Electronic clinical quality measure reporting challenges: findings from the Medicare EHR Incentive Program’s Controlling High Blood Pressure Measure

Abstract

Objective

To identify physician and practice characteristics associated with high clinical and technical performance on the electronic clinical quality measure (eCQM) that calculates the proportion of patients with hypertension who have controlled blood pressure.

Materials and Methods

The study included 268 602 physicians participating in the Medicare Electronic Health Record Incentive Program between 2011 and 2014. Independent variables included delivery reform participation and physician, practice level, and area characteristics. Successful technical performance was a reported eCQM with non-zero values in both the numerator and denominator. Successful clinical performance was a reported eCQM value of ≥70% hypertension control.

Results

Physicians with longer experience using certified health information technology, participants in delivery reform programs, and specialists that traditionally manage hypertension were 5%–15% more likely to achieve 70% control. Physicians in smaller and rural practices and a subset of physicians unlikely to primarily manage hypertension were more likely to submit measures with a zero value in either the numerator or denominator.

Discussion

More physicians are using eCQMs to track and report their quality improvement efforts. This research presents the first examination of national eCQM data to identify physician and practice-level characteristics associated with performance.

Conclusion

With careful selection of measures relevant to the clinician’s specialty, complete data entry, and support for continuous quality improvement, health care professionals can excel technically and clinically. As care delivery transitions from fee-for-service to quality- and value-based models, high performers may realize financial gains and better patient outcomes. These analyses suggest patterns that may inform steps to improve performance.

BACKGROUND AND SIGNIFICANCE

Cardiovascular disease is responsible for nearly 1 in 3 deaths in the United States and remains the nation’s number-one killer.^1,2 Over $300 billion is spent annually on care and lost productivity for this largely preventable condition.¹ Hypertension is a major contributor to cardiovascular disease that can be controlled with lifestyle modifications and medications. Yet, of the 70 million American adults with hypertension, only about half have controlled hypertension.³

To address this burden, in 2012 the US Department of Health and Human Services (HHS) launched Million Hearts^®, a national initiative with the goal of preventing 1 million cardiovascular events by 2017. Million Hearts focuses clinicians and health care systems on improving performance on a small set of evidence-based clinical quality measures (CQMs) for the ABCS: Aspirin when appropriate, Blood pressure (BP) control, Cholesterol management, and Smoking cessation.⁴ Meeting performance targets of ≥70% on the CQMs is critical to achieving the 2017 Million Hearts goal.

Clinical quality measures help assess and track performance on key indicators of care. Electronic CQMs (eCQMs) are based on information extracted from the electronic health record (EHR). Million Hearts^® encourages the use of the eCQM known as NQF0018. The eCQM NQF0018, titled “Controlling High Blood Pressure,” assesses the proportion of patients in the clinician’s EHR system with hypertension who have controlled BP. The denominator for this measure consists of patients ages 18–85 years in the clinician’s EHR who had a diagnosis of hypertension noted before, or within the first 6 months of, the measurement period. The measure numerator includes those patients from the denominator whose most recent office BP indicated adequate control (<140/90 mmHg). Clinicians who choose to report on NQF0018 identify patients within their EHR system with a diagnosis of hypertension (measure denominator) and input those patients’ BP readings (measure numerator). Failure to record either hypertension diagnosis or BP readings results in a measure value of zero.

One growing source of clinician-specific data on NQF0018 and other eCQMs is the Medicare EHR Incentive Program (IP).⁵ Beginning in 2011, IP participants were eligible to receive financial bonuses for their “meaningful use” of certified health IT to improve patient safety, care coordination and quality, efficiency of care, and public health reporting.⁶ As a pay-for-reporting initiative, its purpose is to encourage participants to establish electronic infrastructure and workflows that can help improve health care quality and efficiency. Therefore, because the IP rewards only technical success in reporting the eCQMs, the data can provide insights into clinical performance.

