A 20-year review comparing the use of ‘schizophrenia’ and ‘psychosis’ in UK newspapers from 2000 to 2019: Implications for stigma reduction

BACKGROUND
There have been increasing calls for schizophrenia to be renamed to reduce negative associations with the diagnosis. 'Psychosis' is one proposed alternative, yet some research suggests this has greater negative connotations. This study compared the use of both terms within UK newspapers longitudinally to assess which is more negatively portrayed in the media.


STUDY DESIGN
Using LexisLibrary News, six UK newspaper archives were searched for the terms 'schizophrenia' and 'psychosis' in articles published between 2000 and 2019; the included article total was n = 9802. Articles were analysed using Linguistic Inquiry and Word Count software.


STUDY RESULTS
A two-level multilevel model was created of newspaper articles nested within newspapers. Adding predictors revealed that diagnostic term was a significant predictor of the emotional tone of articles (β = -2.30, SE = 0.51, p < .001, 95 % CI -3.30 to -1.30), with 'psychosis' having a more negative tone than 'schizophrenia.' Tabloids were shown to have a significantly more negative tone than broadsheets (β = -9.32, SE = 1.04, p < .001, 95 % CI -11.36 to -7.28), and emotional tone of writing had become more negative over time (β = -0.26, SE = 0.04, p < .001, 95 % CI -0.34 to -0.18).


CONCLUSIONS
These findings suggest 'psychosis' is associated with greater negativity than 'schizophrenia' within UK newspapers, and may be a poor substitute term to reduce stigma. As article tone worsened for both diagnoses over time, increased stigma reduction campaigns are required to target negativity within the media.


