Long-term outcome of cognitive behaviour therapy clinical trials in central Scotland.

OBJECTIVES: To establish the long-term outcome of participants in clinical trials of cognitive behaviour therapy (CBT) for anxiety disorders and psychosis, examining the effectiveness and cost-effectiveness associated with receiving CBT in comparison with alternative treatments. DESIGN: An attempt was made to contact and interview all of the participants in eight randomised, controlled, clinical trials of CBT for anxiety disorders and two randomised, controlled, clinical trials of CBT for schizophrenia conducted between 1985 and 2001. Case note reviews of healthcare resources used in the 2 years prior to entering the trials and the 2 years prior to follow-up interview were undertaken. SETTING: Mixed rural and urban settings in five localities in central Scotland. Anxiety disorder trials were conducted mainly in primary care and included three with generalised anxiety disorder, four with panic disorder and one with post-traumatic stress disorder (PTSD). The psychosis studies (one on relapse prevention and one with chronic disorder) were conducted in secondary care. PARTICIPANTS: Of the 1071 entrants to the 10 studies, 489 agreed to participate (46% of original entrants, 52% of those available to contact). INTERVENTIONS: Follow-up interviews took place between 1999 and 2003, 2-14 years after the original treatment. Interviews for Trials 1-8 were conducted by a research psychologist blind to original treatment condition. Interviews for Trials 9 and 10 were conducted by community psychiatric nurses also blind to treatment condition. Case note reviews were completed following the interview. MAIN OUTCOME MEASURES: For Trials 1-8 the main interview-based outcome measures were: Anxiety Disorders Interview Schedule-DSM-IV for diagnosis and co-morbidity, Clinical Global Severity (0-8) and the Hamilton Anxiety Rating Scale. The main patient-rated measures were: Brief Symptom Inventory, SF-36 II, Clinical Global Improvement (1-7), and the Positive and Negative Affect Scale. For Trials 9 and 10 the primary outcome measure was the interview-based Positive and Negative Syndrome Scale (PANSS). RESULTS: For the anxiety disorder studies (Trials 1-8), over half of the participants (52%) had at least one diagnosis at long-term follow-up, with significant levels of co-morbidity and health status scores comparable to the lowest 10% of the general population. Only 36% reported receiving no interim treatment for anxiety over the follow-up period with 19% receiving almost constant treatment. Patients with PTSD did particularly poorly. There was a 40% real increase in healthcare costs over the two time periods, mainly due to an increase in prescribing. A close relationship was found between poor mental and physical health for those with a chronic anxiety disorder. Treatment with CBT was associated with a better long-term outcome than non-CBT in terms of overall symptom severity but not with regard to diagnostic status. The positive effects of CBT found in the original trials were eroded over longer time periods. No evidence was found for an association between more intensive therapy and more enduring effects of CBT. Long-term outcome was found to be most strongly predicted by the complexity and severity of presenting problems at the time of referral, by completion of treatment irrespective of modality and by the amount of interim treatment during the follow-up period. The quality of the therapeutic alliance, measured in two of the studies, was not related to long-term outcome but was related to short-term outcome. The cost-effectiveness analysis showed no advantages of CBT over non-CBT. The cost of providing CBT in the original trials was only a very small proportion (6.4%) of the overall costs of healthcare for this population, which are high for both physical and mental health problems. In the psychosis studies (Trials 9 and 10), outcome was generally poor with only 10% achieving a 25% reduction in total PANSS scores from pretreatment to long-term follow-up, also cost-effectiveness analysis showed no advantages of CBT over non-CBT, although healthcare costs fell over the two time periods mainly owing to a reduction in inpatient costs. CONCLUSIONS: Psychological therapy services need to recognise that anxiety disorders tend to follow a chronic course and that good outcomes with CBT over the short term are no guarantee of good outcomes over the longer term. Clinicians who go beyond standard treatment protocols of about 10 sessions over a 6-month period are unlikely to bring about greater improvement. Poor outcomes over the long term are related to greater complexity and severity of presenting problems at the time of referral, failure to complete treatment irrespective of modality and the amount of interim treatment during the follow-up period. The relative gains of CBT are greater in anxiety disorders than in psychosis. Longitudinal research designs over extended periods of time (2-5 years), with large numbers of participants (500+), are required to investigate the relative importance of patient characteristics, therapeutic alliance and therapist expertise in determining the cost-effectiveness of CBT in the longer term.

