TrustArc Webinar - Stay Ahead of US State Data Privacy Law Developments
Proposal2_iVoice
1. iVoice
Summary
The goal of this proposed project is to set up, test, and evaluate an iVoice mobile app that will be used to
capture an individual patient’s perceptions of health improvement and quality of services. The data from
iVoicewill support quality improvement for mental health and substance use services in the city of Philadelphia
Behavioral Health System. We are partnered with staff from the Quality Improvement Center for the City of
Philadelphia Department of Behavioral Health to execute this project. iVoice survey data will be generated at
the point of mental health service. Currently, quality improvement does not factor in the patient’s perspective.
Quality measures generated fromiVoice data includes measures of health behavior, quality of life, access, and
quality of services. The iVoice data will be linked with data from the Aligning Forces program to create a
robust profile of evidence to be used to evaluate quality of services and to support for pay-for-performance
decisions by the Philadelphia Quality Improvement Center team. To incentivize user participation we intend to
contribute points to an individuals’ account when they use iVoice. As points accumulate over time, patients’ can
use points to purchase a wide variety of consumer products. In addition, we propose to gamify the point system
by offering prizes for individuals or clinics based on the highest number of points, thereby encouraging
participation. The purpose of creating patient data is to improve the health and the quality of care for the
vulnerable population with mental and substance use conditions.
The Problem
Chronic illnesses are the leading cause of disability and death and, in the U.S., affect almost half the adult
population, or about 133 million Americans. A recent study places chronic care at 78% of total U.S. healthcare
spending, and forecasts costs of over a trillion dollars per year by 2020.1 Among individuals with chronic illness
who incur the highest costs are those with Serious Mental Illness (SMI) who have multiple layers of physical
and/or mental health problems that interfere with their capacity to socialize, plan, organize, and function in their
life. People with serious mental illness (SMI), approximately 15 million Americans, are in the top 5% of
Medicaid beneficiaries for per capita costs and account for more than 50% of all Medicaid spending, with
annual per person costs of $43,130 - $80,374. Despite this level of investment, services for the SMI have major
quality problems. These statistics speak volumes about the ineffectiveness of current health systems to care for
some of the most vulnerable of populations, and the need for innovative solutions to improve the quality and
management of their care.
The goal of this project is to set up, test, and evaluate an iVoice mobile app that will be used to capture an
individual patient’s perceptions of health improvement and quality of services. The data from iVoice will
support quality improvement for mental health and substance use services in the city of Philadelphia Behavioral
Health System. We are partnered with staff from the Quality Improvement Center for the City of Philadelphia
Department of Behavioral Health to execute this project. iVoice survey data will be generated at the point of
mental health and substance use services. The generated data support public administrators to make evidence
based decisions that improve the quality and cost effectiveness of these services.
Advances in mobile phone technology, and the adoption of these devices across all socio-economic strata, has
created a powerful new potential to collect patient perceptions about their health and health experience for
quality improvement. This platform includes native mobile apps and mobile web sites. Communication from
the population to the care teams, public health practitioners, and care quality improvement organizations is
limited and not systematically collected in health settings other than hospitals. The mobile technology platform
is at an inflection point: the functionality is sufficiently reliable and sophisticated, and the device adoption is
widespread. There are mobile apps that have met with some recent success and motivating certain audiences.
These include Eatery, Lift, RunKeeper, and Zamzee. These apps are popular with the “worried well”, generally
people of higher SES and health status who are looking to improve their fitness. These are not the patients
1
2. associated with the bulk of the country’s health care expense: people suffering from serious diseases that are not
being adequately managed, and often with comorbidities and complications.
Strategy
We developed a prototype of iVoice and presented it to the City of Philadelphia Quality Improvement Team in
the Department of Behavioral Health. Cathy Bolton, PhD is the director of quality improvement. She has a
large team of research analysts and quality improvement specialists. Over the past 10 years, she has worked
with the Department of Behavioral Health to improve the quality of services. Currently, patient satisfaction
data comes for face-to-face surveys. Obviously, data collection is time consuming, expensive, and requires
cleaning and manipulation to link it with claims data. Claims data, the primary data source for quality
improvement, is limited to service utilization. A link with patient’s perception of health improvement and
quality of services would be ideal. Mobile apps offer an opportunity to collect patient data at the point of
service. Adding patient satisfaction data to the quality improvement assessment process provides and
unprecedented opportunity. Additionally, linking with other government data sources to obtain contextual
factors provides a robust dataset for analysis and decision making about managing resources.
