Analysis of CDCR Death Reviews 2006
Public Version
August 20, 2007
Kent Imai, MD
Consultant, California Prison Health Care Receivership

BACKGROUND
The CDCR Death Review Committee (DRC) is a multidisciplinary committee chaired by
the Statewide Medical Director and consisting of MDs, RNs, healthcare administrators,
and correctional officers. The DRC meets 2-4 times monthly to discuss and analyze each
death that occurs in the CDCR. Prior to the multidisciplinary reviews, an MD prepares a
written report on each death. A subcommittee with representation from mental health,
nursing, and custody reviews deaths from suicide. The main DRC reviews all other
deaths.
The majority of the year 2006 death review reports were prepared by CDCR QMAT
physicians. A minority of the reports were prepared by the Statewide Medical Director,
by the Regional CMOs, or by the UCSD contract review physicians.
Each death review report was based on a reading of the patient’s available medical
record. The reviewer attempted to assess the patient’s entire experience with medical
care during his/her period of incarceration. The reports focused specifically on the cause
of death and the quality of care provided to the patient. Upon identifying significant
departures from the community standard of care and potentially problematic providers,
the DRC referred cases to the Professional Practice Executive Committee (PPEC) for
further evaluation of the provider’s fitness for continued service in CDCR. The PPEC
interpretation of community standard considers what a reasonable, similarly credentialed
provider would do, given the situation in which the care in question was rendered.
The death reviews were valuable in identifying potentially unsafe practitioners. As one
step in its practitioner assessments, PPEC conducted pattern of practice reviews for these
individuals. Typically, the reviewer assessed a large sample of patient care interactions
(usually 40-60 patient charts, including the index death case and any other deaths
involving the clinician) for adherence to a community standard of care. After considering
evidence from multiple sources, PPEC took one of several actions:

1. Temporary restriction from practice in the CDCR, pending a complete review of
the clinician’s pattern of practice.
2. A program of remediation, e.g., taking a course in an area of deficiency, followed
by close monitoring
3. Suspension of privileges
4. No adverse action.
Sixty-two CDCR practitioners (56 MDs and DOs and 6 Nurse Practitioners) have had
adverse action taken by the PPEC, from June 2005 to July 2007. Of these, 41 were
initiated by the death reviews.

PURPOSE
Until now, the Death Review Committee and PPEC have been focused on identifying and
sanctioning individual practitioners. There has been little or no emphasis on identifying
systemic deficiencies of care and acting on them.
The purpose of this analysis is to categorize each of the 2006 deaths as non-preventable,
preventable, or possibly preventable, to summarize the major lapses in care (both
individual and systemic) contributing to the patient deaths, and to make recommendations
for quality improvement.

LIMITATIONS
There were significant limitations in the ability of the reviewers to conduct meaningful
death reviews.
A major limitation was the absence of a well-organized, easily navigated medical record.
This same limitation plagues the CDCR providers themselves during care provision. The
physician portion of the CDCR medical records includes hand-written progress notes
suffering from brevity, poorly documented reasoning, and illegible handwriting. The
medical records available to reviewers were often incomplete, making it difficult to
determine an accurate chronology of events or to “tease out” critical pieces of clinical
information. An important laboratory or x-ray result might be misfiled, or the record
might be missing recommendations of consultants or records of emergency room visits
and hospitalizations.
There was variation in the quality of the death review reports, in part because of the
difficulties in the medical record, and in part because there was no template or form for
guiding systematic death review. Some reports were quite brief and superficial. Others
went into great detail and reflected great effort at reconstructing events and determining
clinical reasoning. There was a spectrum of fault finding. Some reports concentrated
only on proximate causes of death and did not address the possibility of an early
opportunity to make a diagnosis that might have affected a patient’s prognosis. Some

