Things I Wish

In the May 2018 editorial in Schizophrenia Bulletin I briefly noted 13 things that I did not like. In this editorial, the reader will find 5 issues where colleagues and I should have made a difference but failed.

1. 1. Platt’s article¹ on the role of hypothesis falsification in science is compelling. But the field seems to have far more “failed” studies than hypothesis falsification. We publish an example of hypothesis falsification involving a brain mechanism for schizophrenia.² Findings supported the mechanism in an identifiable subgroup only. Knowledge was advanced by rejecting the hypothesis for schizophrenia and identifying the relevant subgroup supporting the hypothesis. The disappointment here is not just the low rate of hypothesis testing. Psychotic psychopathology is present in all cases of schizophrenia, but other psychopathologies are not uniformly present. Testing a hypothesis relating to avolition in schizophrenia, for example, may have weak and misleading results if testing schizophrenia versus normal controls. Even if all the participants with schizophrenia were selected with avolition there would be a host of differences with normal controls related to confounds. Comparing schizophrenia participants with and without the hypothesized factor will reduce common confounds such as medication, daily activities, body weight, employment, marriage, and a host of differences that confound interpretation of differences between schizophrenia and normal controls.

2. 2. Persons with schizophrenia often stop antipsychotic medication and often delay or refuse return to antipsychotic medication during disease exacerbation. Alternative treatment, if efficacious, might provide an approach to preventing an initial exacerbation from becoming a relapse. As part of a targeted drug study, we tested a clinical trial relating to treatment initiation in persons with schizophrenia who were medication free but symptoms associated with beginning exacerbation were observed. Patients were randomly assigned to an AP drug, or diazepam, or a placebo. The 2 drugs were similar in efficacy for relapse prevention, and both were superior to placebo. This seemed to be an alternative approach to exacerbation in patients who decline antipsychotic medication. To my knowledge no replication studies³ and no influence on clinical management.

3. 3. For most AP drugs we know a dose range that works but not minimal dosing that is effective. At the time of the following study the main way to address tardive dyskinesia was reduced AP drug exposure. But when does dose reduction become inadequate for psychosis? In a RCT we did a 1 year look at fluphorazine depot administration at recommended level compared to a placebo substitution for 2 out of every 3 injections. Clinical course and outcome were similar in the 2 groups with no evidence of therapeutic disadvantage to the reduce dose group.⁴ Here also the field has been slow in addressing optimal dosing to prevent adverse effects. Davis et al have provided substantial information on minimal dosing for some antipsychotic drugs.⁵

4. 4. The following is just an interesting set of findings with a computational approach. Led by John Bartko in the early 1970s we addressed clinical assessment data in schizophrenia with 3 cluster analytic methods. Each gave interesting results from the same data, but results were not the same. We then used K-means but removed a random 20% of the clinical data, reran the program with different results. Then, keeping all the clinical data and removing 20% of patients resulted in interesting but different findings.⁶ Replication samples are used commonly today to assure validity to findings but wonder if results are sufficiently stable across computational approaches.

5. 5. The most important instance of our research findings failing to adequately inform the field relates to analyses in the data from the International Pilot Study of Schizophrenia.⁷ In 1974, together with John Strauss and John Bartko, we presented data confirming schizophrenia as a clinical syndrome rather than a disease entity⁸ and showed that Schneiderian First Rank Symptoms were not unique to schizophrenia.^9,10 Failure to convince the DSM-III process managers had the result of 30 years of schizophrenia research organized in the schizophrenia versus control framework with the result of very moistest advance given the heterogeneity of schizophrenia and the failure to focus on specific psychopathologies. NIMH made a sharp turn, RDoC, to call for research on specific elements and welcomed transdiagnostic research.¹¹ DSM-5 made explicit that we are dealing with a heterogeneous syndrome and not a disease entity and removed First Rank Symptoms from its unique role in diagnosis of schizophrenia.¹²