Content uploaded by Jean Omalley
Author content
All content in this area was uploaded by Jean Omalley on Nov 14, 2017
Content may be subject to copyright.
Receipt of Preventive Services After Oregon’s Randomized
Medicaid Experiment
Miguel Marino, PhD1,2, Steffani R. Bailey, PhD1, Rachel Gold, PhD, MPH3,4, Megan J.
Hoopes, MPH3, Jean P. O’Malley, MPH2, Nathalie Huguet, PhD1, John Heintzman, MD1,
Charles Gallia, PhD5, K. John McConnell, PhD6, and Jennifer E. DeVoe, MD, DPhil1,3
1Department of Family Medicine, Oregon Health and Science University, Portland, Oregon
2Department of Public Health and Preventive Medicine, Division of Biostatistics, Oregon Health
and Science University, Portland, Oregon
3OCHIN, Inc. Portland, Oregon
4Kaiser Permanente Northwest Center for Health Research, Portland, Oregon
5Office of Health Analytics, Oregon Health Authority, Salem, Oregon
6Center for Health System Effectiveness, Department of Emergency Medicine, Oregon Health
and Science University, Portland, Oregon
Abstract
Introduction—It is predicted that gaining health insurance via the Affordable Care Act will
result in increased rates of preventive health services receipt in the U.S, primarily based on self-
reported findings from previous health insurance expansion studies. This study examined the long-
term (36-month) impact of Oregon’s 2008 randomized Medicaid expansion (“Oregon
Experiment”) on receipt of 12 preventive care services in community health centers using
electronic health record data.
Methods—Demographic data from adult (aged 19–64 years) Oregon Experiment participants
were probabilistically matched to electronic health record data from 49 Oregon community health
centers within the OCHIN community health information network (N=10,643). Intent-to-treat
analyses compared receipt of preventive services over a 36-month (2008–2011) period among
those randomly assigned to apply for Medicaid versus not assigned, and instrumental variable
analyses estimated the effect of actually gaining Medicaid coverage on preventive services receipt
(data collected in 2012–2014; analysis performed in 2014–2015).
Results—Intent-to-treat analyses revealed statistically significant differences between patients
randomly assigned to apply for Medicaid (versus not assigned) for eight of 12 assessed preventive
services. In intent-to-treat[MM1] analyses, Medicaid coverage significantly increased the odds of
receipt of most preventive services (ORs ranging from 1.04 [95% CI=1.02, 1.06] for smoking
assessment to 1.27 [95% CI=1.02, 1.57] for mammography).
Address correspondence to: Miguel Marino, PhD, Oregon Health and Science University, Department of Family Medicine, Mailcode:
FM, 3181 SW Sam Jackson Park Road, Portland OR 97239. marinom@ohsu.edu.
HHS Public Access
Author manuscript
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Published in final edited form as:
Am J Prev Med. 2016 February ; 50(2): 161–170. doi:10.1016/j.amepre.2015.07.032.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Conclusions—Rates of preventive services receipt will likely increase as community health
center patients gain insurance through Affordable Care Act expansions. Continued effort is needed
to increase health insurance coverage in an effort to decrease health disparities in vulnerable
populations.
Introduction
In the U.S., lack of health insurance is associated with decreased access to health care,
including lower receipt of recommended preventive services among uninsured patients
compared with insured patients.1–8 The 2010 Affordable Care Act (ACA) created new
opportunities for millions of uninsured people to obtain health insurance.9,10 It is predicted
that ACA coverage opportunities will positively affect rates of receipt of preventive services
as uninsured patients become insured.11–13 These predictions are largely based on data that
could be influenced by unmeasured external factors. For instance, increased rates of
preventive services receipt among people who gain insurance coverage as a result of a
significant life event (e.g., getting a new job) could confound how change in insurance status
affects preventive care receipt. To estimate the causal effect of gaining health insurance on
receipt of preventive services, researchers examined “natural experiments” in which
individuals gained coverage owing to a policy change such as Massachusetts’ 2006 health
insurance expansion. Most of these studies were observational or quasi-experimental and
relied on self-reported data, which could explain why their findings were inconsistent.14–18
Randomizing patients to receive an intervention provides the strongest design to assess
causal relationships; however, it is nearly impossible to conduct a study that randomizes
insurance coverage. The “Oregon Experiment,” a randomized natural experiment, provided
a unique opportunity to isolate the effect of health insurance on preventive services
receipt.19–22 In 2008, Oregon expanded Medicaid coverage to a limited number of “non-
categorically eligible” individuals (i.e., those not federally mandated to receive Medicaid). It
was anticipated that the number of people that signed up for coverage would exceed the
expansion budget; thus, to most fairly allocate limited resources, interested adults were
added to a list and were randomly selected to apply for Medicaid coverage. From a
“reservation list” of >100,000 entries, approximately 30,000 people were randomly selected
to apply, and approximately 10,000 gained coverage.23 Detailed information about Oregon’s
Medicaid program in 200823,24 and the Oregon Experiment is available elsewhere.19,21,23,25
This study utilized this randomized natural experiment to assess the impact of gaining
Medicaid coverage on receipt of preventive services among community health center (CHC)
patients. The authors hypothesized that those who were selected to apply and gained
Medicaid would receive more preventive services than those who did not gain Medicaid
coverage.
An ideal setting for isolating the effect of insurance, CHCs provide care for millions of
patients, regardless of insurance coverage status or ability to pay.26 CHCs also care for a
high percentage of racial/ethnic minority patients and others likely to have low rates of
preventive services and to experience healthcare disparities.27 Thus, CHC patients would
likely be among those most affected by a policy change to expand Medicaid coverage.
Marino et al. Page 2
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
To limit bias inherent in self-reported measures, this study utilized electronic health record
(EHR) data from 49 CHCs. It assessed participants’ receipt of preventive services, as
documented in the EHR, in the 36 months after the Oregon Experiment. This is the first
study to focus on the impact of the Oregon Experiment on receipt of preventive care services
in CHCs utilizing EHR data.
Methods
Data Sources
This study used EHR data from the OCHIN community health information network, a
501(c)(3) network of health systems that supports >300 CHC clinic sites by providing a
centrally hosted EpicCare EHR with an enterprise-wide master patient index (each patient
has a single medical record available across the network). Originally called the Oregon
Community Health Information Network, its official name became “OCHIN, Inc.” as
membership expanded beyond Oregon. Detailed information about OCHIN and the
suitability of OCHIN’s EHR database for research purposes is available elsewhere.28–30 The
authors identified all Oregon CHC sites in the OCHIN network that were live on EHR as of
March 11, 2008 (N=49), which was the earliest date a patient could have received Medicaid
via the Oregon Experiment. State administrative data, including the Oregon Experiment
reservation list (names, addresses, and other contact details provided to sign up for the
chance to gain Medicaid) and Oregon’s Medicaid enrollment data were also used to assess
periods of Medicaid coverage during the study period.
Study Population
Individuals on the Oregon Experiment “reservation list” (N=100,407) were probabilistically
matched to individual OCHIN patients (N=106,692), using Link Plus software31 and
demographic variables common to both data sets. Two researchers independently performed
a case-by-case review of uncertain matches using additional demographic variables.
Appendix Table 1 provides more details. The authors identified 11,041 matched individuals,
4,205 of whom were selected to apply for coverage, and 6,836 who were not selected. To
preserve randomization, minimal exclusions were applied: patients aged <19 years (n=8)
and >64 years (n=337), patients not alive at the end of the post-period (n=60), and those
with unknown sex (n=1). This led to an exclusion of 156 (3.7%) from the selected group and
242 (3.5%) from the non-selected group. The final study population consisted of 10,643
patients: 4,049 selected to apply for coverage and 6,594 not selected.
Measures
Random selection to apply for Medicaid coverage occurred through eight monthly drawings
held between March and October 2008. Among selected individuals, coverage start dates
were retroactively assigned as the date of selection notification (the “selection date”). For
analyses, a 2008 selection date was randomly assigned to individuals not selected to apply
based on the distribution of observed selection dates among people selected to apply.
Outcomes were assessed in the 36 months after the selection date (post-period).
