To The Editor:
Posttraumatic stress disorder is a serious psychological condition that has
received increasing attention over the past decade. Starr et al. should be
commended for their attempt, in their article "Symptoms of Posttraumatic
Stress Disorder After Orthopaedic Trauma" (2004;86:1115-21), to alert
orthopaedic surgeons to the fact that psychological consequences of a severe
orthopaedic injury are possible and important. Nonetheless, the percentage of
respondents who "met the criteria" for posttraumatic stress
disorder (as measured with the Revised Civilian Mississippi Scale for
Posttraumatic Stress Disorder) was startlingly high. Because the presence of a
serious psychiatric disorder in more than one-half of a traumatized sample is
exceedingly rare, it led me to examine the methods and analytic strategy used
in this report. There are a number of issues that deserve mention.
Posttraumatic stress disorder cannot be diagnosed until at least one month
after the traumatic episode. The authors noted that some respondents had been
injured as few as two days prior to assessment. Individuals seen less than
four weeks after the trauma should have been excluded from the sample.For a diagnosis of posttraumatic stress disorder to be made, the person's
response to the event must involve intense fear, helplessness, or horror
(Criterion A2 of the DSM [Diagnostic and Statistical Manual of Mental
Disorders]-IV). These responses do not appear to have been assessed.According to the DSM-IV, symptoms must be present for one month (Criterion
E). The duration of symptoms does not appear to have been measured in the
present investigation.Criterion F—that the disturbance must cause clinically important
distress or impairment in social, occupational, or other important areas of
functioning—is considered by many to be the hallmark of the disorder.
Again, it does not appear to have been assessed.
Posttraumatic stress disorder cannot be diagnosed until at least one month
after the traumatic episode. The authors noted that some respondents had been
injured as few as two days prior to assessment. Individuals seen less than
four weeks after the trauma should have been excluded from the sample.
For a diagnosis of posttraumatic stress disorder to be made, the person's
response to the event must involve intense fear, helplessness, or horror
(Criterion A2 of the DSM [Diagnostic and Statistical Manual of Mental
Disorders]-IV). These responses do not appear to have been assessed.
According to the DSM-IV, symptoms must be present for one month (Criterion
E). The duration of symptoms does not appear to have been measured in the
present investigation.
Criterion F—that the disturbance must cause clinically important
distress or impairment in social, occupational, or other important areas of
functioning—is considered by many to be the hallmark of the disorder.
Again, it does not appear to have been assessed.
Thus, it is clear that, while the investigators measured symptoms that were
consistent with criteria B, C, and D of the DSM-IV, the absence of a full
assessment of posttraumatic stress disorder required the investigators to be
extremely circumspect about their terminology. In fact, because all DSM-IV
criteria were not assessed (e.g., degree of functional impairment and duration
of symptoms), respondents should not have been assumed to have posttraumatic
stress disorder.
Moreover, an important historical event occurred very close to the
assessment of posttraumatic stress disorder among the respondents in this
study. The September 11, 2001, terrorist attacks had a substantial impact on
the psychological state of individuals across the country—not simply
those who lived in a directly affected community1. Moreover, these
attacks had a clear, demonstrable impact over the six months after the
attacks, with substantial numbers of individuals from a nationally
representative sample showing posttraumatic stress symptoms and elevated
levels of distress1. The fact that the assessment of posttraumatic
stress disorder was conducted within weeks after the attacks at one of the
study sites and within months after the attacks at the second site may have
inflated the results. In fact, many of the items on the Revised Civilian
Mississippi Scale for Posttraumatic Stress Disorder assess trauma symptoms
that are not specific to the orthopaedic trauma or injury (e.g., items 1, 4,
10, 11, 12, 13, 19, 21, 22, 24, 25, 26, 27, 28, 29, 30).