reviewers focused entirely on individual culpability and did not address possible systemic
issues of care.
In early 2007 the DRC created a form for reports, leading to greater uniformity. The
form prompts reviewers to address nursing issues, systemic issues of care, and
preventability of death, in addition to individual practitioner lapses. Only the last 20 of
the year 2006 reports used this template.
The majority of deaths did not trigger autopsies. This is usual in the non-CDCR world as
well, but it makes complete clinical closure elusive, especially in the cases of sudden
cardiac arrest.
There are also inherent limitations in conducting a retrospective, case-based analysis such
as this one. There are no established criteria for attribution of “preventability.” Research
in this area is primarily epidemiological, comparing actual versus expected deaths in
large populations over time. A search of the medical literature revealed no case-based
studies for preventable deaths in adult primary care. Such studies would be difficult
precisely because creating rigorous criteria for preventability would be difficult. Another
limitation of this analysis is that it depends wholly on the judgment of a single reviewer.
For example, several of the sudden cardiac arrests were judged to be possibly preventable
because of a failure of clinicians to evaluate symptoms of syncope or chest pain in the
weeks or months prior to the patient’s death. Another reviewer might have judged these
deaths to have been non-preventable, because there is no assurance that a proper
evaluation of these red flag symptoms would in fact have prevented the patients’ deaths.
Many patients who have complete cardiovascular evaluations, who receive appropriate
medications and who have appropriate interventional procedures nevertheless succumb to
their disease. And without an autopsy, there is less assurance that the patient had a
preventable cardiovascular death. In short, there is no easy methodology that can reliably
quantify preventable deaths.
Despite the limitations in the death review process, it has proven useful in identifying
many egregious examples of individual errors in judgment and failures to perform
commensurate with community standards. This analysis consolidates findings for the
year 2006 deaths.

DEFINITIONS
Non-preventable: The health care system and individual providers probably would not
have been able to prevent the patient’s death. (Homicides and drug overdoses fall
here.)
Preventable: Better medical management or a better system of care would have
prevented death.
Possibly preventable: Better medical management or a better system of care might have
prevented death.

FINDINGS
Total year 2006 CDCR deaths
Suicides (not included in this analysis)
Execution (not included)
Death Reviews unavailable for this report
Death reviews in this analysis
Non-preventable deaths
315
Preventable deaths
18
Possibly preventable deaths
48

426
43
1
1
381

A. Non-Preventable Deaths
1. Causes of non-preventable death
105
53
28
17
17
16
16
14
11
10
7
6
5
5
3
2
4
319

Cancer
End-stage liver disease
Sudden cardiac arrest
AIDS
Drug overdose
Congestive heart failure
Homicide
Coronary artery disease (likely higher, because over 2/3 of cases of sudden
cardiac arrest are attributable in autopsy studies to CAD)
COPD
End-stage renal disease
Stroke
Pneumonia
Upper GI hemorrhage
Cocciodioidomycosis
Sepsis
Pulmonary embolism
1 each of diabetic ketoacidosis, neuroleptic malignant syndrome,
encephalitis, and subarachnoid hemorrhage
Total (Of the 315 cases, several had more than one major cause of death)

2. Lapses in care in cases of non preventable death
Lapses were noted in over half of the cases of non-preventable death. In
many cases, these lapses in care may have contributed to an earlier death or
more suffering in patients who had fatal diagnoses such as cancer or end stage
liver, heart, or kidney disease.
Cases
Lapses
66
Poor primary clinician management – includes instances of clinical
inertia in response to abnormal labs or x-rays, not treating to
established guidelines and targets (blood pressure, blood sugar, etc),
cursory evaluation of signs and symptoms (weight loss, new dementia,
syncope, “can’t walk”, new ascites, chest pain, abdominal pain),
delayed referral to a higher level of care, illegible handwriting, poor
documentation, and fragmented care
15
Poor management of terminal event, including failure to administer
narcan
13
System delays - medical records, delayed access to care, delayed
response to 602 appeals, delays in obtaining tests, etc
13
Delays in diagnosis
9
Patient “refusal” of care/evaluation
9
Delays in obtaining specialty referral
6
Poor “handoffs” between clinicians, including coordination between
inpatient and ambulatory, or at time of inmate transfers
5
Poor palliative care

