Subthreshold depression – concept, operationalisation and epidemiological data. A scoping review

Abstract

Purpose

In diagnostic systems (e.g., DSM-5, ICD-10), depression is defined categorically. However, the concept of subthreshold depression (SD) has gained increasing interest in recent years. The purpose of the present paper was to review, based on a scoping review, the relevant papers in this field published between October 2011 and September 2020.

Materials and methods

Of the 1,160 papers identified, 64 records could be included in further analysis. The scoping review was conducted using both electronic and manual methods.

Results

The main result of the analysis is that the operationalisation criteria used are highly heterogeneous, which also leads to very heterogenous epidemiological data.

Conclusions

Clear conclusions are not possible scrutinising the reported results. Most definitions seem to be arbitrary, with considerable overlap (e.g., between SD and minor depression). The review also revealed that the impact of SD on quality of life and related parameters appear to be in the range of the respective impact of major depression (MD) and therapeutic approaches might be helpful for SD and also for the prevention of conversion from SD to MD. Keeping the presented difficulties in mind, a proposal for the definition of SD is made in the present paper in order to facilitate the discussion leading to more homogeneous criteria.

Keywords:

• Subthreshold depression
• subsyndromal depression
• epidemiology
• operationalisation
• systematic review

Introduction

In both the DSM-5 and ICD-10-system, depression is defined categorially. In DSM-5, the distinction between major and minor depression is given, in the ICD-10, the distinction between mild, moderate and severe depressive episodes. In the ICD-10, depressive adjustment disorders (ICD-10: F43.2) are defined – besides a critical life event as a trigger – as depressive symptoms which do not fulfil the criteria for a mild depressive episode. Regarding symptom severity, the definition of dysthymia in ICD-10 has much in common with depressive adjustment disorders. However, no critical life event can be identified and the duration has to be at least two years.

In the last decades and especially in the last 10–15 years, the concept of subthreshold depression (SD), which is sometimes also called subsyndromal depression, has gained increasing interest. The most widely used definition for SD is a clinically relevant level of depressive symptoms without meeting the diagnostic criteria for a major depressive disorder (MD) in the DSM system (or a depressive episode in the ICD-system). Such conditions can be defined dimensionally (i.e., by the definition of a cut-off score on a validated self-rating depression instrument without fulfilling the criteria for MD) or categorially (i.e., fewer than five symptoms of DSM-MD) (Ebert et al., 2018).

In 2012, Rodríguez et al. published a survey on SD conditions based on the literature published between January 2001 and September 2011 and identified with a PUBMED research. In this survey, the relevant data regarding definitions and associated factors of SD and also minor depression (as defined in DSM-IV) were reviewed. The main results of their work regarding SD are given in Table 1. It should be noted that minor depression (according to the DSM system) is not referred to here, since minor depression is not the topic of the presented paper. However, as the authors point out, there is a huge overlap between SD and minor depression, which can also be seen when scrutinising the respective diagnostic criteria for SD used by the different authors as depicted in Table 1 of their thoroughly written review.

Table 1. Prevalences of subthreshold depression (SD) and related syndromes and settings according to Rodriguez et al. (2012).

Author	Diagnosis	Sample/Population	Prevalence
Rucci et al., 2003	SD	Primary care	9.9%
Forsell, 2007	Subsyndromal symptomatic depression	Community	5% (wave 1), 5.5% (wave 2)
Backenstrass et al., 2006	Nonspecific depressive symptoms	Primary care	9.1%
Vuorilehto et al., 2005	Subsyndromal depression	Primary care	2.9%
da Silva Lima and de Almeida Fleck, 2007	Subsyndromal depression	Primary care	6.2%
Fergusson et al., 2005	SD	Community	7.3%
Goldney et al., 2004	SD	Community	12.9%
Ayuso-Mateos et al., 2010	Subsyndromal depression	Community	2.8%
Regeer et al., 2006	SD	Community	17.2% at baseline
Gómez-Restrepo et al., 2004	Subclinical depression	Community	1.4%
Baumeister and Morar, 2008	SD	Community	0.7% clinical significance criteria 1.8% symptom count

Rodriguez et al. (2012) reported prevalence rates between 2.9 and 9.9% in primary care and 1.4–17.2% in community settings. Results regarding gender and family status remained unclear. It became obvious that quality of life was considerably lower in people with SD, although the quality of life was even more reduced in people with MD. The authors concluded that depression, as a disorder, is better explained as a spectrum rather than a “collection of discrete categories.” Since SD, although falling below the diagnostic categories, is also characterised by severe functional impairments, further research aiming to characterise this condition more precisely and more clearly is urgently needed, and a well-recognized unique definition of SD is still missing.

How important it is to come to a valid definition of SD can also be deduced from the fact that meta-analyses [for example regarding the efficacy of psychotherapeutic interventions (Cuijpers et al., 2014) or comparing mortality rates between MD and SD (Cuijpers et al., 2013)] already exist, summarising different studies. However, as these studies mostly use heterogeneous definitions of SD, the meaning of such meta-analyses is questionable. In the narrative review of Biella et al. (2019) it is therefore also emphasised that, although SD is the most prevalent psychiatric condition in the elderly, no consensus regarding the diagnostic items varying across countries and research groups exists, which consecutively leads to the high variability of the reported results.

However, before a more valid definition of SD can be established, the current definitions and main epidemiological findings should be reviewed, also since 10 years have passed since the review of Rodriguez et al. (2012) and only PUBMED/MEDLINE was used as a literature database in this review.

Thus, the purpose of the presented scoping review is two-fold: first, the systematic literature survey is continued starting from 2012 onwards, and second, the epidemiological data will be summarised and the diagnostic criteria will be extracted and discussed in order to help establish generalised operationalisations for SD instead of very individual approaches as has often been the case over the last 20 years.