Introduction
Schizophrenia is commonly described as one of the most debilitating mental health disorders (Ajinkya et al., 2015;Chaudhury et al., 2006) and it is frequently stigmatised by the public (World Health Organization, 2019), being associated with perceptions of dangerousness and unpredictability (Angermeyer and Matschinger, 2003;Levey and Howells, 1995;Wood et al., 2014). Moreover, when compared to other diagnoses such as depression, schizophrenia is more stigmatised (Norman et al., 2012). The stigma surrounding schizophrenia leads to fears of disclosure, with individuals who have a diagnosis of schizophrenia expressing awareness of the negative connotations surrounding their diagnosis (Dickerson et al., 2002;Howe et al., 2014;Knight et al., 2003). A wide body of literature has discussed the high economic costs and broad personal and social implications of stigma for people diagnosed with schizophrenia (e.g. Royal Australian and New Zealand College of Psychiatrists, 2016); with fears of disclosure and discrimination linked to numerous adverse impacts such as worsening of symptoms and reduced sense of recovery (Vass et al., 2015), diminished self-esteem and self-efficacy (Corrigan and Rao, 2012), and reduced engagement in life opportunities (Thornicroft et al., 2007).
The stigma surrounding schizophrenia has been influenced by mass media, with studies suggesting that the media portrayal of mental health disorders informs public attitudes (Wahl, 1992). Research has found that recalling past media reports about those with mental health problems harming another individual influenced strength of agreement that a person labelled with schizophrenia was dangerous (Reavley et al., 2016). When looking at specific newspaper portrayals, positively toned articles such as informative pieces, articles discussing recovery, or those which challenge stigma for example, have been associated with reduced stigmatising attitudes (Ross et al., 2019). In contrast, negatively toned articles such as those which discuss dangerous, criminal, or violent behaviour, or perpetuate common stigmatising attitudes, have been associated with increased stigma and stereotyping (Ross et al., 2019). With newspapers now being accessible online and reaching a much larger audience, the relationship between negative newspaper articles and stigma is a growing concern, with both historical and recent reviews of newspaper reporting on schizophrenia demonstrating common associations made between schizophrenia, violence, and controversy (Gwarjanski and Parrott, 2018;Wahl, 1996). This pattern is also found in newspaper reviews cross-culturally, with reviews in Brazil, France, Japan, and Portugal finding schizophrenia is associated with danger, crime, and negative metaphorical use in the media (Dubugras et al., 2011;Lampropoulos et al., 2017;Ottewell, 2017;Rodrigues-Silva et al., 2017).
When considering the UK's newspaper coverage of schizophrenia, it is similarly negative, with frequent associations with violence (Bowen et al., 2019), and incorrect use of the term as a negative metaphor (Chopra and Doody, 2007). This varies by newspaper format, with research suggesting that tabloids associate schizophrenia with greater violence and use more stigmatising descriptors compared to broadsheets (Clement and Foster, 2008). Despite increasing guidelines aiming to improve reporting on mental health in the media (Mind, n.d.;National Union of Journalists, 2014;Time to Change, n.d.), UK newspaper coverage of schizophrenia has not improved over time, with the proportion of negative news stories about schizophrenia remaining the same between 1992 and 2008 (Goulden et al., 2011). Moreover, research suggests that metaphorical use of the term 'schizophrenia' increased in tabloids over time, and overall use of stigmatising descriptors did not improve between 1996 and 2005 (Clement and Foster, 2008). More recently, Hildersley et al. (2020) explored UK newspaper coverage between 2008 and 2019 finding that articles about schizophrenia were more likely to be stigmatising compared to those on other diagnoses, with a recent increase in the number of stigmatising articles about schizophrenia, but a decrease for other diagnoses. In the same time period, Li et al. (2021) found that articles in tabloids and articles about severe mental illness were more likely to be stigmatising, compared to articles in broadsheets and articles about common mental disorders. However, they also found that overall the number of stigmatising articles in both tabloids and broadsheets is decreasing over time.
The negative associations and stigma attached to schizophrenia has led to increasing demands for the label to be replaced, with hopes this will reduce discrimination and improve understanding (Brabban et al., 2013;Guloksuz and van Os, 2019;Kingdon et al., 2007;Lasalvia et al., 2015;Lasalvia et al., 2021;Lasalvia and Tansella, 2013). This is not a new idea; in 2002 Japan changed the label of schizophrenia from Seishin Bunretsu Byo ('mind-split-disease'), to the term Togo Shitcho Sho ('integration disorder') in an effort to distance the diagnosis from the pervasive societal stigma attached to it (Sato, 2006). South Korea followed suit and also dropped the term in 2012 (Lee et al., 2013).
In the West, key stakeholders, including service-users and their relations, health care professionals, and academics, have proposed that the term 'schizophrenia' should be replaced with the term 'psychosis' or associated terminology, such as 'psychosis spectrum/susceptibility syndrome' (George and Klijn, 2013;Guloksuz and van Os, 2018;Henderson and Malhi, 2014;Lasalvia et al., 2021;Mesholam-Gately et al., 2021;van Os, 2016). This replacement is already appearing within psychological research, with the journals Schizophrenia Bulletin and Schizophrenia Research adding subtitles to their later volumes in an effort to include the term 'psychosis' (Carpenter, 2016;Keshavan et al., 2017).
Replacing 'schizophrenia' with 'psychosis' to reduce stigma belies the assumption that the term 'psychosis' is less stigmatised than 'schizophrenia', which may not be true. Indeed, Bentall (2013) proposes a replacement will be met with equal stigma, and suggests increased education around schizophrenia may be a viable solution. When exploring the stigma attached to the term 'psychosis', there is limited research, but the studies available highlight poor public understanding, high levels of experienced stigma, and perceived negativity attached to the term (Burke et al., 2016;Chan et al., 2016;Wood et al., 2018;Wood and Irons, 2017), with some suggestion that the amount of stigma towards psychosis is equal to that of schizophrenia (Yang et al., 2013). These negative associations towards psychosis are also present within the media, with commonplace incorrect and derogatory use on social media (Alvarez-Mon et al., 2019), and associations with danger and overall negativity found within newspapers (Delahunt-Smoleniec and Smith-Merry, 2020).
As stigma and negativity have been associated with both diagnoses (Angermeyer and Matschinger, 2003;Burke et al., 2016), it is surprising that there is limited research comparing the stigma attached to the two terms to justify the proposed move from 'schizophrenia' to 'psychosis'. Existing studies have mixed findings; Magliano and Marassi (2018) concluded that 'schizophrenia' is more negatively portrayed than 'psychosis' in Italian newspapers, lending support to the replacement of 'schizophrenia' with 'psychosis' to reduce negativity. In contrast, Passerello et al. (2019) compared the use of the two terms on Twitter and found that 'psychosis' is associated with greater stigma and negativity compared to 'schizophrenia'.
Due to this conflicting evidence, and as the notion of replacing 'schizophrenia' with 'psychosis' is gaining popularity; this study will add to the debate by exploring whether the term 'psychosis' is associated with less negativity than the term 'schizophrenia', utilising newspapers due to their broad reach, and as negativity within them has been suggested to worsen stigmatising attitudes (Ross et al., 2019).
Given the existing findings in the literature, this study has four key aims: 1. The study aimed to assess whether there is a difference between the emotional tone of articles containing the term 'schizophrenia' and articles containing the term 'psychosis'. 2. The study aimed to explore whether there would be a difference in emotional tone depending on newspaper format. Specifically, we hypothesised that tabloids will have a more negative emotional tone of writing than broadsheets. 3. The study also aimed to explore whether there are changes in emotional tone of writing over time, and if this differs for the terms 'schizophrenia' and 'psychosis'. 4. Finally, the study aimed to assess whether the change in tone over time differs by newspaper format. Specifically, we hypothesised that the tone in tabloids will be more negative over time, compared to broadsheets.
We were unable to predict directional hypotheses for all aims, due to the mixed evidence in the existing literature (e.g. on whether schizophrenia or psychosis is associated with more negativity). Where there is consistent evidence in one direction, we have included our prediction on what we expect to find.

