Unmet need in major depressive disorder and acute suicidal ideation or behavior: findings from a longitudinal electronic health record data analysis

Abstract

Aims

Using a national electronic health records (EHR) database, the current study describes treatments, depression severity, and health care resource utilization (HRU) among patients with major depressive disorder (MDD) and acute suicidal ideation or behavior (MDSI) prior to, during, and following a suicide-related event in the United States.

Materials and methods

This retrospective matched cohort study used data collected from the Optum EHR de-identified database for patients with diagnosis codes for MDD and acute suicidal ideation or behavior and a propensity score–matched cohort of patients without MDD or a suicide-related event. The study period was 31 October 2015–30 September 2019. MDD-related treatments and 9-item Patient Health Questionnaire (PHQ-9) scores, when available, were assessed at the first health care encounter for a suicide-related event (index period), 12 months before (pre-period), and 6 months after (post-period). All-cause and MDD-related HRU were assessed during the post-period.

Results

The mean (standard deviation) age of patients with MDSI was 39 (16) years; 55.0% were female. Index events occurred as follows: inpatient stay, 38.9%; observation unit stay, 4.6%; emergency department (ED) visit, 46.5%; and outpatient visit, 10.1%. Antidepressants and psychotherapy were the most common pharmacologic and nonpharmacologic treatments, respectively, prescribed during the pre- (31.3%, 9.5%, respectively) and index (41.2%, 18.7%, respectively) periods. Post-period data (n = 40,261) revealed only 43.4% received an antidepressant and 20.5% had psychotherapy after the suicide-related event. Few patients had PHQ-9 scores recorded during the pre- (4.4%), index (1.3%), and post- (7.6%) periods. During the post-period, 11.8%, 5.0%, and 33.1% of patients had ≥1 all-cause inpatient stay, observation unit stay, and ED visit, respectively; 61.0% had ≥1 all-cause and 33.4% ≥1 MDD-related outpatient visit. Most patients with MDSI and an inpatient encounter or ED visit were discharged to home or self-care (65.4%). Odds of an all-cause hospital encounter during the post-period were higher for patients with versus without MDSI (by 30.1, 33.5, and 33.9 times for inpatient stay, ED visit, and observation unit stay, respectively).

Conclusion

This analysis highlights an opportunity to improve outcomes for this vulnerable population. More complete data on patient outcomes is needed to inform strategies designed to optimize screening and treatment.

Keywords:

• Longitudinal patient care
• major depressive disorder
• suicidal ideation
• EHR database
• mental health screening

JEL CLASSIFICATION CODES:

• I11
• I1
• I
• I12

Introduction

Major depressive disorder (MDD) is among the most common mood disorders, with depression being the leading cause of disability worldwide^1–3. An estimated 19.4 million adults in the United States in 2019 had ≥1 major depressive episode (MDE) in the past year, representing approximately 7.8% of the adult population; of these, nearly two-thirds had an MDE with severe impairment⁴. Particularly severe symptoms, including acute suicidal ideation or behavior, place some patients with MDD at increased risk of poor outcomes⁵. In 2018, 32.0% of adults with MDD reported serious thoughts of suicide, 11.7% reported a suicide plan, and 4.6% reported a suicide attempt⁶. The total economic burden of MDD in the United States is substantial, estimated at $326.2 billion in 2018, including direct medical costs, absenteeism and presenteeism, and lost lifetime earnings due to suicide⁷. Health systems often struggle with providing adequate care for mental health disorders, such as MDD, and there is a need to improve both continuity of care and the patient experience^8,9.

Empirical studies focused on clinical and economic burden of patients with MDD and acute suicidal ideation or behavior (MDSI) have recently been reported using different databases covering different populations, such as the Veterans Health Administration (VHA) system dataset¹⁰, OptumHealth Care Solutions, Inc. database¹¹, Premier Hospital database¹², and online patient community platform, PatientsLikeMe¹³. The findings from these analyses suggest the need for better care, better screening, and more efficacious treatment of MDSI. However, to our knowledge, this is the first study focused on this population receiving care in the integrated delivery network (IDN) setting. Electronic health record (EHR) data potentially offer advantages in the study of conditions like MDD because, in addition to records of outpatient visits, hospitalizations, diagnostic procedures, medications, and laboratory results, they may contain useful information in the clinical notes, including patient-reported outcomes data. One measure of depression captured in the EHR of particular interest to this patient population is the 9-item Patient Health Questionnaire (PHQ-9). The PHQ-9, a patient self-reported instrument, is aligned with the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) symptom criteria for MDD and has been validated as a standardized tool to screen for depressive disorders and to measure depression symptom severity and treatment response¹⁴. The PHQ-9 has shown high accuracy, with sensitivity and specificity of 61.5–92% and 82–88%, respectively^15–17, compared with semi-structured diagnostic interviews for MDD. In 2016, the PHQ-9 was endorsed as a quality measure by the National Committee for Quality Assurance (NCQA) for inclusion in NCQA Healthcare Effectiveness Data and Information Set (HEDIS) measures^14,18. Therefore, HEDIS should incentivize its use and collection. In this study, PHQ-9 was used as a tool to measure depression symptom severity from unstructured data derived from the EHR database.

