Research report
A trajectory-based approach to understand the factors associated with persistent depressive symptoms in primary care

https://doi.org/10.1016/j.jad.2012.12.021Get rights and content

Abstract

Background

Depression screening in primary care yields high numbers. Knowledge of how depressive symptoms change over time is limited, making decisions about type, intensity, frequency and length of treatment and follow-up difficult. This study is aimed to identify depressive symptom trajectories and associated socio–demographic, co-morbidity, health service use and treatment factors to inform clinical care.

Methods

789 people scoring 16 or more on the CES-D recruited from 30 randomly selected Australian family practices. Depressive symptoms are measured using PHQ-9 at 3, 6, 9 and 12 months.

Results

Growth mixture modelling identified a five-class trajectory model as the best fitting (lowest Bayesian Information Criterion): three groups were static (mild (n=532), moderate (n=138) and severe (n=69)) and two were dynamic (decreasing severity (n=32) and increasing severity (n=18)). The mild symptom trajectory was the most common (n=532). The severe symptom trajectory group (n=69) differed significantly from the mild symptom trajectory group on most variables. The severe and moderate groups were characterised by high levels of disadvantage, abuse, morbidity and disability. Decreasing and increasing severity trajectory classes were similar on most variables.

Limitations

Adult only cohort, self-report measures.

Conclusions

Most symptom trajectories remained static, suggesting that depression, as it presents in primary care, is not always an episodic disorder. The findings indicate future directions for building prognostic models to distinguish those who are likely to have a mild course from those who are likely to follow more severe trajectories. Determining appropriate clinical responses based upon a likely depression course requires further research.

Introduction

While mortality rates from cardiovascular disease, stroke and cancer are steadily decreasing, there is no evidence of reduced morbidity or mortality rates due to depression (Insel, 2009). Studies continue to find substantial numbers of people with unmet needs (Parslow and Jorm, 2000, Kendrick et al., 2009, Prins et al., 2011, Coyne et al., 2002) and a large mismatch between individual needs and how the system responds (Gunn et al., 2010, Palmer et al., 2010, Herrman et al., 2002). This is despite the public attention that depression has received (Dumesnil and Verger, 2009) in countries such as the US (U.S. Preventive Services Task Force, 2009), the UK (Rix et al., 1999; Dunion and Gordon, 2005), Australia (Jorm et al., 2005) and New Zealand (Vaughan and Hansen, 2004); the wide scale use of antidepressants (Lockhart and Guthrie, 2011, Moore et al., 2009, Marcus and Olfson, 2010) and the increased use of screening for depression within primary care (Kessler et al., 2005, U.S. Preventive services Task Force, 2009, Ööpik et al., 2006).

Current screening and diagnostic approaches used for depression are limited, especially in the primary care setting, where it is most commonly managed (Brown and Barlow, 2009, Helzer et al., 2006, Katerndahl et al., 2005, Klein, 2008, Lamers et al., 2010, Vuorilehto et al., 2005). Current screening approaches identify large numbers of potential cases. For example, the LIDO study screened 18,489 patients in primary care waiting rooms in six countries and found that 24–55% rated as ‘probably depressed’; scoring 16 or more on the CES-D (Herrman et al., 2002). This level of identification raises important challenges for primary care, which deals with an unreferred, heterogeneous patient population at different stages and severity of depression course.

Most guidelines base treatment advice on the level of severity of symptoms and the degree to which patients meet diagnostic criteria of Major Depressive Disorder (MDD) at one assessment point (Hegarty et al., 2009). Studies in the UK and NZ have identified problems with this approach as they do not assist in guiding management for the large numbers of people with sub-syndromal depression (Thompson et al., 2001, Magpie Research Group, 2005, Magpie Research Group et al., 2006). The forthcoming DSM 5 is likely to adopt a dimensional approach to diagnosis (Regier et al., 2010) supporting the need for in-depth understanding of how symptoms change over time and over the severity spectrum.

Knowledge which charts depression course in primary care or community samples is limited and contradictory. Since the mid 1990s a number of primary care based observational (Herrman et al., 2002, Van Weel-Baumgarten et al., 1998, Van Weel-Baumgarten et al., 2000, Dowrick et al., 1998) and intervention studies (Rost et al., 2001, Aikens et al., 2008) have highlighted the complex nature of depression in primary care. Findings from the Netherlands that, of those who were prescribed antidepressants, 60% had no recurrence in a 10 year period (Van Weel-Baumgarten et al., 1998), contrast with Finnish findings that most cases of depression seen in primary care are recurrences or relapses, rather than new cases (Vuorilehto et al., 2005). Neither of these studies used a trajectory based approach.

