The differences between referred and non-referred patients to a psychiatric consultation-liaison service in a general hospital

Abstract

Objective

The primary aim of this study was to investigate whether there are clinical differences between patients who are referred or not referred for psychiatric consultation and liaison service. The secondary aim was to compare the perspectives of doctors, nurses and patients.

Methods

This naturalistic, prospective and comparative study (N = 294) utilised a control sample of non-referrals (n = 177, consenting 81) and referrals (n = 177, consenting 49). The normality of the data was examined with the Shapiro-Wilk test; bivariate group comparisons were made using Mann-Whitney, Wilcoxon tests and bivariate regression analyses. Statistically adjusted group comparisons were performed with multivariate median regressions.

Results

The sample presented limited representativeness. Referred patients were predominantly women, mostly living alone and not working. Compared to the non-referred patients, their disease episode and length of hospital stay were significantly longer, self-efficacy and quality of life lower and psychological stress was higher. For referred patients, there were no differences between the estimations of mental burden and the need for care among doctors, nurses and patients. Self-efficacy and appraisal of one's own burden were the best predictors of the extent of mental symptoms.

Discussion

Patients in an admission ward for internal medicine referred to a psychiatric consultation-liaison service displayed a more adverse psychosocial profile and were more psychologically burdened than non-referred patients, but they are also relevantly subsyndromal burdened. Identifying and supporting burdened patients is an endeavour that requires collaborative care, especially in the transition to specialised mental health and to primary care.

Keywords

consultation-liaison psychiatry non-referred patients patient’s perspective self-efficacy internal medicine

Introduction

Consultation psychiatry (CLP) has been well established since the 1960s, especially as multidisciplinary consultation-liaison teams. According to the underlying care philosophy,^1–3 CLP can be considered as “psychosomatics” because: i) it represents a comprehensive approach to the clinical needs of patients in general hospitals; ii) the psychosocial perspective is included in multidimensional diagnosis, care planning and secondary respectively tertiary prevention; iii) the bidirectional influences of mental and physical conditions are recognised in every disease, and they are sometimes decisive for an adjustment of treatment perspective.

Referral policy is not universal: it depends on referral barriers and facilitators, resources, awareness of psychosomatic relationships in clinical practice, a cooperative spirit, the experience of referring doctors and core competences of CLP teams.^4,5 Therefore, the range of diagnoses and referral questions are very broad and by no means uniform. Nevertheless, the most frequent diagnoses in consultation services are substance abuse-related mental disorders, age-related disorders like delirium and dementia and a broad scope of adjustment disorders with depression and anxiety symptoms.⁶ Liaison settings focus on special care settings and target programmes, including psychooncology, puerperal disorders, organ transplantation, psychocardiology or chronic pain. This method ensures a multidisciplinary treatment during the hospital stay, for instance after a suicide attempt or with eating disorders, puerperal psychosis, and lack of capacity to consent or straining crisis after serious diagnoses. In the last decades, a wide range of services and subspecialties have been established,⁷ frequently without balanced consideration of real patient’s needs.⁸ Research about real consultation structures^4,9 has become relevant in order to establish care rationales for CLP^10,11 towards efficient collaborative care.

Referral quotas to CLP services can reach up to 5% of admissions in general hospitals; on average, at least a third and up to half of admissions in a general hospital suffer from a relevant psychiatric condition depending on the ward, age and underlying somatic disease.^12–17 A broad scope of investigations has dealt with the effectiveness of CLP; Wood and Wand investigated 40 articles and identified five measurements of effectiveness: cost effectiveness, including length of stay, concordance, staff and patient feedback and follow-up outcome studies.¹⁸ Little is known about the psychological burden and needs of non-referred patients in general hospitals, despite the high prevalence of current mental disorders among internal medical patients (between 30–40%), especially for somatoform disorders (17–20%).^19,20 Despite these prevalence figures, there is a wide gap between performed consultations and quotas of mental disorders in wards for internal medicine. In this investigation, the burden and needs of non-referred and referred patients in an internal medicine ward were dimensionally compared. The importance of this study in the German context is first to improve the awareness facing mental conditions in admitted patients in general hospitals independently of the presence of full criteria for mental disorder; second, to demonstrate the importance of better staffed CLP teams in order to identify care needs in general hospital avoiding referrals to external psychiatric units.

Methods

Objective

The main objective for this study was to determine whether there are clinical differences between the patients referred and not referred to a CLP service and to examine whether patients not referred to the CLP service have a relevant psychosocial burden. The secondary objective was to compare the perspectives from doctors, nurses and patients on burden and support needs.

This issue can be broken down into the following three questions:

Are there psychosocial, medical and care-related differences between patients who are selectively referred to the psychiatric consultation and liaison service and those who are not?

Are there differences in perception of the somatic and psychological burden, as well as the need for psychiatric support beyond standard medical care, among physicians, caregivers and patients?

Are there statistically controlled prognostic associations between mental burden and well-chosen clinical and psychosocial variables?