The purpose of the current study was to leverage a national EHR database to describe treatments, depression severity, and health care resource utilization (HRU) prior to, during, and following a suicide-related event among patients with MDSI. A more complete understanding of the treatment progression and HRU of this patient population may help to identify opportunities to improve treatment pathways and outcomes.

Methods

Study design and data source

This was a retrospective, matched cohort study that used data from the Optum de-identified EHR database. The Optum EHR database is a longitudinal patient-level medical record of >100 million patients treated at >700 hospitals and >7,000 clinics across the United States. This study analyzed data from adult patients (≥18 years of age) enrolled in an IDN (approximately 80% of the full dataset) in order to capture the continuum of care from ambulatory to inpatient/facility settings. Patients from a non-IDN were excluded to avoid missing data from incomplete records. During the study period (31 October 2015–30 September 2019), patients were followed from the index date (defined as the first occurrence of a diagnosis code for acute suicidal ideation or behavior) to the earliest of the following occurrences: death, end of health care activity in the dataset, or end of data availability. Acute suicidal ideation or behavior was defined by International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes for (1) suicidal ideation, (2) suicide attempt, or (3) intentional self-harm suggestive of suicidal behavior, based on the coding classification in Hedegaard et al. (Supplementary Appendix 1)¹⁹. To increase the likelihood that intentional self-harm events were reflective of suicidal behavior, a suicidal ideation diagnosis code was required to be observed on the index date or within 6 months post-index. All patient data contained in the Optum EHR database were de-identified and in compliance with the Health Insurance Portability and Accountability Act. Since this study consisted of secondary data analyses of de-identified patient records, the study was exempt from institutional review board approval.

Patient and public involvement

No patient involved.

Study cohorts

Patients were categorized into two mutually exclusive cohorts (MDSI and non-MDSI) based upon study-defined criteria. The study design incorporated a study period, pre-index period, index diagnosis identification period, and post–index event period (Figure 1). Patients within the MDSI cohort met the following criteria:

• Within the study period, ≥1 diagnosis for MDD as defined using the ICD-10-CM codes F32 and F33. Codes specific for F32.8 (other depressive episodes) and F33.8 (other recurrent depressive disorders) were excluded.

• Within the index period, ≥1 diagnosis code indicating acute suicidal ideation or behavior between 31 October 2016 and 30 September 2019 (Supplementary Appendix 1). The date of the first code for suicidal ideation or behavior determined the index date. Only those events with evidence of health care activity ≥12 months prior to the index date and within the study period were included.

• Within the 12 months prior to or on the index date (pre-period), ≥1 diagnosis for a depressive disorder (ICD-10-CM: F32, F33, F34.1, F43.21), and no diagnosis for bipolar or related disorders, mania, dementia, intellectual disability, schizophrenia, schizotypal, delusional, or other nonmood psychotic disorders during the study period.

Figure 1. Study design. (a) MDSI cohort. (b) Non-MDSI cohort. Abbreviations: MDSI, major depressive disorder and acute suicidal ideation or behavior; MDD, major depressive disorder; Dx, diagnosis. ^a≥1 diagnosis for MDD or other depressive disorder within the 12 months prior to or on the index date. ^b≥12 months of health care activity prior to the index period. ^cThe date of the first occurrence of suicidal ideation is defined as the index date and could occur any time within the index period. ^d≥1 diagnosis code for acute suicidal ideation or behavior during 31 October 2016 to 30 September 2019. ^e≥1 diagnosis for a depressive disorder during the study period.

