Psychiatric Comorbidity and Mortality After Acute

Myocardial Infarction
Thad E. Abrams, MD, MS; Mary Vaughan-Sarrazin, PhD; Gary E. Rosenthal, MD

Background—Prior studies of the impact of psychiatric comorbidity on outcomes after acute myocardial infarction (AMI)
have frequently relied on inpatient secondary diagnosis codes. This study compared associations between psychiatric
comorbidity and AMI outcomes that were derived using secondary diagnosis codes and codes captured from prior
outpatient encounters.
Methods and Results—Retrospective cohort study analyzing 21 745 patients admitted in 2004 to 2006 to Veterans Health
Administration hospitals with AMI using administrative data. Psychiatric comorbidity was identified using (1)
secondary inpatient diagnosis codes from the index hospitalization and (2) diagnoses from prior outpatient encounters.
Outcomes included 30- and 365-day mortality and the receipt of coronary revascularization within 30 days of admission.
Generalized estimating equations and Cox proportional hazards were used to adjust mortality and receipt of
revascularization for demographic and clinical variables. Psychiatric disorders were identified in 2285 (10%) patients
from inpatient secondary diagnosis codes and 5225 (24%) patients from prior outpatient codes. Patients with psychiatric
comorbidity had higher adjusted 30- and 365-day mortality, based on outpatient codes (odds ratios, 1.19 [95% CI, 1.09
to 1.30] and 1.12 [95% CI, 1.03 to 1.22], respectively), but similar mortality based on inpatient codes (odds ratios, 0.89
[95% CI, 0.69 to 1.01] and 0.93 [95% CI, 0.82 to 1.06], respectively). In contrast, patients with psychiatric comorbidity
had lower receipt of coronary revascularization based on outpatient codes (hazard ratio, 0.92; [95% CI, 0.85 to 0.99],
but similar receipt based on inpatient codes (hazard ratio, 1.00 [95% CI, 0.91 to 1.10]).
Conclusions—Inpatient secondary diagnosis codes identified fewer patients with psychiatric comorbidity than prior
outpatient codes. Moreover, associations with AMI outcomes differed for the 2 approaches. These findings raise
potential concerns about the validity and reliability of psychiatric inpatient secondary diagnosis in estimating the impact
of psychiatric comorbidities on AMI outcomes and in developing risk-adjustment models. (Circ Cardiovasc Qual
Downloaded from http://ahajournals.org by on November 9, 2023

Outcomes. 2009;2:213-220.)
Key Words: comorbidity 䡲 myocardial infarction 䡲 hospital mortality

A number of studies over the past 2 decades have

examined associations between the presence of psychi-
atric conditions and adverse outcomes from acute coronary
syndromes. These analyses include prospective trials,1–5 sys-
tematic reviews,6 –9 and observational studies.10 –20 Many of
the latter observational analyses have relied on the use of
large administrative (ie, claims) databases, either wholly or in
part, as a means to identify patients with psychiatric comor-
bidities. Unfortunately, findings across these studies have
revealed inconsistencies. For example, some studies have
shown that patients with a wide spectrum of psychiatric
comorbidities have a higher risk of death after an acute
myocardial infarction (AMI), are less likely to undergo
coronary revascularization, and are subject to lower quality of
care, as assessed by a number of process measures (eg, use of
appropriate medications).14 –18 However, other studies have
failed to detect such relationships.19,20
One potential explanation for the inconsistencies in the
associations between AMI-related hospital outcomes and
psychiatric comorbidities may be the methodological ap-
proach that has been used to identify psychiatric comorbidity
from administrative databases. Therefore, we completed this
study to directly evaluate how different approaches of iden-
tifying psychiatric comorbidities may impact AMI-related
hospital outcomes for a range of psychiatric conditions.
We studied a large cohort of patients admitted to Veterans
Health Administration (VHA) facilities and identified psychi-
atric comorbidity using 2 different approaches that have been
used in prior work—secondary diagnosis codes captured
during the index hospitalization and diagnosis codes captured
during prior outpatient encounters. Our objectives were to (1)
compare the rates of psychiatric diseases identified by 2
approaches and (2) determine whether associations between

Received October 17, 2008; accepted January 27, 2009.

From the Center for Research in the Implementation of Innovative Strategies in Practice (T.E.A., M.V.-S., G.E.R.), Iowa City VA Healthcare System,
and the Department of Internal Medicine (T.E.A., G.E.R.), University of Iowa, Iowa City, Iowa.
The online-only Data Supplement can be found at http://circoutcomes.ahajournals.org/cgi/content/full/10.1161/CIRCOUTCOMES.108.829143/DC1.
Correspondence to Thad E. Abrams, MD, MS, Iowa City VA Medical Center, 601 Hwy 6 West, Mailstop 152, Iowa City, IA 52246-2208. E-mail
thad-abrams@uiowa.edu
© 2009 American Heart Association, Inc.
Circ Cardiovasc Qual Outcomes is available at http://circoutcomes.ahajournals.org DOI: 10.1161/CIRCOUTCOMES.108.829143

