This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Research Protocols, is properly cited. The complete bibliographic information, a link to the original publication on http://www.researchprotocols.org, as well as this copyright and license information must be included.
Severe life events are acknowledged as important etiological factors in the development of clinical disorders, including major depression. Interview methods capable of assessing context and meaning of events have demonstrated superior validity compared with checklist questionnaire methods and arguments for interview approaches have resurfaced because choosing the appropriate assessment tool provides clarity of information about gene-environment interactions in depression. Such approaches also have greater potential for understanding and treating clinical cases or for use in interventions.
(1) To argue that life events need sophisticated measurement not satisfactorily captured in checklist approaches. (2) To review life-events measures and key findings related to disorder, exemplifying depression. (3) To describe an ongoing study with a new online measure and to assess its psychometric properties and the association of life events in relation to disorder and educational outcomes.
The Computerised Life Events Assessment Record (CLEAR) is under development as a tool for online assessment of adult life events. Based on the Life Events and Difficulties Schedule interview, CLEAR seeks to assess life events to self and close others, link these to other events and difficulties, and utilize calendar-based timing, to improve upon checklist approaches.
The CLEAR study is in the preliminary stages and its results are expected to be made available by the end of 2015.
There is currently no sophisticated technological application of social risk factor assessment, such as life events and difficulties. CLEAR is designed to gather reliable and valid life-event data while combating the limitations of interviews (eg, time consuming and costly) and life-event checklists (eg, inability to accurately measure severity and independence of life events). The advantages of using such innovative methodology for research, clinical practice, and interventions are discussed.
Links between life events and clinical disorders have a long history, given the fact that stressful life events are an important predictor of the onset and course of various disorders across the life span, including depression, eating disorders, and psychosis [
Empirical investigation of life events and disorder started with checklist self-report approaches in the 1960s [
This paper outlines the ineffective measurement of life events in many contemporary research studies. It also presents a new online computerized approach—Computerised Life Events Assessment Record (CLEAR)—designed to optimize interview advantages while incurring low cost and being time effective. The ongoing development and future testing of CLEAR will be outlined with a focus on clinical health. It is expected that this new online method will offer an enhanced but readily available life-event measure with important implications for studying disorders.
The development of CLEAR has implications for genetic studies of depression as well as for more effective clinical application. For instance, some individuals are more likely to experience severe life events, because of psychosocial vulnerability (eg, difficulty in relationships resulting in more relationship events) [
Among the different in-person semistructured interviews, probably the most widely used is the LEDS [
This interview is considered the “gold standard” for measuring life stress and is superior to checklist approaches. The disadvantage comes from the time and labor involved [
The early investigation of life events by checklist (eg, Holmes and Rahe questionnaire [
The other aspect involves the estimated severity of the event in terms of a likely stressful and negative emotional response in most people. In checklist approaches, this is generically ascribed. Yet, apart from the worst ones (eg, death of a spouse), almost all are dependent on context for their likely severity. For example, marriage and pregnancy are not inherently stressful unless the context is negative (eg, unplanned pregnancy, unstable partnership, health risks, or financial and housing difficulties), where a much higher stress score is allocated. A more recent checklist identified those events most often scored as severe life events in interview measures [
Interview measures have found that the likely meaning of an experience plays a central etiological role in the development of depression, with life events tied to changes involving loss (of relationship, role, cherished idea, or sense of self), danger (threat of a future loss, conflicts in core social roles, threats to plans you have made), or punishing environments (entrapment, humiliation) being the most predictive of disorder [
The timing of events is critical to determining their etiological role in depression onset. Events that occur after onset can only have a maintenance role at best. Therefore, precise timing of events is required. In addition, other important stressors are chronic, with severity levels that can vary over time. These are termed “difficulties” and comprise problematic situations, which last 4 weeks or more, and can go on for years. These can occur in as many domains as events, and can be antecedent or consequent to the event. An important analysis of such linkages showed that an event preceded by a severe difficulty (hence “matched”) for at least 6 months and in the same domain greatly increased the risk of depression onset [
Life-event interviews also categorize “independence” of the event. This is the extent to which the event is likely to be separate from the actions, planning, or control of the individual, that is, it occurs externally to the individual. Independence allows researchers and clinicians to estimate whether the event is a cause or consequence of disorder. For example, losing a job because the employer has gone bankrupt would be judged totally independent outside of personal control; personal health events are “nearly totally” independent, interactions with close others only “possibly independent,” and intentional acts as “nonindependent” [
Given this context, the inadequacy of checklist life-event questionnaires for etiological study of depression is apparent. Although quick and easy to administer, requiring few resources, they are subject to serious methodological limitations compromising the quality of the data gathered.
