Abstract
This cross-sectional study uses single health system data on acute hospital visits over 6 years to determine the yield of head computed tomography examinations for actionable pathology in common psychiatric presentations and to characterize low-risk scenarios in which imaging may be avoidable.
Head computed tomography (CT) examinations are used in acute settings to exclude structural pathology for patients with various psychiatric presentations.1 Recommendations for neuroimaging in psychiatric illness vary and mostly address first episodes of psychosis.2 To our knowledge, no current guidelines address specific and common psychiatric presentations such as hallucinations, delusions, and suicidal ideation. This study aimed (1) to determine the yield of head CT examinations for actionable pathology in common psychiatric presentations and (2) to characterize any very low–risk scenarios in which imaging may be avoidable.
Methods
The Yale University Institutional Review Board approved this retrospective cross-sectional study and waived informed consent. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Included sites within the Yale New Haven Health system consist of 1541 inpatient beds (>75 000 visits per year) and 4 emergency departments (EDs; approximately 200 000 visits per year). We queried the health system database to identify head CT examinations performed in ED and inpatient settings between March 2014 and February 2020, with an indication of suicidal ideation, homicidal ideation, hallucinations, delusions, paranoia, catatonia, and/or psychosis. We reviewed associated medical records to exclude patients with concurrent nonpsychiatric indications (eg, focal neurologic deficit, trauma, or headache) for imaging.
Acute findings on imaging were defined as those warranting immediate medical attention (eg, ischemic infarct, hydrocephalus, or intracranial hemorrhage). Explanatory findings were any other (nonacute) abnormality that could account for the patient’s presentation (eg, progressive gliosis or encephalomalacia). Diagnostic yield was defined as the proportion of head CT examinations with acute or explanatory findings.
Here, we report the yield for the whole study population and for specific presentations and care settings. 95% CIs were calculated using the binomial “exact” method.
Results
An initial health system database search revealed 546 patients who had undergone head CT examination. The 177 examinations with concurrent nonpsychiatric indications were excluded, resulting in 369 patients (Figure). The mean (SD) patient age was 59.6 (21.8) years; 165 (45.7%) were men and 204 (54.3%) were women. Of 369 head CT examinations, 297 (80.5%) and 72 (19.5%) were performed in the ED and inpatient settings, respectively. The most common indications were isolated hallucinations (223 [60.4%]) and other or unspecified psychotic symptoms (73 [19.8%]). Other scenarios were less common (Figure).
Figure. Flow of Computed Tomography Examinations From Patient Eligibility Through Categorization by Indication.
Psychosis NOS indicates other or unspecified psychotic symptoms.
aThe 19 head computed tomography (CT) examinations performed for more than 1 psychiatric indication were counted toward each component presentation in the indication-specific analysis (Table).
The diagnostic yield was 0.00% (95% CI, 0.00%-0.99%) for the total study population. Indication and setting-specific analyses also resulted in a yield of 0.00% (Table).
Table. Head Computed Tomography Examinations Ordered for Specific Psychiatric Presentations in Differing Settingsa.
| Presentation | Computed tomography examinations ordered, No. (%) | |
|---|---|---|
| Isolated psychiatric presentations (n = 350)b | With concurrent psychiatric presentations (n = 369)c | |
| Hallucinations | 223 (63.7) | 232 (62.3) |
| Delusions | 15 (4.3) | 21 (5.7) |
| Paranoia | 6 (1.7) | 17 (4.6) |
| Catatonia | 7 (2.0) | 7 (1.9) |
| Other or unspecified psychotic symptoms | 65 (18.6) | 73 (19.8) |
| Ideation | ||
| Suicidal | 30 (8.6) | 32 (8.7) |
| Homicidal | 4 (1.1) | 6 (1.6) |
| Multiple psychiatric symptoms | NA | 19 (5.2) |
| Setting | ||
| All emergency department | NA | 297 (80.5) |
| All inpatient | NA | 72 (19.5) |
Abbreviations: ED, emergency department; NA, not applicable.
The numbers of examinations for each isolated presentation and when counting patients with other psychiatric presentations are given. The number of patients in ED and inpatient settings is also provided. The yield for acute or explanatory pathology in all scenarios was 0%.
Percentages differ from those given in the Results, in which isolated symptoms are given as a percentage of all cases rather than only cases with isolated symptoms.
The 19 examinations performed for more than 1 psychiatric indication were counted toward each component presentation in the indication-specific analysis. Therefore, the indication-specific analyses sum to more than the total number of patients.
Discussion
In this cross-sectional study, we observed no explanatory findings for 369 patients with head CT examinations performed for psychiatric presentations over 6 years. In aggregate, the 95% CI for finding actionable pathology was less than 1.00%. Our study builds on the existing literature, suggesting that CT imaging may be avoidable for medical clearance in various scenarios.3
A recent review of 16 studies demonstrated actionable abnormality for 10 of 2312 patients (0.4%) undergoing CT or magnetic resonance imaging for a first psychosis episode.2 The included studies did not screen for nonpsychiatric symptoms and may therefore overestimate yield in isolated psychiatric presentations.2 Few publications and, to our knowledge, no professional society guidelines have provided recommendations for the specific presentations included in our study.4 Our results provide further support for refining guidelines and implementing quality improvement initiatives such as Choosing Wisely.5
This study has some limitations. The single health system setting may not capture the full breadth of practice variation. In addition, our database query was based on key terms and may not capture instances in which alternative wording was documented, perhaps reducing case variation and power. Future work using larger data sets should corroborate our findings.
In conclusion, routine use of head CT examination for common psychiatric presentations may not be warranted if there are no other indications for neuroimaging. Our findings suggest that additional research and guideline refinement may improve the value and efficiency of psychiatric evaluation in the acute setting.
References
- 1.Ng P, McGowan M, Goldstein M, Kassardjian CD, Steinhart BD. The impact of CT head scans on ED management and length of stay in bizarre behavior patients. Am J Emerg Med. 2018;36(2):213-217. doi: 10.1016/j.ajem.2017.07.080 [DOI] [PubMed] [Google Scholar]
- 2.Forbes M, Stefler D, Velakoulis D, et al. The clinical utility of structural neuroimaging in first-episode psychosis: a systematic review. Aust N Z J Psychiatry. 2019;53(11):1093-1104. doi: 10.1177/0004867419848035 [DOI] [PubMed] [Google Scholar]
- 3.American College of Radiology . Acute mental status change, delirium, and new onset psychosis. In: ACR Appropriateness Criteria. American College of Radiology; 2018. [Google Scholar]
- 4.Keepers GA, Fochtmann LJ, Anzia JM, et al. ; (Systematic Review ). The American Psychiatric Association practice guideline for the treatment of patients with schizophrenia. Am J Psychiatry. 2020;177(9):868-872. doi: 10.1176/appi.ajp.2020.177901 [DOI] [PubMed] [Google Scholar]
- 5.Srivastava R, Holmes RD, Noel CW, Lam TV, Shewchuk JR. Reducing neuroimaging in first-episode psychosis by facilitating uptake of choosing wisely recommendations: a quality improvement initiative. BMJ Open Qual. 2021;10(3):e001307. doi: 10.1136/bmjoq-2020-001307 [DOI] [PMC free article] [PubMed] [Google Scholar]