Evidence-based assessment/Step 7: Add more intensive methods to finalize diagnoses and case formulation
Medical disclaimer: This page is for educational and informational purposes only and may not be construed as medical advice. The information is not intended to replace medical advice offered by physicians. Please refer to the full text of the Wikiversity medical disclaimer.
|Steps 1-2: Preparation phase|
|Steps 3-5: Prediction phase|
|Steps 6-9: Prescription phase|
|Steps 10-12: Process/progress/outcome phase|
Prescription Phase: More Intensive Methods to Finalize Diagnoses and Case Formulations[edit | edit source]
Overview[edit | edit source]
Updating Probabilities[edit | edit source]
One way of deciding diagnoses and treatment targets would be to add more information to push some probabilities up into the Treatment Zone. This is what several projects with IBM's Watson are doing (cf. IBM Watson), along with other applications of machine learning.
Structured and Semi-Structured Diagnostic Interviews[edit | edit source]
Comparing the typical clinical to the more structured interview formats reveals several patterns. Agreement about some diagnoses is modest overall, and significantly worse about many. Some of the gaps are likely due to fiscal considerations – many payors will not reimburse for psychosocial treatment for conduct disorder, creating an incentive to diagnose the same behaviors as attributable to something reimbursable. Others are due to gaps or variations in training, such as the wide discrepancies in diagnosing pediatric bipolar disorder or autism. Clinicians also tend to diagnose fewer conditions than a more structured interview detects . Sometimes more structured interviews detect cases that do not have sufficient impairment to warrant a clinical concern; but at least as important is the pattern of clinicians calling off the search as soon as they have confirmed a (reimbursable) diagnosis. This leads them to underestimate rates of comorbidity compared to SDIs, often missing additional treatment targets or factors that might moderate treatment selection. Reviews of notes and interviews also find that clinicians often assign a diagnosis prematurely, without having documented enough symptoms to meet formal criteria for a diagnosis. Adding SDIs improves diagnostic reliability across a gamut of clinical settings ranging from pediatric community mental health settings, emergency rooms  and inpatient units, improves psychosocial outcomes, and provides equally good symptom reduction with less clinician time – allowing more patients to be seen, or less expense per patient. When the clinician’s diagnosis of record happens to be confirmed by an SDI, families engaged more with treatment, had fewer no-shows or cancellations, and had significantly better improvements in internalizing outcomes, indirectly suggesting that more accurate diagnosis leads to better process and outcome.
Despite these advantages, clinicians rarely choose to use semi-structured interviews. They see them as infringing on their professional autonomy and potentially damaging rapport with clients. Ironically, surveys with clients find that they actually prefer semi-structured interviews, believing that that evaluation is more thorough and the clinician gains a more comprehensive understanding of the client’s needs and situation . Recent surveys also indicate that cost and practical barriers may be the main stumbling blocks as attitudes become less of an issue. Development and validation of free SDIs – such as the Child and Adolescent Psychiatric Assessment and the Development and Well-Being Assessment (DAWBA) -- directly address the cost obstacle , as do the growing number of free but validated instruments for other niches. Why not just use the semi-structured interview as the basis for diagnosis, skipping the preamble of the rating scales and risk factors? First, no single interview covers all of the diagnoses and clinical topics that a clinician might encounter, and surprisingly, some of the omissions can be common issues. Older interviews did not ask about ADHD in adults, nor about mania or hypomania in youths. Many versions of the Kiddie Schedule for Affective Disorders and Schizophrenia (KSADS) lack modules for pervasive developmental disorders (e.g., ). The training requirements, and the issue of rater drift, also create practical obstacles to implementation. Finally, a questionnaire or computer may be more likely to get positive responses about sensitive issues difficult for parents or patients to bring up in person. It may be easier for our client to first broach her drinking, her thoughts of self harm, or her confusion about her sexuality in the privacy of a questionnaire or computer session than face-to-face. The discomfort or ease of addressing sensitive topics is likely to vary across cultures, as well, with degree of stigma as a key factor.
A hybrid approach, using rating scales and risk factors to generate the contending hypotheses, can guide the selection of which interview modules to use for follow-up. It becomes possible to “take the best,” picking sections of interviews that are best for each particular purpose, using the KSADS modules for mood disorder , the ADIS for anxiety, and so forth.
The DAWBA already is integrating responses on checklists (specifically the SDQ) as part of the formal diagnostic process, combined with structured interviews and clinical judgment. Intriguingly, the DAWBA offers the possibility of subcontracting to get the clinical judgment component, noting that it may be more cost-effective to pay a small fee for outside expertise rather than trying to train and support personnel for that role in-house (http://www.dawba.info/d0.html, accessed January 19, 2016).
Cognitive and Achievement Testing.
Cognitive and achievement testing may also play an important role at this stage for some cases. If the referral question focuses on academic performance, or if the contending hypotheses include a learning disorder, then cognitive and achievement testing clearly provide valuable data. Sometimes the referral question focuses on possible ADHD or learning disability, and cognitive testing may reveal that the person has average or low average ability and has been functioning in a fast paced or challenging environment by dint of hard work rather than having some other disorder. Similarly, autism and developmental disorder evaluations need to gauge whether social deficits are out of line with verbal and nonverbal cognitive ability. Brief, four-subtest ability tests would be well-suited to recognizing these scenarios, and verbal ability estimates may also help gauge the suitability of more cognitive components in therapy versus emphasizing behavioral interventions. In contrast, despite decades of speculation and clinical lore about subtest scatter or profile analysis being associated with clinical conditions such as ADHD, the EBA framework makes clear that the associations are too weak to be informative for individual decision-making. Deficits in processing speed or working memory are non-specific, meaning that many different clinical conditions can show them, and the pattern does not help to pick between the hypotheses. Ipsative analysis has technical psychometric challenges making it unlikely that apparent strengths and weaknesses will be stable over time or show incremental predictive value. The validity coefficients for checklists are significantly higher for predicting diagnoses such as ADHD or autism than those found with cognitive profile analysis – and checklists can usually be administered and scored much more quickly and at much lower cost. Even those that would benefit from assessment of cognitive ability usually will not need the traditional full battery; in most cases, a four subtest estimate of general ability will provide the most robust and empirically supported information. Specific questions about neuropsychological deficits also would be better addressed by tailored, high-validity batteries, rather than using omnibus cognitive ability tests that attempt to provide neuropsychological Swiss Army knife tools that are not adequate for high stakes use.
Rationale[edit | edit source]
Steps to put into practice[edit | edit source]
Structured Diagnostic Interviews[edit | edit source]
Semi-Structured Diagnostic Interviews[edit | edit source]
Semi-structured interviews require more training to use consistently, and periodic re-calibration to avoid drift. Large grants can support this type of infrastructure, but it is challenging to implement in clinical practice.
- Add links to KSADS Wikipedia Page
- SCID (does it have a Wikipedia Page?)
Tables and figures[edit | edit source]
References[edit | edit source]