Algorithms in Psychological Health Care

On the one hand, I see the value of having general guidelines for health care providers to follow.  But on the other hand, I see that there are more fundamental principles, such as establishing trusting relationships, practicing listening and interviewing skills, etc., that are far more important as standards of care, than following some kind of mechanical algorithm.   Most of the so-called "algorithmic" elements in managing psychological distress are things that most any clinician or therapist would have studied extensively in their long years of schooling, hence it is potentially quite redundant (and wasteful of time) to dwell at length on the preparation of such standards.    But I do think there are many ways in which care standards could be improved in a caring and collaborative community.  For example, I think that regular multidisciplinary "rounds"-style meetings to jointly discuss ways to manage particular problems, could be a fruitful, meaningful,  immediately useful, intellectually stimulating and robust process.  
 

Preamble

Algorithms of care can improve the efficiency of treating disease in a population, particularly when resources are limited, and when individual practices may have idiosyncratic variation.   Good examples of care algorithms which can lead to vast reductions of illness and death, and vast reductions in cost (both in terms of money and of effort), include those for treating cholera or for treating insulin-dependent diabetes.    For cholera, a simple standardized pathway of giving fluid and electrolyte replacement can be readily learned by all caregivers, and can very simply prevent death by dehydration.  For diabetes, standardized glucose monitoring regimes with basic guidance for insulin type and dosing can similarly be learned by all members of a care team (including the patients), leading to great improvements in safety, reductions in diabetes-related medical emergencies, and improvements in long-term morbidity. 

In managing mental illnesses, it can be valuable to consider a similar style of care algorithms. 

Patient Preference

In many cases, a given person may wish to have a certain type of care for a psychiatric problem.  Many patients simply want to talk to someone regularly, and do not necessarily wish to do CBT exercises.   Some patients strongly desire a medication therapy.  Other patients are strongly opposed to having a medication therapy. 

We cannot push patients into a care algorithm which is too rigid to account for patient preferences.  It is, however,  fair to introduce all patients to the various options available. In most cases, different varieties of care (such as different styles of psychotherapy, different specific medications, etc.) have far fewer differences in effectiveness than one might expect.  There are certain generalities for almost all psychiatric syndromes, however:  while all types of psychotherapy are helpful, there is good evidence that ideas from CBT should be encouraged irrespective of the style.  “Formal” CBT is not necessarily superior to “informal CBT,” particularly if a particular patient does not actually wish to have “formal CBT” but rather simply wants a supportive therapist to talk to, or perhaps a trial of psychodynamic therapy.  In practice today, most therapists use eclectic styles, such as a psychodynamically-informed variation of CBT, etc.  

Therapist Preference

Different individual therapists have different backgrounds, personality styles, areas of interest, and strengths.  Some particular therapists may excel in CBT-style therapies.  Other therapists may be experts in meditation.  Others may have a unique eclectic approach.  All of these individual therapist strengths and variations should be nurtured.  While it is good to have some unifying features of care, in the form of care algorithms,  it would be bad for the morale of the staff, and bad for patient care, for all therapists and physicians to have to conform to an identical pathway.  

Once again, patient preference may also guide which therapist would be most suitable; this fact should be respected deeply, especially for such an intimate matter as dealing with a mental health issue. 

Most of us, if were to start seeing a therapist, would want to choose the person we see, based on a variety of personal and professional factors.  

Especially in a university such as UBC which values the notion of diversity and personal autonomy, we should emphasize the ability for students seeing a mental health worker to choose the style of care that they would prefer, within the constraints of the system, as opposed to be sent on a rigidly observed care algorithm. 

 Comorbidities

Some of the most common clinical presentations in mental health care are of people who have so-called comorbidities.  These are people who meet criteria for more than one formal diagnostic category at the same time.  

Prevalence of comorbidity:  According to Brown et al (2001) a patient with an anxiety disorder diagnosis has a 57% chance of having additional DSM-IV Axis 1 comorbidities; a patient with a mood disorder diagnosis has an 81% chance of having additional DSM-IV Axis 1 comorbidities.  This figure does not even account for Axis II (personality), Axis III (physical health), or Axis IV (psychosocial) comorbidity. 

