Scientific Papers

Patient characteristics, validity of clinical diagnoses and Outcomes Associated with Suicidality in Inpatients with Symptoms of Depression (OASIS-D): design, procedures and outcomes | BMC Psychiatry


Study setting

The investigator-initiated OASIS-D study is coordinated by Charité—Universitätsmedizin, Campus Virchow-Klinikum, Department of Child and Adolescent Psychiatry, Berlin, Germany, and funded by Janssen-Cilag GmbH.

Patients are being recruited from the following eight adult mental health centers: Three centers at the Charité – Universitätsmedizin Berlin, i.e. (i) St. Hedwig Hospital, Department of Psychiatry and Psychotherapy, (ii) Charité Campus Mitte, Department of Psychiatry and Psychotherapy, and (iii) Charité Campus Benjamin Franklin, Department of Psychiatry and Psychotherapy, (vi) Clinic for Psychiatry and Psychotherapy at the Ludwig-Maximilians-University Munich, (v) Department of Psychiatry, Psychosomatic Medicine and Psychotherapy at the University Hospital Frankfurt, (vi) Department of Psychiatry and Psychotherapy at the University of Cologne, (vii) Department of Psychiatry and Psychotherapy at the University Hospital Carl Gustav Carus in Dresden, as well as (viii) Central Institute of Mental Health, Department of Psychiatry and Psychotherapy in Mannheim.

Primary inclusion criteria

Inclusion criteria for PP1 are (i) male and female patients; (ii) aged between 18 and 75 years; (iii) admitted to a psychiatric inpatient unit, and (iv) having an ICD-10-based chart diagnosis of single depressive episode (F32) or recurrent depressive episode (F33).

Inclusion criteria for PP2 (originating from PP1) are (i) male and female patients; (ii) aged between 18 and 75 years; (iii) admitted to a psychiatric inpatient unit; (iv) having an ICD-10-based clinical diagnosis of a single or recurrent depressive episode of at least moderate severity, including single depressive episode, moderate episode (F32.1), single depressive episode, severe episode without psychotic symptoms (F32.2), single depressive episode, severe episode with psychotic symptoms (F32.3), recurrent depressive episode, currently moderate episode (F33.1), recurrent depressive episode, currently severe episode without psychotic symptoms (F33.2), and recurrent depressive episode, currently severe episode with psychotic symptoms (F33.3); (v) clinically diagnosed suicidality that persists for ≥ 48 h after inpatient admission; and (vi) sufficient German language skills to understand the purpose and procedures of the study and cooperate with the study procedures; and (vii) providing written informed consent to participate in the study.

Inclusion criteria for PP3 (originating from PP2) are (i) male and female patients; (ii) aged between 18 and 75 years; (iii) admitted to a psychiatric inpatient unit; (iv) having an ICD-10-based research diagnosis according to the Mini-International Neuropsychiatric Interview (M.I.N.I.) [71] of a single or recurrent depressive episode of at least moderate severity (for details, see PP2 above); (v) research interview-based suicidality with a score ≥ 1 on item 2 (passive suicidality) OR item 3 (active suicidality) using the Sheehan–Suicidality Tracking Scale (S-STS) [21]; (vi) written informed consent to participate in the study.

Exclusion criteria

Exclusion criteria for PP1 are (i) psychiatric inpatients younger than 18 years and older than 75 years; (ii) no clinical diagnosis of a unipolar depressive episode; (iii) depressive episode in the context of bipolar disorder.

Exclusion criteria for PP2 and PP3 are (i) psychiatric inpatients younger than 18 years and older than 75 years; (ii) depression secondary to a physical illness, such as hypothyroidism or vitamin B12 deficiency, or substance use; (iii) severe physical illness symptoms that preclude participation in the study; (vi) unable to understand the study procedures; (v) unable to give informed consent; (vi) unable to give consent due to (limited) legal capacity.

Aims and hypotheses

The primary aim in PP1 is to systematically characterize patient, illness, and treatment characteristics of consecutively admitted patients with a clinical diagnosis of a single or recurrent unipolar depressive episode.