Marino et al. Page 3
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
To examine the effect of providing access to apply for health insurance on the receipt of
preventive services, intent-to-treat (ITT) analyses were conducted comparing patients
randomly selected to apply for Medicaid coverage (i.e., intervention group) versus those not
selected to apply (i.e., control group). However, the ITT approach does not provide a causal
estimate of obtaining Medicaid insurance. For example, individuals randomly selected to
apply for insurance did not always follow through, and thus remained uninsured. To
estimate the effect of gaining Medicaid on receipt of preventive services, bivariate probit
instrumental variable (IV) analyses were conducted. To be considered a valid instrument for
IV analyses, the variable(s) must be associated with Medicaid coverage, but not associated
with the receipt of preventive services in the relevant time period except through the
instrument’s effect on Medicaid coverage. Based on these criteria, two instrumental
variables that met the standards for valid instruments were used: (1) selection status in the
Oregon Experiment (randomly selected to apply, or not)20,32; and (2) Medicaid coverage
status in the pre-period (any coverage or no coverage). Both variables were positively and
significantly associated with post-period coverage, but neither would be expected to be
directly associated with post-period preventive service receipt except through their
association with post-period coverage. The treatment variable was having at least 6 months
of continuous Medicaid coverage in the post-period starting from their selection date, as
participants who received Medicaid were covered for 6 months before they had to reapply to
renew coverage. The ITT and IV analyses are presented together in this study to identify the
effects of gaining access to apply for health insurance (ITT) and actually gaining Medicaid
(IV), which are two different experiences.
The primary outcomes were whether or not the patient received preventive care services in
the post-period: screenings for cervical, breast, and colorectal cancer (fecal occult blood
testing and colonoscopy); screenings for diabetes (glucose and hemoglobin A1c [HbA1c]),
hypertension, obesity, smoking; lipid screening; chlamydia testing; and receipt of influenza
vaccination. Codes were used based on EHR Meaningful Use Stage 1 measures.33 These
included ICD-9-CM diagnosis and procedure codes, Current Procedural Terminology and
Healthcare Common Procedure Coding System codes, Logical Observation Identifiers
Names and Codes, and medication codes. The authors also used relevant code groupings and
codes specific to the OCHIN EHR, used for Meaningful Use reporting and internal quality
improvement initiatives.34 Appendix Table 2 provides detailed technical specifications and
patient eligibility criteria for each measure.
For covariates, this study used EHR data to obtain patient age, sex, race/ethnicity, household
income, and baseline health status prior to each patient’s selection date. Patients’ household
income was estimated as the average of available Federal Poverty Level from all visits. To
measure baseline health status, prior diagnosis of five chronic conditions was assessed using
standard Meaningful Use criteria33 or Healthcare Effectiveness Data and Information Set35
codes: asthma, coronary artery disease, diabetes, dyslipidemia, and hypertension. If a
qualifying diagnosis code appeared on the problem list or in two or more encounters prior to
the selection date, the patient was considered to have the condition.
Marino et al. Page 4
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Statistical Analysis
Differences were assessed in the covariates between patients randomly selected to apply for
Medicaid versus those not selected to apply, using chi-square tests for categorical
characteristics and two-sample t-tests for continuous predictors. This was done for every
preventive service outcome separately, and covariates that displayed significant differences
between the selection groups were included in adjusted analyses. Next, ITT analyses were
conducted for each outcome, comparing preventive service receipt in the 36 month post-
period among those randomly selected to apply versus not selected using generalized
estimating equation models with a logit link and robust sandwich variance estimator to
account for the clustering of patients within CHCs.
A maximum-likelihood bivariate probit IV model36 was used, as it has been shown to be
more consistent and less biased for models with binary outcomes and binary endogenous
variables compared with the common two-stage least-squares model.37 The bivariate probit
model controlled for the same covariates included in the ITT models. A robust variance
estimator that account for within-clinic correlation was implemented.38,39 The validity of the
instruments was tested using an over-identification test. All statistical tests were two-sided
and significance was defined as a p-value <0.05. Statistical analyses were performed using
SAS, version 9.3 and Stata, version 12.1 (data collected in 2012–2014; analysis performed
in 2014–2015). This study was approved by the IRB at Oregon Health and Science
University and was registered as an observational study at clinicaltrials.gov
(NCT02355132).
Results
A total of 10,643 participants with an average age of 39.2 years at baseline were followed
for 36 months after random selection to apply for coverage (Table 1). About 59% of
participants had no chronic conditions documented in the EHR in the pre-period, and 60%
were non-Hispanic white. There were no significant differences at baseline between the
selected and not selected groups in gender, age, Federal Poverty Level, or race/ethnicity.
The groups differed on the number of chronic conditions prior to selection date.
Table 2 presents the percentage of participants who received preventive services during the
36-month post-period by selection group and the ORs of receipt of preventive services,
comparing those participants selected to apply versus those not selected. Patients selected to
apply had significantly higher odds of receiving assessments of BMI (AOR[MM2]=1.12,
95% CI=1.10, 1.14), blood pressure (AOR=1.09, 95% CI=1.07, 1.12), and smoking status
(AOR=1.04, 95% CI=1.02, 1.06). Statistically significant increases in the odds of receipt of
preventive services were also observed when comparing selected with not selected groups
for the following outcomes: 15% increase for Pap test, 27% increase for mammography,
15% increase for fecal occult blood testing, and 24% increase for chlamydia testing.
Additionally, participants who were selected to apply had lower odds of receiving HbA1c
testing (AOR=0.79, 95% CI=0.71, 0.88), compared with participants not selected.
Among the 4,049 patients that were selected to apply, 44% actually gained Medicaid
coverage (≥6 months of continuous coverage) in the post-period. Table 3 displays the effects
Marino et al. Page 5
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
of gaining Medicaid coverage on receipt of preventive services estimated by the IV
approach. The bivariate probit model estimates indicated that participants who had Medicaid
coverage in the post-period had significantly higher likelihood of receiving BMI (increase of
12.5%, 95% CI=10.6, 14.4), blood pressure (increase of 10.1%, 95% CI=7.0, 13.3), and
smoking (increase of 6.2%, 95% CI=5.3, 7.1) screenings, compared with those who did not
have Medicaid coverage. Among cancer-related screenings, statistically significant
Medicaid coverage effects were observed for Pap testing (10.3% increase, 95% CI=8.8,
11.7) and mammography (14.5% increase, 95% CI= 10.1, 18.8). A positive Medicaid
coverage effect was also observed for chlamydia testing (increase of 27.3%, 95% CI=14.1,
40.4) and lipid screening (increase of 8.0%, 95% CI=1.0, 15.0). No significant effect of
Medicaid coverage was found on receipt of fecal occult blood testing, colonoscopy, glucose,
or HbA1c screenings.
Discussion
Previous studies of the Oregon Experiment examined the impact of a Medicaid expansion on
self-reported healthcare utilization and service receipt in the general population.21,22,25,40
This study extends that work to evaluate the effect of a Medicaid expansion on receipt of
preventive services in CHCs, a setting likely to be impacted by ACA Medicaid expansions,
as most CHC patients are uninsured or Medicaid recipients.41 This study also expands on
prior examination of the Oregon Experiment by including a longer follow-up (36 months)
and using EHR data from a linked system to objectively measure receipt of preventive
services.20,21 EHR data can overcome potential biases that result when asking patients to
recall service receipt, particularly over a long period of time.42–44 The randomization
component of the Oregon Experiment enabled examination of [MM3]both the effect of
being selected to apply for Medicaid coverage on utilization of CHC services, and the
isolated effect of actually gaining Medicaid coverage.
The findings strengthen the survey-based evidence from other Oregon Experiment studies
regarding the causal link between health insurance and receipt of breast and cervical cancer
screening.20,21 Similar to those studies no significant effect of insurance on influenza
vaccination was found. Interestingly, Medicaid coverage positively affected screenings for
BMI, blood pressure, and smoking (not assessed in previous Oregon Experiment studies),
despite the fact that these services are usually performed at most visits and do not generate a
separate billing charge. Based on post-hoc analyses (results not shown), one possible
explanation for these findings is that insured patients had a higher primary care office visit
rate than those who were uninsured, increasing the odds that these routine services would be
performed. It should be noted that although the authors did not find statistically significant
differences in glucose and HbA1c testing and colorectal cancer screenings by insurance
coverage, there was a trend toward Medicaid coverage having a positive effect on these
outcomes.
Policy Implications
These findings can help inform what to expect as an increasing number of uninsured patients
gain coverage via the ACA insurance expansions.45 CHCs provide critical access to millions
Marino et al. Page 6
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
of uninsured and underinsured Americans and do an excellent job of providing quality
services.27,46–51 However, previous studies show that without insurance coverage, CHC
patients cannot always obtain all recommended services; having a primary care medical
home and health insurance coverage is optimal.4,52–55 The finding that CHC patients who
gained insurance coverage in the Oregon Experiment had increased rates of receipt of many
preventive care services suggests that ACA Medicaid expansions could potentially lead to
better access to preventive healthcare services for many Americans. Another ACA provision
that requires most health plans to cover evidence-based preventive services without cost
sharing will further increase this access: About 71 million Americans with private insurance
gained access to fully covered preventive services in 2010–2011 with no co-pay.56 Without
cost sharing, it is reassuring that discussion of preventive care receipt for insured patients
may no longer have to include whether or not a patient can afford the out-of-pocket costs
that used to be associated with many of these services. However, it is important to remember
that an estimated 30 million Americans might remain uninsured, despite ACA insurance
expansions.57 The findings also suggest that these people are much less likely to receive
many recommended preventive services, and that continued effort is needed to increase
access to insurance and health care in this population.