Barlow’s “Unified Protocol for the Transdiagnostic Treatment of Emotional Disorders” in an example of a therapeutic system which addresses comorbidities, by recognizing what Barlow considered an emotional syndrome which underlies many of the specific diagnostic manifestations.  In their words,

heterogeneity in the expression of emotional disorder symptoms (e.g.,individual differences in the prominence of social anxiety, panic attacks, anhedonia) is regarded as a trivial variation in the manifestation of a broader syndrome. (Farchione et al, 2012)

The example of Barlow’s system carries highly relevant practical wisdom, in terms of running an efficient mental health service:  it is possible to offer a quite similar treatment strategy  to individuals with a broad range of diagnoses and comorbidities. 

In many other cases, a given person may not wish to receive a diagnostic label at all, and a suggested treatment regime for a given diagnosis may be problematic.  Some people may find such labels and ensuing label-specific streams of care to be objectionable or even discriminatory. 

Therefore, given the issue of comorbidities and of clients’ reservations about labeling, it is important to view  “algorithms” with extreme flexibility and sensitivity, and perhaps consider not using them except as a very rough guideline. 

Readiness for Change

A therapeutic philosophy called  “motivational interviewing” is intended to address the fact that many people with the same diagnosis (such as an addiction, a mood disorder, or a relationship problem) may differ in their willingness to participate in a change process, whether this be psychotherapy, medication treatments, or even environmental change (e.g. dropping a course, seeking financial aid, etc.). 

All treatment algorithms must consider the differences between people in their degree of insight about their health concerns, and their willingness or readiness for change. 

It is highly counterproductive to prescribe a change strategy to someone who does not desire it.  And it is also highly counterproductive to simply send such a person away, if they do not choose to participate in a given program of action. 

Therapeutic Alliance

The goodness of the relationship between a patient or client and a caregiver (a therapist, physician, or other support) is strongly related to clinical improvements in all psychiatric conditions.  It is intuitively obvious that this so-called “therapeutic alliance” must be tended to as the highest priority in any care regime.  An algorithm of care must begin by developing a positive, trusting relationship between the patient or client and the caregiver, and the algorithm must not be applied in a mechanical manner which could harm the “therapeutic alliance.”  The research literature about this stretches back for decades.   Martin et al (2000) in a meta-analysis, show that therapeutic alliance is strongly related to outcome.   A more recent research example is Arnow et al (2013), who show that therapeutic alliance is strongly related to improvement in a group of chronically depressed adults; of note, this effect was particularly strong in a subgroup receiving a type of therapy called CBASP, which is similar to the varieties of therapy most commonly recommended in standard care algorithms in the past decade.  

However, it should be noted that problems with the therapeutic alliance are more likely if the severity of symptoms is higher.  In many cases, a factor which impacts care of any serious psychological problem is a difficulty establishing trusting relationships with a caregiver, regardless of the quality of care offered.  Therefore, we may see that therapeutic alliance is excellent in many cases, for particular cohorts, but this may simply be due to the clinical problems in this cohort being mild, rather than the care being somehow exemplary.  Conversely, a clinician dealing with severely symptomatic clients may have lower therapeutic alliance measures, but this could be due to the severity of the clients’ problems, not to problems in the quality or propriety of care.  

But another good recent research paper by DelRe et al (2012) shows that therapeutic alliance is more strongly determined by the therapist than by the client; here is a quote from their conclusion:

In summary, therapist variability in the alliance appears to be more important than patient variability for improved patient outcomes (as assessed with the PTR moderator). This relationship remained significant even when simultaneously controlling for several potential covariates of this relationship. These results suggest that some therapists develop stronger alliances with their patients (irrespective of diagnosis) and that these therapist's patients do better at the conclusion of therapy. (DelRe et al, 2012)

Other recent research shows that a poor therapeutic alliance can not only cause a regime of therapy to be ineffective, it can cause it to be actively harmful.   Goldsmith et al (2015) show that early psychosis patients can benefit from psychotherapy, but are harmed by attending therapeutic sessions with poor therapeutic alliance. 

Therefore, it is important in this “algorithmic” process to remember the massively important issue, which transcends all other issues of technical details, decision trees, etc.–of attending to the therapeutic alliance, by fostering compassionate, wise interpersonal skills in all counseling professionals, as the cornerstone of any algorithm. 