Study design

This study was prospective, comparative and naturalistic in a developed country (Germany). Besides the electively referred patients to the CLS, all patients admitted to the internal medicine ward who did not receive a consultation (control group) were consecutively visited by the psychiatric expert between March and June 2017. Inclusion criteria for participation in the study: i) Internistic disease, which required a hospitalization; ii) Age >18 years; iii) Informed consent to the study; iv) Indication for consultation: naturalistic according to the clinical criteria of the internal medicine doctors. The exclusion criteria were: i) lack of consent; ii) age of minority; iii) no possible understanding; iv) excessive weakening caused by the disease. If the inclusion criteria were met, the survey was performed in a standardised manner in the first 24 hours after admission. All patients were informed that personal data would be collected but that the statistical evaluation of the data would be anonymous. After their written consent, the patients were included in the study.

Almost all referred patients (48 out of 49) showed a mental disorder: 32.6% substance abuse disorder; 28.3% depressive disorder, 17.4% organic mental disorder, and 10.9% somatoform disorder (see Figure 1). 35.4% of referred patients showed at least one additional mental disorder. Therefore, 37.5% of referred subsample displayed a substance abuse disorder. Within the subsample of non-referred patients 18.5% had a mental condition (not necessarily a mental disorder) that required psychopharmacological treatment before admission in a general hospital.

Figure 1.

Flow chart of the sample composition (controls and cases), reasons for non-participation and assessment of representativity of the sample by means of comparison of seven routine variables. “Controls” are non-referred patients from an admission unit of internal medicine; “Cases” are electively referred patients to a consultation-liaison (CLP) team. p = level of significance of testing; n.s.: not a statistically significant difference.

The indication for psychiatric consults was established with daily visits to the senior physicians on the basis of basic psychopathological, behavioural or psychosocial criteria that required clarification and/or treatment. The consultant physician visited the assigned patients at their bedside; the interview was not structured performed. All data collection and consultation occurred at the bedside. The questionnaires were checked for completeness in a timely manner (no later than the following day), so that missing items were added together with the patients and possible problems in understanding could be clarified.

The doctors and nursing staff of the internal medicine ward were informed about the study. The definition of the items was agreed upon with ward staff in order to avoid deviations in interpretation. The patients who met the inclusion criteria but were not included in the study were anonymously assigned to four different categories: a) patients who were not reached; b) patients who explicitly did not consent to the study; c) patients who were unable to participate in the study for health reasons; or d) patients whose knowledge of German was insufficient to complete a survey.

Figure 1 shows the subsamples of controls (patients not referred to the consultation service, n = 177) and cases (patients presented to the consultation service, n = 117). Ninety-six patients were excluded from the control sample, and 68 were excluded from the patients examined by consultation; the respective reasons are listed in Figure 1. To ensure the representativeness of the consenting patients, they were compared with the excluded patients on the basis of seven routine variables: the excluded patients were older and had more diagnoses requiring treatment and additional medications; there were no differences in gender, migration background or length of stay. Thus, on the basis of these variables, it can be said that the sample can only be considered representative to a limited extent.

Assessed variables

Different categories of variables were examined:

comparison of groups: internal medicine patients who were or were not referred to the consultation service (Tables 1 and 2);

Table 1.

Multidimensional profile of the sample and differences between cases and controls.

	Whole sample(n = 130)		Cases (referred patients; n = 49)	Controls (non-referred; n = 81)	Differences (Mann-Whitney U/Chi-square; n = 130)
	M (SD)	Test for normality	M (SD)	M (SD)	z or Chi2	p	Cohen’s d or Cramer’s v
Socio-demographic variables
Age	62.3 (17.9)	<0.001	60.4 (17,8)	63.5	0.94	n.s.	0.17
Gender (proportion women)	54%		69.4%	44.4%	7.64	0.006	0.24
Migrant background	21%		18.4%	22.2%	0.27	n.s.	0.05
Graduation (lower sec. education/high school),	55/17%		67/12%	52/20%	3.89	n.s.	0.17
Vocational qualif. (none/apprenticeship/graduation)	30/57/13%		33/57/10%	28/57/15%	0.68	n.s.	0.07
Occupation (unemployed/employed/pension)	7/25/61%		14/16/54%	3/31/64%	15.3	0.002	0.34
Living alone	39%		51%	32%	4.58	0.032	0.19
Medical variables
Duration of current disease (days)	15.4 (22.9)	<0.001	20.9 (24.3)	12.0 (21.5)	3.33	<0.001	0.39
Number diseases requiring treatment	4.4 (2.6)	<0.001	4.0 (2.2)	4.6 (2.9)	0.91	n.s.	0.27
Current number of drugs	5.2 (3.8)	<0.001	4.8 (3.1)	5.4 (4.2)	0.61	n.s.	0.16
Utilisation of medical resources
Outpatient medical appointments (last 6 months)	5.7 (7.7)	<0.001	59 (5.9)	5.7 (.6)	0.76	n.s.	0.02
Number of hospitalisations (last 6 months)	0.98 (1.77)	<0.001	1.35 (2.02)	0.76 (1.57)	1.76	n.s.	0.33
Number of hospitalisation days (last 6 months)	7.1 (13.5)	<0.001	8.5 (15.1)	6.2 (12.5)	0.92	n.s.	0.16
Number of hospitalisations in lifetime	8.4 (13.3)	<0.001	9.2 (16.0)	7.9 (11.4)	0.36	n.s.	0.09
Length of stay (current hospitalisation)	7.6 (7.3)	<0.001	10.5 (9.7)	5.8 (4.5)	3.92	<0.001	0.69