Limitations
Analyses were limited to the 49 Oregon CHCs that had fully implemented the OCHIN EHR
before March 11, 2008 to be able to fully capture documented data on preventive services
over the 36-month follow-up. This resulted in small sample sizes and reduced power for
some preventive service categories, likely explaining the trend toward Medicaid coverage
having a positive effect on colon cancer and glucose screenings but not reaching the level of
statistical significance. This study was conducted in Oregon CHCs; patients seeking care
outside this state and setting may behave differently. Further, the majority of the sample was
already receiving care at the Oregon CHCs prior to the Oregon Experiment; thus, the results
may not generalize to other patient populations such as those seeking care for the first time
via the ACA, or among patients less engaged in their health care. The observed percentage
receipt of most screening outcomes during the 36-month follow-up was slightly lower than
other studies,51 likely because the authors did not limit the sample to patients with a primary
care visit during the post-period (68% of the selected group and 66% of those not selected
had one or more primary care visit in the post-period). The authors also were unable to
assess the extent to which patients sought care outside the OCHIN network. If a patient
gained insurance and left the OCHIN network, this would diminish the percentage receipt of
preventive services for the Medicaid coverage group and thus bias the effects towards the
null.28 Additionally, gaining Medicaid was defined as having ≥6 months of continuous
Medicaid coverage in the post-period. If subjects in the Medicaid group lost coverage after 6
months, this could adversely affect preventive service receipt later in the study period; thus,
the observed treatment effects may be underestimated.
Conclusions
Utilizing the Oregon Experiment, a randomized natural experiment, this study demonstrates
a causal relationship between Medicaid coverage and receipt of several preventive services
Marino et al. Page 7
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
in CHC patients, including receipt of breast and cervical cancer screenings as well as
screenings for BMI, blood pressure, and smoking, during a 3-year follow-up.
Acknowledgments
This study was supported by grants R01HL107647 from the National Heart, Lung, and Blood Institute and K08
HS021522–02 from the Agency for Healthcare Research and Quality. The funding agencies had no involvement in
the preparation, review, or approval of the manuscript. We thank Heather Angier and Eve Dexter for their
contributions, and gratefully acknowledge the OCHIN community health centers and Practice-Based Research
Network.
References
1. Bailey SR, O’Malley JP, Gold R, Heintzman J, Likumahuwa S, DeVoe JE. Diabetes care quality is
highly correlated with patient panel characteristics. J Am Board Fam Med. 2013; 26(6):669–679.
http://dx.doi.org/10.3122/jabfm.2013.06.130018. [PubMed: 24204063]
2. Freeman JD, Kadiyala S, Bell JF, Martin DP. The causal effect of health insurance on utilization,
outcomes in adults: a systematic review of U.S studies. Med Care. 2008; 46(10):1023–1032. http://
dx.doi.org/10.1097/MLR.0b013e318185c913. [PubMed: 18815523]
3. Buchmueller TC, Grumbach K, Kronick R, Kahn JG. The effect of health insurance on medical care
utilization and implications for insurance expansion: A review of the literature. Med Care Res Rev.
2005; 62(1):3–30. http://dx.doi.org/10.1177/1077558704271718. [PubMed: 15643027]
4. DeVoe J, Fryer G, Phillips R, Green L. Receipt of preventive care among adults: insurance status
and usual source of care. Am J Public Health. 2003; 93:786–791. http://dx.doi.org/10.2105/AJPH.
93.5.786. [PubMed: 12721145]
5. Duru OK, Vargas RB, Kermah D, Pan D, Norris KC. Health insurance status and hypertension
monitoring and control in the United States. Am J Hypertens. 2007; 20(4):348–353. http://
dx.doi.org/10.1016/j.amjhyper.2006.11.007. [PubMed: 17386339]
6. McMorrow S, Kenney GM, Goin D. Determinants of Receipt of Recommended Preventive
Services: Implications for the Affordable Care Act. Am J Public Health. 2014; 104(12):2392–2399.
http://dx.doi.org/10.2105/AJPH.2013.301569. [PubMed: 24432932]
7. Cowburn S, Carlson M, Lapidus J, Heintzman J, Bailey S, DeVoe J. Insurance continuity and
human papillomavirus vaccine uptake in Oregon and California federally qualified health centers.
Am J Public Health. 2014; 104(9):e71–79. http://dx.doi.org/10.2105/AJPH.2014.302007. [PubMed:
25033154]
8. Holden CD, Chen J, Dagher RK. Preventive care utilization among the uninsured by race/ethnicity
and income. Am J Prev Med. 2015; 48(1):13–21. http://dx.doi.org/10.1016/j.amepre.2014.08.029.
[PubMed: 25442235]
9. Kaiser Family Foundation. [Accessed June 10, 2015] The Medicaid program at a glance. Publication
7235-05; 2012.www.kff.org/medicaid/7235.cfm.
10. Kaiser Family Foundation. [Accessed June 10, 2015] How is the ACA Impacting Medicaid
Enrollment?. 2014. http://kff.org/medicaid/issue-brief/how-is-the-aca-impacting-medicaid-
enrollment/.
11. Abraham JM. How might the Affordable Care Act’s coverage expansion provisions influence
demand for medical care? Milbank Q. 2014; 92(1):63–87. http://dx.doi.org/
10.1111/1468-0009.12041. [PubMed: 24597556]
12. Ku L, Jones K, Shin P, Bruen B, Hayes K. The states’ next challenge--securing primary care for
expanded Medicaid populations. N Engl J Med. 2011; 364(6):493–495. http://dx.doi.org/10.1056/
NEJMp1011623. [PubMed: 21268720]
13. Schoen C, Hayes SL, Radley DC, Collins SR. Access to Primary and Preventive Health Care
Across States Prior to the Coverage Expansions of the Affordable Care Act. Commonwealth Fund.
2014; 17 (Issue Brief).
Marino et al. Page 8
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
14. Clark CR, Soukup J, Riden H, et al. Preventive care for low-income women in Massachusetts post-
health reform. J Womens Health (Larchmt). 2014; 23(6):493–498. http://dx.doi.org/10.1089/jwh.
2013.4612. [PubMed: 24798240]
15. Okoro CA, Dhingra SS, Coates RJ, Zack M, Simoes EJ. Effects of Massachusetts Health Reform
on the Use of Clinical Preventive Services. J Gen Intern Med. 2014; 29(9):1287–1295. http://
dx.doi.org/10.1007/s11606-014-2865-2. [PubMed: 24789625]
16. Keating NL, Kouri EM, He Y, West DW, Winer EP. Effect of Massachusetts health insurance
reform on mammography use and breast cancer stage at diagnosis. Cancer. 2013; 119(2):250–258.
http://dx.doi.org/10.1002/cncr.27757. [PubMed: 22833148]
17. Kolstad J, Kowalkski AE. The impact of health care reform on hospital and preventive care:
Evidence from Massachusetts. J Public Econ. 2012; 96:909–929. http://dx.doi.org/10.1016/
j.jpubeco.2012.07.003. [PubMed: 23180894]
18. Van Der Wees PJ, Zaslavsky AM, Ayanian JZ. Improvements in health status after Massachusetts
health care reform. Milbank Q. 2013; 91(4):663–689. http://dx.doi.org/10.1111/1468-0009.12029.
[PubMed: 24320165]
19. Allen H, Baicker K, Finkelstein A, Taubman S, Wright BJ, Group OHS. What the Oregon health
study can tell us about expanding Medicaid. Health Aff (Millwood). 2010; 29(8):1498–1506.
http://dx.doi.org/10.1377/hlthaff.2010.0191. [PubMed: 20679654]
20. Baicker K, Taubman SL, Allen HL, et al. The Oregon experiment--effects of Medicaid on clinical
outcomes. N Engl J Med. 2013; 368(18):1713–1722. http://dx.doi.org/10.1056/NEJMsa1212321.
[PubMed: 23635051]
21. Finkelstein A, Taubman S, Wright B, et al. The Oregon Health Insurance Experiment: Evidence
from the First Year. Q J Econ. 2012; 127(3):1057–1106. http://dx.doi.org/10.1093/qje/qjs020.
[PubMed: 23293397]
22. Taubman SL, Allen HL, Wright BJ, Baicker K, Finkelstein AN. Medicaid Increases Emergency-
Department Use: Evidence from Oregon’s Health Insurance Experiment. Science. 2014;
343(6168):263–268. http://dx.doi.org/10.1126/science.1246183. [PubMed: 24385603]
23. Oregon Division of Medical Assistance Programs. [Accessed June 10, 2015] OHP Standard
Reservation List Final Report. 2008. www.oregon.gov/oha/healthplan/DataReportsDocs/OHP
%20Standard%20Reservation%20List%20Final%20Report.pdf.