Search strategy
Consistent with figures from November 2019, six UK newspapers with the largest circulation (Tobitt, 2019) were included within the archive search, which utilised the LexisLibrary News database. The newspapers sampled were the tabloids: The Sun, Daily Mirror, Daily Mail; and the broadsheets: The Telegraph, The Times, and The Independent. We selected an equal number of each format as they vary in style. Specifically, tabloids are typically smaller, more sensationalistic, and focus on less serious new stories, whereas broadsheets are larger, feature more serious and in-depth reporting, and have a higher news content. Each newspaper's archives were searched for the use of the term 'schizophrenia', then 'psychosis, from November 1, 2000, to November 1, 2019. Where Sunday editions featured as a separate newspaper these were also searched and combined with the daily edition.
Before being included within the sample, articles were checked against the study's exclusion criteria (see Fig. 1). We excluded any duplicate articles. Specifically, where two articles within the same newspaper were exact replicas of each other, or extremely similar, only one copy was retained. Articles with extremely high similarity were assessed to determine which would be retained. If most of the content was identical only one version was retained, with this being the article with the highest word count or the fewest errors (e.g. typographical errors, spelling mistakes). If there were notable differences between the two articles, both were retained. We then excluded any of the remaining articles which contained both 'schizophrenia' and 'psychosis', as they did not allow comparison of the terms.

Quality assessment
The articles were checked to ensure they met the content analysis requirements. Three changes were made to ensure precision in the analysis (Pennebaker et al., 2015): 1. Internet notations were removed (e.g., URLs, email addresses, hashtags, and Twitter handles). 2. Authors' names were removed to ensure that only the headline and main body of the text were analysed. 3. Article errors, typographical errors, and spelling mistakes were corrected where possible (e.g., missing spaces: fortunebehind → fortune behind).

Pre-processing
After the final sample was obtained (see Fig. 2), the articles were uploaded to the Linguistic Inquiry and Word Count software (LIWC) Version 1.6.0 (Pennebaker Conglomerates, Inc) which analysed the content of the uploaded article against its internal dictionary, creating percentages of words within the article which reflect the chosen category. Tone was selected as the category, which is a 0-100 point scale giving a value to the emotional tone of writing in an uploaded article. A score of 50 indicates neutrality whereby a topic is discussed neutrally. Scores below 50 indicates a low, or negative tone (e.g., more anxiety, hostility, or sadness), and above 50 indicates a higher or positive tone (e. g., an optimistic tone of writing; Pennebaker et al., 2015). The tone measures more than just specific word usage, capturing the overall mood of the piece of writing as a whole. LIWC is a lab validated tool, which has been used in over 20,000 published scientific articles (Linguistic Inquiry and Word Count, n.d.). Specifically, the 'emotional tone' category has been used for various research themes, such as linking writing tone with depression and suicidal ideation (Lumontod, 2020). Research has suggested the LIWC achieves acceptable levels of reliability and validity when measuring emotion related words (Landoni et al., 2022), and that it is sensitive in detecting emotional expression in text (McDonnell et al., 2020).