Population: Currently there are 1.5 million people living in Philadelphia; 527,000 receive Medicaid services.
Of these individuals, 224,654 have at least one administrative claim for a mental health diagnosis (ICD-9-CM
295.xx to 316.xx); 76,187 are adults age 18 years and older and the remainder are children. The population that
was treated for a serious mental illness includes 40,738 adult clients; 30% have an affective disorder; 6% have a
schizophrenia diagnosis; 21% have a substance use disorder, and 43% have another functional disability.
Approximately 68% of these adults have a major medical problem ranked in order of occurrence: diabetes
mellitus, essential hypertension, lumbar pain, respiratory disorders and osteoarthritis and allied neurological
disorders. Thus, approximately 76,187 adults individuals could potentially useiVoice.
Design and Implementation
Systems Diagram
The proposed system is mobile web site accessible with any smartphone, tablet, laptop, or desktop. The
specifics of the system, including the data fields, end-user interfaces, and researcher dashboard interfaces,
would be “co-designed” using rapid-prototyping approach with a sample of actual end-users giving iterative
feedback.
2
3. Sample Screen From Mobile Web App: Soliciting Patient Satisfaction Scores
The technical implementation can be readily
handled by Big Yellow Star Inc. during Phase 2 of
this competition. We are accustomed to an agile
approach to deployment of reliable mobile
technology for health systems and public health
informatics. We would use cloud infrastructure
(multi-zone Amazon Web Services), scalable
application architecture (Ruby on Rails and
MongoDB) and mobile interface standards (HTML5
and jQuery).
The main end-user interface is the mobile web app,
which can include three types of layouts that share
the same functionality: smartphone, tablet, and
laptop/desktop. The second user interface is the
interactive text messaging component. Although the
exact use would need to be decided by the co-design
process, an example scenario is that users set up text
message “triggers” that solicit numeric data input.
The users would receive these according their
preferred schedule and then reply to them as they
arrive. In this way the most routine data can be
collected without the user even opening the mobile
web app. And a third interface, added at a later
phase, can be the automatic incorporation of data
from wireless biometric sensors such as the
Withings scale.
Figure 1.0 provides a graph expressing the quality improvement loop from the collection of patient responses
using iVoice to linking iVoice data with government and claims data to develop and evaluate quality indicators
then providing quality improvement reports to various stakeholders such as government officials, facilities, and
providers. Rationale for pay-for-performance decisions is most enhanced with the iVoice data, particularly for
outpatient services which have not been targeted with pay for performance incentives. Figure 1.0 demonstrates
the potential for the proposed data to identify and then support and evaluate quality improvement in particular
areas. Dr. Bolton and her staff have used pay-for-performance for hospital services but the data for outpatient
services is lacking. Adding the robust proposed dataset provides evidence to support pay-for-performance
decisions because the linked data provides new quality measures that align with national quality measures. The
procedures for data collection follows:
1. The secretary of the clinic assists patients to enter data onto an IPad that is securely attached to the
check out desk. The usual patient flow is to check in for their appointment with the secretary; once
service is completed, the patient checks out with the secretary to handling billing and future
appointment. It is at this point, the secretary will instruct the patient to go to the IPad and answer
questions about their visit. She will help the patient to sign on to a mobile website that will have the
following frames:
a. Login. The patient logs in with their Medicaid number. Typically, 80% of patients with serious
mental health problems such as schizophrenia, mood or other mental status problems are covered
by Medicaid. The secretary will provide the patient with their Medicaid number that the patient
3
4. enters onto the login page. If the patient has problems, the secretary will be trained to assist the
patient.
a. The next pages are a series of 5-10 questions about quality, health and services with likert scale
answers. The questions were developed by the New York Care Coordination Program (NYCCP)
and will be used because psychometric studies support reliability and validity. The measures
also translate well with national quality indicators and are functionally relevant. The questions
were developed using individuals with serious mental illness. We added a feature to the app that
allows the Quality Director to periodically change the questions to get at access, quality and
costs issues. We will use the same databases or tools for data collection.
i. The Enrollee Satisfaction Survey allows patients to rate the services they receive.
ii. Quality of life assessment.
b. The data collected from the app includes the Medicaid number as the unique identifier that can
link to other data such as Medicaid Claims where measures of cost and utilization of services can
be obtained.
