Appl Clin Inform 2019; 10(03): 421-445
DOI: 10.1055/s-0039-1692186
Research Article
Georg Thieme Verlag KG Stuttgart · New York

Physicians Voluntarily Using an EHR-Based CDS Tool Improved Patients' Guideline-Related Statin Prescription Rates: A Retrospective Cohort Study

Timothy S. Chang
1  Department of Neurology, University of California, Los Angeles, Los Angeles, California, United States
,
Ashwin Buchipudi
2  Information Services and Solutions, University of California, Los Angeles, Los Angeles, California, United States
,
Gregg C. Fonarow
3  Division of Cardiology, Department of Medicine, University of California, Los Angeles, Los Angeles, California, United States
,
Michael A. Pfeffer
4  Division of General Internal Medicine, Department of Medicine, University of California, Los Angeles, Los Angeles, California, United States
,
Jennifer S. Singer
5  Department of Urology, University of California, Los Angeles, Los Angeles, California, United States
,
Eric M. Cheng
1  Department of Neurology, University of California, Los Angeles, Los Angeles, California, United States
› Author Affiliations
Funding This research was supported by NIH National Center for Advancing Translational Science (NCATS) UCLA CTSI Grant Number UL1TR001881.
Further Information

Address for correspondence

Timothy S. Chang, MD, PhD
Department of Neurology, University of California
Los Angeles, 695 Charles E Young Dr South, Room 2309, Los Angeles, CA 90095
United States   

Publication History

02 January 2019

26 April 2019

Publication Date:
19 June 2019 (online)

 

Abstract

Background In 2013, the American College of Cardiology (ACC) and the American Heart Association (AHA) released a revised guideline on statin therapy initiation. The guideline included a 10-year risk calculation based on regression modeling, which made hand calculation infeasible. Compliance to the guideline has been suboptimal, as many patients were recommended but not prescribed statin therapy. Clinical decision support (CDS) tools may improve statin guideline compliance. Few statin guideline CDS tools evaluated clinical outcome.

Objectives We determined if use of a CDS tool, the statin macro, was associated with increased 2013 ACC/AHA statin guideline compliance at the level of statin prescription versus no statin prescription. We did not determine if each patient's statin prescription met ACC/AHA 2013 therapy intensity recommendations (high vs. moderate vs. low).

Methods The authors developed a clinician-initiated, EHR-embedded statin macro command (“statin macro”) that displayed the 2013 ACC/AHA statin guideline recommendation in the electronic health record documentation. We included patients who had a primary care visit during the study period (January 1–June 30, 2016), were eligible for statin therapy based on the ACC/AHA guideline prior to the study period, and were not prescribed statin therapy prior to the study period. We tested the association of macro usage and statin therapy prescription during the study period using relative risk and mixed effect logistic regression.

Results Subjects included 11,877 patients seen in primary care, who were retrospectively recommended statin therapy at study initiation based on the ACC/AHA guideline, but who had not received statin therapy. During the study period, 125 clinicians used the statin macro command for 389 of the 11,877 patients (3.2%). Of the 389 patients for whom that statin macro was used, 108 patients (28%) had a statin prescribed during the study period. Of the 11,488 for whom the statin macro was not used, 1,360 (13%) patients received a clinician-prescribed statin (relative risk 2.3, p < 0.001). Controlling for patient covariates and clinicians, statin macro usage was significantly associated with statin therapy prescription (odds ratio 2.86, p < 0.001).

Conclusion Although the statin macro had low uptake, its use was associated with a greater rate of statin prescriptions (dosage not determined) for patients whom 2013 ACC/AHA guidelines required statin therapy.


#

Background and Significance

Although many best practice guidelines exist for initiating medication in select patient groups, clinicians prescribe the targeted medications at suboptimal rates.[1] In 2013, the American College of Cardiology (ACC) and the American Heart Association (AHA) released a revised guideline regarding statin therapy initiation[2] (see [Fig. 1] and [Supplementary Material] (available in the online version) for a brief overview of the 2013 ACC/AHA statin guideline). Both the 2013 ACC/AHA statin guideline compliance, and the percentage of patients prescribed statin therapy per the guideline, varied by patient population. In 2015, cardiologists prescribed 2013 guideline-compliant, secondary prevention statin therapy to 91% of their patients with atherosclerotic cardiovascular disease (ASCVD).[3] However, prevention guideline compliance was suboptimal in primary care patients. In the 4 years following 2013 ACC/AHA statin guideline release, 42% of patients with no history of ASCVD and a 10-year ASCVD risk score > 7.5% received statin prescription in accordance with the ACC/AHA statin guideline.[4]

Zoom Image
Fig. 1 Central schematic including simplified version of the 2013 American College of Cardiology/American Heart Association (ACC/AHA) statin guideline and study design. For the statin guidelines, patients with a history of ASCVD, LDL ≥ 190 mg/dL, or DM aged 40 to 75 years old were recommended statin therapy. Patients without DM aged 40 to 75 years old and an LDL of 70 to 189 mg/dL necessitated 10-year ASCVD risk calculation. Patients with a 10-year ASCVD risk calculation ≥ 5% were recommended statin therapy while those with a 10-year ASCVD risk calculation < 5% were not recommended statin therapy. For the study design, patients recommended statin therapy for primary prevention (LDL ≥ 190 mg/dL; DM and 40–75 years old; no DM, LDL 70–189 mg/dL and 10-year ASCVD risk ≥ 5%) but not prescribed statin therapy as of December 31, 2015, were included in this study. We tested the association of macro usage and statin therapy prescription during the study period (January 1, 2016–June 30, 2016). Abbreviations: ASCVD, atherosclerotic cardiovascular disease; ASCVD risk, 10-year ASCVD risk calculation; DM, diabetes mellitus; LDL, low-density lipoprotein; Rx, prescription; YO, years old.

A postulated reason for suboptimal 2013 ACC/AHA statin guideline compliance was its complexity.[5] [6] While previous statin guidelines utilized integer-based risk scores that could be determined by hand,[7] the 2013 ACC/AHA statin guideline utilized four regression-based risk scoring equations that required use of a calculator. Which 2013 ACC/AHA regression equation to use depended on a patient's gender and race. Each regression equation incorporated age, high-density lipoprotein (HDL), total cholesterol, diabetes mellitus (DM) history, systolic blood pressure, antihypertensive medication, and smoking status. The 10-year ASCVD risk calculation determined if statin therapy should be initiated in patients without DM and with low-density lipoprotein (LDL) values between 70 and 189 mg/dL. Although other existing 10-year risk calculators such as Framingham[8] and QRISK[9] were available, the ACC/AHA developed their 2013 new 10-year risk calculator based on regression equations. Concurrent with the 2013 ACC/AHA statin guideline publication, the ACC/AHA released mobile and online ASCVD Risk Estimator calculators.[10] [11] However, clinicians had to manually enter information, which was time consuming.[12] These calculators were available prior to the studies showing suboptimal 2013 ACC/AHA statin guideline compliance.[4] The current study authors hypothesized that a clinical decision support (CDS) tool that automatically retrieved patient data and performed regression calculation could improve 2013 ACC/AHA statin guideline compliance.

Previous CDS tools improved guideline compliance with modest benefit[13] for DM,[14] heart failure,[15] and pneumonia.[16] Childhood vaccination rates improved when implemented within electronic health record (EHR) templates, with preloaded immunization records, and alerts.[17] In randomized control trials, CDS tools with features such as automation, on-screen display, system initiation, and advice to patients as well as clinicians were more effective. Features such as advise within charting or order entry were less effective.[18] [19]

A systematic review identified 34 previous health care intervention tools tested in randomized control trials for lipid management.[20] Five CDS tools were integrated into an EHR. For example, the MayoExpertAdvisor study, based on a Web service with prefilled patient data from the EHR, provided 2013 ACC/AHA ASCVD risk calculation. That study showed decreased time for clinicians to determine a statin therapy recommendation, but did not report whether guideline compliance improved.[21] In a systematic review, a low percentage of CDS tools evaluated clinical benefit.[22]

Within the authors' EHR system, Epic,[23] clinicians can initiate CDS tools during note documentation. Such tools can automatically retrieve patient data and perform calculations. Therefore, study clinicians did not need to exit the EHR to perform calculations on an external platform. Previous macro commands were developed at other institutions for obesity counseling[24] and H1N1 swine flu recommendations.[25] However, those macros did not retrieve patient data or perform calculations.