213
 214 Circ Cardiovasc Qual Outcomes May 2009
psychiatric comorbidity and AMI-related hospital outcomes
differed depending on which method of identifying psychi-
atric comorbidity was used. Based on these findings, we
hoped to provide guidance for future studies on optimal
approaches for identifying psychiatric comorbidity using
administrative data.
Methods
Data Sources
Data were derived from 4 VHA sources: (1) the Patient Treatment
File; (2) the Outpatient Care Files; (3) the Decision Support System
laboratory files; and (4) the Vital Status File. The Patient Treatment
File contains data on all hospitalizations in VHA hospital facilities
nationally. Data elements include demographics, socioeconomic
status, residential zip code, presence of disabilities related to military
service; primary and secondary diagnoses and procedures, as defined
by International Classification of Diseases, 9th Revision, Clinical
Modification (ICD-9-CM) codes; admission sources (eg, transfer
from another hospital, emergency room); admission and discharge
times and dates; discharge destination and vital status. The Outpa-
tient Care Files include administrative data on all outpatient encoun-
ters. Data elements include: dates of visits; type of clinic (eg, primary
care, mental health); and ICD-9-CM diagnoses and procedures for
each encounter. The Decision Support System laboratory files
contain the results of selected all laboratory tests performed on an
inpatient or outpatient basis. The Vital Status File was used to obtain
dates of death of patients after hospital discharge; the validity of this
file, with respect to the National Death Index, has been previously
demonstrated.21,22 Each of the databases includes unique identifiers
that allow merging of patient-level information across the databases.
Patient Population
Downloaded from http://ahajournals.org by on November 9, 2023
The Patient Treatment File was used to identify 29 440 consecutive
patients who were admitted with a principle diagnosis of AMI
(ICD-9-CM 410) during the period October 1, 2003 to September 30,
2006 to all 144 VHA hospitals nationwide. For patients with more
than 1 admission for AMI during the study period, we excluded
(n⫽4561) those admissions that represented repeat AMI admissions
during the study period. We then excluded patients without any VHA
outpatient visits during months 13 to 24 before admission (n⫽3134),
leaving a final study sample of 21 745 patients. We excluded patients
without VHA visits in months 13 to 24 before their admission to
ensure that patients in our sample were eligible for VHA care during
the entire 12-month period before admission. Because the VHA does
not maintain files accessible to investigators that identify individuals
who are eligible for VHA care, this additional step ensured that
individuals included in the analysis would have been eligible for
VHA outpatient services a full 12 months before the date of
admission. Separate analyses demonstrated that the excluded popu-
lation had lower rates observed mortality and lower identification
rates of psychiatric comorbidity.
Study Variables
Study end points included both 30- and 365-day mortality after
admission and the receipt of coronary revascularization within 30
days of admission. ICD-9-CM procedure codes from the index
hospitalization and from subsequent encounters were used to identify
the receipt of revascularization by either percutaneous coronary
intervention (PCI; 36.00 to 36.09) or coronary artery bypass graft
surgery (CABG; 36.10 to 36.19).
Primary independent variables included the presence of 5 psychi-
atric comorbid conditions, considered separately and in aggregate.
These conditions were identified using 2 different methodological
approaches: (1) ICD-9-CM secondary diagnosis codes from the
index hospitalization; and (2) ICD-9-CM diagnosis codes captured
during 1 or more outpatient encounters during the 12 months before
admission. For both approaches, 5 major categories of psychiatric
conditions that have been examined in prior AMI outcomes studies
were identified: (1) depressive disorders (ICD-9-CM 296.20 to 36,
311, 300.4); (2) anxiety disorders (300.00 to 300.02, 293.84, 309.28,
309.21 to 309.23); (3) posttraumatic stress disorder (PTSD)
(309.81); (4) bipolar disorders (296.00 to 296.06, 296.40 to 296.89);
or (5) psychotic disorders, including schizophrenia and schizoaffec-
tive disorders (295). Other psychiatric conditions (eg, substance
abuse disorders, personality disorders) were not examined, given the
frequent overlap of these conditions with the 5 included conditions.
Categorizations were based on all codes captured during the relevant
encounters; thus, patients could be categorized with more than 1
condition.
Other variables that were used to describe the study population
and to adjust for differences in the risk of mortality included: age;
race (categorized as white, black, Hispanic, or missing); gender;
marital status; VHA eligibility criterion (presence of a service-
connected disability or indigent); comorbid medical conditions that
are unlikely to reflect hospital complications, as defined using
ICD-9-CM codes;23 mechanical ventilation on day of admission;
location of infarction;24 and results of 9 selected laboratory tests
(serum creatinine, blood urea nitrogen, albumin, total bilirubin,
glucose, sodium, and troponin [both subtypes I and T], white blood
cell count, and hemoglobin). All laboratory tests were captured
within a 48-hour window surrounding the admission time. We chose
the 48-hour window to optimize availability of laboratory data and to
minimize laboratory scores that may reflect illness developing after
the admission.
For each laboratory value (excluding troponin), we selected the most
abnormal value, based on weights used in the APACHE (Acute
Physiology and Chronic Disease Health Evaluation) III methodology.25
Weights associated with each variable were then summed to create
an overall laboratory severity score. Missing laboratory values were
considered to be normal, consistent with the APACHE III method-
ology. A minimum of 4 nonmissing laboratory values were required
to calculate a laboratory severity score. Patients with scores of 0 and
for whom more than 4 values were missing were considered to have
missing laboratory severity scores. Such patients had a higher
observed mortality than other patients with scores of 0, and were thus
examined in the analyses as a separate indicator variable. Model
discrimination was improved by using a summary score rather than
entering the laboratory values individually.
For troponin, we selected the most abnormal subtype I and T value
for each patient. We then standardized values for each subtype by
creating percentile ranks (1–100). For patients with values for both
subtypes, we used the subtype value with the highest percentile rank.
Percentile ranks were then classified into 4 categories of increasing
risk, based on empirical associations with 30-day morality. Patients
with missing values for both troponin I and T were examined
separately in analyses.
Lastly, the use of coronary catheterization within 30 days of
admission was identified using ICD-9-CM codes (37.22 to 37.23, 29,
88.53 to 88.57).
Analysis
The analysis consisted of several steps. First, agreement between
psychiatric comorbidities measured by inpatient secondary diagnosis
codes and outpatient diagnosis codes were measured using the ␬
statistic. Second, bivariate relationships between the presence of
psychiatric comorbidity (as determined by both inpatient and outpa-
tient diagnosis codes) and other demographic and clinical character-
istics, mortality, and receipt of catheterization and revascularization
were determined using the ␹2 or t tests. Third, other demographic and
clinical factors that were associated (P⬍0.05) with the risk of 30-day
or 365-day mortality and receipt of revascularization within 30 days
were identified. Fourth, these variables (with the exception of
psychiatric comorbidities) were then entered into stepwise multivari-
able regression analyses to identify independent (P⬍0.01) predictors
of mortality, catheterization, and revascularization. (Variables in-
cluded in the risk-adjustment models for 30- and 365-day mortality
and their associated odds ratios are included in the supplemental
materials.) These analyses used logistic regression to model mortal-
ity and proportional hazards to model receipt of coronary catheter-