Methods

The literature review was conducted using both electronic and manual methods. Regarding electronic methods, the search was performed in PUBMED/MEDLINE, PSYCHINFO and EMBASE databases covering October 2011 to September 2020. Search strategy focussed on terms for ‘subthreshold’ and ‘depression’, with ‘epidemiology’, ‘human and economic burden’ and ‘prevention and therapy’. The precise methods were again tailored to each database. The full search strategy for the PUBMED/MEDLINE database was: (minor depression [Title/Abstract] OR subthreshold depression [Title/Abstract] OR subsyndromal depression [Title/Abstract] OR subclinical depression [Title/Abstract] AND (subthreshold depressive symptoms OR subclinical depressive conditions). This search strategy corresponds exactly to that used by Rodriguez et al. (2012). No restrictions for language were applied.’ References of the identified papers were screened manually to find additional related studies.

Eligibility criteria

After reading the abstracts, the eligibility of the selected full-text articles was evaluated by HPV and JS independently without automation tools using the following criteria:

Type of study: For epidemiological studies, only trials with original data published as full articles in peer-reviewed journals were included, no study protocols, diagnostic or methodological papers were considered. Moreover, only studies representative of the general population (including sub-populations of several age groups) or of specific patient populations (e.g., primary care or psychiatry) were included. In cases where two or more articles reported data from the same study sample, only the most relevant article was considered. Reviews and meta-analyses were excluded except for those of outstanding quality, as judged by HPV (e. g. adding additional aspects).

In contrast to the approach regarding epidemiological data, no requirements regarding the type of paper or the diagnostic criteria used had to be met in order to obtain complete information regarding operationalisation approaches for SD. With one exception: if the number of participants in the SD group was too small (i.e., number of participants < 60), then the respective paper was also not included.

Data extraction

Studies’ characteristics were extracted considering the following items: setting (general population, specific patient population, adolescents, adults, elderly, etc.), country, the study period of data collection, study design (longitudinal, cross-sectional, retrospective), sample size, age range, assessment of SD.

Assessment of methodological quality

As in the publication of Haller et al. (2014) and in a comparable paper of our group (Volz et al., 2021), both works, however, dealing with subthreshold anxiety, a scoring system according to Loney et al. (1998) was used in a modified way in order to critically score the epidemiological studies. One point was given for each of the following criteria:

1. Random population sample with an unbiased sampling strategy.

2. Adequate sample size (>1,000).

3. Adequate response rate (>70%).

4. Comparisons between respondents/non-respondents (i.e., those who refused the initial query).

5. Reliable and valid assessment of SD (standardised instruments used).

Less than three points were judged as “low quality,” at least three points as “high quality.”

Data analysis

No formal data analysis was performed and the respective data of the single studies are reported narratively.

Results

Search results

1,160 papers were identified in the database search. By manually checking the reference lists, no further papers could be identified. Of these 1,160 papers, 179 were duplicates. The remaining 981 abstracts were then screened, and 90 studies were read in full text to assess their eligibility for inclusion in this review. Of these, 26 studies were excluded: 11 because n was too low and 15 because SD was not the main focus of the respective paper. See also Figure 1.

Figure 1. Flowchart of study exclusions (n: number).

Subsyndromal depression – operationalisation

As can be seen from Table 2, the operationalisation of SD is very heterogeneous. However, certain patterns can be identified:

Table 2. Different definitions of subthreshold depression.