We implemented the study's statin macro in July 2014, shortly after the November 2013 ACC/AHA guideline and online calculator publications. Unlike earlier statin CDS tools elsewhere that were not integrated into the EHR,[21] the current study's statin macro was accessible during EHR note generation. The previous tools had been printed on paper forms,[26] [27] shown on a separate screen,[14] [28] or emailed to clinicians.[29]


#

Objectives

We determined if the statin macro was associated with statin prescription, regardless of dose, per the 2013 ACC/AHA statin guidelines. Specifically, for patients retrospectively recommended (at the time of study initiation) but not prescribed statin therapy, we investigated whether subsequent statin macro usage was associated with statin prescription. We did not determine if each patient's statin prescription met the ACC/AHA 2013 therapy intensity recommendations (high vs. moderate vs. low). [Fig. 1] shows a central schematic of the study design.


#

Methods

Statin Macro Development

After the November 2013 ACC/AHA statin guideline was published, and prior to the statin macro implementation, we received multiple requests from clinicians to provide a CDS tool for the ACC/AHA statin guideline. We did not assess clinician knowledge of or compliance with the 2013 ACC/AHA statin guideline in our health system. We hypothesized a statin macro could improve knowledge about the 2013 ACC/AHA statin guideline and also guideline compliance.

A multidisciplinary expert panel including cardiologists, internists, neurologists, clinical informaticians, and quality officers developed the statin macro, which was based on the published ACC/AHA statin guideline. Developers tested hundreds of patients for accuracy of the macro's statin guideline recommendations. During statin macro development, expert panel members revised it several times prior to the study. Revisions led to statin therapy recommendations based on 17 scenarios of a patient's ASCVD history, DM history, LDL level, 10-year ASCVD risk, and medication list ([Supplementary Table S1], available in the online version). For example, the macro suppressed a statin recommendation if the patient had a statin allergy. Multiple governance committees, including the institutions' primary care leadership, approved use of the macro. The study institution's Institutional Review Board (IRB) approved a waiver of authorization for this study (IRB#: 16–001676).

The clinician-initiated CDS tool was relevant and manageable. The statin macro made a clear recommendation of statin therapy or no statin therapy. The statin macro contained a hyperlink to the 2013 ACC/AHA statin guideline and the ACC/AHA ASCVD Risk Estimator online calculator.[10] The ASCVD Risk Estimator hyperlink allowed clinicians to manually confirm macro recommendations. Errors found in the statin macro could be logged as a ticket to the EHR help desk.

In the study site's Epic EHR terminology, the CDS tool we developed is known as a Smart Phrase. Authors refer to the Smart Phrase herein by its generic name, “macro command,” to avoid vendor-specific terminology. While documenting a note, clinicians could type an abbreviated phrase such as “CVRISK” to invoke the statin macro directly and incorporate its output into the note. Macros of this sort are clinician-initiated CDS tools that are “pulled” by clinicians. Clinician-initiated CDS tools contrast with system-initiated CDS tools that are automatically “pushed” to clinicians as alerts. [Fig. 2] shows a screenshot of the statin macro within a note. Development of the statin macro was consistent with the GUIDES checklist.[30] The statin macro automatically retrieved patient data from the EHR. Developers tested hundreds of patients for accuracy of patient data retrieval. In addition to activating the statin macro directly while documenting a clinical note, clinicians could alternatively add the macro to a note template, allowing the macro to upload with every use of the encompassing template.

Zoom Image
Fig. 2 Screenshot of the statin macro within a note. Blue highlighted text was variable and specific to each patient. The blue “2013 American College of Cardiology/American Heart Association (ACC/AHA) guideline” text included a hyperlink to the 2013 ACC/AHA statin guideline. The blue “here” text included a hyperlink to the online ASCVD Risk Estimator calculator (Screenshot used with permission from © 2019 Epic Systems Corporation).

While EHRs have high consistency, they do not have 100% completeness or correctness.[31] [32] The physician-targeted statin macro delivered consistent, on demand, and fast recommendations within the clinicians' note documentation workflow. Statin macro versions included those that showed a summary and those that showed all variable values ([Table 1]). Variables were highlighted in blue. Clinicians could customize CDS delivery with these variations.

Table 1

Statin macro examples

Version 1: Recommendation

Hyperlinked 2013 ACC/AHA guideline followed by recommendation and statin status

2013 ACC/AHA guideline recommends moderate or high-intensity statin because 10-year ASCVD risk ≥ 7.5%. Patient is not taking a statin[a]

Version 2: Brief

Hyperlinked 2013 ACC/AHA guideline followed by recommendation and statin status

The second line shows the risk calculator results if it should be calculated

2013 ACC/AHA guideline recommends moderate or high-intensity statin because 10-year ASCVD risk ≥ 7.5%. Patient is not taking a statin. Ten-year ASCVD risk is 9.0% as of 1:15 pm on January 1, 2016

Version 3: Full

Hyperlinked 2013 ACC/AHA guideline followed by recommendation and statin status

The second line shows the risk calculator results if it should be calculated

The third line shows the optimal risk score

The following lines show the values used to calculate ASCVD score

2013 ACC/AHA guideline recommends moderate or high-intensity statin because 10-year ASCVD risk ≥ 7.5%. Patient is not taking a statin

Ten-year ASCVD risk is 9.0% as of 1:15 pm on January 1, 2016

Ten-year ASCVD risk with optimal risk factors is 3.6%

Values used to calculate ASCVD score:

Age: 55 years old

Gender: Male

Race: not African American

HDL cholesterol: 30 mg/dL. HDL cholesterol measured 60 days ago

Total cholesterol: 200 mg/dL. Total cholesterol measured 60 days ago

Systolic BP: 130 mm Hg. BP was measured 2 days ago

The patient is not being treated with medication that influences SBP

The patient is currently not a smoker

The patient does not have a diagnosis of diabetes

Click here for the 2013 ACC/AHA Cardiovascular Risk Estimator tool (online calculator)

Abbreviations: ACC, American College of Cardiology; AHA, American Heart Association; ASCVD, atherosclerotic cardiovascular disease; BP, blood pressure; HDL, high-density lipoprotein; SBP, systolic blood pressure.


a Blue highlighted text was variable and specific to each patient.



#

Statin Macro Implementation

The study institution installed an enterprise-wide version of the EHR from Epic[23] in over 170 urban clinics and 3 hospitals at a large, urban academic medical center from March 2013 to July 2014. Epic was installed in all primary care clinics by July 11, 2013. The statin macro was first accessible to clinicians on July 21, 2014. January 1, 2016 to June 30, 2016 defined the study period.

One of the study sites' biweekly ambulatory EHR email updates described the statin macro and encouraged usage. Clinician awareness of the statin macro also spread through word of mouth. As a clinician-initiated CDS tool, clinicians using the macro were likely motivated and believed its usage would improve patient care. The study did not measure the study clinicians' assessment of their premacro-usage likelihood of prescribing a statin for the patient at hand.

The email advertising the statin macro included instructions. Prior to statin macro implementation, we did not assess factors that would influence guideline compliance. Because the statin macro was a clinician-initiated CDS tool, clinicians were not forced to follow recommendations or explain why they did not follow recommendations. The institution's clinical leadership supported the statin macro but did not provide incentives for statin guideline compliance.


