• Psychiatric readmissions: A reader responds

    About one of my psychiatric readmissions posts, a reader wrote me,

    Your comparison to a readmission for another broken bone seems pretty inappropriate. I don’t know enough about physical medicine/disease to find a “perfect” analogy, but mental health readmissions seem to be more analogous to something like diabetes or certain types of heart disease – chronic conditions that have a high risk of rehospitalization if left untreated, but should be able to be well controlled (and thus few rehospitalizations) if appropriate treatment (such as medication) is determined and adherence is high. Based on having a family member who has been hospitalized several times for anxiety (which MAY be different from diagnoses like bipolar, in this context – but may not be), it seems that a readmission within a couple of years may well mean that the mental health after the first discharge never returned to “100%”, but it was also not “bad enough” during the interim to require rehospitalization. (Medication compliance has been high – but that alone has not been a “cure.”) If you’re looking for multiple readmissions, even 10 years might be reasonable.

    Like the reader, I too am not expert in this area, so I offer these points as speculative and invite subject matter experts to weigh in:

    • It’s essential to be explicit on what use one has in mind for readmission rates. A typical use is as a quality indicator. That’s what I have in mind. What factors of treatment are modifiable such that readmissions (somehow measured) could be reduced? I expressed this more clearly here.
    • My broken bones case did assert a chronic condition of a sort, “bones more prone to fracture than is typical.” In fact, I was getting at the very type of chronic issue that the reader raises. So, I agree that diabetes and cardiovascular conditions apply. Even with that, I’m less certain that multi-year readmission rates are reasonable quality indicators for mental or physical health.
    • As suggested, it is reasonable to expect variation in potentially avoidable and unavoidable readmissions across mental health diseases and even within diseases, by severity and other related factors.
    • The literature supports the sensible idea that medication adherence is a relevant factor. You’ll see this factor mentioned in my recent round-ups (here and here). Is poor adherence a patient or provider factor? At first blush it seems like a patient factor, but consider also attention to side effects and outpatient follow-up and monitoring. So, if adherence is poor — poor enough to lead to a readmission — is this an indication of poor provider quality or not? This is a tricky question. I invite comment.
    • More generally, and acknowledging all the other social and prognostic causal factors, are readmissions and indication of poor quality of either the index hospitalization or subsequent outpatient care? If so, what’s the maximum interval between index and subsequent hospitalization for which it is reasonable, on average, to infer a quality problem?

    About the questions in the last bullet, I refer you to the OECD report, “Selecting Indicators for the Quality of Mental Health Care at the Health Systems Level in OECD Countries” (PDF). The authors considered readmissions at 7 and 30 days but not longer. Unless I missed something, it’s not evident in the report why longer intervals were not considered, but as I wrote, the case for (poor) quality-driven rehospitalization becomes weaker as the interval lengthens. That’s not to say rehospitalizaton at one, two, or ten years later is not related to quality of care. It’s just far more likely that if rehospitalizations do indicate poor quality at all, more proximate rehospitalizations are a stronger indicator. How many other good reasons are there that a patient should return to the hospital in a week or month after discharge? All other things equal, is a hospital with a higher 7- or 30-day rehospitalization rate offering worse inpatient care or transitions to worse follow-up, outpatient care?

    (There are so many issues that I’m not raising here. For a good discussion how the use of rehospitalizations as quality indicators can be problematic see Karen Joynt and Ashish Jha.)

    The OECD report authors write,

    Hospital readmission rates are widely used as proxies for relapse or complications following an inpatient stay for psychiatric and substance use disorders. Since they indicate premature discharge or lack of coordination with outpatient care, high readmission rates have led some inpatient facilities to examine factors associated with readmissions, including patient characteristics, length of stay, discharge planning and linkages with outpatient care. Given the high cost of institutional care, reducing readmission rates can have a substantial effect on mental health spending.

    They also write,

    Studies of the association between readmissions rates and other indicators of quality have been mixed. Rosenheck et al. (1999) found small but significant correlations between 180-day readmission rates and other measures of outcome in a cohort of veterans with service-related posttraumatic stress disorder (PTSD), but had non-significant findings for 14 and 30-day rates. Lyons et al. (1997) found no association between psychiatric readmission rates (30 and 180-day) and clinical measures of outcome in a more diverse cohort of inpatients with psychiatric disorders.

    That’s pretty weak support for a 7- or 30-day readmission rate quality indicator. Apart from the Rosenheck et al. study, it isn’t that clear that there is a large body of work that suggests longer is necessarily better. But, like I said, this is not an area of expertise for me. Can anybody speak more expertly to these issues?