20%
Inpatient Codes Outpatient Codes
15.1%
15%
10%
7.0%
6.9%
5%
5.0%
1.9%
2.6%
2.0%
0.9%
0%
Depression Anxiety PTSD Bipolar
Figure 1. Rates of individual psychiatric illnesses identified by
secondary diagnosis codes from the index inpatient admission
and diagnosis codes from prior outpatient visits.
ization and revascularization, censoring patients who died within 30
days of admission. In the multivariable analyses, laboratory severity
scores and troponin percentile ranks were categorized into discrete
ranges that maximized associations with mortality and analyzed as
separate (n-1) indicator variables; patients with missing values were
analyzed as separate indicator variables.
Next, variables from the multivariable risk-adjustment models
were entered in separate generalized estimating equations (GEE) or
proportional hazards models that included indicator variables either
for psychiatric comorbidity for all conditions in aggregate or for
individual conditions. The GEE models used an exchangeable
working correlation matrix in accounting for the clustering of
patients within hospitals; proportional hazards models used robust
sandwich covariance matrix estimators to account for clustering.
Analyses to test the proportional hazards assumption found no
interactions between terms between psychiatric comorbidity and the
time-to-event. Coefficients associated with the psychiatric comor-
Downloaded from http://ahajournals.org by on November 9, 2023
bidity indicator variables were used to estimate adjusted odds of
death or the hazard revascularization. Separate analyses were con-
ducted for psychiatric comorbidities identified using inpatient sec-
ondary diagnoses and outpatient diagnoses. Models examining
mortality did not include the use of revascularization, given the
potential endogeneity of revascularization and mortality.
Lastly, analyses were conducted in which patients with psychiatric
comorbidity were classified into 3 exclusive groups depending on
whether the comorbidity was identified by (1) both outpatient and
inpatient diagnoses, (2) outpatient diagnosis alone, and (3) inpatient
diagnosis alone. The risk of death in these patients was compared to
patients without psychiatric comorbidity identified by either method.
The authors had full access to and take full responsibility for the
integrity of the data. All authors have read and agree to the
manuscript as written. All analyses were conducted using SAS
statistical software version 9.1. The study was approved by the both
the University of Iowa Institutional Review Board and the Research
and Development Committee at the Iowa City VHA Medical Center.
Results
Overall, study subjects had a mean age (SD) of 68.5 (11.6)
years and 98% were male. Sixty-three percent of patients
were white and 12% were black; race was missing in 24%.
Psychiatric comorbidity was identified in 10% (n⫽2285) of
patients using inpatient secondary diagnosis codes and 24%
(n⫽5225) of patients using prior outpatient diagnosis codes.
Rates of identifying individual psychiatric conditions were
also higher using outpatient codes than inpatient codes for
each of the 5 conditions (Figure 1).
Patients with psychiatric comorbidity, as identified by
inpatient or outpatient diagnoses were younger, had lower
mean laboratory severity scores, were less likely to be
married and to have congestive heart failure, but were more
Abrams et al Psychiatric Disease and Mortality After AMI 215
likely to have chronic obstructive pulmonary disease (Table
1). Mean troponin I and T values were similar in patients with
and without psychiatric comorbidity, although the mean
predicted probabilities of 30-day and 365-day death were
lower in patients with psychiatric comorbidity identified by
either approach.
The agreement between the identification of patients with
psychiatric comorbidity, as identified by inpatient diagnosis
codes and outpatient diagnosis codes was fair26 (␬⫽0.34
1.8%
1.4%
[P⬍0.001]). Of the 5225 patients identified by outpatient
diagnosis codes, 31% (n⫽1625) also were identified by
Psychosis
inpatient diagnosis codes, whereas of the 2287 patients
identified by inpatient diagnosis codes 71% (n⫽1625) also
had an outpatient diagnosis code. ␬ statistics for identifying
individual conditions were 0.22 (P⬍0.001) for depression,
0.15 (P⬍0.001) for anxiety, 0.40 (P⬍0.001) for PTSD, 0.44
(P⬍0.001) for bipolar disorder, and 0.56 (P⬍0.001) for
psychosis.
Patients with psychiatric comorbidity identified by inpa-
tient diagnosis codes had lower unadjusted rates for both
30-day (6.4% versus 12.2% [P⫽0.001]) and 365-day (16.3%
versus 26.7% [P⬍0.001]) mortality than patients without
psychiatric comorbidity. Patients with psychiatric comorbid-
ity identified by outpatient diagnosis also had lower unad-
justed rates for 365-day (23.4% versus 26.3% [P⬍0.001])
than patients without psychiatric comorbidity, but similar
unadjusted 30-day (10.9% versus 11.9% [P⫽0.05]) mortality.
The differences between the associations with mortality
based on inpatient and outpatient diagnoses were generally
consistent in analyses of individual psychiatric conditions
(Table 2).
In GEE analyses, the adjusted odds of death were similar in
patients identified by inpatient diagnoses for both 30-day
(odds ratio [OR], 0.89; 95% CI, 0.69 to 1.01; P⫽0.08) and
365-day (OR, 0.93; 95% CI, 0.82 to 1.06; P⫽0.29), but were
higher among patients identified by outpatient diagnoses for
both 30-day (OR, 1.19; 95% CI, 1.09 to 1.30; P⬍0.001) and
365-day (OR, 1.12; 95% CI, 1.03 to 1.22; P⫽0.007) mortal-
ity. These associations were generally consistent in analy-
ses of individual conditions, with the exception of the
higher risk associated with anxiety disorders as identified
by inpatient diagnoses (Figure 2). Additional analyses
found that the adjusted odds of death for patients who were
identified by either inpatient or outpatient diagnoses was
higher for both 30-day (OR, 1.16; 95% CI, 1.06 to 1.27
[P⫽0.002]) and 365-day (OR, 1.11; 95% CI, 1.02 to 1.21
[P⫽0.01]) mortality.
To further understand the differences in mortality between
outpatient and inpatient diagnoses, separate analyses exam-
ined associations in patients who were identified as having
psychiatric comorbidities by both approaches or by one of the
approaches but not the other (Table 3). These analyses found
the highest risks of death in patients with psychiatric comor-
bidity identified only by outpatient diagnoses. In contrast,
patients with psychiatric comorbidity identified by both
outpatient and inpatient diagnoses had similar mortality as
patients without inpatient or outpatient diagnoses.
The receipt of cardiac catheterization within 30 days of
admission was higher in patients with psychiatric comor-
 216 Circ Cardiovasc Qual Outcomes May 2009

