The Prevalence of Psychiatric Disorders in General Hospital Inpatients: A Systematic Umbrella Review

Background: To inform the future development of consultation-liaison psychiatry services, we need accurate information on the prevalence of psychiatric disorders in the general hospital inpatient setting. Systematic reviews have summarized the literature on speci ﬁ c aspects of this broad topic, but there has been no high-level overview that aggregates their ﬁ ndings and identi ﬁ es gaps in the relevant literature. Objective: We aimed to produce a comprehensive overview of the ﬁ eld, summarizing the research literature on the prevalence of psychiatric disorders (i.e., interview-based psychiatric diagnoses) in general hospital inpatients. We did this using a systematic umbrella review (systematic review of systematic reviews), which is the best and most ef ﬁ cient method for summarizing a broad area of research. Methods: We searched Ovid Medline, Ovid Embase, Ovid PsycINFO, EBSCO CINAHL, and Scopus from database inception to September 2021 for systematic reviews that provided a pooled prevalence estimate, or prevalence range, for interview-diagnosed psychiatric disorders in general hospital inpatients. Two reviewers independently assessed articles and extracted data. The review is registered with PROSPERO, number CRD42019125574. Results: We screened 11,728 articles and included 10 systematic reviews in our umbrella review. We were able to extract pooled prevalence


INTRODUCTION
][3][4][5] Consultation-liaison (C-L) psychiatry services that aim to address this psychiatric morbidity have been set up in many general hospitals. 6To inform the future development of these services, we require accurate information on the prevalence of psychiatric disorders in the general hospital inpatient setting.It is therefore concerning that we have no comprehensive and systematic summary of the relevant research.Systematic reviews summarize the literature on specific aspects of this broad topic; for example, our group has published reviews of the prevalence of depression and anxiety in the inpatient setting. 7,8However, to date, there has been no high-level overview that aggregates the findings from such reviews and identifies gaps in the systematic review literature.
We therefore aimed to produce a comprehensive and systematic overview that summarizes published research on the prevalence of psychiatric disorders (i.e., interview-based psychiatric diagnoses) in general hospital inpatients.We did this using a systematic umbrella review.This type of review is the best and most efficient way of summarizing a broad area of research. 9An umbrella review involves identifying all the systematic reviews that contain relevant information and then extracting and collating data from them. 9,10As a systematic review of systematic reviews, an umbrella review represents one of the highest levels of evidence synthesis available.

Study Design
We conducted a systematic umbrella review (also called a meta-review, overview, or systematic review of systematic reviews).We used procedures that accorded with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines and registered the study protocol with PROSPERO (number CRD42019125574). 11,12

Search Strategy
We identified all relevant systematic reviews by searching Ovid Medline, Ovid Embase, Ovid Psy-cINFO, EBSCO CINAHL, and Scopus from database inception to September 2021.Searches were run for the combination of "prevalence", "general hospital inpatient", "psychiatric disorder" (which included major depression, bipolar disorder, generalized anxiety disorder, panic disorder, phobia, obsessive compulsive disorder, delirium, dementia, substance use disorders, schizophrenia, delusional disorder, acute stress disorder, adjustment disorder, post-traumatic stress disorder, somatoform [somatic symptom] disorders, factitious disorder, dissociative disorders, anorexia nervosa, bulimia nervosa, personality disorders), and "systematic review or meta-analysis" using both standardized subject terms and free-text terms, including synonyms and alternative spellings.
All references were exported (to Endnote X9; Thomson Reuters, New York, NY), and duplicates were removed following the method described by Falconer. 13We provide full details of the searches used in the Appendix 1.We contacted the authors of relevant conference abstracts that we found through our database search in order to obtain any associated publications and also did both manual reference list searches and forward-citation searches for the articles we included.

