THOMAS P. BERESFORD, M.D. FREDERICC. BLOW, Ph.D. RICHARDC.W. HALL, M.D. LINDA O. NICHOLS, Ph.D. JAMES W. LANGSTON, M.D.
CT scanning in psychiatric inpatients: Clinical yield The authors surveyed CT scan use in 4,600 psychiatric admissions over a period of 2 1/2 years. Attending psychiatrists with free access to CT scanning requested scans in 165 (3.6%) of the cases. Review of 156 scans showed a positive result in 50 of them. Scans were ordered most often to rule out pathologic conditions: approximately 86.5% of use. Scan results altered diagnosis in 18% of cases and confirmed suspected diagnoses in 11 %. Of the 50 abnormal scans, 28 (56%) showed focal pathology. Preliminary indications are that mental status testing was more useful in predicting abnormal scan results than were focal neurologic findings. The authors present illustrative cases. ABSTRACT:
The use of CT scans of the brain in psychiatric patients has been the subject of controversy. Since its introduction in 1973, CT scanning has illuminated clinical conditions commonly seen by neurosurgical or neurologic practitioners but has, in the words of one reviewer, yielded "few dramatic findings" in the day-to-day work of the psychiatrist. I Investigators within psychiatry'"
and outside its bounds' have cautioned against overuse of this often expensive technological tool. Some institutions, with apparent distrust of the psychiatrist's clinical judgment or with an eye to cost containment, require a neurologist's authorization before a CT scan can be performed. We have been fortunate to have had few administrative restrictions on CT scan use. The psychiatrists on our staff
Presented at the 31st Annual Meeting of the Academy of Psychosomatic Medicine. Phi/adelphia. November 12.1984. From the VA Medical Center. Memphis; the University of Tennessee Center for the Health Sciences; and the University of Florida. Reprint request to Dr. Beresford. Chief, Psychiatry Service. VA Medical Center. 1030 Jefferson Ave.. Memphis. TN 38104.
FEBRUARY 1986·VOL27·N02
are able to obtain the procedure based on clinical indication alone. Part of our responsibility has been to monitor the results of CT scan use and to begin to find ways of making that use more precise. We report on the former topic here and reserve the latter for subsequent data analysis and discussion. Method With institutional review board approval of our written protocol, we reviewed 30 months of CT scan requests for patients admitted to the psychiatry service of our federally supported teaching hospital. The psychiatry service consists of 180 beds, part of a I,OOO-bed facility associated with the University of Tennessee College of Medicine. Two of the six 30-bed units admit patients for alcoholism or polydrug abuse. The remaining four wards admit patients with all other psychiatric diagnoses. We reviewed the medical records of patients who had received a CT scan of the brain during the 30 months. We collected basic demographic data (age, sex, race, marital status) for each patient as well as pertinent clinical information: admission and dislOS
CTscanning
charge diagnoses, date and result of the CT scan, admission mental state examination, neurologic examination. neurologic consultant's examination. EEG results, psychometric testing results, and abnormal laboratory test data. Finally. we reviewed the medical record data in each case as it pertained to three questions: Did the results of the CT scan change either the diagnosis or subsequent treatment? Did the CT scan rule out a specific diagnostic entity suspected after clinical examination? Did the CT scan confirm a previously suspected entity tentatively diagnosed after the clinical examination (the "rule in" approach)?" The reviews were conducted by two senior psychiatrists (TPB and RCWH). Most records were examined by only one reviewer. A subset of patient clinical data was examined by both reviewers independently to assess interreviewer reliability; their agreement was 89% (r=O.77. P
106
tails with the possible exception of a one- or two-sentence note on the request slip indicating the reason for the scan.