Table 1. Characteristics of Patients With and Without Psychiatric Comorbidity as Identified by Outpatient and Inpatient Diagnoses
Mental Illness Identified by Outpatient Diagnosis Mental Illness Identified by Inpatient Diagnosis

Yes No Yes No
(n⫽5225) (n⫽16 520) P (n⫽2287) (n⫽19 458) P
Age
Mean⫾SD 65.3⫾11.7 69.3⫾11.7 ⬍0.001 62.2⫾11.2 69.0⫾11.5 ⬍0.001
Gender, n (%)
Male 5088 (97.4) 16 263 (98.4) ⬍0.001 2219 (97.1) 19 132 (98.3) ⬍0.001
Female 137 (2.6) 257 (1.6) 66 (2.9) 328 (1.7)
Race, n (%)
White 3388 (64.8) 10 322 (62.5) 0.01 1446 (63.3) 12 264 (63.0) 0.81
Black 539 (10.3) 2088 (12.6) ⬍0.001 265 (11.6) 2362 (12.1) 0.45
Hispanic 53 (1.0) 202 (1.2) 0.22 16 (0.7) 239 (1.2) 0.03
Missing 1223 (23.4) 3902 (23.6) 0.75 547 (23.9) 4578 (23.5) 0.66
Married, n (%) 2503 (47.9) 8604 (52.1) ⬍0.001 963 (42.1) 10 144 (52.1) ⬍0.001
Source of admission, n (%) 111 (2.1) 326 (2.0) 0.50 48 (2.1) 389 (2.0) 0.75
Skilled care facility
Transfer from acute care hospital 519 (9.9) 1735 (10.5) 0.24 235 (10.3) 2019 (10.4) 0.89
Medical comorbidities, n (%)
CHF† 1284 (24.6) 4613 (27.9) ⬍0.001 397 (17.4) 5500 (28.3) ⬍0.001
Diabetes 1962 (37.6) 6258 (37.9) 0.67 762 (33.4) 7458 (38.3) ⬍0.001
Chronic obstructive lung disease 1223 (23.4) 3437 (20.8) ⬍0.001 513 (22.5) 4147 (21.3) 0.21
Hypertension† 2977 (57.0) 9274 (56.1) 0.29 1409 (61.6) 10 842 (55.7) ⬍0.001
Renal failure† 513 (9.8) 1993 (12.1) ⬍0.001 154 (6.7) 2352 (12.1) ⬍0.001
Fluid/electrolyte disorder† 543 (10.4) 1757 (10.6) 0.62 170 (7.4) 2130 (11.0) ⬍0.001
⬍0.001
Downloaded from http://ahajournals.org by on November 9, 2023

Obesity 270 (5.2) 703 (4.3) 0.01 142 (6.2) 831 (4.3)
Laboratory Severity Score,* Mean⫾SD 12.0⫾8.3 13.2⫾8.6 ⬍0.001 10.1⫾7.6 13.2⫾8.6 ⬍0.001
Troponin I
Mean⫾SD 20.5⫾62.2 20.4⫾70.0 0.95 22.3⫾73.1 20.2⫾68.0 0.34
Troponin T
Mean⫾SD 1.6⫾3.3 2.2⫾9.5 0.13 2.7⫾18.0 2.0⫾6.7 0.19
Mean probability of 30-day mortality, %‡ 9.8⫾11.5 12.6⫾13.0 ⬍0.001 7.2⫾9.4 12.4⫾13.0 ⬍0.001
Mean probability of 365-day mortality, %‡ 22.9⫾19.5 26.6⫾18.8 ⬍0.001 17.1⫾15.9 26.5⫾19.6 ⬍0.001
CHF indicates congestive heart failure.
*Values were missing for Laboratory Severity Scores in 1693 patients, troponin I in 7649 patients, and troponin T in 18 098 patients (4529 were missing both
troponin I and T values).
†Medical comorbidities included in final risk adjusted models.
‡Calculated from risk-adjusted models that include all demographical and clinical covariates.

bidity, based on inpatient secondary diagnosis codes
(52.6% versus 46.6%; P⬍0.001) but was similar in pa-
tients with and without psychiatric comorbidity, based on
prior outpatient codes (47.5% versus 47.1%; P⫽0.59).
Similarly, receipt of coronary revascularization with either
PCI or CABG within 30 days of admission was higher in
patients with psychiatric comorbidity, based on inpatient
secondary diagnosis codes (28.0% versus 22.7%; P⬍0.001)
but was similar based on prior outpatient codes (23.9% versus
23.1%; P⫽0.26).