#

Study Criteria

January 1, 2016 to June 30, 2016 defined the study period. We chose this 6-month period based on prestudy estimates for the number of times clinicians used the statin macro. As a retrospective cohort study, clinicians did not know during the study period they would be included in this study. Inclusion criteria were patients:

  • Who had a primary care visit during the study period. Any outpatient visit with a patient's primary care clinician defined a primary care visit.

  • Who were 40 to 75 years old as of January 1, 2016. The 10-year ASCVD risk calculator was developed for this age range.

  • Who did not have a statin prescription before the study period as of December 31, 2015.

  • Who were retrospectively identified as candidates for statin therapy based on the 2013 ACC/AHA statin guideline before the study period as of December 31, 2015.

Exclusion criteria were patients:

  • Who had a statin allergy.

  • Who had a history of ASCVD. Patients with a history of ASCVD were recommended statin therapy based on the 2013 ACC/AHA statin guideline for secondary prevention.

  • Who did have a statin prescription before the study period as of December 31, 2015.

  • Who did not have sufficient data to determine the 2013 ACC/AHA statin guideline recommendation. For example, some patients did not have a LDL measurement or were missing data necessary to calculate the 10-year ASCVD risk (e.g., blood pressure measurement).

  • Who were not recommended statin therapy based on the 2013 ACC/AHA statin guideline before the study period as of December 31, 2015.

We extracted EHR data as of December 31, 2015, necessary to determine the ACC/AHA statin guideline recommendation. Data included the patients' age, gender, smoking status, visit diagnoses, and problem list. We extracted the most recent data prior to December 31, 2015, for blood pressure (looking back to January 1, 2014) and cholesterol (total, low-density, and high-density: looking back to January 1, 2011). We extracted statin and antihypertensive medication (December 31, 2015–June 30, 2016, including start and end date of medications), allergies (as of June 30, 2016), and statin macro usage (January 1, 2016–June 30, 2016). Searching note text for “2013 ACC/AHA guideline*10-year ASCVD risk,” where * indicated additional text that may be between these phrases, identified statin macro usage. All statin macro versions had these phrases.


#

Exposure and Outcome

Any version of statin macro usage during the study period defined the exposure. Statin therapy prescription of any dosage during the study period defined the outcome. A statin prescription with a start date from January 1, 2016, to June 30, 2016, or an end date after June 30, 2016, defined statin prescription during the study period. Prescriptions that were hand written or called into a pharmacy, and not entered in the EHR, were not accounted for.


#

Analyses

For patients meeting the inclusion/exclusion criteria, we compared variables between patients for whom the statin macro was and was not used. The Student's t-test was used for continuous variables (age, systolic blood pressure, total cholesterol, LDL, HDL) and the binomial test was used for binary variables (gender, race, smoking, DM, antihypertensive medication). For clinicians who did and did not use the statin macro, we compared clinician type, gender, and specialty using the binomial test.

We calculated the relative risk of statin therapy prescription for macro usage compared with no macro usage. To control for patient covariates and clinicians, we used a mixed effect logistic regression model. The response variable was statin therapy prescription. The dependent variables included macro usage and variables used for the ACC/AHA statin guideline recommendation. As multiple patients may have been seen by the same primary care clinician, we included patients' clinician as a random effect, where each clinician had a specific random intercept. We used the likelihood ratio test to determine the significance of variables in the model. Wald confidence intervals were calculated for odds ratios of fixed effect variables in the model.[33]

In this primary analysis, patients without complete data were removed. Because missing data precluded statin recommendation for some patients, we repeated analysis on the final imputed data set of a multiple imputation procedure with predictive mean matching[34] [35] [36] using the Hmisc package.[37] See [Supplementary Material] (available in the online version) for further details on missing data analysis. All analyses were performed in R (version 3.5.2).[38]


#
#

Results

From January 1, 2016, to June 30, 2016, 72,315 patients aged 40 to 75 years had a primary care visit. Of those, 60,438 patients were dropped based on the study exclusion criteria ([Fig. 3]). Thus, 11,877 study eligible patients met the inclusion/exclusion criteria. [Table 2] shows baseline characteristics of patients included in the study. DM was significantly less common in the statin macro usage group (11% vs. 16%, p < 0.05). In contrast, LDL levels were higher in the macro usage group (127 vs. 123 mg/dL, p < 0.05).

Zoom Image
Fig. 3 Flow diagram showing study exclusion criteria, inclusion criteria, and results. 72,315 patients aged 40 to 75 years had a primary care visit during the study period. A total of 60,438 patients were excluded. Counts for categories of excluded patients are shown. 11,877 patients were recommended statin therapy based on the 2013 American College of Cardiology/American Heart Association (ACC/AHA) statin guideline but not prescribed statin therapy prior to the study period. Counts for categories of included patients are shown. Statin therapy was prescribed during the study period in 28% (108 of 389) of patients for whom the statin macro was used compared with 13% (1,360 of 11,488) of patients for whom the statin macro was not used. Abbreviations: ASCVD, atherosclerotic cardiovascular disease; DM, diabetes mellitus; LDL, low-density lipoprotein.
Table 2

Baseline characteristics of patients meeting the inclusion/exclusion criteria stratified by statin macro usage (N = 11,877)

Statin macro usage

(N = 389)

No statin macro usage

(N = 11,488)

p-Value

Male

55% (213)

55% (6,322)

0.96

Age (y)

62.8 (8.1)

63.3 (7.7)

0.23

Black

11% (44)

8.9% (1,019)

0.12

Smoke

8.2% (32)

7.1% (810)

0.43

DM

11% (42)

16% (1,863)

0.005

Antihypertensive

39% (151)

39% (4,510)

0.9

Systolic BP (mm Hg)

132 (17)

132 (17)

0.84

Total cholesterol (mg/dL)

210 (35)

207 (37)

0.15

LDL (mg/dL)

127 (29)

123 (31)

0.006

HDL (mg/dL)

57 (16)

59 (19)

0.13

Abbreviations: BP, blood pressure; DM, diabetes mellitus; HDL, high-density lipoprotein; LDL, low-density lipoprotein.


Note: Numbers in parenthesis correspond to total patients or standard deviation for binary and continuous variables, respectively.


A total of 443 primary care clinicians cared for study eligible patients. Of those, 440 clinicians did not use the statin macro on at least one patient while 125 clinicians used the statin macro on at least one patient. Because they used the statin macro for some patients and did not use the statin macro for other patients, 122 clinicians were in both groups. [Table 3] compares type, gender, and specialty of clinicians who did and did not use the statin macro. Most clinicians were internal medicine and family medicine physicians. There were significantly more residents who did not use the statin macro (5%) compared with those who used the statin macro (1%).

Table 3

Characteristics for clinicians of patients who met the inclusion/exclusion criteria stratified by statin macro usage (N = 443)

Statin macro usage on at least 1 patient

(N = 125)

No statin macro usage on at least 1 patient

(N = 440)

p-Value

Type

 Physician

98% (123)

92% (406)

0.023

 Resident

1% (1)

5% (24)

0.047

 Fellow

1% (1)

2% (7)

0.82

 Nurse practitioner

0% (0)

1% (3)

0.82

Sex

 Male

54% (67)

45% (199)

0.12

 Female

46% (58)

49% (215)

0.70

 Unknown

0% (0)

6% (26)

0.011

Specialty

 Internal medicine

66% (83)

71% (310)

0.45

 Family medicine

34% (42)

25% (108)

0.06

 Surgery

0% (0)

2% (7)

0.34

 Neurology

0% (0)

1% (4)

0.64

 Obstetrics and gynecology

0% (0)

1% (3)

0.82

 Unknown

0% (0)

2% (8)

0.28

Note: Numbers in parenthesis correspond to total patients.


