supply-demand balance at the bedside: An exploration of data sources - - PowerPoint PPT Presentation

▶
supply demand balance at the bedside
SMART_READER_LITE
LIVE PREVIEW

supply-demand balance at the bedside: An exploration of data sources - - PowerPoint PPT Presentation

Oct 27, 2023 29 likes •391 views

1 Toward dynamic assessment of supply-demand balance at the bedside: An exploration of data sources and methods Dana Womack, PhD, RN Department of Medical Informatics Oregon Health & Science University 2 When Im really stretched

slide-1
SLIDE 1

1

Toward dynamic assessment of supply-demand balance at the bedside: An exploration of data sources and methods

Dana Womack, PhD, RN Department of Medical Informatics Oregon Health & Science University

slide-2
SLIDE 2

2

“When I’m really stretched I’ll stop offering help to others. I’ll stop trying to build rapport with my patients. I’ll prioritize the essential things… I’ll defer assessments on the patient who is not as sick, or defer clean-ups. If someone is having a blood pressure crisis, the

  • ther patient might have to wait. And it feels

bad, but that’s just what has to happen.”

—Intensive care RN

slide-3
SLIDE 3

3

3

Hospital clinician overload

Clinician overload can lead to

  • elevated patient risk
  • clinician burnout
  • staff turnover

Hospitals cannot staff to peak demand, as labor costs already represent ~ 60% of the operating budget Need: Proactive & targeted response to overload “hot spots” as they emerge

3

slide-4
SLIDE 4

4

4

Workplaces are increasingly quantified

Can we harness ambient data for good?

Image: www.omnicell.com Image: www.vocera.com Image: www.wnyc.org/story/213967-pager Image: www.kronos.com

slide-5
SLIDE 5

5

5

Common pattern of adaptive system failure: Decompensation

Adapted from Woods D, Branlat M. Basic patterns in how adaptive systems fail. In: Resilience Engineering in Practice: A Guidebook; 2011. p. 127-44.

Degraded system performance

Target system performance Compensation

Extra effort exerted to sustain target performance

Decompensation

Performance declines as adaptive capacity is exhausted

Escalating Control Escalating Disturbance

slide-6
SLIDE 6

6

6

Image: https://loveandworktour.files.wordpress.com/2015/08/walking-against-the-wind.jpg

Is it windy today?

slide-7
SLIDE 7

7

7

Image: https://loveandworktour.files.wordpress.com/2015/08/walking-against-the-wind.jpg

Adaptation

Yes!

slide-8
SLIDE 8

8

8

Study questions

  • 1. How do RNs recognize strain today?
  • 2. Do work environments contain “digital echoes”
  • f strain?
  • 3. It is feasible to provide early warning of work

system strain?

slide-9
SLIDE 9

9

9

Methods

Nurse call Phone

Medications Time & attendance

CRISP-DM Data Mining Process Business understanding Data understanding Modeling Data preparation Evaluation

Focus groups Frontline RNs Data Gathering: Interview guide Analysis

  • Development of coding template by

three qualitative researchers

  • Content coding in Quirkos software
  • Environmental scan
  • Data acquisition &

association

  • Activity feature

‒ Definition ‒ Generation ‒ Selection

  • Classification of shifts

by unplanned overtime

(proxy measure of strain)

slide-10
SLIDE 10

10

10

Setting: Two patient care units in an academic medical center

Scope: Day shift

Unit Beds RNs/shift

(mean)

Ratio

(day shift)

Distinct RNs, including floats Distinct patients Medical ICU 16 11 1:1-2 156 1,464 Medical- surgical 20 6 1:4 105 1,485 Per shift Per year

slide-11
SLIDE 11

11

11

  • 1. Business

Understanding

slide-12
SLIDE 12

12

12

What environmental changes occur during times of strain?

Empty nurses’ station

‒ RNs ping-ponging between rooms ‒ Frequent task switching ‒ Delayed response to call lights ‒ No one sitting down to chart ‒ Increase in phone call volume ‒ Many PRN medication requests ‒ Multiple patients with delirium,

  • r 2+ person assist
slide-13
SLIDE 13

13

13

How do RNs adapt to increasing strain?

