Role of Placebos versus Antidepressants and Contribution to the Doctor-patient Relationship

ABSTRACT

---

Current literature provides arguments against and in favor of the discussion on effectiveness of the antidepressant medication. Both the methodology, study models and tools used for comparing effectiveness of medications cause controversy. This debate becomes harder when considering the existence of placebo effects in results that promote or question the use of these drugs. Understanding placebo actions not only allows us to better understand the research results regarding use of antidepressants, but additionally, regarding effectiveness of the various treatments that today are used in mental health.

Key Words: placebo effect, antidepressant agents, mental health.

EFFECTIVENESS OF ANTIDEPRESSANTS

---

The debate regarding antidepressants effectiveness has been an issue since the moment they started to be used massively. Various opinions, both in favor and against it have arisen in time, but when comparing its effect with studies involving placebos, some doubts regarding its usefulness when performing studies aimed to extrapolate results to clinical internship. If we understand that a placebo is a substance with no active potential, we should wonder then why, in randomized studies on antidepressants effectiveness, it seems that placebos have a therapeutical potential that is -many times- similar to that of drugs, or at least it does not allow to prove that active substances are more clinically effective.

Moncrieff´s ⁽¹⁾ Action Model focused on drugs highlights that psychiatric drugs may be considered as psychoactive substances, if we think they pass into the blood-brain barrier thus affecting brain functioning, and causing some specific mental/physical disturbances in any person who uses them. These people may consider this condition is more desirable than the original mental disturbances. According to the foregoing, the role that these drugs should have is not related to treating a mental disturbance from its origin, thus causing long lasting results, but rather leading to bearable mental conditions which allow the patients to have a regular daily life.

One of the reasons why patients use antidepressants is because of the potential adjustment of base neurochemical imbalance which could be root of the pathology and, therefore, lead to that so longed “normality” ⁽¹⁾. However, a review made by Khan et al. ⁽²⁾ reported that a higher effectiveness of antidepressants is achieved against placebos in less than half of the studies made. On the other hand, there is no clear correlation between the medication and specific disease as a treatment ⁽¹⁾. In clinical tests made in 1982, antidepressants were --as an average-- 6 points more effective than placebos when compared by using the Hamilton´s Scale. This difference decreased 3 points –as an average--, in 2008 ⁽³⁾, which leads to questioning what changes happened during the last few decades so that the effect of antidepressant medication has such a huge variation.

An analysis of randomized tests --made on antidepressants effects in the short term-- in placebo-controlled ⁽⁴⁾ major unipolar depression, shows the rate of responders to antidepressants and a placebo is 50% and 30% respectively. The traditional calculation of the antidepressant-placebo difference (50-30=20%) in these pharmacological tests is based on the assumption that placebo responders should be antidepressants responders. Such assertion has never been neither investigated nor proved, given the complexity involving the methodology of the studies aimed to prove such theory.
When making a review of the literature, it is very easy to find studies with contradictory results. The sampling size seems to be the hardest difficulty. In 2004A a review was made by Moncrieff et al. ⁽⁵⁾ including 9 studies, with a total of 751 participants, after clearing the variables which could cause a bias on results, concluded that antidepressants and placebos effects do not have a statistically significant difference. On the other hand, updated reviews, just as that made by Cipriani et al. ⁽⁶⁾ in 2018, comparing effectiveness of 21 antidepressants versus placebo, where 522 studies were included with a total of 116,477 participants, concluded that all antidepressants proved a significant reduction of depressive symptomatology in adults suffering depression. However, results reported by studies made on pathologies, such as depression seem to be influenced more than just the time when they were made or its methodology, as even studies with similar characteristics made in the same years report significant differences in their results, thus leading to questioning if other factors, such as duration time of the studies could have a relevant significance on results. Just as stated by Hegerl and Mergl ⁽⁷⁾, duration of effectiveness studies has changed in time, from an average of 4 to 6 week, therefore, including patients who suffer acute depression conditions, even one of 12 weeks in the studies made during the last two decades, which could impact on clinical tests results, as the natural course of the disease becomes a distractor.

Beneficial effects of ISRS are not yet determined whether if they are superior to adverse effects ⁽¹⁾, although some reviews have been made, such as that of Arroll et al. (8) who regarding depression determined for the ISRS a number necessary to (NNT) from 7-8, and a number necessary to damage, from 20-90. On the other hand, a review made by Bighelli et al. ⁽⁹⁾, of 41 studies, with 8,252 participants, compared antidepressants effects versus placebo for managing panic disorder in adults, reports a slight bias in terms of its higher beneficial effect in antidepressants, but it highlights that NNT was 7. Therefore, it calls our attention that NNT has not had a variation in time, but despite of all this, when analyzing various effectiveness studies against the placebo during the last few years, we still have mismatching results.

Regarding antidepressants effectiveness in severe pathologies, Leuchter et al. ⁽¹⁰⁾ studied cerebral functions of people with a diagnosis of major depression treated with antidepressants or placebo. They observed that both groups obtained similar results by reducing depressive symptoms, with the beginning of the study, however, the two groups did not have any physiological equivalence, even though both affected functions of the prefrontal brain region, their mechanism was contrariwise. The study reported that people responding to placebo had a significant increase of their functions, while those using antidepressants reported a reduction in the functions in such area. This proved that placebos and antidepressants could have various action mechanisms; therefore, the response to each of them could be completely independent among them, and there are no overlapping results; thus generating a bigger problem when trying to interpret results and evaluate the methodology carried out for clinical tests.

Unlike the effect of antidepressants in young adults, whose actual result is well known and widely studied, elderly people have more comorbidities, as well as social determinants having a higher risk of suffering psychiatric pathologies, such as depression ⁽¹¹⁾. For them it is quite important to consider the type of therapies to be used and their effectiveness. The systematic review made by Wilson et al. ⁽¹²⁾, aimed to analyze the effect of antidepressant medication in elderly patients, reported higher effectiveness to reduce depressive symptoms by using drugs, compared with placebos, apart from a similar response to these drugs when comparing with younger adults. Along with this, a more updated review of the literature reported that even though the ISRS were not better than placebos, when reaching disease remission at 8 weeks, did report to have higher effectiveness to prevent recurrence ⁽¹³⁾. Even though there is favorable literature aimed to avoid relapse risk in these patients, that is still not enough. Long term effects of these drugs – both for adults and for elderly people is still under question.