24. Oregon Department of Human Services, Programs DoMA. [Accessed June 2, 2015] Oregon Health
Plan Annual Report: Medicaid and State Children’s Health Insurance Program Section 1115(a)
Medicaid demonstration extension. 2008. www.oregon.gov/oha/healthplan/DataReportsDocs/FFY
%202008%20Annual%20Report.pdf.
25. Baicker K, Taubman SL, Allen HL, et al. The Oregon experiment--effects of Medicaid on clinical
outcomes. N Engl J Med. 2013; 368(18):1713–1722. http://dx.doi.org/10.1056/NEJMsa1212321.
[PubMed: 23635051]
26. National Association of Community Health Centers. [Accessed June 10, 2015] United States
Health Center Fact Sheet. 2012. Available from: www.nachc.org/client//US12.pdf.
27. The White House. The Obama Administration and Community Health Centers. Washington, D.C:
2012. www.whitehouse.gov/sites/default/files/05-01-12_community_health_center_report.pdf.
[Accessed June 10, 2015]
28. Devoe JE, Gold R, McIntire P, Puro J, Chauvie S, Gallia CA. Electronic health records vs
Medicaid claims: completeness of diabetes preventive care data in community health centers. Ann
Fam Med. 2011; 9(4):351–358. http://dx.doi.org/10.1370/afm.1279. [PubMed: 21747107]
29. Devoe JE, Gold R, Spofford M, et al. Developing a network of community health centers with a
common electronic health record: description of the Safety Net West Practice-based Research
Network (SNW-PBRN). J Am Board Fam Med. 2011; 24(5):597–604. http://dx.doi.org/10.3122/
jabfm.2011.05.110052. [PubMed: 21900444]
30. Devoe JE, Sears A. The OCHIN community information network: bringing together community
health centers, information technology, and data to support a patient-centered medical village. J
Am Board Fam Med. 2013; 26(3):271–278. http://dx.doi.org/10.3122/jabfm.2013.03.120234.
[PubMed: 23657695]
Marino et al. Page 9
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
31. U.S. DHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion.
[Accessed June 10, 2015] Registry Plus, a suite of publicly available software programs for
collecting and processing cancer registry data. 2010. www.cdc.gov/cancer/npcr/.
32. Sussman JB, Hayward RA. An IV for the RCT: using instrumental variables to adjust for treatment
contamination in randomised controlled trials. BMJ. 2010; 340:c2073. http://dx.doi.org/10.1136/
bmj.c2073. [PubMed: 20442226]
33. Center for Medicaid and Medicare Services. [Accessed June 10, 2015] 2011-2012 Eligible
Professional Clinical Quality Measures (CQMs). 2012. http://ushik.ahrq.gov/
MeaningfulUseMeasures?system=mu&enableAsynchronousLoading=true.
34. Heintzman J, Bailey SR, Hoopes MJ, et al. Agreement of Medicaid claims and electronic health
records for assessing preventive care quality among adults. J Am Med Inform Assoc. 2014; 21(4):
720–724. http://dx.doi.org/10.1136/amiajnl-2013-002333. [PubMed: 24508767]
35. National Committee for Quality Assurance (NCQA). HEDIS 2013: Technical Specifications for
Physician Measurement. Washington, DC: 2013.
36. Heckman J. Dummy Endogenous Variables in a Simultaneous Equation System. Econometrica.
1978; 46(6):931–959. http://dx.doi.org/10.2307/1909757.
37. Greene, W. Econometric analysis. 5th ed. Upper Saddle River, NJ: Prentice Hall; 2003.
38. Williams RL. A note on robust variance estimation for cluster-correlated data. Biometrics. 2000;
56(2):645–646. http://dx.doi.org/10.1111/j.0006-341X.2000.00645.x. [PubMed: 10877330]
39. Rogers W. Regression standard errors in clustered samples. Stata Technical Bulletin. 1994; 3(13):
19–23.
40. Baicker K, Finkelstein A. The effects of Medicaid coverage--learning from the Oregon experiment.
N Engl J Med. 2011; 365(8):683–685. http://dx.doi.org/10.1056/NEJMp1108222. [PubMed:
21774703]
41. Kaiser Commission on Medicaid and the Uninsured, Kaiser Family Foundation. [Accessed June
10, 2015] Community Health Centers: A 2012 Profile and Spotlight on Implications of State
Medicaid Expansion Decisions. 2014 Sep. http://kaiserfamilyfoundation.files.wordpress.com/
2014/09/8624-community-health-centers-a-2012-profile-and-implications-of-state-medicaid-
expansion-decisions.pdf.
42. Bowling A. Mode of questionnaire administration can have serious effects on data quality. J Public
Health (Oxf). 2005; 27(3):281–291. http://dx.doi.org/10.1093/pubmed/fdi031. [PubMed:
15870099]
43. Choi BC, Pak AW. A catalog of b iases in questionnaires. Prev Chronic Dis. 2005; 2(1):A13.
[PubMed: 15670466]
44. Tisnado DM, Adams JL, Liu H, et al. What is the concordance between the medical record and
patient self-report as data sources for ambulatory care? Med Care. 2006; 44(2):132–140. http://
dx.doi.org/10.1097/01.mlr.0000196952.15921.bf. [PubMed: 16434912]
45. Decker SL. In 2011 nearly one-third of physicians said they would not accept new Medicaid
patients, but rising fees may help. Health Aff (Millwood). 2012; 31(8):1673–1679. http://
dx.doi.org/10.1377/hlthaff.2012.0294. [PubMed: 22869644]
46. Shi L, Stevens GD. The role of community health centers in delivering primary care to the
underserved: experiences of the uninsured and Medicaid insured. J Ambul Care Manage. 2007;
30(2):159–170. http://dx.doi.org/10.1097/01.JAC.0000264606.50123.6d. [PubMed: 17495685]
47. Bruen BK, Ku L, Lu X, Shin P. No evidence that primary care physicians offer less care to
Medicaid, community health center, or uninsured patients. Health Aff (Millwood). 2013; 32(9):
1624–1630. http://dx.doi.org/10.1377/hlthaff.2012.1300. [PubMed: 24019368]
48. Goldman LE, Chu PW, Tran H, Romano MJ, Stafford RS. Federally qualified health centers and
private practice performance on ambulatory care measures. Am J Prev Med. 2012; 43(2):142–149.
http://dx.doi.org/10.1016/j.amepre.2012.02.033. [PubMed: 22813678]
49. Angier H, Hoopes M, Gold R, et al. An early look at rates of uninsured safety net clinic visits after
the affordable care act. Ann Fam Med. 2015; 13(1):10–16. http://dx.doi.org/10.1370/afm.1741.
[PubMed: 25583886]
50. National Association of Community Health Centers. [Accessed June 10, 2015] A sketch of
community health centers. Chart book. 2014. www.nachc.com/client//Chartbook_2014.pdf.
Marino et al. Page 10
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
51. Heintzman J, Marino M, Hoopes M, et al. Using electronic health record data to evaluate
preventive service utilization among uninsured safety net patients. Prev Med. 2014; 67:306–310.
http://dx.doi.org/10.1016/j.ypmed.2014.08.006. [PubMed: 25124279]
52. Gold R, DeVoe J, Shah A, Chauvie S. Insurance continuity and receipt of diabetes preventive care
in a network of federally qualified health centers. Med Care. 2009; 47(4):431–439. http://
dx.doi.org/10.1097/MLR.0b013e318190ccac. [PubMed: 19330890]
53. DeVoe JE, Tillotson CJ, Lesko SE, Wallace LS, Angier H. The case for synergy between a usual
source of care and health insurance coverage. J Gen Intern Med. 2011; 26:1059–1066. http://
dx.doi.org/10.1007/s11606-011-1666-0. [PubMed: 21409476]
54. Gold R, DeVoe JE, McIntire PJ, Puro JE, Chauvie SL, Shah AR. Receipt of diabetes preventive
care among safety net patients associated with differing levels of insurance coverage. J Am Board
Fam Med. 2012; 25(1):42–49. http://dx.doi.org/10.3122/jabfm.2012.01.110142. [PubMed:
22218623]
55. Bailey SR, O’Malley JP, Gold R, Heintzman J, Marino M, DeVoe JE. Receipt of diabetes
preventive services differs by insurance status at visit. Am J Prev Med. 2015; 48(2):229–233.
http://dx.doi.org/10.1016/j.amepre.2014.08.035. [PubMed: 25442228]
56. U.S. DHHS. [Accessed June 10, 2015] Affordable Care Act Rules on Expanding Access to
Preventive Services for Women. HHS.gov/HealthCare. 2013. www.hhs.gov/healthcare/facts/
factsheets/2011/08/womensprevention08012011a.html.