Selection Criteria
We included reviews (in any language) if they met all the following criteria: (1) The review was systematic (defined as a review with clear objectives, predefined eligibility criteria, a systematic search strategy, reproducible methods, and a systematic presentation of findings); (2) it aimed to summarize the prevalence of 1 or more of the psychiatric disorders listed above, using data from diagnostic interview studies (which may have used structured or clinical interviews); (3) it included a pooled prevalence estimate or a prevalence range for the relevant disorder(s) in adult (aged 16 years or older) general hospital inpatients (i.e., patients who had been admitted to any general hospital units, not patients in other settings such as psychiatric hospitals, rehabilitation hospitals, or emergency departments); (4) the pooled prevalence estimate or prevalence range used data from at least 2 primary studies.

Data Extraction
In order to minimize bias, 2 researchers independently screened the titles and abstracts of all the articles identified by the searches, using EndNote and Excel, to determine which might meet our selection criteria.If an article was considered potentially relevant, 2 researchers then reviewed the full text, with the help of a translator where necessary (an electronic translator was Psychiatric Disorders in Inpatients used in the first instance, and a person who was fluent in the relevant language was consulted if further clarification was required).Any disagreements about whether to include an article were resolved in discussion with a third researcher.Using a specially designed, standardized data-extraction form, 2 researchers independently extracted the following data from each included review: focus of the review; population studied; psychiatric disorder(s) studied; number and characteristics of primary studies of the prevalence of psychiatric disorders in general hospital inpatients; pooled prevalence estimate or prevalence range for general hospital inpatients; relevant author comments on the review's findings.Any disagreements about the data extracted were resolved in discussion with a third researcher.

Quality Assessments
To assess the quality of the reviews that we included, 2 researchers independently rated each review using the PASS quality checklist. 14This checklist was designed by our group to help researchers, clinicians, and clinical decision-makers rapidly judge the quality of systematic reviews of the prevalence of medical conditions.It has 4 domains: Planned with a clearly stated aim; All the relevant literature was considered; Selection of included studies was unbiased and transparent; Synthesis of data from included studies was unbiased and informative.Each of these domains has a small number of questions which can be scored as either "yes" or "no/unclear".

Data Synthesis
We described the characteristics of the systematic reviews and their findings in both narrative and table formats.We did not conduct a meta-analysis of the prevalence estimates extracted because this would have "double-counted" information from primary studies which were included in more than 1 review.

Literature Overview
Figure 1 shows the full umbrella review flowchart.The searches identified 18,376 articles.After removing duplicates, we screened 11,728 titles and associated abstracts and then assessed 864 full papers for eligibility.We excluded the majority of these 864 articles because they were not systematic reviews or they had not aimed to estimate the prevalence of a psychiatric disorder.We also excluded a smaller number of articles because, although they were systematic reviews, their prevalence estimates were not based on diagnostic interview studies (i.e., they included studies that used rating scales or medical records to determine the presence of psychiatric disorder).][17][18][19][20][21][22][23][24] The focus of each of the 10 reviews was generally broader than the prevalence of psychiatric disorders in general hospital inpatients (Tables 1-3).Consequently, only some of the primary studies that they summarized were relevant to this umbrella review.The prevalence estimates and ranges that we report below, and in the tables, refer to the subset of studies in each review that were specifically of general hospital inpatients.

Prevalence of Major Depression
We found 6 systematic reviews from which we could extract data on the prevalence of major depression in general hospital inpatients (Table 1).Their findings were reasonably consistent, with all the reported pooled prevalence estimates lying between 12% and 20%.One review reported a pooled prevalence of 12% in studies of general medical and surgical inpatients (and prevalence ranging from 2% to 56% in studies of patients in specialist units or with specific diagnoses). 7The authors noted that there was substantial heterogeneity in the prevalence estimates of the primary studies, but despite a number of exploratory analyses, they were unable to adequately explain this.They concluded that the wide range was likely to be due to differences between the studies in the local populations (e.g., different national and local prevalence of depression), health care systems (e.g., different hospital types and admission pathways), patients (e.g., variability in the characteristics of patients admitted to general medical and surgical units), and methods (e.g., study inclusion criteria, how diagnostic criteria were applied, and the timing of assessments after hospital admission).Two reviews were of the prevalence of major depression in patients with cancer; these reported a pooled estimate of 12.3% and a prevalence range of 4% to 14% for patients with cancer in the inpatient setting. 17,19Two reviews focused on studies of the prevalence of major depression after an acute myocardial infarction or a stroke. 20,21They van Niekerk et al.
reported somewhat higher prevalence estimates, of approximately 18%, for patients hospitalized with these conditions.A review of studies of the prevalence of depression in patients with burn injuries found a much lower prevalence of 4% in hospitalized patients.However, its authors emphasized that this estimate was likely to be "artifactual" as the 2 primary studies it included were of low quality. 22