B. Preventable Deaths
1. Causes of preventable deaths
6
3
2
1
1
1
1
1
1
1
18

Asthma
Sudden cardiac arrest
Congestive heart failure
Acute myocardial infarction
Duodenal ulcer, perforated
Hyperthermia [redacted]
Incarcerated hernia
Acute pancreatitis
Stroke (probable)
Testicular cancer
Total

2. Lapses in care in cases of preventable death

Asthma –failure of clinicians to follow published guidelines and standards
of care in the evaluation and management of asthma, failure of RNs to
appropriately triage sick asthmatics to an MD, failure to ensure timely
follow-up after treatment of an acute exacerbation, failure to recognize the
volatility of symptoms , failure to refer refractory asthma to a
pulmonologist, and a botched handoff in which a steroid dependent
asthmatic did not receive steroids for two days following transfer from a
county prison to a CDCR facility.
Sudden death –failure by MDs and midlevels to adequately evaluate “red
flag” symptoms such as exertional chest pain, chest pain associated with
dizziness, and recurrent syncope occurring weeks to months prior to death
in patients with cardiac risk factors.
Acute myocardial infarction – failure by MD to come in while on call to
evaluate a pt with hypotension and tachycardia, failure to correctly
interpret new edema and shortness of breath, and an 8 hour delay in access
to MD evaluation while experiencing “constant and extreme” chest pain
on the day of death.
Congestive heart failure – midlevel practicing beyond scope of practice in
unsupervised or poorly supervised situations, botched handoff from acute
hospital to CDCR facility, multiple failed appointments because of
dialysis, and MD failure to entertain diagnosis of CHF in a patient with
new orthopnea, exertional dyspnea and edema.
Perforated duodenal ulcer – failure by MDs and RNs to adequately
respond to patient complaint about severe abdominal pain on multiple
occasions over five days, resulting in prolonged delay in diagnosis and
treatment.
Hyperthermia – unsafe transfer of [redacted] patient from one CDCR
facility to another [redacted] resulting in death from hyperthermia.
Incarcerated hernia – five week delay in referral to specialist for a patient
with recurrent severe abdominal pain, vomiting and known bilateral
inguinal hernias.
Acute pancreatitis – failure of RNs and MDs to properly triage, evaluate
and manage a patient who presented nine times over three days with
severe “10/10” abdominal pain, resulting in prolonged delay in recognition
and treatment.
Stroke – midlevel practicing beyond scope in poorly supervised setting
who failed to evaluate a pt who had symptoms of weakness, inability to
walk [redacted] and who was repeatedly known to be “down” for more
than 48 hours.
Testicular cancer – two year delay in diagnosis of testicular cancer in
[redacted] patient with chronic testicular pain, metastatic at time of eventual
diagnosis, botched transfer with inadequate information passed from

facility to facility (lost urology consult), failure of MDs to work up for
cancer in a young man with 17 months of testicular pain.

C. Possibly Preventable Deaths
1. Causes of possibly preventable deaths
5
4
4
3
3
3
2
2
2
2
2
2
2

Sudden cardiac arrest
Coccidioidomycosis
AIDS
Acute myocardial infarction
Bowel perforation
Sepsis
Coronary artery disease
Congestive heart failure
Drug overdose
Gastrointestinal hemorrhage
Subdural hematoma
Colorectal cancer
Opiate toxicity
1 each of COPD, gastric cancer, cholecystitis in end-stage liver
disease, acute renal failure from rhabdomyolysis following trauma,
12 cervical cancer, lung cancer, pneumonia, aortic dissection, drug
induced hepatitis, diabetic ketoacidosis, carcinoma of thymus, seizure
disorder
48 Total

