Full text of DOI:10.1111/j.1553-2712.2001.tb01276.x

ACADEMIC EMERGENCY MEDICINE • February 2001, Volume 8, Number 2
125
Evaluation of Missed Diagnoses for Patients Admitted
from the Emergency Department
MARY
C
HELLIS, MD, JAMES E. OLSON, P
HD, JAMES AUGUSTINE, MD,
GLENN C. HAMILTON, MD, MSM
Abstract. Objective: To define a quality assurance whose ED medical records were scored with errors by
instrument to evaluate errors in diagnostic processes both evaluators. Interevaluator reliability was as-
made by physicians in the emergency department sessed using Cochran’s Q-test with a selected series
(ED). Methods: This was a retrospective clinical in- of medical records. Results: Twenty-eight records
vestigation of inpatient ED records. Over a six-year (0.6%) were found to contain one or more errors in
period, 5,000 medical records of admitted patients the diagnostic process that contributed to misdiag-
were randomly selected for evaluation. Each record nosis. For these patients appropriate diagnosis was
was initially examined by one of five physician eval- not made until one to 16 days after admission. Three
uators. If the primary ED diagnosis differed from the patients of 18 whose records were available for de-
primary discharge diagnosis, the ED record was in- tailed review may have suffered complications that
spected to determine reasons for the misdiagnosis. resulted, in part, from the delay in diagnosis and sub-
The authors considered several aspects of the diag- sequent treatment. Significant interevaluator reli-
nostic process, including patient history, tests or- ability for identification of errors in the diagnostic
dered, interpretation of clinical data, choice and per- process was obtained (p > 0.1). Conclusions: A two-
formance of procedures, injury pattern recognition, tiered evaluation of ED records selected by inconsis-
reasoning, and evaluation. Records that demon- tent initial and final diagnoses can be used reliably
strated errors in the diagnostic process were reeval- to screen for errors in the diagnostic process made by
uated for the same diagnostic process errors by a emergency physicians (EPs). The rate of physician er-
sixth physician. Disagreements regarding suspected ror contributing to a misdiagnosis is very low, sug-
errors in the diagnostic process were settled by dis- gesting that EPs are delivering quality patient care.
cussion. Finally, to determine potential medical con- Key words: errors, medical; diagnostic errors; quality
sequences of the misdiagnosis, one individual re- assurance, health care. ACADEMIC EMERGENCY
viewed the complete medical records of patients MEDICINE 2001; 8:125–130
HE focus of emergency medicine (EM) re- (ED) as one aspect of identifying and reporting
mains delivering high-quality patient care medical errors. An effective method for detecting
and reducing morbidity and mortality. A funda- and identifying reasons for medical errors also
T
mental belief in health care is that providers are
intrinsically motivated to improve patient safety.
However, recent organized analyses of medical er-
rors1 have focused attention on methods of improv-
ing the decisions and practices of health care pro-
viders and the organization of the health care
delivery system. Specifically designed quality as-
surance procedures may be used to access physi-
cian performance in the emergency department
may impact medical-legal consequences and inter-
actions with hospital administrators and insur-
ance agencies.
Methods of evaluating quality assurance and
physician performance in the ED are deficient in
the literature. Many articles discuss medical au-
dits and patient outcomes, but focus primarily on
major trauma systems or primary care facilities
and not the entire spectrum of EM practice.2–5 Two
recent studies documented 3–4% medical errors in
large numbers of patients tracked through the en-
tire health care system from primary care physi-
cian, out-of-hospital and hospital care, and conva-
lescence at home.6,7 In a smaller study, errors in
diagnosis were found to be a major contributor of
mortality in cerebrovascular accidents and pneu-
monia.8
From the Department of Emergency Medicine (MC, JEO,
GCH) and Department of Physiology and Biophysics (JEO),
Wright State University, Dayton, OH; and Premier Health
Care Services, Centerville, OH (JA).
Received May 30, 2000; revision received September 15, 2000;
accepted September 27, 2000. Presented at the ACEP Research
Forum, San Francisco, CA, October 1997, and at the SAEM
Ohio/Midwest Regional Emergency Medicine Research Sym-
posium, Columbus, OH, April 1998.