Between 2011 and 2013, IP participants had to report on 3 required eCQMs and 6 additional eCQMs selected from a list that included NQF0018.⁷ In 2014, to ease the reporting burden, the program shifted to a system whereby participants could select 9 eCQMs from 3 different domains or report eCQMs from 1 of 2 “recommended core” panels. The recommended adult core panel included NQF0018.⁸ IP participants reported NQF0018 values based on the patients included in their EHR system, meaning that across IP participants, there could be patient duplication.

For eCQMs to accurately reflect clinical performance, several conditions must be met: (1) Clinicians must know what measure(s) are being reported; (2) clinicians understand how and where to enter essential data into the EHR (in the case of NQF0018, this includes both the diagnosis and actual BP readings); and (3) the care team reviews the data prior to its submission to gauge its accuracy.

Previously, IP data from 2011 to 2013 were used to show their potential value for public health surveillance.⁷ This new research presents the first examination of these data to identify physician- and practice-level characteristics associated with clinical and technical performance.

METHODS

Study population

This analysis population included 268 602 unique office-based physicians who participated in at least 1 year of the IP between 2011 and 2014. IP reporting captures all adult patients diagnosed with hypertension, not just those in the Medicare population. In 2014, the Centers for Medicare and Medicaid Services (CMS) estimated that there were approximately 393 000 unique Medicare-eligible professionals, which included both physician and non-physician clinicians (optometrists, dentists, chiropractors, and podiatrists). There is no estimate of eligible physicians alone; however, physicians comprised the majority of the eligible professional population. Non-physician clinicians were excluded from the analysis due to low NQF0018 reporting rates.

Outcomes

Two primary constructs were estimated through these analyses: clinical and technical performance (Table 1). Performance was determined from the self-reported aggregate measure score (measure defined above). Successful clinical performance was evaluated as the achievement of ≥70% hypertension control among physicians who reported eCQM scores. Successful technical performance was defined as a reported NQF0018 score with a non-zero value in both the denominator and numerator. Two binomial outcomes were used to assess variations in technical performance failure. The first outcome identified physicians who reported a value of zero in the denominator; the second identified physicians who reported a non-zero value for the denominator but a zero value for the numerator.

Table 1.

Outcome measure descriptions

Construct	Variable	Measure Description
Technical performance	Technical success	Physician reported a non-zero value for both numerator and denominator
	Zero denominator reporting	Physician reported a zero value in denominator; failed to identify patients with hypertension
	Zero numerator reporting	Physician reported a non-zero value in denominator but a zero value in the numerator; failed to identify hypertension control among patients with hypertension
Clinical performance	Successful clinical performance	Physician reported >70% hypertension control among patients with hypertension
	Failed to achieve >70% control	Technical success but NQF0018 score <70%

Independent variables

Independent variables included physician- and practice-level characteristics, community measures of social determinants at the county level to estimate characteristics of the patient populations, and delivery reform participation.

Physician characteristics

Physician characteristics included age, specialty, and number of years the physician reported to the IP using certified health IT. IP data were merged with the 2011 American Medical Association (AMA) Physician Masterfile and CMS’s Provider Enrollment, Chain, and Ownership System (PECOS) using the physician’s National Provider Identifier. Physician age was obtained from the AMA Masterfile and, if not available, data from PECOS were used. Practice size was determined by counting the number of clinicians associated with the physician’s organizational National Provider Identifier in PECOS. The length of time that a physician used certified health IT was based on when he or she first attested in the IP.

Physician specialty was primarily drawn from the AMA Masterfile. If data were missing from that file, the specialty was obtained from the PECOS file. Physician specialty was grouped into 8 categories. Primary care specialties included pediatrics, geriatrics, and family, general, adolescent, and internal medicine.

Practice characteristics

Area characteristics were based on the 2012–13 Health Resources and Services Administration Area Health Resources File and CMS’s zip code–level HHS-designated health professional shortage areas (HPSAs) from 2014, and were based on the physicians’ business addresses reported to the IP. Physicians in rural care settings were those located in non-metropolitan counties. Other area characteristics included HHS regions and county-level demographics, including the percentage of the population that was Hispanic, African American, in poverty, or >65 years of age. Each county-level demographic category was parsed into tertiles of equal distribution within the original dataset.

