Context Matters: Gender and Depression Datasets

In the global mental health gender data challenge, teams will collect data on gender and depression within diverse cultural contexts.

We are in the process of understanding what data to prioritize. We recognize teams must gain sufficient ethnographic understanding of where they are working. We also know teams should collect prevalence and other descriptive epidemiological data.** But we are also interested in capturing contextual variables beyond basic demographics that offer insight into why people are suffering.

    What contextual data would you prioritize? How would you collect that data (indicators, assessments, or technology)?
  • What is your dream dataset?

Hi @Suneetharani, @stephaniel, @farah - Please share your thoughts on what contextual data is essential to understand the reason people are suffering from depression.

@ingmarweber, @KarenBett, @ukarvind - Please share your thoughts on what technology / assessments / indicators can be used to capture contextual data of people suffering from depression.

First, I think we are getting off on the wrong foot by focusing on depression. Depression is just one form of mental illness, albeit a common one. Significantly, it is not the only one that might interfere in a person’s ability to function in a work setting. I do not claim it is common but a very interesting problem is the “impostor syndrome.” People who are suffering from this problem might be spectacularly successful in their work–and then suddenly quit. That is a costly loss to the employer–perhaps more than the under performing individual dealing with depression and some of its effects including alcoholism and drug abuse.

Re: what daata to collect and how. Have you considered a series of focus group studies across societies? Focus groups are small groups of people with well developed opinions on a topic usually because they have steeped themselves in the available literature etc. That is not always the case so care must be taken in forming them. But focus groups are a good way to start on what data to collect and how in various societies for whom our mental health data is scant.

Thanks @boblf029 for sharing your thoughts. We have been using focus groups for our research.

@sarahb, @DrLiliaGiugni, @Aria, @Fatima, @WD_Research - You may have inputs on which contextual data is important to learn the reason of people’s mental illness. Join the discussion to share your thoughts.

Hi @shihei, @AnnalijnUBC, @Pavel, @Tapman, @mfree - We would love to hear your thoughts on Aaron’s questions. Thanks.

I see depression as a combination of biological vulnerability and environmental factors that increase risk. Biological vulnerability can be thought of as genetics but also the changes in the nervous system and the HPA axis that happens when there’s trauma/childhood abuse…

Thanks for the discussion! I agree with @boblf029 in terms of scope of focus. I am not at all a mental health expert, but it might be good to to focus on the areas of mental health where there is a difference in gendered experiences of the issue and treatment. I think anxiety could be a good candidate.

Apart from that, here are some contextual information that I think would be important to capture:

  • Family and carer duties - relationship status, children, aging parents
  • Financial independence and contribution to decision making -
  • Experiences of sexual assault / gendered violence
  • Feelings of freedom / coercion and control
  • Family / cultural expectations

Stacy o has some excellent suggestions. But I think she or whoever wants to work with her concepts will need to spend some time defining the terms such as “gendered violence” operationally keeping in mind she will not be the one collecting the data…

Thanks @staceyo and @boblf029 for sharing your perspective on important contextual information.

Hi @Tsion, @YaelNevo, @panderekha, @pepsicola28, @rana, @Andrea - What do you think about Aaron’s questions and the comments shared so far. share your thoughts.

Hello @Kalpana, @asneves, @aakanksha_k, @munnatic, @shruti, @niki, as you all have a background in GBV, what do you think about Stacey’s comment above. We would love to see few more contextual data added to the list Stacey shared. Join the discussion to share your thoughts.

There is a need to better explore electronic health records and the notes that physicians/clinicians enter into charts because this data is rarely reported/included in research. Much of the mental health data/outcomes are couched inside of text fields. There are numerous reasons why mental health ends up in the notes rather than a clean diagnosis. AI and ML techniques can solve this data issue and give us clearer insight into the real extent of the burden of mental health among women who seek health care.

Agreed - depression is too narrow. Depression overlaps with anxiety and overlaps with suicidal ideation. And would need to define if depression refers to unipolar or bipolar. Suggest sticking with mental health problems - broad and captures the spectrum of disorders.

@ktabb - Welcome Karen! We’re excited to have you join our online community of experts.
Thanks for sharing your thoughts on this important discussion topic.

We agree Electronic Health Records (EHR) are great resources. Is it possible for you to name few EHRs, which we could dive into to further our research.

Also, If say we take into consideration the overall mental health problems, what contextual data should we collect to understand the reason of one’s suffering.