For each clinician who used the statin macro at least once in study eligible or ineligible patients, [Table 4] shows the total number of patients seen, statin macro usage, and statin prescription stratified by patient study eligibility. The number of study eligible patients seen per clinician who used the statin macro at least once ranged from 1 to 284. For study eligible patients, the top 33 clinician statin macro users (of 125 physicians) contributed 62% of all macro usages. For each clinician who never used the statin macro and prescribed statin therapy at least once, [Table 5] shows the number of patients seen and statin prescription in study eligible and ineligible patients. The number of study eligible patients seen per clinician who never used the statin macro and prescribed statin therapy at least once ranged from 1 to 197.

Table 4

For clinicians who used the statin macro at least once, clinician statin macro usage and prescription stratified by patient study eligibility

Study eligible patients

Study ineligible patients

Clinician ID

Total no. of patients

No. of patients macro used

No. of patients macro used more than once

No. of patients statin prescribed (macro used, macro not used)

Total no. of patients

No. of patients macro used

No. of patients macro used more than once

No. of patients statin prescribed (macro used, macro not used)

16419

284

20

0

19 (5, 14)

657

19

0

290 (11, 279)

18983

222

1

1

4 (1, 3)

942

3

0

139 (1, 138)

15692

212

4

0

18 (3, 15)

614

2

0

269 (1, 268)

15026

189

12

2

23 (7, 16)

1,085

43

0

249 (15, 234)

21490

173

1

0

11 (0, 11)

442

3

0

150 (1, 149)

17711

167

1

1

12 (1, 11)

549

2

0

196 (1, 195)

19343

165

3

0

20 (2, 18)

553

2

0

214 (0, 214)

15688

153

2

0

13 (1, 12)

424

5

0

198 (2, 196)

19301

140

9

2

5 (1, 4)

256

9

2

69 (0, 69)

16389

139

0

0

17 (0, 17)

634

2

0

325 (1, 324)

15007

132

1

0

11 (0, 11)

552

2

0

143 (0, 143)

15684

131

11

2

20 (7, 13)

588

25

3

185 (5, 180)

26752

131

1

0

24 (0, 24)

498

6

0

196 (0, 196)

9121

130

1

0

4 (0, 4)

451

2

0

117 (0, 117)

29618

127

3

0

8 (1, 7)

336

4

0

80 (0, 80)

11186

126

13

1

8 (2, 6)

273

9

0

115 (1, 114)

20857

116

1

0

13 (0, 13)

434

2

0

115 (0, 115)

19609

110

3

0

9 (1, 8)

438

2

1

108 (1, 107)

14796

109

1

0

13 (1, 12)

259

1

0

116 (0, 116)

15368

102

5

0

10 (2, 8)

427

6

0

115 (2, 113)

23513

101

0

0

11 (0, 11)

261

1

0

95 (0, 95)

13589

100

3

0

2 (1, 1)

381

14

0

85 (3, 82)

15267

100

0

0

8 (0, 8)

240

1

0

76 (0, 76)

17479

99

2

0

11 (1, 10)

554

5

0

213 (0, 213)

5838

96

11

1

8 (1, 7)

439

29

1

147 (2, 145)

14749

91

3

0

15 (1, 14)

336

13

3

117 (2, 115)

15348

91

5

1

18 (1, 17)

358

15

0

126 (2, 124)

9825

89

11

0

24 (3, 21)

373

15

1

163 (1, 162)

7776

88

0

0

4 (0, 4)

222

1

0

58 (1, 57)

25950

86

4

0

7 (1, 6)

371

9

0

95 (4, 91)

30320

86

0

0

10 (0, 10)

540

2

0

106 (0, 106)

31344

85

7

0

15 (1, 14)

324

7

0

87 (0, 87)

18108

84

5

0

23 (3, 20)

562

11

0

204 (1, 203)

26751

84

1

0

14 (0, 14)

400

6

0

153 (2, 151)

14900

83

4

0

14 (1, 13)

423

6

0

132 (3, 129)

31643

83

0

0

19 (0, 19)

284

1

0

103 (0, 103)

30342

82

0

0

7 (0, 7)

560

1

0

97 (0, 97)

30340

80

3

0

23 (3, 20)

831

31

0

207 (10, 197)

28599

79

4

0

18 (2, 16)

307

5

0

103 (3, 100)

17436

78

0

0

10 (0, 10)

420

4

0

153 (0, 153)

17873

76

9

3

9 (3, 6)

517

18

0

119 (2, 117)

19243

76

6

0

11 (1, 10)

275

1

0

101 (1, 100)

30337

76

5

0

15 (1, 14)

563

18

1

107 (2, 105)

31406

75

3

0

11 (1,10)

411

12

0

128 (1, 127)

19090

74

5

0

8 (1, 7)

315

2

0

63 (1, 62)

6162

71

2

0

8 (1, 7)

474

3

0

180 (1, 179)

14821

70

1

0

8 (0, 8)

186

0

0

74 (0, 74)

27277

70

11

3

3 (0, 3)

322

24

2

98 (5, 93)

30329

68

0

0

11 (0, 11)

687

1

0

178 (0, 178)

27338

67

6

1

5 (0, 5)

333

18

1

81 (5, 76)

25053

66

0

0

7 (0, 7)

344

2

0

108 (1, 107)

26849

66

2

0

3 (0, 3)

185

4

0

51 (0, 51)

27632

63

3

1

10 (0, 10)

283

4

0

53 (1, 52)

15000

62

4

1

5 (2, 3)

144

2

0

70 (0, 70)

28934

62

0

0

14 (0, 14)

305

11

0

110 (6, 104)

29225

62

4

0

7 (1, 6)

367

7

0

118 (6, 112)

22153

61

0

0

3 (0, 3)

237

10

1

70 (4, 66)

14891

59

1

0

6 (0, 6)

221

0

0

64 (0, 64)

15240

59

4

1

12 (1, 11)

259

9

1

115 (4, 111)

28297

59

2

0

15 (1, 14)

286

6

0

81 (1, 80)

18276

58

1

0

9 (0, 9)

82

2

0

54 (2, 52)

19733

58

2

0

6 (1, 5)

366

5

0

124 (3, 121)

30311

56

0

0

4 (0, 4)

520

2

1

63 (0, 63)

30332

56

0

0

9 (0, 9)

526

1

0

77 (0, 77)

28044

54

8

0

13 (1, 12)

388

14

0

139 (5, 134)

30327

54

1

0

6 (0, 6)

512

28

0

87 (2, 85)

30341

53

1

0

5 (0, 5)

659

35

1

100 (6, 94)

12997

52

0

0

0 (0, 0)

154

2

0

60 (0, 60)

27079

52

3

1

6 (2, 4)

132

3

0

59 (2, 57)

30322

52

3

2

9 (0, 9)

532

34

5

112 (9, 103)

15384

51

1

0

2 (0, 2)

252

7

1

88 (1, 87)

18673

51

1

0

0 (0, 0)

527

0

0

35 (0, 35)

19506

51

1

0

9 (0, 9)

245

0

0

61 (0, 61)

26777

51

8

0

9 (1, 8)

294

27

0

77 (11, 66)

31365

50

0

0

5 (0, 5)

204

1

0

55 (1, 54)

5996

50

7

0

7 (2, 5)

427

30

1

148 (2, 146)

8183

50

3

0

6 (0, 6)

239

5

0

71 (0, 71)

20912

49

6

1

8 (0, 8)

145

3

0

74 (0, 74)

25156

48

7

3

15 (3, 12)

224

23

2

96 (6, 90)

25372

48

2

0

7 (1, 6)

170

1

0

55 (0, 55)

25720

48

0

0

4 (0, 4)

316

7

0

61 (2, 59)

30394

48

3

0

15 (1, 14)

442

9

0

119 (3, 116)

17237

47

8

1

12 (2, 10)

350

8

0

104 (3, 101)

30408

47

8

4

16 (3, 13)

209

12

1

88 (3, 85)

31351

47

1

0

2 (0, 2)

261

0

0

58 (0, 58)

24906

46

2

0

14 (0, 14)

297

4

1

63 (0, 63)

30343

46

0

0

6 (0, 6)

552

10

1

116 (4, 112)

