John Done

The Hospital Anxiety and Depression Scale: A meta confirmatory factor analysis

OBJECTIVE To systematically evaluate the latent structure of the Hospital Anxiety and Depression Scale (HADS) through reanalysis of previous studies and meta confirmatory factor analysis (CFA). METHOD Data from 28 samples were obtained from published studies concerning the latent structure of the HADS. Ten models were considered, including eight previously identified models and two bifactor models. The fit of each model was assessed separately in each sample and by meta CFA. Meta CFA was conducted using all samples and using subgroups consisting of community samples, cardiovascular disease samples and samples from studies administering the English language version of the HADS. RESULTS A bifactor model including all items loading onto a general distress factor and two orthogonal anxiety and depression group factors provided the best fit for the majority of samples. Meta CFA provided further support for the bifactor model with two group factors. This was the case using all samples, as well as all subgroup analyses. The general distress factor explained 73% of the covariance between items, with the (autonomic) anxiety and (anhedonic) depression factors explaining 11% and 16%, respectively. CONCLUSION A bifactor structure provides the most acceptable empirical explanation for the HADS correlation structure. Due to the presence of a strong general factor, the HADS does not provide good separation between symptoms of anxiety and depression. We recommend it is best used as a measure of general distress.

A simple solution for model comparison in bold imaging: the special case of reward prediction error and reward outcomes

Conventional neuroimaging techniques provide information about condition-related changes of the BOLD (blood-oxygen-level dependent) signal, indicating only where and when the underlying cognitive processes occur. Recently, with the help of a new approach called “model-based” functional neuroimaging (fMRI), researchers are able to visualize changes in the internal variables of a time varying learning process, such as the reward prediction error or the predicted reward value of a conditional stimulus. However, despite being extremely beneficial to the imaging community in understanding the neural correlates of decision variables, a model-based approach to brain imaging data is also methodologically challenging due to the multicollinearity problem in statistical analysis. There are multiple sources of multicollinearity in functional neuroimaging including investigations of closely related variables and/or experimental designs that do not account for this. The source of multicollinearity discussed in this paper occurs due to correlation between different subjective variables that are calculated very close in time. Here, we review methodological approaches to analyzing such data by discussing the special case of separating the reward prediction error signal from reward outcomes.

Peer victimisation and internalising difficulties: The moderating role of friendship quality

A cross-sectional study is reported in which loneliness and emotional problems are explored in adolescent victims of direct or relational bullying, together with the potentially moderating influence of friendship quality. Adolescents (N=401, mean age 13.5) completed the School Relationships Questionnaire, to identify bullying and victimisation roles, the Friendship Activity Questionnaire (FAQ), the Strengths and Difficulties Questionnaire (SDQ) to assess emotional problems, and the Loneliness and Social Dissatisfaction Questionnaire (LSDQ). Relational victims, but not direct victims reported significantly elevated feelings of loneliness and emotional problems compared to non-victims. Direct victims reported a significantly higher quality of friendship, compared to non-victims. Poor quality of friendship was also associated with high levels of loneliness, and vice versa for direct victims, but not for relational victims. This indicates that the higher quality of friendship found in direct victims is associated with the reduced levels of loneliness found in this group. Given the cross-sectional nature of this study, it is not possible to ascertain whether this association truly reflects the role of friendship quality as a moderator, and hence protective factor against adverse influences of victimisation. The different mechanisms underlying direct and relational victimisation are important for future research and intervention programmes.

Not 2 old 2 TXT: There is potential to use email and SMS text message healthcare reminders for rheumatology patients up to 65 years old

Short message service (SMS) and email reminders have the potential to improve adherence to appointments and medication taking. Within the UK, information and communication technology (ICT) is widely used with a very high proportion of people having access to the internet and mobile phones. Little is known about ICT use by older adults and those with chronic illness. A feasibility survey was carried out with 112 rheumatology patients in Hertfordshire, UK to determine their current use of the internet, email and SMS and their willingness to receive electronic reminders in the future. A high proportion of patients up to age 65 are successfully using ICT despite older age or functional disability caused by rheumatic disease. Forty-four percent would be willing to receive an electronic appointment reminder and 25% a medication reminder. The results suggest that reminders would be welcomed by some patients and extensive patient training would not be needed before implementation.