Digital health interventions are increasingly seen as a way to assess, treat, and prevent psychological disorder and deliver mental health provision. Such Web-based assessments and services have the ability to overcome geographical barriers, lower delivery costs, and reduce workforce demands [
The current project in progress aims to address the need for improved and accessible life stress measurement by developing an online data capture tool (CLEAR) and testing its psychometric properties and its association with disorder and educational outcomes. Currently, the project is in its early stages and CLEAR is still under development. In the following sections, we outline the basic architecture of the CLEAR system and the study to test it once complete.
CLEAR is a new complex measurement tool, and therefore, its feasibility and usability will be assessed by life-event expert and nonexpert volunteers (n=20) across a range of ages. These groups will act as a panel to test out CLEAR before it is finalized. Panelists will rate either their own experience, or case study examples from archived interview data, to determine both user friendliness and whether the full context of the event can be adequately captured. Their feedback will inform improvements to the system.
The project will utilize 3 different samples to develop and test CLEAR. A midlife sample (average age 52) will be recruited from the Depression Case Control (DeCC) study, involving a pool of 2299 respondents from London, Cardiff, and Birmingham, originally studied for gene-stress interaction and depression [
Participants will be approached by letter or email, which will explain the study and enclose an information sheet and consent form. Those who are interested in the study will be sent the CLEAR URL and log-on details, which will allow them to access and complete CLEAR from any Internet-enabled computer or tablet. The validity of CLEAR will be assessed by interviewing 30 participants from each of the samples (10 undergraduates, 10 recurrent depression cases, and 10 unaffected controls) using the in-person LEDS interview and CLEAR in counterbalanced order. The time taken for each participant to complete CLEAR will depend to some extent on how many life events have occurred over the 12-month period. However, the average in-person LEDS takes approximately an hour to complete, and therefore, it is assumed this will be the average time taken to finish CLEAR.
The data generated will be rated blind/reviewed by separate researchers and compared using Cohen kappa and intraclass correlation coefficients (ICCs) for level of agreement between the 2 methods. The total number of life events captured, the domain category, severity rating, and the timing of events will be compared to give an indication of how well CLEAR mimics the in-person method for full reporting and recall. Any further modifications will be made to CLEAR if required.
Test-retest reliability of CLEAR will be undertaken using an additional 20 undergraduates, 20 depression cases, and 20 controls from each of the samples, measured an optimal 3-4 weeks apart for stability in reporting (using Cohen kappa and ICC). CLEAR will be tested on the remaining participants (n=285) and the rates and types of life events and difficulties reported in the 3 samples will be compared and analyzed in relation to sex, social class, and age using chi-square statistics. The association between life events and past/recent depressive disorder and physical illness (DeCC sample), and academic performance (undergraduate sample) will be tested using logistic regression to look at the contribution of life events and indicators of social disadvantage to health and educational outcomes. Once both reliability and validity of CLEAR have been determined, the program will be available for more general use.
