Medical Errors: Causes and Prevention

1
Medical ErrorsCauses and Prevention

• Roger L. Bertholf, Ph.D.
• Associate Professor of Pathology
• University of Florida Health Science
  Center/Jacksonville

2
IOM To Err Is Human Building a Safer Health
System (2000)

• Frequency
• Cost
• Outcomes
• Types
• Causes
• Recommendations

3
Adverse Event vs. Error

• An adverse event is an injury caused by medical
  management rather than the underlying condition
  of the patient. An adverse event attributable to
  error is a "preventable adverse event." Negligent
  adverse events represent a subset of preventable
  adverse events that satisfy legal criteria used
  in determining negligence (i.e., whether the care
  provided failed to meet the standard of care
  reasonably expected of an average physician
  qualified to take care of the patient in
  question).
• An error is defined as the failure of a planned
  action to be completed as intended (i.e., error
  of execution) or the use of a wrong plan to
  achieve an aim (i.e., error of planning).

About half of preventable AEs are considered
negligent
4
Examples of Medical Errors

• Diagnostic error (inappropriate therapy)
• Equipment failure
• Infection (nosocomial, post-operative)
• Transfusion-related injury
• Misinterpretation of medical orders
• System failures that compromise diagnostic or
  treatment processes.

5
Frequency of Medical Errors
Study AEs Errors Fatal Est. Deaths
NY (1984) 2.9 58 13.6 98,000
CO/UT (1992) 3.7 53 6.6 44,000
MVA 43,000 Breast CA 42,000 AIDS 16,000.
8th most frequent cause overall.
6
How reliable is this estimate?

• Includes only AEs producing a specified level or
  harm
• Two reviewers had to agree on whether an AE was
  preventable or negligent
• Included only AEs documented in the patient
  record

Some studies, using other sources of information
about adverse events, produced higher estimates.
7
Cost

• Adverse events 37.6 50 billion
• Preventable adverse events 17 29 billion
• Half of cost is for health care
• Represent 4 (AE) and 2 (errors) of all health
  care costs

lost income, lost household production,
disability, health care costs Exceeds total
cost of treating HIV and AIDS
8
Causes
Medication error 19
Wound infection 14
Technical complications 13
Leape et al. (1991) The nature of adverse events
in hospitalized patients (1,133 AEs studied in
30,195 admissions) Overall frequency
(inpatients) is 3 per 1,000 medication orders 2
per 1,000 considered significant errors
9
AHA List of Medication Errors

• Incomplete patient information
• Unavailable drug information (warnings)
• Miscommunication of medication order
• Confusion between drugs with similar names
• Lack of appropriate drug labeling
• Environmental conditions that distract health
  care providers

10
Most Common Medication Errors
Failure to adjust dosage in response to a change in hepatic/renal function 13.9
History of allergy to the same or related medication 12.1
Wrong drug name, dosage form, or abbreviation on order 11.4
Incorrect dosage calculation 11.1
Atypical or unusual critical dosage consideration 10.8
11
A Comparison of Risks
Risk (per flight) of dying in a commercial airline accident 1 in 8 million
Risk (per hospital admission) of dying from a medical error gt1 in 1,000
1 in 2 million from 1967-1976
12
Six Sigma Quality Control

• Quality Management program designed by Mikel
  Harry and Richard Schroeder in 2000
• Strives to make QM a quantitative science
• Sets performance standards and goals for a
  production process

13
Six Sigma Paradigm DMAIC
14
Six Sigma Process Performance
Target
- Tolerance
Tolerance
Probability
.67
.95
0
1
2
3
-1
-2
-3
4
5
6
-4
-5
-6
SD (?)
15
Six Sigma Performance

• Goal is to achieve lt 1 DPM
• Not all processes can achieve the 6? level of
  performance
• Demings Principle is that fewer defects leads
  to increased productivity, efficiency, and lower
  cost

16
Healthcares Six Sigma Performance
Process Errors Sigma
Preventable adverse events 3.0 2.5
Lab order accuracy 1.8 3.6
Reporting errors 0.048 4.8
False negative PAP 2.4 3.45
Unacceptable specimen 0.3 4.25
Duplicate test orders 1.52 3.65
17
What Causes Accidents?
18
Sidney Dekker
What is striking about many accidents is that
people were doing exactly the sorts of things
they would usually be doingthe things that
usually lead to success and safety. . .Accidents
are seldom preceded by bizarre behavior. From
The Field Guide to Human Error Investigations
(2002)
19
A Primer on Accident Investigation

• Human error as a cause
• Human error as a symptom

20
Human Error

• Bad Apple Theory
• Complex systems are inherently safe
• Human intervention subverts the inherent safety
  of complex systems
• Reaction to failure
• Bad outcome bad decision
• Retrospective, proximal, counterfactual, and
  judgmental

21
The Bad Apple Theory

• The illusion of success
• Bad procedures often produce good results
• Success breeds confidence
• Failure is an aberration
• The system must be safe
• The economical answer
• It is easier to change human behavior than it is
  to change systems