Distinct psychological distress trajectories in rheumatoid arthritis: findings from an inception cohort.

OBJECTIVE As with other chronic physical illness, rates of depressive disorder are high in rheumatoid arthritis (RA). The aim of the current study was to identify distinct trajectories of psychological distress over 10 years in a cohort of RA patients recruited very early in the course of the disease. METHODS Psychological distress as measured by the Hospital Anxiety and Depression Scale total score was assessed annually in a subgroup of 784 patients enrolled in a multi-centre RA inception cohort (Early RA Study). A latent growth mixture modelling (GMM) approach was used to identify distinct psychological distress patterns. RESULTS Four distinct psychological distress trajectories were observed: low-stable (68%), high-stable (12%), high-decreasing (9%) and low-increasing (11%). Symptoms of pain, stiffness and functional impairment were significantly associated with levels of psychological distress at the time of diagnosis and after 3 years; serological markers of disease activity (ESR and CRP) were not. CONCLUSIONS Although the majority of individuals developing RA experience little or no impact of the effects of the disease on their psychological well-being, a significant proportion experience high levels of distress at some point which may be related to their subjective appraisal of their condition. Assessment and treatment of psychological distress should occur synchronously with somatic symptoms.

Attitudes of young people toward driving after smoking cannabis or after drinking alcohol

Objective Currently there is a public welfare debate about the acute effects of cannabis and risk of motor vehicle accidents. This study sought to disclose young peoples attitudes, values, and willingness to drive after smoking cannabis, and their awareness of the potential risks. Design Focus group interviews which contrasted attitudes and beliefs about drinking and driving with those about smoking cannabis and driving. Setting At the college or workplace where young people were either studying or working. Method Five focus groups comprising peers from the same work/study environment, each addressing the same set of key issues. Results Young people appear to be knowledgeable about the risks of drinking and driving, and hold a culture wide value that such behaviour is antisocial. This is in stark contrast to their willingness to smoke cannabis and drive coupled with poorly developed values and knowledge about the risks involved. Conclusion Young people appear to be risk averse when it comes to drink- driving, but willing to take risks with smoking cannabis and driving. The difference probably arises from the well developed public health campaigns and education aimed to discourage drink-driving. It is therefore reasonable to be optimistic that health education could change attitudes and willingness to drive after smoking cannabis.

A 5 item version of the Compliance Questionnaire for Rheumatology (CQR5) successfully identifies low adherence to DMARDs

BackgroundTaking DMARDs as prescribed is an essential part of self-management for patients with Rheumatoid Arthritis. To date, the Compliance Questionnaire for Rheumatology (CQR) is the only self-report adherence measure created specifically for and validated in rheumatic diseases. However, the factor structure of the CQR has not been reported and it can be considered lengthy at 19 items. The aim of this study was to test the factor structure of the CQR and reduce the number of items whilst retaining robust explanation of non-adherence to DMARDs. Such a reduction would increase the clinical utility of the scale, to identify patients with sub-optimal adherence to DMARDs in the clinic as well as for research purposes.MethodsAn exploratory factor analysis was performed to reduce the number of items in the CQR and then a confirmatory factor analysis was run to establish the fit of a 5 item version (CQR5) to the data. A discriminant function analysis was performed to determine the optimal combination of questions to identify suboptimal adherence.ResultsThe factor analyses identified a unidimensional 5 item model that explains 50.3% of the variance in adherence and has good internal consistency and fit to the data. Discriminant function analysis shows that the CQR5 can affectively detect 69% of low adherers to DMARDs using Fisher’s weighted regression equation.ConclusionA shortened version of the CQR increases the clinical utility by reducing the patient burden whilst maintaining a good level of reliability and validity for a short, self-administered, self-report questionnaire.

