IPSQWHIT: Measuring the Quality Improvements Associated with Decision Support in Pediatrics

Slide Presentation from the AHRQ 2008 Annual Conference

On September 8, 2008, Timothy G. Ferris, M.D., M.P.H., made this presentation at the 2008 Annual Conference. Select to access the PowerPoint® presentation (280 KB; Plugin Software Help).

Slide 1

IPSQWHIT: Measuring the Quality Improvements Associated with Decision Support in Pediatrics

Timothy G. Ferris, MD, MPH
Medical Director, MGPO
Associate Professor of Medicine and Pediatrics
Harvard Medical School

Slide 2

Background

Expectations for the ability of electronic health records (EHRs) to improve quality are based on potential of decision support.

• Slow adoption is a barrier:
  Paper → HER → DS
• Evidence of improved quality summarized in Shekelle/AHRQ Evidence Report:
  • Significant improvements in quality.
  • Difficult to aggregate and/or generalize.
• Office of the National Coordinator for Health Information Technology (ONC) Report on measuring quality benefits of Health Information Technology (Health IT):
  http://www.rwjf.org/files/research/3297.31831.hitreport.pdf

Slide 3

IPSQWHIT

Improving Patient Safety and Quality With Health Information Technology

• Funded by the Agency for Healthcare Research and Quality (AHRQ) Health IT Value Request for Applications (RFA):
  • Focused on improvements in safety, quality, and efficiency through Health IT.
  • Several pediatric grants funded.

Slide 4

IPSQWHIT (continued)

• e-Rx decision support: Weight based dosing.
• Reminders (synchronous and asynchronous).
• Results management.
• Templates (acute and chronic conditions).

All specifications and templates available on AHRQ Web site.

Slide 5

Prioritizing Decision Support

• MD use of DS has limits:
  • Need to prioritize what is asked of them.
• Grounds for prioritization:
  • Safety.
  • Clinical impact:
    • Evidence for improvement:
      • Common problem/small impact.
      • Rare problem/large impact.

Slide 6

Design and Setting

• Group randomized trials conducted within Partners Healthcare in Eastern MA.
• 26 pediatric practices:
  • All participants had already adopted the same EHR.
  • Hospital based (1), health center (6), private (19).
  • Participation depended on multiple factors.
  • Once selected, sites were paired by type and randomized to intervention or control.
  • Analyses adjusted for clustering by MD and practice.

Slide 7

Weight Based Dosing Decision Support (WBDDS)

Slide 8

Medication Errors in Pediatrics

• Medication errors among the most common and most injurious of all errors in health care.¹
• Pediatric prescriptions may be more prone to error.
• Limited data on rates of pediatric dosing errors.
• Unclear if computerized decision support in the context of electronic prescribing reduces weight related dosing errors.

¹ Bates DW, et al, JAMA 1998; Dean B, et al. Qual Saf Health Care 2002; Kaushal R, et al. JAMA 2001;
² Sullivan JE, et al. J Surg Oncol 2004

Slide 9

Objectives

• To examine the prevalence of dosing errors in ambulatory pediatrics.
• To examine the effectiveness of weight based dosing decision support in reducing the frequency and severity of dosing errors.

Slide 10

Description of Intervention

• The WBDDS included two components:
  1. Active component: a medication menu allowing selection of a dose based on the child's weight.
  2. Passive component: display of a computer generated total daily dose in mg/kg based on the child's weight.
• Child's most recent weight imported from the EHR.

Slide 11

Weight based dosing (WBD) calculator

Two screen shots showing the dosing calculator for Amoxicillin.

• Active (part 1): Dose calculated based on patient weight.
• Passive: Total daily dose calculated.
• Active (part 2): Select rounded dose from drop-down menu.

Slide 12

Results: Dosing errors as a proportion of child office visits

Screen shot of a diagram showing nested rectangles and squares.

• All visits: n=32942 (100%).
• All visits where any Rx provided: n=17526 (53%).
• All visits where WBD Rx provided: n=3684 (11%).
• All visits with a WBD Rx error: n=285 (.87%).

• Adverse drug events: n=22 (.06%).
• 7.7% of eligible meds had a dosing error.
• 1% had a dosing error >10% from recommended dose.

Slide 13

Results: Rates of dosing errors for weight based dosing medications

Slide 14

Results

• Physicians prescribed antibiotics more than any other type of medication and antibiotics were the most likely medication to include a dosing error.
• The active decision support (WBDDS) was used for approximately 10% of Rx in intervention group:
  • No dosing errors when active decision support was used.

Slide 15

Results (continued)

• Majority of dosing errors (58%) judged to be correctable with use of decision support.
• 22% of dosing errors considered directly attributable to incorrect use of the electronic prescribing software.
• Interviews revealed a number of barriers: technical difficulties, user interface challenges, and negative physician perceptions.

Slide 16

Limitations

• Prescribing software did not accommodate medications requiring variable dosing or combination medications:
  • Significant source of dosing errors.
• Unable to fully assess physician use.
• No systematic assessment for adverse drug events (ADEs).

