routine screening practices during routine health maintenance visits.
In March 2015, we implemented a clinical decision support (CDS) tool, an electronic medical record
(EMR) health maintenance alert, or “flag,” at five clinic
locations in and around St. Paul, Minnesota, within
the HealthPartners healthcare system. This alert within
our EMR flagged providers to order HIV or hepatitis C
screening for eligible patients when accessing their chart
and listed HIV screening or hepatitis C screening as
“due” under the EMR health maintenance section.
Providers adapting this alert at the five sites received
no formal training or education. Eligible patients who
were flagged for HIV screening included patients 18
years of age or older who had never been screened for
HIV within our system. Eligible patients flagged for hepatitis C screening included patients born between 1945
and 1965 who had never had prior screening within our
care system. Two of the five participating clinics serve
urban populations within St. Paul, while the other three
serve suburban populations surrounding St. Paul.
Data regarding screening practices by provider
location and patient age, gender, and race/ethnicity
were analyzed in the 10-month period leading up to
implementation of the EMR alert and in the 10-month
period after implementation. Analysis of this review
focused on patient encounters that were billed as or
associated with a diagnosis code of “routine health
maintenance” in hopes of reviewing one-time screening
trends in our five clinics, as opposed to targeted screening for specific high-risk population groups.
As such, patients screened during routine health
maintenance for HIV or hepatitis C and had already
had one of these tests results prior to that visit were
excluded from that portion of the study. Additionally,
patients over age 65 at the time of the office visit who
were screened for HIV during routine health maintenance were also excluded from the study, as they did
not fit our study population.
We performed statistical analysis of the data
using SAS 9. 4. 24 First, we used descriptive statistics
(frequencies and percentages) to explore distributions of
demographic factors (gender, race) of patients during the
pre- and post-change time periods. Two-by-two contingency tables were created to evaluate the association of
percentage of patients screened with the pre- and post-change periods, and chi-square statistics and associated
p-values were calculated to test this association.
Screening demographics for the HIV group before
and after implementation of the healthcare flag are
listed in Table 1. There were 12,275 screening eligible
patients seen in clinics for routine health maintenance
in the pre-alert time period, while 12,504 patients were
seen in clinic for routine health maintenance post-alert.
Demographically, the percent of each race/ethnicity seen
in the HIV cohort was within 1% pre- and post-EMR
flag. However, significantly more African Americans
were seen in the pre-alert cohort ( 8.5%) than the post-
alert cohort ( 7.5%), while significantly fewer Hispanics
were seen pre-alert ( 2.6%) than post-alert ( 3.1%, p
< 0.01, Table 1). There was no significant difference
between the two genders.
Table 2 shows screening demographics for the
hepatitis C group. There were 7,399 screening eligible
patients seen in clinic for routine health maintenance
pre-alert and 7,797 post-alert. While all races/ethnicities seen were within 1% between the two cohorts, as
with the HIV group, significantly more African Americans were seen in the pre-alert cohort ( 4.5%) than
the post-alert cohort ( 3.8%), while significantly fewer
Hispanics were seen pre-alert ( 1.5%) than post-alert
( 2.0%, p < 0.01, Table 1). There was no significant difference between genders.
Table 3 documents results of HIV screening
practices pre- and post-health maintenance alert, while
results of hepatitis C screening practices pre- and post-health maintenance alert are documented in Table 4.
Overall, significantly more people were screened for
HIV ( 38.2%) and hepatitis C (44%) post-health maintenance alert than pre-health maintenance alert ( 4.9%
and 1%, respectively, p < 0.01). This was true across all
races/ethnicities, and both sexes (p <0.01, Tables 3-4).
This was also true across all five clinic locations, with
significantly more patients screened post-health maintenance alert then pre-health maintenance alert for both
HIV (Table 5) and hepatitis C (Table 6) (p < 0.01).
Our results show that a targeted intervention by
way of an EMR health maintenance flag is successful
at improving screening for HIV and hepatitis C in
compliance with USPSTF and CDC onetime screening
recommendations. We conclude that this intervention
can be effectively implemented across a healthcare
system and can be successfully implemented in both
urban and suburban high-volume outpatient practices
with equal success. Finally, it appears that this is an
effective way to screen appropriate patients of all races,
ethnicities, and genders for HIV and hepatitis C.
EMR prompts have become an important but
underutilized tool in preventative health care. Given the
profound public health consequences that undiagnosed