M: mean; SD: standard deviation; Test for normality: Shapiro-Wilk test, p < 0.05 indicates non-normal distribution of variable’s values; Mann-Whitney: non-parametric test for comparison of non-matched pairs; z: non-parametric test value normalised; Chi²: chi-square test value; p: level of significance of tested differences; Cohen’s d: effect size of rank differences for metric variables; Cramer’s V: effect size of different distribution for categorical variables; n.s.: not statistically significant at a 0.05 level.

Table 2.

Psychosomatic profile of sample and differences between cases and controls.

	Whole sample(n = 130)		Cases(referred patients; n = 49)	Controls(non-referred; n = 81)	Differences (Mann-Whitney or Chi-square; n = 130)
	M (SD)	Test for normality	M (SD)	M (SD)	z orChi²	p	Cohen’s d or Cramer’s V
Psychiatric variables
Psychopathological burden (SCL-K)	11.7 (7.4)	<0.001	14.3 (7.1)	10.2 (7.3)	3.41	<0.001	0.57
Anxiety (HADS-A)	7.5 (4.8)	<0.001	9.8 (5.1)	6.2 (4.1)	4.17	<0.001	0.80
Depression (HADS-D)	7.1 (5.1)	<0.001	9.4 (5.4)	5.6 (4.2)	3.91	<0.001	0.80
Current number of psychopharmaceuticals	0.5 (0.8)	<0.001	0.92 (0.95)	0.21 (0.47)	517	<0.001	1.02
Pesonality dimension (BFI-10)
Extraversion	2.2 (1.0)	n.s.	2.13 (1.08)	2.21 (0.94)	0.46	n.s.	0.08
Agreeableness	2.4 (0.8)	n.s.	2.43 (0.89)	2.34 (0.77)	060	n.s.	0.11
Consciousness	2.9 (0.8)	n.s.	2.96 (0.80)	2.83 (0.82)	0.90	n.s.	0.16
Neuroticism	2.1 (0.8)	n.s.	2.21 (0.83)	1.98 (0.82)	1.56	n.s.	0.28
Openess	2.3 (0.8)	n.s.	2.46 (0.79)	2.30 (0.91)	1.00	n.s.	0.18
Psychological resources
Self-efficacy (GSW-6)	11.1 (3.9)	0.013	9.98 (4.37)	11.8 (9.54)	2.57	0.011	0.46
Physical quality of life	36.8 (10.3)	0.008	33.6 (8.9)	38.7 (10.7)	2.65	0.008	0.59
Psychological quality of life	38.9 (6.7)	0.003	38.6 (6.6)	39.1 (6.8)	0.85	n.s.	0.08

M: mean; SD: standard deviation; Test for normality: Shapiro-Wilk test, p < 0.05 indicates a non-normal distribution of variable’s values; Mann-Whitney: non-parametric test for comparison of non-matched pairs; z: non-parametric test value normalised; Chi²: chi-square test value; p: level of significance of tested differences; Cohen’s d: effect size of rank differences for metric variables; Cramer’s V: effect size of different distribution for categorical variables; n.s.: not statistically significant at a 0.05 level.

psychosocial, medical and utilisation variables (Table 1);

from a basic psychosomatic perspective: relevant symptom burden variables, personality dimensions and psychological resources (self-efficacy and quality of life; Table 2);

the appraisal of physical and psychological burden as well as further need for psychiatric support by the doctor, nurse and patient him-or herself (Table 3), using self-developed items. The patients were asked to rate their current burden on a 10-point scale (“How do you rate the current physical/psychological burden from 0 [none] to 10 [extreme]?”). This assessment was also made for the same patient by the caregiver and the consultant. Finally, an assessment of the need for psychiatric support beyond standard care was made. This assessment was also made on a scale of 0 to 10, whereby “0” indicated no further treatment, while “10” stood for an assessment according to which continuous, intensive psychiatric/psychotherapeutic treatment would be necessary, without which the interviewed patient would suffer harm. This assessment was also made by the patient, the caregiver and the psychiatric expert (consultant).

Table 3.

Differences among doctors, nurses and patients in the appraisal of physical and psychological strain as well as need for support by consultation liaison service (CLS).