Slide 17

Conclusions

• Dosing errors represent a substantial fraction of medication errors in pediatrics:
  • 10% of eligible Rx.
  • National extrapolation: approximately 4,000,000 dosing errors in weight based dosing eligible pediatric prescriptions every year.
• WBDDS reduced dosing errors from 10.0 per 100 scripts to 6.3 per 100 scripts:
  • National extrapolation: reduction of over 150,000 dosing errors per year.
• Real world effectiveness vs. ideal world efficacy.

Slide 18

Conclusions (continued)

• Very few adverse drug events associated with these dosing errors.
• No incorrectly dosed prescriptions when the active form of WBDDS was used.
• Difficulties using software were a major barrier to regular use of the active DS.

Slide 19

Implications

• Weight based dosing decision support led to reductions in the overall dosing error rate and for overdoses in particular.
• New errors caused by electronic prescribing software.
• Full benefit of e-prescribing will require WBDDS designed to accommodate physician workflow.

Slide 20

Alerts & Reminders

Slide 21

Reminders

ADHD: % of patients receiving follow-up care every 6 months.

• Rates: 53.9% (Cont) vs. 70.1% (Int) (p=.04)
  • 33.5% vs. 43.7% at ADHD visit (p=.27)
  • 22.3% vs. 28.2% at Well child check (p=.33)
• Intervention patients were 2.1 times as likely to have had appropriate follow-up.

Slide 22

Reminders (continued)

Chlamydia: annual screening test for patients who are sexually active.

• Rates: 24% (Control) vs. 48% (Intervention).
• 61% of screening tests were ordered by the patient's primary care physician (PCP).

Slide 23

Reminders (continued)

Obesity

• Lipid profile every 2 years for patients with Body Mass Index (BMI) ≥99th percentile:
  • 23 of 200 patients (11.5%) received a lipid profile.
  • No significant difference between control and intervention (13 intervention vs. 10 control).
• Follow-up visit every six months patients with BMI ≥95th percentile:
  • 75% of intervention group patients had visit where nutritional habits were reviewed vs. 71% in the control (p=.5).

Slide 24

Results Management

Slide 25

Table 1. Goals for quality improvement with an electronic results management system

Slide 26

Results Management

Main findings:

• Full adoption practices reported gains in efficiency, reliability, timeliness, and provider satisfaction.
• Some partial adopters reported decreased efficiency and increased risk of lost test results.
• Barriers to ERM adoption included lack of inclusion of all ordered tests in the ERM system, user-interface design issues, and lack of sufficient pediatric customization.

Ferris et al, Pediatrics (in press)

Slide 27

Templates

Slide 28

Templates

ADHD

• Usage:
  • 32% of attention deficit hyperactivity disorder (ADHD) specific visits at intervention clinics.
• Documentation quality:
  • Documentation of symptoms: 96.6% (T) vs. 29%.
  • Treatment effectiveness: 100% (T) vs. 61.3%.
  • Treatment side effects: 96.6% (T) vs. 54.8%.

Slide 29

ARI Smart Form

Usage:

• Successfully used at 561 ARI visits to treat 522 individual patients with 680 primary and secondary diagnoses.
• The Smart form was employed by 39 providers with a median number of uses/user of 18 (range 1-109).
• Used for only 8% of all eligible visits (!).

Slide 30

ARI Smart Form (continued)

Changes in prescribing:

• In the intervention group, fewer antimicrobial prescriptions were written when the SF was used:
  • 31.7% (SF) vs. 39.9% (p <.0007).
• Providers using the SF were less likely to recommend a macrolide antibiotic:
  • 6.2% of ARI visits vs. 9.5%, p=.022.
• Providers also prescribed fewer antibiotics for viral ARI illnesses when utilizing the SF:
  • 12.3% of viral ARI visits versus 18.1% of viral ARI illnesses; p=.0125).

Slide 31

Lessons learned:

Clinical perspective

• Outpatient pediatric workflow necessitates tools designed specifically for that population and setting.
• Clinicians respond with variable frequency to prompts to perform preventive care measures.
• Reminders promote effective management of chronic conditions at well child visits (well child templates might inhibit documentation).
• Smartform lead to increased guideline adherence for acute illness care.

Slide 32

Lessons learned: (continued)

QI/Health IT perspective

• Administrative/organizational barriers are substantial.
• Effective design requires cooperation from practice administrators, IT personnel, network leadership, and clinicians—also iterative modification as guidelines change.
• Variation in clinical workflow across ambulatory settings necessitates the tools that can be easily modified.
• Given the impact of perceived value on use, provider training and education appear an integral component of implementation.

Slide 33

Acknowledgments

AHRQ

• Iris Mabry
• Jon White

Co-Investigators

• John Co
• James Perrin
• David Bates
• Rainu Kaushal
• Eric Poon

Research Assistant

• Sarah Johnson