Removal of eye blinking artifact from the electro-encephalogram, incorporating a new constrained blind source separation algorithm.

A robust constrained blind source separation (CBSS) algorithm has been developed as an effective means to remove ocular artifacts (OAs) from electro-encephalograms (EEGs). Currently, clinicians reject a data segment if the patient blinked or spoke during the observation interval. The rejected data segment could contain important information masked by the artifact. In the CBSS technique, a reference signal was exploited as a constraint. The constrained problem was then converted to an unconstrained problem by means of non-linear penalty functions weighted by the penalty terms. This led to the modification of the overall cost function, which was then minimised with the natural gradient algorithm. The effectiveness of the algorithm was also examined for the removal of other interfering signals such as electrocardiograms. The CBSS algorithm was tested with ten sets of data containing OAs. The proposed algorithm yielded, on average, a 19% performance improvement over Parra's BSS algorithm for removing OAs.

Fatty acid protection from palmitic acid-induced apoptosis is lost following PI3-kinase inhibition.

We have previously shown that saturated fatty acids induce DNA damage and cause apoptotic cell death in insulin-producing beta-cells. Here we examine further the effects of single or combined dietary fatty acids on RINm5F survival or cell death signalling. Palmitate and stearate, but not linoleate, oleate or palmitoylmethyl ester, induced growth inhibition and increased apoptosis in RINm5F cells following 24 h exposure. Co-incubation with inhibitors of ceramide synthesis, myriocin or fumonisin B(1), did not improve viability of palmitic acid treated RINm5F cells. The inhibitor of inducible nitric oxide synthase, 1400 W, similarly had no protective effect. However, linoleic acid protected against palmitic acid-induced apoptotic and necrotic cell death. The specific pharmacological inhibitors of phosphatidylinositol 3-kinase, LY294002 and wortmannin, abolished the protective effect of linoleic acid on apoptosis but not on necrosis. These data show that the growth inhibitory and apoptosis-inducing effect of the saturated fatty acid palmitate on RINm5F cells is prevented by co-incubation with the polyunsaturated fatty acid linoleate but not inhibitors of ceramide or nitric oxide generation. A key role for phosphatidylinositol 3-kinase in mediating the linoleic-acid reduction in apoptosis is suggested.

Does cognitive-behavioural therapy influence the long-term outcome of generalized anxiety disorder? An 8-14 year follow-up of two clinical trials.

BACKGROUND: Generalized anxiety disorder is a common condition of excessive worry and tension which tends to run a chronic course associated with significant psychiatric and medical problems. Cognitive behaviour therapy (CBT) has been shown to be of clinical value in about 50% of cases with treatment gains maintained over follow-up periods ranging from 6 to 12 months. The potential value of CBT over the longer term has not been subject to rigorous investigation. METHOD: Results are reported of 8-14 year follow-up of two randomized controlled trials of cognitive-behaviour therapy for generalized anxiety disorder employing structured interview with an assessor blind to initial treatment condition. Comparison groups included medication and placebo in one study based in primary care, and analytical psychotherapy in the other based in secondary care. Follow-up samples (30% and 55% of trial entrants) were broadly representative of the original cohorts. RESULTS: Overall, 50% of participants were markedly improved of whom 30-40% were recovered (i.e. free of symptoms). Outcome was significantly worse for the study based in secondary care in which the clinical presentation of participants was more complex and severe. For a minority (30-40%), mainly from the secondary care study, outcome was poor. Treatment with CBT was associated with significantly lower overall severity of symptomatology and less interim treatment, in comparison with non-CBT conditions, but there was no evidence that CBT influenced diagnostic status, probability of recovery or patient perceptions of overall improvement. CONCLUSIONS: Both CBT and the complexity and severity of presenting problems appear to influence the long-term outcome of GAD.

Heuristic pattern correction scheme using adaptively trained generalized regression neural networks.

In many pattern classification problems, an intelligent neural system is required which can learn the newly encountered but misclassified patterns incrementally, while keeping a good classification performance over the past patterns stored in the network. In the paper, an heuristic pattern correction scheme is proposed using adaptively trained generalized regression neural networks (GRNNs). The scheme is based upon both network growing and dual-stage shrinking mechanisms. In the network growing phase, a subset of the misclassified patterns in each incoming data set is iteratively added into the network until all the patterns in the incoming data set are classified correctly. Then, the redundancy in the growing phase is removed in the dual-stage network shrinking. Both long- and short-term memory models are considered in the network shrinking, which are motivated from biological study of the brain. The learning capability of the proposed scheme is investigated through extensive simulation studies.