	Subsamples			Differences between perspectives for each subsample(Wilcoxon matched-pair tests)
	Cases	Controls	Difference	For cases (n = 49)			For controls (n = 81)
	n = 49	n = 81	N = 130	Doctor-Nurse	Doctor-Patient	Nurse-Patient	Doctor-Nurse	Doctor-Patient	Nurse-Patient
	M (SD)	M (SD)	p (Cohen’s d)	z (p)	z (p)	z (p)	z (p)	z (p)	z (p)
Appraisal of physical strain
Doctor	5.28 (2.27)	4.76 (1.87)	n.s. (0.26)	n.s.	2.11 (0.034)	2.74 (0.006)	n.s.	2.07 (0.038)	2.85 (0.004)
Nurse	4.90 (2.36)	4.41 (1.98)	n.s. (0.23)
Patient	6.29 (2.72)	5.38 (2.48)	n.s. (0.35)
Appraisal of psychological strain
Doctor	6.67 (1.34)	4.51 (2.29)	<0.001 (1.09)	n.s.	n.s.	n.s.	2.26 (0.023)	n.s.	2.62 (0.009)
Nurse	6.12 (1.89)	3.84 (2.14)	<0.001 (1.11)
Patient	5.92 (2.84)	4.77 (2.75)	0.024 (0.41)
Appraisal of support need by CLS
Doctor	6.84 (2.05)	3.59 (2.61)	<0.001 (1.34)	n.s.	n.s	n.s.	2.09 (0.036)	n.s.	n.s.
Nurse	6.77 (2.03)	2.90 (2.37)	<0.001 (1.72)
Patient	6.44 (2.47)	2.92 (2.63)	<0.001 (0.37)

M: mean; SD: standard deviation; Wilcoxon matched-pair test: non-parametric test for comparison matched pairs; z: non-parametric test value normalised; p: level of significance of tested differences; Cohen’s d: effect size of rank differences for metric variables; CLS: consultation liaison service; n.s.: non statistically significant at a 0.05 level.

Description of the psychometric tools

Generalized Self-Efficacy (GSW-6)

This scale measures the optimistic expectation of competence: confidence in oneself to master a difficult situation. The test consists of six items that are scored on a Likert scale from “not at all true” (1) to “exactly true” (4). The sum score ranges from 4 to 24.²¹

Big Five Inventory (BFI-10)

This tool comprises the five-dimensional assessment of personality (Extraversion, Neuroticism, Agreeableness, Openness, Consciousness). The test consists of 10 questions (two per dimension, one with reversed polarity) on a Likert scale that ranges from “does not apply at all” (1) to “applies completely” (5). The scale values are averaged. The range per scale is 1–5; there are normalised population values.²²

Hospital Anxiety and Depression Scale (HADS)

This self-assessment scale is used as a screening procedure for assessing in somatic patients in general hospitals. The HADS consists of 14 questions, seven each for the anxiety (HADS-A) and depression (HADS-D) subscales. Each item on the questionnaire is scored from 0 to 3 (0 = not at all; 1 = from time to time, occasionally; 2 = a lot of the time; 3 = most of the time); each scale ranges from 0 to 21 points. A sum > 10 can be considered abnormal.^23,24

Symptom Checklist, Short Version (SCL-K-9)

The psychopathological burden of the patients was assessed using the German short version (SCL-K-9) of the “Symptom Checklist 90” (SCL-90-R).²⁵ The patients were asked to assess the extent to which they had experienced various complaints within the last 7 days. The SCL-K-9 consists of one item each of the nine subscales of SCL-90-R (somatisation, obsessive-compulsivity, insecurity in social contact, depression, anxiety, aggression/hostility, phobic anxiety, paranoid thinking and psychoticism) and thus comprises a total of nine items (an example item on anxiety: “How much did you suffer from the feeling of being tense or agitated in the last 7 days?’). The answers to the items are given on a five-level rating scale (from 0 = “not at all” to 4= “very strong”).²⁶

Short Form-12 of the Health Survey-36 (SF-12)

This is a two-dimensional instrument for measuring health-related quality of life.²⁷ The test consists of 12 items, six each for a basic physical and psychological dimension of health-related quality of life. The questions are dichotomous (items 4–7), or they have three values (2 and 3), five values (items 1, 8 and 12) or six values (9–11). For each question, a dummy variable is calculated and weighted with a physical factor and a psychological factor. The sum of the six somatic values plus a somatic constant results in a somatic cumulative value (KSK), and the six psychological values plus a psychological constant results in a psychological cumulative value (PSK).²⁸