Two markers of delivery reform participation were included. The first was receipt of National Committee for Quality Assurance Patient-Centered Medical Home (PCMH) certification. The second was participation in a Center for Medicare and Medicaid Innovation Pioneer Accountable Care Organization (ACO).

Analytic methods

For outcomes that measure rare events, an odds ratio will approximate a relative risk value; however, using logistic regression to generate relative risk estimates for outcomes with non-rare incidence can produce invalid point estimates and overly wide confidence intervals.^9–12 Other direct relative risk estimates can be obtained through binomial regression, but convergence problems can occur. For these reasons, Poisson regression with a robust variance estimate was used to directly estimate the adjusted relative risks. All analyses were performed using SAS version 9.3.

RESULTS

Physician-level Reporting on NQF0018

NQF0018 reporting includes physicians who reported with technically successful eCQM scores and physicians reporting a zero denominator or numerator. Between 2011 and 2013, slightly more than one-quarter of all physicians participating in the IP reported on NQF0018 (Figure 1 ). Between 2012 and 2013, the number of physicians reporting to the IP increased, but the proportion of those IP participants reporting on NQF0018 remained stable. In 2014, NQF0018 reporting more than doubled to 66.6%. Of the more than 17 million patients represented in NQF0018 reporting in 2014, 62.1% had controlled BP (Table 2).

Figure 1.

Physician reporting of NQF0018^a, Medicare EHR Incentive Program, 2011–2014

Table 2.

Physician reporting through the Medicare EHR incentive program on the number of patients with hypertension and associated hypertension control rate^a

	Hypertension control rate % (n)	Total adult population with hypertension
2011	65.2 (1 199 541)	1 840 162
2012	64.7 (5 007 615)	7 719 936
2013	61.5 (10 135 741)	16 492 203
2014	62.1 (10 724 882)	17 279 552

^aPhysicians participating in the Medicare EHR Incentive Program may report on the optional clinical quality measure NQF0018 (blood pressure control). Physician reporting includes all adult patients with hypertension seen during the reporting period (not limited to just the Medicare population). This table demonstrates the population of patients with hypertension seen by physicians who reported on NQF0018 through the Medicare Incentive Program, and among those, what proportion have controlled hypertension. Number of patients can include duplicated patients, since counts were based on aggregate counts from individual physicians.

The proportion of physicians who reported on NQF0018 but reported no patients with hypertension (zero denominator reporting) increased from <1% in 2011–13 to 11.9% in 2014 (Figure 2 ). There was a similar, although smaller, increase in the proportion of physicians who reported a zero numerator.

Figure 2.

Physician reporting and performance trends for NQF0018^a, within the Medicare EHR Incentive Program, 2011–2014

Clinical performance

Physician reports of ≥70% hypertension control dropped from a high of 40.6% in 2011 to 29.1% in 2014. Multivariate regression analyses indicated that clinical performance was positively associated with length of use of certified health IT, delivery reform participation, and physician specialty (Table 3). Physicians using certified health IT for 4 years were more likely to report >70% hypertension control than those in their first year of use (relative risk [RR] = 1.15; 95% confidence interval [CI], 1.12–1.18). Physicians in PCMH-certified practices and those participating in Pioneer ACOs were more likely to report >70% hypertension control than their corresponding counterparts (RR = 1.19 and 1.06, respectively).

Table 3.

Physician and practice characteristics associated with clinical and technical performance on NQF0018^a

