistic regression or logistic regression hybrid models had higher sensitivity than XGBoost or XGBoost hybrid models (0.837-0.857 vs 0.704-0.735) but a lower positive predictive value (0.122-0.141 vs 0.340-0.367) and specificity (0.793-0.821 vs 0.952-0.958). The hybrid versions of the logistic regression and XGBoost models had slightly higher areas under the precision-recall curve, specificity, and positive predictive values than the respective models that did not include any unsupervised features. Conclusions: XGBoost provided more accurate predictions of measles cases at the county level compared with logistic regression. The threshold of prediction in this model can be adjusted to align with each county’s resources, priorities, and risk for measles. While clustering pattern data from unsupervised machine learning approaches improved some aspects of model performance in this imbalanced data set, the optimal approach for the integration of such approaches with supervised machine learning models requires further investigation.


  • Open Access

    Predicting Measles Outbreaks in the United States: Evaluation of Machine Learning Approaches (Preprint)

    by Boshu Ru, Stephanie Kujawski, Nelson Lee Afanador, Richard Baumgartner, Manjiri Pawaskar, Amar Das

    BACKGROUND Measles, a highly contagious viral infection, is resurging in the United States, driven by international importation and declining domestic vaccination coverage. Despite this resurgence, measles outbreaks are still rare events that are difficult to predict. Improved methods to predict outbreaks at the county level would facilitate the optimal allocation of public health resources. OBJECTIVE We aimed to validate and compare extreme gradient boosting (XGBoost) and logistic regression, 2 supervised learning approaches, to predict the US counties most likely to experience measles cases. We also aimed to assess the performance of hybrid versions of these models that incorporated additional predictors generated by 2 clustering algorithms, hierarchical density-based spatial clustering of applications with noise (HDBSCAN) and unsupervised random forest (uRF). METHODS We constructed a supervised machine learning model based on XGBoost and unsupervised models based on HDBSCAN and uRF. The unsupervised models were used to investigate clustering patterns among counties with measles outbreaks; these clustering data were also incorporated into hybrid XGBoost models as additional input variables. The machine learning models were then compared to logistic regression models with and without input from the unsupervised models. RESULTS Both HDBSCAN and uRF identified clusters that included a high percentage of counties with measles outbreaks. XGBoost and XGBoost hybrid models outperformed logistic regression and logistic regression hybrid models, with the area under the receiver operating curve values of 0.920-0.926 versus 0.900-0.908, the area under the precision-recall curve values of 0.522-0.532 versus 0.485-0.513, and F2 scores of 0.595-0.601 versus 0.385-0.426. Logistic regression or logistic regression hybrid models had higher sensitivity than XGBoost or XGBoost hybrid models (0.837-0.857 vs 0.704-0.735) but a lower positive predictive value (0.122-0.141 vs 0.340-0.367) and specificity (0.793-0.821 vs 0.952-0.958). The hybrid versions of the logistic regression and XGBoost models had slightly higher areas under the precision-recall curve, specificity, and positive predictive values than the respective models that did not include any unsupervised features. CONCLUSIONS XGBoost provided more accurate predictions of measles cases at the county level compared with logistic regression. The threshold of prediction in this model can be adjusted to align with each county’s resources, priorities, and risk for measles. While clustering pattern data from unsupervised machine learning approaches improved some aspects of model performance in this imbalanced data set, the optimal approach for the integration of such approaches with supervised machine learning models requires further investigation.


  • Open Access

    Rates of respiratory syncytial virus (RSV)-associated hospitalization among adults with congestive heart failure—United States, 2015–2017

    by Stephanie A. Kujawski, Michael Whitaker, Matthew D. Ritchey, Arthur L. Reingold, Shua J. Chai, Evan J. Anderson, Kyle P. Openo, Maya Monroe, Patricia Ryan, Erica Bye, et al.

    Background: Respiratory syncytial virus (RSV) can cause severe disease in adults with cardiopulmonary conditions, such as congestive heart failure (CHF). We quantified the rate of RSV-associated hospitalization in adults by CHF status using population-based surveillance in the United States. Methods: Population-based surveillance for RSV (RSV-NET) was performed in 35 counties in seven sites during two respiratory seasons (2015–2017) from October 1–April 30. Adults (≥18 years) admitted to a hospital within the surveillance catchment area with laboratory-confirmed RSV identified by clinician-directed testing were included. Presence of underlying CHF was determined by medical chart abstraction. We calculated overall and age-stratified (<65 years and ≥65 years) RSV-associated hospitalization rates by CHF status. Estimates were adjusted for age and the under-detection of RSV. We also report rate differences (RD) and rate ratios (RR) by comparing the rates for those with and without CHF. Results: 2042 hospitalized RSV cases with CHF status recorded were identified. Most (60.2%, n = 1230) were ≥65 years, and 28.3% (n = 577) had CHF. The adjusted RSV hospitalization rate was 26.7 (95% CI: 22.2, 31.8) per 10,000 population in adults with CHF versus 3.3 (95% CI: 3.3, 3.3) per 10,000 in adults without CHF (RR: 8.1, 95% CI: 6.8, 9.7; RD: 23.4, 95% CI: 18.9, 28.5). Adults with CHF had higher rates of RSV-associated hospitalization in both age groups (<65 years and ≥65 years). Adults ≥65 years with CHF had the highest rate (40.5 per 10,000 population, 95% CI: 35.1, 46.6). Conclusions: Adults with CHF had 8 times the rate of RSV-associated hospitalization compared with adults without CHF. Identifying high-risk populations for RSV infection can inform future RSV vaccination policies and recommendations.