Factors Associated with Improved MCH Epidemiology Functioning in State Health Agencies Deborah Rosenberg • Amy Herman-Roloff • Joan Kennelly • Arden Handler Published online: 17 September 2010 Ó Springer Science+Business Media, LLC 2010 Abstract This paper discusses characteristics that are associated with enhanced maternal and child health (MCH) epidemiology functioning in state health agencies. The concept of the ‘‘MCH Epidemiology Effort’’ is introduced as ‘‘the epidemiologic work carried out by multiple units and agencies aimed at informing program planning and policy development on behalf of women, children and families.’’ This concept focuses attention on MCH epide- miology functioning at the organizational level rather than on individual MCH epidemiologists. The analysis used data from all 50 states and the District of Columbia. Each state participated in a telephone interview and submitted material that demonstrated the breadth, depth, and capacity of its MCH Epidemiology Effort. Several organizations, including the Council for State and Territorial Epidemiol- ogists, the Health Resources and Services Administration/ Maternal and Child Health Bureau, and the Centers for Disease Control and Prevention provided additional sec- ondary data. The outcome for analysis was a three-category measure of MCH epidemiology functioning. The findings are consistent with, and add specificity to, those from prior assessments. In a multivariable model, agenda-setting by consensus, involvement of external stakeholders, the total of doctorally trained staff, and accessing CDC assignees or other staff were all significantly related to higher level MCH epidemiology functioning (ORs of 6.1, 6.6, 2.5, and 6.4, respectively; P \ 0.05). Organizational visibility of the MCH Epidemiology Effort and a data environment marked by routine data-sharing and data integration were marginally related. We provide recommendations for action at the state and federal level for advancing evidence- based decision-making in maternal and child health. Keywords MCH epidemiology Á Public health agency capacity Á Analytic capacity Á Capacity assessment Background Maternal and child health (MCH) epidemiology emerged as a distinct field over the past 20 years as state and local health agencies transitioned from the delivery of personal health services to carrying out the core functions of public health. The transition requires agencies to commit them- selves to data-based decision-making for program planning, evaluation, policy development, and advocacy, to hire individuals with different types of skills and abilities, and to provide additional training to current staff. To support this transition, the Health Resources and Services Administra- tion/Maternal and Child Health Bureau (HRSA/MCHB) and the Division of Reproductive Health/Centers for Dis- ease Control and Prevention (DRH/CDC) and its partners have implemented a variety of initiatives to enhance the analytic capacity of state and local health agencies. Pivotal among these initiatives was the creation of the CDC/HRSA MCH Epidemiology Program (MCHEP), which has assigned MCH epidemiologists to public health agencies since 1986 [1]. These individuals serve as senior scientists to provide state MCH agencies with the analytic leadership necessary to engage in data-based D. Rosenberg (&) Á A. Herman-Roloff Epidemiology and Biostatistics, School of Public Health, University of Illinois at Chicago, 1603 W. Taylor, Chicago, IL 60612, USA e-mail: [email protected]J. Kennelly Á A. Handler Community Health Sciences, School of Public Health, University of Illinois at Chicago, 1603 W. Taylor, Chicago, IL 60612, USA 123 Matern Child Health J (2011) 15:1143–1152 DOI 10.1007/s10995-010-0680-x
10
Embed
Factors Associated with Improved MCH Epidemiology Functioning in State Health Agencies
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Factors Associated with Improved MCH EpidemiologyFunctioning in State Health Agencies
Deborah Rosenberg • Amy Herman-Roloff •
Joan Kennelly • Arden Handler
Published online: 17 September 2010
� Springer Science+Business Media, LLC 2010
Abstract This paper discusses characteristics that are
associated with enhanced maternal and child health (MCH)
epidemiology functioning in state health agencies. The
concept of the ‘‘MCH Epidemiology Effort’’ is introduced
as ‘‘the epidemiologic work carried out by multiple units
and agencies aimed at informing program planning and
policy development on behalf of women, children and
families.’’ This concept focuses attention on MCH epide-
miology functioning at the organizational level rather than
on individual MCH epidemiologists. The analysis used
data from all 50 states and the District of Columbia. Each
state participated in a telephone interview and submitted
material that demonstrated the breadth, depth, and capacity
of its MCH Epidemiology Effort. Several organizations,
including the Council for State and Territorial Epidemiol-
ogists, the Health Resources and Services Administration/
Maternal and Child Health Bureau, and the Centers for
Disease Control and Prevention provided additional sec-
ondary data. The outcome for analysis was a three-category
measure of MCH epidemiology functioning. The findings
are consistent with, and add specificity to, those from prior
assessments. In a multivariable model, agenda-setting by
consensus, involvement of external stakeholders, the total
of doctorally trained staff, and accessing CDC assignees or
other staff were all significantly related to higher level
MCH epidemiology functioning (ORs of 6.1, 6.6, 2.5, and
6.4, respectively; P \ 0.05). Organizational visibility of
the MCH Epidemiology Effort and a data environment
marked by routine data-sharing and data integration were
marginally related. We provide recommendations for
action at the state and federal level for advancing evidence-
based decision-making in maternal and child health.