Considering that NNT of drugs antidepressants has not has any significant changes during the last 10 years, and according to Moncrieff statements that there is no direct correlation between psychiatric medication and specific diseases  ⁽¹⁾, it is necessary to evaluate, in the public health area, if the benefit of these drugs is enough to consider its use in primary care as unique therapy, chosen before other therapeutical , such as psychotherapy or complementary therapies  (acupuncture, physical exercise, among others). When focusing only on if antidepressant effect is higher than the placebo effect, there are many doubts whose grounds are questioning the use of placebos, and the difficulty to correlate this to the current clinical reality ⁽⁷⁾.

Findings made by Li et al. ⁽¹⁴⁾ provide an interesting perspective stating that more severe patients are more willing to remain ins studies with placebos, while -on the other hand- Hegerl and Mergl state that the fact of not knowing what they are consuming, along with the participation of patients coming from countries having a poor health coverage –where there is a belief that if they participate in clinical tests that could mean to potentially receive that so longed active component. This situation would make that participants an adulterate the information provided in the Hamilton´s Scale, in order to look like having a more severe range of symptoms; therefore, this could imply they would be included in the studies; and on the other hand this would influence the patient´s attitude to receive the pills ⁽⁷⁾.

PLACEBO RELEVANCE

---

If there is a great deal of research casting doubts on the effectiveness of antidepressant medication, then why do these work? McCormack and Korownyk ⁽¹⁵⁾ made a paper commenting about Cipriani et al´s review. ⁽⁶⁾. They stated that in the placebo groups the average improvement response ranges between 30 to 40%. They interpret that an Odds Ratio of 1.6 means 10-12% of higher improvement in the treatment group against the placebo group. In other words, if 10 patients with suffering moderate and severe depression take antidepressants during 8 weeks, five of them (50%: 40% per placebo plus benefit of 10%) would report to feel better, but in four of them this improvement is not due to the drugs. ¿What allowed that those 4 people felt better with the placebo?

In other words, this effect occurs in a medical setting. a physician gives a patient a pill, but the patient does not know that pill is made of sugar only. This is the placebo. The patient´s health further feels better, as the patient believed the pill had a pharmacological agent, which is good for his/her condition. This is the placebo effect ⁽¹⁶⁾.

Many studies have reported the relevance and extension of this phenomenon. Most of them have focused on the effectiveness in subjective discomforts, such as pain, anxiety and depression; however, there are numerous investigations describing its physiological effects: heart functions effects, hyperlipidemia ⁽¹⁷⁾, healing of wounds ⁽¹⁸⁾ and even extended life of patients suffering cancer ⁽¹⁹⁾, just to name a few of these studies.

THEORIES ABOUT ACTION MECHANISMS

---

Even though the placebo effect, probably has always existed in medicine (20), studies regarding its action mechanism have not devised an agreed explanation so far; instead of that, only various hypothesis are available, and each of them assume to be the right one ⁽²¹⁾.

From a psychological point of view there are many mechanisms contributing to the placebo effect; these include expectations, conditioning, learning history, memory, motivation, somatic aspects, reward, anxiety reduction, and senses ⁽²²⁾. Even though there are more research regarding these placebo effect mechanisms, the conditioning hypothesis and the expectations hypothesis are two theories currently competing in the research area.

The first psychophysiological hypothesis uses the operant conditioning theory, where beneficial situations are learned and memorized; and then returned under the same conditions, except for one: the active product. With this theory we could explain the placebo effect and specially all the phenomenon known as placebo-sag, where accumulation of negative experiences with various therapies reduces the placebo effect ⁽²³⁾. However, this hypothesis does not explain how the placebo effect may happen in the first situation, with no prior conditioning, or why we can see the phenomenon of familiarization in chronic diseases treatments, where with the same dosage, the effect decreases more and more and the placebo responds even worse as well, and their effect shoud be contrariwise. No doubt, we can see how this theory explains part of this mechanism, but it does not explain, the anticipatory trait and not only the reactive trait of the placebo response.

The second hypothesis is the theory of expectations. Currently it has become the most popular theory about this matter. It is based on the hypothetical expectation the patient has about the using a product would cause on him/her. Patient´s expectations, hope and his/her eagerness to heal may arise the placebo effect. This theory is better understood as the cognitivist approach to the psychological explanation of the phenomenon. This theory provides a reflexive consciousness for generating expectations, therefore, the placebo effect would become an intentional and anticipatory act, consciously accessible, where the placebo effect may happen due to conditioning tests by means of a product, but also by means of other sources, such as verbal/social signs ⁽¹⁶⁾. Even though this hypothesis has become prevailing in the psychological theory of the placebo effect, and it allows to deeply outline the physician´s role in the patient´s expectations, --which will be further reviewed—there are still some incomplete aspects in its drafting. The first element of discussion is the placebo effect, included within bodily activities classified as skillful, and activities named as irreflexive. The skillful part is developed within a learning process. This is why psychologists build the Expectations Theory by means of cognitive processes involving cultural/social learning mediated/directed by conditioning tests. The problem is the difficulty to start a physiological response, by means of a conceptual or linguistic representation. In our daily life thinking about a specific part of our body or brain area is not enough for significantly alter its conduct. In this sense it is possible to argue that an agent, in this case the placebo, may have a representation of its action, but this functional representation works hidden. This is how the placebo response is suggested to use implicit/unconscious expectations ⁽²⁴⁾. This leads us to the second component of the discussion. If expectations are not explicit constructs, its demonstration is subject to criticism and cannot be forged. if expectations measurements depend on the verbal self-report and expectations are defined as implicit, regardless of the result of an experiment, any researcher can challenge data reliability.

Even though the Theory of Expectations has limitations in terms of research, its foundation allows us to put the focus of the placebo effect, not on the placebo itself, but rather in the psychological/physiological response meaning, which is found in the origins or in the treatment of a disease ⁽²⁵⁾.