13939

45

1

0

4 (1, 3)

128

0

0

66 (0, 66)

19338

45

0

0

9 (0, 9)

218

3

0

116 (2, 114)

22879

45

0

0

1 (0, 1)

376

3

0

82 (0, 82)

25225

43

2

0

3 (0, 3)

222

5

0

62 (0, 62)

26841

43

0

0

4 (0, 4)

465

1

0

43 (0, 43)

24399

42

0

0

6 (0, 6)

85

2

0

43 (1, 42)

31430

42

2

0

9 (0, 9)

161

0

0

36 (0, 36)

32271

42

0

0

5 (0, 5)

143

3

0

68 (2, 66)

9187

42

1

0

3 (0, 3)

129

2

0

60 (0, 60)

19189

41

1

0

8 (1, 7)

334

7

0

93 (2, 91)

14961

40

1

0

4 (0, 4)

332

4

0

64 (0, 64)

27642

39

0

0

9 (0, 9)

181

1

0

63 (1, 62)

31098

39

0

0

2 (0, 2)

414

2

0

37 (1, 36)

30335

37

0

0

9 (0, 9)

301

1

0

65 (1, 64)

23235

36

1

0

5 (0, 5)

286

1

0

51 (0, 51)

15866

35

3

0

10 (2, 8)

244

19

1

115 (6, 109)

18117

35

2

0

9 (0, 9)

192

11

3

41 (0, 41)

31644

35

2

0

6 (1, 5)

244

17

0

54 (4, 50)

11213

34

1

0

8 (0, 8)

106

1

0

48 (0, 48)

22866

34

2

0

5 (1, 4)

250

3

0

62 (0, 62)

27268

34

1

1

8 (1, 7)

211

7

0

55 (2, 53)

23603

33

1

0

0 (0, 0)

163

0

0

20 (0, 20)

31840

33

2

0

11 (1, 10)

168

11

0

32 (2, 30)

13272

32

0

0

7 (0, 7)

482

2

0

69 (0, 69)

29296

32

2

0

11 (2, 9)

123

4

0

44 (1, 43)

8494

32

1

0

6 (0, 6)

75

1

0

40 (1, 39)

24383

30

2

1

4 (0, 4)

141

5

0

46 (0, 46)

6308

30

4

0

5 (0, 5)

72

4

0

54 (2, 52)

Missing

29

1

0

3 (0, 3)

160

4

0

46 (0, 46)

12428

29

2

0

5 (1, 4)

282

24

3

68 (7, 61)

19384

29

4

0

6 (0, 6)

96

3

0

57 (0, 57)

10399

28

0

0

5 (0, 5)

243

1

0

42 (0, 42)

13600

27

1

0

1 (0, 1)

79

1

0

39 (1, 38)

15251

27

2

1

7 (1, 6)

161

1

0

55 (0, 55)

19691

27

0

0

3 (0, 3)

69

1

0

25 (1, 24)

30324

27

0

0

2 (0, 2)

443

4

0

75 (0, 75)

17448

25

0

0

2 (0, 2)

176

5

0

39 (0, 39)

30336

25

0

0

4 (0, 4)

419

20

1

35 (3, 32)

27839

24

3

0

3 (0, 3)

64

0

0

25 (0, 25)

25241

23

0

0

6 (0, 6)

132

5

1

63 (1, 62)

28109

22

1

0

9 (0, 9)

182

15

1

55 (4, 51)

31420

22

1

0

2 (1, 1)

331

10

1

35 (0, 35)

9849

22

1

0

2 (0, 2)

45

3

0

15 (1, 14)

26833

21

0

0

2 (0, 2)

147

2

0

38 (0, 38)

32040

21

0

0

3 (0, 3)

361

1

0

29 (0, 29)

15286

20

1

0

2 (0, 2)

110

4

1

24 (1, 23)

19168

20

0

0

1 (0, 1)

66

1

0

24 (0, 24)

21774

20

1

1

3 (0, 3)

128

2

0

27 (0, 27)

27355

20

0

0

2 (0, 2)

367

1

0

50 (0, 50)

28133

19

3

2

7 (1, 6)

90

7

0

35 (1, 34)

28428

17

0

0

0 (0, 0)

66

1

0

6 (0, 6)

26839

16

0

0

1 (0, 1)

119

2

0

25 (1, 24)

26042

15

2

0

2 (1, 1)

65

1

0

1 (0, 1)

6806

15

1

0

1 (0, 1)

24

0

0

16 (0, 16)

26041

14

1

0

6 (1, 5)

79

4

0

23 (1, 22)

9719

14

1

0

4 (1, 3)

33

3

0

11 (1, 10)

22922

13

0

0

2 (0, 2)

69

1

0

21 (1, 20)

25948

13

0

0

1 (0, 1)

56

5

0

25 (1, 24)

28117

13

0

0

1 (0, 1)

41

4

0

6 (2, 4)

90000005

13

0

0

1 (0, 1)

109

1

0

26 (0, 26)

16367

12

4

1

4 (0, 4)

32

4

0

9 (0, 9)

28040

12

1

0

1 (0, 1)

75

6

0

20 (2, 18)

31014

12

1

0

2 (0, 2)

104

5

0

16 (1, 15)

28039

11

0

0

1 (0, 1)

89

5

0

18 (0, 18)

21032

10

1

0

4 (0, 4)

47

0

0

18 (0, 18)

22142

10

2

1

1 (0, 1)

22

0

0

12 (0, 12)

26838

10

3

0

1 (1, 0)

49

3

0

19 (1, 18)

12413

9

2

0

0 (0, 0)

32

1

0

16 (1, 15)

16570

9

0

0

0 (0, 0)

41

1

0

8 (0, 8)

27912

9

2

0

1 (1, 0)

26

3

1

9 (0, 9)

28857

9

0

0

1 (0, 1)

33

1

0

5 (1, 4)

29484

9

0

0

3 (0, 3)

64

2

0

17 (1, 16)

18575

8

0

0

1 (0, 1)

26

1

0

9 (0, 9)

18661

8

1

0

1 (0, 1)

27

3

0

14 (1, 13)

31977

8

0

0

3 (0, 3)

19

1

0

10 (1, 9)

5944

8

1

0

0 (0, 0)

9

0

0

6 (0, 6)

13089

7

1

1

1 (0, 1)

35

3

0

11 (1, 10)

21782

7

1

0

0 (0, 0)

10

1

0

5 (1, 4)

32336

7

1

0

2 (0, 2)

12

0

0

3 (0, 3)

31975

6

0

0

0 (0, 0)

202

1

0

11 (0, 11)

32626

6

0

0

3 (0, 3)

34

1

0

9 (1, 8)

27207

5

1

0

2 (0, 2)

28

2

0

2 (0, 2)

28878

5

0

0

0 (0, 0)

119

13

0

32 (3, 29)

31849

5

0

0

3 (0, 3)

93

2

0

23 (1, 22)

28883

4

0

0

1 (0, 1)

21

1

0

6 (0, 6)

30488

4

0

0

0 (0, 0)

50

1

0

10 (0, 10)

27471

3

1

0

0 (0, 0)

9

0

0

1 (0, 1)

10682

2

0

0

1 (0, 1)

2

1

0

2 (1, 1)

13088

2

0

0

0 (0, 0)

19

2

0

6 (0, 6)

16535

2

0

0

1 (0, 1)

9

1

1

5 (1, 4)

19413

2

0

0

1 (0, 1)

47

1

0

8 (0, 8)

29736

2

0

0

1 (0, 1)

7

1

0

6 (0, 6)

30640

2

1

0

0 (0, 0)

9

2

0

6 (2, 4)

30647

2

0

0

1 (0, 1)

10

2

0

1 (0, 1)

30656

2

1

0

1 (1, 0)

5

0

0

3 (0, 3)

31340

2

0

0

0 (0, 0)

54

2

0

9 (1, 8)

31574

2

1

1

2 (1, 1)