Psychometric properties of the Parental Bonding Instrument and its association with psychological distress in a group of incarcerated young offenders in Scotland.

BACKGROUND: The Parental Bonding Instrument (PBI) was developed in order to help quantify the parental contribution to psychological distress. It has subsequently been shown to be of value in examining the influence of parents in many diverse areas including delinquency. METHOD: A shortened form of the PBI and the Hospital Anxiety and Depression Scale (HADS) were administered to a group of young offenders held in custody in Scotland. RESULTS: High levels of psychological distress were linked with low parental care, but there were no associations between psychological distress and parental control. However, an analysis of the factor structure found that, although the care factor of the PBI showed good internal reliability, the control factor was less well defined in this group. CONCLUSIONS: It is suggested that a three-factor solution representing care, control with regard to independence and protective control may be more appropriate.

Relationships between the A priori and A posteriori errors in nonlinear adaptive neural filters.

The lower bounds for the a posteriori prediction error of a nonlinear predictor realized as a neural network are provided. These are obtained for a priori adaptation and a posteriori error networks with sigmoid nonlinearities trained by gradient-descent learning algorithms. A contractivity condition is imposed on a nonlinear activation function of a neuron so that the a posteriori prediction error is smaller in magnitude than the corresponding a priori one. Furthermore, an upper bound is imposed on the learning rate eta so that the approach is feasible. The analysis is undertaken for both feedforward and recurrent nonlinear predictors realized as neural networks.

On the choice of parameters of the cost function in nested modular RNN's.

We address the choice of the coefficients in the cost function of a modular nested recurrent neural-network (RNN) architecture, known as the pipelined recurrent neural network (PRNN). Such a network can cope with the problem of vanishing gradient, experienced in prediction with RNN's. Constraints on the coefficients of the cost function, in the form of a vector norm, are considered. Unlike the previous cost function for the PRNN, which included a forgetting factor motivated by the recursive least squares (RLS) strategy, the proposed forms of cost function provide "forgetting" of the outputs of adjacent modules based upon the network architecture. Such an approach takes into account the number of modules in the PRNN, through the unit norm constraint on the coefficients of the cost function of the PRNN. This is shown to be particularly suitable, since due to inherent nesting in the PRNN, every module gives its full contribution to the learning process, whereas the unit norm constrained cost function introduces a sense of forgetting in the memory management of the PRNN. The PRNN based upon a modified cost function outperforms existing PRNN schemes in the time series prediction simulations presented.

Prediction of outcome in intensive care unit trauma patients: a multicenter study of Acute Physiology and Chronic Health Evaluation (APACHE), Trauma and Injury Severity Score (TRISS), and a 24-hour intensive care unit (ICU) point system.

