capacity for public health informatics among local health departments j. mac mccullough, phd, mph...
TRANSCRIPT
Capacity for Public Health Informatics among Local Health Departments
J. Mac McCullough, PhD, MPHAssistant Professor
School for the Science of Health Care DeliveryArizona State University
Acknowledgements
• Co-author: Kate Goodin, MPHProgram Manager, Epidemiology & Data ServicesMaricopa County Department of Public Health
2
Public Health Informatics
• The application of IT and IS to public health practice, research, and learning.
• An evidence-based way of strengthening the work of a public health department– Enhance capacities to perform surveillance,
monitor outbreaks, respond to emergencies, etc.– Can interact with IT from clinical sector to further
boost capacity
3
Current Use of PH Informatics
Informatics System Proportion of LHDs Currently Using
Immunization registry 85.8%Electronic disease registry 75.8%Electronic syndromic surveillance system 66.5%Electronic lab reporting 51.4%Electronic health records (EHRs) 25.1%Health information exchanges (HIEs) 13.9%
4
What’s at Stake
2005 2008 2010 20130%
10%
20%
30%
40%
50%
60%
70%
80%
5
Proportion of Physicians Using EHRs Proportion of LHDs Using EHRs
Not all types of PH informatics are rapidly diffusing through the public health system.
What’s Known & What’s Not Known
• Certain LHDs are more likely to use specific informatics functionalities.
• Not known whether LHDs tend to adopt only specific systems or whether LHDs invest in broader informatics capacities– May be interactions between systems (e.g., use of
EHR, participation in HIE)– May be economies of scale—most hospitals have a
CIO
6
Study Objective
• Objective: to test for patterns in the presence of public health informatics functionalities within LHDs – Accomplished through the creation of an empirical
classification of LHD informatics capacities. – This empirical classification can then be used to
explore correlates of informatics capacity.
7
Methods
• Used secondary data from 2013 NACCHO Profile Survey of Local Health Departments– NACCHO data are the single largest source of data
on LHDs– Conducted regularly, contain data on LHD structure,
finance, services, …. , informatics.– Content can change across years
• Data available on informatics usage from n=505 LHDs from across U.S.
8
Creating a Typology
• Hierarchical cluster analysis used to categorize LHDs according to public health informatics capacity. – Calculated via Ward’s
Method.• Three-cluster measure
was determined to provide optimal combination of data fit and parsimony.
050
100
150
200
250
L2sq
uare
d di
ssim
ilarit
y m
easu
re
G 1 G 2 G 3 G 4 G 5 G 6 G 7 G 8 G 9 G 1 0 G 1 1 G 1 2 G 1 3 G 1 4 G 1 5 G 1 6 G 1 7 G 1 8 G 2 0 G 2 1 G 2 2 G 2 3 G 2 4 G 2 5 G 2 6 G 2 7 G 2 8 G 2 9 G 3 0 G 3 1 G 3 2 G 3 3 G 3 4 G 3 5
Dendrogram for Cluster Analysis
Low HighMid 9
Predictors of Interest: LHD Characteristics
Finances• Per capita revenues• Clinical revenues• State-sources• Federal-sources
Workforce
• FTEs per capita• Any informatics
personnelServices Offered• Provision of ~40
different public health services
Leadership/Governance
• Local board of health
• Freestanding versus part of health and human services agency
• Single county vs. other jurisdiction
• Authority to impose fees
• State vs. local governance
10
After developing typology, use chi2 and t-tests to explore category composition according to:
All data came from 2013 NACCHO Profile
11
Findings:
Type of FunctionalityPercent With Functionality Difference Between Groups
Total Low(n=112)
Mid(n=92)
High (n=255)
Low vs. Mid
Low vs.
High
Mid vs.
HighImmunization Registry 85.8% 49.1% 98.9% 97.3% *** ***Electronic Disease
Registry 75.8% 18.8% 93.7% 93.3% ** ***Electronic Syndromic
Surveillance System 66.5% 47.3% 60.9% 76.9% ***Electronic Lab Reporting 51.4% 17.9% 0.0% 84.7% *** ***Electronic Health Records 25.1% 17.9% 19.6% 30.2% *** *Health Information
Exchange 13.9% 5.4% 6.5% 20.4% *** ***
* p < .05 ** p < .01 *** p < .001
• LHD informatics capacity was clustered into three distinct groups.
• The LHDs with the lowest level of informatics usage had significantly lower levels of usage for all six functionalities assessed.
Characteristics of Low, Mid, High Capacity LHDs
• High capacity LHDs:• Disproportionately serve large populations (> 500,000)• Receive significantly higher revenues from Medicare/Medicaid (likely
means they engage more in direct services and thus bill CMS)• More likely to employ IT personnel
• Low capacity LHDs:• More likely to be multi-county or other complex jurisdiction types• Less likely to have an executive director with a clinical background
12
Public Health Services Offered vs. Informatics Capacity
• Low-capacity LHDs provided significantly fewer public health services than LHDs with mid-or high-levels of informatics capacity (p < .01).