Hypothesis 1

We hypothesize that MDD at inpatient admission is characterized by (i) most frequent referral by a primary care physician, (ii) female patient predominance of about 60% [5, 6], (iii) older patients aged approx. 45 years [5, 6], (iv) admission due to psychiatric emergency in approx. 25% of the sample, (v) occurrence of treatment-resistant depression in approx. 20% of the sample [11], (vi) occurrence of at least passive suicidal ideation in at least 50% of the sample [6, 72], (vii) lifetime history of a suicide attempt in approx. 20% of the sample [14], and (viii) preponderance of medication treatment with selective serotonin reuptake inhibitors (SSRIs) [9, 10].

Hypothesis 2

The primary aim in PP2 is to test the validity of the clinical diagnosis of at least moderately severe MDD in patients with clinically documented suicidality using the M.I.N.I. and compare predictors (diagnosis, demographic and illness characteristics, treatment) between patient groups with concordant and discordant clinical and research diagnosis.

Hypothesis 2A: We hypothesize that in < 50% of patients the clinical diagnosis will not be confirmed by the structured research diagnosis [38].

Hypothesis 2B: We hypothesize that the patient group with unvalidated vs validated clinical diagnosis of at least moderately severe MDD by the research interview compared will be associated with (i) younger age (less historical information and more dynamic emergence of (co)morbidities) [39], (ii) female sex (possible diagnostic bias), (iii) lower number of previous depressive episodes (less historical validation), (iv) lower number of index admission in the context of a psychiatric emergency (less historical validation), (v) higher number of comorbid personality disorders (more diagnostic overlap) [40, 41], and (vi) lower suicidal severity (less severe psychiatric condition).

Hypothesis 3

The primary aim of PP3 is a 6-month structured recording of the duration and frequency persistence and recurrence of suicidality and its correlates in 315 patients diagnosed with at least moderately severe MDD by the M.I.N.I. and with confirmed presence of at least passive suicidal ideas by the S-STS ≥ 48 h after inpatient admission.

Hypothesis 3A: We hypothesize that the time to remission of suicidal symptoms after the baseline is a major determinant of shorter overall duration of passive or active suicidal ideation (PASI) during the entire 6-month observation period.

Hypothesis 3B: We hypothesize that the rate of remission of suicidal symptoms after baseline is a major determinant of lower recurrence of PASI during the 6-month observation period.

Hypothesis 3C and 3D: We hypothesize that a longer total duration of PASI (Hypothesis 3C) and a greater risk of recurrence of PASI after complete remission (Hypothesis 3D) during the 6-month observational phase is associated with the following factors at time T0 (i.e., “moderators”): greater severity of (i) depressive symptoms [28, 36, 42, 43], (ii) manic symptoms (below the threshold of (hypo)mania as part of MDD), (iii) anxiety symptoms [28, 36, 45], (iv) anger and hostility [43], (v) quality of life impairment [48, 49], (vi) global symptoms of illness and (vii) suicidality [28, 36, 42, 47], (viii) higher number of lifetime depressive episodes, (viii) index admission as part of a psychiatric emergency, (ix) longer duration of current depressive episode [44], (x) higher non-adherence to pharmacological and non-pharmacological treatments, (xi) comorbid borderline personality disorder [35, 46], and (xii) comorbid substance use diagnosis [28].

Exploratory study aim 1 of PP3 is to describe the association between PASI and suicide attempts (S-STS item 14) as well as with the utilization of outpatient and inpatient care services during the 6-month observation period.

Exploratory hypothesis 1A, 1B and 1C: We hypothesize that a higher number of days with PASI (exploratory hypothesis 1A), a higher number of days with active suicidal ideation (exploratory hypothesis 1B) and a higher number of suicide attempts (exploratory hypothesis 1C) during the 6-month observation period are significantly associated with a higher number of (i) outpatient appointments, (ii) emergency department visits, (iii) inpatient hospitalizations, (iv) hospitalizations as part of psychiatric emergencies, (v) psychiatric hospital bed days, and (vi) suicide-related hospital bed days.

Exploratory study aim 2 of PP3 is to assess risk factors for different levels of active suicidality, consisting of active suicidal ideation (S-STS item 3), active suicidality with preparatory action (S-STS 12), self-injurious behavior without suicidal intent (S-STS item 13), or suicide attempts (S-STS item 14).