Security is a key concern of CLEAR; the CLEAR servers are built from CentOS Linux 5.4, which is a secure variant of Linux, has no services or ports installed, and includes only what is strictly necessary for CLEAR. In addition, a firewall is installed to further restrict access to the server. All data are entered into CLEAR under a unique log-on, and therefore, no names or contact details are entered on to the CLEAR system. The data are stored on a secure MySQL database that is updated whenever a participant enters information through the CLEAR interface. The log-ons will be stored in a password-protected file with the participant’s study ID numbers. A separate password-protected file will contain the ID numbers and any identifying respondent information (eg, contact details). Therefore, for this study there is the ability to recontact participants if needed.
The CLEAR study is in the preliminary stages and its results are expected to be made available by the end of 2015.
Respondents complete CLEAR by providing demographic information; information about close others; and life events and difficulties over the past year in 12 domains (education, work, reproduction, housing, money, crime, health, romantic, other relationships, children, death, and miscellaneous). The assessment also includes a fixed battery of measures, a depression questionnaire (General Health Questionnaire, GHQ, [
The logic-driven menus guide the respondent based on their prior answers. For example, if a respondent chooses the “education” category, this presents them with a menu of options (eg, selection interview, examination results) and depending on response, a following different set of options is provided (eg, application rejected/accepted versus passed/failed exam). Once respondents have completed these, they describe and score various aspects of the event including the event context, independence, and their feelings about the event. For each event, respondents are asked if this relates to any other event or difficulty and a menu (which continually updates) is presented with all previously entered events and difficulties. Thus, they are also able to link events and difficulties. This creates a dynamic feedback system in which more links between events and difficulties can be added as CLEAR is completed. Throughout this process, there are detailed instructions (including video) and domain-specific examples to inform the respondent. Important aspects, such as level of threat/unpleasantness, are given labels benchmarking the target level to encourage appropriate ratings.
CLEAR is scored using a precoded algorithm to produce a rating of “severe” life event as well as “D-matching” events and other indices. For analysis of the in-person LEDS interview, a derived variable of “severe life event” is one, which is rated (1) “marked” or “moderate” on long-term contextual threat/unpleasantness (ie, objective assessment, present at 10 days after the start); (2) “self” or “joint” focused; and (3) is not “illness related” (ie, part of the disorder investigated such as treatment/hospitalization or symptom related such as suicide attempt). The same algorithm for combining these 3 scales in producing a binary severe life-event variable will be precoded in the CLEAR online version from the data entered and made available for the report produced, or downloading to SPSS (SPSS Inc., Chicago, IL, USA) for further analysis. For “matching difficulty events,” a stated link to rated difficulties of “very marked,” “marked,” or “high moderate” severity is required, in the same domain (eg, work or marital), and of 6 months duration prior to the event. This will similarly be precoded consistent with the regular interview analysis of data.
Information can be pooled from various sources to assign the likely negative meaning of the event for the respondent based on demographics in combination with objective ratings of the event circumstances. The logic-driven menus provide detail about the basic event type and circumstances that may apply (eg, for moving house a submenu is provided where an individual can choose an option “forced to move” and from the following menu tick options that may apply such as “large cost of moving” and “neighborhood less desirable”), and the self-report data provide demographic information including current circumstances (eg, employment status, number of dependents) and historical data (eg, education and employment history). The system also requires self-assessed threat/unpleasantness ratings of events and difficulties. Together, these will be used to produce an overall objective severity rating. In addition, the written descriptions provide further surrounding detail that can be reviewed by researchers to check for reliability. Furthermore, using both the open-ended text-box answers and scores can help researcher review each case in depth, which also allows for quality control checking and enabling extended qualitative analysis if needed, or in a minority of cases recontacting respondents.