Studies documenting depression courses using trajectories based on longitudinal data sets are emerging from the United States of America (Beard et al., 2008, Dunn et al., 2011, Interian et al., 2011, Klein et al., 2008, Kuchibhatla and Fillenbaum, 2011, Lincoln and Takeuchi, 2010, Stoolmiller et al., 2005, Tomey et al., 2010, Mora et al., 2009, Costello et al., 2008, Cui et al., 2008, Yaroslavsky et al., 2012), Canada (Brendgen et al., 2005, Brendgen et al., 2010, Colman et al., 2011), Switzerland (Merikangas et al., 2003), Singapore (Hong et al., 2009), Taiwan (Chen et al., 2011, Huang et al., 2011), the UK (Colman et al., 2007), the Netherlands (Dekker et al., 2007, Rhebergen et al., 2012), and Norway (Skipstein et al., 2010). These studies include selective samples of community dwelling children and adolescents (Brendgen et al., 2010, Brendgen et al., 2005, Dekker et al., 2007, Stoolmiller et al., 2005, Yaroslavsky et al., 2012), recent mothers (Mora et al., 2009, Skipstein et al., 2010), women in mid-life (Tomey et al., 2010), women after breast cancer diagnosis (Dunn et al., 2011), patients with heart failure (Kuchibhatla and Fillenbaum, 2011), older people (Cui et al., 2008, Chen et al., 2011, Huang et al., 2011) and specific cultural groups (Interian et al., 2011, Beard et al., 2008, Merikangas et al., 2003). Wide variation in the interval between symptom measurements is reported, with most being six months or longer between follow-up times. This is particularly problematic when depression symptom measurement tools commonly ask about the past two weeks or month. The trajectories reported in the identified studies may not represent the true underlying trajectories due to the long time between measurement intervals. Two studies addressed this issue; Dunn et al. (2011), measured symptoms using the CES-D, monthly for six months, in a group of 389 women following breast cancer surgery and Kuchibhatla and Fillenbaum (2011) measured depressive symptoms with the Hamilton Depression Rating Scale at two-weekly intervals for 14 weeks in 469 patients with heart failure who met criteria for major depression. Both studies are limited by the fact that all participants have recently received major treatment interventions for serious health conditions.

Two studies have recruited a primary care sample; one from a Latino population (n=220) (Interian et al., 2011) and another from people aged over 65 years (n=392) (Cui et al., 2008). The findings from these studies are limited by their selective population approach, modest sample sizes, length of time between follow-up intervals (6 months to 1 year) and reliance on retrospective measurement tools.

The Netherlands Study of Depression and Anxiety (NESDA) consists of a large, mixed cohort from primary care, outpatient and population samples and has demonstrated, using latent class growth analysis, that current DSM categories do not adequately represent course trajectories for those with current Major Depressive Disorder (MDD) and/or Dysthymia (Rhebergen et al., 2012). This finding highlights the need to expand the focus beyond those who reach diagnostic thresholds for MDD or Dysthymia to include those with depressive symptoms across the severity spectrum. Understanding depressive symptom trajectories, the factors associated with a particular course and how these relate to functioning and quality of life is of major clinical importance. Such knowledge would help to identify those at low and higher risk of poorer outcomes. This identification could assist in treatment decisions and service planning; especially when deciding what to do with the large numbers who screen positive for probable depression yet have symptoms at sub-syndromal level.

Our main objectives were to (1) identify depressive symptom trajectories in a large cohort of primary care attendees participating in the diamond longitudinal study (Gunn et al., 2008) and (2) examine associations between depressive symptom trajectories and a wide range of factors related to socio–demography, co-morbidity, health service use and treatment which could be used to inform clinical care. We document how we have used latent class growth mixture modelling (GMM) to identify naturally occurring groups based upon depressive symptoms and classify individuals into subgroups based on the similarity of symptom levels over time (Muthén and Muthén, 2000).

Section snippets

Design

Data were collected as part of the diamond prospective longitudinal cohort study investigating what happens to people with depressive symptoms over time. It is one of the largest primary care depression cohort studies worldwide. Diamond is informed by a social model of health and full details of study methods are published elsewhere (Gunn et al., 2008, Potiriadis et al., 2008, Boardman et al., 2011). Diamond seeks to investigate in-depth and over time how primary care responds to people with

Results

Sample characteristics: Table 1 presents the baseline characteristics of the cohort and for the entire screening sample. The cohort was similar in age to those screened yet cohort participants were more likely to be female, not married, to live alone and to report increased levels of social disadvantage and poor health. The cohort had a mean age of 48 years (range from 18 to 75 years; median=48). More than two-thirds of the participants were female and most were born in Australia. The mean