Results Quantitative data. A total of 165 CT scans of the brain were performed over the 30 months. Nine records were lost to follow-up. five due to transfer to other facilities and four due to loss by our clerical service. We thus reviewed a total of 156 scans. Over this same time period, 4.600 patients were admitted to the inpatient service. We, therefore. report the results of CT scans for 3.6% of all patients admitted and 94.5% of all scans requested by our service during this period. As might be expected in a veterans' hospital population, the majority of the study sample was male (95.5%). The age range was 23 to 89 years with a median age of 52 years; the age distribution was bimodal, with peaks at 30 and 60 years. The largest number of individuals were married (35.3%). One fourth were single (25.0%) while approximately one fifth (22.4%) were separated or divorced. The remaining subjects were widowed (4.5%) or of unknown marital status (12.8%). Only four patients had been referred to the radiologist from the alcohol and drug rehabilitation units. The remaining 152 had been admitted to general
inpatient wards. Table I shows the overall results of the CT scans. Approximately three of every ten scans ordered by the psychiatrists were read as abnormal. Somewhat more than half (56%) ofthe abnormal scans showed a focal finding. while the remainder (44%) presented with more generalized findings. More than two thirds of the patients referred for CT scanning had scans read as within normal limits. Table 2 lists the specific abnormalities found. They vary in clinical importance from small to very serious, the latter evidenced by tumor. hematoma, and infarction. Because the primary clinical usefulness of CT scanning derives from its ability to identify possible organic mental syndromes. we sorted the patients in this study into three categories with respect to diagnosis. Table 3 shows the number of patients who presented signs or symptoms of organic mental states. of functional diagnoses. or of both categories, in the judgment of the admitting physician. By the time of discharge. various organic or functional diagnoses had been established or the signs and symptoms of each remained, placing patients in the both category. As can be seen. the number of patients in the organic and functional categories increased, while that in the both category lessened. This change was highly significant statistically (P<. 000 I). With respect to ordering practices, in an overwhelming majority of the cases (86.5%) the physicians requested the CT scan to rule out a suspected pathologic condition (Table 4). In only an additional 10.9% were the scans obtained to confirm a previously suspected clinical diagnosis. With regard to clinical impact. in the opinion of the reviewers. the information from the scan changed or should have changed the diagnosis for
PSYCHOSOMATICS
about one patient out of every six who received the examination. This is comparable with the rate found in previous studies.' Table 5 provides an accounting of the 28 patients in whom we believed the CT scan result had had a significant impact on diagnosis. In all but one of the cases, the CT result taken with the history, physical examination, and laboratory data superseded a functional diagnosis. In the case of patient 14, marked generalized atrophy considered in the context of his clinical history argued for a more
pessimistic diagnosis of schizophrenia, undifferentiated type. In the more dramatic cases, CT scanning yielded a definite diagnosis. But in most of the cases (18 of 28) diagnosis was confined to syndromes listed in DSM-III': organic affective disorder, dementia, and organic delusional disorder. For statisticians and for the clinician yearning for diagnostic certainty, the number of definitive cases is small. From our point of view, the information appeared nonetheless valuable in guiding further diagnostic
procedures, the use of medicines, and follow-up care. The question arises as to whether any preexisting clinical signs or symptoms predicted the outcome of the CT studies in this sample of patients. We assessed two clinical parameters: the presence or absence of focal neurologic signs at physical examination and the presence or absence of altered results of the mental status examination on admission. With respect to focal signs, our assessment revealed an association between a patient's neuro-
Table 2-Neuroradiologlc Interpretations: Bilateral (N = 22), Right-Sided (N = 13), and Left-Sided (N = 15) Lesions Findings
N
Bilateral lesions Cerebral atrophy Cerebral and cerebellar atrophy Ventricular enlargement Ventricular outflow obstruction Atrophy of the superior cerebellar vermis Righ frontal and left occipital lesions (old cerebrovascular accident) Bifrontal atrophy Focal atrophy about the Sylvian fissures Patchy encephalomalacia In the posterior fossa Enlarged third ventricle Small frontal hematomas or hygromas Old surgical lesion Brain stem stroke
6
2 2 2 2
FEBRUARY 1986·VOL27·N02
N
Focal infarct of the putamen Lacunar infarcts of the putamen and Internal capsule Lacunar infarct of the head of the caudate nucleus Old craniotomy site Mucocele of the orbit
left-sided lesions