22
Assigning Blame

• Retrospective

23
Retrospective Analysis
Time
24
Assigning Blame

• Retrospective
• Proximal

25
Proximity

• It is intuitive to focus on the location where
  the failure occurred
• Sharp end vs. Blunt end
• The sharp end is the point at which the failure
  occurs
• The blunt end is the set of systems and
  organizational structure that supports the
  activities at the sharp end

26
Retrospective Analysis
Sharp End
Time
Institution Systems Procedures Organization
Blunt End
27
Assigning Blame

• Retrospective
• Proximal
• Counterfactual

28
What Might Have Been. . .

• In retrospect, it is always easy to see where
  different actions would have averted a bad
  outcome
• In retrospect, the outcome of any potential
  action is already known
• Counterfactuals pose alternate scenarios, which
  are rarely useful in determining the true cause

29
Assigning Blame

• Retrospective
• Proximal
• Counterfactual
• Judgmental

30
The Omniscient Perspective

• As an investigator, you always know more than the
  participants did
• It is very difficult, if not impossible to judge
  fairly the reactions of those who had less
  information than you
• Investigators define failure based on outcome

31
Lessons for Investigators

• There is no primary cause
• Every action affects another
• There is no single cause
• Errors in complex systems are nearly always
  multi-focal
• A definition of human error is elusive
• Definition of error
• Humans operate within complex systems

32
Failure Mode and Effects Analysis

• Everything will eventually fail
• Humans frequently make errors
• The cause of a failure is often beyond the
  control of an operator

33
10 Steps for FMEA

1. Review the process
2. Brainstorm potential failure modes
3. List potential effects of each failure mode
4. Assign a severity rating
5. Assign an occurrence rating
6. Assign a detection rating
7. Calculate the risk priority number for each
   effect
8. Prioritize these failure modes based on the RPN
   and severity
9. Take action to reduce or eliminate the high-risk
   failure modes
10. Recalculate the RPN

34
Ranking the Failure Modes

• Calculate the RPN
• Rate Severity, Occurrence, and Detection on a
  scale of 1 10
• RPN S x O x D (maximum 1000)
• Prioritize Failure modes
• Not strictly based on RPN
• Severity of 9 or 10 should get priority
• Goal is to reduce RPN

35
STRETCH!
36
Case Exercise 1
A 91-year-old female was transferred to a
hospital-based skilled nursing unit from the
acute care hospital for continued wound care and
intravenous (IV) antibiotics for
methicillin-resistant Staphylococcus aureus
(MRSA) osteomyelitis of the heel. She was on IV
vancomycin and began to have frequent, large
stools.
37
Case Exercise 1
The attending physician ordered a test for
Clostridium difficile on Friday, and was then off
for the weekend. That night, the test result came
back positive. The lab called infection control,
who in turn notified the float nurse caring for
the patient. The nurse did not notify the
physician on call or the regular nursing staff.
Isolation signs were posted on the patient's door
and chart, and the result was noted in the
patient's nursing record. Each nurse who
subsequently cared for this patient assumed that
the physician had been notified, in large part
because the patient was receiving vancomycin.
However, it was IV vancomycin (for the MRSA
osteomyelitis), not oral vancomycin, which is
required to treat C. difficile.
38
Case Exercise 1
On Monday, the physician who originally ordered
the C. difficile test returned to assess the
patient and found the isolation signs on her
door. He asked why he was never notified and why
the patient was not being treated. The nurse on
duty at that time told him that the patient was
on IV vancomycin. The float nurse, who had
received the original notification from infection
control, stated that she had assumed the
physician would check the results of the test he
had ordered. Due to the lack of follow-up, the
patient went three days without treatment for C.
difficile, and continued to have more than 10
loose stools daily. Given her advanced age, this
degree of gastrointestinal loss undoubtedly
played a role in her decline in functional status
and extended hospital stay.
39
Case Exercise 1

• What are the systems/processes involved in this
  incident?
• What were the failure points?

40
Analysis

• MD failed to check the result of an ordered test
• Float RN wrongly assumed that MD had been
  notified of the result
• RN incorrectly assumed that IV vancomycin was
  adequate therapy

41
Failure Points

• Laboratory system for reporting critical results
• Is a positive C. difficile culture considered a
  panic result?
• To whom are panic values reported?
• RN/MD communication
• Does the institution foster an environment where
  RNs can comfortably question MD orders?