Finally, a more traditional way to analyze these data in order to examine
demographic and injury-related predictors of the presence or absence of
posttraumatic stress symptoms would have been to use logistic regression. The
nontraditional analytic strategy employed in this study may have masked
factors that, in combination, could have assisted the orthopaedic surgeon in
identifying at-risk individuals who might benefit from psychological
referral.
Dr. Silver raises several good points. First, she points out that,
according to the DSM-IV, posttraumatic stress disorder cannot be diagnosed
until at least one month after the trauma, and she suggests that patients who
were evaluated less than four weeks after the trauma should have been excluded
from our sample. We considered excluding such patients but decided against it.
In our sample, patients who were seen at a longer interval after the injury
had more symptoms of posttraumatic stress disorder. Exclusion of patients seen
less than four weeks after the trauma would have made the apparent prevalence
of the illness even higher. If we excluded those assessed less than thirty
days after the injury, the rate of posttraumatic stress disorder would have
jumped to 55%. If a 50% rate of the illness seems startlingly high, 55% would
be even worse.
Since our goal was to measure the prevalence of illness among orthopaedic
trauma outpatients, we decided to include those assessed soon after injury.
Patients who return to their orthopaedic surgeon's office two weeks after
injury with symptoms of posttraumatic stress disorder may not meet rigid
criteria for the illness, but the symptoms are still present. Our goal was to
record those symptoms and to bring them to the attention of other orthopaedic
surgeons. For that reason we wanted to be as inclusive as possible.
Next, Dr. Silver raises concerns about our failure to assess other criteria
listed in the DSM-IV and states: "In fact, because all DSM-IV criteria
were not assessed (e.g., degree of functional impairment and the duration of
symptoms), respondents should not have been assumed to have posttraumatic
stress disorder." The question of impairment is very important. In fact,
a search for causes of impairment after orthopaedic trauma was one thing that
led us to do this study.
If one asks orthopaedic trauma patients, "Are you impaired?"
many answer with a resounding "Yes!" We assumed, perhaps
incorrectly, that the fact that the patients were seeking treatment at an
orthopaedic clinic meant that their injury had caused a "clinically
important impairment." It seems probable that some patients' impairment
is due to their physical injury, but, for others, impairment may arise from
psychological distress. In fact, it may be difficult to tell whether impaired
function is caused by physical injury, by psychological distress, or by some
combination of the two. Psychological distress is strongly associated with
poor functional outcome scores among patients who have sustained high-energy
lower-extremity
trauma2. Could the
same be true for less severely injured orthopaedic trauma patients? And how
common are symptoms of posttraumatic stress disorder among orthopaedic trauma
patients? As best as we can tell, nobody knows the answer to those
questions.
The goal of our study was to try to estimate the prevalence of
posttraumatic stress disorder in our patient population. In the future, we
hope to find out if functional impairment can be reduced by treating
psychological distress.
As we noted, "a diagnosis of posttraumatic stress disorder based on a
questionnaire is not the same as a clinical diagnosis made by a mental health
professional. A more rigorous diagnosis may reveal different results."
Dr. Silver, a mental health professional, may be right when she says that we
should have been more circumspect with our terminology. However, if we assume
that our patients answered the questions honestly, it is hard to ignore their
responses.
Perhaps, in the interest of diagnostic rigor, it would be more accurate to
say that the patients in our sample did not meet all of the criteria necessary
to make the diagnosis of posttraumatic stress disorder, they just had lots of
posttraumatic stress symptoms.
Given that we failed to adopt the one-month criterion for symptom duration
and that we assumed that attendance at an orthopaedic trauma clinic
constituted evidence of a clinically important impairment, Dr. Silver's
criticism of our assignment of the diagnosis is probably deserved.
Dr. Silver also raises the question of the impact of the September 11
terrorist attacks on our patient sample and cites research carried out by her
and her colleagues1,
a web-based survey of 933 people residing outside New York, NY. The sample of
people assessed by Silver et al. included only one patient personally injured
in the attacks. Thirty-eight percent of the respondents had no exposure to the
attacks as they occurred, and another 60% reported watching them occur live on
TV. Only 2% of the sample had direct firsthand exposure to the attacks.