1	Source	Definition	Epidemiological study/Quality (1. random population sample/sample size > 1,000 2. Response rate > 70 3. Comparison between respondent/non-respondent/ 4. reliable and valid assessment of Depression/ 5. standardised assessment scale used)	Time	Number	Time and number	Scale
PUBMED
	Ebert et al., 2018	CESD ≥ 16 and no DSM-IV MD (also not in the past 6 months), bipolar disorder, or psychotic disorder with the help of SCID					X
	Gosling et al., 2018	PHQ-9 between 5 and 19, no MD, no lifetime bipolar affective disorder (MINI)					X
	Hill et al., 2014	Depressed mood or loss of interest or pleasure ≥ 1 week and ≥ 2 of the 7 DSM-IV-MD-symptoms but no lifetime history of dysthymic disorder, MD, or bipolar disorder	X (1/1/1/1/1)	X	X	X
	Jinnin et al., 2016	Not having a MD during the previous 12 months and no lifetime history of bipolar disorder and BDI-II ≥ 18	X (0/1/0/1/1)				X
	Mokhtar et al., 2020	CESD-R ≥ 16, but no DSM-IV-symptoms of MD					X
	Pietrzak et al., 2013	Two leading questions regarding mood and loss of interest. If yes, other symptoms of major depression asked. Yes to at least one leading question and a total of maximal 4 symptoms: subsyndromal depression: limited symptoms; yes to at least one leading question and a total of at least 5 symptoms, but not meeting the distress or impairment criterion for major depression: subsyndromal depression: no distress group; DSM-IV	X (1/1/1/0/0)	X	X	X
	Poppelaars et al., 2016	Above the 70^th percentile of the RADS-2 of the included population					X
	Schmitz et al., 2014	PHQ-9 between 5 and 14 and scored ≥ 1 on either of the first two PHQ-9 items	X (1/?/0/1/1)				X
	Topuzoğlu et al., 2015	CIDI 2.1: less than 5 MD-symptoms or duration less than 2 weeks or non-interfering impairment due to symptoms	X (1/1/0/1/0)	X	X	X
	Weisel et al., 2019	CESD ≥ 16 and no MD according to MINI and no MD in the past 6 months					X
	Xiao et al., 2016	At least 2 DSM-depression symptoms for at least 2 weeks that are associated with social dysfunction, not meeting DSM-criteria for minor depression, MD or dysthymia (details published by Xiao et al., 2013)	X (1/0/0/0/1)	X	X	X
PsychInfo
	Vulser et al., 2018	In the past 4 weeks: ≥ 3 DSM-IV-symptoms of MD including ≥ 1 core symptom (depressed or irritable mood, loss of interest) not fulfilling DSM-IV-criteria for MD		X	X	X
	Takagaki et al., 2018, 2016	BDI-II ≥10, no MD according CIDI					X
	Wang et al., 2018	PHQ-9 between 9 and 14 and scored ≥ 1 on either of the first two PHQ-9 items	X (0/?/0/1/1)				X
	Wong et al., 2018	PHQ-9 between 5 and 9; no MD within past 6 month or dysthymia with subdepressive symptoms lasting longer than 2 years					X
	Xiang et al., 2018	CESD ≥ 3 and less than 5 depressive symptoms according CIDI and no formal psychological or psychiatric treatment	X (0/?/?/1/1)				X
	Sigström et al., 2018	SD ≥ 2 of the 9 DSM-5-depression symptoms without fulfilling DSM-5-criteria for minor depression or MD (according to Judd et al., 1994)	X (0/1/0/1/1)	X	X	X
	Singhal et al., 2018	CDI between 14 and 24					X
	Moldovan et al., 2013	BDI between 10 and 16, not being in psychotherapy or on psychotropic medication	–				X
	Lee et al., 2013	CESD > 9, no MD (DIS)	X (0/0/0/1/1)				X
	Jeong et al., 2013	DSM-IV-depression symptoms criterion A ≥ 2, one of these symptoms depressed mood or anhedonia for more than a half of day or more than seven days during 2 weeks, no MD or minor depression according DSM-IV	X (1/?/0/1/1)	X	X	X
	Gledhill and Garralda, 2013	MFQ ≥ 17 on the MFQ, but no K-SADS diagnosis of depression	X (0/0/0/1/1)				X
	Furukawa et al., 2012	K6 ≥ 9, BDI-II ≥ 10, no MD (CIDI), no lifetime history of BD					X
	Morgan et al., 2012	PHQ-9 2–4 symptoms of depression experienced more than half the days or nearly every day for 2 or more weeks		X	X	X	X
	van der Aa et al., 2015a, b	CESD ≥ 16, no DSM-IV mood or anxiety disorder according to MINI	X (0/0/1/1/1)				X
	Laborde-Lahoz et al., 2015	One core symptom of MD (i.e., depressed mood or anhedonia) and up to only four depressive symptoms or five depressive symptoms being not clinically significant or causing functional impairment	X (1/1/0/0/1)		X
	Balázs et al., 2013	BDI-II < 20 and sadness or loss of pleasure according to DSM-IV-TR-MD	X (1/1/1/0/1)				X
	Vancampfort et al., 2017; Stubbs et al., 2016, 2017a,2017b, 2018	At least one criterion B (ICD-10-DCR) based on CIDI, total number of symptoms maximal 3, duration at least 2 weeks		X	X	X
	Zarski et al., 2018	CESD ≥ 16 and no DSM-IV MD (also not in the past 6 months), bipolar disorder, or psychotic disorder with the help of SCID					X
	Jeuring et al., 2018	CESD ≥ 16, but without a past-year diagnosis of MD according to DIS	X (0/?/0/1/1)				X
	Buntrock et al., 2017	CESD ≥ 16 and no DSM-IV MD (also not in the past 6 months), bipolar disorder, or psychotic disorder with the help of SCID					X
	Ludvigsson et al., 2016	GDS-15 between 3 and 5	X (0/0/0/1/1)				X
	Camacho et al., 2014	CESD between 8 and 15					X
	Almeida et al., 2015	PHQ-9 between 5 and 9, relevant symptoms present at least a week in the past 2 weeks	X (1/0/0/1/1)	X	X	X	X
	Wyman et al., 2020	CESD ≥ 16 and no DSM-IV-diagnosis of MD or dysthymia with the help of SCID	X (0/0/0/1/1)				X
	Ludvigsson et al., 2019, 2018	GDS-15 between 3 and 5					X
	van Zoonen et al., 2016	K 10 > 19, no MD, dysthymia, or anxiety disorder in the past 6 months using CIDI		X			X
	Horiuchi et al., 2017	CESD ≥ 16 and no DSM-IV-MD as assessed MINI					X
	Buntrock et al., 2015, 2016	CESD ≥ 16 and no DSM-IV MD (also not in the past 6 months), bipolar disorder, or psychotic disorder with the help of SCID					X
	Kang et al., 2015	DSM-IV-criteria (evaluated with the K-SADS): One mood disturbance symptom (either depressed mood, loss of interest, or irritability) and 1–3 other MD-symptoms either at threshold (5–7 days weekly, more than half of awake time) or subthreshold level (3–4 days weekly, more than half of awake time)		X	X	X
	Karsten et al., 2013	IDS-SR30 ≥ 14 and absence of full-syndromal depressive disorder (CIDI, 2.1)	X (1/1/0/1/1)				X
	Brooks et al., 2019	PHQ-9 between 5 and 9	X (1/?/0/0/1)				X
	Wu et al., 2020; Johansson et al., 2019, Gonçalves-Pereira et al., 2019	EURO-D 4 or 5, no ICD-10-criterion for depression, determined with the help of the GMS-AGECAT	X (1/?/0/1/1)				X
	Karp et al., 2019	PHQ-9 at least 1 and at least one of the cardinal symptoms of depression (low mood, anhedonia) for at least several days for the past two weeks; no MD and no history of MD in the past 12 months (SCID)		X	X	X	X
	Philipson et al., 2020	BDI-C ≥ 16 or CESD-C ≥ 30 and BDI ≥ 11, or a self-reported suicide attempt and no other depressive episode at the time being or in the past (DICA-R-A)	X (1/0/0/1/1)				X
	Oh et al., 2020	MINI-K ≥ 2 DSM-IV-depression symptoms, at least 1 depressed mood or anhedonia within the same 2-week period; no diagnoses of MD or minor depression	X (1/0/0/1/1)	X	X	X	X
EMBASE
	Shirazian et al., 2016	PHQ-9 between 5 and 14	X (0/0/0/0/1)				X
	Pols et al., 2017, 2018	PHQ-9 ≥ 6, no MD (DSM-IV) according MINI					X
	Peters et al., 2015	Life time depressed mood or loss of interest or pleasure lasting at least 2 weeks + at least 2 of the 7 other MD-DSM-IV-depression symptoms, but never experienced any full-syndrome mood disorder	X (1/1/0/0/0)	X	X	X
	Ho et al., 2016	GMS-AGECAT: 1–2	X (1/1/0/1/0)	X	X	X
	Vaccaro et al., 2017	GDS ≥ 6, one key question for depression (depressed mood, anhedonia) positive and no MD or dysthymia	X (1/1/1/1/1)	X	X	X	X
	Zhang et al., 2020	GDS-15: 1–5	(0/0/0/0/1)				X
	Pickett et al., 2014	MD: depressed mood or anhedonia + at least four other depressive symptoms for at least 2 weeks M/SD (minor depression, sub-syndromal depression): depressed mood or anhedonia and 1–3 other depressive symptoms for at least two weeks Dysthymia according DSM-IV SD: at least two depressed symptoms threshold or subthreshold by SCID-criteria (one being depressed mood or anhedonia), not meeting criteria for 1. or 2.	(0/0/0/1/1)		X X