• Differences in service provision:• Most pronounced for Population Focused services
(e.g., STD screening, tobacco prevention, unintended pregnancy)
• Least pronounced for Individual Focused services (e.g., behavioral health, HIV tx, obstetrical care, substance abuse)
13
Discussion
• A diverse matrix of factors appear to impact an LHD’s informatics capacity: • Setting, finances, governance, leadership, and services offered.
• High- and low-capacity LHDs differed across all six informatics capacities• This consistent pattern across all six systems suggests a deficit of
informatics capacities in certain LHDs relative to others.
14
Discussion
• Commonly state-supported applications (e.g., immunization registries) saw higher levels of use among mid-capacity LHDs: • LHD therefore operates more akin to information consumers than
information brokers. • State-level involvement may promote broader informatics capacity
among LHDs.• Association between service provision and informatics capacity
especially prevalent for population-focused public health services • May emphasize the role that informatics plays for specific public
health services and the symbiotic nature of broad-based capacity for public health informatics and broad-based provision of population-focused services.
15
Limitations
• Cross sectional study: study explored associations and correlates and did not seek to ascribe causality.• Partitioned data into training and validation sets. Typology
characteristics remained highly consistent across these two iterations.
• Self-reported data: systematic over- or under-reporting possible, though previous studies found longitudinal consistency in NACCHO informatics data.
• No measures available for intensity or effectiveness of services provided.
16
Conclusion
• Typology represents a new conceptualization of department-wide informatics capacity.
• Some LHDs have strong, broad capacity for informatics, others are lagging.• How can low-capacity LHDs can maximize the value of informatics to their
work and the communities they serve, given their lower levels of service provision relative to high-informatics capacity LHDs?
• A third group is doing well with common (and commonly state-supported) applications but lags in more advanced system capacity.• How can we work to promote adoption of less common technologies?
Consideration to state-level factors may be especially important for these LHDs.
• Future studies might explore the direction and causal nature of the relationship between service provision and informatics capacity.
17
Public Health Services Examined
Individual-focused Population-focusedBasic Home health care Basic Chronic disease programs
Adult immunizations Oral health Blood lead Maternal and child health
Child immunizations Prenatal care Communicable/inf. disease Physical activity
EPSDT Primary care HIV/AIDS STDs
Family planning School health Nutrition Specialized
MCH home visits Well-child clinic Other STDs Chronic disease
WIC Specialized Tuberculosis Injury
Expanded Behavioral or mental Tuberculosis Injury
Cancer HIV/AIDS Tobacco Mental illness
Cardiovascular disease Obstetrical care Unintended pregnancy Substance abuse
Diabetes School-based clinics Expanded Syndromic surveillance
High blood pressure Substance abuse Behavioral risk factors Violence
From Bekemeier et al., Classifying local health departments on the basis of the constellation of services they provide.American Journal of Public Health. 2014;104(12):e77-82.
Transformational PH Informatics: Surveillance
• Traditional disease surveillance– Physicians report specific diseases upon diagnosis Health
department follows up on reported cases.– Even with timely and accurate reporting, not a good method of
identifying emerging outbreaks. What about unreported cases??• Informatics-based surveillance
– E.g., BioSense, automated surveillance system that receives data from hundreds of hospitals nationwide.
– Can mine free-text chief complaint fields to identify disease patterns in nearly real-time.
• Can use a well-functioning disease surveillance system for fundamentally different purposes.
20
Maturation of PH Informatics2005 2008 2010 2013
Use of IT in the field Use of IT in the field
Wireless access to LPHA
Wireless access to LPHA
IT disaster recovery planning
IT disaster recovery planning
Federal IT standards initiatives
Federal IT standards initiatives
Electronic health records (EHRs)
Electronic health records (EHRs)
Electronic health records (EHRs)
Electronic health records (EHRs)
Health Information Exchanges (HIEs)
Health Information Exchanges (HIEs)
Health Information Exchanges (HIEs)
Immunization registry Immunization registry
National health information network
Practice management system
Electronic Disease Reporting system
Electronic Lab reporting
Electronic syndromic surveillance system
Creating a Typology
• Cluster Analysis: assembling observations into groups based on their similarity/dissimilarity on selected measures.
22
Creating a Typology
23
• Hierarchical cluster analysis: method that generates a tree-like structure based on distance/ similarity between observations
• Example: NFL Teams, clustered through 2015 season statistics• 1) Good offensive teams• 2) Good overall teams• 3) Mediocre teams• 4) Good defenses• 5) Bad teams• 6) Inconsistent teams