Surprisingly, at two months following the attacks, 17% of the respondents
reported September 11-related posttraumatic stress symptoms; 5.8% did so at
six months.
It may be that the September 11 attacks inflated the results of our study.
Or it may be that direct personal injury, such as that sustained by our
patients, is more likely to cause posttraumatic stress symptoms than indirect
exposure to an event such as the September 11 attacks.
Dr. Silver also notes that "many of the items on the Revised Civilian
Mississippi Scale for Posttraumatic Stress Disorder assess trauma symptoms
that are not specific to the orthopaedic trauma or injury." Our patient
sample was composed entirely of people who had sustained an orthopaedic injury
and were seen for follow-up in an orthopaedic trauma clinic. The cover sheet
for our questionnaire carried the title "Study of Stress after
Orthopaedic Trauma" and stated, "You are being asked to complete
this questionnaire because you have sustained an injury. Our goal with this
study is to see how injury affects orthopaedic patients emotionally or
psychologically." Questions 1, 4, 10, 11, 12, and 13 from the Revised
Civilian Mississippi Scale for Posttraumatic Stress Disorder were altered by
us to include references to "the injury," "my injury,"
or "since I was injured," instead of "the event," as
originally written by Norris and Perilla, the questionnaire's
developers3.
Questions 19, 21, 22, 24, 25, 26, 27, 28, 29, and 30 were used verbatim from
the questionnaire by Norris and Perilla. Our thought was that the cover sheet
and the questions made it clear that the goal of the questionnaire was to
assess how injury affected patients psychologically or emotionally. There is a
possibility that symptoms arising from the September 11 attacks inflated our
results. Since we did not address the attacks directly, we have no way to know
if this is the case.
Finally, Dr. Silver suggests that a more traditional analytic strategy
might have assisted us in identifying at-risk individuals who might benefit
from psychological referral. Previously, we had performed a multiple
independent variable logistic regression analysis with backward elimination,
initially including those variables that were significant (p < 0.05) at a
univariate level. The intent was to ascertain if combinations of significant
variables were good predictors. When we used this technique, ISS (Injury
Severity Score) remained in the model, whereas the summed Extremity
Abbreviated Injury Score and elapsed time since the injury were dropped.
Motivated by her suggestion, we increased the complexity of the model to
include other variables and numerous first-order interactions. We found that
age and ISS remained in an additive model with the predicted probability of
posttraumatic stress disorder increasing with a higher ISS and a lower age.
However, the area under the associated ROC (receiver-operating characteristic)
curve was 0.57, a value in the same range as was reported for single variables
in the paper. Thus, we think logistic regression adds little useful
information to the analysis of the data.
As for identifying "at-risk" individuals, we think we have.
Orthopaedic trauma patients are at risk for posttraumatic stress disorder, or
at least for posttraumatic stress disorder symptoms. And, at least in our
sample, patients who said that the emotional problems caused by their injury
were more difficult than the physical problems were at increased risk. We
think that this simple question may serve as a screening tool for identifying
patients who may benefit from further screening or treatment.
Silver RC, Holman EA, McIntosh DN,
Poulin M, Gil-Rivas V. Nationwide longitudinal study of psychological
responses to September 11. JAMA.2002;
288: 1235-44.2881235
2002
[PubMed][CrossRef]
McCarthy ML, MacKenzie EJ, Edwin D,
Bosse MJ, Castillo RC, Starr A; LEAP study group. Psychological distress
associated with severe lower-limb injury. J Bone Joint Surg Am.2003;85:
1689-97.851689
2003
[PubMed]
Norris FH, Perilla JL. The revised
Civilian Mississippi Scale for PTSD: reliability, validity, and cross-language
stability. J Trauma
Stress.1996;9:
285-98.9285
1996
[CrossRef]