Address for correspondence and reprints: James E. Olson,
PhD, Department of Emergency Medicine, Cox Institute, 3525
Southern Boulevard, Kettering, OH 45429. Fax: 937-296-4287;
e-mail: james.olson@wright.edu
We sought to use a method that would focus the
scope of assessment on the emergency physician
(EP) diagnostic process. To this end, we created a
chart review process that evaluated areas of diag-

126
MISSED DIAGNOSES
Chellis et al. • DIAGNOSIS ERRORS IN THE ED
of 5,000 admissions from the ED were chosen for
evaluation (Fig. 1). A convenience sampling was
performed by the evaluators who selected days to
examine all records of patients admitted over a 24-
hour period.
Study Protocol. Each medical record was ini-
tially examined by one of five board-certified EP
evaluators to determine whether the primary ED
admitting diagnosis differed from the primary dis-
charge diagnosis. If a discrepancy was found, the
ED record then was reviewed using a set of criteria
to determine whether an error in the diagnostic
process contributed to the inaccurate initial diag-
nosis. These criteria were defined from the basic
tenets of diagnostic process relevant to the practice
of EM.9,10 We examined several aspects of the di-
agnostic process, including history and physical ex-
amination, tests ordered, physician judgment and
reasoning, injury pattern recognition, procedures
performed, and workup (Table 1). These classifi-
cations are not exclusive of each other. In general,
physician errors were scored when, as determined
by review of the medical record, information avail-
able or obtainable at the time of ED presentation
was not properly used or was ignored. Our criteria
developed for this purpose are similar to those
classified by Leape et al. as diagnostic or treatment
errors.11
Because of the subjective nature of this review
process, we used a second tier of evaluation to im-
prove consistency in error reporting. Records that
were found in the initial review to contain errors
in the diagnostic process were evaluated by a sixth
physician using the same criteria. Disagreements
between the first and second evaluators were re-
solved by discussion. Finally, one physician re-
viewed the complete medical record of patients
whose ED charts were scored with an error to de-
termine potential medical consequences of the in-
itial misdiagnosis.
Figure 1. Flow chart of the medical record review pro-
cess. The number of medical records evaluated at each
tier of review is indicated in light gray boxes. Individ-
uals reviewing medical records are indicated by dark
gray boxes.
nosis important for quality assurance and that ex-
amined resulting patient morbidity. Our objective
was to examine the utility of a quality assurance
instrument to define errors in diagnostic processes
made by physicians in the ED. We used a two-
tiered physician evaluation system to review
charts of patients admitted to the hospital from the
ED. When discrepancies between the admitting
and the discharge diagnoses were found, patient
data were examined in detail to determine what,
if any, mistakes were made by the EP in the di-
agnostic process, and whether those mistakes af-
fected patient management and outcome.
Interevaluator reliability using our criteria for
diagnostic process errors was assessed using a se-
lected series of medical records. Forty charts with
no errors in the diagnostic process were randomly
pulled from the 5,000 previously reviewed. To this
pool were added ten that were known to contain
errors in diagnostic process. Using the criteria
shown in Table 1, the five physician reviewers who
participated in the first tier of evaluation indepen-
dently examined these 50 charts unaware of which
ones contained diagnostic errors.
M
ETHODS
Study Design. We performed a retrospective
clinical review of the records of patients admitted
from the ED. The study was approved by all rele-
vant institutional review boards.
Study Setting and Population. This study was
performed in an urban, trauma level 1 hospital
with an annual ED census of more than 70,000.
The site is a community hospital and teaching fa-
cility affiliated with Wright State University Data Analysis. The number of records contain-
School of Medicine and the Integrated Emergency ing diagnostic process errors are expressed as a
Medicine Residency Program of Wright State Uni- percentage of the total number of records. Reli-
versity. Over a six-year period (1992–1998), a total ability was estimated using Cochran’s Q-test.

ACADEMIC EMERGENCY MEDICINE • February 2001, Volume 8, Number 2
127
T
ABLE 1. Criteria Used to Detect Errors in the Diagnostic
RESULTS
Process*
Errors in the history taken or physical examination performed.
This includes errors of omission of key portions of the physical
examination relevant to the final diagnosis or missing the find-
ings or importance of historical factors available at the time of
presentation.
