Abstract
Video Directly Observed Therapy (VDOT)- A useful approach to tuberculosis (TB) treatment during a natural disaster
Alexis Medrano, MPH, Dana Wiltz-Beckham, DVM, MPH, MBA, Vishaldeep Sekhon, MPH, Kimberly Fields, RN, MSN, FNP, Brian Reed, MD, Les Becker, MBA and Umair A. Shah, MD, MPH
Harris County Public Health, Houston, TX
APHA's 2018 Annual Meeting & Expo (Nov. 10 - Nov. 14)
Background:
Harris County Public Health (HCPH) is the local health department serving Harris County, Texas. The tuberculosis (TB) case rate in Harris County jurisdiction (4.4 per 100,000) is 1.5 times the US rate. In addition to the ongoing standard of care for TB treatment, Directly Observed Therapy (DOT), HCPHâs TB Elimination Program implemented asynchronous Video Directly Observed Therapy (VDOT) in 2014. This smartphoneâbased technology allows individuals to remotely record medication doses and submit videos on a HIPAA compliant platform.
Hurricane Harvey made landfall in Harris County on August 26, 2017. Harvey posed a threat for the interruption of TB treatment and services due to impassable/closed roads and road hazards with transportation for patients and staff.
Objective:
To illustrate the usefulness of VDOT for the continuation of TB treatment during a natural disaster.
Methods:
During Hurricane Harvey, 108 patients were receiving TB treatment in Harris County. Of those, 61 (57%) were on VDOT and 47 (43%) were on DOT. VDOT patients were advised to continue treatment by submitting their videos to the HIPPA compliant platform, which were reviewed daily by trained TB staff. Traditional DOT patients were advised to self-medicate, which does not count toward their treatment completion.
Result:
During the hurricane, 59 out of 61 (97%) of VDOT patients were able to continue treatment with no interruptions. Two patients with interruptions resumed treatment after 2 and 12 days, respectively. The 47 DOT patients were required to self-medicate, but unavailability of DOT services lead to prolongation of their treatment for 16 days on average.
Conclusion:
Hurricane Harvey marks the first time HCPH has used VDOT in a natural disaster of this magnitude. By utilizing VDOT in accordance with HCPH guidelines and preparation plans, TB patients successfully avoided treatment disruption in a natural disaster because of VDOTâs flexibility, asynchronous and remote nature.
Clinical medicine applied in public health Communication and informatics Protection of the public in relation to communicable diseases including prevention or control Provision of health care to the public
Abstract
Using association rules mining technique in analyzing qualitative medical health record notes to improve health care deliveries
Chieko Maene, MS1, Karen Kim, MD, MS1, Matt Johnson, MPH2, Akushika Kwesi3, Manasi Jayaprakash, MD, MPH4, Michael Quinn, PHD1 and Nathan Bostick, ScD, MA, MPP1
(1)University of Chicago, Chicago, IL, (2)Asian Health Coalition, Chicago, IL, (3)University of Chicago, (4)Center for Asian Health Equity/Asian Health Coalition, Chicago, IL
APHA's 2018 Annual Meeting & Expo (Nov. 10 - Nov. 14)
Background: Unstructured qualitative data such as informal surveys or short interviews can be easily collected by health care providers and used to improve health care once it is analyzed. Proprietary qualitative data analysis applications are available but associated costs and/or capacity to deploy can limit use.
Objective: This paper reviews the use of a rather simple data mining approach, association rules mining, in extracting meaningful information from existing qualitative data using an open source software, R.
Method: In 2016, the University of Chicago Medical Center hired a non-nurse patient navigator (PN) to facilitate efficient colonoscopy screening for patients. Input data is the sample medical free text notes collected by the PN during the inception of the navigations (N=19), which contain up to 6 interactions per patient. The notes included demographic information, past history, thoughts and concerns expressed by patients, responses and services provided by the PN. The recorded notes were transformed into coded topics and examined by frequencies and by the importance of associations in R. Output association measures were then used to reduce the number of co-occurring sets and also to discover meaningful associations.
Results: Among recurring coded themes, the most often discussed pertained to bowel preparations. The distant second was of transportation arrangements. Actions most often taken by the PN were split dose bowel preparation instructions and a recommendation to follow a low residue diet which were found to be always present when a patient sought procedure information (confidence=1). Among the patients who had previously canceled procedures, a fear of having the procedure has shown the strongest association (lift=1.69).
Conclusions: In the data analysis, quantified association measures helped to both quantify and understand the actions taken by the PN regularly and under certain circumstances. This data mining approach illustrates the use of existing qualitative data in developing theoretical framework for improving health care.