57. Kaiser Family Foundation. [Accessed June 10, 2015] Adults who remained uninsured at the end of
2014. 2015 Jan 29. http://kff.org/report-section/adults-who-remained-uninsured-at-the-end-
of-2014-issue-brief/
Appendix Table 1
Demographic Characteristics of Patients on the Medicaid ‘Reservation List’ That Were
Probabilistically Matched to OCHIN EHR Data and Included in the Study Sample Versus
Those That Were Not
Study sample
N=11,041
Not in study
sample
N=89,366
no. (column %) no. (column %)
Gender
Female 6,034 (54.7) 47,080 (52.7)
Male 5,007 (45.3) 42,286 (47.3)
Age
Mean (SD)a39.9 (12.5) 39.8 (13.3)
Language
English 9,195 (83.3) 72,222 (80.8)
Spanish 542 (4.9) 3,918 (4.4)
Other/Unknown 1,304 (11.8) 13,226 (14.8)
Urban-rural status
Urban 10,809 (97.9) 83,635 (93.6)
Rural 197 (1.8) 5,441 (6.1)
Unknown 35 (0.3) 290 (0.3)
Note: To identify individuals common to both the Medicaid reservation list and the OCHIN patient population, we used
LinkPlus software to probabilistically compare demographic variables contained in both datasets. Matching variables
included first and last name, date of birth, gender, street address, city, Oregon Medicaid identification number, and
preferred language. The software generates a “match score” indicating each pair’s likelihood of being a match. For pairs of
Marino et al. Page 11
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
uncertain match status based on match score, we conducted double clerical review by independent reviewers. We also
completed several rounds of quality assurance analyses to verify the validity of our match results.
aTwo-sample t-test
Appendix Table 2
Outcome Measure Specifications
Metric Denominator Numerator Areas of EHR
included in search
BMI assessment All patients Patients in the denominator
with at least one of the
following documented in
measurement period: MU
ICD-9-CM grouper1
Standard Concept Id N_c160
(physical exam finding: BMI
percentile); ICD-9-CM
diagnosis codes V85.0-
V85.4; Height and weight
recorded at same encounter.
Encounter vital signs,
Encounter diagnoses,
Problem list
Blood pressure assessment All patients Patients in the denominator
with systolic and diastolic
blood pressure recorded at
same encounter(s) in study
period.
Encounter vital signs
Smoking status assessment All patients Patients in the denominator
with smoking status recorded
at one or more encounters in
measurement period.
Social history2
Cervical cancer screening Female patients ages 21–
64.
Exclusions: EHR
documentation of
hysterectomy.
Hysterectomy: MU
CPT grouper1 Standard
Concept Id N_c273
(procedure performed:
hysterectomy);
MU ICD-9-CM
grouper1 Standard
Concept Id N_c274
(procedure performed:
hysterectomy);
hysterectomy noted as
reason for no last
menstrual period in
encounter record.
Patients in the denominator
with at least one of the
following documented in
measurement period: MU
ICD-9-CM grouper1
Standard Concept Id N_c279
(laboratory test result: pap
test); MU CPT grouper1
Standard Concept Id N_c277
(laboratory test result: pap
test); MU HCPCS grouper1
Standard Concept Id N_c278
(laboratory test result: pap
test).
Hysterectomy:
Encounter diagnoses,
Problem list,
Procedures, Medical
history, Surgical
history;
Screening codes:
Encounter diagnoses,
Labs, Procedures,
Problem list, Health
maintenance3
Breast cancer screening Female patients age ≥40.
Exclusions: EHR
documentation of
bilateral mastectomy.
Mastectomy: MU CPT
grouper1 Standard
Concept Id N_c79
(procedure performed:
unilateral mastectomy)
Patients in the denominator
with at least one of the
following documented in
measurement period: MU
CPT grouper1 Standard
Concept Id N_c72 (diagnostic
study performed: breast
cancer screening); MU
HCPCS grouper1 Standard
Concept Id N_c73 (diagnostic
study performed: breast
cancer screening); MU
ICD-9-CM grouper1
Standard Concept Id N_c74
(diagnostic study performed:
breast cancer screening);
OCHIN internal use codes
for equivalent procedures and
referrals.
Bilateral mastectomy:
Surgical history;
Screening codes:
Encounter diagnoses,
Procedures, Referrals,
Health maintenance3
Marino et al. Page 12
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Metric Denominator Numerator Areas of EHR
included in search
Colorectal cancer screening Patients age ≥50.
Exclusions: EHR
documentation of
colorectal cancer history,
total colectomy,
completed colonoscopy
within 10 years, or
completed flexible
sigmoidoscopy within 5
years.
Colorectal cancer
history: MU ICD-9-CM
grouper1 Standard
Concept Id N_c520
(diagnosis active/
inactive/resolved:
colorectal cancer). Total
colectomy: MU CPT
grouper1 Standard
Concept Id N_c36
(procedure performed:
total colectomy).
Colonoscopy and
flexible sigmoidoscopy:
see Numerator column.
Patients in the denominator
with at least one of the
following documented in
measurement period: MU
CPT grouper1 Standard
Concept Ids: N_c18
(procedure performed:
colonoscopy), N_c29
(procedure performed:
flexible sigmoidoscopy),
N_c13 (laboratory test
performed: FOBT); MU
HCPCS grouper1 Standard
Concept Ids: N_c32
(procedure performed:
colonoscopy), N_c30
(procedure performed:
flexible sigmoidoscopy),
N_c17 (laboratory test
performed: FOBT); HCPCS
code G0120; MU ICD-9-CM
grouper1 Standard Concept
Id N_c15 (laboratory test
performed: FOBT); MU
LOINC grouper1 Standard
Concept Id N_c16 (laboratory
test performed: FOBT);
OCHIN internal use codes
for equivalent labs and
referrals.
Colorectal cancer
history and total
colectomy: Encounter
diagnoses, Procedures,
Problem list, Medical
history, Surgical
history; Screening
codes: Encounter
diagnoses, Labs,
Procedures, Problem
list, Surgical history,
Referrals, Health
maintenance3
Colonoscopy Patients age ≥50.
Exclusions: EHR
documentation of
colorectal cancer history,
total colectomy,
completed colonoscopy
within 10 years, or
completed flexible
sigmoidoscopy within 5
years.
Colorectal cancer
history: MU ICD-9-CM
grouper1 Standard
Concept Id N_c520
(diagnosis active/
inactive/resolved:
colorectal cancer). Total
colectomy: MU CPT
grouper1 Standard
Concept Id N_c36
(procedure performed:
total colectomy).
Colonoscopy and
flexible sigmoidoscopy:
see Numerator column.
Patients in the denominator
with at least one of the
following documented in
measurement period: MU
CPT grouper1 Standard
Concept Id N_c18 (procedure
performed: colonoscopy);
MU HCPCS grouper1
Standard Concept Id N_c32
(procedure performed:
colonoscopy); OCHIN
internal use codes for
equivalent referrals.
Colorectal cancer
history and total
colectomy: Encounter
diagnoses, Procedures,
Problem list, Medical
history, Surgical
history; Screening
codes: Procedures,
Surgical history,
Referrals, Health
maintenance3
Fecal occult blood test
(FOBT) Patients age ≥50.
Exclusions: EHR
documentation of
colorectal cancer history,
total colectomy,
completed colonoscopy
within 10 years, or
completed flexible
sigmoidoscopy within 5
years.
Colorectal cancer
history: MU ICD-9-CM
grouper1 Standard
Concept Id N_c520
(diagnosis active/
inactive/ resolved:
Patients in the denominator
with at least one of the
following documented in
measurement period: MU
CPT grouper1 Standard
Concept Id N_c13 (laboratory
test performed: FOBT); MU
HCPCS grouper1 Standard
Concept Id N_c17 (laboratory
test performed: FOBT); MU
ICD-9-CM grouper1
Standard Concept Id N_c15
(laboratory test performed:
FOBT); MU LOINC
grouper1 Standard Concept
Colorectal cancer
history and total
colectomy: Encounter
diagnoses, Procedures,
Problem list, Medical
history, Surgical
history; Screening
codes: Encounter
diagnoses, Procedures,
Labs, Problem list,
Referrals, Health
maintenance3
Marino et al. Page 13
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Metric Denominator Numerator Areas of EHR
included in search
colorectal cancer). Total
colectomy: MU CPT
grouper1 Standard
Concept Id N_c36
(procedure performed:
total colectomy).
Colonoscopy and
flexible sigmoidoscopy:
see Numerator column.
Id N_c16 (laboratory test
performed: FOBT); OCHIN
internal use codes for
equivalent labs.
Chlamydia screening Sexually active female
patients ages 19–24.