Right-sided lesions Frontal lobe atrophy Focal parietal lobe atrophy Subdural hematoma Encephalomalacia adjacent to lateral ventricle, old Encephalomalacia in the middle cerebral artery distribution Parietal and temporal lobe atrophy Occipital lobe infarct including visual cortex
Findings
2 1 1
Infarct of the putamen Atrophy in the middle cerebral artery distribuhon with mass effect Frontal lobe infarct Frontal encephalomalacia Midfrontal hypodense lesion from old hematoma Frontal and temporal lobe encephalomalacia due to recent skull fracture Temporal lobe stroke Temporal lobe atrophy, worse about the Sylvian ftssure Metastatic parietal lobe lesion, renal cell carcinoma Anterior parietal lobe encephalomalacia Occipital lobe encephalomalacia (infarct) Cerebral and cerebellar encephalomalacia from old intracerebral hematoma Calciftcatlon, parietal lobe
2 2 1 1
107
CTscanning
Table 3-Dlagnostlc Groupings Before and After CT SCan
I
Admission N(%)
Discharge
Organic
25(16.0)
42(26.9)
Functional
52 (33.3)
64(41.0)
Both
79 (50.7)
50(32.1)
N(%)
Table 4-Cllnlcallmpact of CT SCans
I Changed diagnosis Ruled out pathology specific nonspecific Confirmed diagnosis
I
N
%
28
17.9
69 66
44.2 42.3
17
10.9
logic status and the laterality of concurrent cr scan findings. In addition, an abnormal mental status examination, especially one characterized by cognitive loss, appeared to predict positive CT findings (49%). We will offer a more detailed analysis of this at a later time. Case descriptions. To provide the clinician with more detailed examples of our cases, we report four that we found particularly germane to the point at issue in this report: the clinical utility of information from CT scan studies. The examples are taken from Table 5. Case 12. An 89-year-old married man was brought to the hospital by his family, who alleged that he had become agitated and had struck his aging wife, whom he accused of being unfaithful. The triage physician referred the patient to the psychiatry service 108
and he was admitted to a general unit. He immediately began accusing the hospital staff of trying to hurt him, continued in an agitated state, and refused to cooperate with mental status testing. He nonetheless appeared to be striking out at visual phantoms that the staff took to be hallucinations. He appeared to be inattentive. His judgment was impaired and he appeared to be unable to remember recent events. He had no history of psychiatric hospitalization or disturbance. Physical examination revealed no focal findings or other neurologic disorder. Laboratory examination demonstrated anemia with low serum vitamin B". An EEG was read as diffusely abnormal "consistent with a degenerative disorder ... The patient's CT scan is pictured in Figure I and demonstrates an infarct of the occipital lobe. This, along with pernicious ane-
mia, constituted his final diagnosis, and he was referred for treatment and rehabilitation of these disorders. Case 8. A 40-year-old man recently separated from his wife was admitted to the psychiatry service after swallowing "half a bottle of pain pills" in a suicide attempt. His estranged wife noted that he had been intermittently depressed, a condition punctuated by episodes of violent behavior. Both the violence and the apparent depression had begun one year prior to admission, after the patient had suffered a head injury. There was no loss of consciousness and no amnesia at that time. Since then, however, the patient had apparent amnesia associated with the violent episodes, according to his wife. The unpredictable nature of the violence, the amnesia, and an increasing frequency of both had prompted her to leave. The patient's history was positive for alcoholism and he described occasional vision changes bordering on hallucination. The results of the mental status examination were otherwise unremarkable and no focal neurologic findings were detected by the psychiatrist or the consulting neurologist. An EEG without sleep deprivation and without nasopharyngealleads was read as normal. This patient's scan is shown in Figure 2. It reveals a lesion on the posterior limb of the right internal capsule that was read as an old cerebrovascular accident. This information resulted in a provisional diagnosis of temporal lobe epilepsy. The patient left the hospital against medical advice before further evaluation and medication trial could be effected. Case 16. This 62-year-old man with a history of chronic paranoid schizophrenia was brought to the hospital by his wife because of what appeared to be a sudden change in personality. He had suffered from schiPSYCHOSOMATICS
Table 5-Dlagnoatlc Changes Affected by CT scan Results Pt.
Age
1 2 3 4 5
50 64
CTftndlng*
Adm....on diagnosis
74 58
FA FA eSF block Metastasis GA
Bipolar. manic Major depression Major depression Major depression Major depression
6 7
65 50
FA GA
Dysthymic disorder SchiZophrenia. paranoid
8
40
FE
9
49
eSF block
Antisocial personality disorder Schizophrenia
10
64
FA
11 12
73 89
FE FE
13
52
14 15
35
16
62
17
63
Subdural hematoma GA FE FE FE
18
63
FE
19
88
eSF block
20 21 22
71 51
FA FE FE FE
71
60
Generalized anxiety disorder Bipolar disorder. atypical Organic mental syndrome with paranoia Alcohol inte»
23
69 40
24 25
58
FA FE
26
35
GA
undi terentiated Alcoholism
27
42
GA. FE
Alcoholism
28
31
FE
Alcohol dependence
61
Schiz~renia.