OBJECTIVE: To conduct a multicenter study to validate the accuracy of the Acute Physiology and Chronic Health Evaluation (APACHE) II system, APACHE III system, Trauma and Injury Severity Score (TRISS) methodology, and a 24-hour intensive care unit (ICU) point system for prediction of mortality in ICU trauma patient admissions. METHODS: The study population consisted of retrospectively identified, consecutive ICU trauma admissions (n = 2,414) from six Level I trauma centers. Probabilities of death were calculated by using logistic regression analysis. The predictive power of each system was evaluated by using decision matrix analysis to compare observed and predicted outcomes with a decision criterion of 0.50 for risk of hospital death. The Youden Index (YI) was used to compare the proportion of patients correctly classified by each system. Measures of model calibration were based on goodness-of-fit testing (Hosmer-Lemeshow statistic less than 15.5) and model discrimination were based on the area under the receiver operating characteristic curve (AUC). RESULTS: Overall, APACHE II (sensitivity, 38%; specificity, 99%; YI, 37%; H-L statistic, 92.6; AUC, 0.87) and TRISS (sensitivity, 52%; specificity, 94%; YI, 46%; H-L statistic, 228.1; AUC, 0.82) were poor predictors of aggregate mortality, because they did not meet the acceptable thresholds for both model calibration and discrimination. APACHE III (sensitivity, 60%; specificity, 98%; YI, 58%; H-L statistic, 7.0; AUC, 0.89) was comparable to the 24-hour ICU point system (sensitivity, 51%; specificity, 98%; YI, 50%; H-L statistic, 14.7; AUC, 0.89) with both systems showing strong agreement between the observed and predicted outcomes based on acceptable thresholds for both model calibration and discrimination. The APACHE III system significantly improved upon APACHE II for estimating risk of death in ICU trauma patients (p < 0.001). Compared with the overall performance, for the subset of patients with nonoperative head trauma, the percentage correctly classified was decreased to 46% for APACHE II; increased to 71% for APACHE III (p < 0.001 vs. APACHE II); increased to 59% for TRISS; and increased to 62% for 24-hour ICU points. For operative head trauma, the percentage correctly classified was increased to 60% for APACHE II; increased to 61% for APACHE III; decreased to 43% for TRISS (p < 0.004 vs. APACHE III); and increased to 54% for 24-hour ICU points. For patients without head injuries, all of the systems were unreliable and considerably underestimated the risk of death. The percentage of nonoperative and operative patients without head trauma who were correctly classified was decreased, respectively, to 26% and 30% for APACHE II; 33% and 29% for APACHE III; 33% and 19% for TRISS; 20% and 23% for 24-hour ICU points. CONCLUSION: For the overall estimation of aggregate ICU mortality, the APACHE III system was the most reliable; however, performance was most accurate for subsets of patients with head trauma. The 24-hour ICU point system also demonstrated acceptable overall performance with improved performance for patients with head trauma. Overall, APACHE II and TRISS did not meet acceptable thresholds of performance. When estimating ICU mortality for subsets of patients without head trauma, none of these systems had an acceptable level of performance. Further multicenter studies aimed at developing better outcome prediction models for patients without head injuries are warranted, which would allow trauma care providers to set uniform standards for judging institutional performance.

Exploiting inherent relationships in RNN architectures.

We provide the relationship between the learning rate and the slope of a nonlinear activation function of a neuron within the framework of nonlinear modular cascaded systems realised through Recurrent Neural Network (RNN) architectures. This leads to reduction in the computational complexity of learning algorithms which continuously adapt the weights of such architectures, because there is a smaller number of independent parameters to optimise. Results are provided for the Gradient Descent (GD) learning algorithm and the Extended Recursive Least Squares (ERLS) algorithm, using a general nonlinear activation function of a neuron. The results obtained degenerate into the corresponding results for single RNNs, when considering only one module in such cascaded systems.

Toward an optimal PRNN-based nonlinear predictor.

We present an approach for selecting optimal parameters for the pipelined recurrent neural network (PRNN) in the paradigm of nonlinear and nonstationary signal prediction. Although there has recently been progress in terms of algorithms for training the PRNN, no account has been made of some inherent features of the PRNN. We therefore provide a study of the role of nesting, which is inherent to the PRNN architecture. The corresponding number of nested modules needed for a certain prediction task, and their contribution toward the final prediction gain (PG) give a thorough insight into the way the PRNN performs, and offers solutions for optimization of its parameters. In particular, nesting, which is a contractive function by its nature, allows the forgetting factor in the cost function of the PRNN to exceed unity, hence it becomes an emphasis factor. This compensates for the small contribution of the distant modules to the prediction process, due to nesting, and helps to circumvent the problem of vanishing gradient, experienced in RNN's for prediction. The PRNN, with its parameters chosen based upon the established criteria, is shown to outperform the linear least mean square (LMS) and recursive least squares (RLS) predictors, as well as previously proposed PRNN schemes, at no expense of additional computational complexity.

An enhanced NAS-RIF algorithm for blind image deconvolution.

We enhance the performance of the nonnegativity and support constraints recursive inverse filtering (NAS-RIF) algorithm for blind image deconvolution. The original cost function is modified to overcome the problem of operation on images with different scales for the representation of pixel intensity levels. Algorithm resetting is used to enhance the convergence of the conjugate gradient algorithm. A simple pixel classification approach is used to automate the selection of the support constraint. The performance of the resulting enhanced NAS-RIF algorithm is demonstrated on various images.

Seventh survey of audiometric practices in the United States.

A 5-page questionnaire was sent to 500 audiologists who were randomly selected from the directory of members of the American Academy of Audiology to assess the clinical practices most commonly used by certified audiologists in the United States. Survey content was limited to only those practices and procedures routinely used by audiologists across the nation. The current results were then compared to those obtained in similar studies conducted in 1971, 1972, 1978, 1985, 1989, and 1994 to determine which clinical procedures are being retained, modified, or replaced.

Complex regulation of the BRCA1 gene.

