Using Health IT to Coordinate Care and Improve Quality in Safety Net Clinics

Abstract

Background:

Health centers provide care to vulnerable and high-need populations. Recent investments have promoted use of health information technology (HIT) capabilities for improving care coordination and quality of care in health centers. This study examined factors associated with use of these HIT capabilities and the association between these capabilities and quality of care in a census of U.S. health centers.

Methods:

Cross-sectional secondary data from the 2015 Health Resources and Services Administration’s Uniform Data System was used to examine 6 measures of HIT capability related to care coordination and clinical decision support and 16 measures of quality (12 process measures, 3 outcome measures, 1 composite measure) for all U.S. health centers. We used adjusted logistic regressions to examine health center characteristics associated with use of HIT capabilities and adjusted linear regressions to examine associations between HIT capabilities and quality of care.

Results:

Many health centers reported using HIT for care coordination activities, including coordinating enabling services (67.3%) or engaging patients (81.0%). Health center size and medical home recognition were associated with significantly greater odds of using HIT for enabling services and engaging patients. These HIT capabilities were associated with higher overall quality, and higher rates of six process measures (adult screening and maternal and child health) and HbA1c control.

Conclusions:

Use of HIT for activities like arranging enabling services and engaging patients are underleveraged tools for care coordination. There may be opportunities to further improve quality of care for vulnerable patients by promoting health centers’ use of these HIT capabilities.

Background

Health centers provide affordable and comprehensive primary health care and supportive services to over 24 million patients in low-income communities in the U.S.¹ Health centers, which include Federally Qualified Health Centers (FQHCs) that receive funds from the Health Resources and Services Administration (HRSA) Health Center Program, receive enhanced reimbursements from Medicare and Medicaid for providing care to medically-underserved populations. FQHCs must offer a sliding fee schedule to all patients and establish a governing board that is composed primarily of patients. Other health centers receive HRSA-funding to enhance access to care and to help provide comprehensive, coordinated, and patient-centered care.² Health center patients experience high rates of chronic conditions, periods of uninsurance, and additional challenges, such as homelessness and unemployment.^2,3 Health centers are part of the U.S. health care “safety net” because they provide care to patients regardless of their ability to pay. Health centers utilize various care coordination strategies to care for their complex patient population, including providing enabling services and patient-centered medical home (PCMH) implementation.⁴ Enabling services are defined as non-clinical services that promote better access to and continuity of care and may include transportation to and from appointments, case management, and health education.⁵ The adoption of electronic health records (EHRs) and enhanced health information technology (HIT) capabilities has been promoted as a strategy to enhance care coordination, including the coordination of enabling services, and improve quality of care.^6–8

EHR adoption among health centers is high, having increased considerably in recent years. In 2010, 29.8% of health centers had EHRs compared to 92.4% in 2015.^1,9 Investment in EHRs and HIT capabilities in safety-net settings has been facilitated by legislation providing incentives for adoption and meaningful use of EHRs and funding the establishment of the Regional Extension Centers to assist small and safety-net providers with EHR adoption¹⁰ and Health Center Controlled Networks to promote use of EHRs and HIT to improve quality.¹¹ In particular, the 2009 Health Information Technology for Economic and Clinical Health Act authorized incentive payments to clinicians and hospitals to promote meaningful use of EHRs. “Meaningful Use” objectives were then developed to monitor achievement of core objectives, such as data capture and sharing, and other enhanced HIT capabilities, intended to promote adherence to evidence-based practices and eventually improve outcomes.¹¹ From 2010 to 2012, health centers increased adoption of all enhanced HIT capabilities related to meaningful use, with the greatest improvements observed in health centers’ capability to provide clinical summaries to patients following visits (52.1% vs. 85.1%) and share electronic versions of health information with patients (41.1% vs. 71.0%).⁹ Adoption of some enhanced HIT capabilities, however, remained low. For example, in 2012, 55.1% of health centers could electronically exchange clinical information, while only 36.1% could electronically report to immunization registries.⁹ Furthermore, despite high rates of EHR adoption, health centers are less likely than multispecialty or HMO-owned practices to report having enhanced HIT capabilities,¹² which are critical for coordinating care for complex patients. While resources exist to help health centers develop HIT capabilities to support care coordination activities, little is known about the factors associated with adoption of specific HIT capabilities for improving care coordination in health centers.