Prevalence of Anxiety Disorders
We found 2 systematic reviews from which we could extract information on the prevalence of anxiety disorders in general hospital inpatients (Table 2).Only 1 of these reviews reported pooled prevalence estimates using data from studies of a wide range of hospital units; these were 8% for any anxiety disorder, 5% for   generalized anxiety disorder, and 3% for panic disorder. 8The authors noted that there was often a lack of clarity about when during the hospital admission the anxiety assessments had been conducted.They also noted that prevalence estimates may have been influenced by whether a diagnostic hierarchy had been used (i.e., whether an anxiety disorder was recorded if another disorder such as depression was also present) and how judgements had been made regarding the attribution of patients' symptoms to an anxiety disorder, rather than an adjustment disorder or their medical condition.The other review was of the prevalence of anxiety disorders in patients with chronic obstructive pulmonary disease. 23This review reported prevalence ranges of 10% to 55% for any anxiety disorder, 10% to 33% for generalized anxiety disorder, and 0% to 41% for panic disorder.The authors suggested that these wide ranges may have resulted from differences in the characteristics of study participants (e.g., their mean ages, the percentage of female participants, and the severity of patients' pulmonary disease).

Prevalence of Delirium
We only found 1 systematic review from which we could extract data on the prevalence of delirium in general hospital inpatients.This reported a pooled prevalence of 15% at the time of hospital admission (Table 3). 15The primary studies in this review were of patients who had been admitted to acute medical or geriatric medicine units (including stroke, respiratory, and oncology units).The authors of the review noted that there was a wide range of prevalence estimates in the primary studies.They also commented that many of these primary studies had excluded patients who were severely ill or lacked capacity to consent, suggesting that the reported pooled prevalence may be an underestimate.

Prevalence of Dementia
We also found only 1 systematic review of the prevalence of dementia in general hospital inpatients (Table 3). 16This review did not determine a pooled prevalence estimate but reported a wide prevalence range (2.8% to 63.0%) from studies of patients aged $55 years who had been admitted to medical and surgical units.The authors of the review noted that higher prevalence estimates were reported by studies done in geriatric units and of older patients (the prevalence range in studies done in geriatric units was 26.8% to 63.0%).They commented that (1) few primary studies had screened patients for delirium or depression to determine whether these were the cause of the participants' cognitive impairment and that this may have led to an overestimate of dementia prevalence; (2) many studies had excluded patients who were severely ill, had communication difficulties, or were unable to consent van Niekerk et al.
to participation, which may have led to an underestimate of dementia prevalence.

Prevalence of Other Psychiatric Disorders
We found no systematic reviews from which we could extract a pooled prevalence estimate or prevalence range for the other psychiatric disorders included in our searches (bipolar disorder, phobia, obsessive compulsive disorder, substance use disorders, schizophrenia, delusional disorder, acute stress disorder, adjustment disorder, post-traumatic stress disorder, somatoform [somatic symptom] disorders, factitious disorder, dissociative disorders, anorexia nervosa, bulimia nervosa, or personality disorders).We are therefore unable to report on the prevalence of these disorders.

Systematic Reviews
The full quality assessments of the 10 systematic reviews, using the PASS checklist, are shown in Table 4.In summary, the quality of the reviews was variable.They all had clearly stated aims and had considered all the relevant literature.However, it was often unclear exactly how these primary studies had been selected for inclusion and how the data extracted from them had been synthesized in the review.

