Is there a tipping point in acute care where the likelihood of patient harm increases precipitously

1
Is there a tipping point in acute care where the
likelihood of patient harm increases
precipitously?

• Alberta Pedroja, Ph.D.
• ATP Healthcare Services
• November 20, 2007
• SCAHRM

2
How does surge in volume affect patient care?

• Do we make more errors?
• Is there an increase in harm to patients?
• Why are these related?
• Staff fatigue?
• Registry?
• Delays in service?
• Systems capacity?
• What does the data say?

3
What do we know so far?

• Weissman et al Brigham and Womens had occupancy
  rates exceeding 100 found an increase in adverse
  events.
• Cook and Rasmussen Systems behave differently
  depending on how many patients are moving through
  them.
• McManus et al Systems break down at a point of
  overload.
• Litvak Mathematical predictions are possible for
  systems with fixed capacity and variable demand.

4
Adjustments to systems capacity

• Hallway beds
• ED Boarding
• Observation Units
• Use of volunteers
• Process redesign to reduce rework and
  inefficiencies

5
Empiracism

• Definition observable relationships
• Theoretical understanding follows
• Some patterns are observed first and theoretical
  basis follows
• 80/20 Rule
• Everyone is, on average, six people away from
  anyone

6
The Story of Tipping Point

• Major incidents routinely occurred at
  approximately the same point in patient volume
• No increase in temporary workers
• No staff fatigue
• No difference in nurse to patient ratios
• Delays in service

7
Adjustment to the relationship of high volume and
high profitability

• If a major incident occurs routinely at high
  volume, this wipes out profitability for that day
  and many days before and after.

profit
profit
8
The Big Questions

• Risk Management Challenge How do we prevent
  these events from occurring?
• At what point in patient volume do our systems
  become vulnerable to loss?
• What do we do when we are approaching that point?
• Patient Safety Perspective Since the risk of
  harm is never 0, how do we push the errors that
  we know will occur, further and further from the
  patient decreasing the likelihood of patient harm?

9
Methodology

• Phase I Exploratory Phase
• Operations leaders considered the empirical
  observations
• Scenarios posed that explained the phenomenon
• Measures of volume suggested
• Phase II Quantitative Analysis
• Comparison of high volume days against high error
  days.

10
Exploratory Phase

• More mistakes occur during high volume times. We
  dont know how busy the system can become before
  errors begin.
• All departments, clinical and non-clinical, have
  war stories associated with spikes in the
  census.
• Leaders have different measures of volume
  depending on the department and the exigencies of
  the hospital.
• Volume increases that occur simultaneously have
  an exponential effect especially on the support
  services.
• The relationship between volume and error is
  non-linear. It feels like one more patient in
  the ED or one additional laboring woman and chaos
  ensues.

11
Quantitative Phase Error Measure

• Incident reports with measure of harm to patients
  (no injury, minor injury, major injury, death)
• Harm assessed on a geometric scale 1, 4, 9, and
  16 (rather than 1, 2, 3, 4)
• Each day had a measure of harm which was the sum
  of the injuries on that day.
• All measures converted to z scores (mean0,
  standard deviation1)
• High volume defined as z 1.5.

12
Measures of Volume

• Med/Surg
• Census
• Change in census from the day before/shift
  before/last hour
• Number of admissions to the Med/Surg unit in a
  given period of time
• Number of Surgeries/Post-operative patients
  admitted in a given period of time
• Surgical Services
• Number of surgeries
• Number of add-ons
• Number of surgeons
• Number of Anesthesiologists

• Emergency Department
• Patients seen in the ED by day
• Patients in the ED by hour of day
• ED LOS by day/by hour of day
• Left w/out being seen by day/by hour of day
• Ambulance runs by day/by hour of day
• Number of acute psych patients
• Errors
• Number of errors
• Severity of errors
• Multiplier for severe errors
• Time of day errors occur

13
Quantitative Phase Volume Measures

• Total Census The midnight census
• Number of Surgeries The total number of
  surgeries performed that day, scheduled and
  unscheduled.
• Add-ons The number of unscheduled surgeries
• Percent Add-ons The number of add-ons as a
  percentage of the total surgeries performed.
• Behavioral Health Admissions The number of
  behavioral health patients admitted each day from
  the Emergency Department. In this facility,
  behavioral health admissions were a stress to the
  systems.
• of high volume measures on a given day

14
Methodology

• Each day had
• volume measures
• measure of harm
• of high volume measures on that day where high
  volume defined as z 1.5
• Comparison number of high volume measures
  against the percentage of high error days

15
Findings
16
Discussion

• Harm to patients is related to the number of high
  volume areas in the hospital.
• As the number of high volume areas increases, so
  does the likelihood of patient harm.
• Systems capacity explanations hold where support
  services are overwhelmed as the number of high
  volume areas increases.

17
Methodological Issues

• Is there any measure more sensitive than census?
• Is there one measure or a combination of measures
  that will predict harm/ likelihood of error?
• Is there a constraint on the number of volume
  indicators that we can use?
• How do we account for hard-wired vs.
  site-specific measures of volume?
• Are there signals in addition to volume that we
  should attend?

18
Need for further study

• How do we define high volume insofar as it
  relates to patient harm?
• What measures of high volume are most sensitive?
• The number of high error days is small. (11 out
  of 515 days had 3 or more high volume areas)
• Is there a fixed number or is capacity a moving
  target?
• Can we replicate these findings?

19
Implications

• Communication among the departments with respect
  to volume is critical for support services to
  adequately serve the patient care areas.
• The dialogue regarding systems capacity must
  begin
• Data-driven
• Avoid excuses
• Provide a rational system of upward flexing

20
Operational Improvements

• High volume procedures e.g. code purple
• Low cost solutions for expanding systems
  capacity, e.g. students to run specimens
• SWAT Nurses
• Bed meetings that include all the departments to
  increase communication
• PI activities to reduce inefficiencies and expand
  system capacity

21
References
Cook, R, Raumssen J, Going solid a model of
system dynamics and consequences for patient
safety. Qual Saf Health Care. 2005
14130-134. Flynn EA, Barker KN, Pepper GA,
Bates DW, Mikeal RL. Comparison of methods of
detecting medication errors in 36 hospitals and
skilled nursing facilities. Am J Health-Syst
Pharm 2002 59(5) 436-446. Litvak E, Buerhaus
PI, Davidoff F, Long MC, McManus ML, Berwick DM.
Managing unnecessary variability in patient
demand to reduce nursing stress and improve
patient safety. Joint Commission Journal on
Quality and Patient Safety. 200531(6)330-338. M
cManus ML, Long MC, Cooper A, Mandell J, Berwick
DM, Pagano M, Litvak E, Variability in surgical
caseload and access to intensive care services.
Anesthesiology. 2003 Jun98(6)1491-6 Reason,
J.T., Managing the risks of organizational
accidents. Croft Road, Aldershot, Hampershire,
England Ashgate Publishing Ltd., 19979.
Vickers A. If the normal distribution is so
normal, how come my data never are? Medscape,
Posted 05/08/2007. Weissman JS, Rothschild JM,
Bendavid E et al. Hospital Workload and Adverse
Events. Medical Care 2007 45448-445.