Karthik Natarajan

Redesigning electronic health record systems to support public health

Current electronic health record systems are primarily clinical in focus, designed to provide patient-level data and provider-level decision support. Adapting EHR systems to serve public health needs provides the possibility of enormous advances for public health practice and policy. In this review, we evaluate EHR functionality and map it to the three core functions of public health: assessment, policy development, and assurance. In doing so, we identify and discuss important design, implementation, and methodological issues with current systems. For example, in order to support public healths traditional focus on preventive health and socio-behavioral factors, EHR data models would need to be expanded to incorporate environmental, psychosocial, and other non-medical data elements, and workflow would have to be examined to determine the optimal way of collecting these data. We also argue that redesigning EHR systems to support public health offers benefits not only to the public health system but also to consumers, health-care institutions, and individual providers.

Diagnosis code assignment: models and evaluation metrics

Background and objective The volume of healthcare data is growing rapidly with the adoption of health information technology. We focus on automated ICD9 code assignment from discharge summary content and methods for evaluating such assignments. Methods We study ICD9 diagnosis codes and discharge summaries from the publicly available Multiparameter Intelligent Monitoring in Intensive Care II (MIMIC II) repository. We experiment with two coding approaches: one that treats each ICD9 code independently of each other (flat classifier), and one that leverages the hierarchical nature of ICD9 codes into its modeling (hierarchy-based classifier). We propose novel evaluation metrics, which reflect the distances among gold-standard and predicted codes and their locations in the ICD9 tree. Experimental setup, code for modeling, and evaluation scripts are made available to the research community. Results The hierarchy-based classifier outperforms the flat classifier with F-measures of 39.5% and 27.6%, respectively, when trained on 20 533 documents and tested on 2282 documents. While recall is improved at the expense of precision, our novel evaluation metrics show a more refined assessment: for instance, the hierarchy-based classifier identifies the correct sub-tree of gold-standard codes more often than the flat classifier. Error analysis reveals that gold-standard codes are not perfect, and as such the recall and precision are likely underestimated. Conclusions Hierarchy-based classification yields better ICD9 coding than flat classification for MIMIC patients. Automated ICD9 coding is an example of a task for which data and tools can be shared and for which the research community can work together to build on shared models and advance the state of the art.

Changes in Tdap and MCV4 Vaccine Coverage Following Enactment of a Statewide Requirement of Tdap Vaccination for Entry Into Sixth Grade

OBJECTIVES We evaluated changes in tetanus toxoid, reduced diphtheria toxoid, acellular pertussis (Tdap), and tetravalen meningococcal (MCV4) vaccine coverage following enactment of a New York State mandate requiring Tdap before entering sixth grade. METHODS Using data from a hospital-based immunization registry, we measured Tdap and MCV4 coverage among youths aged 11 to 14 years in New York City at 3 time points: premandate, mandate year 1, and mandate year 2. RESULTS Among overlapping cohorts of 4316 (premandate), 4131 (mandate year 1), and 3639 (mandate year 2) youths, Tdap coverage increased steadily over time (29%, 58%, and 83%, respectively). Increases were observed among all ages. Across the same time points, MCV4 coverage also increased (10%, 30%, and 60%, respectively). Most adolescents did not receive MCV4 during the same visit they received Tdap. CONCLUSIONS A Tdap school-entry mandate was associated with substantial increases in immunization coverage, even in age groups not directly affected by the mandate. At the postmandate time points, MCV4 coverage remained lower than Tdap coverage. Provider education should emphasize the importance of reviewing vaccine records and administering all recommended vaccines at every clinical encounter.

Influenza vaccination coverage and timeliness among children requiring two doses, 2004–2009

OBJECTIVE To assess influenza vaccination coverage and timeliness among children requiring two doses in a season. METHODS This study examined seasonal influenza vaccination of 17,800 children from five academically-affiliated clinics in New York City using hospital and city immunization registries. Eligible children were 6 months-8 years and needed two influenza vaccine doses in a given season between 2004-05 and 2009-10. Any (≥ 1 dose) and full (2 doses) vaccination coverage by December 15 and March 31 as well as interval between doses were calculated. Vaccination trends over time, determinants, and missed opportunities were assessed. RESULTS Children were primarily Latino and publicly insured. Full coverage by March 31 increased between the 2004-05 and 2009-10 seasons (9% vs. 29%, p<0.001). Few children received both doses by December 15 (2-13%). The interval between doses was almost twice as long as recommended and increased over time (2004-05: 52 days; 2009-10: 64 days; p<0.001). Older age and Latino ethnicity were negative predictors of full vaccination by March 31. Missed opportunities for the second dose were common. CONCLUSION Despite improvements, low-income, minority children requiring two influenza vaccine doses remain at risk of incomplete and delayed vaccination. Barriers to and strategies for timely full vaccination should be explored.