Community Health Sciences, School of Public Health,
University of Illinois at Chicago, 1603 W. Taylor, Chicago,
IL 60612, USA
123
Matern Child Health J (2011) 15:1143–1152
DOI 10.1007/s10995-010-0680-x
decision-making [1, 2]. Other federal initiatives include
the provision of trainings and workshops, publication of
analytic methods workbooks, support of pre-doctoral and
doctoral training as well as student internships, and the
establishment of the CDC/Council of State and Territorial
Epidemiologists (CSTE) MCH Epidemiology Fellows
Program [3]. In parallel to these workforce development
activities, the Title V Block Grant has provided supple-
mental funding to states through the State Systems
Development Initiative (SSDI) to facilitate improvement in
the data infrastructure.
Over the past decade, there have been several efforts to
assess the capacity of the state MCH epidemiology enter-
prise [2, 4–9]. While the tools developed for these
assessments acknowledged the multi-dimensional nature of
MCH epidemiology functioning, they focused on describ-
ing existing capacity and not on identifying factors related
to the variation in capacity across the states. The analysis
presented here takes the next step by examining factors—
singly and in combination—that are associated with
improved MCH epidemiology functioning.
Methods
Conceptual Framework
This analysis used a version of the conceptual framework
described by Handler et al. [10] to examine public health
performance. Figure 1 illustrates the adaptation of this
framework to MCH epidemiology, including examples of
indicators which correspond to the structure-process-output
components of the framework. Note that structure is divided
into: (1) Health Agency structure, and (2) MCH Epidemi-
ology Effort structure. We defined the state MCH Epide-
miology Effort as the epidemiologic work carried out by
multiple units and agencies aimed at informing program
planning and policy development on behalf of women,
children and families.
Data Collection
We conducted telephone interviews lasting 1–1.5 hours with
all 50 states and Washington, DC between March and July,
2006. (The word ‘‘state’’ is used to refer to all 51 jurisdic-
tions.) The interview questions focused on organizational as
opposed to individual competencies. Typically, 2 study
personnel and 3–5 state staff participated in the interview,
including an MCH data spokesperson and the Title V
Director, or designee. Additionally, each state submitted a
portfolio of data sharing agreements, reports, presentations,
policy briefs, a listing of routine data linkages, databases
used in the Title V Needs Assessment, and other documents
illustrating the breadth, depth, and capacity of its MCH
Epidemiology Effort.
We also accessed other capacity-related data sources
[11] to inform the development of our interview questions.
A few data elements from these other sources were directly
incorporated into analysis, but the findings presented here
are predominantly from our primary data collection effort.
MA
CR
O C
ON
TE
XT
Population Context
Public Health
Context
MCH-Epi Effort Process
1.Provides analytic direction2.Provides expertise for data integration*3.Conducts high-level data analysis
Output
1.Disseminates reports*2.Generates policy brief s
Health Status
Intermediate Outcome
1.Program improvement*2.Policy development
Health Agency Structure
1.Sufficient funding for continuing education and staff development2.Updated database and web servers, statistical analysis software,
GIS software, etc.