Statistical analysis
Analyses were conducted using IBM SPSS Statistics for Windows, Version 25 and MLwiN Version 3.04 (Centre for Multilevel Modelling, University of Bristol).
To address the research hypotheses multilevel modelling was utilised. The dependent variable for all analyses was the emotional tone of writing in the article, with lower tones indicating greater negativity on the topic. To address each aim, several models were created. For each model, model fit to the data was assessed by comparing the difference between the new and previous model's 2*loglikelihood value, against a Chi-Squared distribution. A two-level model (articles nested within newspapers) was tested against a null model (mean of articles only) to examine superiority. Explanatory variables were then added individually to address each hypothesis. The hypotheses required the level 1 explanatory variables of diagnostic term (created to show which term is featured in the article, coding of 0 = 'schizophrenia' or 1 = 'psychosis') and year of article (continuous variable created to reflect year of article, with 0-19 coding to represent 2000-2019), the level 2 explanatory variable of newspaper format (created to show which newspaper type the article came from, coding of 0 = 'broadsheet' or 1 = 'tabloid'), and same-level and cross-level interactions between these variables. Where a predictor was significant in a previous model, it was retained in subsequent models, with the significance of a predictor being determined by its z-score (β/SE).

Multilevel modelling
For all models, the dependent variable was emotional tone. To ensure a two-level model was required a single-level and two-level null model were fitted and compared, to determine which was a better fit to the data. The first model was a single-level model, with a mean article tone of β 0 = 25.96 (SE = 0.25). The second model was a two-level model, with articles nested within newspapers (β 0j = 24.42, SE = 1.94). By comparing the 2*loglikelihood values, it was confirmed that the twolevel model was a better fit to the data than the single-level model (χ 2 (1) = 330, p < .001). When looking at the unexplained variance, most was within newspapers at the article level (96.33 %), with the intra-class correlation coefficient (ICC) confirming only 3.67 % of the unexplained variance was between newspapers (r = 0.04).
For mean article tone for each newspaper and diagnostic term, see Table 1.

Emotional tone of articles containing 'schizophrenia' versus 'psychosis'
We aimed to explore whether there will be a difference between the emotional tone of articles containing the term 'schizophrenia' and articles containing the term 'psychosis'. To address this, diagnostic term was added to the two-level null model. This model was a better fit to the data than the null model (χ 2 (1) = 21, p < .001). Diagnostic term was a significant predictor of tone (β = − 2.30, SE = 0.51, z = − 4.51, p < .001, 95 % CI − 3.30 to − 1.30), with the term 'psychosis' having a lower (more negative) average tone than the term 'schizophrenia'. The addition of this predictor reduced the article-level variance, with diagnostic term explaining 0.12 % of the total variance from the null model. However, the ICC remained the same (r = 0.04). For full results see Table 2, Model 1.

Emotional tone of tabloids versus broadsheets
We hypothesised tabloids will have a more negative emotional tone of writing than broadsheets. To test this, a binary level 2 explanatory variable of newspaper format was added to Model 1. The addition of this variable to the diagnostic term-only model significantly improved the model fit to the data (χ 2 (1) = 16, p < .001). Newspaper format was a significant predictor of tone (β = − 9.32, SE = 1.04, z = − 8.96, p < .001, 95 % CI − 11.36 to − 7.28). This indicates that while keeping diagnostic term constant, tabloids had a lower (more negative) average tone than broadsheets. The addition of this predictor greatly reduced the betweennewspaper variance partition from 3.76 % to 0.21 % (r = 0.002). Moreover, adding newspaper format explained 3.56 % of the total variance from Model 1. For full results see Table 2, Model 2.