Interpretation of published meta-analytical studies affected by implementation errors in the GingerALE software

GingerALE (http://brainmap.org/ale/) is a widely used, freely distributed software package used to undertake co-ordinate based activation likelihood estimation (ALE) meta-analysis of neuroimaging data. The developers of the software (Eickhoff et al., 2017) have recently reported their discovery of two implementation errors which affected versions of the software prior to version 2.3.6 (released in April 2016). These errors, which have been discussed previously in Neuroscience and Biobehavioral Reviews (Tanasescu et al., 2015; Tench et al., 2016) affected the multiple comparisons correction procedure resulting in the application of more liberal statistical thresholds than should have been the case. The first error, involving calculation of the threshold for the False Discovery Rate (FDR) correction, was amended in GingerALE V2.3.3 (May, 2015) but affected all earlier versions of the software. The second error, in the cluster-level Familywise Error (FWE) correction process dating from V2.2 (May 2012), was corrected in April 2016 in V2.3.6. Several hundred published meta-analysis studies (http://www. brainmap.org/pubs/) have used versions of the GingerALE software affected by these errors. This number includes two neuroimaging metaanalyses by the present authors, published before the errors came to light (Garrison et al., 2013; Zmigrod et al., 2016). The GingerALE developers have recommended that the authors of affected studies repeat their analyses with the latest version of the software, and compare their results with the original findings (Eickhoff et al., 2017). Consistent with a few other authors of studies that used versions of GingerALE now known to have been affected by these implementation errors (e.g. Smith and Delgado, 2017), we have done this, and we summarise our findings below. We also use our experience to make suggestions for the interpretation of other published meta-analytical studies affected by the GingerALE software errors, and discuss the implications for interpreting statistical analyses more generally that may be affected by similar problems relating to the use of non-open-source, third party software products. The implementation error in the GingerALE FDR code affected calculation of the statistical threshold for determining activation significance, meaning that clusters that would otherwise have been excluded were falsely shown to have achieved significance (Eickhoff et al., 2017). Importantly, this error did not affect the calculation of individual activation likelihood effect sizes, nor the application of the statistical threshold once it had been calculated. As such, reported uncorrected ALE p values calculated from the modelled activation maps are unaffected, as are the peak locations identified in the analysis, with the implementation error impacting only on which peaks were designated as being significantly above threshold (Eickhoff et al., 2017). However, the scale of the error is variable and dependent on the particular properties of the data, being affected by both the number of neuroimaging experiments in the dataset and the number of foci in each experiment: smaller datasets being typically more affected than larger ones (Eickhoff et al., 2016; M Fox. personal communication). The effect of correcting this error on data from our two published ALE analyses was a large reduction in the number of clusters that exceeded the statistical threshold. Our first study, a meta-analysis of neuroimaging data relating to prediction error in reinforcement learning (Garrison et al., 2013), was based on a full dataset of 35 experiments and 445 foci. The significance threshold used, FDR correction with p < .05, implemented in GingerALE V2.1.1, pN (a conservative setting making no assumption about data correlation), and a minimum cluster size of 50 mm, had been chosen to mirror similar meta-analyses published a few years previously (e.g. Liu et al., 2010). Re-analysis of the prediction error data revealed that for the top level ‘All Studies’ prediction error analysis, only four of the originally reported 33 activation peaks survived correction using these FDR settings when implemented in the corrected version of the software (GingerALE V2.3.6). The impact of the error on smaller datasets was similar, so for example only three activation peaks survived for the instrumental and reward analyses using these FDR settings (previously 21 peaks each). In light of current arguments that FDR may not, in any event, be an optimal correction method for ALE analyses (Eickhoff et al., 2016, 2012), we further analysed the All-Studies prediction error data with GingerALE V2.3.6 using FWE voxel correction (p < 0.05), and cluster-level FWE correction (cluster-forming threshold of p < .001, cluster-level correction of p < 0.05) as recommended in the GingerALE manual (http://www.brainmap.org/ale/manual.pdf). Four activation peaks survived correction using FWE and five for cluster level correction.