Table 2. Unadjusted 30-Day and 365-Day Mortality Rates in Patients With and Without Individual Psychiatric Disorders, as
Identified by Prior Outpatient Diagnoses and Inpatient Secondary Diagnoses
Depression Anxiety PTSD Bipolar Psychosis

Dependent Variable Yes No P Yes No P Yes No P Yes No P Yes No P

30-day mortality
Outpatient diagnosis, % 11.1 11.7 0.36 12.7 11.5 0.17 6.7 11.9 ⬍0.001 7.7 11.7 0.01 15.2 11.5 0.03
Inpatient diagnosis, % 5.8 11.9 ⬍0.001 10.3 11.6 0.41 3.5 11.8 ⬍0.001 1.6 11.7 ⬍0.001 8.6 11.7 0.10
365-day mortality
Outpatient diagnosis, % 24.4 25.3 0.09 26.5 25.5 0.39 15.2 26.3 ⬍0.001 16.9 25.8 ⬍0.001 29.1 25.5 0.11
Inpatient diagnosis, % 16.0 26.1 ⬍0.001 18.4 25.7 ⬍0.001 8.5 26.0 ⬍0.001 10.8 25.7 ⬍0.001 23.2 25.6 0.34
 Abrams et al
Figure 2. Adjusted odds of 30-day mortality for individual psy-
chiatric disorders as identified by other either secondary inpa-
tient diagnosis codes or prior outpatient diagnosis codes.
In proportional hazards regression analyses, adjusting for
age, illness severity, and comorbidity, the receipt of revascu-
larization was lower in patients with psychiatric illness
identified by outpatient diagnoses (hazard ratio [HR], 0.92;
95% CI, 0.85 to 0.98; P⫽0.02), but was similar among
patients with inpatient psychiatric diagnoses (HR, 1.00; 95%
CI, 0.92 to 1.09; P⫽0.91). Lastly, in analyses restricted to
patients who underwent cardiac catheterization, the receipt of
revascularization was similar in patients with psychiatric
Downloaded from http://ahajournals.org by on November 9, 2023
comorbidity, based on both outpatient (HR, 0.99; 95% CI,
0.92 to 1.07; P⫽0.81) and inpatient (HR, 1.06; 95% CI, 0.96
to 1.18; P⫽0.23) diagnoses.
Discussion
The current study compared associations between psychiatric
comorbidity and AMI outcomes that were derived using 2
alternative methods of identifying psychiatric comorbidity
form administrative data in a cohort of veterans. We empha-
size the following 3 findings. First, rates of psychiatric
comorbidity differed depending on the identification ap-
proach and were substantially higher using diagnosis codes
Table 3. Adjusted Odds of Death for Patients With Both an Outpatient Diagnosis of Psychiatric Comorbidity and With
an Inpatient Diagnosis or With Either an Outpatient Diagnosis or an Inpatient Diagnosis, Relative to Patients With Neither
an Outpatient Diagnosis nor an Inpatient Diagnosis
30-Day Mortality
Subgroup n Adjusted OR
Yes outpatient diagnosis*
Yes inpatient diagnosis 1625 0.87
Yes outpatient diagnosis
No inpatient diagnosis 3600 1.29
No outpatient diagnosis
Yes inpatient diagnosis 660 0.92
No outpatient diagnosis†
No inpatient diagnosis 15 860 1.00
*Diagnosis of depression, anxiety, psychosis, PTSD, or bipolar disorder on one or more outpatient encounters.
†Referent group.
Psychiatric Disease and Mortality After AMI 217
from prior outpatient encounters than secondary diagnosis
codes from inpatient admissions. Moreover, the rates of
psychiatric comorbidity found using prior outpatient diag-
noses were more consistent with estimates of psychiatric
illness (eg, depression) in AMI patients in prospective or
cross-sectional studies that used patient interviews to identify
such conditions.3,27–29 Second, patients identified by second-
ary inpatient codes tended to have lower severity, as exem-
plified by lower overall predicted mortality rates, lower rates
of most comorbidities, and lower laboratory severity scores.
Third, associations between psychiatric comorbidity and
mortality and receipt of revascularization differed among
patients identified using outpatient and inpatient diagnoses.
For example, patients identified from prior outpatient encoun-
ters had a modestly elevated risk of both 30-day and 365-day
adjusted mortality, whereas patients identified by secondary
inpatient diagnosis codes had similar 30- and 365-day ad-
justed mortality as patients without such codes.
The differences in identification rates, differences in pa-
tient characteristics, and the different associations with AMI-
related outcomes suggest that these approaches may be
identifying somewhat different illness constructs.
The lack of an association between AMI related out-
comes and patients identified by inpatient codes may
represent several possibilities. First, secondary inpatient
codes may be more likely to represent incident (new or
acutely recognized) psychiatric illness associated with the
admission illness, although, to our knowledge, no studies
have specifically examined this possibility. Second, sec-
ondary inpatient codes may be more likely to represent
transient or less severe psychiatric illness or to represent more
remote psychiatric disease. Such conditions would have less
prognostic or disease management implications than ongoing
psychiatric disease. However, one exception to this finding
was the higher risk of death associated with an inpatient
secondary diagnoses of anxiety. This finding may indicate
that symptoms of anxiety are a marker of more severe
disease, undertreated cardiac ischemia, or patients’ awareness
of a more complicated hospital course. This explanation has
received some support by Moser et al,30 who reported that
365-Day Mortality
95% CI P Adjusted OR 95% CI P
0.71–1.06 0.13 0.94 0.82–1.08 0.11
1.16 –1.43 ⬍0.001 1.19 1.08 –1.31 ⬍0.001
0.65–1.32 0.67 1.02 0.78 –1.31 0.94
... ... 1.00 ... ...
 218 Circ Cardiovasc Qual Outcomes May 2009
AMI patients with higher levels of measured anxiety for
patients had increased rates of cardiac complications (eg,
arrhythmias, ventricular fibrillation, ongoing ischemia, and
reinfarction).
In contrast, patients identified by outpatient psychiatric
diagnoses in the prior 12 months may be more likely to have
chronic or unresolved illnesses, which, in turn, may impact
outcomes of acute medical conditions by decreasing rates of