Evidence regarding the relationship between adoption of EHRs, enhanced EHR capabilities to support care coordination activities, and quality of care is mixed. Health centers and safety net hospitals have reported that shared access to patient data is a crucial component for providing coordinated care. ¹³Among six health centers in New York during 2008–2010, EHR adoption was associated with ongoing improvement in quality of care for three years following initial adoption.¹⁴ A 2009 survey of health centers reported that health centers with higher HIT capacity, measured as use of EHRs and 15 other HIT capabilities, were more likely to receive discharge summaries, use a patient notification system for preventive and follow-up care, and easily make timely specialist appointments for uninsured patients.¹⁵ Additionally, a 2017 study of an integrated delivery system for Medicaid enrollees composed of hospitals, physician offices, and health centers found that use of real-time electronic alerts, deployed along with other care coordination activities, resulted in increased rates of primary care visits following emergency room visits and fewer hospital stays.¹⁶ A 2014 systematic review of studies examining the association between HIT and quality of care reported that most studies were focused on clinical decision support or computerized provider order entry, examined process measures of quality, and reported at least some positive impacts on quality and safety, but that more work was needed to understand the context that leads to improvements.¹⁷ To our knowledge, no study has examined the association between enhanced HIT capabilities related to care coordination and process and outcome measures of quality of care in the context of health centers.

Given recent investments in health centers and increasing number of patients served by them, this study sought to address the gap in knowledge regarding factors associated with adoption and use of enhanced HIT capabilities supporting care coordination activities using the 2015 Uniform Data System (UDS), a nationwide census of health centers from HRSA. Additionally, to understand the relationship between use of these enhanced HIT capabilities and quality of care in HC settings with high levels of EHR adoption, we examined the association between HIT capabilities for coordinating enabling services and engaging patients with 12 process measures, 3 outcome measures, and 1 composite measure of overall clinical quality of care.

Methods

Data.

Secondary data on health centers were obtained from HRSA’s 2015 UDS. The UDS includes annual data submitted by health centers on patient characteristics, staffing, finances, clinical quality performance, and indicators of HIT capabilities. All HRSA-funded health centers are required to submit UDS data. HRSA releases detailed reporting instructions annually and staff members review submitted data to ensure compliance with program requirements, report accomplishments, and identify opportunities to improve performance.¹⁸ UDS data are reported at the level of the health center grantee; grantees with multiple sites of care report aggregate information. We excluded from analysis health centers that were seasonal, mobile, or otherwise not permanent. The Kaiser Family Foundation provided information on state Medicaid expansion status.¹⁹ ZIP-code level estimates of community characteristics were obtained from the American Community Survey (ACS) 5-year estimates (2011–2015).²⁰

Measures.

Building on the Donabedian model, we organized our measures into structure, process, and outcome categories.²¹ Structural measures describe characteristics of organizations and providers. We examined six dichotomous measures of enhanced HIT capability, which align with the Centers for Medicare & Medicaid Services (CMS) Meaningful Use criteria. For each of the following items, which were taken verbatim from the UDS, health centers were asked to report if their EHR system had and they used the following care coordination capabilities:

• Uses HIT to coordinate or to provide enabling services such as outreach, language translation, transportation, case management, or other similar services.

• Engages patients through HIT such as patient portals, kiosks, secure messaging (i.e., secure email) either through the EHR or through other technologies;

• Provides patients with electronic summaries of office visits or other clinical information when requested;

• Uses computerized, clinical decision support such as alerts for drug allergies, checks for drug-drug interactions, reminders for preventive screening tests, or other similar functions;

• Exchanges clinical information electronically with other key providers/health care settings such as hospitals, emergency rooms, or subspecialty clinicians;

• Sends prescriptions to the pharmacy electronically (Not including faxing).

Additionally, we included continuous measures of the number of full-time equivalent (FTE) primary care team members and enabling service providers per 10,000 patients. We constructed a categorical measure of the number of sites with an installed EHR system in use (none, some, or all) and a dichotomous measure indicating PCMH recognition for one or more sites. We did not adjust for Meaningful Use stage because the HIT capabilities we examined are closely aligned with Meaningful Use objectives.