The non-MDSI cohort was chosen to be the comparative cohort as a reflection of the general population without the condition of the case cohort (i.e. MDSI) and was constructed in two steps. First, the index date of each patient in the MDSI cohort was used to select a sample of patients (100 per each patient in the MDSI cohort) with health care activity documented on that same date who also met the following criteria:

• No diagnosis indicating MDD or acute suicidal ideation or behavior at any time during the study period.

• No diagnosis for a depressive disorder on or within 12 months prior to their index event, and ≥12 months of health care activity prior to the index event (pre-period).

• No diagnosis for bipolar or related disorders, mania, dementia, intellectual disability, schizophrenia, schizotypal, delusional, or other nonmood psychotic disorders during the study period.

Due to the nonrandomized nature of this noninterventional study, comparisons of HRU between these two cohorts may be confounded by differences in the populations at baseline. Thus, a statistical procedure was employed to minimize the potential influence of this issue on the comparisons and focus on economic differences arising from differences in clinical characteristics. Specifically, patients in the MDSI cohort were matched to patients in the non-MDSI cohort using a 1:1 propensity score–matching method to balance differences in patient demographics, regional trends in care, and changes in care over time among the two cohorts^11,20–22. The propensity score was defined as the probability of being in the MDSI cohort versus the non-MDSI cohort and was estimated using a logistic regression model based on the following baseline characteristics as covariates: age group, sex, race, ethnicity, insurance type, geographic region, geographic division, and the year of index date. Any characteristics related to the state of health were excluded from the propensity score matching. A patient with MDSI was matched to a patient without MDSI whose propensity score was closest to that of the patient with MDSI. Once a patient without MDSI was matched to a patient with MDSI, that patient without MDSI was no longer eligible for consideration as a match for other patients with MDSI. Balance on baseline covariates was assessed with standardized differences, with an absolute standardized difference ≥0.1 chosen as the threshold for potentially meaningful imbalance²³.

Outcome measures

Patient demographics and insurance type were assessed for both the MDSI and non-MDSI cohorts during the 12 months prior to the index date (pre-period) and were used in the statistical matching procedure. Mental health comorbidities (based on the DSM-5) were also evaluated during the pre-period and compared between the MDSI and non-MDSI cohorts²⁴.

The use of MDD-related pharmacologic and nonpharmacologic therapies was assessed for the MDSI cohort during the pre-period, index period, 1-month post-period, and 6-month post-period. MDD-related pharmacologic and nonpharmacologic therapies, obtained from prescription and procedure records, respectively, were summarized into the following categories: use of antidepressants (i.e. selective serotonin reuptake inhibitors, serotonin norepinephrine reuptake inhibitors, serotonin modulators, norepinephrine-serotonin modulators, tricyclic and tetracyclic antidepressants, miscellaneous antidepressants, and monoamine oxidase inhibitors), adjunctive treatments for MDD (i.e. anticonvulsants/mood stabilizers, atypical/second-generation antipsychotics, psychostimulants, thyroid hormone, buspirone, and lithium), other psychotherapeutic drugs (e.g. benzodiazepines), and use of nonpharmacologic therapy (i.e. psychotherapy, transcranial magnetic stimulation, electroconvulsive therapy, and vagus nerve stimulation).

PHQ-9 scores were used to categorize depression symptom severity and were obtained (if available) from unstructured data derived from the EHR note fields using Optum’s proprietary natural language processing algorithm. Among patients in the MDSI cohort with PHQ-9 scores recorded in their medical records, the scores were evaluated during the pre-period, index period, and 6-month post-period. Severity categories were defined as: none/minimal, score of 0–4; mild, 5–9; moderate, 10–14; moderately severe, 15–19; and severe, 20–27¹⁵. For patients with multiple PHQ-9 scores during a period, only the most recent one during a period was used. Among patients with ≥2 PHQ-9 score records (one during the pre-period and another during the 6-month post-period), the change in depression severity was calculated.

All-cause and MDD-related HRU were reported by category (i.e. inpatient stay, observation unit stay, emergency department [ED] visit, and outpatient visit) and assessed during the 1-month post-period and 6-month post-period. The same four categories were used to describe the care setting of the index event. Using patient discharge codes, discharge disposition was examined for those patients in the MDSI cohort with an index care setting of inpatient stay, observation unit stay, or ED visit.

Statistical analysis

Continuous variables were represented using means and standard deviations (SDs), and categorical variables were represented with counts and percentages. Differences in proportions of categorical variables were assessed using Chi-square tests. Statistical analyses were performed using SAS Enterprise Guide, version 7.1.