2. Lapses in care in cases of possibly preventable death Cases
Lapses
30
Errors by individual physicians, nursing and midlevel staff – includes
failure to adequately evaluate clinical “red flag” signs and symptoms,
(chest pain, abdominal pain , weight loss, seizures, altered mental
status, fever and tachycardia, poorly resolving pneumonia, joint
effusion, history of significant trauma), failure to adequately pursue
abnormal test results (leucopenia, abnormal blood sugars, abnormal
radiology studies), failure to transfer patients to appropriate higher
levels of care, inadequate clinical surveillance of known conditions
(cervical cancer, immune compromised patients)
11
Delayed referrals for specialty care or special tests – (cardiology,
gastroenterology, vascular surgery, stress tests, etc.)
9
Delays in access (delayed response to patient requests for care –
“7362s”)
7
Poor provider communication, including failure to act on specialist

6
5
3
2

recommendations and lost medical information when patients undergo
interfacility transfers
Missed abnormal test results (chest x-rays, CT scan, blood sugars,
positive stress tests)
Fragmentation of care, multiple providers with no individual
ownership of a patient’s complaint or abnormal finding
Poor response to emergency or “man down” situations
Surgical or procedural complications (colonoscopy and herniorhaphy
resulting in perforated bowel)

DISCUSSION
A. Lapses in Care
Significant lapses in care were noted in more than half of the death reviews. These can
be divided into individual practitioner lapses, systemic lapses, and “no-fault” lapses

1. Individual practitioner errors in judgment or attitude
-

Failure to appreciate potentially serious signs and symptoms (exertional chest
pain, new onset shortness of breath and dizziness, unexplained tachycardia as
harbingers of cardiac events, severe abdominal pain and abdominal distention as
signs of acute abdominal catastrophe, increased use of inhalers as prelude to
status asthmaticus),

-

Failure to tailor the pace of evaluation to the clinical situation (rectal bleeding,
testicular pain, indicating rapid workup to detect potentially curable cancers ),

-

Failure to perform the basic history and physical examination,

-

Failure to follow well established guidelines for care (asthma, diabetes mellitus,
hypertension, coccidioidomycosis)

-

Failure to apply critical thinking or to enlist help in difficult cases

-

Superficial or no documentation to indicate thought processes.

-

Failure to take individual responsibility for patient outcomes

2. Systemic lapses
-

A system that allows delays in triaging and processing patient requests for care
resulting in patients with red flag symptoms not being evaluated in a timely
manner.

-

A system that allows fragmentation of care and clinical inertia, leading to lack of
individual practitioner responsibility and accountability for each patient.

-

Systemic and pervasive prolonged delays in specialty referrals

-

No system for flagging abnormal test results,

-

Incomplete medical records

-

Poorly managed transfers of care – when patients move from one facility to
another, there is increased risk of medical error.

-

Practices which place mid level providers in vulnerable clinical situations, poorly
supported or unsupported, with little or no mentoring.

-

Practice environments (noisy, unkempt, crowded, lacking privacy) and patient
characteristics (high rate of dual diagnosis, chronic pain, and manipulation for
secondary gain) and other cultural factors which promote practice isolation and
discourage collegiality and professionalism.

3. “No-fault” lapses
-

Patient “non-adherence” to suggested treatment

-

Patient “refusal” of care or evaluation (sometimes masking frustration with the
system of care or reflecting poor provider – patient communication)

B. Trends in Preventable Deaths Over Time
There was no clear trend indicating an increase or decrease in the number of preventable
deaths over time. Unlike the situation in hospitals, in which quality improvements can
lead to aggregate decreases in mortality within the space of a year, improvements in
primary care may take longer before yielding mortality decreases.
Month
January
February
March
April
May
June
July
August
September
October
November
December

Deaths

Preventable

42
35
46
31
36
39
40
36
25
32
32
29

0
3
1
1
1
0
2
2
1
3
0
4

Possibly
All preventable and possibly
preventable
preventable deaths
3
3
4
7
4
5
8
9
4
5
6
6
3
5
1
3
3
4
6
9
3
3
3
7