Primary Studies Included in the Systematic Reviews
The reviews report limitations in the quality of the primary literature.The primary studies included were generally small, with median sample sizes ranging from 34 to 245.Although 3 of the 10 reviews had excluded primary studies that did not meet their basic quality criteria, the authors noted that many of the studies included still had important methodologic shortcomings. 7,8,19Of the remaining 7 reviews, 1 reported that the primary studies they summarized all had substantial methodologic limitations 22 ; the others assessed their primary studies to be of only moderate quality.Examples of methodologic shortcomings in the primary studies, described by review authors, include small samples, often recruited from single centers; convenience sampling; exclusion of patients with severe medical problems, documented history of psychiatric disorder, or lacking capacity to consent to participation; poor participation rates; lack of standardized application of diagnostic criteria; and lack of clarity regarding the timing of assessments in relation to the date of hospital admission.

Main Findings
We aimed to produce a comprehensive and systematic overview of the research literature on the prevalence of psychiatric disorders in general hospital inpatients, summarizing the data from relevant systematic reviews.
We were able to include data from 10 reviews in our systematic umbrella review.The pooled prevalence estimates extracted from these reviews for each disorder were as follows: major depression 12% to 20%, any anxiety disorder 8%, generalized anxiety disorder 5%, panic disorder 3%, delirium 15%.For dementia, we were only able to extract a prevalence range, which was 3% to 63%.The authors of the 10 reviews noted substantial variability in the prevalence estimates reported by the primary studies they included.We found no systematic reviews from which we could extract prevalence data for the other psychiatric disorders that we included in our searches, indicating important gaps in the literature.

Discussion of Main Findings and Relevant Literature
6][27][28] Of the disorders we sought evidence about, major depression was the focus of the greatest number of reviews.Although these reviews had addressed different clinical populations, their findings were similar.The prevalence estimates were 2 to 4 times the average 12-month general population prevalence of 5% as reported in international studies. 25The reviews of anxiety disorders also found them to be common in general hospital inpatients, but with a prevalence only modestly higher than that in the general population, 26 perhaps because diagnostic criteria for anxiety disorders require symptoms of relatively long duration that are judged not to be attributable to a medical condition. 8The prevalence of delirium, estimated at almost 1 in 6 inpatients, was unsurprisingly high but might have been even higher if the relevant systematic review had included studies done in surgical and intensive care units.Dementia, as might be expected, was found to Psychiatric Disorders in Inpatients have a very wide estimated prevalence range that was noted to be particularly high in geriatric units.We found no systematic reviews from which we could extract prevalence data for other psychiatric disorders.This lack of reviews is particularly concerning for disorders which we know are common from clinical experience and primary studies in the general hospital setting; for example, substance use and adjustment disorders. 29,30We also did not have sufficient data to be able to comment on the relative prevalence of psychiatric disorders in different hospital units.This is because reviews (and primary studies included in these) have tended to study either samples of patients with specific medical conditions or samples from a combination of units.

Strengths and Limitations of This Umbrella Review
The strengths of this umbrella review include (1) the use of information from systematic reviews, the highest level of evidence available; (2) data from systematic reviews of interview-based studies, in order to provide pooled prevalence estimates and ranges for definite psychiatric diagnoses; (3) the use of a comprehensive, sensitive search for articles on a wide range of psychiatric disorders with no restriction on language; (4) clearly defined inclusion criteria in order to minimize bias in article selection; (5) article selection and data extraction by 2 reviewers, working independently with a procedure for dealing with disagreements, to minimize bias.
The main limitations of the review are (1) the variable and sometimes limited quality of both the systematic reviews and the primary studies they summarized; (2) variation in the types of diagnostic interviews (different types of structured interviews and also clinical interviews) and the way that diagnostic criteria were applied in the primary studies; (3) the variable, and often unclear, timing of the psychiatric assessments in relation to the hospital admission date in the primary studies; (4) the different ways that reviewers defined "general hospitals" and "general hospital inpatient units"; (5) the use of an umbrella review approach which limited our inclusion of information from primary studies to only those that had been included in published systematic reviews; (6) our focus on only interview-diagnosed psychiatric disorders means that we did not include information on the prevalence of clinically significant symptoms as measured using rating scales.