Exploratory study hypothesis 2: Active suicidality as defined above is associated with (i) slower rate of remission of suicidal symptoms after baseline, greater baseline severity of (ii) depressive symptoms, (iii) manic symptoms (below the threshold of (hypo)mania as part of MDD), (iv) anxiety symptoms, (v) anger and hostility, (vi) quality of life impairment (vii) global illness symptoms, and (viii) suicidality, (ix) higher number of lifetime depressive episodes, (x) duration of current depressive episode, (xi) index admission as part of psychiatric emergencies, (xii) higher non-adherence to pharmacological and non-pharmacological treatments, (xiii) comorbid borderline personality disorder, and (xiv) comorbid substance use diagnosis.

Assessments and timeline

In PP1, sociodemographic, family, illness, and treatment-related data are collected to characterize patients with MDD at the time of inpatient admission (T0), i.e., age (coded in 5 year intervals), sex, ICD-10 F-diagnosis, current and lifetime suicidality, psychiatric emergencies, current and previous treatments, medication resistance, and nonadherence.

In PP2, sociodemographic, illness and treatment data (i.e., sex, Body-Mass-Index (BMI), family psychiatric disorders, education/ work, pathways to admission, F-diagnosis of ICD-10, illness duration, number of episodes, current and lifetime suicidality, psychiatric emergency, current and previous treatment, medication treatment-resistance and non-adherence, substance use) are collected at ≥ 48 h after inpatient admission (T1). M.I.N.I. will be used as an interview to determine the research diagnosis of at least moderately severe MDD, and S-STS will be used to assess suicidality.

In PP3, assessments are performed at time points ≥ 48 h after admission (T1), inpatient discharge (T2), 3 months after T1 (T3, by telephone), and 6 months after T1 (T3). The following investigator interviews and rating scales will be performed: For suicidality the S-STS, for depression the Montgomery-Asberg Depression Rating Scale (MADRS) [73], for global improvement change the Clinical Global Impression-Change (CGI-C) [74], for global severity of illness the Clinical Global Impression-Severity (CGI-S) [75], for imminent suicide risk the Clinical Global Impression of Imminent Suicide Risk (CGI-I), and for resolution of suicide risk the Clinical Global Impression of resolution of Suicide Risk (CGI-SR-R), for mania the Young Mania Rating Scale (YMRS) [76], for personal and social performance the Personal and Social Performance Scale (PSP) [77], for service use the Service Use and Resource Form (SURF) [78], for borderline disorder the ICD-10 borderline criteria (Table 1). Additionally, information obtained from patients regarding psychotropic medications (reason of initiation, discontinuation, switching, effectiveness, safety), electroconvulsive therapy, substance use, and presence of specific psychiatric emergency as reason for hospitalization. Furthermore, the following self-rating questionnaires will be obtained from the patients: Quick Inventory of Depressive Symptomatology-Self Report (QIDS-S) [79], Beck Depression Inventory-II (BDI-II) [80], Beck Anxiety Inventory (BAI) [81], Quality of Life in Depression Scale (QLDS) [82], European Quality of Life Group, 5-Dimension, 5-Level (EQ-5-DL) [83], 36-item Short-Form Health Survey (SF-36) [84], Work Productivity and Activity Impairment (WPAI) [85], Patient Health Questionnaire 9-item (PHQ-9) [86], Beck Hopelessness Scale (BHS) [87], Munich Chronotype Questionnaire (MCTQ) [88], State -Trait Anger Expression Inventory (STAXI) [89], Aggression Questionnaire-Buss Perry (AF-BP) [90]) (Table 1).