Discharge diagnosis
Organic affective syndrome Dementia Hydrocephalus Renal cell carcinoma. metastasIs Organic delusional syndrome, probable alcohol dementia Organic affective disorder Dementia, mild; organic hallucinosis Possible temporal lobe epilepsy Delirium, seizures, hydrocephalus Dementia, frontal lobe syndrome Organic affective syndrome Occipital lobe infarct Subdural hematoma Schizophrenia. undifferentiated Toxic delirium, mild dementia Organic delusional disorder Dementia probably due to old eVA Occipital lobe Infarct Normal pressure hydrocephalus Organic affective disorder Organic affective disorder Organic affective disorder Dementia, frontal lobe syndrome Dementia Organic affective disorder, possible dementia Dementia, probably multi-infarct with alcohol component Dementia, post old intracerebral hematoma Organic affective disorder
"FA: toeal atrophy GA: generalized atrophy FE: toeal encephalomalacia
FEBRUARY 1986'VOL27'N02
109
CTscanning
zophrenia since his late teens. For over two years prior to admission he had been well maintained on thioridazine. 500 mg/d. One year before admission he experienced a witnessed grand mal seizure. After a negative evaluation. he was begun on phenytoin. At admission. his phenytoin level was within the therapeutic range. His only other medicines were an oral hypoglycemic agent for diabetes mellitus diagnosed five years earlier and the thioridiazine. The patient's wife corroborated the patient's statement of compliance in regularly taking his antipsychotic medicine. The patient presented with a depressed mood but without flattened affect. His speech was slow. He de-
scribed delusions of paranoia and vague hallucinatory noises but not the voices that he had usually heard. He presented with poor recent memory and poor concentration but with an otherwise clear mental status. He had no focal or lateralizing physical findings. No EEG was ordered during the present admission. An EEG done three months earlier had revealed diffuse slowing. His scan, in Figure 3, shows an old lacunar infarction in the right internal capsule and the putamen. This patient was continued on his medicine regimen and underwent gradual improvement to his fonner state as his cerebrovascular status improved. Case 13. A 52-year-old man sepa-
FIGURE I-The arrow points to an infarcted area in the left occipital lobe, detected by CT scan in an elderlyagitated patient. In the figures, you are looking down at the patient's head and R denotes right side ofhis head.
110
rated from his wife came to the hospital complaining of depression. He had a long history of alcoholism and reported having been in an automobile accident about one week before admission. He could not recall having lost consciousness at that time. He appeared intoxicated but a breath test for alcohol was negative. He complained of difficulty seeing and of headache. He had no history of seizures. His mental status was characterized by expressionless affect and an indifferent mood. Short-tenn memory was poor and he found unexplainable gaps in his long-tenn memory. His concentration was poor. On physical examination, he had a shuffling, broadbased gait. and positive snout and
FIGURE 2-The arrow points to a lesion on the posterior limb of the right internal capsule in a violent middle-aged patient who later received a diagnosis ofpossible temporal lobe epilepsy.
PSYCHOSOMATICS
suck reflexes bilaterally. There were no focal signs. The results of the laboratory examination were nonnal and no EEG was done. Figure 4 presents his CT scan showing a recent subdural hematoma and a newer subdural bleed medial to the first hematoma. There is also a dramatic mass effect with compression of the right lateral ventricle and obstruction of the left lateral ventricle outflow. This patient was taken to surgery for aspiration of his hematomas and subsequently recovered both affect and memory.
Discussion Current lore suggests that, left to themselves, psychiatrists will request
expensive tests, such as CT scans, without regard to cost or to clinical usefulness of the infonnation to be gained. Our data do not bear this out. Left to themselves, eight separate attending psychiatrists in a VA medical center over 21f2 years ordered CT scans for a very small percentage of their inpatients. Contrary to the possible advantages of the "rule-in" approach, most ofthe CT scans were ordered to rule out possible pathologic entities. Given the nonspecificity of most psychiatric symptoms,6 the need to rule out subtle disease processes remains, in our view, the only justifiable approach to psychiatric diagnosis. This is generally accepted, as evidenced by the in-
FIGURE 3-The arrow points to an old lacunar infarction involving the right internal capsule and putamen. This patient was found not to have worsening ofschizophrenia but instead an organic delusional disorder.
