Identifying predictors of patient outcomes evaluated over time may require modeling interactions among variables while addressing within-subject correlation. Generalized linear mixed models (GLMMs) and generalized estimating equations (GEEs) address within-subject correlation, but identifying interactions can be difficult if not hypothesized a priori. We evaluate the performance of several variable selection approaches for clustered binary outcomes to provide guidance for choosing between the methods.

We conducted simulations comparing stepwise selection, penalized GLMM, boosted GLMM, and boosted GEE for variable selection considering main effects and two-way interactions in data with repeatedly measured binary outcomes and evaluate a two-stage approach to reduce bias and error in parameter estimates. We compared these approaches in real data applications: hypothermia during surgery and treatment response in lupus nephritis.

Penalized and boosted approaches recovered correct predictors and interactions more frequently than stepwise selection. Penalized GLMM recovered correct predictors more often than boosting, but included many spurious predictors. Boosted GLMM yielded parsimonious models and identified correct predictors well at large sample and effect sizes, but required excessive computation time. Boosted GEE was computationally efficient and selected relatively parsimonious models, offering a compromise between computation and parsimony. The two-stage approach reduced the bias and error in regression parameters in all approaches.

Penalized and boosted approaches are effective for variable selection in data with clustered binary outcomes. The two-stage approach reduces bias and error and should be applied regardless of method. We provide guidance for choosing the most appropriate method in real applications.

Tobacco smoking remains one of the leading causes of preventable illness and death and is heritable with complex underpinnings. Converging evidence suggests a contribution of the polygenic risk for smoking to the use of tobacco and other substances. Yet, the underlying brain mechanisms between the genetic risk and tobacco smoking remain poorly understood.

Genomic, neuroimaging, and self-report data were acquired from a large cohort of adolescents from the IMAGEN study (a European multicenter study). Polygenic risk scores (PGRS) for smoking were calculated based on a genome-wide association study meta-analysis conducted by the Tobacco and Genetics Consortium. We examined the interrelationships among the genetic risk for smoking initiation, brain structure, and the number of occasions of tobacco use.

A higher smoking PGRS was significantly associated with both an increased number of occasions of tobacco use and smaller cortical volume of the right orbitofrontal cortex (OFC). Furthermore, reduced cortical volume within this cluster correlated with greater tobacco use. A subsequent path analysis suggested that the cortical volume within this cluster partially mediated the association between the genetic risk for smoking and the number of occasions of tobacco use.

Our data provide the first evidence for the involvement of the OFC in the relationship between smoking PGRS and tobacco use. Future studies of the molecular mechanisms underlying tobacco smoking should consider the mediation effect of the related neural structure.

There seems to be geographical differences in decisions about breast conserving surgery (BCS) in breast cancer patients. This study was to evaluate patients’ attitude to BCS and to assess the factors affecting cancer practice in West China.

A structured questionnaire was distributed to 184 patients, eliciting information about the patients’ characteristics, occupation, education, family life, recognition of illness, knowledge about BCS, the main means of gaining surgery information, selecting surgery approaches, preferences to breast reservation.

In all, 163 patients completed the questionnaire. The results indicated that only 7.4% of patients received BCS and 23% of the remaining patients desired to have BCS and the affecting factors were significantly associated with their family life, recognition of illness and the main means of gaining surgery information (P < 0.05). No associations were between BCS selecting and the other variables studied. The most frequent reasons for selecting BCS were keeping the female shape and improving quality of life (71%), the second most were postoperative recovery, minimal influence of physical function (47%) and patients’ knowledge about BCS (42%). The most frequent reasons for not selecting BCS were uncertainty about BCS results and worry about recurrence (81%), the second most was the elderly age unnecessary for BCS (40%).

The findings indicate that breast cancer patients in West China do not take BCS as the first choice as the best treatment method. It is warranted that further study of more patients, attitude of patients’ partners and physicians to BCS.