HPV vaccination: are we initiating too late?

BACKGROUND Human papillomavirus (HPV) vaccination is recommended in early adolescence. While limited data suggest that patients frequently delay initiation of the three-dose series, age-based variability in initiation of HPV vaccination and its clinical relevance are not well described. Thus, this study aims to characterize HPV vaccination delay among adolescent and young adult females. METHODS This retrospective cohort study examined age at HPV vaccination initiation and missed opportunities for receipt of the first vaccine dose (HPV1) among 11-26 year-old females (n=22,900) receiving care at 16 urban academically-affiliated ambulatory care clinics between 2007 and 2011. Predictors of timely vaccination and post-licensure trends in age at HPV1 receipt were assessed using multivariable logistic regression and a generalized linear mixed model, respectively. Chlamydia trachomatis and Papanicolaou screening before HPV vaccination initiation, as markers of prior sexual experience and associated morbidity, were examined in a subcohort of subjects (n=15,049). RESULTS The proportion of 11-12 year-olds who initiated HPV vaccination increased over time (44.4% [2007] vs. 74.5% [2011], p<0.01). Initiation rates also improved among 13-26 year-olds. Thus, the mean age at HPV1 receipt remained unchanged between 2007 and 2011 (16.0 ± 2.7 vs. 15.9 ± 4.0 years, p=0.45). Spanish language was a positive predictor (AOR 1.62, 95% CI 1.05-2.48) of HPV vaccination initiation among 11-12 year-olds in 2011. The majority (70.8-76.4%) of unvaccinated subjects experienced missed vaccination opportunities. Of the subcohort, 36.9% underwent Chlamydia screening before HPV1 receipt (19.1% with ≥ 1 positive result). Of those with prior Papanicolaou screening (16.6%), 32.1% had ≥ 1 abnormal result. CONCLUSIONS These low-income, minority females frequently delayed initiation of HPV vaccination. Many had evidence of prior sexual experience and associated morbidity, placing them at risk of HPV-related complications. Promoting timely HPV vaccination and reducing missed vaccination opportunities are crucial.

Immunization Data Exchange With Electronic Health Records.

OBJECTIVE: To assess the impact of exchange of immunization information between an immunization information system (IIS) and an electronic health record on up-to-date rates, overimmunization, and immunization record completeness for low-income, urban children and adolescents. METHODS: The New York City Department of Health maintains a population-based IIS, the Citywide Immunization Registry (CIR). Five community clinics in New York City implemented direct linkage of immunization data from the CIR to their local electronic health record. We compared immunization status and overimmunization in children and adolescents 19 to 35 month, 7 to 10 year, and 13 to 17 year-olds with provider visits in the 6-month period before data exchange implementation (2009; n = 6452) versus 6-months post-implementation (2010; n = 6124). We also assessed immunization record completeness with and without addition of CIR data for 8548 children and adolescents with visits in 2012–2013. RESULTS: Up-to-date status increased from before to after implementation from 75.0% to 81.6% (absolute difference, 6.6%; 95% confidence interval [CI], 5.2% to 8.1%) and was significant for all age groups. The percentage overimmunized decreased from 8.8% to 4.7% (absolute difference, −4.1%; 95% CI, −7.8% to −0.3%) and was significant for adolescents (16.4% vs 1.2%; absolute difference, −15.2%; 95% CI, −26.7 to −3.6). Up-to-date status for those seen in 2012 to 2013 was higher when IIS data were added (74.6% vs 59.5%). CONCLUSIONS: This study demonstrates that data exchange can improve child and adolescent immunization status. Development of the technology to support such exchange and continued focus on local, state, and federal policies to support such exchanges are needed.