FEBRUARY 1986·VOL27·N02
junction included for every DSM-III diagnosis: one must rule out any underlying physical pathology that might account for the patient's symptoms.' The usefulness of this approach can be questioned from two points of view: the impact of the CT scan information on diagnosis and treatment, and the cost of the test for each positive result. One in every three patients for whom a CT scan was requested had a positive finding. For half of these (that is, one in every six in the study sample), the CT scan finding changed the diagnosis and therefore had an impact on treatment. In our view, this is a high rate of return for any specialized clinical test.
FIGURE 4--The upper arrow points to a newer hemorrhage with an older one lying lateral to it. The lower arrow points to the right lateral ventricle. now displaced to the left ofthe midline. The left lateral ventricle is lying posteriorly. Following surgeryfor aspiration ofthe hematomas. this patient recovered both affect and memory. III
CTscanning
If we assume, for purposes of discussion, that the positive scan patients were the only ones with positive CT findings among the total number admitted to our service, the rate of positive scan findings for the entire study population would be a little more than 1% (50 out of 4,600). If one ordered a CT scan as a screening test for every patient admitted to the hospital, a theoretical rate of positive yield would be about 1%. Using standard clinical judgment, our staff increased that rate of yield by some 30 times. As Table 5 and the case descriptions suggest, patients presenting with psychiatric symptoms offer a difficult clinical challenge. The CT scan provided the diagnosis in cases 12 and 13. Without this noninvasive test. patient 12 could have been properly diagnosed only by visual field testing-an impossibility in his agitated state-or possibly by arteriography. For patient 13. the diagnosis might have been suspected only when his hematoma worsened to the point of long track signs or more serious symptoms. For patient 8, the CT findings together with the clinical history pointed out the need for an aggressive evaluation and perhaps a medicine trial for someone who might otherwise have been dismissed as having a character
disorder. Similarly. had his physician not heeded the DSM-II1 requirement of ruling out organic pathology, patient 16 might have been regarded as having an exacerbation of his schizophrenia. an indication for higher doses of his neuroleptic or for a change in medicine regimen. In this case. either course of action would have been needless. Because of instances like these. we find it unavoidable to conclude that the CT scan is a highly useful test in the hands of trained psychiatrists. whether one considers the quality of the information it can provide or the frequency with which significant radiologic change can be found. No discussion would be complete without considering cost. In our hospital each CT scan costs approximately $110. excluding the cost of the radiologist's time. The total cost for the 156 patients in our sample was $17.160. The cost for each positive finding (total cost/number of patients with a positive CT result) was about $343. For a comparable private hospital in our city. adding debt and overhead costs to the CT scan bill. a single scan might cost $448. In that setting. the total bill would have been $69.888 and the cost per positive scan about $1.397. From our perspective in a
public hospital setting. the cost of a positive scan is small relative to the clinical usefulness of the infon,nation derived. not to mention the reassurance such information can give both patient and physician. The challenge for clinical researchers in attempting to evaluate and to improve the day-to-day use of technologic advances such as CT scanning lies in the need for clearer indications for the use of specialized tests. The high frequency of symptoms of both organic mental disorders and classic psychiatric disorders among the patients in our sample underscored the need for greater precision in clinical observation. In the present study two traditional examinations. neurologic focality and mental status change. appeared worth evaluating with respect to their use in predicting a meaningfully positive CT scan result. While our current data require further sorting and analysis. our impression is that focal findings on physical examination are probably much less useful in predicting CT scan outcome than was previously thought and that certain aspects of mental status testing. most notably cognitive functioning. will probably retain their usefulness. We shall report these data and analyses subsequently. 0
3 Tsai L. Tsuang M: How can we avoid unnec· essary CT scanning for psychiatric patients? J Clin Psychiatry 42:453-454. 1981 4. Larson EB. Mack LA. Watts B. et al: Computed tomography in patients with psychiatric ill· nesses: Advantage of a 'rule-in' approach. Ann Intern Med95:360-364. 1981.
5. Diagnostic and Statistical Manual of Mental Illnesses. ed 3 Washington. DC. American Psychiatric Association. 1980. 6. Hall RCW: Psychiatric Presentations of Medical //fness. New York. SP Medical & Scientific Books. 1980.
REFERENCES 1. ROberts JKA. Lishman WA The use of the CAT. head scanner in clinical psychiatry. Br JPsychiatry 145:152-158. 1984 2. Evans NJR: Cranial computerized tomography in clinical psychiatry: 100 consecutive cases. Compr Psychiatry 23 445-450. 1982
112
PSYCHOSOMATICS