Of the 5,000 charts reviewed, 28 were judged by
both tiers to have errors in the diagnostic process
(0.6%). Five records judged by a first-tier evaluator
to contain errors were eliminated following review
by the second-tier physician. The list of patients
with their admitting and discharge diagnoses are
shown in Table 2. Gastrointestinal bleeding/duo-
denal ulcer was the most common diagnosis
missed (5), followed by digoxin toxicity (3), and
pneumonia (3).
Errors by way of omitting tests or in the interpretation of tests
ordered. These are errors of omission when appropriate tests
that would have aided in making the correct diagnosis are not
ordered, or errors in performance when results of tests ordered
are misinterpreted.
Errors in judgment or reasoning that may lead to a misdirected
evaluation. This occurs when physicians neglect to consider all
possible diagnoses consistent with available information.
Following this review of the 5,000 medical
charts, 18 of the 28 complete inpatient records
were obtained for further review by one of the phy-
sicians to determine what consequences may have
resulted from the initial diagnostic errors. The
other ten records were not reviewed further either
because they could not be located due to inconsis-
tencies in medical record numbers or because per-
tinent files had been lost from the record. Of the
remaining 18 records, only three (17%) were found
to have suffered delay in proper treatment due to
the errors in the diagnostic process made by the
EPs. These treatment delays ranged from 24 hours
Errors in recognition of a predictable pattern of injury. This
includes overlooking medical problems or injuries that are typ-
ically associated with the presenting complaint or illness or are
documented by findings obtained from the physical examina-
tion or history.
Errors due to improperly performed procedures. These errors
of performance are limited to those executed by the physician
that would have directed the evaluation toward the correct di-
agnosis if the procedure had been performed properly.
*Satisfying one criterion was sufficient to document an error
to 16 days. No patient in this group of 16 suffered in the diagnostic process.
a long-term complication, and all were discharged
to home. One patient (#18) died 24 hours after ad-
mission; however, this death was not related to the
error in diagnosis. The final diagnosis of this pa-
tient was made at autopsy. This detailed review
also disclosed that all of the 18 patients reviewed
with errors in the diagnostic process arrived to the
ED by emergency medical services (EMS). The ma-
jority of this group were elders. No errors due to
improperly performed procedures were docu-
mented in this group of patients.
The assessment of interevaluator reliability for
the review of 50 selected charts gave Cochran’s Q
statistic of 4.89 with p > 0.1. Thus, we cannot ex-
clude the null hypothesis that all physician eval-
uators classified the charts similarly.
intestinal bleeding, initially diagnosed as chest
pain. Though both complaints necessitate
a
workup for cardiac etiologies, this review demon-
strated that gastrointestinal etiologies can easily
be missed if not considered by the EP. We found
that incomplete histories and physical examina-
tions, as well as errors in reasoning, led to many
of the identified diagnostic errors. Other diagnoses
that were inappropriately made upon admission,
such as a perforated diverticulum diagnosed as a
pneumomediastinum, resulted from errors in the
workup and thought processes of the physician.
Our audit further confirmed that more common di-
agnoses that can manifest as ambiguous signs and
symptoms, such as urosepsis and digoxin toxicity,
can be easily missed if the appropriate tests are
not ordered and the diagnoses not considered.
Review of inpatient charts for those patients
who were admitted found three of 18 (17%) who
may have suffered complications from the delay in
diagnosis. Of those charts reviewed, the correct di-
agnosis and treatment (assuming the discharge di-
agnosis as the criterion standard) were usually in-
stituted within 24 hours of admission. In many of
these instances, the admitting physicians per-
formed a more complete history and physical ex-
amination, thus eliciting information to enable a
more accurate diagnosis. Our study confirms that
DISCUSSION
In this investigation, two levels of physicians re-
viewing 5,000 charts of patients admitted from a
single ED identified only 28 (0.6%) involving sig-
nificant errors in the diagnostic process. This in-
cludes only physician errors that contributed to an
error in diagnosis. Conceivably errors may have
been made in cases that received the correct di-
agnosis by the physician. The magnitude of the er-
ror rate for this limited population of hospitalized
patients is compatible with previous error analyses
of the entire health care delivery system, which significant consequences to patient care may result
documented error rates in 3–4% of patients.6,7
from information missed during the initial ED
The most common missed diagnosis was gastro- evaluation.