Statistics

First, the metric variables were examined for a normal distribution pattern using the Shapiro-Wilk test. Only the personality dimension variables were normally distributed (see Tables 1 to 3), so nonparametric tests were used for comparisons and associations. To answer the first question, unpaired Mann-Whitney U tests (or Student’s t-tests for the personality dimensions) were used to compare the subsamples on the basis of metric variables and chi-square tests for nominal variables. The effect size of the differences was calculated using Cohen's d for metric variables and Cramer's V for categorical variables (Tables 1 and 2). To answer the second question regarding the concordance or discordance of perspectives among doctors, nurses and patients, Wilcoxon matched-pair tests were used to explore the extent of appraisal differences, and bivariate regression analyses were employed to investigate the tendency of concordance among perspectives (Table 3 and Figure 2). The third question was examined using median regression because the dependent variables were not normally distributed (tests on mental stress). Eleven independent variables were selected for adjustment. The main independent variable was the group variable, namely referred vs. non-referred patients. “Age”, “gender” and “living alone” were selected as the clinically relevant psychosocial variables, while the “number of diagnoses requiring treatment”, “patient perspective on their own psychological and physical burden” and “duration of the current episode of illness” were chosen as clinical burden variables. The “number of hospital stays throughout life” was selected as a surrogate variable for a chronic course, while “neuroticism” and “self-efficacy” were selected as the personality variables relevant in the literature for a somatic course (Table 4). Bootstrapped median regression delivered by means of robust estimates and standard errors presented similar significance levels for associations compared to regular median regression (not displayed). The authors calculated the Post-hoc power (1-β) assuming effect size f²=0.15; α = 0.05; n = 130; and 11 predictors. The power was 0.83 .

Figure 2.

Bivariate linear regression analyses between doctor’s and patient’s appraisal of physical strain (left), psychological strain (middle) and need for consultation (right). The upper row corresponds to controls, while the lower row corresponds to cases. The straight line represents a fitted regression, while the sinuous line indicates a smoother (lowess) for residuals. A significant association occurred when p < 0.05, linearity when hatsq (Pregibon test) > 0.05.

Table 4.

Median multivariate regression analyses for target psychological and quality of life variables.

	Psychological burden (SCL-K-9)			Anxiety (HADS-A)					Depression (HADS-D)				Physical QoL (SF-12-K)			Psychological QoL (SF-12-P)
	coef	p	95% CI	coef	p		95% CI		coef	p	95% CI		coef	p	95% CI	coef	p	95% CI
Group (referred vs. no-referred)	n.s.			2.33	0.027		0.27 to 4.38		2.72	0.006	0.80 to 4.64		n.s.			n.s.
Age	n.s.			n.s.					n.s.				n.s.			n.s.
Gender (1 = men; 0 = women)	n.s.			n.s.					n.s.				n.s.			n.s.
Living alone	n.s.			n.s.					n.s.				n.s.			n.s.
Number diagnoses	n.s.			n.s.					n.s.				n.s.			n.s.
Physical strain (pat.)	n.s.			n.s.					n.s.				−1.09	0.025	−2.05 to −1.14	n.s.
Psychological strain (pat.)	0.73	0.004	0.23 to 1.23	0.54	0.008		0.14 to 0.94		0.51	.007		0.14 to 0.88	n.s.			n.s.
Duration current disease	0.08	0.003	0.03 to 0.13	n.s.					n.s.				n.s.			n.s.
Hospitalisations lifetime	0.16	0.001	0.07 to 0.25	n.s.					n.s.				n.s.			n.s.
Neuroticism (BFI-N)	1.51	0.047	0.02 to 3.00	n.s.					n.s.				n.s.			−1.76	0.047		−3.52 to −0.02
Self-efficacy (GSW-6)	−0.36	0.021	−0.67 to −0.05	−0.38		0.003		−0.63 to −0.13	−0.46	<0.001		−0.68 to −0.22	0.62	0.023	0.09 to 1.16	n.s.
N	129			130					130				130			130
RSD/MSD	763/490			487/354					542/341				1060/843			665/622
Pseudo R²	0.35			0.27					0.37				0.20			0.06
Post-hoc power(1-β)	0.83			0.83					0.83				0.83			0.83

SCL-K-9: Symptom Checklist, short version; HADS: Hospital Admission and Depression Scale; SF-12-K: Short form-12 of the Health Survey-36; QoL: quality of life; Coef: median regression coefficient; p: level of statistical significance of association between independent and dependent variables; 95% CI: confidence interval at a 95% level; n.s.: non statistically significant at a 0.05 level; BFI-N: Big Five Inventory-10, dimension Neuroticism; GSW-6: self-efficacy test “Generalisierte Selbstwirksamkeit”; (pat) : patient’s appraisal; N= observations in the model with values for all variables; RSD = raw sum of deviations; MSD: Minimum sum of deviations; Pseudo R²: model fit according to McFadden. Post-hoc power (1-β) calculated assuming effect size f2 = 0.15; α = 0.05; n = 130; 11 predictors.

Results

The average age of the participants was over 60. Further, more than a third were living alone (39%), and there was almost an equal number of males and females. Most of the subjects had an intermediate vocational education and training (57%) and were pensioners (61%). On average, the surveyed patients showed more than four diseases that required treatment, had almost six outpatient appointments in the last half year and more than eight hospitalisations in their lifetime (Table 1). The personality dimensions ranged from 2.1 (Neuroticism) to 2.9 (Consciousness). The psychological burden measured by means of a very short version of SCL-90 was 11.7 (range: 0–36), by HADS-A it was 7.5 (range: 0–21) and by HADS-D it was 7.1 (range: 0–21). The average self-efficacy score was 11.1 (range: 6–24), physical quality of life was 36.8 (range: 10.6–64.4) and psychological quality of life was 38.9 (range: 8.5–70.1 in a German representative norm sample).