Codes indicative of
sexually active woman:
MU CPT grouper1
Standard Concept Id
N_c207 (procedure
performed: procedures
indicative of sexually
active woman); MU
HCPCS grouper1
Standard Concept Id
N_c208 (procedure
performed: procedures
indicative of sexually
active woman); MU
ICD-9-CM grouper1
Standard Concept Id
N_c580 (diagnosis
active: sexually active
woman); MU LOINC
grouper1 Standard
Concept Id N_c210
(laboratory test
performed: Laboratory
tests indicative of
sexually active woman);
Internal OCHIN
grouper “Diagnosis
Concept: Sexually
Transmitted Disease”;
Social History2 sexually
active flag.
Patients in the denominator
with at least one of the
following documented in
measurement period: MU
LOINC grouper1 Standard
Concept Id N_c219
(laboratory test result:
chlamydia screening);
Internal OCHIN grouper
“Health Maintenance –
Chlamydia Satisfying
Procedure”.
Sexually active codes:
Encounter diagnoses,
Labs, Procedures,
Problem list, Social
history2; Screening
codes: Labs,
Procedures, Health
maintenance3
Cholesterol screening Patients age ≥20. Patients in the denominator
with at least one of the
following documented in
measurement period: MU
CPT grouper1 Standard
Concept Ids: N_c156
(laboratory test performed:
LDL), N_c183 (laboratory
test performed: HDL),
N_c180 (laboratory test
performed: Total
Cholesterol), N_c186
(laboratory test performed:
Triglycerides); MU LOINC
grouper1 Standard Concept
Ids: N_c157 (laboratory test
performed: LDL), N_c184
(laboratory test performed:
HDL), N_c181 (laboratory
test performed: Total
Cholesterol), N_c187
(laboratory test performed:
Triglycerides).
Procedures, Labs,
Health maintenance3
Influenza vaccine Patients age ≥50.
Exclusions:
Documentation of
vaccine allergy/
Patients in the denominator
with at least one of the
following documented in
measurement period: CPT
codes 90653, 90654, 90655,
Exclusion: Allergies,
Immunizations;
Vaccine codes:
Immunizations,
Marino et al. Page 14
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Metric Denominator Numerator Areas of EHR
included in search
contraindication or
patient declined. 90656, 90657, 90658, 90659,
90660, 90661, 90662, 90663,
90664, 90666, 90667, 90668,
90672, 90685, 90686, 90687,
90688, G0008, Q2038;
OCHIN internal use codes
for equivalent procedures and
immunizations.
Procedures, Health
maintenance3
Pneumococcal vaccine Patients with a diagnosis
of diabetes, asthma, or
coronary artery disease
by start of measurement
period. Exclusions:
Documentation of
vaccine allergy/
contraindication or
patient declined.
Diabetes: MU ICD-9-
CM grouper1 Standard
Concept Id N_c47
(diagnosis active:
diabetes). Asthma: MU
ICD-9-CM grouper1
Standard Concept Id
A_c221 (diagnosis
active: asthma). CAD:
MU ICD-9-CM
grouper1 Standard
Concept Id A_c122
(diagnosis active:
Coronary Artery Disease
includes MI); MU CPT
grouper1 Standard
Concept Id A_c169
(procedure performed:
Cardiac Surgery).
Patients in the denominator
with at least one of the
following documented in
measurement period: MU
RxNorm grouper1 Standard
Concept Id N_c421
(medication administered:
pneumococcal vaccination);
CPT codes 4040F, 90669,
90670, 90732, G0009, S0195;
OCHIN internal use codes
for equivalent procedures.
Exclusion: Allergies,
Immunizations;
Diabetes, asthma, and
CAD diagnoses:
Encounter diagnoses,
Problem list, Surgical
history; Vaccine codes:
Procedures,
Medications,
Immunizations, Health
maintenance3
Glucose screening Patients age ≥45. Patients in the denominator
with at least one of the
following documented in
measurement period: LOINC
codes 1492–8, 1494–4,
1496-9, 1499-3, 1501-6,
1502-4, 1504-0, 1507-3,
1508-1, 1514-9, 1515-6,
1518-0, 1530-5, 1531-3,
1533-9, 1554-5, 1557-8,
1558-6, 6749-6, 9375-7,
10450-5, 14753-8, 14754-6,
14756-1, 14757-9, 14759-5,
14764-5, 14765-2, 14771-0,
14995-5, 17865-7, 20436-2,
20437-0, 20438-8, 25666-9,
26554-6, 30251-3, 30265-3,
30267-9, 32320-4, 40285-9,
40286-7, 41024-1, 49134-0,
51597-3, 55351-1, 55381-8,
10449-7, 12610-2, 12646-6,
1521-4, 2345-7, 25428-4,
27353-2, 1527-1, 1469-9,
1539-6, 1542-0, 2348-1,
2349-9, 6760-3; CPT codes
80047, 80048, 80053, 80050,
80069, 82947, 82950, 82951,
82948, 82952; OCHIN
internal use codes for
equivalent procedures.
Procedures, Labs
HbA1c measurement Patients with a diagnosis
of diabetes. Diabetes:
MU ICD-9-CM
grouper1 Standard
Patients in the denominator
with at least one of the
following documented in
measurement period: MU
Diabetes diagnosis:
Encounter diagnoses,
Problem list; Screening
codes: Procedures,
Marino et al. Page 15
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Metric Denominator Numerator Areas of EHR
included in search
Concept Id N_c47
(diagnosis active:
diabetes).
LOINC grouper1 Standard
Concept Id N_c87 (laboratory
test result: HbA1c test); N
groupers: “Health
Maintenance – Hemoglobin
A1C procedures”; “Health
Maintenance – Diabetes
HgbA1c satisfying
procedures”
Labs, Health
maintenance3
CPT, Current Procedural Terminology; EHR, electronic health record; HCPCS, Healthcare Common Procedure Coding
System; LOINC, Logical Observation Identifiers Names and Codes; MU, Meaningful Use
1Meaningful Use groupers created for OCHIN reporting based on specified Standard Concept identifier from Clinical
Quality Measures for Eligible Professionals. www.ushik.ahrq.gov/mdr/portals.
2Social history is the area of the EHR used to record sexuality and substance use
3Health maintenance is the EHR’s preventive health tool that is used to remind patients and providers when appropriate
preventive services are due
Appendix Table 3A
Demographic Characteristics for BMI, Blood Pressure, and Smoking Screening
Subpopulation Study Sample by Oregon OCHIN Patients Selected to Apply for Health
Insurance Coverage via Oregon Experiment vs. Not Selected to Apply (N=10,643)
Selected
N=4,049 Not selected
N=6,594 p-valuea
no. (column %) no. (column %)
Gender 0.734
Female 2,231 (55.1) 3,611 (54.8)
Male 1,818 (44.9) 2,983 (45.2)
Age
Mean (SD)b39.2 (11.7) 39.5 (11.9) 0.185
Race/Ethnicity 0.263
Hispanic, any race 548 (13.5) 867 (13.2)
Non-Hispanic, white 2,447 (60.4) 3,949 (59.9)
Non-Hispanic, other 795 (19.6) 1,412 (21.4)
Unknown 259 (6.4) 366 (5.6)
Average FPLc0.827
<100% 2,911 (71.9) 4,772 (72.4)
≥100% 1,101 (27.2) 1,766 (26.8)
Missing/Unknown 37 (0.9) 56 (0.9)
Number of chronic conditions diagnosed prior to selection dated<0.001
0 2,394 (59.1) 3,936 (59.7)
1 567 (14) 1,083 (16.4)
2 261 (6.5) 442 (6.7)
3–5 135 (3.3) 272 (4.1)
No data to assess 692 (17.1) 861 (13.1)
Marino et al. Page 16
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL, federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s
exact test due to low cell counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Appendix Table 3B
Demographic Characteristics for Pap Screening Subpopulation Sample by Oregon OCHIN
Patients Selected to Apply for Health Insurance Coverage via Oregon Experiment vs. Not
Selected to Apply (N=4,931)
Selected
N=1,872 Not selected
N=3,059 p-valuea
no. (column %) no. (column %)
Gender NA
Female 1,872 (100.0) 3,059 (100.0)
Male 0 (0.0) 0 (0.0)
Age
Mean (SD)b37.9 (10.9) 38.6 (11.4) 0.032
Race/Ethnicity 0.263
Hispanic, any race 338 (18.1) 516 (16.87)
Non-Hispanic, white 1,061 (56.7) 1,710 (55.9)
Non-Hispanic, other 364 (19.4) 663 (21.7)
Unknown 109 (5.8) 170 (5.6)
Average FPLc0.970
<100% 1,325 (70.8) 2,167 (70.8)
≥100% 531 (28.4) 868 (28.4)
Missing/Unknown 16 (0.9) 24 (0.8)
Number of chronic conditions diagnosedprior to selection dated0.008
0 1,197 (63.9) 1,934 (63.2)
1 237 (12.7) 467 (15.3)
2 110 (5.9) 183 (6.0)
3–5 52 (2.8) 106 (3.5)
No data to assess 276 (14.7) 369 (12.1)
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL, federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s
exact test due to low cell counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Marino et al. Page 17
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Appendix Table 3C
Demographic Characteristics for Mammography Screening Subpopulation Study Sample by