We have analyzed the promoter region of the human BRCA1 gene in detail and demonstrate that the expression of the BRCA1 gene is under complex regulation. First, its transcription is under the control of two promoters generating two distinct transcripts alpha and beta, and second, promoter alpha is shared with the adjacent NBR2 gene and is bi-directional. Both promoter alpha and promoter beta are responsive to estrogen stimulation. We also discerned that there are striking differences in both the genomic organization and immediate cis-control elements of the BRCA1 gene between humans and mice.

Isolation and characterisation of the NBR2 gene which lies head to head with the human BRCA1 gene.

To study the regulation of BRCA1 gene expression and the potential importance of dysregulation of this gene in breast and ovarian cancer, we have examined the 5' region of the human BRCA1 gene in detail. We have identified a new gene, NBR2, which is partially related to the NBR1 gene (formerly known as 1A1-3B and mapping directly adjacent to the pseudo-BRCA1 gene) and which lies head to head with the BRCA1 gene. The physical distance between the transcription start sites of the NBR2 and BRCA1 genes is 218 bp, suggesting that regulation of the expression of both genes may be co-ordinated through a bi-directional promoter. The NBR2 gene contains five exons spanning a genomic region of approximately 30 kb between the BRCA1 and pseudo-BRCA1 genes. Northern analysis showed that the NBR2 gene is expressed in all the tissues examined. The NBR2 cDNA contains an open reading frame of 112 amino acids and is predicted to encode a protein of approximately 12 kDa. Single-strand conformation polymorphism (SSCP) analysis of the NBR2 gene failed to identify any mutations in either breast or ovarian cancer, suggesting that if the NBR2 gene is involved in the development of these cancers, other mechanisms for tumorigenesis may exist. Hybridisation of NBR2 probes to zoo blots showed that the NBR2 gene is present in human and other primates. No hybridisation to DNA from other species was observed, suggesting that genomic elements controlling BRCA1 expression may differ between species.

Mutations and alternative splicing of the BRCA1 gene in UK breast/ovarian cancer families.

BRCA1 is a tumour suppressor gene located on chromosome band 17q21. It is estimated that mutations in the BRCA1 gene account for approximately 45% of the breast cancer families and almost all of the breast/ovarian cancer families. We have used single strand conformation polymorphism analysis, direct sequencing, allele specific oligonucleotide hybridisation, and reverse transcription polymerase chain reaction (RT-PCR) to look for mutations in the BRCA1 gene in 49 breast or breast/ovarian cancer families. Five distinct mutations, three novel and two previously observed, were detected in seven families. Each novel mutation was identified in one family: 3896delT in exon 11, a splicing mutation in the intron 9-exon 10 junction, and an inferred regulatory mutation. The 185delAG in exon 2 was found in three families sharing the same haplotype, but this haplotype is different from that shared by the Ashkenazi Jewish families, suggesting that the 185delAG in our families may have arisen independently. Another previously reported mutation, the 3875del4 in exon 11, was identified in one family. Of the 49 families examined, linkage analyses for both the BRCA1 and the BRCA2 regions were performed on 33 families, and mutations in the BRCA1 gene were identified in all but one family that have a lod score above 0.8 for BRCA1. All of the mutations cause either a truncated BRCA1, or loss of a BRCA1 transcript, thus are likely to be functionally disruptive. In addition, we found that alternative splicing is a common phenomenon in the processing of the BRCA1 gene. Seven variant BRCA1 transcripts were identified by RT-PCR; all but one maintained the BRCA1 open reading frame. We believe that alternative splicing may play a significant role in modulating the physiological function of BRCA1.

Isolation of the murine Nbr1 gene adjacent to the murine Brca1 gene.

The human NBR1 cDNA has previously been identified using polyclonal sera to CA125, an ovarian tumor antigen used in monitoring ovarian cancer. The gene was mapped to the BRCA1 region on chromosome 17q21 and subsequently found to lie in close proximity to the recently identified BRCA1 gene. The NBR1 protein has a B-box motif but the function of the protein is as yet unknown. To investigate the function and importance of this gene, we have studied the conservation of this gene in other species and in particular in the mouse. We have isolated murine Nbr1 cDNA and genomic clones. Translation of the cDNA sequence indicates that the protein is highly conserved, being 89% similar and 84% identical to the human. Analysis of the murine Nbr1 genomic clones indicates that it maps less than 1 kb from the Brca1 gene and that, unlike that in human, this region is not duplicated.