PHYSICIANS AS HEALING AGENTS

---

Before color, shape or administration way, placebo effectiveness is first/mainly determined by the information provided by the physician. This is how medicine becomes significant for the patient, and the physician recognizes its valuation in his/her role as subject involved in a relationship with the patient.

The physicians´ empathy capacity; their conviction in what they do; their confidence dedicated to their patients; their capacity to feel they can control the situation ⁽²¹⁾; In short, the quality of the personal relationship he/she generates; all of them are quite crucial for the healing process, as these subjective aspects may, oftentimes and by themselves cause a physiological change expected from the medication or, to a great extent, reinforce the verum action.

From the patient´s point of view, the most important component for his/her improvement are the general qualities of the professional, whether he/she is a physician or a psychologist ⁽²⁶⁾. Patients experiencing a cooperative/participative relationship is key for creating and keeping a partnership aimed to favor patients´ improvement. Understanding this, both for the professional --who is willing to listen to as an active assistance component-- and for the physician --who provides the drugs for discomfort relief of the patient-- must be the foundation to believe it is possible to make a significant change in the life of others.

DISCUSSION

---

Currently, one of the most relevant health concerns is the need to limit expenses in this area. The most used method regarding this issue is to demand from treatments to prove their effectiveness. The approach that has most contributed to this need is evidence-based medicine.

From the 60s, for depression treatment, the standard implemented for approval of a psychotropic drug has been randomized clinical trial (RCT). RCTs have allowed to provide clinical orientation on the effect of medications and to limit existing variations in medical practice. However, this type of tests –used to learn about the specific effects of a treatment--

Have unexpectedly reasserted, the relevance and extension of this unexpected guest for biological psychiatry as placebo effects are.

As discussed at the beginning of this article, there is plenty of research that has proved a significantly poor/inexistent effectiveness regarding prevalence of the treatment with antidepressants and placebo-controlled groups. Just as stated by Khan et al. ⁽²⁷⁾ in a systematic review, antidepressants are reported to work only slightly better than placebos. This low effectiveness is even weaker when studies are performed with a double blind, rather than when the researcher knows what is the control group. But even more relevant when questioning the role of antidepressants as the basis of depression treatment is the comparison with other treatments, such as psychotherapy or alternative treatments, where antidepressants report the same effectiveness than other types of treatment and a slight increase, compared with the placebo effect.

These data reveal the need to perform a critical analysis of the prevalence of one treatment compared with the other, but even more, it highlights the need to deeply understand the action mechanisms underpinning interventions in mental health.

If we consider that the placebo effect appears in all type of interventions as the expectation the patient has to receive the benefits of a treatment or intervention, we must understand that placebos –which are inert substances--, has become a part of any potential healing process. In this sense, we think it is necessary to make a part of the current discussion regarding treatment for mental health pathologies, the various unspecific factors determining the success/failure of a treatment. Ignore these facts and insist on a biological psychiatry, not committing the patient in his/her relationship with the professionals, in the sense of his/her discomfort, in the development of an improvement expectation and in the importance of the therapeutical, means to quit demands for help to alternative approaches not considering, for instance, the need to reduce reincidence, people´s direct expenses or its cost-effectiveness.

CONCLUSIONS

---

There are still many unexplored aspects in depression treatment and the implications to perform studies involving a placebo. Evidence-based medicine must not ignore elements hard to be quantified or explicit. Those so called unspecific factors may be very specific, such as the impact of the relationship or the impact on the patient regarding beliefs or expectations, and psychological mechanisms which from the base may be fruitful in all mental health interventions. Generation of scales aimed to better measure the placebo effect, and methods aimed to reduce thereof in this health area, could provide benefits for future clinical tests aimed to test the effect of psychiatric medication. The evidence we have gathered to date effectively proves that antidepressants have an active effect on patients, but in terms of public health, the effect could not be significant enough as to justify its funding above other therapeutical alternatives.