Oregon OCHIN Patients Selected to Apply for Health Insurance Coverage via Oregon
Experiment vs. Not Selected to Apply (N=3,661)
Selected
N=979 Not selected
N=1,682 p-valuea
no. (column %) no. (column %)
Gender NA
Female 979 (100.0) 1,682 (100.0)
Male 0 (0.0) 0 (0.0)
Age
Mean (SD)b49.0 (5.8) 49.4 (6.0) 0.128
Race/Ethnicity 0.340
Hispanic, any race 105 (10.7) 161 (9.6)
Non-Hispanic, white 633 (64.7) 1,065 (63.3)
Non-Hispanic, other 186 (19) 366 (21.8)
Unknown 55 (5.6) 90 (5.4)
Average FPLc0.652
<100% 692 (70.7) 1,206 (71.7)
≥100% 278 (28.4) 465 (27.7)
Missing/Unknown 9 (0.9) 11 (0.7)
Number of chronic conditions diagnosed prior to selection dated0.126
0 442 (45.2) 788 (46.9)
1 192 (19.6) 337 (20.0)
2 106 (10.8) 191 (11.4)
3–5 59 (6.03) 120 (7.1)
No data to assess 180 (18.4) 246 (14.6)
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL, federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s
exact test due to low cell counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Marino et al. Page 18
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Appendix Table 3D
Demographic Characteristics for FOBT and Colonoscopy Screening Subpopulation Sample
by Oregon OCHIN Patients Selected to Apply for Health Insurance Coverage via Oregon
Experiment vs. Not Selected to Apply (N=2,531)
Selected
N=951 Not selected
N=1,580 p-valuea
no. (column %) no. (column %)
Gender 0.060
Female 445 (46.8) 790 (50.0)
Male 506 (53.2) 790 (50.0)
Age
Mean (SD)b54.4(3.2) 54.6(3.4) 0.141
Race/Ethnicity 0.060
Hispanic, any race 56 (5.9) 129 (8.2)
Non-Hispanic, white 651 (68.5) 1,012 (64.1)
Non-Hispanic, other 192 (20.2) 355 (22.5)
Unknown 52 (5.5) 84 (5.3)
Average FPLc0.963
<100% 686 (72.1) 1,145 (72.5)
≥100% 256 (26.9) 421 (26.7)
Missing/Unknown 9 (1.0) 14 (0.90)
Number of chronic conditions diagnosed prior to selection dated<0.001
0 359 (37.8) 601 (38.0)
1 188 (19.8) 374 (23.7)
2 125 (13.1) 223 (14.1)
3–5 71 (7.5) 157 (9.9)
No data to assess 208 (21.9) 225 (14.2)
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL=federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s
exact test due to low cell counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Marino et al. Page 19
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Appendix Table 3E
Demographic Characteristics for Chlamydia Screening Subpopulation Sample by Oregon
OCHIN Patients Selected to Apply for Health Insurance Coverage via Oregon Experiment
vs. Not Selected to Apply (N=366)
Selected
N=133 Not selected
N=233 p-valuea
no. (column %) no. (column %)
Gender NA
Female 133 (100.0) 233 (100.0)
Male 0 (0.0) 0 (0.0)
Age
Mean (SD)b20.1 (0.8) 20.1 (0.8) 0.972
Race/Ethnicity 0.859
Hispanic, any race 26 (19.6) 38 (16.3)
Non-Hispanic, white 67 (50.4) 124 (53.2)
Non-Hispanic, other 34 (25.6) 62 (26.6)
Unknown 6 (4.5) 9 (3.9)
Average FPLc0.223
<100% 102 (76.7) 164 (70.4)
≥100% 31 (23.3) 69 (29.6)
Missing/Unknown 0 (0.0) 0 (0.0)
Number of chronic conditions diagnosed prior to selection dated0.051
0 117 (88.0) 203 (87.1)
1 13 (9.8) 30 (12.9)
2 3 (2.3) 0 (0)
3–5 0 (0) 0 (0)
No data to assess
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL, federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s
exact test due to low cell counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Marino et al. Page 20
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Appendix Table 3F
Demographic Characteristics for Lipid Screening Subpopulation Study Sample by Oregon
OCHIN Patients Selected to Apply for Health Insurance Coverage via Oregon Experiment
vs. Not Selected to Apply (N=10,407)
Selected
N=3,958 Not selected
N=6,449 p-valuea
no. (column %) no. (column %)
Gender 0.732
Female 2,174 (54.9) 3,520 (54.6)
Male 1,784 (45.1) 2,929 (45.4)
Age
Mean (SD)b39.6 (11.4) 39.9 (11.6) 0.184
Race/Ethnicity 0.070
Hispanic, any race 538 (13.6) 848 (13.2)
Non-Hispanic, white 2,399 (60.6) 3,868 (60.0)
Non-Hispanic, other 769 (19.4) 1,372 (21.3)
Unknown 252 (6.4) 361 (5.6)
Average FPLc0.702
<100% 2,843 (71.8) 4,673 (72.5)
≥100% 1,078 (27.2) 1,722 (26.7)
Missing/Unknown 37 (0.9) 54 (0.8)
Number of chronic conditions diagnosed prior to selection dated<0.001
0 2,310 (58.4) 3,814 (59.1)
1 562 (14.2) 1,071 (16.6)
2 261 (6.6) 442 (6.9)
3–5 135 (3.4) 272 (4.2)
No data to assess 690 (17.4) 850 (13.2)
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL, federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s
exact test due to low cell counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Marino et al. Page 21
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Appendix Table 3G
Demographic Characteristics for Flu Vaccine Subpopulation Study Sample by Oregon
OCHIN Patients Selected to Apply for Health Insurance Coverage via Oregon Experiment
vs. Not Selected to Apply (N=2,505)
Selected
N=948 Not selected
N=1,557 p-valuea
no. (column %) no. (column %)
Gender 0.156
Female 446 (47.1) 778 (50.0)
Male 502 (53.0) 779 (50.0)
Age
Mean (SD)b54.4(3.2) 54.6(3.4) 0.151
Race/Ethnicity 0.061
Hispanic, any race 57 (6.0) 126 (8.1)
Non-Hispanic, white 652 (68.8) 998 (64.1)
Non-Hispanic, other 188 (19.8) 350 (22.5)
Unknown 51 (5.4) 83 (5.3)
Average FPLc0.994
<100% 687 (72.5) 1,127 (72.4)
≥100% 252 (26.6) 416 (26.7)
Missing/Unknown 9 (1.0) 14 (0.9)
Number of chronic conditions diagnosed prior to selection dated<0.001
0 358 (37.8) 599 (38.5)
1 185 (19.5) 359 (23.1)
2 124 (13.1) 223 (14.3)
3–5 73 (7.7) 151 (9.7)
No data to assess 208 (21.9) 225 (14.5)
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL, federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s
exact test due to low cell counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Marino et al. Page 22
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Appendix Table 3H
Demographic Characteristics for Glucose Screening Subpopulation Study Sample by
Oregon OCHIN Patients Selected to Apply for Health Insurance Coverage via Oregon
Experiment vs. Not Selected to Apply (N=4,082)
Selected
N=1,506 Not selected
N=2,576 p-valuea
no. (column %) no. (column %)
Gender 0.352
Female 718 (47.7) 1,267 (49.2)
Male 788 (52.3) 1,309 (50.8)
Age
Mean (SD)b51.7(4.5) 51.7(4.7) 0.838
Race/Ethnicity 0.385
Hispanic, any race 109 (7.2) 209 (8.1)
Non-Hispanic, white 1,008 (66.9) 1,660 (64.4)
Non-Hispanic, other 305 (20.3) 564 (21.9)
Unknown 84 (5.6) 143 (5.6)
Average FPLc0.709
<100% 1,096 (72.8) 1,894 (73.5)
≥100% 396 (26.3) 663 (25.7)
Missing/Unknown 14 (0.9) 19 (0.7)
Number of chronic conditions diagnosed prior to selection dated<0.001
0 615 (40.8) 1,123 (43.6)
1 303 (20.1) 554 (21.5)
2 174 (11.6) 305 (11.8)
3–5 98 (6.5) 215 (8.4)
No data to assess 316 (21.0) 379 (14.7)
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL, federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s
exact test due to low cell counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Marino et al. Page 23
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Appendix Table 3I
Demographic Characteristics for HbA1c Screening Subpopulation Study Sample by Oregon
OCHIN Patients Selected to Apply for Health Insurance Coverage via Oregon Experiment
vs. Not Selected to Apply (N=728)
Selected
N=248 Not selected
N=480 p-valuea
no. (column %) no. (column %)
Gender 0.717
Female 121 (48.8) 241 (50.2)
Male 127 (51.2) 239 (49.8)
Age
Mean (SD)b47.4 (8.9) 46.8792 (9.8) 0.445
Race/Ethnicity 0.371
Hispanic, any race 47 (19.0) 94 (19.6)
Non-Hispanic, white 142 (57.3) 245 (51.0)
Non-Hispanic, other 52 (21.0) 126 (26.3)
Unknown 7 (2.8) 15 (3.1)
Average FPLc0.856
<100% 186 (75) 363 (75.6)
≥100% 62 (25) 117 (24.4)
Missing/Unknown 0 (0.0) 0 (0.0)
Number of chronic conditions diagnosed prior to selection dated0.909
0 0 (0.0) 0 (0.0)
1 58 (23.4) 108 (22.5)
2 81 (32.7) 151 (31.5)
3–5 108 (43.6) 220 (45.8)
No data to assess 1 (0.4) 1 (0.2)
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL, federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s
exact test due to low cell counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Marino et al. Page 24
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Figure 1.