Odds ratios (ORs) generated using a logistic regression model were used to compare the proportion of patients with ≥1 encounter of all-cause or MDD-related HRU between the MDSI and non-MDSI cohorts.

Results

Patient demographics and mental health comorbidities

A total of 71,161 patients with MDSI were included in this analysis. The mean (SD) age was 39 (16) years, and 55.0% of patients were female. Index events were distributed across health care settings as follows: inpatient stay, 38.9%; observation unit stay, 4.6%; ED visit (with no inpatient admission), 46.5%; and outpatient visit, 10.1%. Propensity score matching yielded a well-matched non-MDSI cohort (Table 1, Supplementary Appendix 2).

Table 1. Baseline characteristics in the MDSI cohort, matched non-MDSI cohort, and standard differences in propensity score–matched cohorts.

Parameter	MDSI cohort (n = 71,161)		Non-MDSI cohort (n = 71,161)		|Std diff^a|
	n	%	n	%	%
Age group, years
18–29	27,393	38.5	27,422	38.5	0.1
30–39	13,278	18.7	13,211	18.6	0.2
40–49	11,077	15.6	11,017	15.5	0.2
50–59	10,659	15.0	10,674	15.0	0.1
60–69	5,837	8.2	5,846	8.2	0.0
70–79	2,092	2.9	2,159	3.0	0.6
≥80	825	1.2	832	1.2	0.1
Female	39,144	55.0	39,164	55.0	0.1
Race
African American	8,893	12.5	9,006	12.7	0.5
Asian	625	0.9	660	0.9	0.5
Caucasian	56,009	78.7	55,780	78.4	0.8
Other/unknown	5,634	7.9	5,715	8.0	0.4
Ethnicity
Hispanic	4,670	6.6	4,697	6.6	0.2
Not Hispanic	61,981	87.1	61,793	86.8	0.8
Unknown	4,510	6.3	4,671	6.6	0.9
Region
Midwest	40,550	57.0	40,506	56.9	0.1
South	11,102	15.6	10,987	15.4	0.4
West	5,960	8.4	5,998	8.4	0.2
Northeast	8,378	11.8	8,446	11.9	0.3
Other/unknown	5,171	7.3	5,224	7.3	0.3
Insurance
Commercial	23,841	33.5	23,760	33.4	0.2
Medicaid	14,343	20.2	14,358	20.2	0.1
Medicare	4,916	6.9	4,956	7.0	0.2
Multiple	4,722	6.6	4,745	6.7	0.1
Other payer type	2,592	3.6	2,575	3.6	0.1
Uninsured	4,530	6.4	4,524	6.4	0.0
Unknown	11,804	16.6	11,819	16.6	0.1
Missing	4,413	6.2	4,424	6.2	0.1

Abbreviations: MDSI, major depressive disorder and acute suicidal ideation or behavior; Std diff, standard difference.

^a|Std diff| ≥10% was considered as significant.

Among patients with MDSI, the most common DSM-5–related comorbidity during the pre-period was anxiety disorder (30.3%). The top 10 most prevalent pre-period mental health comorbidities (based on the MDSI cohort) are presented in Supplementary Appendix 3; all comorbidities were significantly more common among patients in the MDSI cohort than among those in the non-MDSI cohort (all p <0.0001).

Pharmacologic and nonpharmacologic therapies

Among patients in the MDSI cohort, antidepressants were the most prescribed medication during the pre-period (31.3%), index period (41.2%), 1-month post-period (29.1%), and 6-month post-period (43.4%), with selective serotonin reuptake inhibitors being the most prescribed antidepressant across time periods (Table 2).

Table 2. Depression and other mental health–related treatments recorded for patients with MDSI.