Abbreviations: AGECAT: Algorithm Automated Geriatric Examination for Computer Assisted Taxonomy; BD: Bipolar disorder BDI (-II): Beck Depression Inventory (second edition); BDI-C: Beck Depression Inventory Child; CDI: Children’s Depression Inventory; CESD(-R): Centre for Epidemiological Studies Depression Scale (Revised); CESD-C: Centre for Epidemiological Studies Depression Scale for Children; CIDI: Composite International Diagnostic Interview; DICA-R-A: Diagnostic Interview for Children and Adolescents in the Revised Form according to DSM-III-R for Adolescents; DIS: Diagnostic Interview Schedule; DSM: Diagnostic and statistical manual of mental disorders; GDS: Geriatric Depression Scale; GMS: Geriatric Mental State; HAMD-17: Hamilton Depression Scale, 17 items; ICD-10-DCR: ICD-10 Diagnostic Criteria for Research; IDS-SR₃₀: Inventory of Depressive Symptomatology-Self Report; K6 (Short Screening Scale, Self-Report Questionnaire, 6 items); K10: Kessler 10; K-SADS: Kiddle Schedule for Affective Disorders and Schizophrenia; MD: major depression; MFQ: The Mood and Feeling Questionnaire; MINI: Mini International Neuropsychiatric Interview; MIN-K: Korean Version of the Mini International Neuropsychiatric Interview; PHQ (-9): Brief Patient Health Questionnaire Mood Scale (9 Item); RADS-2: Reynolds Adolescent Depression Scale; SADS: Schedule for Affective Disorders and Schizophrenia; SCID: Semi-Structured Clinical Interview; SD: subthreshold depression.

? - No information available 

• Combination of a validated instrument (mostly used: Centre for Epidemiological Studies Depression Scale-Revised [CESD], Brief Patient Health Questionnaire Mood Scale [PHQ-9] or Beck Depression Inventors [BDI-II]) to identify depressive symptoms and the exclusion of DSM-IV MD (exclusion sometimes also operationalised, e.g., via Composite International Diagnostic Interview [CIDI], Semi-Structured Clinical Interview [SCID] or Mini International Neuro-Psychiatric Interview [MINI]).

• Only some, but not all DSM-(IV) symptoms of MD, exclusion of DSM (-IV) MD sometimes operationalised via CIDI, SCID or MINI.

By far the most often used instrument was the CESD, which was used in 16 of the listed 61 trials in Table 2. The CESD is a 20-item measure (an eight-item version is also used), originally published by Radloff (1977). The patient or the caregiver is asked to rate depressive symptoms of the previous week like restless sleep, poor appetite, depressed mood, or feeling lonely (= rarely or none of the time, 1 = some or little of the time, 2 = moderately or much of the time, 3 = most or almost all the time). Scores range from 0 to 60, with high scores indicating pronounced depressive symptoms.

The most commonly used definition for SD using CESD is CESD ≥ 16. However, other cut-off values were also used, e.g., > 9 or between 8 and 15. Besides the different cut-off values, it is sometimes difficult to compare such defined populations across different studies, as different versions of the CESD exist.

The second most frequently used instrument was the PHQ-9 (Kroenke et al., 2001). This is a self-administered depression version of the PRIME-MD (Primary Care Evaluation of Mental Disorders). The patient must score every single symptom of the nine DSM symptoms regarding the past 14 days according to not at all (0), several days (1), more than half the days (2) and nearly every day (3). Thus, the instrument can not only be used to get a score representing the severity of depression but also to find a DSM diagnosis. Here, the most used operationalisation for SD is a score between five and 14, but five to nine or ≥ 9 are also used.

In the elderly, the most often used instrument is the Geriatric Depression Scale (GDS) (Montorio & Izal, 1996). This scale is a self-report measure of depression in the elderly. The respondent can answer “yes” or “no,” with each answer indicating depressive complaints being awarded 1 point. Two versions exist: firstly, the 30-item version called the GDS-long form (GDS-L). Secondly, 15 items showing a high correlation with depressive symptoms were chosen from these 30 items for the GDS-short form (GDS-S). However, the most used definition in the trials included is ≥ 16.

Epidemiology

Prevalence

General population

Here, epidemiological studies found are described, with only a few exceptions (Pickett et al., 2014; Wyman et al., 2020). Two large-scale representative studies examining the life-time resp. the 12-months prevalence of SD could be identified:

Pietrzak et al. (2013) used data from a representative sample of 43,093 adult participants of the NESARC (National Epidemiological Survey on Alcohol and Related Conditions) interviewed during a so-called wave I of this study, performed in the years 2001 and 2002. The lifetime prevalence of SD in this trial was 11.6%.