Process measures described care delivered to patients and outcome measures reflect the impact of that care on the health of patients. These UDS measures are consistent with the Agency for Healthcare Research and Quality’s National Quality Strategy²² and other national performance measurement initiatives²³ and were reported by health centers using an EHR for the universe of eligible patients or, when not possible, a random sample of 70 paper charts.¹⁸ Full definitions of these measures are available in the 2015 UDS Reporting Instructions.¹⁸ Twelve process measures related to chronic conditions (HIV linkage to care, pharmacologic therapy for patients with asthma, lipid-lowering therapy for coronary artery disease (CAD), and aspirin for ischemic vascular disease (IVD)), adult screening (tobacco use screening and cessation intervention, adult weight screening and follow-up, cervical cancer screening, depression screening and follow-up, and colorectal cancer screening), and maternal and child health (entry into prenatal care during first trimester, childhood immunization, and child and adolescent weight assessment and counseling). We also constructed a standardized composite measure of performance with 10 of the aforementioned process measures, excluding HIV linkage to care and early entry into prenatal care, by calculating z-scores to standardize each measure, summing all measures, and then standardizing the composite (Cronbach’s alpha=0.76).²⁴ Finally, three outcome measures included percentages of low birthweight births, patients with diabetes and hemoglobin A1c (HbA1c) <8%, and patients with hypertension and blood pressure (BP) <140/90 mm Hg.

We examined the following characteristics of patients, organizations, and communities. Patient characteristics included three continuous measures of the percentage of minority race/ethnicity (i.e., American Indian/Alaska Native, Asian, black, Native Hawaiian, other Pacific Islander, Multiracial, and Hispanic/Latino patients), uninsured, and homeless patients. Organizational characteristics included continuous measures of total patients and net revenue and a categorical measure for region (West, Northeast, Midwest, or South). We constructed dichotomous measures for type(s) of grant funding received (community, migrant, homeless, and public housing) and rural location. A dichotomous variable, obtained from the Kaiser Family Foundation, indicated if the state expanded Medicaid before or during 2015. ¹⁹ Three continuous measures of community characteristics that reflect each health center’s service area were derived from the ACS, including median annual household income, percentage of individuals aged 25 or older with less than a high school diploma, and percentage of unemployed males aged 16 and older. ²⁰ These measures were constructed by taking a weighted average of ZIP Code Tabulation Area (ZCTA)-level measures from the ACS, where the weights were the number of patients served by each health center grantee in each ZIP code.

Analysis.

Descriptive statistics were generated for all variables in the analytic sample. For three variables with a large number of missing observations (i.e., percent minority, homeless, and uninsured), we imputed values based on the mean. For the two measures of enhanced HIT capability with sufficient variation (i.e., coordinating enabling services and engaging patients), we estimated logistic regression models to identify factors associated with use of each enhanced HIT capability, adjusting for characteristics of patients, organizations, and communities. For select continuous explanatory variables, we reported marginal effects in terms of change in predicted probability to facilitate interpretation of results. We used linear regression to examine the association between HIT capabilities for coordinating enabling services and engaging patients and the process and outcome measures, adjusting for the aforementioned characteristics. Analyses were performed using Stata 13 (Statacorp College Station, TX). This study was determined to be exempt by our institutional review board.

Results

Characteristics of sample.

Among the 1,375 HRSA-funded health centers in 2015, we excluded 30 centers located outside of the 50 states and the District of Columbia and health centers missing information on patients (n=9), primary care team members (n=13), or revenue (n=1). Thus, 4% of observations were excluded for geography or missing information. Nearly all health centers had EHRs installed and in use at all sites (93.7%) and over two-thirds had obtained PCMH recognition (69.7%) (Table 1). Among health center patients, on average 40.0% were a minority race or ethnicity, 27.0% were uninsured and 7.6% were homeless. In communities served by health centers, 18.4% of individuals aged 25 and older did not have a high school diploma on average.

Table 1.