Degraded system performance

Increasing strain

slide-14
SLIDE 14

14

14

  • 2. Data Understanding
slide-15
SLIDE 15

15

15

Ambient activity data is not yet integrated

Patient

Lab Pharmacy Imaging Documentation Surgery Others Electronic health record Integrated

Team

Paging

Medication dispensing Phone

Time & attendance Nurse call Others Non-integrated 2 patient care units 366 shifts >400,000 records Ambient data

slide-16
SLIDE 16

16

16

  • 3. Data Preparation
slide-17
SLIDE 17

17

17

Raw log files shift & staff-centric data

Exemplary data from a single medical-surgical work shift

Medication log files Phone transactions Shift ID Consistent staff IDs Disparate staff IDs

slide-18
SLIDE 18

18

Time Meds Phone Nurse call

  • Unit
  • Employee ID

RN name

  • Dates worked
  • Total hours worked
  • Start & end (float RNs
  • nly)
  • Pay codes
  • Device unit
  • Patient name
  • Date-time of dispenses
  • Dispensing RN name
  • Drug name

Duplicate records Lack of start/end time for regular staff RN name only – outdated, inconsistent format Dispensing role not indicated

  • RN name
  • Date-time of call
  • Call type

Manual ID mapping

  • Room number
  • Date-time of call
  • Call type

No patient-room link No RN-room link

Data required significant preprocessing

slide-19
SLIDE 19

19

19

Data exploration: Observable shift rhythms

Medication count Hour of day

Medical-surgical unit 366 day shifts

slide-20
SLIDE 20

20

20

Adaptive strategy: Help passing meds within a single shift

RN 1 RN 2 RN 3 RN 4 RN 5 RN 6

Received med help

Gave med help

Nurse

Medication count

Meds given to RN’s

  • wn

pts

Managers do not have access to this info today due to lack of data integration Key:

slide-21
SLIDE 21

21

21

  • 4. Modeling
slide-22
SLIDE 22

22

22

Unplanned overtime >30 min. to <3 hours Overtime ≤ 30 min. excluded from definition

Unplanned overtime: Binary proxy measure

  • f work system strain

Count, RN instances Overtime duration in hours, MICU

Frequency of unplanned overtime by unit, FY 2016 Unit % Hours % Shifts MICU 1.16% 52% Med-surg 0.17% 8%

Overtime >3 hours: likely an extra shift

slide-23
SLIDE 23

23

23

Visualization of differences across outcomes

Cross-assignment meds (help) Aggregate minutes spent on phone

No overtime Unplanned overtime

Source: MICU Shifts: 366 day shifts

No overtime Unplanned overtime

slide-24
SLIDE 24

24

24

Image: https://loveandworktour.files.wordpress.com/2015/08/walking-against-the-wind.jpg

MICU selected features

Activity & workplace characteristics % IV push medications % patients assigned to same RN as yesterday Differences in activity across RNs, patients % Experienced, non-float RNs Adaptive Strategies Aggregate minutes on phone # cross-assignment med admin (helping)

slide-25
SLIDE 25

25

25

  • 5. Evaluation
slide-26
SLIDE 26

26

26

Repeatable process for insight creation

Define features that reflect activity & adaptation Feature definition, generation Feature selection Machine learning classification Train & test a support vector machine (SVM) algorithm Identify features that differ by

  • utcome state

X 4

0-12 Full shift 0-10 0-8 0-6 Hour 0-4

25% Test 92 shifts 75% data for training 274 MICU shifts

Repeat for multiple timeframes, to assess ability to predict outcome during a work shift

4-fold cross validation train & test Predict unplanned

  • vertime

Present Absent

slide-27
SLIDE 27

27

27

MICU Shift hour 1 2 3 4 Accuracy

  • Conf. Int.