REFERENCES

---

1. Moncrieff J. Un approach alternativo of the pharmacological treatment en psychiatry. Rev Asoc Esp Neuropsiq 2018; 38: 181-193. doi: 10.4321/S0211-57352018000100010.
2. Khan A, Khan S, Brown WA. Are placebo controls necessary to test new antidepressants and axiolytics?. Int J Neuropsychopharmacol 2002; 5: 193-197. doi: 10.1017/S1461145702002912.
3. Khan A, Bhat A, Kolts R, Thase ME, Brown W. Why has the antidepressant-placebo difference in antidepressant clinical trials diminished over the past three decades?. CNS Neurosci Ther 2010; 16: 217-226. doi: 10.1111/j.1755-5949.2010.00151.x.
4. Rihmer Z, Gonda X. Is drug-placebo difference in short-term antidepressant drug trials on unipolar major depression much greater than previously believed?. J Affect Disord 2008; 108: 195-198. doi: 10.1016/j.jad.2008.01.020.
5. Moncrieff J, Wessely S, Hardy R. Active placebos versus antidepressants for depression. Cochrane Database Syst Rev 2004; 1: CD003012. doi: 10.1002/14651858.CD003012.pub2.
6. Cipriani A, Furukawa TA, Salanti G, Chaimani A, Atkinson LZ, Ogawa AND, et al. Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major depressive disorder: a systematic review and network meta-analysis. Lancet 2018; 391: 1357-1366. doi: 10.1016/S0140-6736(17)32802-7.
7. Hegerl OR, Mergl R. The clinical significance of antidepressant treatment effects cannot be derived from placebo-verum response differences. J Psychopharmacol (Oxf) 2010; 24: 445-448. doi: 10.1177/0269881109106930.
8. Arroll B., Elley CR, Fishman T, Goodyear-Smith FA, Kenealy T, Blashki G, et al. Antidepressants versus placebo for depression in primary care. Cochrane Database Syst Rev 2009; 3: CD007954. doi: 10.1002/14651858.CD007954.
9. Bighelli I, Castellazzi M, Cipriani A, Girlanda F, Guaiana G, Koesters M, et al. Antidepressants versus placebo for panic disorder in adults. Cochrane Database Syst Rev 2018; 4: CD010676. doi: 10.1002/14651858.CD010676.pub2.
10. Leuchter AF, Cook IA, DeBrota DJ, Hunter AM, Potter WZ, McGrouther CC, et al. Changes in brain function during administration of venlafaxine or placebo to normal subjects. Clin EEG Neurosci 2008; 39: 175-181. doi: 10.1177/155005940803900405.
11. Alexopoulos GS. Depression in the elderly. Lancet 2005; 365: 1961-1970. doi: 10.1016/S0140-6736(05)66665-2.
12. Wilson K, Mottram P, Sivanranthan A, Nightingale A. Antidepressant versus placebo for depressed elderly. Cochrane Database Syst Rev 2001; 2: CD000561. doi: 10.1002/14651858.CD000561.
13. Tham A, Jonsson OR, Andersson G, Söderlund A, Allard P, Bertilsson G. Efficacy and tolerability of antidepressants in people aged 65 years old or older with major depressive disorder A systematic review and meta-analysis. J Affect Disord 2016; 205: 1-12. doi: 10.1016/j.jad.2016.06.013.
14. Li F, Nasir M, Olten B, Block MH. Meta-analysis of placebo group dropout in adult antidepressant trials. Prog Neuropsychopharmacol Biol Psychiatry 2020 marzo [citado 4 of mayo of 2020]. Disponible en URL: https://www.sciencedirect.com/science/article/abs/pii/S0278584619302180. doi: 10.1016/j.pnpbp.2019.109777.
15. McCormack J, Korownyk C. Antidepressants effectiveness. BMJ 2018 marzo [citado 15 of febrero of 2020]. Disponible en URL: https://www.bmj.com/content/360/bmj.k1073. doi: 10.1136/bmj.k1073.
16. Stewart-Williams S, Podd J. The placebo effect: dissolving the expectancy versus conditioning debate. Psychol Bull 2004; 130: 324-340. doi: 10.1037/0033-2909.130.2.324.
17. Dobrilla G, Scarpignato C. Placebo and placebo effect: their impact on the evaluation of drug response in patients. Dig Dis 1994; 12: 368-377. doi: 10.1159/000171471.
18. Sullivan MD. Placebo controls and epistemic control in orthodox medicine. J Med Philos 1993; 18: 213-231. doi: 10.1093/jmp/18.2.213.
19. Kelkar P, Ross MA. Natural history of disease and placebo effect. Perspect Biol Med 1994; 37: 244-246. doi: 10.1353/pbm.1994.0021.
20. Macedo A, Farré M, Baños JE. Placebo effect and placebos: what are we talking about? Some conceptual and historical considerations. Eur J Clin Pharmacol 2003; 59: 337-342. doi: 10.1007/s00228-003-0612-4.
21. Tilmans-Cabiaux C. Effet placebo: the part of l’esprit dans the guérison. En: Dagneaux I, ed. Une prescription sur mesure: Moins mais mieux sied au grand âge. Louvain-the-neuve: Presses universitaires of Louvain; 2010. Disponible en URL: http://books.openedition.org/pucl/1677. Consultado the 10 of septiembre of 2019.
22. Price DD, Finniss DG, Benedetti F. A comprehensive review of the placebo effect: recent advances and current thought. Annu Rev Psychol 2008; 59: 565-590. doi: 10.1146/annurev.psych.59.113006.095941.
23. Peck C, Coleman G. Implications of placebo theory for clinical research and practice in pain management. Theor Med Bioeth 1991; 12: 247-270. doi: 10.1007/BF00489609.
24. Hahn R, Kleinman A. Biomedical Practice and Anthropological Theory: Frameworks and Directions. Annu Rev Anthropol 1983; 12: 305-333.
25. Moerman OF, Jonas WB. Deconstructing the placebo effect and finding the meaning response. Ann Intern Med 2002; 136: 471-476. doi: 10.7326/0003-4819-136-6-200203190-00011.
26. Botella L, Pacheco M, Herrero OR, Corbella S. Main results of the research en psychotherapy. En: Botella L, comp. Construcciones, narrativas and relaciones: aportaciones constructivistas and construccionistas a the psychotherapy. 1ra ed. Barcelona: Edebé; 2006.
27. Khan A, Faucett J, Lichtenberg P, Kirsch I, Brown WA. A systematic review of comparative efficacy of treatments and controls for depression. PLoS One 2012 julio [citado 5 of mayo of 2020]. Disponible en URL: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0041778. doi: 10.1371/journal.pone.0041778.

Abstract

---

Introduction: The need for early intervention in psychosis has led to searching for useful biomarkers in its prediction, where the analysis of spoken language stands out for its easy obtaining and low economic cost. In this systematic review, we analyze the main speech disturbances in patients at ultra-high risk for psychosis (UHR), evaluating their prevalence and their relationship with transition to psychosis.

Methods: A search was carried out in PubMed and Embase databases for studies in English or Spanish, as well as the reference lists of the articles found.

Results: Of 140 articles identified, 15 were included. The variables of the language analyzed were Linguistic Cohesion, Semantic Coherence, Syntactic Complexity, Metaphorical Production, Emotional Prosody and Formal Thought Disorder (FTD). The prevalence found ranged between 21% and 85% for disturbances included within the FTD, with not prevalence measures for the rest of the variables. The global accuracy for UHR transition to psychosis ranged from 70% to 100% across studies.

Conclusion: There is a wide range of speech disorders in UHR patients, where the FTD is the most studied one. The prevalence of these alterations seems to be high, especially with regard to FTD. The analysis of those speech alterations in UHR patients appears as an excellent tool to predict transition to psychosis, particularly through interview transcription and the use of Artificial Intelligence.

Key words: Ultra-high risk for psychosis; Psychosis; Speech.

Introduction

---

Language disturbances have been deeply studied in mental pathologies, such as major depressive disorder, mania and specially schizophrenia ^{(1) (2)}. Currently language is deemed as a predictive biomarker of psychosis ^{(3) (4)}. One of the main sources of language studying schizophrenia is the so called Formal Thought Disorder (FTD), which comprises various symptoms centrally manifested in psychotic disorders, whose presence has been correlated with poor functioning and evolution of the disease ^{(5) (6)}. FTD Clinical Evaluation, was proposed by Andreasen, in 1979 with the Scale named Thought, Language and Communication (TLC) who reformulated it, along with Grove, in 1986 ^{(7) (8)}. TLC gathers language and thought disturbances arising in FTD, in Subscales (positive, positive objective, negative, disorganized), and in turn in 18 factors ranged from (0) absent or not observed to (4) severe. However, in the literature ⁽⁹⁾ these divisions vary, according to subscales.