Compared with the non-referred patients, the referred ones were mainly women (69.4% vs. 44.4%), less employed (16% vs. 31%), often living alone (51% vs. 32%), had been suffering from the current disease for a longer time (20.9 vs. 12.0 days), showed an overall greater psychological burden (general psychopathology, anxiety, depression and number of psychopharmaceuticals), displayed lower self-efficacy (9.98 vs. 11.8) and presented a lower physical quality of life (33.6 vs. 39.1). The effect sizes of differences were only relevant for psychological burden (Cohen’s d 0.57–1.02; Tables 1 and 2). There were no differences concerning personality dimensions, age, migration background, graduation, vocational qualification, psychological quality of life, number of diseases and current drugs as well as all measured medical utilisation variables, with the exception of a significantly longer length of the hospital stay in the group of referred patients (Tables 1 and 2).

Concerning subjective appraisal of psychological as well as physical strain and psychiatric care needs beyond care as usual, the referred patients showed higher levels of strain and psychiatric care needs overall and in the three assessed perspectives compared to the non-referred patients. These differences were only statistically significant for psychological strain and for appraisal of psychiatric support needs among all perspectives. For referred patients, there were only a few significant differences in the pairwise comparison of perspectives: the patients reported a higher level of physical strain compared to nurses and doctor. There were no differences for psychological strain. For non-referred patients, the doctors scored the patients’ psychological strain and psychiatric support needs higher than nurses, and nurses rated the psychological strain higher than patients (Table 3). As a complementary analysis to the score level differences, associations in scoring tendencies were assessed using bivariate regression models between doctor’s and patient’s perspectives. For referred patients, both perspectives exhibited the same tendency independent of the scored level. For non-referred patients, this association was significant for appraisal of psychological strain and psychiatric support needs (Figure 2).

The third objective was to assess possible predictors for psychological burden and quality of life within patients hospitalised in an internal medicine ward. In multivariate median regression models, there were differences between referred and non-referred patients regarding depression and anxiety but not psychopathological burden in general or quality of life (Table 4). The best predictors for higher psychopathological levels were the patients’ appraisal of their own psychological strain and lower self-efficacy. Most predictors were found for general psychopathological burden assessed by a very short form of SCL-90 (Table 4). The best model fit was identified for general psychological burden (pseudo R² = 0.35) and depression (pseudo R² = 0.37).

Discussion

The most relevant results in this study revealed that patients in an internal medicine ward who were referred to a CLP displayed an adverse psychosocial profile and were psychologically burdened. Nevertheless, non-referred patients also showed a negative psychosomatic profile that comprised subsyndromal anxiety and depression, low psychological quality of life, lower levels of self-efficacy and relevant subjective psychological as well as physical strains.

Considering the seven analysed variables, the current sample is only minimally representative because the non-participants are older and medically more burdened. The fact that most metric variables were not normally distributed is due to the general unfavourable medical conditions that led to hospitalisation. Therefore, this sample does not represent the general population. The whole sample showed a negative psychosocial profile, particularly denoted by the fact that 39% of all participants live alone, especially referred patients (51%). Loneliness is a risk factor for psychiatric morbidity and implies an unfavourable prognosis for medical conditions.^29–31 A patient’s physical burden and medical history are congruent with an indication for hospital admission. All personality dimensions in the sample showed lower scores compared to a survey in the general population.²² The consciousness and neuroticism results might be an artifact due to the reduction of vitality because of acute physical strain. The physical as well as the psychological quality of life were reduced in the whole sample because of weakening due to medical conditions, age and an unfavourable psychosocial profile.

This investigation demonstrated descriptively—as well as by means of bivariate and multivariate tests—that patients in an internal medicine ward referred to a CLP service were in fact psychosocially and psychologically strained. Thus, the referrals were performed accurately; the lower self-efficacy of patients and the loneliness may reinforce the appraisal by doctors of need of a psychiatric consultation. On the contrary, one cannot assume that non-referred patients were not burdened because the assessed variables indicated subsyndromal psychological strain and no differences compared to referred patients with regard to psychological quality of life. In addition, the subjective appraisal of the doctors, nurses and patients demonstrated a relevant burden and psychiatric support needs independent of the provided resources. The length of the hospital stay was on average double in the referred compared to non-referred patients. These data indicate that at least perceived mental care requires a longer hospital stay, as demonstrated in a wide range of investigations,^32,33 but displaying a broader spread due to the very individual conditions. In multivariate models - which are more restrictive due to statistical adjustment - there were no statistical differences between referred and non-referred patients in general psychopathological burden measured by the SCL-K-9 or in the physical and psychological quality of life measured by a short version of SF-36. These findings suggest that hospitalised patients in an acute internal medicine ward are on average relevantly burdened. Further, the CLP referral decision depends on objective symptoms as well as subjective appraisal of adverse psychosocial conditions, interpersonal resonance and helplessness. Indeed, lower self-efficacy was the best predictor in multivariate models for psychological burden and lower quality of life.