dietary and medication adherence, poorer disease manage-
ment skills, and higher rates of tobacco use and other
substance abuse.31–33 It is possible that such patients may be
at higher risk of adverse events after AMI.
Finally, secondary inpatient diagnosis codes for psychiatric
comorbidity may be a marker for patients with less compli-
cated hospital courses or with less comorbid illness. This may
reflect the fixed number of secondary diagnosis codes that are
captured in administrative data and the possibility that psy-
chiatric comorbidities are less likely to be captured in the
presence of other complex diagnoses or treatment complica-
tions. This possibility is supported by Finlayson et al,34 who,
counterintuitively, found lower mortality associated with
diabetes, prior AMI, and chronic obstructive lung disease
among patients undergoing pancreaticoduodenectomy or ab-
dominal aneurysm repair. Other studies by Iezzoni et al35 and
Jencks et al36 have made similar observations in medical
populations.
Our findings of lower disease severity among patients with
psychiatric comrobidity identified by secondary inpatient
codes are consistent with 2 recent studies by Druss et al who
Downloaded from http://ahajournals.org by on November 9, 2023
analyzed administrative supplemented by clinical data col-
lected in the Cooperative Cardiovascular Project. These
studies found that individuals with psychiatric comorbidity,
identified by secondary diagnosis codes, had lower predicted
mortality, lower rates of many medical comorbidities and
shock, and lower laboratory severity.14,16
In contrast, in a study of younger privately insured patients
admitted with AMI, Jones et al19 found that patients with
psychiatric comorbidity, as identified by prior outpatient
claims, had either similar or higher rates of most medical
comorbidities (eg, peripheral vascular disorders, hyperten-
sion, chronic obstructive pulmonary disease, obesity, neuro-
logical disorders).
Finally, the lower rate of psychiatric comorbidity we found
using inpatient secondary diagnosis codes is also consistent
with the results of earlier studies of AMI that relied either on
secondary inpatient14,16,17 codes or prior outpatient diagnosis
codes19 to identify psychiatric conditions. Importantly, the
current study is the first study to demonstrate the differences
in rates of identification in the same patient population and
the first to specifically examine whether associations between
psychiatric comorbidity and AMI outcomes depend on the
method of identifying psychiatric comorbidity.
It is important to acknowledge several potential limitations.
First, our findings are based on a largely older male veteran
population and may not be generalizable to females or other
non-VHA populations. Generalizability may also be affected
by potential selection biases related to unique benefits (eg,
compensation for service related injuries, medication pre-
scription benefits) offered by the VHA healthcare system.
Second, our use of claims data may underestimate rates of
psychiatric comorbidity; as such illnesses are likely to be
underdiagnosed. Additionally, our use of claims data cannot
estimate the acuity or severity of the psychiatric comorbidity
and may be subject to misdiagnosis between PTSD and
depression. For example, among patients with an outpatient
diagnosis of PTSD, only 43% had a concomitant diagnosis of
depression, notably lower than rates reported in some studies
of veterans.37,38 Third, the effect sizes of the associations
between psychiatric comorbidity and mortality that we ob-
served were relatively modest and somewhat lower than the
effects observed in prior prospective studies of depression
and AMI related mortality.1,2,4
Fourth, the sensitivity and specificity of ICD-9-CM codes
in administrative databases may vary across individual diag-
nosis.39,40 Formal studies comparing accuracy of administra-
tive data to other clinical data has found varying levels of
agreement.35,41,42 Nevertheless, much work has used admin-
istrative data to examine the quality of care42 including
several studies of veterans.43– 45 Moreover, work by Krakauer
et al46 and Ash et al47 suggest that claims data may yield
similar associations with mortality as data abstracted from
patients medical records. Moreover, our models were further
strengthened by including specific cardiac injury markers and
other general laboratory data.
Finally, the exclusion of patients who did not have a visit
in months 13 to 24 before the admission may have excluded
some previously healthy VA users who did not have any
outpatient visits in the prior 24 months or patients who only
intermittently use the VA.
Despite these limitations, the current study holds implica-
tions for future research and policy. First, although the
different approaches of identifying psychiatric comorbidity
may identify different disease constructs, the low rates of
psychiatric comorbidity identified by secondary diagnosis
suggests that such diagnoses may have limited sensitivity.
Moreover, the somewhat different associations with mortality
and revascularization suggest that inpatient secondary diag-
nosis codes should not be used in isolation in studies to
examine the prognostic effects of psychiatric comorbidities
on outcomes of hospitalization or in the development of
risk-adjustment models to standardize hospital outcomes for
AMI and other acute conditions. Ideally, studies using ad-
ministrative data to study psychiatric illnesses should exam-
ine multiple approaches for ascertaining comorbidity. Sec-
ond, the current findings highlight the need for more in-depth
studies to examine the clinical complexity and chronicity of
psychiatric illness that is detected using diagnoses from prior
outpatient encounters and, perhaps most importantly, using
inpatient secondary diagnosis codes. Such work should
