Thursday, April 3, 2014

More on Geriatric Depression and Overprescribing Antidepressants in Primary Care

A recent article in the New England Journal of Medicine adds some more epidemiological data to the issue of the treatment of geriatric depression.  The centerpiece of the article by Ramin Mojabai, MD is a graphic that is a combination of data from the National Survey on Drug Use and Health or NSDUH and the U.S. National Health and Nutrition Examination Survey or NHANES.  His central point is that the majority of people diagnosed with depression in primary care clinics do not meet diagnostic criteria for major depression.  The actual numbers for the elderly are 18% of those diagnosed with depression and 33% of those diagnosed with major depression actually have a diagnosis of major depression as assessed by rating scales or structured interviews.  The bar graphs in the A panel illustrate that most people over the age of 35 who are taking antidepressants do not meet criteria for major depression.  The opposite is true for the 18-34 year olds where antidepressant prescriptions are less than the prevalence of depression.  Panel B illustrates that the prevalence of people who were told by their clinician that they had depression and did or did not meet criteria for major depression.  In all cases the clinicians involved estimated non-major depression as being more prevalent than major depression.  Can we learn anything from these graphs?

The striking feature in Panel A is the dissociation of the total number of people taking antidepressants from the people with a diagnosis of major depression.  I can see that happening for a couple of reasons.  I would expect the number of people who are stable on antidepressant therapy to accumulate over time.  Most of them would have major depression in stable remission and would no longer meet the criteria.  A related issue is the atypical presentations of depression with increasing age.  I have seen many cases of depression presenting as pseudodementia, Parkinson's syndrome, and polyarthritis or a similar chronic pain syndrome.  In all cases, the symptoms responded to antidepressant medication but they would not meet criteria for major depression and most often the evaluation would resemble an evaluation for a medical problem.  There is also the problem of depression in the aging population who have a form of dementia.  At the upper end of this age distribution that may involve as many as 5% of the 65 year old population and they are likely overrepresented in primary care settings.  Lastly there is the problem of suicide in the elderly.  I reviewed a recent paper in the American Journal of Geriatric psychiatry that documented a decreased risk for suicide in elderly men and women who were taking antidepressants and the increased suicide risk in that group.  It is likely that many primary care physicians are concerned about that higher level of risk and this may influence prescribing for this group.  The other interesting comparison is that using different methodologies the ballpark antidepressant use in the elderly in Denmark approximates the antidepressant use on the US.  It is probably a few percentage points lower, but the study in Denmark used a more robust marker of antidepressant use (refilling the actual prescription) rather than survey questions.

The author addresses the issue of antidepressants being used for other applications like headaches and chronic pain chronic pain and states from an epidemiological perspective that two thirds of the prescriptions are for "clinician diagnosed mood disorder."  The standard used in this study of DSM major depression criteria is too strict to use as a marker for antidepressant use since there are other valid psychiatric indications that primary care physicians are aware of and treat.  Panic disorder, generalized anxiety disorder, social anxiety disorder and dysthymia are a few.  There are also more fluid states like adjustment disorders that seem to merit treatment based on severity, duration, or in many cases by the fact that there are no other available treatment modalities.  These are all possible explanation for the author's observation that the majority of people diagnosed with depression in primary care clinics do not meet criteria for major depression.

Diagnostic complexity is another issue in primary care settings.  Patients are often less severely depressed, have significant anxiety, may have an undisclosed problem with drugs or alcohol, and have associated medical comorbidity.  In an  ideal situation, a diagnosis of depression is not necessarily an easy diagnosis to make.  It takes the full cooperation of a patient who is a fairly accurate historian with regard to symptom onset and dates.  They are harder to find than the literature suggests.  The epidemiological literature often depends on lay interviewers using structured interviews like the DIS or SADS to make longitudinal diagnoses.  This approach will not work for a large number of patients and a significant number will not be able to recall events, dates, medications or prior treatments with any degree of accuracy.  With that level of uncertainty, antidepressant prescription often comes down to a therapeutic trial so that the patient and physician can directly observe what happens between them as the only available reliable data.