Discussion

Using growth mixture modelling and the PHQ-9 as a continuous measure, we identified five distinct depressive symptom trajectories within this primary care cohort. The most striking findings are the degree to which symptom trajectories remained static over time and the associations between the “severe” group and the high levels of anxiety and abuse. These findings suggest that depression is not always an episodic disorder, especially as it presents in primary care and raises the possibility that

Conclusions

Cross-sectional categorical diagnostic approaches provide little guidance as to the likely course of depressive symptoms (Rhebergen et al., 2012). The findings presented here indicate future directions for building prognostic models to distinguish those who are likely to have a mild course from those who are likely to follow more severe trajectories. Such models would assist clinical decision making and better targeting of interventions and will require prognostic tools which are easy to apply

Role of funding source

The diamond study was initiated with pilot funding from the beyondblue Victorian Centre of Excellence and the main cohort has received project grant funding from the Australian National Health and Medical Research Council (IDs 299869, 454463, 566511 and 1002908). The one year Computer Assisted Telephone Interview was funded by a Stream 3 grant from the Australian Primary Health Care Research Institute (APHCRI). No funding body had a role in study design; the collection, analysis, and

Conflict of interest

No conflict declared.

Acknowledgements

The named authors submit this publication on behalf of the diamond study investigators which include: Professor Jane Gunn, Professor Helen Herrman, Professor Mike Kyrios, A/Professor Kelsey Hegarty, Professor Christopher Dowrick, Dr. Gail Gilchrist, A/Professor Grant Blashki, Professor Dimity Pond, Dr. Patty Chondros, A/Professor Renata Kokanovic and Dr. Victoria Palmer. We acknowledge the 30 dedicated GPs, their patients and practice staff for making this research possible. We thank the cohort

References (86)

  • M.N. Kuchibhatla et al.

    Trajectory classes of depression in a randomized depression trial of heart failure patients: a reanalysis of the SADHART-CHF Trial

    American Journal of Geriatric Pharmacotherapy

    (2011)
  • K.M. Rost et al.

    The quality improvement for depression collaboration: general analytic strategies for a coordinated study of quality improvement in depression care

    General Hospital Psychiatry

    (2001)
  • K. Tomey et al.

    Physical functioning among mid-life women: associations with trajectory of depressive symptoms

    Social Science and Medicine

    (2010)
  • H. Akaike

    A new look at the statistical model identification

    IEEE Transactions on Automatic Control

    (1974)
  • T. Armstrong et al.

    Physical Activity Patterns of Australian Adults: Results of the 1999 National Physical Activity Survey

    (2000)
  • F.E. Baum et al.

    Epidemiology of participation: an Australian community study

    Journal of Epidemiology and Community Health

    (2000)
  • M. Brendgen et al.

    Links between friendship relations and early adolescents' trajectories of depressed mood

    Developmental Psychology

    (2010)
  • M. Brendgen et al.

    Relations with parents and with peers, temperament, and trajectories of depressed mood during early adolescence

    Journal of Abnormal Child Psychology

    (2005)
  • T. Brown et al.

    A proposal for a dimensional classification system based on the shared features of the DSM-IV anxiety and mood disorders: implications for assessment and treatment

    Psychological Assessment

    (2009)
  • I. Colman et al.

    Predictors of long-term prognosis of depression

    Canadian Medical Association Journal

    (2011)
  • D.M. Costello et al.

    Risk and protective factors associated with trajectories of depressed mood from adolescence to early adulthood

    Journal of Consulting and Clinical Psychology

    (2008)
  • J.C. Coyne et al.

    Emotional disorders in primary care

    Journal of Consulting and Clinical Psychology

    (2002)
  • J. Das-Munshi et al.

    Public health significance of mixed anxiety and depression: beyond current classification

    British Journal of Psychiatry

    (2008)
  • M.C. Dekker et al.

    Developmental trajectories of depressive symptoms from early childhood to late adolescence: gender differences and adult outcome

    Journal of Child Psychology and Psychiatry

    (2007)
  • C. Dowrick et al.

    Outcomes of Depression International Network (ODIN). Background, methods and field trials. ODIN Group

    British Journal of Psychiatry

    (1998)
  • H. Dumesnil et al.

    Public awareness campaigns about depression and suicide: a review

    Psychiatric Services

    (2009)
  • Dunion, L., Gordon, L., 2005. Tackling the attitude problem. The achievements to date of Scotland's ‘see me’...
  • L. Dunn et al.

    Identification of distinct depressive symptom trajectories in women following surgery for breast cancer

    Health Psychology

    (2011)
  • P. Elliott et al.

    Patterns of treatment response in chronic posttraumatic stress disorder: an application of latent growth mixture modeling

    Journal of Traumatic Stress

    (2005)
  • G. Gilchrist et al.