clearly examine associations of these conditions with impor-
tant outcomes.
Finally, this work emphasizes that, despite a growing
awareness about the impact of psychiatric illness on AMI
outcomes and possible mediators of such differences14,16
differences in outcomes may persist. These differences, in
turn, may represent opportunities for targeting strategies to
improve hospital care.
 Abrams et al
In conclusion, as the recognition of psychiatric comorbid-
ity improves with better screening approaches and a higher
public awareness, the monitoring of health care outcomes will
become increasingly important. It is likely that much of this
work will continue to use administrative data. However, in
relying on administrative data, both investigators and policy
makers should encourage the use of claims data from both
outpatient and inpatient encounters to identify psychiatric
illness. Moreover, in the absence of studies of the sensitivity
and specificity of different identification methods, both par-
ties should remain circumspect in interpreting findings that
rely on secondary inpatient codes for identifying psychiatric
comorbidity.
Sources of Funding
The research reported here was supported by the Department of
Veterans Affairs, Veterans Health Administration, Health Services
Research and Development Service through the Center for Research
in the Implementation of Innovative Strategies in Practice (CRIISP;
HFP 04-149). Dr Abrams is an associate supported by additional
VHA funding though the Office of Academic Affairs.
Disclosures
The views expressed in this article are those of the authors and do not
necessarily represent the views of the Department of Veterans
Affairs. The authors report no conflicts of interest in regards to this
study.
References
1. Parashar S, Rumsfeld JS, Spertus JA, Reid KJ, Wenger NK, Krumholz
HM, Amin A, Weintraub WS, Lichtman J, Dawood N, Vaccarino V.
Downloaded from http://ahajournals.org by on November 9, 2023
Time course of depression and outcome of myocardial infarction. Arch
Intern Med. 2006;166:2035–2043.
2. Ferketich AK, Schwartzbaum JA, Frid DJ, Moeschberger ML.
Depression as an antecedent to heart disease among women and men in
the NHANES I study. National Health and Nutrition Examination Survey.
Arch Intern Med. 2000;160:1261–1268.
3. Penninx BW, Beekman AT, Honig A, Deeg DJ, Schoevers RA, van Eijk
JT, van Tilburg W. Depression and cardiac mortality: results from a
community-based longitudinal study. Arch Gen Psychiatry. 2001;58:
221–227.
4. Ariyo AA, Haan M, Tangen CM, Rutledge JC, Cushman M, Dobs A,
Furberg CD. Depressive symptoms and risks of coronary heart disease
and mortality in elderly Americans. Cardiovascular Health Study Collab-
orative Research Group. Circulation. 2000;102:1773–1779.
5. de Jonge P, van den Brink RH, Spijkerman TA, Ormel J. Only incident
depressive episodes after myocardial infarction are associated with new
cardiovascular events. J Am Coll Cardiol. 2006;48:2204 –2208.
6. Nicholson A, Kuper H, Hemingway H. Depression as an aetiologic and
prognostic factor in coronary heart disease: a meta-analysis of 6362
events among 146 538 participants in 54 observational studies. Eur
Heart J. 2006;27:2763–2774.
7. van Melle JP, de Jonge P, Spijkerman TA, Tijssen JG, Ormel J, van
Veldhuisen DJ, van den Brink RH, van den Berg MP. Prognostic asso-
ciation of depression following myocardial infarction with mortality and
cardiovascular events: a meta-analysis. Psychosom Med. 2004;66:
814 – 822.
8. Rugulies R. Depression as a predictor for coronary heart disease. a review
and meta-analysis. Am J Prev Med. 2002;23:51– 61.
9. Barth J, Schumacher M, Herrmann-Lingen C. Depression as a risk factor
for mortality in patients with coronary heart disease: a meta-analysis.
Psychosom Med. 2004;66:802– 813.
10. Sorensen C, Brandes A, Hendricks O, Thrane J, Friis-Hasche E, Haghfelt
T, Bech P. Depression assessed over 1-year survival in patients with
myocardial infarction. Acta Psychiatr Scand. 2006;113:290 –297.
11. Lawrence DM, Holman CD, Jablensky AV, Hobbs MS. Death rate from
ischaemic heart disease in Western Australian psychiatric patients
1980 –1998. Br J Psychiatry. 2003;182:31–36.
Psychiatric Disease and Mortality After AMI 219
12. Carney CP, Jones L, Woolson RF. Medical comorbidity in women and
men with schizophrenia: a population-based controlled study. J Gen
Intern Med. 2006;21:1133–1137.
13. Daumit GL, Pronovost PJ, Anthony CB, Guallar E, Steinwachs DM, Ford
DE. Adverse events during medical and surgical hospitalizations for
persons with schizophrenia. Arch Gen Psychiatry. 2006;63:267–272.
14. Druss BG, Bradford DW, Rosenheck RA, Radford MJ, Krumholz HM.
Mental disorders and use of cardiovascular procedures after myocardial
infarction. JAMA. 2000;283:506 –511.
15. Kisely S, Smith M, Lawrence D, Maaten S. Mortality in individuals who
have had psychiatric treatment: population-based study in Nova Scotia.
Br J Psychiatry. 2005;187:552–558.
16. Druss BG, Bradford WD, Rosenheck RA, Radford MJ, Krumholz HM.
Quality of medical care and excess mortality in older patients with mental
disorders. Arch Gen Psychiatry. 2001;58:565–572.
17. Young JK, Foster DA. Cardiovascular procedures in patients with mental
disorders. JAMA. 2000;283:3198; author reply 3198 –9.
18. Petersen LA, Normand SL, Druss BG, Rosenheck RA. Process of care
and outcome after acute myocardial infarction for patients with mental
illness in the VA health care system: are there disparities? Health Serv
Res. 2003;38:41– 63.
19. Jones LE, Carney CP. Mental disorders and revascularization procedures
in a commercially insured sample. Psychosom Med. 2005;67:568 –576.
20. Plomondon ME, Ho PM, Wang L, Greiner GT, Shore JH, Sakai JT, Fihn
SD, Rumsfeld JS. Severe mental illness and mortality of hospitalized
ACS patients in the VHA. BMC Health Serv Res. 2007;7:146.