MCH-Epi Effort Structure
1.Sufficient staff for data collection, analysis, dissemination, etc.2.Participation in the leadership of the agency
Examples: Organizational Feedback Loop
Integrating / linking data is a process ; when the process is ongoing, it is also a structural feature.Dissemination is an output; when diissemination is expected and routine, it is also a structural feature.
Program improvement is an intermediate outcome ; when improvement is an agency imperative, it is also a structural feature.
*
*
*
Fig. 1 Conceptual framework
1144 Matern Child Health J (2011) 15:1143–1152
123
Analytic Approach
We created a summary measure to characterize each state’s
level of MCH epidemiology functioning based on the
review of: (1) the interview responses, and (2) the materials
in the state portfolios. Four members of the research team
independently assigned two scores to each state–one based
on the interview, and the other based on the portfolio; these
8 scores (2 assessments by 4 reviewers) were averaged. We
then coded the state averages into three categories, classi-
fying each state as ‘‘below average’’, ‘‘average’’, or ‘‘above
average’’, reflecting relative levels of MCH epidemiology
functioning. This is the outcome variable in all analyses.
During the interview, we asked about several factors
hypothesized to be related to the level of MCH Epidemi-
ology functioning. These independent variables included
the following structural (#1–5), process (#6–7), and output
(#8–9) factors:
1. An identifiable MCH Epidemiology Unit
2. An identifiable leader of the MCH Epidemiology
Effort
3. A collaborative approach to setting the MCH Epide-
miology Effort agenda
4. High-level staff with access to continued training
5. Incorporation of CDC assignees, fellows, and interns
6. Internal/external data sharing and routine data integra-
tion (linkage)
7. Sophisticated epidemiologic analysis
8. Well-developed approaches for dissemination
9. High-level interpretation and translation of data
We assessed the crude relationship between each of
these factors and the outcome measure of MCH epidemi-
ology functioning, and in addition used crude cumulative
logistic and crude generalized logistic regression models to
further assess whether the magnitude of any association
was consistent (or close to consistent) across the range of
the outcome measure. In the generalized logistic models,
we separately examined the dichotomous contrasts of
above average versus average and average versus below
average functioning. Finally, we built multivariable mod-
els, again using cumulative and generalized logistic
regression to consider level of functioning as either an
ordinal or nominal outcome, with attention to the test of
proportional odds. We first specified models with subsets
of related variables and then combined the important
variables from each into a single overall model. Population
size and resources of a state (defined by median household
income, percent college graduates, percent foreign born,
and percent of children under 5 years who are uninsured)
were also considered.
We present tests of statistical significance, but they
should be considered with caution. With a total of 51
observations, statistical power to detect differences is
constrained, particularly for multivariable modeling. In
fact, statistical power dictated that only results of cumu-
lative logit models be presented, with findings from sepa-
rate comparisons of above average versus average, and
average versus below average only described when rele-
vant. Given the inherent limitations in sample size, we
considered associations with odds ratios of 1.5 or greater,
and p-values of less than 0.20 of interest to report.
Results
Table 1 shows the distribution of states on key indicators.
Thirty-nine reported having an identifiable MCH Epide-
miology Effort, with or without a named organizational
home; twelve reported a weak or nonexistent MCH Epi-
demiology Effort. In 36 states the MCH epidemiology
agenda was established through consensus; in twenty of
these, partners included external stakeholders. Fifteen
states reported agenda-setting driven solely by a few key
staff or no unified agenda-setting process. In 36 states there
was an identified leader of the MCH Epidemiology Effort;
more than half had hosted CDC assignees, fellows, or
interns. Only eighteen states were classified as having a
well-developed data infrastructure.
Sixteen states reported routinely conducting multivari-
able analysis, and thirty reported that MCH epidemiolo-
gists provide interpretation and translation of data. Of the
21 states reporting limited data translation, ten answered
that translation is not the job of MCH epidemiologists.
Approximately half of the states reported that evidence
produced by the MCH Epidemiology Effort guided pro-
gram development and modification.
On the summary outcome measure of MCH epidemi-
ology functioning, approximately one-third of the states
were classified as above average, average, or below aver-
age. (Note: the authors were under no constraints regarding
the categorization of states on the range of functioning.)