Physician/Practice Characteristic	Percent (RR, 95% CI) of physicians with >70% hypertension controlb	Percent (RR, 95% CI) who reported zero denominatorc	Percent (RR, 95% CI) who reported zero numeratord
No. years reporting with certified health IT
1 year (Ref)	32	20.5	8.5
2 years	31	8.2	6.2
	(1.00, 0.98–1.02)	(0.75, 0.72–0.79)	(0.68, 0.62–0.74)
3 years	36	14.8	4.3
	(1.09, 1.07–1.11)	(0.40, 0.36–0.44)	(0.52, 0.45–0.58)
4 years	39	6.5	6.3
	(1.15, 1.12–1.18)	(0.35, 0.30–0.41)	(0.34, 0.24–0.44)
Rural setting
No (Ref)	34.7	12.2	2.8
Yes	35.8	7.7	5.2
	(0.94, 0.91–0.97)	(0.83, 0.76–0.91)	(0.78, 0.66–0.89)
Specialty
Primary care (Ref)	42.0	3.7	1.9
Other	27.5	21.7	8.8
	(0.72, 0.70–0.74)	(5.33, 5.26–5.40)	(4.61, 4.53–4.70)
Obstetrics/gynecology	41.7	10.2	6.3
	(1.17, 1.14–1.20)	(2.49, 2.38–2.59)	(2.75, 2.62–2.88)
Neurology	29.7	8.9	5.8
	(0.81, 0.76–0.85)	(2.43, 2.30–2.55)	(3.40, 3.26–3.54)
Nephrology	22.3	2.4	3.2
	(0.61, 0.57–0.66)	(0.58, 0.34–0.81)	(1.07, 1.19–1.56)
Endocrinology	42.2	2.8	2.5
	(1.10, 1.05–1.14)	(0.78, 0.51–1.04)	(1.07, 0.80–1.33)
Cardiovascular surgery	26.0	7.0	5.6
	(0.75, 0.70–0.80)	(1.80, 1.66–1.95)	(2.86, 2.71–3.01)
Cardiology	36.3	3.5	1.5
	(1.00. 0.97–1.02)	(0.91, 0.78–1.04)	(0.78, 0.59–0.93)
Practice Size
Solo practice (Ref)	39.7	17.0	6.2
2–5 clinicians	36.3	18.0	5.9
	(0.94, 0.90–0.97)	(1.08, 1.02–1.14)	(1.02, 0.92–1.13)
6–10 clinicians	36.6	16.0	6.3
	(0.95, 0.91–0.99)	(0.99, 0.92–1.05)	(0.91, 0.96–1.19)
11–50 clinicians	34.4	14.4	5.3
	(0.89, 0.86–0.92)	(0.88, 0.82–0.94)	(0.91, 0.82–1.01)
51+ clinicians	34.6	8.4	5.8
	(0.86, 0.83–0.89)	(0.54, 0.49–0.60)	(0.61, 0.52–0.70)
NCQA Patient-Centered Medical Home certified
No (Ref)	33.5	12.6	5.3
Yes	48.7	2.2	1.5
	(1.18, 1.16–1.21)	(0.62, 0.44–0.80)	(0.48, 0.25–0.70)
Pioneer Accountable Care Organization participant
No (Ref)	34.7	12.1	5.1
Yes	37.2	5.2	3.2
	(1.05, 1.01–1.08)	(0.67, 0.55–0.79)	(0.67, 0.53–0.82)

^aNQF0018 is a clinical quality measure, titled “Controlling High Blood Pressure,” which is defined as the percentage of all patients ages 18–85 years who had a diagnosis of hypertension within the first 6 months of the measurement period, or any period of time before the measurement period, whose BP was adequately controlled (<140/90 mmHg) during the measurement period. The clinical target is ≥70% hypertension control. This table provides regression results that controlled for all characteristics listed in the Supplementary File 1. ^bIncludes all physicians reporting to the Medicare EHR incentive program who reported on NQF0018, excluding those who reported on NQF0018 with clinically meaningless results (zero numerator or denominator). ^cRegression among all physicians who reported on NQF0018. ^dZero numerator is considered if physicians reported a non-zero denominator; regression among all physicians who reported on NQF0018.

Ref = referent group; NCQA = National Committee for Quality Assurance; RR = relative risk; CI = confidence interval.

Only obstetricians/gynecologists (RR, 1.17; 95% CI, 1.14–1.20) and endocrinologists (RR, 1.10; 95% CI, 1.10–1.14) were more likely to report >70% hypertension control than primary care physicians. Primary care physicians and cardiologists performed equally well.