During FTD exploration and in other language disturbances, as psychosis biomarkers, various applications for Artificial Intelligence have been developed. By means of training non supervised learning models, better known as Natural Language Processing (NLP), successful automatic identification of some psychosis predicting characteristics in population at risk has been achieved, thus becoming a promising tool for supporting early diagnosis of severe mental diseases ^{(10) (11)}.

During the last decades the scientific community has focused on identifying predictors at the prodromal phase of psychosis, where some disturbances in psychosocial areas may appear, along with nonspecific symptomatology ⁽¹²⁾. This stage becomes an opportunity for early intervention and reduction of the transition rate to psychosis ⁽¹³⁾. This has led to the development of specific criteria aimed to identify people with high-risk clinical characteristics, for further development of psychosis, known as High risk mental condition HMRC. Various instruments for determining EMAR have been designed, among them we can find semi structured psychometric interviews, such as CAARMS (Comprehensive Assessment of At-Risk Mental States) ⁽¹⁴⁾ and SIPS/SOPS (Structured Interview for Prodromal Symptoms/Scale of Prodromal Symptomatology) (15). Both instruments have proved to have high sensitivity to detect EMAR. However, these have a low specificity, which implies that only 29% of EMAR patients transitioned to psychosis within 2 years, reaching up to 36%, after 3 years of follow-up ^{(16) (17)}.

The main objective of this systematic review is to propose an analysis of studies including spoken language disturbances in EMAR population. Likewise, the specific objectives are: 1) to identify the main speech disturbances in EMAR patients, 2) to determine prevalence of speech disturbances and 3) to evaluate the correlation among speech disturbances and transition to psychosis.

Methods

---

Sources of information

This systematic review was carried out by following PRISMA Guidelines, by searching in PubMed and Embase, on July 12th, 2020, with no restrictions in year of publication. Additionally, manual search is performed in the reference lists of the articles found, in order to identify relevant publications. 

Search Strategy

Key Words and the specific strategy for each data base is further detailed in Table 1. 

 

Table 1: Specific Search Strategies

 

Eligibility Criteria

Inclusion criteria for the studies were as follows: 1) cross sectional design (for Prevalence Rates) or longitudinal (for the rates of transition to psychosis), 2) EMAR diagnosis made by using one of the following instruments: Comprehensive Assessment of at Risk Mental States (CAARMS) (14), or Structured Interview for Prodromal Syndromes/Scale of Prodromal Symptoms (SIPS/SOPS) (15), 3). These are instruments for diagnosing transition to psychosis and for analyzing clearly specified language disturbances, and 4) Studies in English or in Spanish.

Only studies measuring speech variables will be included. Analysis of written language or gestures will not be considered. Additionally, only studies where open answers will be chosen are evaluated, excluding multiple choice tasks, with a limited number of potential answers.

Studies were not restricted per age of the participants, duration of the language evaluation strategy, follow-up duration (in longitudinal studies), or according to relation with antipsychotic treatment.

Data Retrieval

For each paper, the following data key were assessed: 1) study design; 2) sample characteristics; 3) EMAR diagnosis criteria and transition to psychosis; 4) characteristics of the language evaluation strategy; 5) language variables analyzed; 6) main results and statistics accuracy; and 7) software used (if available).

Recorded Variables

i. Oral Language: Specifically dealing with language structure disturbances and/or discursive content regarding interaction.

ii. EMAR: Criteria to define EMAR patients include 3 symptomatic groups: a) “Attenuated Psychotic Symptoms”  (APS) with positive psychotic symptoms with an intensity or under-threshold frequency, b) “Brief Limited and Intermittent Psychotic symptoms” (BLIPS) with brief psychotic episodes, less than 1 week of duration, which remits spontaneously with no need to use antipsychotic drugs, and c) “Genetic risk” (GRD) including people with schizotypal personality disorder and/or who have a first-grade relative with some psychotic disorders, combined with  a significant decrease in the functioning level.

iii. Instruments for EMAR diagnosis: CAARMS and SIPS/SOPS are psychometric interviews validated and spread all over the world (14) (15). Both contain various sections comprising positive, negative, cognitive, emotional/affective symptoms, among others.

iv. Prevalence and correlation with transition: Both measures were obtained by statistical analysis of the sample.

Results

---

Studies Selection

The flow chart of the studies selected in the systematic review is described in the Figure #1. In total, 235 records were found (225 from the search made in PubMed and Embase, plus 10 crossed references from the identified articles). Eligibility of 35 full text articles was evaluated. From all of them, 20 did not meet the criteria, therefore, they were excluded from the qualitative analysis. As a result, the studies included in this systematic review turned out to be 15.

Studies Characteristics

The list of studies included and their main characteristics are described in the Table 2.

 

All of these studies had a longitudinal design; 10 of them were prospective; 5 of them were retrospective, using transcribed interviews or scores obtained from other studies. 11 of them used SIPS/SOPS and 4 CAARMS for EMAR diagnosis. The size of the samples analyzed varied between 10 to 744 EMAR patients.

For the diagnosis of transition to psychosis (EMAR+) various criteria were used. SIPS/SOPS, CAARMS, DSM-IV, SCID-I, BPRS/CASH and CIE-10, alone or in combination.