Table 1 Assessments and timeline

Sample size and statistical analyses

Regarding PP1, the projected number of approximately n = 3,000 of the epidemiologic sample of hospitalized patients with MDD in the included age range (18–75 years) is based directly on the number of these patients in the eight participating study centers that were sampled in preparation of the study in 2017. Since only descriptive results are analyzed for patient population 1, no formal power analysis and case number calculation are provided. Descriptive statistics will be used to characterize a systematically and consecutively included epidemiologic sample of patients with MDD. Two interim analyses are performed using the same described descriptive statistics in PP1 at the epidemiologic sample size time points of 500 and 2000, respectively. These interim analyses are performed to identify potential data gaps and opportunities for the participating centers to review and address data capture and recording procedures of routine clinical information relevant to the characterization of patients hospitalized with MDD. Due to a higher observed patient dropout (30–35%) in PP3 between discharge and 6-month follow-up than projected in the initial protocol (20%), the subject number in PP3 was increased (see below) and the study duration was extended, which also provides more time for recruitment of patients into PP1, i.e., until the date of the last assessment of the last patient in PP3, increasing the projected number of patients to approximately n = 4500.

The hypotheses pertaining to PP1 will be analyzed exploratorily by using appropriate descriptive statistics, as well as univariate tests (chi-square tests).

Regarding PP2, based on clinical experience and questionnaire information from the recruitment centers as part of the design preparations, an estimated 33% of 3331 patients (n = 1,099) with a clinical diagnosis of at least moderately severe MDD are assumed to be still suicidal ≥ 48 h after admission. Approximately 60% of this group are expected to consent to participate in the interview-based study (and the naturalistic, 6-month follow-up study, should they also fulfill criteria for PP3, determined during participation in PP2). The following analyses of congruence between clinical diagnosis and research diagnosis (gold standard) of at least moderately severe MDD will take place:

  • (a) two groups will be formed (primary analysis), i.e., positive vs. negative validation of the clinical diagnosis with the research-based interview M.I.N.I.,

  • (b) use of descriptive statistics to estimate proportions of patients meeting the gold standard of research diagnosis and their confidence intervals (see a, hypothesis 2A), as defined by diagnostic concordance with the clinical diagnosis at the time of hospitalization; and.

  • (c) to perform multiple mixed logistic regression analysis with backward elimination to compare patients with positively validated and non-validated at least moderately severe MDD (see a, hypothesis 2B), with random factors to account for clustering of patients in centers.

Regarding PP3, the OASIS-D study was initiated with the assumption that approximately 50% of the 520 patients surveyed in PP2 would not meet ICD-10 criteria for the inclusion criteria of at least moderately severe MDD, so that260 patients would be enrolled in PP3. Assuming a 20% drop out rate, this would yield n = 208 with data at 6 months. During the interim analysis 2, in September 2022, we noted a higher drop-out rate of 34% in PP3 than the 20% expected attrition rate at the time of the study design. Due to this finding, the number of patients in PP3 was increased to n = 315, so that with an attrition rate of 34%, the number of patients with data at month 6 would remain at n = 208. Since there was a higher than initially projected transition rate from PP2 to PP3, so that the n of patients entering PP2 did not need to be increased.

Hypotheses 3A and 3C will be analyzed by means of methods for count data, such as Poisson or Negative Binomial models, depending on whether overdispersion is present in the data or not. The dependent variable will be overall duration of PASI during the 6-month observation period and the independent variable of main interest will be time to remission of suicidal symptoms after baseline.

Hypotheses 3B and 3D will be analyzed by using Cox proportional hazard models, where the dependent variable will be time-to-first-recurrence of PASI and the independent variable of main interest will be time to remission of suicidal symptoms after baseline.

Further independent variables for regression analysis will include patient, illness, and treatment variables, including severity of depressive symptoms (MADRS, Quality of Life in Depression Scale (QIDS-S) [79]), manic symptoms (YMRS), anxiety (BAI), anger/hostility (STAXI, AF-BP), global illness symptoms (CGI-S), quality of life (EQ-5-DL), and suicidality (S-STS); presence of psychiatric emergencies; number of lifetime depressive episodes; duration of current depressive episode; nonadherence to pharmacological and nonpharmacological treatments; and comorbid borderline personality and substance abuse. Additionally, a random factor accounting for clustering of patients in centers will be added to the models.

Assuming a minimum requirement of 10 patients per predictor variable in regression analyses [91,92,93] and approximately 20 independent patient, illness, and treatment variables that would be tested as potential correlates of time to remission of suicidality, we estimated that a sample size of at least 200 would be required.

Additionally, as a secondary analysis, we will divide patients into two groups using the median time up to the first S-STS value of zero, which are either below or above the median time to first complete remission of suicidality.