6

0

0

1 (0, 1)

31583

2

0

0

0 (0, 0)

6

1

0

4 (1, 3)

31598

2

1

0

0 (0, 0)

8

1

0

5 (1, 4)

32041

2

0

0

0 (0, 0)

28

1

1

0 (0, 0)

32285

2

0

0

1 (0, 1)

33

1

0

10 (0, 10)

32340

2

0

0

0 (0, 0)

49

4

0

9 (0, 9)

32462

2

0

0

0 (0, 0)

5

1

0

3 (0, 3)

3860

2

0

0

0 (0, 0)

185

1

0

0 (0, 0)

28858

1

0

0

0 (0, 0)

9

1

0

4 (1, 3)

29806

1

0

0

0 (0, 0)

11

2

0

6 (1, 5)

30664

1

0

0

1 (0, 1)

12

1

0

7 (1, 6)

30666

1

0

0

1 (0, 1)

11

2

0

5 (0, 5)

30694

1

1

0

1 (1, 0)

1

1

0

0 (0, 0)

31560

1

1

0

1 (1, 0)

2

0

0

0 (0, 0)

31586

1

0

0

0 (0, 0)

7

1

0

5 (1, 4)

31596

1

0

0

0 (0, 0)

10

1

0

2 (0, 2)

32504

1

0

0

0 (0, 0)

1

1

0

1 (1, 0)

7097

1

1

0

1 (1, 0)

4

0

0

3 (0, 3)

10003848

0

0

0

0 (0, 0)

3

1

0

1 (1, 0)

19799

0

0

0

0 (0, 0)

1

1

0

1 (1, 0)

21750

0

0

0

0 (0, 0)

17

1

0

9 (0, 9)

28938

0

0

0

0 (0, 0)

2

1

0

0 (0, 0)

29800

0

0

0

0 (0, 0)

2

1

0

0 (0, 0)

30326

0

0

0

0 (0, 0)

41

1

0

2 (0, 2)

30584

0

0

0

0 (0, 0)

3

1

0

3 (1, 2)

30652

0

0

0

0 (0, 0)

10

2

0

4 (2, 2)

30662

0

0

0

0 (0, 0)

5

1

0

2 (1, 1)

30672

0

0

0

0 (0, 0)

8

1

0

5 (1, 4)

Totals

9,515

389

43

1 218 (108, 1,110)

47,345

1,112

46

13, 269 (272, 12,997)

Abbreviation: ID, identification.


Table 5

For clinicians who never used the statin macro and prescribed statin therapy at least once, clinician statin macro usage and prescription stratified by patient study eligibility