Characteristics of health centers, UDS 2015 (N=1,322)

	Mean (standard deviation) or %
Structural measures
Primary care team FTE per 10,000 patients1	13.8 (42.5)
Enabling service FTE per 10,000 patients2	4.7 (10.5)
PCMH recognition for one or more sites	69.7%
Currently have an EHR system installed and in use
None	1.4%
Some sites	4.9%
All sites	93.7%
Contextual factors
% minority patients	40.0% (29.2%)
% uninsured patients	27.0% (18.5%)
% homeless patients	7.6% (19.5%)
Total patients	116,332 (253,703)
Health center net revenue	$649,606 ($293,562)
Type of funding
Community health center	92.7%
Migrant health center	12.2%
Health care for the homeless	21.0%
Public housing primary care	6.7%
Rural	54.5%
Region
West	29.1%
Northeast	17.4%
Midwest	19.7%
South	33.8%
State expanded Medicaid	65.1%
% of individuals with less than high school diploma, aged 25 and older	18.4% (7.7%)
% of males unemployed, aged 16 and older	6.8% (2.1%)
Median annual household income	$45,262 ($11,083)

^1.The following types of providers were included in the primary care team: physicians, nurse practitioners, physician assistants, certified nurse midwifes, and nurses.

^2.Enabling service providers included case managers and health educators.

Nearly all health centers reported the capability to send prescriptions to pharmacies electronically, exchange clinical information electronically with other providers, use computerized, clinical decision support, and provide patients with electronic summaries of visits (Figure 1). In contrast, fewer health centers reported using HIT to coordinate enabling services, such as transportation and case management (67.3%) or engage patients with secure email or patient portals (81.0%).

Figure 1.

Describes the prevalence of enhanced HIT capabilities in health centers (UDS 2015)

For all chronic condition process measures, average rates exceeded 65%, including the percentages of eligible patients who received pharmacologic therapy for IVD (67.6%) and CAD (69.8%) (Figure 2). Although health centers screened 61.0% of adults for tobacco use, rates for the other adult screening measures were all less than 40%. Maternal and child health measures exhibited variation, with 76.2% of health centers achieving early entry into prenatal care for patients, but only 35.0% of children receiving weight assessment and counseling. When examining outcome measures, on average, BP control for patients with hypertension was 62.7% and HbA1c control <8% for patients with diabetes was 47.0%. Few health centers births were infants less than 2500 grams (8.9%).

Figure 2.

Describes the prevalence of health center process and outcome quality of care measures (UDS 2015)

Results for the regression models estimating the adjusted odds of having enhanced HIT capabilities are displayed in Table 2. Health centers with PCMH recognition had significantly greater odds of using an EHR or HIT to coordinate enabling services (OR=2.45, P<0.001) and engage patients (OR=2.86, P<0.001). For a health center with mean values for all characteristics, having 50,000 more patients was associated with a 0.9 percentage point increase (P=0.026) in the probability of using HIT for enabling services and 1.5 percentage point increase (P=0.003) in the probability of using HIT to engage patients. Receipt of migrant health center funding was associated with significantly greater odds of using HIT to coordinate enabling services (OR=1.76, P=0.007). Number of uninsured patients was associated with decreased probability of using HIT to engage patients. For a health center with mean values for all characteristics, having an additional 10 percentage points of uninsured patients was associated with a 1.6 percentage point decrease in the predicted probability of using HIT to engage patients (P=0.013).

Table 2.

Regression results providing the adjusted odds of having enhanced HIT capabilities

Variables	HIT capability to coordinate or provide enabling services Odds ratio (standard error)	HIT capability to engage patients through HIT Odds ratio (standard error)
Structural measures
Primary care team FTE per 10,000 patients	0.997 (0.003)	0.99 (0.01)
Enabling service FTE per 10,000 patients	1.01 (0.01)	1.02 (0.01)
PCMH recognition for one or more sites	2.45*** (0.33)	2.86*** (0.45)
Contextual factors
% minority patients	1.23 (0.32)	0.77 (0.24)
% uninsured patients	0.81 (0.32)	0.32* (0.15)
% homeless patients	0.62 (0.33)	0.25* (0.15)
Total patients (in 10,000)	1.01* (0.01)	1.02** (0.01)
Health center net revenue (in $10,000)	1.00 (0.0003)	1.00 (0.0004)
Type of funding
Community health center	0.99 (0.34)	0.71 (0.29)
Migrant health center	1.76** (0.38)	1.27 (0.33)
Health care for the homeless	1.27 (0.25)	1.31 (0.33)
Public housing primary care	0.93 (0.25)	1.11 (0.37)
Rural	0.86 (0.13)	0.86 (0.16)
Region (reference: West)
Northeast	1.18 (0.24)	1.07 (0.26)
Midwest	1.03 (0.20)	1.14 (0.27)
South	0.79 (0.15)	1.70* (0.41)
State expanded Medicaid	1.00 (0.16)	1.01 (0.20)
% of individuals with less than HS diploma, aged 25 and older	1.23 (1.27)	0.27 (0.33)
% of males unemployed, aged 16 and older	0.10 (0.38)	0.01 (0.07)
Median annual household income (in $10,000)	1.08 (0.08)	1.01 (0.08)
Observations	1,322	1,296