42
Lisa Belkin
. . . it is virtually impossible for one mistake
to kill a patient in the highly mechanized and
backstopped world of a modern hospital. A cascade
of unthinkable things must happen, meaning
catastrophic errors are rarely a failure of a
single person, and almost always a failure of a
system. From How Can We Save the Next Victim?
(NY Times Magazine, June 15, 1997)
43
Case Exercise 2
An 81-year-old female maintained on warfarin for
a history of chronic atrial fibrillation and
mitral valve replacement developed asymptomatic
runs of ventricular tachycardia while
hospitalized. The unit nurse contacted the
physician, who was engaged in a sterile procedure
in the cardiac catheterization laboratory (cath
lab) and gave a verbal order, which was relayed
to the unit nurse via the procedure area nurse.
Someone in the verbal order process said "40 of
K." The unit nurse (whose past clinical
experience was in neonatal intensive care) wrote
the order as "Give 40 mg Vit K IV now."
44
Case Exercise 2
The hospital pharmacist contacted the physician
concerning the high dose and the route and
discovered that the intended order was "40 mEq of
KCl po." The pharmacist wrote the clarification
order. However, the unit nurse had already
obtained vitamin K on override from the Pyxis
MedStation (an automated medication dispensing
system) and administered the dose intravenoustly
(IV). The nurse attempted to contact the
physician but was told he was busy with
procedures. A routine order to increase warfarin
from 2.5 mg to 5 mg (based on an earlier INR) was
written later in the day and interpreted by the
evening shift nurse as the physicians response
to the medication event. The physician was not
actually informed that the vitamin K had been
administered until the next day. Heparin was
initiated and warfarin was re-titrated to a
therapeutic level. The patients INR was
sub-herapeutic for 3 days, but no untoward
clinical consequences occurred.
45
Case Exercise 2

• What are the systems/processes involved in this
  incident?
• What were the failure points?

46
Analysis

• Verbal orders
• Third party messengers
• Use of abbreviations
• Failure to question unusual orders
• Lack of control over medication availability

47
Failure Points

• Hospital policy for medication orders
• Read Back requirement
• Ability to circumvent pharmacist review

48
J.C.R. Licklider (1915-1990)
It seems likely that the contributions of human
operators and computers will blend together so
completely in many operations that it will be
difficult to separate them neatly in
analysis. From Man-Computer Symbiosis (1960)
49
Anatomy of a Laboratory Error
50
Phase I A failed calibration

• Recalibration of the acetaminophen assay was
  prompted by a QC failure
• Recalibration was followed by acceptable QC
  results

51
Phase II QC failures

• Subsequent QC measurements produced an error code
  indicating the result was above the linear limit
  of the method
• QC failures went unnoticed, since the LIS did not
  display the error code
• Several patient specimens were reported
  incorrectly, resulting in inappropriate treatment

52
Phase III Discovery

• The ED staff contacted the laboratory to question
  the high acetaminophen result on a patient who
  denied recent ingestion of the drug
• Investigation revealed the QC failures, and the
  assay was successfully recalibrated

53
Phase IV InvestigationPrincipal Questions

• Why was an acceptable QC result obtained
  immediately after a failed calibration?
• Why didnt the technologists notice subsequent QC
  failures?
• Should the clinicians have been more suspicious
  of unusually high results?

54
The Process
55
Failure Points in The Process
56
Unrecognized calibration failure

• Roche modular
• Throughput/timing algorithm

57
Unnoticed QC failures

• Interface through Digital Innovations box
• Error codes are rare in QC results
• Supervisory review does not occur regularly on
  weekends

58
Lack of clinical suspicion

• History is often unreliable in overdose cases
• An antidote for acetaminophen exists
• Symptoms of acetaminophen toxicity may not appear
  until after the window of therapeutic opportunity
  has passed

59
Conclusions

• An unexpected error occurred in the calibration
  algorithm encoded in the instrument software
• The failure of information to cross the
  instrument/LIS interface masked the erroneous
  control results
• Suspect results were not immediately apparent to
  clinicians

60
Lessons

• Complex technologies always have unexpected
  failure modes
• Interfaces between systems and operators are
  opportunities for distortion or loss of important
  information
• The fallacy of the un-rocked boat

61
Richard I. Cook
Recognizing hazard and successfully manipulating
system operations to remain inside the tolerable
performance boundaries requires intimate contact
with failure. From How Complex Systems Fail
(2002)
62
How Complex Systems Fail

• Complex systems are intrinsically hazardous
  systems
• Complex systems are heavily and successfully
  defended against failure
• Catastrophe requires multiple failuressingle
  point failures are not enough
• Complex systems contain changing mixtures of
  failures latent within them

63
How Complex Systems Fail

• Catastrophe is always just around the corner
• Post-accident attribution to a root cause is
  fundamentally wrong
• Human operators have dual roles as producers and
  as defenders against failure
• Human practitioners are the adaptable element of
  complex systems

64
How Complex Systems Fail

• Change introduces new forms of failure
• Safety is a characteristic of systems and not of
  their components
• Failure-free operations require experience with
  failure

65
IOM Recommendations

• Establish national focus
• Identify and learn from medical errors through
  mandatory reporting
• Raise standards and expectations
• Implement safe practices

66
AHRQ Safety Recommendations for Patients

• Ask questions if you have doubts or concerns
• Keep and bring a list of ALL the medicines you
  take
• Get the results of any test of procedure
• Talk to your doctor about which hospital is best
  for your health needs
• Make sure you understand what will happen if you
  need surgery