Implications for Clinical Practice
While substantial variation is likely between hospitals and hospital units, it is clear from our findings that psychiatric disorders are very common in the general hospital inpatient setting.The data that we have summarized indicate this is the case for major depression, anxiety disorders, delirium, and dementia.There are also a number of other disorders that have not yet been the subject of a systematic review but which we expect from clinical experience to be common in this setting.These include adjustment disorder, substance use disorders, somatoform (somatic symptom) disorders, and post-traumatic stress disorder.While we do not have sufficient data to tell us the total prevalence of psychiatric disorders in the general hospital inpatient setting, we can make a conservative estimate of it by simply adding the lowest pooled estimates for the disorders for which we do have data.This addition suggests that as many as one-third of inpatients have a psychiatric disorder.Although such a crude calculation must be interpreted with caution as it makes the questionable assumption of no comorbidity between disorders, 31 it is still likely to be a substantial underestimate as it does not include either dementia (for which we do not have a pooled prevalence estimate) or any of the psychiatric disorders for which we have no systematic review data; nor does it include other presentations, including subthreshold disorders, which may nevertheless require psychiatric consultation. 32][35] These high prevalence estimates present a challenge for C-L psychiatry.We know that, although many of the third of patients who have psychiatric disorder need a psychiatric consultation, 36 referral rates to C-L psychiatry services are relatively low and typically less than 6%. 37We also know that the patients who are referred tend to be those with conspicuous disorders, such as hyperactive delirium, rather than those with less obvious problems such as depression. 37We therefore need to develop C-L psychiatry services to better meet the need identified by this review.This could be done by providing more population-based care, using screening to systematically identify patients with psychiatric disorders, as done by proactive and integrated approaches. 38We might also tailor services to better serve patients with the common, but currently less frequently Psychiatric Disorders in Inpatients referred, disorders, such as major depression. 39The sheer scale of the need we have highlighted also suggests that, in addition to providing direct care, C-L psychiatry has an important role in educating other general hospital clinicians to provide psychiatric care themselves, as advocated more than half a century ago. 40plications for Future Research While we are able to report some clear findings, our umbrella review also highlights the limitations of the existing literature.Consequently, there is an urgent need for more and better research.We need highquality systematic reviews that study the psychiatric disorders for which we currently lack pooled prevalence estimates.These reviews should be designed to meet the basic quality criteria we have described in our PASS checklist and specifically focus on the general hospital inpatient setting. 14We also need more, and better quality, primary studies of the prevalence of psychiatric disorders in the general hospital inpatient setting, especially in clearly specified hospital units.These studies would ideally exclude fewer patients by using recruitment protocols that allow patients who are unable to consent for themselves to take part; have clearly specified procedures for the application of diagnostic criteria, especially for disorders that require symptoms of relatively long duration (such as anxiety disorders) or collateral information (such as dementia); study a clearly specified list of psychiatric disorders; and report the timing of the assessments in relation to hospital admission.

CONCLUSIONS
We provide a summary of the prevalence of psychiatric disorders in general hospital inpatients, as described in relevant systematic reviews.Although the data are surprisingly limited, they do indicate that major depression, anxiety disorders, delirium, and dementia are common in this setting, and it is likely that at least a third of general hospital inpatients have a psychiatric disorder.While future planning of C-L psychiatry will benefit from more research, we already know enough to justify well-resourced psychiatric services for patients who have been admitted to general hospitals.These services should be developed to address both the scale and profile of need in the units that they serve.

FIGURE 1 . 1 Full
FIGURE 1. Prevalence of Psychiatric Disorders in General Hospital Inpatients: systematic umbrella review flowchart,

TABLE 1 .
Systematic Reviews of the Prevalence of Major Depression in General Hospital Inpatients * Calculated from data in the paper.vanNiekerk et al.

TABLE 2 .
Systematic Reviews of the Prevalence of Anxiety Disorders in General Hospital Inpatients

TABLE 3 .
Systematic Reviews of the Prevalence of Delirium and Dementia in General Hospital Inpatients

TABLE 4 .
PASS Quality Assessment of Systematic Reviews Included in the Umbrella Review