Missed and Delayed Diagnoses in the Ambulatory Setting:
A Study of Closed Malpractice Claims
Tejal K. Gandhi, MD, MPH; Allen Kachalia, MD, JD; Eric J. Thomas, MD, MPH; Ann Louise Puopolo, BSN, RN; Catherine Yoon, MS;
Troyen A. Brennan, MD, JD; and David M. Studdert, LLB, ScD
Background: Although missed and delayed diagnoses have become
an important patient safety concern, they remain largely unstudied,
especially in the outpatient setting.
Objective: To develop a framework for investigating missed and
delayed diagnoses, advance understanding of their causes, and
identify opportunities for prevention.
Design: Retrospective review of 307 closed malpractice claims in
which patients alleged a missed or delayed diagnosis in the ambu-
Setting: 4 malpractice insurance companies.
Measurements: Diagnostic errors associated with adverse out-
comes for patients, process breakdowns, and contributing factors.
Results: A total of 181 claims (59%) involved diagnostic errors that
harmed patients. Fifty-nine percent (106 of 181) of these errors
were associated with serious harm, and 30% (55 of 181) resulted
in death. For 59% (106 of 181) of the errors, cancer was the
diagnosis involved, chiefly breast (44 claims [24%]) and colorectal
(13 claims [7%]) cancer. The most common breakdowns in the
diagnostic process were failure to order an appropriate diagnostic
test (100 of 181 [55%]), failure to create a proper follow-up plan
(81 of 181 [45%]), failure to obtain an adequate history or perform
an adequate physical examination (76 of 181 [42%]), and incorrect
interpretation of diagnostic tests (67 of 181 [37%]). The leading
factors that contributed to the errors were failures in judgment (143
of 181 [79%]), vigilance or memory (106 of 181 [59%]), knowl-
edge (86 of 181 [48%]), patient-related factors (84 of 181 [46%]),
and handoffs (36 of 181 [20%]). The median number of process
breakdowns and contributing factors per error was 3 for both
(interquartile range, 2 to 4).
Limitations: Reviewers were not blinded to the litigation outcomes,
and the reliability of the error determination was moderate.
Conclusions: Diagnostic errors that harm patients are typically the
result of multiple breakdowns and individual and system factors.
Awareness of the most common types of breakdowns and factors
could help efforts to identify and prioritize strategies to prevent
Ann Intern Med. 2006;145:488-496.
For author affiliations, see end of text.
current diagnostic process in health care is complex, cha-
otic, and vulnerable to failures and breakdowns. For exam-
ple, one third of women with abnormal results on mam-
mography or Papanicolaou smears do not receive follow-up
care that is consistent with well-established guidelines (1,
2), and primary care providers often report delays in re-
viewing test results (3). Recognition of systemic problems
in this area has prompted urgent calls for improvements (4).
However, this type of error remains largely unstudied
(4). At least part of the reason is technical: Because omis-
sions characterize missed diagnoses, they are difficult to
identify; there is no standard reporting mechanism; and
when they are identified, documentation in medical
records is usually insufficiently detailed to support detailed
causal analyses. The result is a relatively thin evidence base
from which to launch efforts to combat diagnostic errors.
Moreover, conceptions of the problem tend to remain
rooted in the notion of physicians failing to be vigilant or
up-to-date. This is a less nuanced view of error causation
than careful analysis of other major patient safety prob-
lems, such as medication errors (5, 6), has revealed.
Several considerations highlight malpractice claims as
a potentially rich source of information about missed and
delayed diagnoses. First, misdiagnosis is a common allega-
tion. Over the past decade, lawsuits alleging negligent mis-
diagnoses have become the most prevalent type of claim in
issed and delayed diagnoses in the ambulatory set-
ting are an important patient safety problem. The
the United States (7, 8). Second, diagnostic breakdowns
that lead to claims tend to be associated with especially
severe outcomes. Third, relatively thorough documenta-
tion on what happened is available in malpractice insurers’
claim files. In addition to the medical record, these files
include depositions, expert opinions, and sometimes the
results of internal investigations.
Previous attempts to use data from malpractice claims
to study patient safety have had various methodologic con-
straints, including small sample size (9, 10), a focus on
single insurers (11) or verdicts (9, 10) (which constitute
?10% of claims), limited information on the claims (8–
11), reliance on internal case review by insurers rather than
by independent experts (8, 11), and a general absence of
Editors’ Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
Editorial comment. . . . . . . . . . . . . . . . . . . . . . . . . . 547
Summary for Patients. . . . . . . . . . . . . . . . . . . . . . . I-12
Conversion of tables and figures into slides
Annals of Internal Medicine
488 © 2006 American College of Physicians
robust frameworks for classifying types and causes of fail-
ures. To address these issues, we analyzed data from closed
malpractice claims at 4 liability insurance companies. Our
goals were to develop a framework for investigating missed
and delayed diagnoses, advance understanding of their
causes, and identify opportunities for prevention.
Four malpractice insurance companies based in 3 re-
gions (northeastern, southwestern, and western United
States) participated in the study. Collectively, the partici-
pating companies insured approximately 21 000 physi-
cians, 46 acute care hospitals (20 academic and 26 nonac-
ademic), and 390 outpatient facilities, including a wide
variety of primary care and outpatient specialty practices.
The ethics review boards at the investigators’ institutions
and at each review site approved the study.
