Primary care workload: linking problem density to medical error

1
Primary care workloadlinking problem density to
medical error
WREN

• Jon Temte, MD/PhD, Mike Grasmick, PhD,
• Peggy OHalloran, Lisa Kietzer, Bentzi Karsch,
  PhD, Beth Potter, MD, John Beasley, MD, Paul
  Smith, MD, and Betsy Doherty, MS-2
• AHRQ Grant 1 R03 HS016026-01

2
Study in a Nutshell

• This AHRQ-funded WREN study
• examine 600 clinical encounters
• conducted by 30 clinicians
• to assess interactions of problem number, MWL and
  error
• Data collection completed with 31 clinician and
  615 visits
• Relationships between clinician MWL and patient
  age and sex, continuity status, number of
  problems per encounter (NPPE) and perceived
  medical error (PME) were assessed using ANOVA and
  correlation analyses.
• Analysis of covariance used to assess potential
  differences among the 31 clinicians.

3
Basic Study Demographics

• Four Primary Care Clinics affiliated with WREN
• 2 urban and 2 rural
• Multiple clinicians (Goal 30)
• Mix of FPs, IMs, MDs, PAs, and NPs
• Quasi-randomly selected patients
• 6 random time periods per day
• Age gt 18, mentally competent
• Current Patient Demographics
• Mean age 54.6 /- 17.5 years
• 63.5 female

4
Social support
Experience
Long-term outcomes
Immediate outcomes
Mediators
Individualfactors
Affect
Poor Communication
Memorycapacity
Provider- Disease- Burnout- Lowquality
Provider- Stress- Errors- Delays
Number ofproblems
Fatigue
MWL
Mentaldemands
Demands
Complexity
Emotionaldemands
x
More slips
Patient-Stress- Poorhealth- Reducedtrust
Patient- Stress- Harm- Dissatis-faction
Temporaldemands
Difficulty ofproblems
Baddecisions
Workschedule
Worksystemfactors
Socialenvironment
Supporttechnology
Control factors
Affect
Perceived Locusof control
Copingstrategies
Supporttechnology
Rest breaks
Decisionauthority
Notes1.The above components are merely examples.
Clearly, others may be added and this is all
amenable to modification.2.This model, despite
its many components, is probably a simplification
of the true nature of mental workload. However,
this model (or something like it) can serve as a
conceptual base camp from which studies are
launched. The boxes with shaded backgrounds
represent variables that can potentially be
measuredalbeit not all in the initial study.
However, I would make the case that many of them
can be measured with minimal intrusion and time
demand on the docs. Some, like experience,
memory capacity, social support, coping
strategies, etc. can be measured only once or can
be obtained without any effort from the doc
(RICHARD JOHN HOLDEN, 2005 rholden_at_students.wisc.
edu).
5
Patient arrives at clinic
DE1 Demographic data (age, sex) Patients
anticipated number of concerns
DE2 Clinicians reported number of problems
(NPPE)
Patient placed in exam room by medical
assistant Informed consent
DE3 Clinicians mental workload (NASA TLX)
Clinician evaluates and manages patient and
problems
DE4 Clinicians estimate of likelihood of
error
DE5 Time spent in direct patient contact
Medical assistant exits patient
Clinician dictates and photocopies clinical
note
DE6 Audit of note for quality measures
DE7 Patients satisfaction, assessment of
level to which concerns were addressed during
visit and estimate of error
6
Results

• Measures of Problem Density
• Number of problems per encounter
• Measures of Mental Workload
• Mean
• Variation
• Range
• Estimates of Completeness and Error

7
Encounter Problem Density

• Number of Problems per Encounter
• Mean 3.30 /- 1.96 (sd)
• Range 1 12
• Significant differences among clinicians
• ANOVA Plt0.001
• Number of Problems per Scheduled Time
• Mean 10.39 /- 6.89 (sd) problems per hour
• Range 2.0 42.0
• Significant differences among clinicians
• ANOVA Plt0.001

8
Managing Multiple and Potentially Competing
Problems(current study n 609 visits)
Mean 3.30 Std. Dev. 1.96
9
Effect of Patient Age onNumber of Problem per
Encounter
r 0.237 P lt 0.001
10
Effect of Patient Sex andContinuity Status on
NPPE
11
Mental Workload in Primary Care(n 598 mean
47.6 18.4)
12
Relative Contributions to EffortNO TIME TO
THINK!
13
Distribution of Subscores20 highest visits
14
Distribution of Subscores20 lowest visits
15
Mental Workload in Primary Care

• Composite NASA-TLX
• n 598
• Range 5.00 to 95.3
• Mean 47.6
• Std dev 18.4
• Individual Variation
• N 31 clinicians
• ANOVA Plt0.001

Clinician Average
16
Effect of Patient Age on Workload
r 0.152 P lt 0.001
17
Effect of Patient Sex, Continuity Status, and
Presenting Problem on Workload
18
Workload Rises over the Week(ANOVA P0.002)
19
MWL is Related to Complexity (TLX 36.3
3.45NPPE r2 0.134)
20
Workload Increases with Additional Medical
Problems
21
Emergent Themes for Outlier Analysis of Clinical
Visits with Lower and Higher than Expected Work
Load
22
Distribution of Perceived Medical Error

• Mean 6.9 /- 2.2 (sd) ? relatively low
• Range 3 16 ? moderate variation
• Significant differences among clinicians
• ANOVA Plt0.001

23
Medical Error is related MWL (PME 5.64
0.026TLX r2 0.044)
24
Conclusion

• Primary care encounters are complex
• Mean of 3.3 problems per visit
• Visits are associated with moderately high
  workloads with a tremendous range
• Workload is associated with
• Complexity and type of visit
• Patient, clinician and workplace factors
• Relationships
• Errors is associated with level of workload
• Some components are not modifiable
• Time factors and frustration can be modified

25
(No Transcript)