Timeliness of Pediatric Influenza Vaccination Compared With Seasonal Influenza Activity in an Urban Community, 2004–2008

OBJECTIVES We assessed pediatric influenza vaccination in relation to community influenza activity. METHODS We examined seasonal influenza vaccination in 34,012 children aged 6 months through 18 years from 5 academically affiliated clinics in northern Manhattan, New York (an urban low-income community) during the 2004-2008 seasons using hospital and city immunization registries. We calculated the cumulative number of administered influenza vaccine doses and proportion of children with any (≥ 1 dose) or full (1-2 doses per age recommendations) vaccination at the onset and peak of community polymerase chain reaction-confirmed influenza activity according to state surveillance reports and by March 31 each season. RESULTS Influenza vaccine administration began before October 1, peaked before influenza activity onset, and declined gradually over each season. Coverage at influenza activity onset, peak, and by March 31 increased over the 5 seasons. However, most children lacked full vaccination at these time points, particularly adolescents, minorities, and those requiring 2 doses. CONCLUSIONS Despite early initiation of influenza vaccination, few children were fully vaccinated when influenza began circulating. Interventions should address factors negatively affecting timely influenza vaccination, especially in high-risk populations.

Timeliness of 2009 H1N1 vaccine coverage in a low-income pediatric and adolescent population

Despite being at highest risk for 2009 H1N1 virus morbidity and mortality, many children were not immunized with the vaccine. Identification of factors that put certain children at higher risk for under-immunization could reveal populations who may need to be specifically targeted for vaccination interventions in future pandemics. Little is known about the prevalence of, or factors associated with, 2009 H1N1 vaccine coverage in low-income, urban pediatric populations. This study evaluated 2009 H1N1 vaccination coverage in 19,643 children aged 6months to 18years receiving care at one of five community clinics associated with an academic medical center in a low-income community. Any (≥1 dose) and full coverage (1 dose for children ≥10years old, 2 doses for those <10years) was determined as of December 1, 2009 and the end of vaccination period (June 30, 2010). Multivariable analyses were used to assess the impact of race/ethnicity, age, insurance, gender, and language on vaccine coverage and timeliness. By December 1, only 16.6% of children had received one dose, and 5.3% had full coverage. By the end of the vaccination period, 36.2% had received at least one H1N1 dose and 23.6% had full coverage. On multivariable analysis, older age, minority race/ethnicity, and private insurance were negatively associated with vaccination by December 1 and end of vaccination period, even after accounting for attendance at a clinic visit. In future pandemics, when timely receipt of a new vaccine in large populations may be imperative, general vaccination programs as well as special targeted education and vaccine reminders for these at risk groups may be warranted.

Vaccination Coverage of Adolescents With Chronic Medical Conditions

INTRODUCTION Adolescents with chronic medical conditions (CMCs) are at increased risk of vaccine-preventable infections. Little is known about their vaccine uptake. METHODS This retrospective cohort study included 3,989 adolescents aged 11-17 years receiving care at academically affiliated pediatric clinics between August 2011 and June 2013. Data were abstracted from the medical centers electronic health record and immunization registry in 2014. Vaccination coverage, timeliness, and missed opportunities were evaluated and analyzed in 2015-2016. RESULTS Adolescents with CMCs had lower human papillomavirus vaccination initiation than those without CMCs (81.3% vs 85.0%), although this difference was only observed in stratified analysis among males (adjusted relative risk=0.90, 95% CI=0.85, 0.96), aged 13-17 years (adjusted relative risk=0.94, 95% CI=0.91, 0.98), and those with more primary care visits (adjusted relative risk=0.94, 95% CI=0.91, 0.98). Adolescents with CMCs had greater influenza vaccination coverage and timeliness than those without CMCs (2011-2012 season: 66.9% vs 50.1%; adjusted hazards ratio=1.27, 95% CI=1.15, 1.40; 2012-2013 season: 73.8% vs 64.5%; adjusted hazards ratio=1.20, 95% CI=1.10, 1.31). Only 32.1% and 18.2% of eligible adolescents had received pneumococcal polysaccharide and 13-valent pneumococcal conjugate vaccines, respectively. Missed opportunities were higher among adolescents with versus without CMCs for human papillomavirus vaccination initiation (4.2 vs 2.7, p<0.001), meningococcal vaccination (4.0 vs 2.9, p<0.001), and influenza vaccination (2011-2012 season: 2.1 vs 1.7, p<0.001; 2012-2013 season: 2.0 vs 1.6, p<0.001). Missed opportunities for pneumococcal vaccination were common. CONCLUSIONS Pockets of undervaccination and missed opportunities exist among adolescents with CMCs. Greater, more timely influenza vaccination suggests that optimal vaccination of high-risk adolescents is possible.