128
MISSED DIAGNOSES
Chellis et al. • DIAGNOSIS ERRORS IN THE ED
Criteria for Errors
TABLE 2. Diagnostic Error Summary
Admission Diagnosis
Discharge Diagnosis
Patient 1
Patient 2
Patient 3
Patient 4
Abdominal pain
Back pain
*Bronchiectasis
Gastrointestinal bleeding, gastritis
Multiple myeloma, fractures
Ureteral colic
N/A‡
N/A
History and physical exam
N/A
Congestive heart failure, urinary Pneumonia
tract infection
Patient 5
Patient 6
Patient 7
Patient 8
Patient 9
*Chest pain, rule out myocardial Duodenal ulcer
infarction
Chest pain, rule out myocardial Dissecting thoracic aneurysm
infarction
Chest pain, rule out myocardial Ulcer disease, gastrointestinal bleeding History and physical exam
infarction
Physical exam
History and physical exam
Chest pain, rule out myocardial Congestive heart failure, digoxin toxic- N/A
infarction
ity
Chest pain, angina
Duodenal ulcer, gastrointestinal bleed- History and physical exam
ing
Patient 10 Chronic obstructive pulmonary Pneumonia
disease exacerbation
History and physical exam
Patient 11 Confusion
Hypocalcemia, hyperglycemia
N/A
Patient 12 Dehydration, acute renal failure
Patient 13 Dizziness, anemia
Patient 14 Elbow, face abrasion
Digoxin toxicity
Gastrointestinal bleeding
Elbow, face abrasion and sternal frac- Physical exam, omitted test
ture, rule out cardiac contusion
Omitted tests, reasoning
N/A
Patient 15 Febrile illness
Pseudomembranous colitis
Omitted tests, predictable pattern of
injury
Patient 16 Gastroenteritis, chronic renal
failure
Patient 17 *Hypotension
Patient 18 †Mental status change
Patient 19 Mental status change
Patient 20 Pelvic fracture
Patient 21 Pneumomediastinum
Patient 22 Pulmonary edema
Patient 23 Scalp, face lacerations
Pancreatitis
History and physical exam, omitted
test, predictable pattern of injury
Omitted test, reasoning
Reasoning
Urosepsis
Pulmonary embolus, death
Urosepsis
Bronchogenic carcinoma
Perforated diverticulitis
Atrial fibrillation, theophylline toxicity N/A
N/A
Omitted tests, misinterpreted test
History and physical exam
Scalp, face lacerations, and pneumotho- N/A
rax
Patient 24 Shortness of breath
Patient 25 Vomiting
Congestive heart failure, anemia, leu- Misinterpreted test, physical exam
kemia
Pancreatitis
History and physical exam, omitted
test, predictable pattern of injury
Reasoning, workup, predictable pat-
tern of injury
Patient 26 Weakness, anemia
Gastrointestinal bleeding
Patient 27 Weakness
Patient 28 Weight loss
Pneumonia
Digoxin toxicity
Misinterpreted test
N/A
*Denotes three patients whose delay in diagnosis may have resulted in complications for the patient.
†Denotes one patient who died 24 hours after admission.
‡N/A denotes medical records unavailable for review of outcome following diagnostic errors.
The number of subsequent complications was specific presentations that could subsequently be
low at our institution partly because of reassess- used to further physician education and improve
ments by in-hospital resident physicians and staff. the quality of patient care. These presentations
The risk of not identifying the correct diagnoses then could be correlated with other patient data.
and instituting proper treatment may be higher in Data from our sample suggest that elder patients
nonteaching hospitals. This problem may be exac- or those patients arriving to the ED by EMS may
erbated if attending physicians admit patients by be considered high-risk groups for diagnostic er-
telephone and do not perform a diagnostic evalu- rors. In particular, geriatric patients require a
ation until the next day. The possibility of compli- more extensive evaluation because this population
cations from errors in diagnosis is even larger for can little tolerate delays in proper treatment as a
those patients whom we discharge to home. Thus, result of misdiagnoses. Deficiencies were found in
future studies of errors in the diagnostic process documentation of complete physical examinations,
might include a review of patients discharged from particularly in rectal exams of patients with ab-
the ED.
dominal pain. In addition, documentation and in-
By documenting the most common errors made terpretation of laboratory results, and processes of
by EPs in the diagnostic process, we determined medical decision making, were found lacking.