Kappa Sensitivity | Specificity

Hours 1-4 7am – 11 am 53.7% 54.2% 51.6% 50% 52 52.4% .4% .41 .41 – .62 .62 .05 .05 .46 .46 | | .59 .59 Hours 1 - 6 7am – 1 pm 61.1% 57.7% 59.7% 58.1 59 59.2% .2% .47 .47 – .68 .68 .16 .16 .57 .57 | | .60 .60 Hours 1 – 8 7am – 3 pm 60% 59.4% 58.9% 52.5% 57 57.7% .7% .46 .46 – .67 .67 .15 .15 .52 .52 |.63 |.63 Hours 1 – 10 7am – 5 pm 64.9% 59.1% 66.5% 54.6% 61 61.3% .3% .50 .50 – .71 .71 .22 .22 .58 .58 | | .65 .65 Hours 1 – 12 7am – 7pm 63.3% 70.3% 59.4% 61.1% 63.5% .5% .55 .55 – .74 .74 .30 .30 .58 .58 – .68 .68 Closed training: 72 72.3% .3% .68 .68 - .77 .77 .45 .45 .64 .64 | .77 .77

Classification of workplace strain – MICU

4-fold train-test Overall results (mean) Shift hours

slide-28
SLIDE 28

28

28

Conclusions

slide-29
SLIDE 29

29

29

Key Findings

Ambient data contains echoes of signs of strain

‒ Environmental & behavioral

Ambient data is underutilized for purposes of care improvement

‒ Produced in real-time, provides granular observability of work

It is feasible to use ambient data to characterize system strain

‒ Unplanned overtime predicted 8 – 10 hours into a shift (MICU)

slide-30
SLIDE 30

30

30

Limitations

As is true for any recall methodology, discussion of past workplace events carries potential for participant recall bias Study conducted on 2 units at a single facility, findings may not generalize to other units or facilities Derived and manually associated ambient data may contain inaccuracies

slide-31
SLIDE 31

31

31

Implications

Dynamic workplace monitoring could augment charge nurse decision making re: mid-shift resource requests Integration of ambient data is needed

  • Create staff & team-centric views of data
  • Facilitate use of ambient data in healthcare

delivery research

Data exploration of temporal patterns of activity can identify improvement opportunities

  • Level-loading work
  • Development of strain-relieving interventions
slide-32
SLIDE 32

32

32

  • 1. Expand the collection of computable signs of strain
  • Additional data sources, feature types & time granularities

2.

Employ higher fidelity outcome data

  • Hourly RN to patient ratios, periodic subjective ratings
  • Additional outcomes e.g. missed care, delayed discharge, others

3.

Transition to field-based, participatory research

  • Engage frontline staff in refining activity features, developing

real-time data displays

  • Develop early warning systems & organizational interventions

Recommendations for future studies

slide-33
SLIDE 33

33

33

Thank you!

National Library of Medicine Pre-doctoral trainee funding Training Grant T15-LM007088. Dissertation advisory committee

  • Paul Gorman, MD
  • Nancy Vuckovic, Ph.D.
  • Michelle Hribar, Ph.D.
  • Deborah Eldredge, Ph.D.

Helpful discussions & suggestions

  • Karen Eden, Ph.D.
  • Brian Womack, Ph.D.
  • Mike Rayo, Ph.D.
  • Emily Patterson, Ph.D.
  • NLM Fellows

Hospital Information providers

  • Jake McFarland, Pharm.D., Pharmacy
  • Angela Baltz, MS, Clinical Engineering
  • Chris Black, Info. Technology Group
  • Sean Farrell, Security
  • Amy Do, Nursing Administration
  • Reese Glasscock, Nrsg Admin.

Hospital Leaders

  • Deborah Eldredge, Ph.D.
  • Mariah Hayes, RN, MN
  • Adrienne McDougal, MSN
  • Geary Gardner, MSN
  • Stephanie Nonas, MD
  • Randy O’Donnell, BS

Collaborative qualitative data analysis

  • Nancy Vuckovic, Ph.D.
  • Linsey Steege, Ph.D.

And many others – thank you!

slide-34
SLIDE 34

34

34

Questions & Discussion

Contact info: womacda@ohsu.edu

slide-35
SLIDE 35

35

35

Backup

slide-36
SLIDE 36

36

36

MICU selected features, by shift-hour

Activity features Shift Hour % IV push medications 4, 6, 8, 10, 12 Sum, aggregate phone minutes 4, 6, 8, 10, 12 Sum, dispensed medications 4, 6, 8 Skewness, medications across RNs 4, 6, 8, 12

  • Std. deviation, nurse calls across rooms

4, 6 Sum, nurse call minutes 8 Sum, cross-assignment meds 10, 12 Sum, within-assignment medications Sum, cross-assignment medications % non-float RNs with 2+ yrs. experience % Pts. assigned to same RN in past week % pts present on unit yesterday % analgesic medications 12