But how to do this?   There are many ideas, but in a collaborative model, it would be a good idea to focus on collaborative teaching and feedback between different clinicians who have varying degrees of experience and skill, as an important element of any care pathway.  

Does Conformity to a “manualized” standard improve clinical outcome? 

There are many so-called “manualized” therapy techniques.  These are designed as an attempt to standardize care, and are particularly useful in research, to determine and measure whether particular styles or techniques are actually better or worse than alternatives.

Yet, existing evidence does not support the notion that variations in therapeutic style strongly impact clinical outcome.  While it is wise for therapists to follow and learn new therapy ideas, such as CBT, the most important thing, once again, is for therapists to develop ways to optimize the therapeutic alliance, rather than focus on particular details from a manualized approach. 

This is also an evolving area of research, one example being Tschuschke et al (2015), who demonstrate that therapists’ adherence to a prescribed treatment regimen should be flexible, particularly for people who have more severe symptoms or problems.  According to the authors, such flexibility is more consistently present in more experienced therapists, and may reflect, in general, the degree of competence in the therapist. 

We can speculate that therapists might have to make sure that the therapeutic process can continue and that the relationship is improving or at least stabilizing on an acceptable level, so as to assure that the treatment can continue. This probably includes therapists easing their treatment protocol temporarily. Thus, treatment adherence in psychotherapy is not always a stable factor but instead depends on therapists’ level of professional experience, clients’ abilities to establish a good enough working alliance, and the climate of the therapeutic cooperation in the dyad, although it might, on average, remain on a relatively low level in most sessions. Nevertheless, the flexibility of therapists treatment adherence reactions seems to impact treatment outcomes substantially if clients’ severity of psychological problems hampers the working alliance. (Tschuschke et al, 2015)

Therefore, with respect to algorithms of care, it should be emphasized that flexibility must be called for in their interpretation, particularly for the many clinical situations in which there are complications or difficulties due to higher levels of severity, complexity, therapeutic alliance problems, or limitations due to low readiness for change. 

References

Brown, T. A., Campbell, L. A., Lehman, C. L., Grisham, J. R., & Mancill, R. B. (2001). Current and lifetime comorbidity of the DSM-IV anxiety and mood disorders in a large clinical sample. Journal of abnormal psychology, 110(4), 585.

Del Re, A. C., Flückiger, C., Horvath, A. O., Symonds, D., & Wampold, B. E. (2012). Therapist effects in the therapeutic alliance–outcome relationship: A restricted-maximum likelihood meta-analysis. Clinical Psychology Review, 32(7), 642-649.

Farchione, T. J., Fairholme, C. P., Ellard, K. K., Boisseau, C. L., Thompson-Hollands, J., Carl, J. R., ... & Barlow, D. H. (2012). Unified protocol for transdiagnostic treatment of emotional disorders: a randomized controlled trial. Behavior therapy, 43(3), 666-678

Goldsmith, L. P., Lewis, S. W., Dunn, G., & Bentall, R. P. (2015). Psychological treatments for early psychosis can be beneficial or harmful, depending on the therapeutic alliance: an instrumental variable analysis. Psychological medicine, 45(11), 2365-2373.

Martin, D. J., Garske, J. P., & Davis, M. K. (2000). Relation of the therapeutic alliance with outcome and other variables: a meta-analytic review. Journal of consulting and clinical psychology, 68(3), 438.

Tschuschke, V., Crameri, A., Koehler, M., Berglar, J., Muth, K., Staczan, P., ... & Koemeda-Lutz, M. (2015). The role of therapists' treatment adherence, professional experience, therapeutic alliance, and clients' severity of psychological problems: Prediction of treatment outcome in eight different psychotherapy approaches. Preliminary results of a naturalistic study. Psychotherapy Research, 25(4), 420-434.

Addendum (March 9, 2017):

Here's an interesting Pew Research article, looking at the pros and cons of algorithms in health care and other areas:  http://www.pewinternet.org/2017/02/08/code-dependent-pros-and-cons-of-the-algorithm-age/
The authors argue that algorithmic approaches, while improving efficiency in some ways, also carry the risk of deepening divides and creating filter bubbles.  They can rely on biased data, may particularly have negative effects on people who are poor, less educated, and disadvantaged, and can limit freedom of choice.