The logic-driven menus guide individuals toward the type of events likely to be stressful, from more general to more specific event types. There is evidence to suggest that inclusion of detailed instructions of different event types in each category gives better test-retest reliability with less “fall off” of event reporting over time, and greater agreement between respondent and co-informant [
Each stem menu of events leads down a path until the options are no longer relevant. However, at each stage, the respondent is given the option of rating “something else/other.” In this way, stressful events that do not fit into proscribed categories or criteria can also be included. This ensures that the specification of events does not make the definition of events too narrow [
Difficulties with recall can be a problem for both interview and checklist methodology, even over a 12-month period [
Recall is also helped through a personalized calendar that is updated as life events and anchoring anniversaries or social occasions (eg, holidays, birthdays) are added to the system. The timing of important psychopathology-related timings (eg, peak depression) can be added to the calendar. Events are often linked to other events in autobiographical memory; therefore, the use of calendars can lead to better quality (ie, more complete and accurate) retrospective reports of events, even after several years [
The CLEAR system aims to be as personalized as possible. In general, simple approaches such as addressing individuals by their name can sufficiently personalize a message to heighten attention to the information provided [
Lastly, the online system can be completed in private. Compared with interviews, self-administered measures can elicit more events that may be sensitive, embarrassing, or have the potential to bring about negative consequences [
The CLEAR system will also be programmed to provide basic reports for clinicians/health professionals on individuals in health settings with appropriate permissions. Health professionals can be provided with unique log-ons to CLEAR to access the data-generated reports from the database. The reports will provide a summary of each life event (severity score, date, classification from the menus, and written context from the respondent), a calendar denoting sequence and timing, and the scores from the GHQ and VASQ with appropriate description of resulting classification. The data from CLEAR can also be downloaded into SPSS files or specific data can be downloaded based on applied filters (eg, all events in the housing category). The data are a mixture of quantitative variables (eg, event category, threat, age, relationship to person close to them) and qualitative variables (eg, event written description and emotional reaction). The provision of such automated reports, once tested for their informative and useable characteristics, will be a major benefit of the measure to ongoing practice.
The first observations concerning life events occurred in the early 20th century in the clinical field when understanding the experience of depressed patients [
Severe life events are relatively common, but only a minority of individuals exposed develop depression. Therefore, the role of personal vulnerability is important. Studies including both low self-esteem and negative interpersonal relationships (ie, conflict with partner or child or lack of close support) as vulnerability indices showed interaction effects with stressful life events in the development of depression [
This paper argues that life events are complex phenomena not currently served by the most commonly used measurement approach, that is, checklists. This is potentially damaging research investigating the etiology of depression: problematic measurement must surely lead to problematic results. It is hoped that CLEAR’s technological advances will produce a useful compromise between life-event checklists and interview approaches, overcoming some of the limitations of questionnaires while reducing the burden inherent in face-to-face interviews. CLEAR should have the capacity to capture life-event details and context, different attributes of the event, timing of the event, and linkages between events and difficulties. Although it will not provide exhaustive coverage of all possible events, it is presumed that the majority of events will be captured and that most respondents will be able to rate the bulk of their events accurately given the guidance provided through the menus, examples, and appropriate benchmarking. Thus, this new method of measuring life events may be able to gather high-quality data, hopefully with reliability and validity comparable to the gold-standard interview approach, overcoming the problems inherent in relying on checklist approaches in etiological research.
It is also hoped that the CLEAR approach to assessing life events and difficulties will aid those in clinical practice. The provision of recent life charts of events labeled in terms of their likely stressful nature and with attributes relating to loss, danger, humiliation, and entrapment will allow clinicians to consider the level of stressor experienced in seeking to estimate patient appraisal and coping capacity.
Computerised Life Events Assessment Record
Depression Case Control
General Health Questionnaire
gene-environment interactions
intraclass correlation coefficients
Life Events and Difficulties Schedule
Vulnerable Attachment Style Questionnaire
The CLEAR “Stress online: Developing a reliable and valid interactive online method for measuring stressful life events and difficulties” project is funded by the Economic and Social Research Council (ESRC ES/K00638X/1).
Dr Taylor is funded in part by the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London, which played no part in the preparation of this manuscript. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health. Dr Taylor has received personal fees from Bristol-Myers Squibb and Otsuka and has a family member who is an employee of GlaxoSmithKline.