^***p<0.001

^**p<0.01

^*p<0.05

Results for the regression models examining associations between enhanced HIT capabilities and quality of care are reported in Table 3. Having and using the EHR or HIT capability to coordinate or provide enabling services was associated with significantly higher rates of all adult screening measures, specifically tobacco use screening and cessation intervention (6.19 percentage points, P=0.005), adult weight screening and follow up (3.98 percentage points, P=0.03), and screening for cervical cancer (6.44 percentage points, P<0.001), depression (6.49 percentage points, P=0.001), and colorectal cancer (5.57 percentage points, P<0.001). Among the maternal and child health measures, use of HIT for enabling services was significantly associated with an increase in only child weight assessment and counseling (4.42 percentage points, P=0.02). Use of HIT for enabling services was associated with a 0.20 standard deviation (P=0.001) increase in overall quality for process measures, which is a relatively small effect.²⁵ Among the outcome measures, use of HIT for enabling services was associated with a 2.72 percentage point (P=0.04) increase in HbA1c control for patients with diabetes.

Table 3.

OLS estimates of the association between enhanced HIT capability and quality of care

	HIT capability to coordinate or provide enabling services			HIT capability to engage patients through HIT
Dependent variables	Coefficient	Standard Error	N	Coefficient	Standard Error	N
Composite measure of quality for process measures (measured in units of standard deviation)1	0.20***	(0.06)	1,245	0.34***	(0.07)	1,224
Process measures
Chronic condition measures
HIV linkage to care	3.97	(2.74)	718	−1.96	(3.51)	708
Pharmacologic therapy for patients with asthma	2.84	(1.72)	1,309	2.42	(2.08)	1,284
Lipid-lowering therapy for coronary artery disease	2.08	(1.45)	1,310	2.78	(1.75)	1,284
Aspirin for ischemic vascular disease	2.48	(1.52)	1,308	3.37	(1.84)	1,283
Adult screening measures
Tobacco screening & cessation intervention	6.19**	(2.21)	1,319	12.15***	(2.66)	1,294
Adult weight screening and follow-up	3.98*	(1.84)	1,321	9.96***	(2.21)	1,295
Cervical cancer screening	6.44***	(1.53)	1,320	9.19***	(1.86)	1,295
Depression screening and follow-up	6.49***	(1.89)	1,319	8.28***	(2.29)	1,294
Colorectal cancer screening	5.57***	(1.25)	1,318	6.13***	−1.52	1,293
Maternal and child health measures
Early entry into prenatal care	1.09	(1.08)	1,137	1.18	(1.32)	1,120
Children with up to date immunizations	−0.1	(1.73)	1,261	1.86	(2.11)	1,239
Child weight assessment and counseling	4.42*	(1.87)	1,301	9.09***	(2.26)	1,276
Outcome measures
HbA1c control <8% for patients with diabetes	2.72*	(1.29)	1,316	3.42*	(1.56)	1,291
BP control for patients with hypertension	−0.56	(0.61)	1,311	1.25	(0.74)	1,288
Low birth weight (<2500 grams)	−0.81	(0.810)	1,138	−1.01	(1.00)	1,121

^***p<0.001

^**p<0.01

^*p<0.05.

Models adjusted for all covariates listed in Table 1.

^1.We constructed a composite measure of quality by standardizing each process measure, summing all the measures, and then standardizing the composite. The composite measure includes all process measures listed in the table except HIV linkage to care and early entry into prenatal care due to lower levels of correlation with the composite measure.

Similarly, having and using HIT for patient engagement was associated with significantly higher rates of all adult screening measures, including tobacco use screening and cessation intervention (12.15 percentage points, P<0.001), adult weight screening and follow up (9.96 percentage points, P<0.001), and screening for cervical cancer (9.19 percentage points, P<0.001), depression (8.28 percentage points, P<0.001), and colorectal cancer (6.13 percentage points, P<0.001). Among the maternal and child health measures, patient engagement using HIT was significantly associated with an increase in only child weight assessment and counseling (9.09 percentage points, P<0.001). Using HIT for patient engagement was associated with a 0.34 standard deviation (P<0.001) increase in overall quality for process measures—a small to medium effect size.²⁵ Among the outcome measures, using HIT for patient engagement was associated with a 3.42 percentage point (P=0.03) increase in HbA1c control for patients with diabetes. Full regression results are available in the appendix (see table, Supplemental Digital Content 1).