Results of speech variables were obtained from 1) records and tasks transcription, open oral interviews, semi structured interviews or from induced free speeches, or 2) ranking of semi structured interviews. 60% ⁽⁹⁾ studies used this second methodology, as a language evaluation strategy, by analyzing the score of one specific item from CAARMS or SIPS/SOPS. CAARMS “Disorganized language” is an item aimed to state an approximate valuation of speech performance. In a very general manner, lexical-semantic adequacy of the answers is evaluated, according to the  interviewee skills for devising ideas and omit irrelevant information for the communication context ⁽¹⁴⁾. On is part, the item “Disorganized Communication” included in SIPS/SOPS provides a general evaluation of the verbal communication and speech coherence during the interview. It reviews difficulties for keeping communication objectives, observing adequacy level of linguistic structures used ⁽¹⁵⁾. Both items are focused on detecting FTD –related disturbances, such as tangentiality, content poorness, incoherence, among others. For full development of the items, both subjective appreciation questions made to the interviewee, as well as an objective evaluation made by the interviewer are required, thus resulting in a final score with a significant subjective component. This situation differentiates the studies that used these items as a Language Evaluation Method of the remaining 40% of studies, which performed a direct analysis of the transcribed answers from EMAR patients. Additionally, the strategy duration ranking items is undetermined, as the analysis is performed during the whole interview, and there is no standardized time for such. That is why we have decided to present the results in 2 sections, in the Table 2, studies using strategies for evaluating language by means of transcripted interviews (Section A) and studies using as a strategy the score analysis of an item of its diagnosis instrument for EMAR (Section B).

Duration of the strategies in Section A ranged from 8-minute structured tasks to narrative open interviews ~60 minutes.

Language Variables

Section A: Linguistic Cohesion, Semantic Coherence, Syntactic Complexity, Metaphoric Production, Emotional Prosody and FTD variables were analyzed. 
Section B: An item of CAARMS or SIPS/SOPS was used. Both items are FTD indicators and show Content Poorness, Loss of Associations, Tangentiality, Circumstantiality and/or Incoherence along the interview.

Prevalence of Language Disturbances

Section A: A study reported the FTD Prevalence levels in EMAR, according to the TLI (Thought and Language Index). It is an index aimed to evaluate 8 speech abnormalities, in response to the Rorschach´s Test or to pictures in the Thematic Perception Test. For the Subscale “TLI Positive/Disorganized” it was 75%; and for the “TLI Negative” Subscale it was 21% for scores ≥0.5 in a scale ranging from 0.25 (questionable disturbance) to 1 (clearly altered) according to its severity. Within the Subscale “TLI Positive/Disorganized” the items “Use of Peculiar Sentences” and “Use of Peculiar Words” had a Prevalence of 72% and 25% respectively, for scores ≥0.5 in severity.

Section B: Four studies reported FTD Prevalence in EMAR; 26.6% for scores ≥4 in severity or ≥3 in frequency of CAARMS and 29.7%, 37.1% and 44% for scores ≥ 3, in SIPS/SOPS severity.

Transition to Psychosis and Language Disturbances

For the statistical analysis of the correlation with the transition were applied various statistic models; logistic regression (with its results, such as OR Odds Ratio), Cox´s regression (with its results, such as HR Hazard Ratio) and convex envelope (convex hull, with its results, such as global accuracy percentage).

Section A: Two studies obtained mixed results. Some items were correlated with the transition and other were not. Among those not related, the full LTI score, the “TLI Negative” Subscale and the “TLI Positive/Disorganized” Subscale are reported. On the other hand, the items “Speech Poorness” and “Frequency of use of Determinants” were not a statistically significant transition predictor (p>0.05).
In all the studies of this section, at least, one transition predictor item was found. In a study, the frequency of the item “Illogical Thought” was a statistically significant transition predictor (p=0.023, OR=4.64; 95% IC=1.24-17.41). For the remaining studies, global accuracies were reported between 70% to 100% for the transition to psychosis. Including within the analysis, 2 combined items as predictors, up to a set of 5 items.
Section B: statistically significant transition predictors were found (p<0.05) for the FTD variable of the SIPS/SOPS and the CAARMS, with various results, according to the model of analysis and the cutting score, for the variable severity. When scores ≥3 and ≤5 were considered, an HR=8.9; 95% IC=2.49-42.41 was obtained. For scores ≥3 an HR=2.63; 95% IC=1.18-5.85 was obtained. Four studies analyzed the same variable, but with no minimum severity score, thus obtaining an OR=1.43;95% IC=1.04-1.97 and OR=2.1; 95% IC=1.03-4.27. Additionally, an HR=1.27; 95% IC=1.08-1.50 and HR=1.69; 95% IC=1-39-2.05.
Additionally, a study classified severity scores of the FTD variable in 4 paths, according to their evolution within 30 months of follow-up. The Path “Persistently High” of the variable was reported to be a statistically significant transition predictor (p<0.05, HR=2.23; 95% IC=1.01-4.93).

Severity of the item “Disorganized Communication /Conceptual Disorganization” was not a statistically significant transition predictor (p>0.05) for 2 studies. 

Discussion

---

Abstract

Use of AI to perform an automated language analysis in psychiatric pathologies is an innovating tool for universal clinical application, as may be seen with this systematic review. Results state that when searching for language anomalies in EMAR population, some variables are more specific for evaluating speech and others are less sharp. The analysis was made with various instruments or softwares and by using various methodologies. The language variables searched are mainly, Linguistic Cohesion, Semantic Coherence, Syntactic Complexity, Metaphoric Production, Emotional Prosody and FTD. According to the analysis, a Prevalence from 21% up to the 85% was reported for disturbances included within FTD. There are no prevalence measures for the remaining variables.

In studies using the score of an item for the research, speech disturbances were found or items which, by means of the retrospective analysis of the samples, significantly predicted transition to psychosis, by using logistic regression measures and Cox´s Regression. 

For those studies using transcriptions for their research, in less than 60 minutes these could obtain an amount of speech enough as to perform transition predictions with an accuracy ranging from 70% to 100%. Even more, some authors estimate that application and measurement of the CAARMS item “Disorganized language” (unspecific measurement) could be made in less than 10-15 minutes, while for applying full CAARMS nearly 120 minutes are required ⁽⁴⁷⁾. These characteristics could assign to language analysis a significant role in EMAR patients follow-up.