Change in emotional tone over time for 'schizophrenia' and 'psychosis'
We aimed to assess whether there will be a change in emotional tone of writing over time, and if this differs for the terms 'schizophrenia' and 'psychosis'. To test the first element of this, year was added as an articlelevel predictor to the previous model, creating Model 3a (see Table 2). The addition of this predictor to Model 2 significantly improved model fit to the data (χ 2 (1) = 35, p < .001). Year was a significant predictor of tone (β = − 0.26, SE = 0.04, z = − 6.50, p < .001, 95 % CI − 0.34 to − 0.18). This indicates that a one unit increase in year was associated with a 0.26-point decrease in article tone, while keeping diagnostic term and newspaper format constant. Both the article-level and newspaperlevel variance decreased through this addition, with the between- Table 1 Descriptive statistics comparing articles' emotional tone for different newspaper formats, newspapers, and diagnostic terms (mean (SD)). To address the second element, an interaction term was added to Model 3a. The interaction was a single-level interaction at the article level, between year and diagnostic term. Adding the interaction to Model 3a did not improve model fit to the data (χ 2 (1) = 2, p = .157). Additionally, it was a nonsignificant predictor of tone based upon our two-tailed hypothesis (β = 0.15, SE = 0.09, z = 1.67, p = .095, 95 % CI − 0.03 to 0.33), with the interaction showing that the terms 'schizophrenia' and 'psychosis' both decreased in tone over time. For full results see Table 2, Model 3b.
As this interaction was nonsignificant and did not improve model fit, it was not included within future models.

Changes in emotional tone over time for tabloids and broadsheets
Finally, we predicted the tone in tabloids will be more negative over time, compared to broadsheets. To address this, a cross-level interaction term was added between year and newspaper format. The addition of the interaction to Model 3a significantly improved model fit to the data (χ 2 (1) = 10, p = .002). This interaction was a significant predictor of tone (β = 0.29, SE = 0.09, z = 3.22, p = .001, 95 % CI 0.11 to 0.47).
Interpreting the interaction showed that broadsheets have a steeper decrease in tone over time compared to tabloids (see Fig. 3). The articlelevel variance decreased through the addition of this interaction, with the between-newspaper variance partition increasing slightly from 0.18 % to 0.19 % (r = 0.002). Adding this interaction term to the model explained 0.10 % of the total variance from Model 3a. For full results, see Table 2, Model 4.  Fig. 3. Trend in mean tone over time for tabloids and broadsheets. The interaction term year by newspaper format is significant (p = .001).