Sensitivity analyses.

We used the Benjamini-Hochberg correction to account for multiple hypothesis testing, finding that all associations with the two HIT capabilities and quality measures remained significant.²⁶ Additionally, we examined whether use of chart review rather than EHRs for reporting quality measures influenced our findings. Among health centers, 46% used a sample derived from chart review rather than EHRs for one or more quality measures. When we adjusted for use of chart review in regression models examining quality of care, results for our HIT explanatory variables were unchanged in direction, but no longer significant in models examining adult screening measures (i.e., tobacco use, adult weight, and, for EHR engagement only, depression and colorectal cancer). Health centers’ use of EHRs for reporting is highly correlated with other HIT capabilities and thus highly correlated with our key explanatory variables, which is why we did not include this adjustment in our main analyses.

We also examined the potential influence of PCMH recognition on our examination of the association between EHR capabilities related to care coordination and quality of care. Among health centers with PCMH recognition, 74.4% reported using EHRs to coordinate enabling services (compared to 51.1% of non-PCMH health centers) and 87.1% reported using EHRs to engage patients (compared to 66.4% of non-PCMH health centers). We examined whether the effect of enhanced EHR capability on the composite quality measure varied according to PCMH recognition, but found that the interaction was not statistically significant for either of the two types of EHR capabilities examined in Table 3 in any models.

Discussion

Federal investment in health centers has grown in recent years, as they have received assistance in developing HIT capabilities from Regional Extension Centers, HRSA Quality Improvement Awards, and participation in Health Center Controlled Networks.^10,11,27 Although health centers have adopted EHRs at high rates, there is variation in the use of HIT and enhanced EHR capabilities that can be used for care coordination activities and little information about how use of these capabilities might improve quality of care and patient outcomes at health centers, an ongoing goal of HRSA.¹¹ We found high rates of adoption and use of HIT capabilities in health centers and a strong association between use of HIT for enabling services and patient engagement and higher performance on multiple measures of quality as well as a composite quality measure.

Despite high rates for several enhanced HIT capabilities, use of HIT for enabling services and patient engagement are underleveraged tools for care coordination, two important activities that are not reimbursable services. Studies suggest that health center patients could benefit from these technologies. A 2013 San Francisco study reported that 71% of health center patients were interested in communicating via e-mail with health care providers.²⁸ Additionally, a 2011 New York City study reported that 49% of health center patients used a patient portal two or more times over two years.²⁹ While more research is needed to understand interest and use of these capabilities among patients in rural communities, a study of health centers in California reported that rural location increased the odds of adoption of EHRs and enhanced capabilities.³⁰ Although the California study did not specifically ask about health centers’ engagement with Regional Extension Centers, which are funded to assist small and safety net providers with achievement of meaningful use objectives, the authors hypothesized about the centers’ likely beneficial role. These studies and our finding that health centers with fewer patients were less likely to utilize HIT care coordination capabilities suggests that there appears to be a continued role for Regional Extension Centers to provide additional technical assistance for adoption of care coordination technologies.

Using HIT for enabling services and patient engagement was consistently associated with improvements in screening measures, but had weaker or no associations with other process measures, such as those related to care for chronic conditions. Health centers may have primarily adopted these types of enhanced HIT capabilities to promote visits, during which these screenings occur, which may help explain why we observed an association with all of the adult screening measures and the child weight assessment and measure. Many health centers use HIT for case management, patient outreach, and other ways to help patients get to their clinic. ^6,31 Four of the adult screening measures we examined had rates of adherence below 35%, suggesting there is much room for improvement, and for that reason the positive association we observed for these measures and enhanced HIT capabilities is particularly promising.