Data were extracted from random samples of closed
claim files from each insurer. A claim is classified as closed
when it has been dropped, dismissed, paid by settlement,
or resolved by verdict. The claim file is the repository of
information accumulated by the insurer during the life of a
claim. It captures a wide variety of data, including the
statement of claim, depositions, interrogatories, and other
litigation documents; reports of internal investigations,
such as risk management evaluations and sometimes root-
cause analyses; expert opinions from both sides; medical
reports detailing the plaintiff’s preevent and postevent con-
dition; and, while the claim is open, medical records per-
taining to the episode of care at issue. We reacquired the
relevant medical records for sampled claims.
Following previous studies, we defined a claim as a
written demand for compensation for medical injury (12,
13). Claims involving missed or delayed diagnoses were
defined as those alleging an error in diagnosis or testing
that caused a delay in appropriate treatment or a failure to
act or follow up on results of diagnostic tests. We excluded
allegations related to pregnancy and those pertaining to
care rendered solely in the inpatient setting.
We reviewed 429 diagnostic claims alleging injury due
to missed or delayed diagnoses. Insurers contributed to the
study sample in proportion to their annual claims volume
(Appendix, available at www.annals.org). The claims were
divided into 2 main categories based on the primary setting
of the outpatient care involved in the allegation: the emer-
gency department (122 claims) and all other locations (for
example, physician’s office, ambulatory surgery, pathology
laboratory, or radiology suites) (307 claims). The latter
group, which we call ambulatory claims, is the focus of this
Study Instruments and Claim File Review
Physicians who were board-certified attendings, fel-
lows, or third-year residents in internal medicine reviewed
sampled claim files at the insurers’ offices or insured facil-
ities. Physician-investigators trained the reviewers in the
content of claim files, use of the study instruments, and
confidentiality procedures in 1-day sessions at each site.
The reviewers also used a detailed manual. Reviews took
on average 1.4 hours per file. To test review reliability, a
second reviewer reviewed a random selection of 10% (42
of 429) of the files. Thirty-three of the 307 ambulatory
claims that are the focus of this analysis were included in
the random blinded re-review. A sequence of 4 instruments
guided the review. For all claims, insurance staff recorded
administrative details of the case (Appendix Figure 1,
available at www.annals.org), and clinical reviewers re-
corded details of the adverse outcome the patient experi-
enced, if any (Appendix Figure 2, available at www.annals
.org). Reviewers scored adverse outcomes on a 9-point
severity scale ranging from emotional injury only to death.
This scale was developed by the National Association of
Insurance Commissioners (14) and has been used in pre-
vious research (15). If the patient had multiple adverse
outcomes, reviewers scored the most severe outcome. To
simplify presentation of our results, we grouped scores on
Efforts to reduce medical errors and improve patient safety
have not generally addressed errors in diagnosis. As with
treatment, diagnosis involves complex, fragmented pro-
cesses within health care systems that are vulnerable to
failures and breakdowns.
The authors reviewed malpractice claims alleging injury
from a missed or delayed diagnosis. In 181 cases in which
there was a high likelihood that error led to the missed
diagnosis, the authors analyzed where the diagnostic pro-
cess broke down and why. The most common missed di-
agnosis was cancer, and the most common breakdowns
were failure to order appropriate tests and inadequate fol-
low-up of test results. A median of 3 process breakdowns
occurred per error, and 2 or more clinicians were involved
in 43% of cases.
The study relied on malpractice claims, which are not rep-
resentative of all diagnostic errors that occur. There was
only moderate agreement among the authors in their sub-
jective judgments about errors and their causes.
Like other medical errors, diagnostic errors are multifacto-
rial. They arise from multiple process breakdowns, usually
involving multiple providers. The results highlight the chal-
lenge of finding effective ways to reduce diagnostic errors
as a component of improving health care quality.
Missed or Delayed Diagnoses
3 October 2006 Annals of Internal Medicine Volume 145 • Number 7 489
Current Author Addresses: Drs. Gandhi and Kachalia and Ms. Yoon:
Division of General Internal Medicine, Brigham and Women’s Hospital,
75 Francis Street, Boston, MA 02114.
Dr. Thomas: University of Texas Medical School at Houston, 6431
Fannin, MSB 1.122, Houston, TX 77030.
Ms. Puopolo: CRICO/RMF, 101 Main Street, Cambridge, MA 02142.
Dr. Brennan: 151 Farmington Avenue, RC5A, Hartford, CT 06156.
Dr. Studdert: Harvard School of Public Health, 677 Huntington Ave-
nue, Boston, MA 02115.
Author Contributions: Conception and design: T.K. Gandhi, E.J.
Thomas, T.A. Brennan, D.M. Studdert.
Analysis and interpretation of the data: T.K. Gandhi, A. Kachalia, C.
Yoon, D.M. Studdert.
Drafting of the article: T.K. Gandhi, A. Kachalia, T.A. Brennan, D.M.
Critical revision of the article for important intellectual content: T.K.
Gandhi, A. Kachalia, E.J. Thomas, T.A. Brennan, D.M. Studdert.
Final approval of the article: T.K. Gandhi, A. Kachalia, E.J. Thomas,
Provision of study materials or patients: A.L. Puopolo.
Statistical expertise: D.M. Studdert.
Obtaining of funding: D.M. Studdert.
Administrative, technical, or logistic support: A. Kachalia, A.L. Puopolo,
Collection and assembly of data: A. Kachalia, A.L. Puopolo, D.M.
Annals of Internal Medicine
W-136 3 October 2006 Annals of Internal Medicine Volume 145 • Number 7