To compare therapeutic efficacy, social function, discontinue rate, relapse and recurrence rate of the depression outpatients with first episode between Venlafaxine extended release and Fluoxertine hydrochloride treatment. Methods In this 48 week natural parallel follow-up study, total 188 patients who meet ICD-10 criteria for a major depressive episode were admitted and assigned to receive either Venlafaxine Extended Release (Venlafaxine XR group) (n=89) or Fluoxertine hydrochloride(Fluoxertine group) (n=99).At baseline,week2,8,12,16,24,32,48,Hamilton Rating Scale for Depression (HAMD)-17 item was used to value disease severity, and Social Disability Screening Schedule(SDSS)for social disability, and the discontinue, relapse and recurrence rates were compared. Results (1) At week 24 Venlafaxine XR group had much lower HAMD17 total score than Fluoxertine group (P<0.05). (2)The remission rate and response rate between two groups had no statistical difference (P>0.05). (3) At week 12, Venlafaxine XR group had a higher SDSS score than Fluoxertine group (P<0.05).(4)At week 12, 16, 24, 32,48,Venlafaxine XR group displayed lower discontinue rates (P<0.05). Venlafaxine XR group had a longer treatment course than Fluoxertine did [(30.99±15.98) weeks vs. [(22.57±15.26) weeks] (P<0.01). (5) The relapse and recurrence rates of two groups had no statistical difference (P>0.05). Conclusions In the acute phase, Venlafaxine XR has a better effect for social function and treatment adherence than Fluoxertine hydrochloride. In the continued phase and sustained phase, Venlafaxine XR performs better for symptoms relief and treatment adherence.Venlafaxine XR has parallel performance with Fluoxertine hydrochloride by the terms of therapeutic efficacy, social function restore, relapse and recurrence rate.

Multiple neurotrophic factors, including vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF)-2, nerve growth factor (NGF) and insulin-like growth factor(IGF)-1, have been shown to play important roles in the pathophysiology of mood disorders. However, insufficient clinical data supporting the importance of these neurotrophic factors in mood disorders, especially manic episode, have made inconclusive to make a connection between these factors and the disorder.

This study intended to investigate possible peripheral biomarkers in serum of manic episode of bipolar disorder.

We aimed to investigate whether or not serum levels of VEGF, FGF-2, NGF and IGF-1 varied in manic state.

Serum levels of VEGF, FGF-2, NGF and IGF-1 were examined in 70 drug-naïve patients with manic episode of bipolar disorder (BM) as well as 50 healthy controls, using an ELISA method.

The mean serum levels of VEGF, FGF-2, NGF and IGF-1 were 168.13±225.61pg/ml, 279.09±378.62pg/ml, 61.38±171.67pg/ml and 162.01±72.00ng/ml in BM patients, and 140.80±143.71pg/ml, 275.46±235.29pg/ml, 36.34±15.14pg/ml and 138.90±80.11ng/ml in healthy controls, respectively. Serum levels of FGF-2, NGF and IGF-1 in patients were significantly higher than those in healthy controls (Z=−2.896, P=0.004; Z=− 2.050, P=0.040; Z=−2.188, P=0.029; respectively), although there was no statistical difference in the serum levels of VEGF between two groups (Z=-0.468, P=0.639). Moreover, serum levels of NGF in patients correlated with the duration of disorder (rs=−0.241, P=0.044).

The increase in serum levels of FGF-2, NGF and IGF-1 in manic state may reflect a neuroprotective role for these factors, and these factors may be considered biological markers for manic episode.

Previous studies have shown that the polymorphisms in COMT gene and environmental factors affect the risk of drug dependence, but there’s no research shown in relapse of heroin dependence, and the mechanism underlying remains uncertain.

Examine the interaction between allelic variants of the catechol-O- methlytransferase (COMT) gene and environmental factors (encountering drug-related environmental situations, social support) in contribution to relapse in heroin dependence.

Construct the gene-environment interaction model in order to understand the mechanism for relapse in heroin dependence.

The 249 heroin dependent subjects who followed up at one year after abstinent by using the natural history interview (NHI), social support rateing Scale (SSRS), and other questionnaires were genotyped for eight tagging single nucleotide polymorphisms (SNP) on the COMT gene. General Multifactor Dimensionality Reduction (GMDR) was used to construct the gene-environment interaction model which impacting relapse in heroin dependence.