Assessment of individual and community needs for health education Communication and informatics Implementation of health education strategies, interventions and programs Planning of health education strategies, interventions, and programs Public health or related research
Abstract
Breast cancer survival in Missouri, 2004â2012
Chester Schmaltz, PhD1, Kala Visvanathan, MHS, FRACP, MD2, Avonne Connor, MPH, PhD2, Jeannette Jackson-Thompson, MA, MSPH, PhD3 and Jiang Du, MA4
(1)Missouri Cancer Registry and Research Center, University of Missouri-Columbia, Columbia, MO, (2)Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, (3)University of Missouri (MU) School of Medicine and MU Institute for Data Science & Informatics, Columbia, MO, (4)University of Missouri, Columbia, Columbia, MO
APHA's 2018 Annual Meeting & Expo (Nov. 10 - Nov. 14)
Background: The Missouri Cancer Registry (MCR) collects population-based cancer incidence and 1st course treatment data on Missouri residents diagnosed with malignant neoplasms (192.650â192.657 RSMo) and has complete survival information with follow-up through 31 December 2015. As a preliminary analysis for a project in collaboration with outside researchers to incorporate comorbidity information from hospital discharge data, MCR analyzed breast cancer (BC) survival.
Purpose: To present preliminary results on survival using MCR data.
Methods: Women age 18+ diagnosed with BC in Missouri during 2004â2012 were selected from the MCR database. Cases were excluded from the survival analysis that were diagnosed after death (autopsy / death certificate only cases) or that were a second or later tumor in her lifetime. Three end-points for survival were examined: death from any cause, BC death, and non-BC death. Covariates include age at diagnosis (modeled with splines), year of diagnosis (modeled with splines), race, stage at diagnosis, grade at diagnosis, marital status, histology, Rural-Urban Continuum Code 2003 (RUCC2003), and ER/PR status. BC-specific survival was additionally analyzed stratified by race (whites & blacks), metro v non-metro, and age group. Cox (proportional hazards) regression was utilized to model the survival.
Results: For any-cause and BC-specific survival: age and stage were very important predictors for length of survival, and marital status and RUCC2003 moderately so. Additionally, for BC-specific survival: low grade and ER+/PR+ status were associated with longer survival. For non-BC survival the effect sizes for stage, grade, histology, and ER/PR status were greatly diminished. The effect of RUCC2003 was also moderately diminished for non-BC survival. Full results including the stratified analyses will be presented.
Discussion: The results of this preliminary analysis largely cohere with what would have been expected for BC cases but provide a useful baseline in which to compare the results of incorporating comorbid conditions.
Biostatistics, economics Epidemiology Public health or related research
Abstract
Breast cancer and comorbidities in Missouri, 2002â2012
Kala Visvanathan, MHS, FRACP, MD1, Jeannette Jackson-Thompson, MSPH, PhD2, Avonne Connor, MPH, PhD3, Chester Schmaltz, PhD4 and Jiang Du, MA5
(1)Johns Hopkins University, Baltimore, MD, (2)University of Missouri (MU) School of Medicine and MU Institute for Data Science and Informatics, Columbia, MO, (3)Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, (4)University of Missouri-Columbia, Columbia, MO, (5)University of Missouri, Columbia, Columbia, MO
APHA's 2018 Annual Meeting & Expo (Nov. 10 - Nov. 14)
BACKGROUND: Comorbidities such as diabetes and hypertension can significantly impact mortality among women with a recent diagnosis of breast cancer; the extent to which comorbidities can explain disparities in survival is unknown. Missouri has high rates of mortality from female breast cancer as well as a variety of cardiovascular risk factors (diabetes, hypertension and obesity). Comorbidity rates are higher among individuals with certain characteristics (low household income, less education, African American or elderly).
The Missouri Cancer Registry (MCR) has population-based data on cancer incidence and survival but not on comorbidities. By linking MCR data with population-based hospital discharge data, two groups of researchersâone from the Missouri Cancer Registry and Research Center (MCR-ARC) and one from Johns Hopkins University (JHU)âcollaborated on a pilot project for which JHU obtained funding.
OBJECTIVE: Conduct a pilot project to test the hypothesis that the co-existence of breast cancer and comorbidities contributes to disparities in mortality.
METHODS: We developed inclusion criteria for women (Nâ55,000) diagnosed with breast cancer (â¥18, diagnosed 1/1/2002â12/31/2012, Missouri resident) and conducted a preliminary analysis on MCR data. We merged data extracted from the MCR database (previously linked with Missouri death files and the National Death Index) with individual-level data from the Missouri Patient Abstract System (PAS; hospital discharge database for 1/1/2001â12/31/2014). After linkage, MCR researchers removed identifiers and sent the linked MCR-PAS dataset to JHU researchers for further analysis to assess mortality/survivorship by age, race, geographic location (urban v. rural) and income.
RESULTS: Results of preliminary analysis of MCR data are presented separately [abstract ID 419125]. Results of linked dataset analysis, conducted from April through October 2018, will be presented.
DISCUSSION: This study involved secondary analysis of two existing datasets. We underestimated time needed to obtain approvals for multi-step linkage but such linkage offers opportunities for disparity reduction.
Chronic disease management and prevention Communication and informatics Conduct evaluation related to programs, research, and other areas of practice Diversity and culture Epidemiology Public health or related research