Stubbs et al. (2017a) described a 12-month SD prevalence of 2.5% (the respective finding for a brief depression episode was 2.7%, for a depressive episode 6.5%) in a population of 237,952 people. In this publication and others of this group, special attention was paid to the association between depressive symptoms and pain. The prevalence of severe pain in the last 30 days in the just mentioned trial of Stubbs et al. (2017a) across the entire sample was 10.7%, and the mean (SD) pain scores were 23.3 (26.0), 43.0 (26.2), 41.7 (26.8), and 49.9 (26.29) for no depression, subsyndromal depression, brief depressive episode, and depressive episode, respectively, while the prevalence of severe pain for these four conditions was 8.0, 28.2, 20.2, and 34.0%, respectively. Whereas the frequency of depressive episodes increased with pain intensity, this was not the case for brief depressive episodes or for SD. A paper published somewhat earlier (Stubbs et al., 2016), with the same approach and the same cohort included, focussed on back pain, showing an odds ratio of 22.21 for SD participants compared to controls. Similar prevalence rates reported by Stubbs et al. (2018) used a very similar sample to explore the association between depression and smoking. The prevalence of current smoking was higher in depressed participants, with odds ratios between 1.36 and 1.49 for SD, brief depressive episodes, and MD compared to people without depression. Also in this sample, Stubbs et al. (2017b) could show that physical health was worse in depressed participants than in controls, also in the SD-group.

Peters et al. (2015) found 3,901 participants with a life-time history of SD in a representative sample of 34,923 participants of the first wave of the NESARC in the United States. Special risk factors for developing MD three years later (diagnosed at the second wave of this trial) were low education, substance use, and younger age. These factors marginally increased the risk of MD.

Topuzoğlu et al. (2015) reported, based on a sample of 3,514 inhabitants of Izmir, Turkey, a 12-month prevalence rate of SD of 4.2% (SD: 0.3), of clinical depression of 8.2% (0.4). Jeuring et al. (2018) looked at the development of the rates of SD and MD over 20 years (1992, 2002, 2012) and found an increase in MD (2.1, 3.9, 3.8%, respectively), which was not present for SD (7.2, 8.7, and 6.2%). Higher socioeconomic and psychosocial conditions are regarded as protective against MD and SD.

Special populations

Diabetes mellitus

Schmitz et al. (2014) described in 1,064 patients with diabetes type 2 that nearly half of the participants suffered from at least one episode of SD. The more episodes of SD were present the higher the risk of poor functioning/impaired health-related quality of life was. Wang et al. (2018) reported a point prevalence for SD of 11.6% in 808 patients hospitalised due to diabetes type 2.

Elderly

On the basis of a representative sample of 10,409 participants being at least 55 years of age, Laborde-Lahoz et al. (2015) reported that 13.8% had a lifetime prevalence of SD (13.7% of MD). Participants with SD had significantly increased odds ratios (OR) for lifetime mood (adjusted odd-ratios 3.65–10.55), anxiety (1.61–2.50), and any personality (1.62) disorders. They also had increased odds of developing new-onset MD (1.44) as well as an anxiety disorder (1.52) from wave 1 to wave 2 (which was performed 2 years later).

In elderly patients, Xiao et al. (2016) investigated 1,068 elderly inhabitants of Shanghai, China, and found a point prevalence of SD of 1.7% (MD: 2.4%) in home care recipients (n = 811). Xiang et al. (2018) found a point prevalence of 38.7% for SD and 13.4% for MD in 811 community-dwelling adults aged 60 and over.

A very interesting study was presented by Sigström et al. (2018). In a population-based sample, 563 participants were investigated at the age of 70, 450 “survivors” at the age of 75 and 79 years. The cumulative 9-year-prevalence rate for SD was 30.9% (for minor depression 27.6, for MD 9.3%). Lee et al. (2013) found practically no differences regarding several demographic parameters between SD and MD patients in a non-representative sample of 315 elderly using community long-term care.

In a population-based sample, Ludvigsson et al. (2016) looked at participants born in 1922 and living in a certain region in Sweden. Of the 650 subjects identified, 496 responded. The authors reported a prevalence rate of 27% for SD (and 6% for MD). 22% of the SD persons received antidepressant treatment compared to only 17% of the MD persons.

Ho et al. (2016) used data collected from the National Mental Health Survey of the Elderly (NMHS-E), which was a nationally representative population-based survey sample of older adults. In 1,092 people, they found a point prevalence rate of 9.0% for SD (and 5.1% for MD).

Looking at a cross-sectional approach on a subgroup of the National Health and Nutrition Examination Survey (NHANES) being older than 50 years and taking opioids (n = 1,036), Brooks et al. (2019) reported a point prevalence rate for subthreshold depression of 22.8 (for moderate and severe depression of 13.0 and 11.8%).

Wu et al. (2020) found a point prevalence of 11.6% for SD and 5.1% for MD in a population-based cohort study using different catchment areas in 8 different low to middle-income countries including 15,991 people older than 64 years.

Using a similar sample as Wu et al. (2020), but a different definition of SD, Johansson et al. (2019) found a point prevalence of 5.4% for ICD-10 depression and 20.6% for SD in 11,474 older adults. Using the same design, Gonçalves-Pereira et al. (2019) found a point prevalence rate for SD of 18.0 and for ICD-10 depression of 4.4% among 1.405 older people in Southern Portugal.

In 6,640 elderly people in Korea (community-based, random sample), Oh et al. (2020) reported a point prevalence rate of 9.24% for SD, which was 2.4fold higher than that of syndromal depression. The incidence rate of subsyndromal depression was 21.70 per 1,000 person-years.

Vaccaro et al. (2017) estimated a point prevalence of 15.71% (95% confidence interval [C.I.]: 13.70–17.72) for SD and 5.58% (4.31–6.95) for MD in a very carefully performed epidemiological trial in a group of 1,321 elderly people (70–74 years), drawn from the responders of the cross-sectional phase of InveCe.Ab (Invecchiamento Cerebrale in Abbiategrasso, Brain Ageing in Abbiategrasso).