Although approximately 1 percentage point more physicians in rural areas reported >70% hypertension control than physicians in urban areas, when controlling for other factors in the regression model rural-based physicians were less likely to achieve the clinical target than urban physicians. Physicians in primary care HPSAs were less likely to report >70% hypertension control than physicians not in HPSAs (RR, 0.80; 95% CI, 0.74–0.86) (Supplementary File 1). There were also differences in clinical performance by region and social determinants of health. For example, physicians located in counties with higher percentages of the population in poverty were less likely to report >70% hypertension control (RR, 0.88; 95% CI, 0.86–0.90).

Zero denominator reporting

“Zero denominator” reporting of NQF0018 occurred when the hypertension diagnosis was not detected by the eCQM logic used by the EHR for measure calculation. In multivariate regression models, physician specialty was most strongly associated with zero denominator reporting; physicians designated as other medical or surgical specialties, obstetricians/gynecologists, and neurologists were more likely to report zero denominators than primary care physicians (RR, 1.80–5.3) (Table 3). Physicians in practices with >10 clinicians and practices in rural areas were less likely to report a zero denominator than their respective counterparts (RR, 0.54–0.88). Longer use of certified health IT (RR, 0.35–0.75) and participation in ACO (RR, 0.67; 95% CI, 0.55–0.79) and PCMH (RR, 0.62; 95% CI, 0.44–0.80) programs were also associated with a lower likelihood of reporting a zero denominator.

Zero numerator reporting

“Zero numerator” reporting of NQF0018 occurs when BP readings are not captured by the eCQM logic used for measure calculation by the EHR. Physicians designated as other medical or surgical specialties (RR, 4.61; 95% CI, 4.53–4.70), neurologists (RR, 3.40; 95% CI, 3.26–3.54), obstetricians/gynecologists (RR, 2.75; 95% CI, 2.62–2.88), cardiovascular surgeons (RR, 2.86; 95% CI, 2.71–3.01), and nephrologists (RR, 1.38; 95% CI, 1.19–1.56) were more likely than primary care physicians to report a zero in the numerator (Table 3). Physicians with longer use of certified health IT (RR, 0.34–0.68), who were in PCMH-certified practices (RR, 0.48; 95% CI, 0.25–0.70), or who participated in a Pioneer ACO (RR, 0.67; 95% CI, 0.53–0.82) were less likely to report a zero numerator compared to their respective counterparts.

DISCUSSION

These results demonstrate population-level variations among physicians in technical and clinical performance for NQF0018 and suggest steps to inform both.

NQF0018 scores were highest in the first year of the IP. Because of the time it takes to adopt and implement certified health IT, it is likely that physicians skilled in hypertension management or those with a significant number of patients with the condition selected this measure in their inaugural year of the IP. It is also likely that overall clinical performance declined as clinicians who were newer health IT adopters began reporting on NQF0018. This hypothesis is supported by these results: physicians with more experience using certified health IT systems were more successful both technically and clinically on NQF0018 than physicians without that experience. This finding suggests that long-term participation in the IP may be a valuable contributor to better outcomes for patients.

For reporting years 2011–13, NQF0018 was one of many optional eCQMs available in the IP. As a result, physicians who reported on NQF0018 during those years may have actively selected the measure for the reasons stated earlier (common diagnosis, frequently managed) or because it was part of an ongoing practice improvement activity. With the introduction of the adult “recommended core” panel in 2014 that allowed participants to report on a preselected set of eCQMs, many physicians or, in some cases, their health care organizations, may have opted to report on the panel rather than selecting individual measures that were relevant to the clinician’s specialty or practice. In this scenario, some participants reporting on the adult panel may have been unaware that they were submitting a measure associated with controlling BP. This is supported by the fact that physicians who reported only in 2014 were more likely to report zero values for either denominator or numerator and were less likely to report >70% hypertension control. The increase in zero reporting likely represents an unintended consequence of an effort to reduce physician reporting burden through use of a panel of preselected measures.