ACADEMIC EMERGENCY MEDICINE • February 2001, Volume 8, Number 2
129
To the best of our knowledge, no previous report study. For example, patients with final diagnoses
has documented errors in the diagnostic process of pulmonary embolus or congestive heart failure
made by EPs regarding their admitted patients; were initially diagnosed as having congestive heart
however, a variety of areas of medical quality as- failure or cholecystitis, respectively.
surance including diagnostic errors have previ-
Guly14,15 created a numerical scale to measure
ously been evaluated in clinical settings other than the severity of diagnostic errors in EDs. The ‘‘mis-
the ED. Three studies from the same institution diagnosis severity score’’ (MSS) combined scores
assessed quality assurance in trauma systems.2–4 that evaluated treatments that should have been
In these studies, trauma nurse coordinators and given and the disposition of the patient. Although
trauma directors preselected patients to be evalu- the MSS was used to describe only the severity of
ated from their own initial screening. A committee trauma diagnostic errors, the author believed it
composed of various members of trauma teams and could be generalized to all areas. Other scales have
other related institutions then reviewed the pa- been designed for specific diagnostic problems,
tients’ records for complications, delays, and such as radiographic diagnoses.16
deaths. Morbidity and mortality of trauma pa-
Our main goal was to evaluate EPs’ performance
tients were analyzed using many different criteria, by determining their accuracy in the diagnostic pro-
including errors in diagnoses. A study done by cess and identifying reasons for diagnostic errors.
Fischer et al.5 looked at errors in diagnosis as one Our study used a broad population of all patients
aspect of an overall audit of adverse events from a admitted to the hospital from the ED. We used a
primary care setting. Two reviewers subjectively specific set of criteria to identify reasons for the di-
categorized the errors into one of four types. Dis- agnostic errors. Although a high degree of interob-
crepancies were resolved by two of the study au- server reliability was observed for detection and
thors, who then further evaluated each case using characterization of diagnostic process errors, our
the individual reviewers’ comments. The authors use of a secondary review by one individual pro-
reviewed hospital records in only six particularly vided a greater degree of uniformity. While our
difficult cases. These reviews focused record anal- method will not identify systematic errors in health
ysis on patients who may have had adverse events care delivery, the method may be broadly applied to
associated with their medical care. In the present ascertain the level of diagnostic process errors made
study, we selected records of all patients who were by physicians in a variety of clinical settings.
determined to have diagnostic errors, whether or
not a significant medical consequence resulted.
Thus, a greater range of physician errors may be
included in our data.
L
IMITATIONS AND FUTURE QUESTIONS
Because we reviewed only patients admitted from
the ED, we may have missed errors made during
the evaluation of patients who were discharged. A
vastly redesigned study would be necessary to
study errors in this population due to the lack of a
definitive diagnosis and inherent problems with
patient follow-up. In this study we assumed the
final diagnoses made upon hospital discharge were
accurate, and thus, we used this as our criterion
standard with which we compared the ED diag-
noses. No attempt was made to verify this diag-
nosis. We would not have selected data from pa-
tients if both diagnoses were in error. Our inability
to perform a detailed review of all 28 patient
records identified to contain errors in the diagnos-
tic process limits the power of our conclusions.
However, we anticipate misfiling of data in the
medical records occurred randomly and did not in-
troduce a systematic bias to our results. Finally,
despite our use of defined criteria to select errors
in the diagnostic process, and the high degree of
interobserver reliability, there exists some ob-
server-to-observer variability in the choice of pa-
tient records with errors in diagnostic process.
Kirch and Schafii12 studied rates of misdiag-
noses over four decades with a hypothesis that ad-
vances in diagnostic procedures have reduced mis-
diagnosis rates. Using a two-tiered review process,
these authors identified misdiagnosis using infor-
mation on death certificates as the criterion stan-
dard. They concluded that despite technological
advances, the rate of misdiagnosis has remained
constant. Similar to our conclusions, they also
found that accurate patient histories and physical
examinations remain the most important infor-
mation for reaching an accurate diagnosis.
O’Connor et al.13 used two EPs, one consulting
physician and one consulting surgeon, to evaluate
400 charts for mistakes in diagnoses of admitted
patients. While no specific criteria were used, er-
rors were subjectively classified as minor, moder-
ate, and major. Their misdiagnosis rates (10% and
12%) were much higher than those reported in the
present study. Differences in characteristics of the
study site and diagnostic tests available to their
EPs may have contributed to the higher rate of
misdiagnosis. Despite these differences in study
design, these authors note diagnostic difficulties in Thus, some charts may have been dismissed that
similar categories to those found in the present actually contained errors.