The  author notes that the primary intervention for depression in primary care is the prescription of antidepressants.  He talks about the ethical concerns about exposing patients especially the elderly to antidepressant drugs if it is not warranted, but he is using the major depression diagnosis here as the standard for treatment.  He makes the same observation that I have made here that mass screening for depression is not warranted based on the concern about false positives.  That stance is supported by the Canadian Task Force on Preventive Health Care.  The U.S. Preventive Services Task Force recommends screening "when staff assisted depression care supports."  My position is that screening, especially in medical populations is problematic not only from the false positive perspective but also because the screening checklist is often used as the diagnosis and an indication for starting antidepressant medications.  Screening checklists are also political tools that are used to manipulate physicians.  The best example I can think of is using serial PHQ-9 scores as a marker of depression treatment in primary care clinics even though it has not been validated for that application. As an extension of that application the PHQ-9 is used as a quality marker in clinics treating depression over time even though there is no valid way to analyze the resulting longitudinal data.

The author makes recommendations to limit the overuse of antidepressants and uses the stepped care approach with an example from the UK National Institute for Clinical Excellence or NICE.  These guidelines suggest support and psychoeducation for patient with subsyndromal types of depression.  A fuller assessment is triggered by very basic inquiries about mood and loss of interest.  Amazingly the PHQ-9 is brought up as an assessment tool at that point.  More monitoring and encouragement is suggested as a next step with a two week follow up to see if the symptoms remit spontaneously.  Medications are a third step for longstanding depressions or those that do not remit with low level psychosocial interventions.  An expert level of intervention is suggested for patient with psychosis, high risk of suicide, or treatment resistance.  That seems like a departure for NICE relative to their guideline for the treatment of chronic neuropathic pain.  In that case the referral for specialty care was contingent on a specific prescribing consideration (opioids) and the pain specialist was considered the gatekeeper for opioid prescriptions in this situation.  Antidepressants are seen as overprescribed drugs but no gatekeeper is necessary.  I suppose the argument could be made that there are not enough psychiatrists for the job, but are they really fewer than pain specialists who prescribe opioids for chronic neuropathic pain in the UK?  

This model is only a slight variation on the Minnesota HMO model of screening everyone in a primary care clinic with a PHQ-9 and treating them as soon as possible with antidepressants.  The driving factor here is cost.  With a month of citalopram now costing as little as $4.00 - there is no conceivable low level psychosocial intervention that is more "cost effective".  I have also been a proponent of computerized psychotherapy as a useful intervention and it is not likely that the Information Technology piece needed to deliver the psychotherapy would be that inexpensive.  Another well known correlate of depression in the elderly is isolation and loneliness.  I was not surprised to find that there were no interventions to target those problems since it would probably involve the highest cost.  In the article standard research proven psychotherapies were recommended on par with the medical treatment of depression, but the question is - does anyone actually get that level of therapy anymore?  My experience in assessing patients who have gone through it is that it is crisis oriented and patients are discharged at the first signs of improvement.  That may happen after 2 or 3 sessions.

I doubt that the stepped care approach will do very much to curb antidepressant prescribing.  This study suggests that overprescribing is a problem using a strict indication of major depression.  There are always problems with how that is sorted out.  I have not seen any studies where a team of psychiatrists goes into a primary care clinic and does the typical exhaustive diagnostic assessment that you might see in a psychiatric clinic.  It would probably be much more relevant to the question at hand than standardized lay interviews or checklists.  There is also a precedent for interventions to curb over prescribing of medications and that is the unsuccessful CDC program to reduce unnecessary antibiotic prescriptions.  If clear markers of a lack bacterial infection can be ignored, what are the chances that an abstract diagnostic process will have traction?

And finally the stepped care interventions seem very weak.  This is a good place for any number of professional and public service organizations to intervene and directly address the psychosocial aspects of depression in the elderly.  Public education on a large scale may be useful.  The psychoeducation pieces can be included in relevant periodicals ahead of time rather than as a way to avoid the use of medications.  Environmental interventions to decrease isolation and loneliness is another potential solution.  From a medical perspective, if the concern is medication risk every clinic where antidepressants are prescribed should have a clear idea of what those risks are and how to assess and prevent them.  Patients who are at high risk from antidepressants should be identified and every possible non medication intervention (even the moderately expensive ones) should be exhausted before the prescription of antidepressant medication.   Primary care prescribing patterns that potentially impact the patient on antidepressants should also be analyzed and discussed.  A focus on risks and side effects can have more impact on the prescription of antidepressants than psychosocial interventions and waiting for the depression to go away.

George Dawson, MD, DFAPA

Supplementary 1:  Permission and credit for the graphic:

"From New England Journal of Medicine, Ramin Mojtabai, Diagnosing depression in older adults in primary care. Volume No 370, Page No. 1181, Copyright © (2014) Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society."

No comments:

Post a Comment