Hello @sadiew, @DNAtimes, @AlexandraW, @suzannewertheim and @luisbenveniste - What do you think about Stacey’s comment. Please join the discussion to share your thoughts.

Hello Shashi,

  1. The contextual data at the larger/collective level as well as at the personal level has to be taken into consideration. A vulnerable section that is subjected to constant threats and violence might suffer from repression. The collective context could be, apart from gender, community, class, region, age, occupation, ideology etc. While the personal context is affected by all the above, individual contexts such as status of/in the family, support systems, lifestyle, influences, hierarchy of credibility, image of one’s conduct, disability, relationships, pastime, desirability, acceptance have to be taken into consideration.
  2. Self-perception could in fact give us a good idea about the complexes people might be suffering from. It is common knowledge that self-perception is shaped by the expectations of the world which can subject individuals to severe stress and trauma leading to psychological issues.
  3. Technology can be used to find out the status of thoughts of individuals and groups by accessing their comments, of course the openly made comments meant for the consumption of a large number of audience, in social media.

Thanks @Suneetharani for sharing your thoughts on this important discussion. All strong points

@Aaron_Denham - What do you think about Suneetha’s comment.

In addition to the above, we will need to understand the health system within which services might be provided - but in LMIC often seldom accessed by women. Is it safe for women to attend. Must they be accompanied by men. Is it feasible for women to travel to them and return within a day. Can they afford the service (what is the cost, do they have access to any money etc).

A lot of LMIC services are ‘turn up and wait’. Hundreds of people are seen each day with a very minimal consult.

So there is a need to understand local/village access, community trained health workers and how they are used, how they are trained. That is, what is the quality of the services available.

I think depression is a good starting point. It is comorbid with so many other conditions it is a good way in.

  • understand the available health service(s)
  • how far away are they
  • what do they cost
  • what treatments are available (drugs/talking therapies) and what do they cost
  • stigma - we will need to understand how that plays out
  • other terms/idioms used instead of depression or anxiety etc so they we can find what we think we are looking for.

Hi @Shashi and @Aaron_Denham - apologies for the delay. Some things that I think provide important context are:

-> Inciting incident, if known (e.g., when did the mental health problem start? Was it preceded by a stressful external situation, or did it just start with no inciting incident?). If there’s a clear inciting incident/event or series of events, then symptoms may not necessarily represent a mental health disorder but instead a normal physiological/neurological response to an extreme or untenable situation.

-> Has the mental health problem occurred before; if so, how many times or how often (or all the time)?

-> If the mental health problem has occurred before, was there an inciting situation then or not (and if so, was it similar to or different than any inciting situation this time around)?

Possible inciting situations: food insecurity, warfare or societal upheaval, disease outbreak, death of family member, financial catastrophe, abuse, etc. Presence of these situations may indicate a mental health issue is a response to a point-in-time situation and may resolve when the situation resolves (i.e., may not be a persistent disorder).

If the inciting situation was long ago and is now resolved or has subsided but the mental health problem persists, that is more likely a disorder.

If there was no inciting situation identified (e.g., “I’ve always been like this” or “it just happened’), or if the magnitude of the mental health issue is far out of proportion to the magnitude of the inciting situation, that is more likely a disorder.

It’s probably important to note that an inciting situation can be long-standing and severe/systemic/structural, such as armed conflict or abuse that goes on for years or decades.

I hope this helps.

Thanks @stephaniel for sharing your thoughts on contextual data. All strong points. Is there way to collect details of inciting incident using technology. If not, how best we can collect this data.

Hi @Shashi - this could involve a combination of:

-> Gathering stories from patients about when mental health struggles began and subsided (if applicable) via mobile phone surveys and health provider interviews;
-> Objectively correlating those dates against provider health records; and
-> Also correlating those dates against known local and global events (in the case of wars, food shortages, disease outbreaks, societal upheaval, and so on) through data matching with news archives that should have ample details.

If there are patterns of mental health struggles correlating with inciting incidents, an interesting question would be what percentage or portion of those mental health problems subsided (or not) after the incidents resolved/subsided. The portion that did not subside probably includes three components: 1.) Some may have been triggered by the inciting incident and then become chronic; 2.) Some may have arisen by chance during the incident; 3.) Some smaller percentage may have been present before the incident occurred and misremembered in the telling (those patient stories might not correlate with provider health records, which could indicate they fall into this category).