Almeida et al. (2015) published a cross-sectional study of 1,649 community-dwelling men aged at least 80 years. Of these, 14.5% suffered from SD (7.3% had clinically significant depression). Decreased interest in sex and anxiety before sex was associated with SD.

Adolescents

In the trial of Balázs et al. (2013), 12,395 adolescents aged between 14 and 16 years from the Saving and Empowering Young Lives in Europe (SEYLE) study were included. The primary aim of the SEYLE study was to compare the efficacy, cost-effectiveness and cultural adaptability of suicide-prevention strategies in schools (Wasserman et al., 2010) in a randomised trial. In each of the 11 participating countries, 1,000 participants from different schools were recruited. No diagnoses were established, and the level of depression (and anxiety) was measured via self-rating scales. The authors found 29.2% of the population to be subthreshold depressive and 10.5% depressed with high comorbidities regarding anxiety and subthreshold anxiety. Functional impairment, measured with the Strengths and Difficulties Questionnaire (SDQ), was reduced, and suicidality (Paykel Suicide Scale [PSS]) had increased already in the subthreshold depressive group, but even more, pronounced in the depressive group.

On the basis of 2,022 adolescents recruited from eight state-subsidized schools located in the northern part of Santiago, Chile, Crockett et al. (2020) reported a point prevalence of 16.5% for SD and 17.7% for MD. A major risk factor was the female sex.

Philipson et al. (2020) used data drawn from the Uppsala Longitudinal Adolescent Depression Study, a community-based cohort study initiated in Uppsala, Sweden, in the early 1990s. Comprehensive diagnostic assessments were conducted at the age of 16–17 years and in follow-up interviews 15 years later (these results will be reported under “3.2.2 Course”). In 321 adolescents, the authors found 64 with SD.

In a prospective cohort-based study comprising 274 primary care attenders (13–18 years), Gledhill and Garralda (2013) found 25% to be suffering from SD. In a meta-analysis comprising a total of 24 studies (Wesselhoeft et al., 2013), the different criteria to operationalise SD were listed (yielding considerable heterogeneity). Regarding epidemiological data, the authors summarised the results of 5 studies reporting estimates of life-time prevalence of SD and MD between 5.3–12.0 resp. 1.1–14.6%. 5 studies reported one-year prevalence rates to be 1.0–20.7 resp. 0.8–13.0%. In 6 out of 12 suited studies comparing prevalence rates for SD and MD, higher prevalence rates for SD were reported. The results regarding risk factors for SD were very heterogeneous, female gender was identified most consistently.

Hill et al. (2014) investigated 1,709 adolescents aged between 14 and 18 years with a mean age of 16.6 (SD: 1.2) years. The participants were randomly selected from nine high schools in Western Oregon. Of these 1,709 adolescents, 424 suffered from SD.

Chronic kidney disease

Shirazian et al. (2016) investigated a sample of 2,500 participants, drawn from data from the National Health and Nutrition Examination Surveys (NHANES). This is a multistage, stratified, clustered probability sample of the US civilian non-institutionalized population conducted in 2-year cycles by the National Centre for Health Statistics (NCHS). These surveys examine disease prevalences and trends over time. Chronic kidney disease (CKD) was defined as an estimated glomerular filtration rate of 15–60 ml/min/1.73 m² or urine albumin-to-creatinine ratio of ≥ 30 mg/g. For the 2,500 participants with CKD, the weighted point prevalence was 21.4% for SD symptoms and 3.1% for severe depressive symptoms. Awareness (of the CKD and resulting consequences) was significantly associated with depressive symptoms.

Visual impaired

In 914 visual-impaired patients attending outpatient low-vision rehabilitation centres in the Netherlands and Flanders, van der Aa et al. (2015a) found a point prevalence of 47.6% for participants suffering either from subthreshold depressive or subthreshold anxiety symptoms.

Course

General population

In the trial of Pietrzak et al. (2013) already reported above, two waves were performed: wave 1 in the years 2001 and 2002, and wave 2 in the years 2004 and 2005. In individuals showing an SD at wave 1, the risk of developing a major depression until wave II was increased (ORs 1.72–2.05) compared to those participants showing no depression. The same was true for dysthymia, social phobia, and generalised anxiety disorder (ORs 1.41–2.92). Among respondents with SD at wave 1, Cluster A and B personality disorders and worse mental health status were associated with an increased likelihood of developing incident major depression at wave 2.

A very interesting and comprehensive meta-analysis in this context was published by Lee et al. (2019). In a systematic literature review, the group identified 16 studies (total n = 67,319) investigating the question of whether the incidence of MD after SD is elevated. The incidence rate ratio (IRR) for people with SD to develop MD was 1.95 (95% C.I.: 1.28–2.97). Subgroup analyses estimated similar IRR for different age groups (youth, adults and elderly) and sample types (community-based and primary care). Sensitivity analyses showed the robustness of the results with regard to different sources of study heterogeneity.

Some reports focussed on the effect of different interventions to lower the conversion rate from SD to MD. Cuijpers et al. (2016) reported in a meta-analysis comprising 5 studies that interpersonal psychotherapy lowers the conversion rate of SD into MD (follow-up time 3–18 months). The odds ratio compared to the control group was 0.30 (95% C.I.: 0.10–0.88). Similar results were described by Reins et al. (2021) who meta-analytically evaluated the results of 7 trials with a total of 2,186 participants for internet-based psychotherapy. Krishna et al. (2013) reported in their review consisting of 4 trials that group psychological interventions in older adults with SD had a significant effect on depressive symptomatology, which is not maintained on follow-up. Buntrock et al. (2016) investigated in a controlled study looking at the effect of a web-based guided self-help intervention whether the conversion rate of SD to MD in a follow-up of 1 year could be changed. After one year, 27% of participants in the intervention group developed MD, compared to 41% in the control group. The same group (Krishna et al., 2015) reported on meta-analytical results of 8 trials. These were similar to the first published review with only 4 trials. The authors additionally reported that the conversion rate from SD into MD was not reduced by group psychotherapy. In a controlled clinical trial investigating the effect of a stepped care approach in SD patients with diabetes mellitus type 2 and/or coronary heart disease with a follow-up of 12 months, no difference regarding the conversion rate to MD between the intervention and the control group could be found (Pols et al., 2017). This was also true after an observation period of 2 years (Pols et al., 2018).