In multivariate analyses that controlled for physician and practice characteristics, rural-area physicians were less likely than urban-area physicians to report >70% hypertension control, although rural-area physicians’ unadjusted performance was higher. This is most likely due to the fact that many rural-area physicians specialize in primary care. Similar shifts after controlling for physician specialty have been reported among rural-area clinician populations when analyses controlled for participation in technical assistance programs and practice size.¹³ Rural clinician performance on other aspects of the IP have been linked to physician specialty and practice size.^14,15 Other work has demonstrated, among small-practice physicians, improved performance on a hypertension control measure associated with pay-for-performance programs.¹⁶ While this study did not control for technical assistance program participation, it did control for practice size, physician specialty, and participation in delivery system reform programs. It may be that those characteristics were driving performance among rural physicians and their clinical performance declined in the absence of those characteristics. Further exploration of what is driving clinical performance among rural clinicians is warranted.

Although these analyses identify a small but increased number of physicians reporting zero values in the numerator and denominator, the national hypertension control findings reported here were similar to other research on clinical populations.^17,18 This suggests that physicians who reported valid numerator and denominator values were reporting data that were consistent with their practice’s overall rate of BP control.

Both clinical and technical performance were strongly associated with physician specialty, certified health IT experience, and participation in delivery system reform activities. Data in these analyses do not provide insights into patient-level characteristics that make hypertension control more or less challenging, but it is likely that there are differences in patient populations among these practices that contribute to variations in clinical performance. These findings suggest 4 domains into which physicians could be categorized based on clinical and technical performance and identify areas for future research. We propose a framework that highlights each domain’s primary challenge: (1) maintain excellence, which applies to physician cohorts with high clinical and technical performance; (2) improve clinical performance, which applies to physician cohorts with high technical performance but low clinical performance; (3) improve technical performance, which applies to physician cohorts that generally have low technical performance, but when reporting non-zero numerators and denominators have high clinical performance; and (4) improve both technical and clinical performance, which applies to physician cohorts with low clinical and technical performance (Figure 3 ).

Figure 3.

Electronic clinical quality measure performance evaluation model

The physician populations identified in group 1 were more likely to demonstrate high clinical and technical performance. This “maintain excellence” group was more likely to be physicians who have used certified health IT for >2 years, participated in delivery reform efforts (PCMH and Pioneer ACO), or were primary managers of their patients' hypertension (primary care specialists, endocrinologists, and cardiologists).

The physician populations that comprise group 2 demonstrate low zero reporting, but also low clinical performance. These populations tend to be physicians in rural areas or large practices. Physicians in this group most likely understand the measure and the way their EHR system calculates it, but are challenged in achieving high BP control rates. Coupling regular review of practice data with clinical quality improvement (CQI) strategies may be of particular value for this group. These strategies could include engaging patients in self-monitoring of BP using electronic exchange of BP readings and treatment advice, implementing standardized hypertension treatment protocols that permit the care team to help patients achieve control, coordinating within large practices to ensure that patients have a treating physician for their hypertension, and utilizing community health workers for additional patient support.

The group 3 cohorts were more likely to have low technical performance; however, when these physician populations were technically successful by reporting non-zero numerators and denominators, they were more likely to report >70% hypertension control. This “improve technical performance” group is populated by physicians with similar characteristics who were on opposite sides of the technical and clinical performance spectrum. The first cohort of group 3 were more likely to report zero numerator or denominator, yet the second cohort, who were technically successful and reported non-zero numerator and denominator values, appeared to manage the condition well. Data input challenges faced by this group could be mitigated through EHR system training, periodic data reviews to ensure that information is captured as necessary, and updating existing workflows, procedures, and policies.

The final physician population has the highest need. Group 4 represents those who failed to achieve either technical or clinical success: physicians in this group were more likely to report zeroes in the numerator and denominator fields, and those who were able to succeed technically were less likely to report >70% hypertension control. Specialists who are unlikely to directly manage hypertension and physicians in small practices primarily constitute this group. For these physicians, the focus should not be on changing their clinical practice to manage hypertension, but rather on selecting and reporting other, more clinically relevant measures and on ensuring patients with hypertension have a treating physician. Among physicians in small practices, possible explanations for these findings include a lack of team members to assist in managing hypertension and a lack of support for updating existing workflows, procedures, and policies to account for the changes that arise as a result of using a health IT system. Although solo practitioners were more likely to report >70% hypertension control than practices with ≥11 physicians, they also had much higher proportions of zero reporting of both numerator and denominator values. Some of the lower clinical performance among physicians in larger practices may be a result of “dilution” from reporting by specialty physicians who do not typically manage hypertension or reporting on subsets of patients with particularly challenging hypertension. The data available in this study are insufficient to identify the causes of differences in outcomes based on practice size; other research has yielded mixed results when exploring the impact of practice size on care quality. Further work should be done to explore this important issue.

