JCCAP FDF/2017/Bipolar disorder

From Wikiversity
Jump to navigation Jump to search

This is the landing page created at the First JCCAP Future Directions Forum to help organize information about publicly available data sets as well as some suggestions for best practices in designing and reporting research looking at these types of variables. There were four keynote addresses: Dr. Eric Youngstrom discussing future directions in assessment, Dr. Matthew Nock discussing suicidal and self injurious behavior, Dr. Mary Fristad discussing bipolar disorder, and Dr. Daniel Shaw discussing trajectories and treatment for conduct problems. Each of these was the focus for 2-3 smaller breakout discussion sessions led by content experts. There are a set of four pages that gather the ideas and resources related to these sessions.

Dr. Mary Fristad[edit]

Dr. Anna Van Meter: Assessment[edit]

Introductions[edit]

Current state of assessment tools in bipolar disorder[edit]

  • There are many screening tools for mania/bipolar disorder, but they are not all well-known and clinicians may not feel confident selecting a tool and/or administering it
    • Youngstrom et al. (2015)[1] multivariate meta-analysis of discriminative validity of caregiver, youth, and teacher rating scales of PBD (free full text) is a good step toward better understanding the available tools, so that an efficient measure can be selected
      • The results show that caregiver report produces significantly larger effects than youth or teacher report.\
      • Three top performing measures are public domain and freely available on Wikipedia
  • Screening tools are helpful for narrowing clinical hypotheses, but are not appropriate for diagnosis. Structured (or semi-structured) interviews are important in order to improve the accuracy of diagnosis
    • Two key obstacles to the accurate diagnosis of bipolar disorder are the episodic nature of the illness and high rates of comorbidity. [Semi]Structured interviews are more likely to pick-up on both
  • Once the clinical data (screening measures, interview, family history, etc) are collected, clinicians tend to combine this information in their heads to arrive at a diagnosis. Research has consistently shown that this method results in less accurate diagnostic decisions, relative to using actuarial methods to combine data
    • The nomogram is a quick, and easy, way to combine clinical data and arrive at the probability that an individual has bipolar disorder (or other diagnosis of interest)

Opportunities for assessment research[edit]

Secondary data analysis[edit]

Research can be conducted using existing data. This includes both meta analysis and the use of existing datasets. Secondary data analysis is a good way for trainees to work on papers likely to have higher impact (relative to data they would collect on their own). Some data are publicly available, other times, the principal investigator must be contacted. This is best done after reading extensively about the study you are interested in, and after drafting a brief proposal, including hypotheses and a timeline for completion of the project. Although contacting a PI can be intimidating, there are benefits including: networking opportunities, better probability of a high impact paper, and the chance to learn from a more advanced scientist. If possible, include the PI on the paper, so you have access to her/his expertise, data help, and higher profile reputation.

  • Potential topics for secondary data analysis related to assessment include:
    • Evaluating existing measurement tools using ROC analysis and investigating potential moderators of scale performance:
      • Informant
      • Age, sex
      • Other moderators?
    • Comparing phenomenology / other characteristics across diagnostic categories to help validate a diagnosis or demonstrate ways in which it is similar/dissimilar to other disorders
    • Testing how sensitive measures are to the effects of treatment
    • Using item response theory to develop brief measures or to investigate how certain questions might vary across demographic groups (DIF)

New data collection[edit]

Developing and executing a new study is exciting, but finding participants is challenging. If possible, try to capitalize on existing cohorts - investigators are often open to adding a measure or paradigm to their existing study, which can save a lot of time and money (and initiate a new collaboration!).

  • Clinics
    • If there isn't an existing study that can facilitate your data collection, seek out specialty clinics where you are likely to find more people who meet your eligibility requirements. Keep in mind that clinicians are busy and your study will not be their priority, so you need to make it as easy as possible for them to help you. Be prepared to spend time at the clinic recruiting and running participants. T
  • Medical records
    • University hospitals often have electronic medical records that can be queried to aid research. With IRB approval, you may be able to identify hospital patients who meet your criteria and contact them to invite them to participate. This is a good way of doing targeted recruitment, which can save a lot of time.
  • Community
  • If you don't have access to a clinical sample, remember that RDoC encourages a transdiagnostic approach, and think in terms of spectrum. Participants from a community or university sample are likely to represent a range of characteristics from healthy to pathological, as long as you assess and document this, you can study phenomena related to psychopathology in a more generalizable sample.
  • Online sampling
    • Participants can be recruited online for both in-person and digital studies. Posting online requires both IRB approval and, often, approval from the site where you are posting. Craigslist, Facebook, and reddit can all be helpful for finding people, though you should plan to get a low number of successful participants relative to the number of people you will communicate with.
  • Amazon's Mechanical Turk is also a helpful tool that can be used both to recruit participants and to conduct your research (if based online). Once you learn the interface, it can be a great tool, but is not always intuitive, and you should plan to pilot test to work out the glitches before posting all of your HITS.
    • Read Chandler et al. (2016) article [2] on doing clinical research with MTurk samples
    • create a Wiki for sharing about MTurk tips.

Technology[edit]

  • New technologies (EMA, apps, actigraphy) present an opportunity to capture symptoms more accurately
  • Though exciting, little is known about how these new methods of data collection compare to more traditional measures
    • An important research opportunity is validating technology against traditional assessment to establish validity
  • Big data from sources like medical records and intensive sampling of participant mood/energy/etc... have created opportunities for machine learning and other advanced statistical methods to identify patterns of behavior (or health characteristics, etc...)