Main findings
Increasing calls to rename schizophrenia (e.g. Guloksuz and van Os, 2019;Lasalvia et al., 2021) led the current study to assess whether replacement with 'psychosis' will engender the desired reductions in stigma and negative associations, through longitudinal analysis of UK newspapers. Exploring our primary aim, we found a difference between the emotional tone of articles containing 'schizophrenia' versus 'psychosis'. Specifically, it was found that 'psychosis' was utilised with a slightly more negative tone than 'schizophrenia' in newspaper reporting. This finding is consistent with Passerello et al.'s (2019) findings when comparing their use on Twitter, but contrary to Magliano et al.'s (2018) findings that 'schizophrenia' was more negatively portrayed than 'psychosis' in Italian newspapers.
Our additional aims assessed whether our results support previous findings from reviews of the UK's newspaper portrayal of schizophrenia. We predicted that tabloids would have a more negative emotional tone of writing than broadsheets, and this was found in all models within the analysis. This is consistent with previous research exploring the UK newspaper portrayal of mental health disorders, with greater violence, bad news stories, and stigma found in tabloids compared to broadsheets (Clement and Foster, 2008;Goulden et al., 2011;Li et al., 2021).
Additionally, we explored if there was a change in emotional tone of writing over time, and this rate of change differed for 'schizophrenia' and 'psychosis'. It was found that there was greater negativity as years increase. However, a difference was not found in the rate of change, with both terms being discussed with a more negative tone over time. Previous UK newspaper reviews of schizophrenia indicated minimal changes in the negative coverage over time (Clement and Foster, 2008;Goulden et al., 2011), or possibly even reductions in the number of stigmatising articles about severe mental illness over time (Li et al., 2021). This is inconsistent with our finding of worsened tone over time for the term 'schizophrenia', but the difference may be accounted for by the variations in coding methods (e.g. coding as bad news vs. good news). Further research has suggested that stigmatising articles about schizophrenia have increased in recent years (Hildersley et al., 2020), which is more consistent with our finding that there is an increased negative tone in articles containing 'schizophrenia' over time. Our novel exploration of the portrayal of 'psychosis' revealed this is also worsening over time, which is a concerning finding. This demonstrates that over the 20 years studied, the media's reporting on the terms 'schizophrenia' and 'psychosis' has worsened over time, which is likely to have a resultant negative influence on public stigma (Ross et al., 2019).
Furthermore, we predicted the tone in tabloids will be more negative over time, compared to broadsheets, but it was found that broadsheets had a steeper decline in tone over time compared to tabloids. Our finding that tabloids have a more negative tone over time aligns with previous research, which suggests there has been increasing negative metaphorical use of the term 'schizophrenia' in tabloids over time (Clement and Foster, 2008). However, our finding that broadsheets had a steeper decline in tone compared to tabloids is inconsistent with previous reviews. One possible explanation for this may be that tabloids had a more negative tone to begin with, resulting in a less steep decline comparatively.
Our results indicated that while 'psychosis' had slightly greater negativity associated, 'schizophrenia' was also presented negatively within UK newspapers. This is concerning as negative media portrayals inform public attitudes and can influence stigmatisation (Ross et al., 2019;Wahl, 1992). However, the finding supports a strong body of research suggesting there are violent and negative associations towards schizophrenia in the media (Bowen et al., 2019;Gwarjanski and Parrott, 2018) and that schizophrenia is highly stigmatised (Angermeyer and Matschinger, 2003;World Health Organization, 2019). Therefore, our findings do not oppose the suggestion that a name change from the term 'schizophrenia' may be required (Brabban et al., 2013;Guloksuz and van Os, 2019;Kingdon et al., 2007;Lasalvia et al., 2015;Lasalvia et al., 2021;Lasalvia and Tansella, 2013), as we found supporting evidence that the term is associated with negativity.
However, our findings do dispute the suggestion that replacement with the term 'psychosis' would result in reduced negativity (George and Klijn, 2013;Magliano and Marassi, 2018). Previous research has shown that psychosis already has negativity and stigma associated to it (e.g. Wood et al., 2018) which is consistent with our finding that 'psychosis' was utilised slightly more negatively in articles than 'schizophrenia'. This suggests replacement with 'psychosis' may be met with greater negativity, supporting the fears of those who are apprehensive about this name change (Bentall, 2013;Sommer and Carpenter, 2016). Parallels can be drawn with Hong Kong, where 'schizophrenia' was replaced with 'Si Jue Shi Tiao' a descriptive term incorporating the positive symptoms of psychosis, and dysregulation, with the hopes that this would be less stigmatising than the traditional descriptions of "mind-split disease" (Chan et al., 2017). Yet, while there was a good acceptance of the new term and a reduction of use of the old term, and the new term maintained comparative positivity and neutrality, media portrayal of the new term was still negative after the change (Chan et al., 2017).
Alternatively, where a novel alternative term has been used to replace 'schizophrenia', it has not always achieved the desired attitude changes. For example, in South Korea the term 'attunement disorder' replaced 'schizophrenia', to reduce stigma caused by the translation of the diagnosis, and create positive associations with recovery (Cho et al., 2018;Lee et al., 2014). It is suggested that this new label reduced discrimination and stigma (Kim et al., 2012), but public awareness of the name change remained low (Cho et al., 2018). Japan's replacement term 'integration disorder' has resulted in lower associations with criminality (Takahashi et al., 2009). However, despite some improvements in the media's reporting since the rename, there is still negativity and violence associated with the new term (Aoki et al., 2016;Koike et al., 2016). These examples demonstrate that even novel replacement labels, without preconceptions, may not achieve the desired impact of reducing stigma, which calls into question the effectiveness of replacing the term 'schizophrenia'.
Therefore, although our results support the idea of replacing the term 'schizophrenia' to reduce negativity and stigma, the UK newspaper portrayal of the term 'psychosis' suggests that it may not be a suitable replacement term to achieve this aim. Instead, this study proposes that if an alternative replacement term is desired, it must be carefully devised, with considerations needed on stigma already associated with suggested replacement labels. Moreover, it is crucial that research continues to consider the opinions of individuals with a diagnosis of schizophrenia, health care professionals, and the public (Lasalvia et al., 2021;Mesholam-Gately et al., 2021), to ensure any new terminology will be met with support and understanding. It must also be considered whether a name change will ever truly be effective at reducing stigma. As research from countries who have replaced the term 'schizophrenia' has shown, there is still a negative media portrayal and a lack of awareness towards replacement terms (Aoki et al., 2016;Cho et al., 2018;Koike et al., 2016). Therefore, it is vital that before implementing a replacement label, insight is gained from the methods used by these countries, and proposals consider how to widen public and media support towards new terminology, such as through greater activism, contact, and education (Bentall, 2013;. Moreover, as this study has shown the terms 'schizophrenia' and 'psychosis' are both negatively portrayed and worsening over time, the study has implications for UK journalists. The emotional tone of articles which feature these diagnostic labels must improve to prevent stigmatising attitudes increasing (Ross et al., 2019). Although campaigns have been attempted previously (e.g. Time to Change, n.d.), the increasingly negative newspaper reporting on these terms show that further stigma reduction campaigns are required. Like Hildersley et al. (2020); Li et al. (2021), we suggest that these should be tailored to the newspaper format, and should be specific for each diagnosis. We suggest these campaigns should feature more heavily in guidelines, training, and education for UK journalists, to increase ethical reporting and languageuse, and ensure reduced negativity when reporting on these diagnoses, and any future replacement term.