    Observational studies of depression in primary care: what do we know?

    BMC Family Practice

    (2007)
  • J. Gunn et al.

    The association between chronic illness, multimorbidity and depressive symptoms in an Australian primary care cohort

    Social Psychiatry and Psychiatric Epidemiology

    (2012)
  • J. Gunn et al.

    Embedding effective depression care: using theory for primary care organisational and systems change

    Implementation Science

    (2010)
  • J.M. Gunn et al.

    Who is identified when screening for depression is undertaken in general practice? Baseline findings from the Diagnosis, Management and Outcomes of Depression in Primary Care (diamond) longitudinal study

    Medical Journal of Australia

    (2008)
  • K. Hegarty et al.

    How could depression guidelines be made more relevant and applicable to primary care? A quantitative and qualitative review of national guidelines

    British Journal of General Practice

    (2009)
  • K. Hegarty et al.

    A multidimensional definition of partner abuse: development and preliminary validation of the Composite Abuse Scale

    Journal of Family Violence

    (1999)
  • J.E. Helzer et al.

    The feasibility and need for dimensional psychiatric diagnoses

    Psychological Medicine

    (2006)
  • H. Herrman et al.

    Longitudinal investigation of depression outcomes in primary care in six countries: the LIDO study. Functional status, health service use and treatment of people with depressive symptoms

    Psychological Medicine

    (2002)
  • S.-I. Hong et al.

    Structural relationships between social activities and longitudinal trajectories of depression among older adults

    The Gerontologist

    (2009)
  • A.V. Horwitz et al.

    The Loss of Sadness: How Psychiatry Transformed Normal Sorrow Into Depressive Disorder

    (2007)
  • T.R. Insel

    Translating scientific opportunity into public health impact: a strategic plan for research on mental illness

    Archives of General Psychiatry

    (2009)
  • A.F. Jorm et al.

    The impact of beyondblue: the national depression initiative on the Australian public's recognition of depression and beliefs about treatments

    Australian and New Zealand Journal of Psychiatry

    (2005)
  • R.E. Kass et al.

    Bayes factors

    Journal of the American Statistical Association

    (1995)
  • D.A. Katerndahl et al.

    Reflections on DSM classification and its utility in primary care: case studies in “Mental Disorders”

    Primary Care Companion to the Journal of Clinical Psychiatry

    (2005)
  • Cited by (38)

    • Depressive symptom trajectories and polygenic risk scores in individuals with an immune-mediated inflammatory disease

      2022, General Hospital Psychiatry
      Citation Excerpt :

      This semiparametric technique was used to identify subgroups of individuals with distinct longitudinal trajectories for PHQ-9 scores. This method has been used to model longitudinal depressive symptoms in primary care and in the older population [31,45,46]. Individuals were included if they had ≥2 PHQ-9 scores to capture the change in depressive symptoms over time.

    • Clinical efficacy of a Decision Support Tool (Link-me) to guide intensity of mental health care in primary practice: a pragmatic stratified randomised controlled trial

      2021, The Lancet Psychiatry
      Citation Excerpt :

      Mental health problems are most often identified and managed in primary care,1 with anxiety and depression accounting for the majority of presentations.2 The nature of these presentations and the illness trajectory varies widely.3 Managing both the volume and heterogeneity of mental health problems that patients present with is a substantial challenge, and both undertreatment and overtreatment are common.4

    • Development of a prognostic model for predicting depression severity in adult primary patients with depressive symptoms using the diamond longitudinal study

      2018, Journal of Affective Disorders
      Citation Excerpt :

      Using the diamond cohort provided a strong study design to develop the prognostic model (Moons et al., 2009a). We were able to prospectively map the natural course of depressive symptoms over time, (Gunn et al., 2013) and use predictors that were well defined and reproducible which increases the generalisability of the model (Moons et al., 2009a). Unlike many prognostic models which are developed in secondary care and then applied to primary care (Moons et al., 2009a), our prognostic model was specifically developed for use in general practice using data collected from general practice attendees.

    • Analysis of depression trajectory patterns using collaborative learning

      2016, Mathematical Biosciences
      Citation Excerpt :

      Existing methods of disease trajectory analysis came from diverse disciplines such as statistics, machine learning, and engineering prognostics [11–15]. However challenges remain for processing the heterogeneous and irregular depression trajectory data: depression has complex dynamics embedded in individual's disease trajectory that is further complicated by the widely reported heterogeneity of the dynamics in the population [7,8]. Thus, we assemble a good combination of statistical tools, tailored for the data challenges that we encounter in the depression EHR dataset, to discover the representative patterns for the heterogeneous depression dynamics and predict individual depression trajectories.

    View all citing articles on Scopus
    View full text