Karsten et al. (2013) used data from the Netherlands Study of Depression and Anxiety (N = 1,266, aged 18–65). Linear mixed models were used to identify predictors of functional impairment at baseline, 1- and 2-year follow-up. The authors found a higher functional impairment in participants with SD at baseline. At the follow-up after 1 and 2 years, this functional impairment improved but remained much higher than in controls.

Special populations

Adolescents

In the investigation of Hill et al. (2014) (see above), the participants (mean age: 16.6 years) were seen again after one year, at the age of 24, and at the age of 30 years. 144 of the original sample of 1.709 adolescents developed MD, the most important risk factors for this development were poor friend support and a history of a (comorbid) anxiety or substance use disorder.

Jinnin et al. (2016) screened 2,494 first-year undergraduate students (2,281 responded) from Hiroshima University in Japan with the help of the Beck Depression Inventory (2^nd edition, BDI-II). The students were 18 or 19 years old and had not had an MD in the past 12 months. The total group was separated into three subgroups: low symptom group (BDI-II-score ≤ 10), middle symptom group (BDI-II-score between 11 and 17), and high symptom group (BDI-II-score ≥ 18); the latter group was considered to suffer from SD. Originally, the numbers of students belonging to the different classes were 1,918, 276 resp. 87. After contacting and performing other procedures, 68, 57 resp. 66 participants remained. These participants were examined every two months for 1 year. Using different statistical methods (mainly growth mixture modelling [GMM]), three classes could be identified. In the “increasing group” (n = 26), by far the most (n = 22) belonged to the SD group. Three students in the SD group developed MD during the 12-months follow-up.

Gledhill and Garralda (2013) (see above) reported that, after 6 months, 57% of adolescents suffering from SD still showed persistent depressive symptoms and 12% had developed MD. In a 15-years follow-up, Philipson et al. (2020) (see above) found that SD was, in contrast to persistent depressive disorder, not linked to lower income.

Elderly

In an 8-year follow-up of data of 496 very old participants of a prospective cohort study (Ludvigsson et al., 2016, see above), Ludvigsson et al. (2019) showed that mortality was increased in a group of MD-participants but not in the SD-group compared to controls. Morbidity, especially activities of daily living and quality-of-life parameters were increased in the MD- and the SD-group compared to controls. Ludvigsson et al. (2018) could show in a similar study that direct costs in people with SD were 1.48-fold higher than in controls.

Zhang et al. (2020) found in 1,188 older people without dementia (455 with mild cognitive impairment [MCI], 733 without MCI), a subgroup of the Alzheimer’s Disease Neuroimaging Initiative (ADNI), that SD at baseline was associated with a significant cognitive decline in cognition during the 5 years follow-up. Also, SD participants showed significantly accelerated atrophy in the hippocampus and middle temporal gyrus.

In the already reported trial of Sigström et al. (2018), the cross-sectional prevalence rate for SD (and minor depression) increased over time. If participants had suffered from SD at the age of 70 years, 47.2% had SD at follow-up and 36.1% had MD or minor depression. In this population-based sample, more than half of the participants experienced some degree of depression during their eighth decade of life. Most new episodes of MD occurred in people who were already in a depressive state (SD, minor depression) at baseline. As the authors stated, this could have implications for late-life depression strategies.

In a 10 years follow-up of the already reported trial of Ho et al. (2016), no higher mortality risk for SD could be identified. For MD, higher mortality risk was mainly due to increased cardiovascular and stroke mortality. Based on data from randomly selected 1,000 participants older than 65 years, Jeong et al. (2013) reported that MD was associated with higher mortality in the follow-up phase of this trial, but SD was not.

At the follow-up (average follow-up time of 3.9 years) of the participants of the already reported Wu et al. (2020) trial, it was found for MD (and subthreshold anxiety), but not for SD, that higher risk mortality existed (hazard ratio 1.45, 95% C.I.: 1.24–1.70).

The research group of the already mentioned trial reported by Johansson et al. (2019) conducted a follow-up after four years and found that people with depression and with SD showed a higher risk for the development of dementia (hazard ratio 1.63, 95% C.I.: 1.26–2.11 resp. 1.28, 1.09–1.51).

Visual impaired

van der Aa et al. (2015b) (main results see above, van der Aa et al., 2015a) described that in visually impaired participants suffering either from subthreshold depressive or subthreshold anxiety symptoms (47.6% of the sample), 34% recovered from their depressive/anxiety symptoms, while 18% developed a depressive and/or anxiety disorder after a watchful waiting period of 3 months.

Discussion

The main results of this paper are:

• SD has a higher prevalence than MD, although a precise estimation is not possible, since different definitions of SD in huge epidemiological trials have been used. All evaluated subgroups are concerned; in the elderly and adolescents, the prevalence seems to be even higher than in the general population. This might be due to the particular, albeit different, stress factors in these age groups.

• The impact of SD on quality of life and related parameters are in the range of the respective impact of MD, but in most trials, in which such parameters were evaluated, it was slightly weaker than in MD.

• In a considerable number of cases, conversion from SD to MD takes place.

• Therapeutic options to treat SD were not the main topic of this paper. However, it should be noted that a group of trials could demonstrate that therapeutic approaches might be helpful for SD and might also be helpful for the prevention of conversion from SD to MD.

Since SD, as shown, is such an important issue having significant implications, early diagnosis and treatment is necessary, but this is currently not taken into account. The importance of subthreshold diagnoses can also be seen in the fact that most suicide attempters have also subthreshold psychiatric disorders (Balázs et al., 2000). However, before recommendations for early diagnosis and treatment can be made, a precise definition/description of the disease entity, used by most clinicians and researchers, is needed.