21. Dominitz JA, Maynard C, Boyko EJ. Assessment of vital status in
Department of Veterans Affairs national databases. comparison with state
death certificates. Ann Epidemiol. 2001;11:286 –291.
22. Sohn MW, Arnold N, Maynard C, Hynes DM. Accuracy and com-
pleteness of mortality data in the Department of Veterans Affairs. Popul
Health Metr. 2006;4:2.
23. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures
for use with administrative data. Med Care. 1998;36:8 –27.
24. Krumholz HM, Chen J, Wang Y, Radford MJ, Chen YT, Marciniak TA.
Comparing AMI mortality among hospitals in patients 65 years of age
and older: evaluating methods of risk adjustment. Circulation. 1999;99:
2986 –2992.
25. Knaus WA, Wagner DP, Draper EA, Zimmerman JE, Bergner M, Bastos
PG, Sirio CA, Murphy DJ, Lotring T, Damiano A. The APACHE III
prognostic system. Risk prediction of hospital mortality for critically ill
hospitalized adults. Chest. 1991;100:1619 –1636.
26. Landis JR, Koch GG. The measurement of observer agreement for cate-
gorical data. Biometrics. 1977;33:159 –174.
27. Dickens C, McGowan L, Percival C, Tomenson B, Cotter L, Heagerty A,
Creed F. New onset depression following myocardial infarction predicts
cardiac mortality. Psychosom Med. 2008;70:450 – 455.
28. Amin AA, Jones AM, Nugent K, Rumsfeld JS, Spertus JA. The prev-
alence of unrecognized depression in patients with acute coronary
syndrome. Am Heart J. 2006;152:928 –934.
29. Bankier B, Januzzi JL, Littman AB. The high prevalence of multiple
psychiatric disorders in stable outpatients with coronary heart disease.
Psychosom Med. 2004;66:645– 650.
30. Moser DK, Riegel B, McKinley S, Doering LV, An K, Sheahan S. Impact
of anxiety and perceived control on in-hospital complications after acute
myocardial infarction. Psychosom Med. 2007;69:10 –16.
31. Ziegelstein RC, Fauerbach JA, Stevens SS, Romanelli J, Richter DP,
Bush DE. Patients with depression are less likely to follow recommen-
dations to reduce cardiac risk during recovery from a myocardial
infarction. Arch Intern Med. 2000;160:1818 –1823.
32. Kronish IM, Rieckmann N, Halm EA, Shimbo D, Vorchheimer D, Haas
DC, Davidson KW. Persistent depression affects adherence to secondary
prevention behaviors after acute coronary syndromes. J Gen Intern Med.
2006;21:1178 –1183.
33. Rieckmann N, Kronish IM, Haas D, Gerin W, Chaplin WF, Burg MM,
Vorchheimer D, Davidson KW. Persistent depressive symptoms lower
aspirin adherence after acute coronary syndromes. Am Heart J. 2006;152:
922–927.
34. Finlayson EV, Birkmeyer JD, Stukel TA, Siewers AE, Lucas FL,
Wennberg DE. Adjusting surgical mortality rates for patient comor-
bidities: more harm than good? Surgery. 2002;132:787–794.
35. Iezzoni LI, Foley SM, Daley J, Hughes J, Fisher ES, Heeren T. Comor-
bidities, complications, and coding bias. Does the number of diagnosis
codes matter in predicting in-hospital mortality? JAMA. 1992;267:
2197–2203.
 220 Circ Cardiovasc Qual Outcomes May 2009
36. Jencks SF, Williams DK, Kay TL. Assessing hospital-associated deaths
from discharge data. The role of length of stay and comorbidities. JAMA.
1988;260:2240 –2246.
37. Spiro A,III, Hankin CS, Mansell D, Kazis LE. Posttraumatic stress
disorder and health status: the veterans health study. J Ambul Care
Manage. 2006;29:71– 86.
38. Hankin CS, Spiro A,III, Miller DR, Kazis L. Mental disorders and mental
health treatment among U.S. Department of Veterans Affairs outpatients:
the Veterans Health Study. Am J Psychiatry. 1999;156:1924 –1930.
39. Hsia DC, Krushat WM, Fagan AB, Tebbutt JA, Kusserow RP. Accuracy
of diagnostic coding for Medicare patients under the prospective-payment
system. N Engl J Med. 1988;318:352–355.
40. Waterstraat FL, Barlow J, Newman F. Diagnostic coding quality and its
impact on healthcare reimbursement: research prospectives. J Am Med
Rec Assoc. 1990;61:52–59.
41. Iezzoni LI. The risks of risk adjustment. JAMA. 1997;278:1600 –1607.
42. Hannan EL, Racz MJ, Jollis JG, Peterson ED. Using Medicare claims
data to assess provider quality for CABG surgery: does it work well
enough? Health Serv Res. 1997;31:659 – 678.
43. Department of Veterans Affairs, Veterans Health Services Research
Administration. A report on the quality of surgical care in the Department
of Veterans Affairs: The Phase III Report. R&D No. IL 10-87-9.
Downloaded from http://ahajournals.org by on November 9, 2023
4-21-1991. Washington, DC, Administrator of Veterans Affairs (findings
also found in Stremple JH, Bross DS, Davis CL, McDonald DO. Com-
parison of postoperative mortality and morbidigy in VHA and nonfederal
hospitals. J Surg Res. 1994;56:405– 416.
44. Department of Veterans Affairs, Veterans Health Services Research
Administration. A report on the quality of surgical care in the Department
of Veterans Affairs: The Phase II Report. R&D No. IL 10-87-8.
4-12-1989. Washington, DC, Administrator of Veterans Affairs (findings
also found in Stremple JH, Bross DS, Davis CL, McDonald DO. Com-
parison of postoperative mortality in VHA and private hospitals. Ann
Surg. 1993; 272– 85.
45. Department of Veterans Affairs, Office of Policy and Planning. Program
Evaluation of Cardiac Care Programs in the Veterans Health Adminis-
tration. Part 1: Acute Myocardial Infarction (AMI) and Percutaneous
Coronary Interventions (PCI) Cohort Analyses. Contract Number V101
(93) 1444. 4-11-2003. Washington DC, available at: http://www.va.gov/
opp/organizations/progeval.htm.
46. Krakauer H, Bailey RC, Skellan KJ, Stewart JD, Hartz AJ, Kuhn EM,
Rimm AA. Evaluation of the HCFA model for the analysis of mortality
following hospitalization. Health Serv Res. 1992;27:317–335.
47. Ash AS, Posner MA, Speckman J, Franco S, Yacht AC, Bramwell L.
Using claims data to examine mortality trends following hospitalization
for heart attack in Medicare. Health Serv Res. 2003;38:1253–1262.