Furthermore, comparisons between the slower and faster remitting group of continuous variables will be performed using multiple comparison procedures (MCPs) with 3 measurement time points (T1, T2, and T4). Because only the primary outcome and very limited other parameters are collected by telephone at T3, data at T3 are not included in these analyses. Because the timing at T2 (hospital discharge) is variable, and this timing is not independent of the outcome measured, the duration between T1 and T2 is included as a covariate in the analyses. Cross-group comparisons of dichotomized or categorical variables are performed using chi-square statistics. For ease of comparison, continuous outcomes, such as depressive symptoms and medication use are calculated. All analyses are two-sided, with alpha = 0.05. Due to the exploratory nature of the study, no adjustment for multiplicity between different endpoints is conducted. Hence, p-values and confidence intervals need to be interpreted in a hypothesis-generating manner. In addition, multivariable regression analysis with backward selection will be performed to identify independent moderators or mediators of continuous and categorical outcomes, respectively. Moreover, missing continuous data will be treated using multiple imputation using chained equations and random effects for the clustering of patients in different centers.

Among the exploratory objectives, the patient, illness, and treatment factors that influence the risk of different levels of active suicidality are identified: i) active suicidality (S-STS item 3), ii) active suicidality with preparatory action (S-STS item 12), iii) self-injurious behavior without suicidal intent (S-STS item 13), and iv) suicide attempt (S-STS item 14) during the observation period of inpatient treatment and within 6 months of study inclusion (T1). Target parameters i)-iv) are associated with slower initial resolution of suicidality (lower than median time to S-STS score = 0); and with: Depressive symptom severity (MADRS, QIDS-S), manic symptoms (YMRS), anxiety (BAI), anger/hostility (STAXI, AF-BP), and global illness symptoms (CGI-S); higher number of lifetime depressive episodes; more frequent psychiatric emergencies; longer duration of current depressive episode; non-adherence to pharmacological and non-pharmacological treatments; comorbid borderline personality; and substance abuse diagnoses.

All analyses will be performed as intent-to-treat analyses, independent of non-adherence with clinically prescribed treatments or drop out from clinical care.

Recruitment and informed consent

Data of PP1 will be obtained via chart review from clinical routine data of patients with MDD (without informed consent) consecutively admitted as inpatients at the eight participating sites. If pseudonymous transmission is prohibited due to the lack of patient consent and the respective state law, the data will be transmitted anonymously, i.e. pooled in an excel file to the study coordinating center.

PP1 serves as a screening for the recruitment of PP2, i.e. individuals in PP2 are recruited from PP1, and individuals in PP3 are recruited from PP2, based on the respective in- and exclusion criteria. Prior to enrollment in PP2, each patient is informed by study staff about the nature, aims, expected benefits and potential risks of the study verbally and in writing. Patients must be given sufficient time and opportunity to decide about study participation. The written informed consent form must be signed by the patient and study staff.

Data management

Data collection takes place via electronic CRFs. For this purpose, the study software secuTrial® of the company interActive Systems GmbH (iAS) is used. The study data are collected online/offline and transferred directly to the database of the study server that is housed at the Charite University medicine in Berlin, Germany. The data transfer between the workstation computer (in each recruitement center) and the study server takes place via a secured connection (SSL encryption), so that the transferred study data cannot be manipulated. The data are stored in the database (Oracle). At the end of the study, the database will be closed after all entries have been entered. These patient data are only stored pseudonymously. The unique assignment to the patient is done via a paper printout, which is filed in the study folder of the respective recruitment site. The originals of all central study documents including documentation sheets will be stored in the recruitment site for at least 10 years after completion of the study. Medical records, paper report forms and original data should be retained for the longest possible period allowed by each participating center.

In some hospitals, the transmission of PP1 data to the study center takes place anonymously for data protection reasons. For this purpose, anonymized pooled data will be transmitted to the coordinating study center in an excel file.

Dissemination plans

Publication of study results will occur regardless of how the nature of the results. Besides poster and/or oral presentations at scientific meetings, at least one main publication related to each of the 3 study populations according to the respective objectives and hypotheses will be prepared and submitted. Secondary publications will also be prepared and submitted.



Source link