2. Platforms for moving, storing, and archiving data are still to be determined and depend on privacy and
security issues. The Quality Improvement team is familiar with the rigors of HIPAA laws of privacy
and security and several attorneys’ work with their program on these challenges.
4
5. Data Measures
Quality measures will be created from linking the Medicaid number of the patient and zip code.
Quality Measures and Data Sources
Measure Data Sources
Health behaviors: tobacco use, diet and exercise, Behavioral Risk Factor Surveillance
alcohol use, sexual activity National Center for Chronic Disease Prevention and
Health-related Quality of Life Health Promotion
iVoice
Health Conditions: HIV, diabetes, obesity. Behavioral Risk Factor Surveillance
Morbidity: poor or fair health, poor physical health iVoice
days, poor mental health days, injury, Sexually Medicaid
transmitted infections
Clinical Care: access, quality of services; utilization iVoice
of hospital, outpatient, emergency room, community Medicaid data
support services and costs of care.
Social and economic factors: education, employment, Behavioral Risk Factor Surveillance
income, family support, safety (violent crime rate) FBI data on violent crime rates
iVoice
Physical environment: air pollution, ozone days, U.S. Environmental Protection Agency
access to services
Analytic Questions
Analytic questions using the integrated data sets will focus on access, quality, and costs. Pay-for-performance
analysis will use patient satisfaction, access, health improvement, and quality. Analysis of cost benefit will
factor patient health improvement and types and quantity of services. Longitudinal questions can be assessed
because the patient is uniquely identified each time iVoiceis used. The County Health Rankings and Measures
can be used to assess population parameters and establish reasonable estimates for expected change. Hot spots
that require targeted quality control can be identified from iVoicethen validated using the County Health data.
Most important, exemplar places and providers of services can be identified and rewarded.
Game/gamification design characteristic that drive engagement and experience
We plan to use a point system for gamifying the design so that participants are engaged in the experience.
Similar to point systems used by airlines or credit cards, the patient would accumulate points each time he/she
uses iVoice. Early implementation may be further incentivized by adding ways for individuals or even clinics to
compete for addition points and buying power. Decisions of the type of gamification incentives will be
reviewed with consumers and other stakeholders to ensure that incentives are indeed attractive to the player.
5
6. Plan for adoption in the community
Figure 3.0 provides a graph of the steps we will take and the timeline for implementing and evaluating the
performance of the app and the data that is generated. The process is iterative until all stakeholders are satisfied
that the app and data meet the goal of the project which is to create and implement an app that captures an
individual patient’s perceptions of health improvement and quality of services.
6
7. Task Start End
Manage system platform, data storage, and data access for QI 5/7/2013 8/22/2014
Build multidisciplinary team of Patients, Quality staff, and clinicians to provide
feedback on feasibility and design 5/22/2013 7/7/2014
Privacy and Security Issues 5/7/2013 6/25/2013
Solve for Access to a unique person system: QR? 6/7/2013 7/26/2014
Set up reward systems; 6/24/2013 8/25/2013
Beta test 1.0 iVoice and point system among Quality Control Staff 6/24/2013 12/25/2013
Correct issues identified in previous beta test 1.0 8/8/2013 11/9/2013
Select one clinic and 5 patients for beta test 2.0 11/14/2013 12/17/2013
Ensure web access 6/24/2013 3/25/2014
Educate secretaries in their role assisting patients to use iVoice. 6/8/2013 6/11/2014
Instruct patients in the use of iVoice, noting usability issues 1/26/2014 1/27/2014
Correct bugs from beta test 2.0 2/2/2014 3/4/2014
Continue to beta test until all bugs are worked out. Vary the clinics; 4/8/2014 5/8/2014
Linking Data and create reports for QI. 5/9/2014 8/10/2014
Test reward points 5/5/2014 5/5/2014
Schedule full role out for all behavioral health clinics 1/8/2014 3/9/2014
Beta test in medical clinics reiterating the process until bugs worked out 5/21/2014 5/21/2014
Identify maintenance tasks 7/21/2014 8/22/2014
7