Limitations
Despite the addition of our explanatory variables, the article-level variation did not largely reduce. The residual unexplained variation was likely due to the articles within each newspaper being unique in content and topics. Our hypotheses did not explore specific subcategories of articles (e.g., crime). However, if future research selects to explore a specific subcategory, it would likely result in less unexplained article-level variation. Additionally, the study used a time-period which resulted in incomplete years for 2000 and 2019. Although obtaining complete records for these years would have ensured greater accuracy, we did not have access to all archives before November 2000 and after November 2019 at the time of data extraction. Finally, although there were benefits in using LIWC to determine article tone, the software does not have a category for violence-related language which would have been interesting to explore in relation to previous research findings. The development of a violence-related language dictionary within the LIWC may provide an avenue for further research into specific stigmatising attitudes around schizophrenia and psychosis.

Conclusions
In summary, this study has demonstrated that UK newspapers use a negative tone when reporting on the terms 'schizophrenia' and 'psychosis', and that this has worsened over time, despite increasing pressures for better reporting on mental health disorders. This is concerning, as negativity within the media has been associated with increased public stigma. This suggests a keen need to increase ethical reporting and language-use when reporting on these diagnoses. The negative portrayal of the term 'schizophrenia' provides support for proponents of renaming the diagnosis, to try to reduce negative connotations. However, as articles containing 'psychosis' were shown to be slightly more negatively toned than articles containing 'schizophrenia', a replacement with the term 'psychosis' is unlikely to be successful at reducing negative associations. Instead any alternative replacement term proposed should be carefully considered, with research into pre-existing stigma and the opinions of those it impacts being vital to conduct before any proposed replacement to terminology can be considered. However, without increased contact, increased education around the name change terminology, and targeted stigma reduction campaigns for both the public and the media, it is highly possible renaming 'schizophrenia' may not be an effective way to reduce stigma in the long-term, and any replacement term may still be associated with stigma and a negative media portrayal.

Role of the funding source
This work did not have a funding source.

Declaration of competing interest
None.