CONSORT diagram of the study.
*Subset of 515,575 total OCHIN patients sent for linkage with an encounter at a clinic live
on EHR by the earliest study date (March 11, 2008)
Marino et al. Page 25
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Marino et al. Page 26
Table 1
Demographic Characteristics by Selected Versus not Selected to Apply for Medicaid Coverage via Oregon
Experiment
Selected
N=4,049 Not Selected
N=6,594 p-valuea
Total N=10,643 no. (column %) no. (column %)
Gender 0.734
Female 2,231 (55.1) 3,611 (54.8)
Male 1,818 (44.9) 2,983 (45.2)
Age
Mean (SD)b39.2 (11.7) 39.5 (11.9) 0.185
Race/Ethnicity 0.063
Hispanic, any race 548 (13.5) 867 (13.2)
Non-Hispanic, white 2,447 (60.4) 3,949 (59.9)
Non-Hispanic, other 795 (19.6) 1,412 (21.4)
Unknown 259 (6.4) 366 (5.6)
Average FPLc0.827
<100% 2,911 (71.9) 4,772 (72.4)
≥100% 1,101 (27.2) 1,766 (26.8)
Missing/Unknown 37 (0.9) 56 (0.8)
Number of chronic conditions diagnosed prior to selection dated<0.001
0 2,394 (59.1) 3,936 (59.7)
1 567 (14.0) 1,083 (16.4)
2 261 (6.5) 442 (6.7)
3–5 135 (3.3) 272 (4.1)
No data to assess 692 (17.1) 861 (13.1)
Note: Boldface indicates statistical significance (p<0.05).
ap-values for chi-square test unless otherwise noted
bTwo-sample t-test
cFPL, federal poverty level; values ≥1,000% FPL were set to missing (affected less than 1% of observations). Fisher’s exact test due to low cell
counts in missing/unknown category.
dChronic conditions assessed: asthma, coronary artery disease, diabetes, dyslipidemia, hypertension
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Marino et al. Page 27
Table 2
Preventive Services Receipt by Selection Status to Apply for Medicaid Coverage via the Oregon Experiment
UNADJUSTED ADJUSTED
Screening N %
Receipt Difference
in % OR (OR 95% CI) OR (OR 95% CI)
BMIb
Selected 4,049 54.3 +3.6 1.14 (1.12–1.17) 1.12 (1.10–1.14)
Not selected 6,594 50.7 Ref
Blood pressureb
Selected 4,049 65.6 +3.0 1.10 (1.07–1.13) 1.09 (1.07–1.12)
Not selected 6,594 62.6 Ref
Smokingb
Selected 4,049 59.2 +2.3 1.07 (1.04–1.10) 1.04 (1.02–1.06)
Not selected 6,594 56.9 Ref
Pap testa,b
Selected 1,872 39.3 +3.4 1.16 (1.02–1.32) 1.15 (1.01–1.30)
Not selected 3,059 35.9 Ref
Mammographyc
Selected 979 45.7 +6.0 1.27 (1.02–1.57) 1.27 (1.02–1.57)
Not selected 1,682 39.7 Ref
FOBTb
Selected 951 20.0 +1.5 1.17 (0.98–1.40) 1.15 (1.01–1.32)
Not selected 1,580 18.5 Ref
Colonoscopyb
Selected 951 10.8 +1.2 1.10 (0.95–1.27) 1.04 (0.90–1.20)
Not selected 1,580 9.6 Ref
Chlamydia testb
Selected 133 38.4 +5.8 1.28 (1.11–1.49) 1.24 (1.07–1.44)
Not selected 233 32.6 Ref
Lipid screeningb
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Marino et al. Page 28
UNADJUSTED ADJUSTED
Screening N %
Receipt Difference
in % OR (OR 95% CI) OR (OR 95% CI)
Selected 3,958 36.1 +2.0 1.10 (0.92–1.32) 1.10 (0.93–1.31)
Not selected 6,449 34.1 Ref
Influenza vaccinationb
Selected 948 37.7 +0.0 1.00 (0.83–1.20) 1.02 (0.85–1.21)
Not selected 1,557 37.7 Ref
Glucoseb
Selected 1,506 58.6 +2.4 1.08 (1.00–1.17) 1.05 (0.94–1.17)
Not selected 2,576 56.2 Ref
HbA1cc
Selected 248 69.0 −4.8 0.79 (0.71–0.88) 0.79 (0.71–0.88)
Not selected 480 73.8 Ref
Note: Boldface indicates statistical significance (p<0.05). Statistics were estimated from intent-to-treat generalized estimating equation logistic models. Each model controlled for imbalances between
selection groups (Appendix Tables 3A–3I show tables comparing patient demographics by selection group for each outcome):
aModel adjusted for age
bModel adjusted for number of chronic conditions diagnosed prior to selection date
cNo covariate adjustment made. No differences in covariate distribution between groups.
Definition of Outcome Denominators:
BMI: all adults in the study
Blood pressure: all adults in the study
Smoking status: All adults in the study
Pap test: Females age 21–64, no history of hysterectomy
Mammography: Females age ≥40, no history of bilateral mastectomy
Colorectal cancer screen: age ≥50, no history of colorectal cancer or total colectomy
Chlamydia test: Sexually active females age 19–24
Lipid screening: age ≥20
Influenza vaccination: age ≥50, without indication of vaccine allergy/contraindication or declined
Glucose: age ≥45
HbA1c: diagnosis of diabetes (diagnosis requires ICD9 code to appear on problem list OR in 2+ separate encounters prior to request date)
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
Author Manuscript Author Manuscript Author Manuscript Author Manuscript
Marino et al. Page 29
Table 3
Estimated Treatment Effect of Medicaid Coverage via the Oregon Experiment on Preventive Services Receipt
Screening outcome N
Mean value (%)
in control group
(95% CI) Change (%) with
Medicaid coverage (95% CI) p-value
BMI 10,643 49.1 (45.6, 52.6) 12.5 (10.6, 14.4) <0.001
Blood pressure 10,643 61.5 (59.1, 63.8) 10.1 (7.0, 13.3) <0.001
Smokinga10,643 56.4 (53.0, 59.7) 6.2 (5.3, 7.1) <0.001
Pap test 4,931 34.4 (32.5, 36.3) 10.3 (8.8, 11.7) <0.001
Mammography 2,661 38.2 (32.7, 43.7) 14.5 (10.1, 18.8) <0.001
FOBT 2,531 19.1 (11.8, 26.4) −0.2 (−5.1, 4.7) 0.933
Colonoscopy 2,531 9.4 (7.3, 11.4) 2.7 (−1.7, 7.1) 0.235
Chlamydia 366 28.7 (26.5, 31.0) 27.3 (14.1, 40.4) <0.001
Lipid screeninga10,407 32.9 (27.3, 38.4) 8.0 (1.0, 15.0) 0.026
Influenza vaccination 2,505 37.8 (31.5, 44.0) −0.4 (−8.3, 7.5) 0.922
Glucose 4,082 55.9 (51.5, 60.3) 4.8 (−3.0, 12.7) 0.227
HbA1c 728 71.9 (69.5, 74.4) 0.8 (−4.0, 5.7) 0.732
Note: Boldface indicates statistical significance (p<0.05). Estimated statistics using bivariate probit instrumental variable model. Models adjusted
for same covariate set as intent-to-treat models.
aSelection status in Oregon Experiment was the only instrument in this model based on test of overidentifying restrictions.
Am J Prev Med. Author manuscript; available in PMC 2017 February 01.