    • I am shaking my head a bit over this whole discussion. In my experience, availability of mental health treatment, in-patient or out-patient, is extremely limited. It is common that people with serious mental health problems can often have a lot of difficulty getting treatment at all. When I lived in Kalamazoo, Michigan, a colleague with serious depression couldn’t even get on a waiting list to see a psychiatrist. He ended up getting a prescription for Prozac from his primary care physician, with no supportive services whatsoever. When I lived in Ann Arbor, I was on the board of a charity that supported people with a variety of mental health problems, and again many of our clients could never access regular care. You had to be an immediate danger to yourself or to others in order to be admitted for in-patient treatment. To worry about nuances in care, when the whole system seems to be breaking down, just doesn’t seem profitable.

      • I hear you. At the same time, with capacity so constrained, don’t we want to know how best to treat those who do have an inpatient stay so they are less likely to return and use resources that might otherwise be spent caring for others? Moreover, considering the hardship of acute psychiatric events, don’t we want patients to receive care that is more likely to keep them in the community?

        I don’t think we can so easily separate capacity and access with quality of care. That does not mean we do not have an access problem. That’s just not the subject of my current focus.

    • As someone who does work in this area and works in a research program (I’m in Canada BTW), I think it’s important to note that this is not about mental health in general, but severe mental illness. The population hospitalized has typical diagnoses of chronic depression, schizophrenia or other psychoses, personality disorders, and bipolar mood disorder. They have chronic diseases to be sure, but it’s important not to assume that routine hospitalization is a fact of chronicity (though it increases the risk of course).

      A flaw with using readmission as a proxy for clinical outcomes and quality of care is that many hospitals (including the US) have appropriate protocols in place NOT to admit for certain severe events that are to some extent part of the disease process, even though lay people would consider such events worthy of admission. For example, repeated suicidal gestures by patients with borderline personality disorder, episodes of uncontrolled behavior by schizophrenics, or aggressive behavior in adolescents. While those around the patient may object to such protocols, these are cases where hospitalization is often of little benefit. In other words, there are many instances of relapse that do not lead to admission,

      Also, readmission does not cover emerg visits (unless I missed that in your excellent posts), a metric that might capture similar issues of chronicity, nonadherence, etc. I assume a majority of the readmissions in these studies are the result of emerg visits.

    • I think it’s worth pointing out the difference between admissions and readmissions here — measuring admissions for mental illness (or, as another commenter pointed out, any chronic disease) probably reflects access to outpatient care and quality of outpatient care delivered, in addition to social support and ability to care for oneself effectively. Readmissions, on the other hand, are (at least as currently framed) supposed to reflect whether or not the care provided during a distinct inpatient episode was adequate or flawed. In that context, short-term readmissions are certainly a better measure of one distinct inpatient episode’s efficacy, but their ability to measure the efficacy of that episode declines over time. As you get out to longer-term “readmissions”, you’re really measuring outpatient, not inpatient, care. In an ideal healthcare system (ACOs?), the inpatient and outpatient settings would be linked, and accountable together in a meaningful way. In the current system, a hospital that discharges a patient may be able to control when their first follow-up appointment is scheduled, but beyond that, the rest is pretty much out of their hands. Both admissions and readmissions may be useful measures, but I suspect they measure different things, and each has its own set of pros and cons.

    • I echo several of the comments on your excellent discussion, and agree that short-term readmissions partly reflect quality of care during the preceding inpatient episode whereas longer-term readmissions are far more influenced by quality of outpatient rather than inpatient, care.
      My views on this topic are influenced by the fact that I am a psychiatrist with experience in the ED and in the OP treatment of people with severe mental illnesses and I am also a services researcher with an interest in quality of mental health care. My read is that the scientific evidence is limited by the difficulty in controlling for the multiple factors that drive both admissions and re-admissions (a previous admission is the best predictor of re-admission). Those include patient factors (clinical, social), admitting physician factors (experience, training), and system-level factors (payment system, quality of OP and IP care; quality of the social safety net).
      In the US and increasingly in other cost-conscious places of the world, inpatient admission is being reserved as the absolute last resort and only for people who are about to hurt themselves or others, which is not the exclusive province of people with severe mental illnesses. As soon as their “lethality” is ameliorated, they are discharged to outpatient care that is unable to meet the patient’s needs. Further, in the US and other places with inadequate social safety nets, people with extremely limited social resources and desperate for a roof or a hot meal may seek admission claiming they are suicidal or homicidal just to gain admission. Although many may be diagnosed with a personality disorder plus minus a substance abuse disorder, they do not have a severe mental illness.