Parameter^a	Pre-period of MDSI event (n = 71,161)		Index MDSI event (n = 71,161)		1-month post-period of MDSI event (n = 57,058)		6-month post-period of MDSI event (n = 40,261)
	n	%	n	%	n	%	n	%
Pharmacologic therapy
Antidepressants	22,303	31.3	29,292	41.2	16,581	29.1	17,469	43.4
SSRI	15,101	21.2	17,106	24.0	10,391	18.2	11,316	28.1
SNRI	5,979	8.4	6,303	8.9	3,936	6.9	4,718	11.7
Serotonin modulators	5,479	7.7	13,244	18.6	5,684	10.0	6,178	15.3
Norepinephrine-serotonin modulator	1,710	2.4	3,393	4.8	1,973	3.5	2,137	5.3
Tricyclic and tetracyclic antidepressants	1,981	2.8	1,225	1.7	737	1.3	1,077	2.7
Bupropion, maprotiline	3,631	5.1	3,365	4.7	2,043	3.6	2,784	6.9
Monoamine oxidase inhibitors	17	<0.1	16	<0.1	6	<0.1	11	<0.1
Agents used to augment antidepressant treatment	9,952	14.0	14,316	20.1	6,934	12.2	8,245	20.5
Lithium	146	0.2	288	0.4	141	0.2	198	0.5
Second-generation antipsychotics^b	3,681	5.2	10,031	14.1	3,855	6.8	4,364	10.8
Anticonvulsant	1,549	2.2	1,728	2.4	962	1.7	1,251	3.1
Thyroid hormone	2,652	3.7	2,113	3.0	1,243	2.2	1,679	4.2
Psychostimulants	2,032	2.9	865	1.2	826	1.4	1,115	2.8
Buspirone	2,152	3.0	1,918	2.7	1,191	2.1	1,759	4.4
Other psychotherapeutic drugs	20,531	28.9	30,035	42.2	9,601	16.8	11,760	29.2
Other second-generation antipsychotics	830	1.2	2,405	3.4	838	1.5	992	2.5
Nonstimulant ADHD treatments^c	2,440	3.4	3,651	5.1	1,232	2.2	1,488	3.7
Benzodiazepines	19,030	26.7	27,741	39.0	8,287	14.5	10,463	26.0
Sleep aids^d	3,277	4.6	4,150	5.8	1,437	2.5	1,961	4.9
Nonpharmacologic therapy
Psychotherapy	6,760	9.5	13,320	18.7	8,203	14.4	8,260	20.5
Transcranial magnetic stimulation	14	<0.1	5	<0.1	10	<0.1	22	0.1
Electroconvulsive therapy	92	0.1	221	0.3	223	0.4	226	0.6
Vagus nerve stimulation	26	<0.1	0	0.0	2	<0.1	7	<0.1

Abbreviations: MDSI, major depressive disorder and acute suicidal ideation or behavior; SSRI, selective serotonin reuptake inhibitor; SNRI, serotonin norepinephrine reuptake inhibitor; ADHD, attention-deficit/hyperactivity disorder; MDD, major depressive disorder.

^aPatients may receive more than one therapy or type of therapy per period. Percentages were calculated out of the total number of patients receiving a therapy within each period.

^bSecond-generation antipsychotics with US Food and Drug Administration approval for MDD: aripiprazole, brexpiprazole, olanzapine in combination with fluoxetine (specifically treatment-resistant MDD), and quetiapine XR.

^cClonidine, guanfacine, and atomoxetine.

^dEszopiclone, zaleplon, and zolpidem.

Atypical antipsychotics indicated for adjunctive use in MDD were the most commonly prescribed augmentation agents during the pre-period (5.2%), index period (14.1%), 1-month post-period (6.8%), and 6-month post-period (10.8%). The proportions of patients receiving nonpharmacologic therapy were lower than the proportions prescribed pharmacologic therapy, with psychotherapy being the most common nonpharmacologic therapy (ranging from 9.5% of patients during the pre-period to 20.5% of patients during the 6-month post-period).

PHQ-9 scores

Overall, very few patients in the MDSI cohort had PHQ-9 scores recorded during the pre-period, index period, and 6-month post-period (4.4%, 1.3%, and 7.6%, respectively; Table 3).

Table 3. PHQ-9 scores for the MDSI cohort^a.

Parameter	Pre-period of MDSI event (n = 71,161)		Index MDSI event (n = 71,161)		6-month post-period of MDSI event (n = 40,261)
	n/Mean	%/SD	n/Mean	%/SD	n/Mean	%/SD
Patients with PHQ-9 score	3,141	4.4	941	1.3	3,062	7.6
PHQ-9 total score	13.0	7.5	17.0	7.2	12.1	7.5
PHQ-9–derived depression severity^b
None/minimal	523	16.7	70	7.4	593	19.4
Mild	543	17.3	113	12.0	641	20.9
Moderate	697	22.2	118	12.5	648	21.2
Moderately severe	661	21.0	231	24.5	571	18.6
Severe	717	22.8	409	43.5	609	19.9

Abbreviations: PHQ-9, 9-item Patient Health Questionnaire; MDSI, major depressive disorder and acute suicidal ideation or behavior; SD, standard deviation.