We identified eight states as small and with fewer resour-
ces, and we hypothesized that they may be meaningfully
different from the other states. Four of these states were
classified in the average category and the other four in the
below average category—none were classified in the above
average group (data not shown).
Table 2 shows results for four cumulative logit models.
Having an increasingly formalized MCH Epidemiology
Effort in the state agency is associated with a higher level
of functioning, as is having an agenda setting process based
on consensus. Having a lead MCH epidemiologist was not
associated with increased functioning (Table 2a). The total
number of doctorally trained key staff, and involving fel-
lows and interns were both statistically significant aspects
Matern Child Health J (2011) 15:1143–1152 1145
123
Table 1 Percent distribution on
select characteristics of MCH
epidemiology capacity 50 states
and the District of Columbia
# %
Organizational position of the MCH Epidemiology Effort
Named unit in Title V or in epidemiology section 21 41.2
No named unit, but recognized presence in Title V
and/or agency-wide
18 35.3
Diffuse/weak or no MCH Epidemiology Effort 12 23.5
Total 51 100.0
Process for setting the agenda of the MCH Epidemiology Effort
Consensus: MCH EPI staff, Title V director,
and external partners
20 39.2
Consensus: MCH EPI staff and Title V director 12 23.5
Consensus: MCH EPI staff 4 7.8
Lead MCH EPI and Title V director 7 13.7
Diffuse, program specific 8 15.7
Total 51 100.0
Identified leader of the MCH Epidemiology Effort
Yes 36 70.6
No 15 29.4
Total 51 100.0
CDC assignee some time during 1987–2005?
Yes 19 37.3
No 32 62.8
Total 51 100.0
Ever have assignees, fellows, and interns
Yes 29 56.9
No 22 43.1
Total 51 100.0
Permissive data sharing environment and routine data integrationa
Yes 18 35.3
No 33 64.7
Total 51 100.0
Data analysis: incorporate multivariable methods
Very frequently or frequently 16 31.4
Sometimes, rarely, or never 35 68.6
Total 51 100.0
Involvement in interpretation and translation of data
Provide interpretation and recommendations 30 58.8
Provide little interpretation or recommendations 21 41.2
Total 51 100.0
Translation is part of the job of MCH epidemiologists
Yes 41 80.4
No 10 19.6
Total 51 100.0
Data used for program development
Very frequently or frequently 25 49.0
Sometimes, rarely, or almost never 26 51.0
Total 51 100.0
Data used for program modification
Very frequently or frequently 28 54.9
Sometimes, rarely, or almost never 23 45.1
Total 51 100.0
1146 Matern Child Health J (2011) 15:1143–1152
123
of the MCH Epidemiology Effort. Having CDC assignees
and[40% of MCH Epidemiology Effort staff identified as
epidemiologists also appeared to be associated with higher
functioning, although odds ratios were unstable (Table 2b).
An environment that promotes and permits data sharing
both internally and externally, and routine data integration,
were both markers of higher level functioning (Table 2c.).
While the use of multivariable analysis by the MCH Epi-
demiology Effort was not found to be significantly asso-
ciated with higher level functioning, 55.6% of the 18 states
with a well-developed data environment frequently used
multivariable methods, while only 18.1% with a less con-
ducive data environment frequently did this work (data not
shown). Finally, publishing at least sometimes in the peer
reviewed literature and submitting abstracts to (and pre-
sumably presenting at) the annual MCH Epidemiology
Conference were both associated with higher levels of
functioning (Table 2d).
Table 3 synthesizes the results described in Table 2. The
variables in this final model are structural indicators—no
process, output, or intermediate outcome indicators showed
a strong enough association with level of functioning to be
included. An increasingly formal organizational presence,
having a consensus approach to agenda-setting, specifically
involving external stakeholders in agenda-setting, having
staff with advanced training, supplementing staff with
external assignees, fellows, and interns, and having a more-
developed data infrastructure were all associated with
improved MCH epidemiology functioning after mutual
adjustment. It is noteworthy that involvement of external
partners contributed to higher functioning, both in terms of
participation in agenda-setting (OR 6.6, 95% CI 1.3–33.2)
and possibly as a marker of an advanced data infrastructure