There are several fundamental questions that arise from these results. Are cut-offs adequate for organising consultation psychiatry in general hospitals? Are psychosocial variables of similar importance for care decisions than psychopathology? Considering that non-referred patients still displayed a relevant psychological burden and psychiatric care needs, how can general hospitals and CLP services ensure equitable care? Which psychiatric resources are necessary to ensure that care is provided in accordance with real needs and how should they be used most effectively? Finally, how can clinicians and caregivers shape an effective transfer of psychosocial recommendations from hospital to general practitioners in order to ensure a secondary or tertiary prevention in follow-up? This study demonstrates, that reports of patients about their psychological strains are equally relevant alongside the viewpoint of physicians and the fact that nurses accurately grasp psychological burden and care needs. This result is given both for indicative referrals to CLS and for not referred patients; this information has to be considered in further risk assessment and care planning for all patients, especially after discharge. The preponderance of barriers and the scarcity of facilitators, as well as the care continuity problems at the interface between primary and secondary care, are well identified.⁸ Comorbid patients in general hospitals deserve more than what might be lost in transition. Therefore, a smooth transfer of information and advises about psychological frailty and needs of patients to community services is of importance, not only for treatment adherence, but perhaps also for secondary and tertiary prevention, especially to general practitioners in charge of the patient, as already suggested in other investigations.^34–37

The most important limitations for the study are the partial representativity of the sample and its small size due to the difficulty in engaging acutely ill and older patients in an extensive investigation; despite of this limitation, non-participants were assessed by means of some routine variables and compared with participants. The scarce information about mental disorder and psychiatric or psychotherapeutic treatment prior to hospitalisation can be considered as a limitation because these variables cannot be included as control variables. Of epidemiological interest is to assess the difference between reported psychological strain or mental illness before admission and after consultation in order to capture the additional psychological strain due to hospitalisation itself or because of prior underdiagnosing. Referred and not-referred patients could be better compared in larger samples including more control variables or condensing then in propensity scores. In order to balance protective and risk factors, structural equation modelling might be adequate. Longitudinal studies will be important in order to identify factors for a better transfer, adherence to treatment and prognosis comparing it with the CLS assessment during hospitalisation.

Footnotes

Authors’ note

Since the study goes beyond the collection of routine data, approval was obtained from the ethics committee of the University of Ulm (file number 335/16) and also from the local ethics committee of the Hospital Friedrichshafen (Germany).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Juan Valdés-Stauber

References

Stern

Freudenreich

Smith

, et al. (eds). Massachusets general hospital handbook of General Hospital Psychiatry. 7th ed. Edinburgh, London, New York: Elsevier, 2018, pp.1–22.

Levenson

(ed.). Psychosomatic medicine and consultation-liaison psychiatry. 3rd ed. Washington: APA Publishing, 2019, pp.3–24.

Fogel

Greenberg

(eds). Psychiatric care of the medical patient. 3rd ed. Oxford: Oxford University Press, 2015, pp.3–19.

Chen

Evans

Larkins

Why are hospital doctors not referring to consultation-liaison psychiatry? A systematic review. BMC Psychiatry 2016; 16:390.

Lavakumar

Gastelum

Choo

, et al. Parameters of consultee satisfaction with inpatient academic psychiatric consultation service: a multicenter study. Psychosomatics 2015; 56:262–267.

Sánchez-González

Baillès-Lázaro

Bastidas-Salvadó

, et al. Clinical profile of inpatients referred to a consultation-liaison psychiatry service: an observational study assessing changes over a 10-year period. Actas Esp Psichiatr 2018; 46:58–67.

Walker

Barrett

Lee

, et al. Organisation and delivery of liaison psychiatry services in general hospitals in England: results of a national survey. BMJ Open 2018; 8:e023091.

Jasmin

Walker

Guthrie

, et al. Integrated liaison psychiatry services in England: a qualitative study of the views of liaison practitioners and acute hospital staffs from four distinctly different kinds of liaison service. BMC Health Serv Res 2019; 19:522.

Guthrie

McMeekin

Thomasson

, et al. Opening the “black box”: liaison psychiatry services and what they actually do. BJPsych Bull 2016; 40:175–180.

10.

Leentjens

Rundell

Diefenbacher

, et al. Psychosomatic medicine and consultation-liaison psychiatry: scope of practice, processes, and competencies for psychiatrists working in the field of CL Psychiatry or Psychosomatics. A consensus statement of EACLPP and APM. Psychosomatics 2011; 52:19–25.

11.

House

Guthrie

Walker

, et al. A programme theory for liaison mental health services in England. BMC Health Serv Res 2018; 18:742.

12.

Rothenhäusler

HB.

Mental disorders in general hospital patients. Psychiatr Danub 2006; 18:183–192.

13.

Wancata

Meise

Sachs

Prevalence of psychiatric disorders in elderly inpatients of medical and surgical departments. Psychiat Prax 2000; 27:170–175.