While enhanced HIT capabilities were associated with higher overall quality, we observed mixed results when examining the impact on care for chronic conditions. Prior studies report that use of secured email communication between patients and providers in non-safety net settings can improve quality of care, including improvements in HbA1c control for patients with diabetes and BP control for patients with hypertension,^32,33 however less is known about this relationship for patients cared for in safety-net settings. Interviews with health center patients in San Francisco suggest that patient barriers to portal use includes security concerns, lack of technical skills, and preference for in-person communication.³⁴ Furthermore, a recent systematic review reported insufficient evidence on the impact of patient portals on health outcomes, cost, and use.³⁵ More research is needed to understand how health centers currently use patient portals and secure messaging for patient engagement and how these technologies can potentially be used to improve health center patient outcomes.

As observed among CHCs, EHR adoption rates by hospitals have increased over time, with 75% of hospitals having at least a basic EHR system in 2014.³⁶ Also similar to CHCs, evidence for the relationship between hospital adoption of EHRs and enhanced EHR capabilities and quality of care is mixed. For example, while Brice and colleagues reported no difference in 30-day readmission rates for hospitals who met meaningful use stage 1 objectives compared to those who did not,³⁷ Adler-Milstein reported improvements in patient satisfaction and adherence to process measures when comparing performance in hospitals before and after EHR adoption.³⁸ Limited research has focused on hospitals’ use of HIT for providing enabling services and engaging patients or its impact on quality. In 2014, 64% of hospitals reported that patients had the ability to view, download, and transmit information, a meaningful use objective related to patient engagement.³⁶ A 2013 systematic review found only 17 studies focused on patient engagement in inpatient settings.³⁹ Nearly all studies examined patient self-efficacy or satisfaction and reported improvements in these measures. Only three studies examined health outcomes and reported improvements. Thus, there is a need for additional research to better understand the role of HIT for patient engagement and enabling services in the inpatient setting.

About 69.7% of health centers have PCMH recognition for at least some of their sites, suggesting that adoption of enhanced HIT capabilities and quality of care should be viewed in this context. The PCMH model emphasizes enhanced access along with comprehensive and coordinated primary care, characteristics aligned with the goals of meaningful use. We found that having PCMH recognition at one or more sites was associated with significantly greater odds of using HIT for enabling services and patient engagement. Whereas the pursuit of PCMH recognition may encourage health centers to adopt one of more of these enhanced HIT capabilities, the effect of these HIT functionalities we observed on quality are above and beyond any effect of these functionalities that may be due to PCMH recognition given that these HIT variables remained significant in regression models even after taking into account each health center’s PCMH recognition status. Thus, adoption and use of these HIT capabilities should be encouraged in both health centers with and without PCMH recognition.

Limitations include our inability to examine associations between quality of care and four measures of enhanced HIT capability due to limited variation and the use of cross-sectional data. Future analyses could attempt a longitudinal approach to examine length of adoption of the HIT capability on quality. However, identifying an appropriate data source may be challenging because although HRSA has collected information on EHR use and HIT capabilities since 2010,⁹ the specific measures collected have changed over time. Additionally, health centers reported process and outcome measures using an EHR or a random sample, with the vast majority using EHRs. The reporting mechanism was highly correlated with enhanced HIT capabilities suggesting that our estimates may be a blend of both the influence of specific capabilities and a measure of overall HIT capability. Additionally, whereas health centers are instructed to report only EHR capabilities they use, it is unknown if reports imply consistent use of the capability for all sites and all patients. Predictors of HIT capabilities included those captured in administrative data, but not other characteristics such as, quality improvement activities, strength of leadership, and practice culture. Thus, our estimates may conflate HIT capabilities with other factors that are correlated with these capabilities.

Conclusion

Health centers serve a patient population with complex health and socioeconomic needs, making care coordination activities essential and often complicated. While gains have been made in adoption of several enhanced HIT capabilities, including those that facilitate communication among disparate providers in the safety-net, some HIT capabilities that are critical for care coordination, including patient engagement and coordination enabling services, are underleveraged tools. We found that use of HIT to engage patients and provide enabling services was associated with improvements in rates of adult screening and overall quality, thus there may be an opportunity to further improve quality of care by promoting the adoption and expanded use of these HIT capabilities. This study has several strengths, including the examination of a nationwide census of health centers and the uniform capture of both HIT capabilities and quality. Additional work is needed to better understand how health centers can utilize these enhanced HIT capabilities to improve additional measures of quality.

Supplementary Material

Supplemental Digital Content 1.

Includes regression results from the fully adjusted models examining the association between HIT capabilities and quality measures.