Clinician ID

Study eligible patients

Study ineligible

Total no. of patients

No. of patients statin prescribed

Total no. of patients

No. of patients statin prescribed

14823

197

14

550

207

19002

152

12

356

136

15255

117

9

382

127

26250

93

5

451

168

9561

91

4

301

101

30321

70

8

543

102

9102

70

9

492

118

31052

69

12

315

86

14892

67

3

207

63

25268

61

7

209

55

19052

59

3

289

53

7389

56

12

575

151

30319

54

13

658

190

3502

50

3

123

32

6020

46

6

144

77

25083

43

4

313

40

31022

38

5

416

40

32078

38

7

116

53

23205

32

1

59

39

13661

30

3

157

42

29620

28

2

47

26

31184

26

4

220

39

24257

25

3

146

79

31148

25

2

98

34

26023

24

2

198

33

28660

24

3

193

49

31011

23

2

341

23

4294

23

3

63

34

28747

22

3

122

28

30291

22

3

122

23

14704

19

2

36

13

26731

17

2

48

31

30318

17

2

146

30

21251

16

1

357

24

24646

16

2

32

16

28050

16

1

104

22

31754

16

5

82

31

30292

14

4

298

41

32385

14

5

58

19

28048

13

2

135

32

13583

12

1

61

32

18591

12

0

27

12

27779

11

1

47

14

25376

10

4

82

8

26830

10

1

12

7

30382

10

1

37

15

32005

10

0

125

7

14767

9

2

31

18

17325

9

1

53

32

29279

9

0

19

12

30288

9

2

229

14

6494

9

0

29

14

27456

8

0

13

12

8251

8

0

10

5

10381

7

0

28

13

16407

7

2

19

11

21366

7

1

17

8

24204

7

1

23

13

30379

7

0

39

23

31794

7

0

41

4

32278

7

0

25

9

16391

6

0

5

4

17319

6

2

95

30

17792

6

1

24

7

18273

6

0

14

9

2619

5

0

15

3

28089

5

2

30

10

28611

5

1

4

3

30317

5

0

32

11

30677

5

2

5

1

32357

5

0

13

6

32961

5

0

11

6

14275

4

0

26

3

16953

4

1

2

0

24969

4

0

5

2

26389

4

0

37

16

29741

4

0

20

5

29816

4

1

25

14

30651

4

1

10

4

31591

4

0

10

2

13656

3

2

40

24

14756

3

1

4

3

18769

3

0

19

14

27330

3

0

6

3

27820

3

0

37

3

29649

3

0

8

4

29809

3

0

5

2

30210

3

1

49

1

30644

3

0

10

3

30650

3

1

5

1

31398

3

1

8

2

31566

3

0

4

3

31577

3

0

6

6

10887

2

1

7

5

13093

2

1

4

2

14765

2

0

1

1

15448

2

1

8

4

17383

2

0

7

1

17651

2

0

5

1

20928

2

1

9

5

21472

2

0

4

1

21971

2

0

16

1

26465

2

1

4

0

28148

2

0

8

5

28617

2

0

7

2

29794

2

1

13

5

29818

2

0

8

4

30626

2

0

7

3

30636

2

0

7

1

30641

2

0

9

3

30665

2

0

5

1

31573

2

0

6

2

32474

2

1

4

3

32966

2

0

29

6

33145

2

0

16

7

7020

2

0

61

10

7773

2

0

4

2

9183

2

0

15

1

10001334

1

1

0

0

10115

1

1

0

0

11416

1

0

16

1

15404

1

0

3

1

18103

1

0

7

4

20170

1

0

1

1

20373

1

0

3

1

21033

1

0

3

3

21134

1

0

8

1

21746

1

0

10

1

22769

1

0

10

4

22802

1

0

2

1

23153

1

0

6

3

23424

1

1

2

1

23436

1

0

16

2

25144

1

0

3

2

25963

1

0

10

2

28045

1

0

3

1

28677

1

0

11

5

28914

1

0

4

3

29738

1

0

2

1

29742

1

0

2

2

29743

1

0

2

2

29786

1

0

7

1

29788

1

0

2

2

29795

1

0

1

1

29829

1

0

3

3

30333

1

1

119

11

30581

1

0

2

1

30587

1

0

2

1

30646

1

0

10

3

30653

1

1

9

3

30659

1

0

6

6

30660

1

0

6

3

30661

1

0

13

3

30674

1

0

14

8

30678

1

0

10

7

30679

1

0

8

4

31173

1

0

2

1

31556

1

0

3

2

31559

1

1

6

3

31568

1

1

3

1

31570

1

1

4

3

31571

1

0

6

2

31578

1

0

4

1

32004

1

0

24

5

32290

1

0

10

2

32467

1

0

2

1

32472

1

0

4

2

32482

1

1

2

2

32487

1

0

8

4

32497

1

0

2

2

32674

1

0

10

3

33009

1

0

3

1

4392

1

0

3

3

5606

1

0

2

1

8995

1

0

122

7

90000014

1

0

5

3

10000141

0

0

1

1

10000390

0

0

1

1

10000739

0

0

1

1

10001141

0

0

1

1

10004278

0

0

1

1

10004338

0

0

1

1

10004367

0

0

2

1

10004906

0

0

1

1

10007567

0

0

1

1

10009733

0

0

1

1

10011977

0

0

1

1

10022155

0

0

1

1

10022969

0

0

1

1

10030040

0

0

2

1

10033676

0

0

1

1

10034012

0

0

1

1

10041394

0

0

1

1

10044676

0

0

1

1

10047841

0

0

1

1

11000316

0

0

1

1

11000496

0

0

1

1

11000872

0

0

1

1

11001242

0

0

1

1

11006889

0

0

3

1

11011013

0

0

1

1

11012330

0

0

1

1

11013507

0

0

1

1

11016417

0

0

1

1

11017524

0

0

1

1

11029330

0

0

1

1

11035221

0

0

2

1

11044881

0

0

2

1

11045323

0

0

2

2

11046508

0

0

1

1

11050138

0

0

1

1

11051979

0

0

1

1

12070

0

0

1

1

12446

0

0

4

1

12547

0

0

1

1

12908

0

0

3

1

14744

0

0

1

1

14797

0

0

3

2

17838

0

0

1

1

17850

0

0

3

2

18107

0

0

1

1

18109

0

0

5

2

18125

0

0

6

3

18771

0

0

1

1

18772

0

0

1

1

19131

0

0

1

1

20094

0

0

3

1

20279

0

0

5

1

20402

0

0

3

1

20445

0

0

1

1

20550

0

0

5

3

20637

0

0

2

1

21429

0

0

1

1

21731

0

0

3

3

21735

0

0

1

1

22913

0

0

1

1

22947

0

0

2

2

23497

0

0

2

1

23586

0

0

1

1

23600

0

0

3

2

23719

0

0

5

1

23800

0

0

13

2

23888

0

0

7

3

24287

0

0

1

1

24405

0

0

2

2

24735

0

0

2

2

24765

0

0

3

2

24947

0

0

1

1

25157

0

0

1

1

25748

0

0

2

1

25882

0

0

2

1

26022

0

0

1

1

26325

0

0

2

2

26973

0

0

1

1

27257

0

0

10

2

27290

0

0

3

1

27937

0

0

4

3

28110

0

0

1

1

28146

0

0

2

1

28600

0

0

2

2

28722

0

0

1

1

28936

0

0

2

1

29251

0

0

2

2

29418

0

0

2

1

29420

0

0

14

4

29494

0

0

2

1

29502

0

0

6

1

29526

0

0

1

1

29733

0

0

1

1

29735

0

0

1

1

29737

0

0

5

2

29739

0

0

2

1

29798

0

0

1

1

29801

0

0

2

1

29802

0

0

3

1

29803

0

0

1

1

29804

0

0

5

3

29817

0

0

3

1

29830

0

0

6

2

29831

0

0

5

1

30586

0

0

2

1

30589

0

0

4

2

30591

0

0

2

2

30592

0

0

2

2

30670

0

0

6

1

30671

0

0

10

3

30673

0

0

6

4

30675

0

0

7

4

30693

0

0

2

1

30696

0

0

2

2

31010

0

0

5

1

31093

0

0

1

1

31275

0

0

1

1

31541

0

0

1

1

31544

0

0

1

1

31557

0

0

8

3

31558

0

0

6

3

31562

0

0

6

2

31567

0

0

6

3

31575

0

0

14

6

31576

0

0

4

1

31587

0

0

3

1

31597

0

0

6

1

31599

0

0

1

1

32450

0

0

1

1

32464

0

0

1

1

32466

0

0

1

1

32470

0

0

1

1

32485

0

0

5

3

32490

0

0

3

2

32491

0

0

5

4

32493

0

0

4

4

32494

0

0

3

2

32498

0

0

1

1

32500

0

0

1

1

33011

0

0

4

2

3722

0

0

1

1

4050

0

0

1

1

505804

0

0

1

1

5279

0

0

1

1

5512

0

0

4

2

5777

0

0

1

1

6319

0

0

5

1

6845

0

0

3

3

7231

0

0

1

1

7421

0

0

1

1

9782

0

0

1

1

Abbreviation: ID, identification.


For study eligible patients, the statin macro was used in 3.2% (389 of 11,877) of patients and not used in 11,488 patients. Clinicians prescribed statin therapy during the study period in 28% (108 of 389) of patients for whom the statin macro was used compared with 13% (1,360 of 11,488) of patients for whom the statin macro was not used ([Fig. 3]). The relative risk of statin therapy prescription for macro usage compared with no macro usage was 2.3 (95% confidence interval [CI], 1.9–2.8, p < 0.001). Statin therapy prescription was significantly more likely in patients for whom the statin macro was used (odds ratio 2.86, 95% CI, 2.24–3.65, p < 0.001) while controlling for gender, age, race, smoking status, ASCVD, DM, systolic blood pressure, antihypertensive medication, total cholesterol, LDL, HDL, and clinician ([Table 6]). Clinician was modeled as a random effect and had a significant variance among clinicians (odds ratio 1.36, p < 0.001).

Table 6

Multivariate adjusted mixed effect logistic regression model for statin therapy prescription during study period

Odds ratio (95% CI)[a]

p-Value[b]

Statin macro usage

2.86 (2.24–3.65)

< 0.001

Male

1.19 (1.04–1.36)

0.011

Age

1.03 (1.02–1.04)

< 0.001

Black

0.88 (0.70–1.09)

0.22

Smoke

1.25 (1.00–1.55)

0.054

DM

2.61 (2.26–3.02)

< 0.001

Antihypertensive

1.32 (1.17–1.49)

< 0.001

Systolic BP

1.00 (1.00–1.00)

0.62

Total cholesterol

1.01 (1.00–1.01)

< 0.001

LDL

1.00 (1.00–1.00)

0.25

HDL

0.98 (0.98–0.99)

< 0.001

Abbreviations: BP, blood pressure; CI, confidence interval; DM, diabetes mellitus; HDL, high-density lipoprotein; LDL, low-density lipoprotein.


Note: Bold indicates variables with significant p-values.


a Odds ratio of variables from mixed effect logistic regression.


b p-Value based on likelihood ratio test while controlling for other variables in the model. Clinician was included as a random effect.


We investigated only the clinicians who had evidence of statin macro usage. From [Table 4], clinicians that used the statin macro at least once in study eligible or ineligible patients saw a total of 9,515 study eligible patients. These clinicians prescribed statin therapy during the study period in 28% (108 of 389) of patients for whom the statin macro was used compared with 12% (1,110 of 9,126) of patients for whom the statin macro was not used. The relative risk of statin therapy prescription for macro usage compared with no macro usage was 2.3 (95% CI, 1.9–2.7, p < 0.001). Statin therapy prescription was significantly more likely in patients for whom the statin macro was used (odds ratio 2.77, 95% CI, 2.16–3.54, p < 0.001) while controlling for covariates and clinician in a mixed effect model. Also from [Table 4], for the study ineligible patients, those same physicians used the statin macro 1,112 times to generate 272 statin prescriptions (24%).

Missing data imputation analysis yielded similar results with our primary analysis. Although more patients met the inclusion/exclusion criteria (11,877 primary analysis vs. 20,240 imputed analysis), the relative risk of statin therapy prescription for macro usage and statin macro odds ratio in the mixed effect logistic regression were similar (see [Supplementary Material], available in the online version).


#

Discussion

This study described the implementation of a CDS tool, the statin macro, for the 2013 ACC/AHA statin guideline. For patients recommended but not prescribed statin therapy before the study period, statin macro usage was significantly associated with increased statin prescription during the study period, although the study did not determine if the statin dosages were guideline compliant. This study is the first to show that a 2013 ACC/AHA statin guideline CDS tool was associated with improved statin guideline-related prescription rates. The study cannot establish a cause and effect relationship, in that clinicians might have used the statin macro more frequently after already having decided to prescribe a statin.

Baseline characteristics including DM and LDL were significantly different in patients for whom the statin macro was and was not used. The large sample size lead to statistical significance, but the clinical significance of a 4 mg/dL difference in LDL levels (123 vs. 127 mg/dL) is somewhat trivial. Clinicians may have used the statin macro less often for DM patients as they knew such patients should receive statin therapy.