Nevertheless, as another main result of the presented review, it can be stated that such a well-recognized definition does not exist. The opposite is the case: There is a huge heterogeneity of definitions and no consensus can be derived from the published literature.

Before going into a little more detail, it is important to be aware of the field in which such definitions are used:

• Clinical issues, i.e., finding patients suffering from SD in a quick and reliable manner to decide whether a treatment is necessary or not.

• Research issues, i.e., precisely defining a target population for research.

Regarding clinical issues, the routine application of a scale – even a self-evaluation scale like the PHQ-9 – is not feasible, since clinicians do not use scales in the routine setting. In different settings, however, this is strongly recommended, especially to document long-term changes. In this context, a definition based on the DSM- or ICD-system is needed, since clinicians are familiar with these diagnostic categories. As SD is, like MD, probably an episodic disorder, the same definition as for dysthymia (in ICD-10) could be used in order to describe symptom severity (“A chronic depression of mood […] which is not sufficiently severe […] to justify a diagnosis of severe, moderate, or mild recurrent depressive disorder). However, the duration should be at least 14 days (as is the case for depressive episodes) but shorter than two years (to be distinguished from dysthymia). The disadvantage of this proposal based purely on clinical categories is the overlap with the DSM definition of minor depression (i.e., the total number of symptoms not exceeding 4).

Regarding research issues, the use of scales to define the target population is a very common approach. On the other side, as has been shown in this paper, the most frequently used scales such as CESD or PHQ-9 are not primarily suited to determine symptom severity of SD, because they are more or fewer operationalizations of DSM symptoms. Moreover, these scales are completed by the patient or the caregiver. In our view, it is very astonishing that in not a single identified research paper was an observer-based scale such as the Montgomery-Ӑsberg Depression Rating Scale (MDRS) used to define symptom severity more precisely. This is, of course, related to the nature of most of the trials identified, which is epidemiological research where observer-based scales are due to pragmatic reasons (high number of included patients) not feasible. But such an approach was also not chosen in small-scale trials. Self and observer-rating scales, however, measure in part different symptom dimensions (e.g., Seemüller et al., 2022). Another research deficit is that intervention studies have rarely been performed, especially with regard to pharmacological approaches.

To return to the main topic of this paragraph, a clear-cut, well-recognized and generally used definition of SD is lacking. However, the main parameters of such a definition are clear:

• No MD (DSM) or depressive episode (ICD) (and also no history of such conditions).

• A certain number of symptoms is mandatory, also comprising core symptoms of a depressive episode like depressed mood, anhedonia or loss of energy for a certain time span, e.g., 14 days.

• The definition should (for research issues) not only be based on the presence of certain symptoms but should also be based on an additional patient- or caregiver-based instrument in order to enhance the reliability of the number-of-symptoms-based approach; even better would be an observer-(physician/psychologist) rated scale.

• For intervention studies, an observer-based instrument seems to be mandatory.

Conclusions

Scrutinising the reported results and other papers in the field, clear conclusions are not possible. Most definitions are arbitrary, the overlap (as stated, e.g., between SD and minor depression) is considerable, and the epidemiological results are, mainly for this reason, highly variable.

Nevertheless, keeping these points in mind and incorporating the main results of our review paper, one proposal for the definition to achieve consensus in this highly important field could be:

• Presence of least two DSM depressive symptoms for at least two weeks, one symptom of depressed mood, no MD or minor depression.

• PHQ-9 between five and nine and/or CESD at least 16.

• MDRS between ten and 18 (for 2 weeks).

Hopefully, this proposal leads to a discussion in the field resulting in much more homogeneous inclusion criteria; otherwise, the so important research in this field is of reduced impact only.

Keypoints section

A clear-cut, well-recognized and generally used definition of SD is lacking. Nevertheless, keeping these points in mind and incorporating the main results of our review paper, one proposal for a definition to achieve consensus in this highly important field could be:

• Presence of least two DSM depressive symptoms for at least two weeks, one symptom of depressed mood, no MD or minor depression.

• PHQ-9 between five and nine and/or CESD at least 16.

• MDRS between ten and 18 (for 2 weeks).

Disclosure statement

Hans-Peter Volz has served as a consultant or on advisory boards for Astra/Zeneca, Eli Lilly, Lundbeck, Pfizer, Schwabe, Janssen, Otsuka, Angelini, and Sage and has served on speakers’ bureaus for Astra/Zeneca, Eli Lilly, Lundbeck, Schwabe, Janssen, Bayer, Recordati, and neuraxpharm. Siegfried Kasper has received grants/research support, consulting fees and/or honoraria within the last three years; grant/research support from Lundbeck; he has served as a consultant or on advisory boards Celegne, IQVIA, Janssen, Lundbeck. Mundipharma, Recordati, Takeda and Schwabe; and he has served on speakers bureaus for Angelini, Aspen Farmaceutica S.A., Janssen, Krka Pharma, Lundbeck, Medichem Pharmaceuticals Inc., Neuraxpharma, OM Pharma, Pierre Fabre, Sanofi, Servier, Schwabe, Sun Pharma. Hans-Jürgen Möller has received grant/research support, consulting fees and honoraria within the last years from AstraZeneca, Bristol-Myers Squibb, Eli Lilly. GlaxoSmithKline, Janssen Cilag, Lundbeck, MSD, Novartis, Organon, Otsuka, Pfizer, Schwabe, Sepracor, Servier, and Wyeth. Erich Seifritz has received honoraria from Schwabe GmbH for educational lectures. He has further received educational grants and consulting fees from Janssen Cilag, Lundbeck, Angelini, Otsuka, Servier, Recordati, Vifor, Sunovion, and Mepha.

Additional information

Funding

This research and its publication were supported by the Dr. Willmar Schwabe GmbH & Co. KG.