Limitations 

FTD analysis, based on the items “Disorganized language” and “Disorganized Communication” of EMARS diagnosis interviews have a significant subjective component, which makes them partly hard to use in statistical analysis. For instance, one of the questions included in CAARMS subjective evaluation is: “¿Do you use words that are not relevant, or are totally impertinent?” ⁽¹³⁾. It is important to highlight that the answer to this question mostly depends on the context, both the patient´s context and the interviewer´s context. In order to mitigate this effect, in the studies inter-interviewer variability scores are usually obtained, which allow to objectify criteria, in a certain manner, when interviews are ranked, as additionally, no accurate limits exist to determine each value within the scores (Example: 0-6 in CAARMS, 0.25-1 in TLI).

An EMAR subject may remain in this phase of under-threshold symptoms, to make his/her transition to psychosis or have remission from his/her condition ⁽⁴⁸⁾. It follows from this that clinical characteristics could evolve similarly; therefore, prevalence measurements of language disturbances will depend on the time the subject is evaluated by Health Professionals, on the follow-up duration; and on the frequency of controls. We could expect higher prevalence rates in studies with a longer duration. On the other hand, the statistical analysis performed by the studies, in order to determine the predictive power of a language variable regarding transition are diverse and consider various mathematic models, which can make general interpretation more complex.

Conclusions

There is a wide range of speech disturbances in EMAR patients’ speech. The most studied disturbance is Formal Thought Disorder (FTD). Prevalence of language disturbances in EMAR, especially regarding FTD, seem to be high, although it is highlighted when these are measured all together. Language disturbances analysis, specially made by means of interviews transcription and computing methods seems to be an excellent tool for predicting transition to psychosis.

It is necessary to keep developing instruments for specifically measuring language disturbances in EMAR population, and also to perform studies focused on language analysis in other foreign languages, other than English, as these offer a cost-effective alternative, are easy to obtain and provide promising results, plus having a great potential to be used in clinical internship.