14.

Welch

McCluskey

Wilson

, et al. Geriatric Medicine Research Collaborative. Delirium is prevalent in older hospital inpatients and associated with adverse outcomes: results of a prospective multi-centre study on World Delirium Awareness Day. BMC Med 2019; 17:229.

15.

Niecke

Lemke

Goeser

, et al. Comorbid mental disorders in gastroenterology and hepatology: prevalence and psychosocial needs in tertiary care. Psychother Psych Med 2019; 69: 29–37.

16.

Fink

Hansen

Søndergaard

, et al. Mental illness in new neurological patients. J Neurol Neurosurg Psychiatry 2003; 74: 817–819.

17.

Kuhnt

Brähler

Faller

, et al. Twelve-month and lifetime prevalence of mental disorders in cancer patients. Psychother Psychosom 2016; 85: 289–296.

18.

Wood

Wand

AP.

The effectiveness of consultation-liaison psychiatry in the general hospital setting: a systematic review. J Psychosom Res 2014; 76: 175–192.

19.

Hansen

Fink

Frydenberg

, et al. Mental disorders among internal medical inpatients: prevalence, detection, and treatment status. J Psychosom Res 2001; 50: 199–204.

20.

Fink

Hansen

Oxhøj

M-L.

The prevalence of somatoform disorders among internal medical patients. J Psychosom Res 2004; 56: 413–418.

21.

Romppel

Herrmann-Lingen

Wachter

, et al. A short form of the General Self-Efficacy Scale (GSE-6): development, psychometric properties and validity in an intercultural non-clinical sample and a sample of patients at risk for heart failure. Psychosoc Med 2013; 10Doc01. doi: 10.3205/psm000091.

22.

Rammstedt

The 10-Item Big Five Inventory (BFI-10): norm values and investigation of sociodemographic effects based on a German population representative sample. Eur J Psychol Assess 2007; 23: 193–201.

23.

Snaith

Zigmond

AS.

Hospital Anxiety and Depression Scale (HADS). In: AJ Rush Jr. (Ed.) Handbook of psychiatric measures. Washington, DC:

American Psychiatric Association, 2000, pp.547–548.

24.

Herrmann-Lingen

Buss

Snaith

RP.

HADS-D Hospital Anxiety and Depression Scale – German Version. 3rd ed. Bern: Hans Huber-Hogrefe, 2011.

25.

Derogatis

LR.

SCL-90-R: Administration, scoring & procedures manual for the R(evised) version. Baltimore: Johns Hopkins University, School of Medicine, 1977.

26.

Klaghofer

Brähler

Konstruktion und teststatistische prüfung einer kurzform der SCL-90–R (Construction and statistical testing of a short form of the SCL-90-R). Z für Klin Psychol Psychiatr Psychother 2001; 49: 115–124.

27.

Ware

Kosinski

Turner-Bowker

, et al. How to score version 2 of the SF-12 Health Survey. Lincoln, RI: QualityMetric, 2002.

28.

Andersen

Mühlbacher

Nübling

, et al. Computation of standard values for physical and mental health scale scores using the SOEP version of SF-12v2. J Appl Soc Sci Stud 2007; 127: 171–182.

29.

Richard

Rohrmann

Vandeleur

, et al. Loneliness is adversely associated with physical and mental health and lifestyle factors: results from a Swiss national survey. PLoS One 2017; 12: e0181442.

30.

Beutel

Klein

Brähler

, et al. Loneliness in the general population: prevalence, determinants and relations to mental health. BMC Psychiatry 2017; 17: 97.

31.

Caarls

van Schijndel

Berk

GVD

, et al. Factors influencing the admission decision for Medical Psychiatry Units: a concept mapping approach. PLoS One 2019; 14: e0221807.

32.

Sockalingam

Alzahrani

Meaney

, et al. Time to consultation-liaison psychiatry service referral as a predictor of length of stay. Psychosomatics 2016; 57: 264–272.

33.

Vulser

Vinant

Lanvin

, et al. Association between the timing of consultation-liaison psychiatry interventions and the length of stay in general hospital. Br J Psychiatry Epub ahead of print 13 November 2019. doi: 10.1192/bjp2019.233.

34.

Richards

Bower

Chew-Graham

, et al. Clinical effectiveness and cost-effectiveness of collaborative care for depression in UK primary care (CADET): a cluster randomised controlled trial. Health Technol Assess 2016; 20: 1–192.

35.

O’Neil

Sadosty

Pasupathy

, et al. Hours and miles: patient and health system implications of transfer for psychiatric bed capacity. WestJEM 2016; 17: 783–790.

36.

Velligan

Roberts

Martinez

, et al. Following AACP guidelines for transitions in care: the transitional care clinic. Psychiatr Serv 2016; 67: 259–261.

37.

Reeves

Pelone

Harrison

, et al. Interprofessional collaboration to improve professional practice and healthcare outcomes. Cochrane Database Syst. Rev 2017; 6: CD000072.