Most clinicians were internal medicine and family medicine physicians. Other specialties were represented because the primary care clinician listed in the EHR can be from any specialty. As our study was performed at a tertiary referral center, neurologists and surgeons may be the EHR-listed primary care clinician.

The statin macro is a clinician-initiated CDS tool. Our results may be confounded as clinicians who used the statin macro may be more technologically proficient or more compliant to statin guidelines. However, almost all the 125 clinicians who used the statin macro for patients also did not use the statin macro for other patients. Therefore, our results are not due to a small number of clinicians who were the only users of the statin macro.

The statin macro was used for a low percentage (3.2%) of patients recommended but not prescribed statin therapy. Other studies of clinician-initiated CDS tools showed very low uptake, which was 0% during some months or at some clinical sites.[39] [40] Aside from low advertisement, the statin macro may have had low uptake as not all primary care visits were focused on prevention. Some visits may have focused on acute issues. CDS features associated with low uptake included clinician perception of loss of autonomy, lack of EHR integration, poor transparency of CDS developers, lack of clinical leadership endorsement, lack of financial incentive, and changing guidelines.[41] [42] Ongoing systematic reviews will further delineate these features.[43] Future strategies to improve statin macro uptake include advertisements describing developers, emphasizing clinical leadership endorsement, and financial incentives.

There are many reasons that clinicians may not prescribe statin therapy according to the 2013 ACC/AHA guideline. In 2015, less than half of surveyed clinicians read the guideline, knew the patient groups recommended statin therapy, or knew the definition of statin intensity.[44] The 2013 ACC/AHA statin guideline was met with controversy as there were no LDL target goals and the number of patients recommended statin therapy would significantly increase compared with previous guidelines.[45] [46] There was also concern that the risk equation overestimated the 10-year ASCVD risk.[47] The 2013 ACC/AHA guideline considered only LDL-C rather than other lipoprotein measures such as LDL particle number or lipoprotein (a), which were associated with cardiovascular risk.[48] [49] Of clinicians who were knowledgeable about the 2013 ACC/AHA statin guidelines, many disagreed with statin intensity definitions and the groups' recommended statin therapy.[50]

Given the benefits of the statin macro, one could consider developing a system-initiated statin CDS tool. Previous studies showed system-initiated CDS tools improved clinical outcomes compared with clinician-initiated CDS tools.[18] However, system-initiated CDS tools, such as best practice alerts, may lead to alert fatigue.[51] [52] A future randomized control trial comparing clinician-initiated versus system-initiated statin CDS tools could compare this CDS feature and provide a greater certainty of the statin CDS tool effect size.

There were limitations to this study. We calculated the 2013 ACC/AHA statin guideline recommendation for all patients before the study period as of December 31, 2015. We could not determine the statin recommendation for each patient at the time of statin macro usage or if the statin macro was used in a template versus on demand due to limitations of EHR data archiving. We did not determine if each patient's statin prescription met the ACC/AHA 2013 therapy intensity recommendations (high vs. moderate vs. low). We did not determine if clinicians using the statin macro tool initiated its use because they already decided to prescribe a statin; if so, interpretation of study results may be affected. We plan to collect time-specific patient data for future studies with macros and could implement CDS monitoring.[53] Since initiation of the study, a new 2018 ACC/AHA statin guideline was recently published.[54] Future versions of the statin macro should include updated 2018 ACC/AHA statin guidelines.


#

Conclusion

Statin macro usage was associated with improved 2013 ACC/AHA statin guideline compliance at the level of statin prescription versus no statin prescription. We did not determine if each patient's statin prescription met the ACC/AHA 2013 therapy intensity recommendations (high vs. moderate vs. low). Macro CDS tools may improve compliance to other societal guidelines.


#

Clinical Relevance Statement

Although many best practice guidelines exist for initiating medication in select patient groups, these medications are prescribed at suboptimal rates. Clinical decision support tools may improve guideline compliance. Use of a statin macro was associated with improved statin guideline compliance.


#

Multiple Choice Questions

  1. The following baseline characteristics were significantly different in patients for whom the statin macro was and was not used:

    • High-density lipoprotein and race.

    • Systolic blood pressure and total cholesterol.

    • Race and systolic blood pressure.

    • Gender and age.

    • Diabetes mellitus and low-density lipoprotein.

    Correct Answer: The correct answer is option e. Baseline characteristics including diabetes mellitus and LDL were significantly different in patients for whom the statin macro was and was not used. However, the large sample size lead to statistical significance with somewhat trivial clinically significant differences.

  2. The following variables were significantly associated with statin prescription in patients who were recommended a statin based on guidelines but not previously prescribed a statin:

    • Statin macro usage, diabetes mellitus.

    • Race, diabetes mellitus.

    • Statin macro usage, race.

    • Statin macro usage, systolic blood pressure.

    Correct Answer: The correct answer is option a. Statin macro usage and diabetes mellitus were significantly associated with statin prescription while controlling for other covariates.


#
#

Conflict of Interest

None declared.

Acknowledgments

We thank the UCLA Health System, its clinicians, and the patients.

Protection of Human and Animal Subjects

The University of California, Los Angeles Institutional Review Board approved a waiver of authorization for this study (IRB#: 16–001676).


Supplementary Material


Address for correspondence

Timothy S. Chang, MD, PhD
Department of Neurology, University of California
Los Angeles, 695 Charles E Young Dr South, Room 2309, Los Angeles, CA 90095
United States   


Zoom Image
Fig. 1 Central schematic including simplified version of the 2013 American College of Cardiology/American Heart Association (ACC/AHA) statin guideline and study design. For the statin guidelines, patients with a history of ASCVD, LDL ≥ 190 mg/dL, or DM aged 40 to 75 years old were recommended statin therapy. Patients without DM aged 40 to 75 years old and an LDL of 70 to 189 mg/dL necessitated 10-year ASCVD risk calculation. Patients with a 10-year ASCVD risk calculation ≥ 5% were recommended statin therapy while those with a 10-year ASCVD risk calculation < 5% were not recommended statin therapy. For the study design, patients recommended statin therapy for primary prevention (LDL ≥ 190 mg/dL; DM and 40–75 years old; no DM, LDL 70–189 mg/dL and 10-year ASCVD risk ≥ 5%) but not prescribed statin therapy as of December 31, 2015, were included in this study. We tested the association of macro usage and statin therapy prescription during the study period (January 1, 2016–June 30, 2016). Abbreviations: ASCVD, atherosclerotic cardiovascular disease; ASCVD risk, 10-year ASCVD risk calculation; DM, diabetes mellitus; LDL, low-density lipoprotein; Rx, prescription; YO, years old.
Zoom Image
Fig. 2 Screenshot of the statin macro within a note. Blue highlighted text was variable and specific to each patient. The blue “2013 American College of Cardiology/American Heart Association (ACC/AHA) guideline” text included a hyperlink to the 2013 ACC/AHA statin guideline. The blue “here” text included a hyperlink to the online ASCVD Risk Estimator calculator (Screenshot used with permission from © 2019 Epic Systems Corporation).
Zoom Image
Fig. 3 Flow diagram showing study exclusion criteria, inclusion criteria, and results. 72,315 patients aged 40 to 75 years had a primary care visit during the study period. A total of 60,438 patients were excluded. Counts for categories of excluded patients are shown. 11,877 patients were recommended statin therapy based on the 2013 American College of Cardiology/American Heart Association (ACC/AHA) statin guideline but not prescribed statin therapy prior to the study period. Counts for categories of included patients are shown. Statin therapy was prescribed during the study period in 28% (108 of 389) of patients for whom the statin macro was used compared with 13% (1,360 of 11,488) of patients for whom the statin macro was not used. Abbreviations: ASCVD, atherosclerotic cardiovascular disease; DM, diabetes mellitus; LDL, low-density lipoprotein.