^aPHQ-9 total score reported as mean and SD; all other parameters reported as n and %.

^bDepression severity categories were defined as follows: none/minimal, score 0–4; mild, 5–9; moderate, 10–14; moderately severe, 15–19; and severe, 20–27. The mean (SD) PHQ-9 scores during these periods were 13.0 (7.5), 17.0 (7.2), and 12.1 (7.5), respectively. Among those with a PHQ-9 score available during the index period, approximately 80% had scores indicating moderate-to-severe depression.

Less than 2% of patients (n = 1,139) had PHQ-9 scores during both the pre-period and 6-month post-period (Table 4).

Table 4. Shifts in PHQ-9 score categories for patients with MDSI with scores during pre- and 6-month post-periods (n = 1,139)^a.

Parameter	Depression severity during the 6-month post-period
	None/minimal		Mild		Moderate		Moderately severe		Severe
	n	%	n	%	n	%	n	%	n	%
Pre-period depression severity
None/minimal (n = 161, 14.1%)	76	47.2	30	18.6	23	14.3	20	12.4	12	7.5
Mild (n = 217, 19.1%)	44	20.3	83	38.2	49	22.6	26	12.0	15	6.9
Moderate (n = 259, 22.7%)	25	9.7	71	27.4	94	36.3	37	14.3	32	12.4
Moderately severe (n = 240, 21.1%)	24	10.0	40	16.7	53	22.1	82	34.2	41	17.1
Severe (n = 262, 23.0%)	25	9.5	35	13.4	54	20.6	46	17.6	102	38.9

Abbreviations: PHQ-9, 9-item Patient Health Questionnaire; MDSI, major depressive disorder and acute suicidal ideation or behavior.

^aDepression severity categories were defined as follows: none/minimal, score 0–4; mild, 5–9; moderate, 10–14; moderately severe, 15–19; and severe, 20–27.

Among these patients, 22.7%, 21.1%, and 23.0% had moderate, moderately severe, and severe depression, respectively, during the pre-period. Of note, although 61.1% of patients with severe depression during the pre-period shifted to a less severe level of symptomatology during the 6-month post-period, 38.9% remained in the severe range.

HRU

For patients in the MDSI cohort (n = 40,261), 61.0% of patients had an all-cause HRU in the outpatient setting 6 months after a suicide-related event, whereas 33.4% of patients had an MDD-related HRU in the same period (Figure 2). Notably, that leaves 39% of patients without an outpatient visit 6 months following a suicide-related event. Visits to the ED 6 months after an event were less frequent but still high, with 33.1% of patients having an all-cause HRU in the ED and 11.1% of patients having an MDD-related HRU. Inpatient stays were similar between the two groups, with 11.8% and 7.0% of patients having an all-cause or MDD-related HRU after a suicide-related event, respectively.

Figure 2. Post-period HRU in the MDSI cohort. (a) Proportion of patients with all-cause HRU. (b) Proportion of patients with MDD-related HRU. Abbreviations: HRU, health care resource utilization; MDSI, major depressive disorder and acute suicidal ideation or behavior; ED, emergency department; MDD, major depressive disorder. ^aHRU reported for the 57,058 patients in the MDSI cohort with ≥1 month of data in the post-period. ^bHRU reported for the 40,261 patients in the MDSI cohort with ≥6 months of data in the post-period.

Compared to those in the non-MDSI cohort (n = 34,109), patients in the MDSI cohort (n = 40,261) had higher odds of an all-cause hospital encounter during the 6-month post-period (Figure 3). Specifically, the OR of an inpatient admission, ED visit, or observation unit stay during the 6-month post-period was 30.1, 33.5, and 33.9 for MDSI versus non-MDSI, respectively, compared with 1.2 for an outpatient visit.

Figure 3. All-cause HRU: MDSI versus non-MDSI cohorts. (a) During the 1-month post-period. (b) During the 6-month post-period. Abbreviations: HRU, health care resource utilization; MDSI, major depressive disorder and acute suicidal ideation or behavior; OR, odds ratio; CI, confidence interval; ED, emergency department.