Dr. Robin Nusslock: Neuroscience[edit]

Hirschfeld, et al (2003). Journal of Clinical Psychiatry shows that 60% of those with Bipolar Disorder have been misdiagnosed at some point with unipolar depression. There is a 10 year standard deviation for the correction of this unipolar depression diagnosis. Why is this dangerous? Because there is evidence that SSRIs may actually trigger manic episodes in some people with bipolar disorder. Nevertheless, prolonged misdiagnosis does not bode well for the patient.

Distinct reward signaling in mood disorders[edit]

There is evidence that, especially in the context with unipolar depression, there is reduced activity in the reward-related brain area of the brain. However, this same brain anomaly is not seen in those who are just experiencing general stress-related depression. On the other side, a manic brain shows an over activation in this area. Risk for major depression vs bipolar disorder characterized by extreme and opposite profiles of reward signaling (Alloy, Nusslock, & Boland, 2016; Nusslock & Alloy, in press; Nusslock, Walden, & Harmon-Jones, 2015; Nusslock et. al, 2014).

But, can we use this data as a predictor of bipolar before they develop bipolar symptoms? Alloy, Young, Damme,...Nusslock, (in prep) suggests that, yes, we can! In this study they find that those with the elevated activation in these reward centers in the brain have a higher risk of developing bipolar disorder in the future. Why? Those with bipolar disorder engage the cortex in the way that amplifies (over amplifies) the interaction between the cortex and subcortical cortex of rewards. Essentially, the brain is on reward overdrive instead of reward cruise-break.

Brain stress test (discussion)[edit]

Is there a way to stress the brain in a way that can show us if there are those abnormalities that indicate higher risk for mania and depression? In our conversation, we looked at the promises, pitfalls, and challenges of using biomarkers to facilitate diagnosis?

  • Something that needs to be better validated are the psychometric properties of this kind of test before we can start making claims and justifying the cost of the scan.
  • What are the brain scan differences of adult bipolar patients and pediatric bipolar patients? And if we are going to use brain stress tests to see if the brain exhibits the abnormalities, what are the developmentally appropriate stresses?
  • What happens to those who have those activated areas in the brain, but don't want to be treated for the mania, because they don't find it impairing? What is the cost-benefit analysis for them?
  • Similarly, what about those that have this elevated activity, but haven't been diagnosed with bipolar disorder and haven't experienced a manic episode yet? It is similar to the dilemma of whether it is better or worse for someone to find out that they have a gene that increases their risk of breast cancer, but they don't have breast cancer yet. The discussion continued with early intervention and the benefits and cons of early interventions.
  • There is also the issue that, using brain scans as diagnostic tools in general, the "normal" brain may not be representative of the actual "normal" brain. Example: say a person has a slightly higher than normal thyroid activity, but their thyroid activity has always been at that level, unchanging, and has not presented any problems. So this "higher than normal" activity, is actually "normal" activity for this person.
  • Is it feasible to have a team of neuroscientists and statisticians at every clinic to review these brain scans and come up with diagnostic likelihoods?
  • There are stigma challenges, challenges linking brain regions to brain systems (and brain systems to physical behaviors), and phenomenology does not map onto biology.
  • Take home point, regardless, is that we are not at the point that we can do this. It is difficult to subject children to these scans and get reliable results since adolescents, in general, have more activated brains that may be normal for them, but "manic" for adults. Additionally, we would want better sensitivity and specificity, considering the extremely negative outcomes in misdiagnosing bipolar disorder.

Dr. Nusslock: ideally, some day, a child comes into the clinic with concerns of depression or bipolar. After getting the family history and psychiatric history, it becomes aware that there is a family history of bipolar disorder. After gathering this data, a cost-benefit analysis is run to see if it is appropriate to get a brain stress test scan done, keeping in mind that this is a tool and not a diagnosis.

Training the future (discussion)[edit]

How are we supposed to master all the domains of clinical neuroscience while still being a student with all the responsibilities of a student?

  • You cannot be an expert in everything. You need a that will support you. Collaborations and teamwork is absolutely critical, not only as a student, but also as a faculty.
  • Your research needs to be a priority, and you need to follow the field and your research frequently. This is a fast paced environment with new discoveries and new changes being made every day.
  • You choose one, and you stick with it. Don't be a jack of all trades (doing coursework, doing clinical training, taking on learning multiple new techniques at the same time). It may not be fun for a little while, but you will have a strong skill set that will make you an expert in a particular area that another team may need. You go to that team with your skill, and there you can start learning a different skill. So sequential learning is key.
  • Post-docs: start thinking about it 2 years before graduating. Look into training grants. Look to work on a faculty member's RO1 or PO1 so you can be working on a team, getting publications, gaining skills, etc for multiple years... Write an NRSA (which requires a two year maturation period anyways, so look for a lab to work for multiple years in)
  • How do you find that faculty member to mentor with? Go to conferences, make connections, and tell your story. Doing this takes time, which is why you want to start 2 years before graduating.

Dr. Amy West: Treatment[edit]

<add notes here>

References[edit]

  1. 10.1037/arc0000024
  2. 10.1146/annurev-clinpsy-021815-093623