130
MISSED DIAGNOSES
Chellis et al. • DIAGNOSIS ERRORS IN THE ED
audit committee: composition, cost, and results. J Trauma.
1987; 27:866–72.
5. Fischer G, Fetters MD, Munro AP, Goldman, EB. Adverse
events in primary care identified from a risk-management da-
tabase. J Fam Pract. 1997; 45:40–6.
Future studies of errors in the diagnostic pro-
cess might include a review of patients discharged
from the ED.
6. Brennan TA, Leape LL, Laird NM, et al. Incidence of ad-
verse events and negligence in hospitalized patients: results of
the Harvard Medical Practice Study. N Engl J Med. 1991; 324:
370–6.
7. Thomas EJ, Studdert DM, Newhouse JP, et al. Costs of
medical injuries in Utah and Colorado. Inquiry. 1999; 36:255–
64.
8. Dubois RW, Brook RH. Preventable deaths: who, how often
and why? Ann Intern Med. 1988; 109:582–9.
9. Hamilton GC. Introduction to emergency medicine. In:
Hamilton GC, Sanders AB, Strange GR, Trott AT (eds). Emer-
gency Medicine: An Approach to Clinical Problem-Solving.
Philadelphia: W. B. Saunders, 1991, pp 3–18.
10. Dailey RH. Approach to the patient in the emergency de-
partment. In: Rosen P (ed). Emergency Medicine: Concepts and
Clinical Practice. St. Louis: Mosby, 1992, pp 22–37.
11. Leape L, Lawthers AG, Brennan TA, Johnson WG. Pre-
venting medical injury. Qual Rev Bull. 1993; 19:144–9.
12. Kirch W, Schafii C. Misdiagnosis at a university hospital
in 4 medical eras, report of 400 cases. Medicine. 1996; 75:29–
40.
13. O’Connor PM, Dowey KE, Bell PM, Irwin ST, Dearden CH.
Unnecessary delays in accident and emergency departments:
do medical and surgical senior house officers need to vet ad-
missions? J Accid Emerg Med. 1995; 12:251–4.
14. Guly HR. A scale for measuring the severity and diagnostic
errors in accident and emergency departments. J Accid Emerg
Med. 1997; 14:290–2.
C
ONCLUSIONS
Our survey suggests that EPs make diagnostic
process errors in only 0.6% of their patients. In a
high-volume ED similar to the site of this study, a
two-tiered chart audit based on ED and final di-
agnoses can be a useful instrument for ED quality
assurance. Valuable information can be obtained to
improve both physician performance and subse-
quent patient care. This method can be used to
evaluate and compare other EP practices.
References
1. Kohn LT, Corrigan JM, Donaldson MS (eds). To Err Is Hu-
man: Building a Safer Health System. Report of the Institute
of Medicine. Washington, DC: National Academy Press, 2000,
p 312.
2. Davis JW, Hoyt DB, McArdle MS, et al. An analysis of er-
rors causing morbidity and mortality in a trauma system: a
guide for quality improvement. J Trauma. 1992; 32:660–6.
3. Davis JW, Hoyt DB, McArdle MS, Mackersie RC, Shackford
SR, Eastman AB. The significance of critical care errors in
causing preventable death in trauma patients in a trauma sys-
tem. J Trauma. 1991; 31:813–9.
15. Guly HR. The misdiagnosis severity score and doctors’ per-
ception of the severity of diagnostic errors. J Accid Emerg Med.
1997; 14:293–4.
4. Shackford SR, Hollingsworth-Fridlund P, McArdle M, East- 16. Wardrope J, Chennells PM. Should all casualty radio-
man AB. Assuring quality in a trauma system—the medical graphs be reviewed? BMJ. 1985; 290:1638–40.
ⅷ
Where to Find AEM Instructions for Authors
For complete instructions for authors, see the January or July issue of
Academic Emergency Medicine; visit the SAEM web site at www.saem.org/
inform/autinstr.htm; or contact SAEM via e-mail at aem@saem.org, via
phone at 517-485-5484, or via fax at 517-485-0801.