The majority of patients with MDSI and an index event classified as either an inpatient encounter or ED visit (n = 64,005) were discharged to home or self-care (65.4%). Other discharge dispositions included discharge to a psychiatric hospital (17.8%) and other types of facilities (11.7%). The disposition was unknown for 3.7% of patients, and 1.3% of patients left against medical advice.

Discussion

This was a real-world study analyzing the continuum of care of patients experiencing major depression and suicidal ideation and behavior. There have been limited studies within this patient population to date. This study used nationally representative data from an EHR database to gain insight into how patients with MDSI are managed in real-world clinical practice, with practical findings to help inform optimization of treatment pathways for this vulnerable population. Findings demonstrate that a substantial proportion of patients did not receive standard-of-care therapies to treat their underlying depressive disorder, including oral antidepressants and psychotherapy. The majority of patients with MDSI (∼90%) required hospitalization (i.e. inpatient stay, observation unit stay, or ED visit), which clearly highlights the fragility of this patient population. Furthermore, among the subset of the population with PHQ-9 scores available, the presence of possible ineffective treatment options, treatment inertia, treatment-resistant depression, and medication nonadherence were suggested by the observation that patients’ depression category often stayed the same or worsened after the suicide-related event. Despite the recognition of high rates of recurrence of suicidal ideation, suicide attempts, and hospitalization in patients with MDD in the 6 months following a suicide-related event¹², a sizeable proportion did not have an all-cause or MDD-related outpatient visit 1 month and 6 months after the index event. In totality with the aforementioned findings, it was consistent to observe high levels of HRU among patients with MDSI as well as significantly higher odds of an all-cause hospital encounter compared to patients without MDSI in the periods after the suicide-related event.

These results expand upon previous findings from a cross-European study of data from the National Health and Wellness Survey, which showed that, among patients with MDD, those with suicidal ideation were at greater risk of HRU events compared with those with MDD but without suicidal ideation (relative risk: health care provider visits, 7.84; general/family practitioner visits, 3.88; hospitalizations, 2.48; ER visits, 2.38)²⁵. The study also showed that HRU was generally higher among patients with MDD, regardless of suicidal ideation, compared with the general population²⁵.

Despite significant contact with health care practitioners among patients with MDSI observed in the dataset, sizeable proportions did not receive any antidepressant therapy or psychotherapy, even after a documented suicidal ideation or suicide attempt. This is consistent with previous studies that reported inadequate treatment for patients with MDD or suicidal ideation. In an analysis of data from the National Survey on Drug Use and Health, more than 30% of participants who made suicide plans and/or attempted suicide received no mental health treatment either before or after planning or attempting suicide²⁶. Patients with mental health conditions may face numerous barriers to receiving care. A recently published study evaluated patients with suicidal ideation and found certain factors, such as stigma, anticipation of provider overreaction, and loss of autonomy, precluded patients from disclosing their suicidal ideations to health care providers²⁷.

The PHQ-9 is recognized as an important quality measure by NCQA¹⁸; however, its use in routine clinical practice varies. EHR studies report the PHQ-9 form was used to evaluate 13.5–73.7% of patients with depression in primary care^28–31. Other studies have reported incomplete PHQ-9 questionnaires in an analysis of depression in EHR records³². In the current study, only a small proportion of patients had PHQ-9 scores recorded in their medical records (<10%), and among the subgroup of patients with PHQ-9 scores during both the pre-period and 6-month post-period, results suggest a need for further investigation of the reasons for symptom improvement (and lack thereof). For example, clinical inertia, a well-recognized phenomenon in the treatment of depression, has been explored in previous studies, although not in the MDSI subpopulation specifically. Research using data from the Quality Improvement for Depression collaboration found that only 29% of patients receiving care for depression achieved a full response to therapy (defined as >50% improvement in symptoms), while treatment adjustments were made for 64% of patients³³. Another study assessed clinician adherence to depression practice guidelines and found low adherence to nearly half the recommendations (e.g. assessment of symptoms and history of depression, management of suicide risk, and treatment adjustment for patients who do not respond to initial treatment)³⁴.