A variety of technical assistance programs exist that provide health IT implementation support and help clinicians select eCQMs that are relevant to their specialty and then develop CQI strategies to improve on those clinically relevant measures. For example, the Office of the National Coordinator for Health IT’s (ONC) Regional Extension Center programs help practices track and measure success on Million Hearts^® eCQMs and transition to PCMHs. ONC’s Workforce Development program funds higher education programs to train clinicians on optimal use of health IT.¹⁹ An Agency for Healthcare Research and Quality initiative, EvidenceNOW, is helping clinicians in smaller practices excel on the Million Hearts^® eCQMs through onsite facilitation and coaching, expert consultation, learning collaboratives, and specific health IT assistance.²⁰ CMS offers several programs: Quality Innovation Network–Quality Improvement Organizations work with clinicians to improve health care services through education, outreach, sharing best practices, and using data to measure improvement; the Transforming Clinical Practices Initiative offers CQI support through peer-based learning networks and workforce development programs; and additional funding to support small practice clinicians succeed in the Quality Payment Program.^21–23 For practices with some experience in CQI methods, Million Hearts^® offers evidence-based strategies for improving hypertension control in the “Hypertension Control Change Package for Clinicians.”²⁴ Qualified clinical data registries work with clinicians to choose quality metrics that are most appropriate to their practice. The qualified clinical data registries then extract the appropriate data from the practice’s health IT system, provide interim reports on performance, and submit final reports to the appropriate organization(s). The American Medical Group Foundation’s Measure Up/Pressure Down^® campaign resources help clinicians focus on care processes known to improve hypertension control.²⁵ State and local health departments may have staff trained in CQI approaches who can identify hypertension control strategies for clinicians and patients.²⁶ Ensuring that the clinicians who struggle with the technical and clinical performance of eCQMs have access to the best practices and lessons learned from these programs will be crucial to the overall success of delivery reform programs.

Limitations

Participating clinicians might not be representative of non-Medicare, or nonparticipant, physician populations.⁷ Physicians reporting on NQF0018 may have selected the measure knowing they performed well on it, although there was no financial benefit for better performance. By 2014, however, two-thirds of participating physicians reported on this measure, providing a large amount of data on physician NQF0018 performance.

CQM data were self-reported and were not individually validated by an ONC Authorized Certification Body. Data reported using 2014 edition health IT products, however, had to be generated using health IT that was certified to calculate that measure.²⁷

Between 2011 and 2014, NQF0018 relied on the use of International Classification of Diseases, Ninth Revision, Clinical Modification code 401, Essential hypertension, to generate the measure denominator. Therefore, patients who were classified as having Secondary hypertension (code 405) and patients who exhibited clinical characteristics of hypertension but lacked an official hypertension diagnosis were not counted.^28–30 This may have resulted in an overestimate of hypertension control.

CONCLUSIONS

Physician achievement of hypertension control among their patient population was associated with length of use of certified health IT, physician specialty, and participation in delivery reform efforts. In 2014 there was a striking decrease in technical performance. With careful selection of measures relevant to the physician’s specialty, complete data entry, and support for continuous quality improvement, clinicians can excel both technically and clinically. As care delivery transitions from fee-for-service to quality- and value-based models, these high performers can realize not only financial gains, but also their primary mission of better outcomes for their patients.

DISCLAIMER

Because the data are either publicly available or exempt under 45CFR 46.101(b)(4), this study was exempt from review by the Institutional Review Boards at ONC and CDC. The findings and conclusions in this article are those of the authors and do not necessarily represent the views or official position of the Centers for Disease Control and Prevention. The authors received no financial support for this publication.

SUPPLEMENTARY MATERIAL

Supplementary material is available at Journal of the American Medical Informatics Association online.