C. Comment on CDCR Environment of Care
CDCR medical staff has been working in an environment of care characterized by
crowded and poorly equipped clinical areas. The medical record systems are outdated
and medical information is difficult to retrieve. The dispensing of prescribed drugs is
often delayed, and there is an unreliable system for refilling medications for the treatment
of chronic medical diseases such as diabetes, hypertension, asthma and coronary heart
disease. The drug profile information is unreliable. Practices in many of the prisons
focus on episodic care rather than continuity of care and preventive medicine. The
environment does not guarantee patient confidentiality, and the culture does not promote
patient advocacy.
The patient population has a number of unfavorable characteristics, such as a high
incidence of dual diagnosis (serious mental illness coexisting with physical illness),
chronic hepatitis, HIV infection, drug and alcohol addiction, and skillful manipulation for
secondary gain.
Despite these barriers, it is noteworthy that 167 of the death reports contained no serious
lapses in medical care. This is a reassuring indication that there are many conscientious
providers and RNs who are doing a good job despite the environment in which they find
themselves.

RECOMMENDATIONS
The CDCR must create a culture of patient safety, in which clinicians readily identify
mistakes and system vulnerabilities and in which all staff share in the responsibility for
optimal patient outcomes. Systems should be reviewed or redesigned to support this end.
To that end, the Death Review Committee should continue in on-going fashion the
analyses piloted in this analysis, identifying not only individual performance issues but
also the most common systemic lapses in care. The Committee should begin to
standardize a list of the lapses and vulnerabilities that contribute to preventable deaths.
The Joint Commission provides examples of how to proceed in this area, e.g., in
categorizing the causes of sentinel events or specifically the causes of delays in treatment
(see the Sentinel Event Alert of June 17, 2002). The Committee should continue its
efforts to standardize its methodology for classifying preventable deaths.
These overall recommendations and most of the specific recommendations which follow
are contained in Goals B, C, and D of the California Receiver’s Plan of Action of May
2007 (POA). Where applicable, relevant POA goals and objectives follow each
numbered recommendation below.

1. Continue PPEC evaluation of individual practitioners referred by the Death
Review Committee.
2. Develop and circulate a Clinical Newsletter in order to improve communication,
educate CDCR providers about important findings of the Death Reviews and to
make meaningful clinical suggestions for improving care. (C.8.1, A.8.5.2)
3. Develop a system wide quality initiative focusing on the management of asthma.
(B.2.5, B.2.6.1)
4. Develop system-wide quality initiatives on the recognition and management of
“red flag” clinical signs, and other subjects, using death review cases as
indicators. (C.1.1, C.1.2, C.6.1)
5. Pilot practitioner “daily reports” at each prison for purposes of peer collaboration
and discussion of problem cases, mistakes, “near misses,” cases of patient nonadherence or refusal of care, local system process redesign, development of
collegiality, and shared responsibility for patient care. (C.5, C.6, C.8)
6. Redesign CDCR processes for mid-level credentialing, privileging, supervision
and mentoring. (A.8.5)
7. Redesign CDCR systems of care (including scheduling) to promote individual and
shared responsibility for patient care outcomes, and to reduce fragmentation of
care wherever possible. (B.3)
8. Redesign process of RN triaging of form 7362s to eliminate delays in care.
(B.7.1, C)
9. Develop systems for tracking and following up abnormal laboratory and other test
results. (B.12.1, B.12.2)
10. Create new templates for managing requests for specialty services in order to
meet minimum standards for emergency (24 hour) urgent (7-14 day) and routine
(60 day) priorities, and to ensure that consultation results are seen by ordering
clinicians within one week of service. (B.3.1.8)
11. Review process for response to emergencies. (B.1)
12. Design and implement system-wide integrated health information systems. (goal
D)
13. Redesign the environment of care to promote efficiency, teamwork,
professionalism, and respect for patients, creating an ethically-based system of
care. (B.3, B.10)
14. Wherever applicable, develop the standard quality metrics to support the
foregoing recommendations. (POA, page 47)