Inadequate treatment of mental health disorders has led to evaluation of varying strategies, such as enhanced mental health resources and collaborative care models, to improve the quality of care^35,36. Currently, there are no standard-of-care clinical practice guidelines for MDSI. Related guidelines include the 2019 Veterans Affairs and Department of Defense (VA/DoD) Clinical Practice Guideline for the Management of Patients at Risk for Suicide³⁷, the 2019 American Psychiatric Association (APA) Clinical Practice Guideline for the Treatment of Depression Across Three Age Cohorts³⁸, and the 2019 National Action Alliance (NAA) Best Practices in Care Transitions for Individuals with Suicide Risk³⁹. For individuals with a history of suicide risk who are transitioning from suicide-related hospitalization to outpatient care, these guidelines recommend a team-based approach, such as scheduling a follow-up appointment within 24 hours of discharge and no later than 7 days post-discharge³⁹. In this study, substantial proportions of patients did not receive any follow-up care or symptom screening within 30 days. The PHQ-9 presents an opportunity for IDNs to improve treatment for patients with MDSI by collection of relevant depression data. Given its inclusion in multiple quality measures, including HEDIS, the PHQ-9 is a practical way to screen for and assess depression symptom severity in the real world⁴⁰. Additionally, the PHQ-9 was referenced in guidelines or endorsed by multiple health systems as a tool to screen for and manage depression^41–43, making it even more relevant to clinicians and further supporting its use in the management of depression. Identification of individuals at risk for MDD and with suicidal ideation or behaviors provides a chance to improve the treatment and management of their depression. Routine use of the PHQ-9 presents an opportunity to identify these patients more effectively in a real-world setting.

Several limitations should be considered when interpreting results of this study. First, the study was unable to account for assessments that might have been made outside of the clinics/hospitals that contribute data to the Optum EHR database. In an effort to address this issue, the analysis focused on only patients who were enrolled in an IDN; however, this action may limit the generalizability of these findings to patients from other health care settings. Additionally, within any EHR database, data can be inaccurate or incomplete, or patients can drop out of the dataset. Second, prescription data reflect the prescriptions written by the health care provider and thus do not necessarily indicate the medication was filled and taken as prescribed; alternatively, patients may have received medication outside of the IDN. Third, this study used propensity score matching to adjust for confounding; however, the possibility of unmeasured or residual confounding cannot be excluded. Fourth, because no scheduling data were reported in this dataset, the number of patients who met the follow-up appointment was not available. Fifth, the proportion of outpatient visits for the MDSI population in this study was relatively small; visits were likely under-recorded due to stigma or the perception of accidental injury^27,44–46. In addition, the entire population-based distribution could be underestimated, based on suicidal ideation and suicide attempt codes, if the intent or cause of an injury was unknown. Finally, PHQ-9 scores were available for a relatively small subset of the population with MDSI. The analysis of changes in depression severity and post-event treatments were therefore limited in this study. Additionally, it is possible that there is systematic bias in why some patients might receive a PHQ-9 assessment compared to others. Thus, it cannot be inferred that the findings are generalizable to the entire population in this sample or externally. Expanding the study population of patients with MDSI may be considered in future studies.

Conclusions

This study documents a high level of HRU among patients with MDSI as well as higher odds of an all-cause hospital encounter relative to patients without MDSI, with the greatest number of health care encounters occurring in the outpatient setting. Few patients had PHQ-9 scores recorded, and sizeable portions of patients did not receive pharmacologic and/or nonpharmacologic therapy, even after a documented suicidal ideation or suicide attempt. To our knowledge, this is the first study focused on the MDSI population receiving care in the IDN setting. This study illustrates the urgency for health systems to dedicate greater resources to studying treatment patterns and outcomes for patients with MDSI. Results of this study suggest there is an urgent need for improved screening, data collection, treatment, and follow-up within health care systems for this vulnerable population.

Transparency

Competing interests

CN, YC, MS, and JV are employees of Janssen Scientific Affairs, LLC, and are stockholders in Johnson & Johnson. MS is an employee of Mu Sigma, which received consultancy fees from Janssen Scientific Affairs, LLC, for performing the study.

Author contributions

All authors contributed to the study design and concept, data analysis and review, data interpretation, and manuscript review.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Previous presentation

Neslusan C, et al. Poster presented at: Neuroscience Education Institute (NEI) MAX! Virtual Congress; 5–8 November 2020.

Acknowledgements

Medical writing support was provided by Caryn Gordon, PharmD, and Courtney St. Amour, PhD, of Lumanity Communications Inc., and was funded by Janssen Scientific Affairs, LLC.

Data sharing statement

The data that support the findings of this study are not publicly available due to privacy or ethical restrictions.

Additional information

Funding

This study was sponsored by Janssen Scientific Affairs, LLC, Titusville, NJ (Ringgold ID: 155044; grant number: not applicable).