References

---

1. Kuperberg GR. Language in schizophrenia part 1: an introduction. Lang Linguist Compass. 2010;4:576-589.
2. Docherty NM. Cognitive Impairments and Disordered Speech in Schizophrenia : Thought Disorder , Disorganization , and Communication Failure Perspectives.  J Abnorm Psychol. 2005;114:269–278.
3. de Boer JN, Brederoo SG, Voppel AE, Sommer IEC. Anomalies in language as a biomarker for schizophrenia. Curr Opin Psychiatry. 2020;33(3):212-218.
4. Argolo F, Magnavita G, Mota NB, Ziebold C, Mabunda D, Pan PM, et al. Lowering costs for large-scale screening in psychosis: a systematic review and meta-analysis of performance and value of information for speech-based psychiatric evaluation. Braz J Psychiatry. 2020;42(6):673-686.
5. Bleuler E. (Trans. J. Zinkin), Zinkin J. Dementia Praecox or the Group of Schizophrenias. International Universities Press; New York: 1950. 
6. Norman RM, Malla AK, Cortese L, Cheng S, Diaz K, McIntosh E, et al. Symptoms and cognition as predictors of community functioning: a prospective analysis. Am J Psychiatry 1999; 156: 400-5.
7. Andreasen NC. Thought, language, and communication disorders. I. Clinical assessment, definition of terms, and evaluation of their reliability. Arch Gen Psychiatry. 1979;36(12):1315-1321.
8. Kircher, T., Krug, A., Stratmann, M., Ghazi, S., Schales, C., Frauenheim, et al. (2014). A rating scale for the assessment of objective and subjective formal Thought and Language Disorder (TALD). Schizophr Res. 2014;160(1-3):216-221.
9. Andreasen NC. Scale for the assessment of thought, language, and communication (TLC). Schizophr Bull. 1986;12(3):473-482.
10. Corcoran CM, Mittal VA, Bearden CE, E Gur R, Hitczenko K, Bilgrami Z, et al. Language as a biomarker for psychosis: A natural language processing approach. Schizophr Res. 2020;1:S0920-9964(20)30247-4.
11. Gupta T, Hespos SJ, Horton WS, Mittal VA. Automated analysis of written narratives reveals abnormalities in referential cohesion in youth at ultra high risk for psychosis. Schizophr Res. 2018;192:82-88.
12. Yung AR, McGorry PD. The prodromal phase of first episode psychosis: past and current conceptualisations. Schizophrenia Bulletin. 1996; 22:353–370.
13. Van der Gaag M, Smit F, Bechdolf A, French P, Linszen DH, Yung AR, et al. Preventing a first episode of psychosis: meta-analysis of randomized controlled prevention trials of 12 month and longer-term follow-ups. Schizophr Res 2013; 149: 56–62.
14. Yung AR, Yuen HP, McGorry PD, Phillips LJ, Kelly D, Dell’Olio M, et al. Mapping the onset of psychosis: the Comprehensive Assessment of At-Risk Mental States. Aust N Z J Psychiatry. 2005;39(11–12):964–71.
15. Miller TJ, McGlashan TH, Rosen JL, Cadenhead K, Cannon T, Ventura J, et al. Prodromal assessment with the structured interview for prodromal syndro- mes and the scale of prodromal symptoms: predictive validity, interrater reliability, and training to reliability. Schizophr Bull 2003;29(4):703–15.
16. Fusar-Poli P, Cappucciati M, Rutigliano G, Schultze-Lutter F, Bonoldi I, Borg- wardt S, et al. At risk or not at risk? A meta-analysis of the prognostic accuracy of psychometric interviews for psychosis prediction. World Psychiatry 2015;14(3):322–32.
17. Fusar-Poli P, Bonoldi I, Yung AR, Borgwardt S, Kempton MJ, Valmaggia L, et al. Predicting Psychosis: Meta-analysis of Transition Outcomes in Individuals at High Clinical Risk. Arch Gen Psychiatry. 2012;69(3):220–229.
18. Goldsmith DR, Crooks CL, Walker EF, Cotes RO. An Update on Promising Biomarkers in Schizophrenia. Focus (Am Psychiatr Publ). 2018;16(2):153-163.
19. Bearden CE, Wu KN, Caplan R, Cannon TD. Thought disorder and communication deviance as predictors of outcome in youth at clinical high risk for psychosis. J Am Acad Child Adolesc Psychiatry. 2011;50(7):669-680.
20. Caplan R, Guthrie D, Fish B, Tanguay PE, David-Lando G. The Kiddie Formal Thought Disorder Rating Scale: clinical assess- ment, reliability, and validity. J Am Acad Child Adolesc Psychiatry. 1989;28:408-416.
21. Halliday MAK, Hassan R. Cohesion in Spoken and Written English. London: Longman’s; 1976.
22. Caplan R, Guthrie D, Foy JG. Communication deficits and formal thought disorder in schizophrenic children. J Am Acad Child Adolesc Psychiatry. 1992;31:151-159.
23. Bedi, G., Carrillo, F., Cecchi, G. Slezak DF, Sigman M, Mota NB, et al. Automated analysis of free speech predicts psychosis onset in high-risk youths. NPJ Schizophr. 2015;1:15030.
24. Landauer TK, Dumais ST. A solution to Plato's problem: the latent semantic analysis theory of acquisition, induction, and representation of knowledge. Psychol Rev 1997;104:211–240.
25. Natural Language Toolkit. NLTK, disponible en http://www.nltk.org/
26. Bird S, Klein E, Loper E. Natural Language Processing with Python. O'Reilly Media: Sebastopol, CA, USA, 2009.
27. Demjaha A, Weinstein S, Stahl D, Day F, Valmaggia L, Rutigliano G, et al. Formal thought disorder in people at ultra-high risk of psychosis. BJPsych Open. 2017;3(4):165-170.
28. Murray HA. Thematic Apperception Test. Harvard University Press, 1943.
29. Liddle PF, Ngan ET, Caissie SL, Anderson CM, Bates AT, Quested DJ, et al. Thought and Language Index: an instrument for assessing thought and language in schizophrenia. Br J Psychiatry. 2002;181:326–30.
30. Gutierrez, E. D., Cecchi, G. A., Corcoran, C., & Corlett, P. Using automated metaphor identification to aid in detection and prediction of first-episode schizophrenia. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing. 2017. p. 2923-2930.
31. Dinh ED, Gurevych I. Token-level metaphor detection using neural networks. In Proceedings of the Fourth Workshop on Metaphor in NLP. 2016. p. 28-33.
32. Socher R, Huval B, Manning CD, Ng AY. 2012. Semantic compositionality through recursive matrix-vector spaces. In Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning. 2012. p. 1201– 1211.
33. Corcoran C. M., Carrillo F., Fernández‐Slezak D., Bedi G., Klim C., Javitt D. C. et al. Prediction of psychosis across protocols and risk cohorts using automated language analysis. World Psychiatry. 2018;17(1):67-75.
34. Rezaii N, Walker E, Wolff P. A machine learning approach to predicting psychosis using semantic density and latent content analysis. NPJ Schizophr. 2019;5(1):9.
35. Addington J, Liu L, Buchy L, Cadenhead, K. S., Cannon, T. D., Cornblatt, B. A, et al. North American Prodrome Longitudinal Study (NAPLS 2): The Prodromal Symptoms. J Nerv Ment Dis. 2015;203(5):328-335.
36. Addington J, Liu L, Brummitt K, Bearden, C. E., Cadenhead, K. S., Cornblatt, B. A. et al. North American Prodrome Longitudinal Study (NAPLS 3): Methods and baseline description. Schizophr Res. 2020;S0920-9964(20)30217-6.
37. Klein D, Manning CD. Accurate unlexicalized parsing. In Proceedings of the 41st annual meeting of the association for computational linguistics. 2003. p. 423-430.
38. Computerized Propositional Idea Density Rater. Disponible online en: www.covingtoninnovations.com/software.html
39. Cannon TD, Cadenhead K, Cornblatt B, Woods S. W., Addington J., Walker E. et al. Prediction of psychosis in youth at high clinical risk: a multisite longitudinal study in North America. Arch Gen Psychiatry. 2008;65(1):28-37.
40. Nelson B, Yuen HP, Wood SJ, Lin A., Spiliotacopoulos D., Bruxner A, et al. Long-term follow-up of a group at ultra high risk ("prodromal") for psychosis: the PACE 400 study. JAMA Psychiatry. 2013;70(8):793-802.
41. Tarbox SI, Addington J, Cadenhead KS, Cannon T. D., Cornblatt B. A., Perkins D. O, et al. Premorbid functional development and conversion to psychosis in clinical high-risk youths. Dev Psychopathol. 2013;25(4 Pt 1):1171-1186.
42. Addington J, Cadenhead KS, Cannon TD, Cornblatt B., McGlashan T. H., Perkins D. O. et al. North American Prodrome Longitudinal Study: a collaborative multisite approach to prodromal schizophrenia research. Schizophr Bull. 2007;33(3):665-672.
43. DeVylder JE, Muchomba FM, Gill KE, Ben-David S., Walder D. J., Malaspina D. et al. Symptom trajectories and psychosis onset in a clinical high-risk cohort: the relevance of subthreshold thought disorder. Schizophr Res. 2014;159(2-3):278-283.
44. Cornblatt BA, Carrion RE, Auther A, McLaughlin, D., Olsen R. H., John M, et al. Psychosis prevention: a modified clinical high risk perspective from the recognition and prevention (RAP) program. Am J Psychiatry. 2015;172:986–994.
45. Cornblatt BA. The New York high risk project to the Hillside recognition and prevention (RAP) program. Am J Med Genet. 2002;114(8):956-966.
46. Armando M, Pontillo M, De Crescenzo F, Mazzone L., Monducci E., Cascio N. L, et al. Twelve-month psychosis-predictive value of the ultra-high risk criteria in children and adolescents. Schizophr Res. 2015;169(1-3):186-192.
47. Rekhi G, Rapisarda A, Lee J. Impact of distress related to attenuated psychotic symptoms in individuals at ultra high risk of psychosis: Findings from the Longitudinal Youth at Risk Study. Early Interv Psychiatry. 2019;13(1):73-78.
48. Kotlicka-Antczak M, Karbownik MS, Stawiski K, Pawełczyk A., Żurner N., Pawełczyk T., et al. Short clinically-based prediction model to forecast transition to psychosis in individuals at clinical high risk state. Eur Psychiatry. 2019;58:72-79.

Annexes

---