The relapse group had higher frequencies of encountering drug-related environment (EDE) and G allele and GG genotype frequencies on COMT gene rs4680 locus and less Social Support Scale scores than that in the abstinence group. Logistic regression analysis showed that encountering more drug-related environment and GG genotype carriers were the risk factors for relapse in heroin dependence. GMDR analysis showed that the COMT gene was interact with the frequency of EDE and social support level to impact the relapse in heroin dependence.

Gene-environment interaction between COMT gene and the frequency of EDE and social support were related to heroin dependence relapse.

The debates about depressive disorder and cognition impairment are supported by controversial data.

We launched the study to investigate cognitive change in first onset depressive patients over a 6 year period.

A prospective cohort was performed. Participants included 206 cases of first onset depressive Chinese outpatient aged from 17 to 60 year old and followed from Apr. 2003-Feb. 2004 to Apr. 2009-Feb. 2010 in Shanghai. During the first 48 weeks, case management service was delivered. Participants were assessed by 17-HAMD and HAMA scale at baseline, week 12, week 32, week 48, and year 6. Cognitive changes were assessed using the WMS-RC, WAIS-RC, and WCST at baseline (n = 116), week 12 (n = 80) and year 6 (n = 24), 41 normal participants as control.

1. (1) During the first depressive onset, cognitive performance deteriorated comparing with those of control group (P < 0.01).

2. (2) At week 12,effective medication could relieve symptom and improve cognition function (P < 0.05), cognitive performance compared between patient and control with no obvious difference (P > 0.05). While patient group had a significantly larger proportion (whose Memory Quotient below 85) than control group did (χ2 = 5.66, P < 0.05).

3. (3) Using a general linear mixed model to estimate cognitive change,patients with more severe depressive retardation and lower education had a worse short-term memory over 6 years (estimate = -1.65, SE = 0.80, P < 0.05;estimate=1.63, SE = 0.30, P < 0.01), adjusting for demographics and medical status.

The first onset depressive patient has cognitive defects. Memory does not full recovery after symptom relief. Short-term memory impairment persists over 6 years with relative factors.

Finding the prediction factors for the risks of post-stroke depression (PSD) is important to stroke survivors. However, most existing studies focused only on general clinical data, which limited the predictive ability. To improve the predictive ability, this study proposed a comprehensive PSD risk prediction model with social psychological factors, neurological, cognitive functional factors and general clinical factors.

The study recruited 188 stroke patients. Patients were diagnosed by DSM-IV criteria. Predictors were collected within a week after stroke. Boosted regression trees (BRT) was used to classify these predictors, and then a predictive model was constructed based on the selected predictors. The receiver operating characteristic (ROC) curve was used to determine the performance of the predictive model .

The risk prediction model was constructed with 6 factors: Body Mass Index (BMI), cerebral infraction history (CI), Social Support Rating Scale (SSRS), Eysenck Personality Questionnaire-Neuroticism (EPQ-N), factor 1 of the 20 items Toronto Alexithymia Scale (TAS-F1) and Snaith-Hamilton-Pleasure Scale (SHARPS). In the contribution of risk prediction factors, social psychological factors was more than 0.60. ROC curve of prediction model was 0.826 (p<0.001; 95% CI) and the accuracy of prediction was 0.81 (p<0.001). Transforming the prediction model to a tree diagram, it was convenient to clinic operation.

A PSD risk prediction model with good prediction performance was constructed to achieve diagnose concisely and clearly. The social psychological factors play an important role for diagnosing PSD in the early period.

To explore the feature of functional connectivity of default mode network (DMN), central-executive network(CEN), and salience network (SN) in patients with schizophrenia during a resting state by functional magnetic resonance imaging (fMRI).

The SPM8, DPARSFA, conn, REST softwares combined with data-driven region of interest analysis were used to compare the functional connectivity (FC) of the DMN, CEN, and SN in 74 patients with schizophrenia(SZ) and 79 age- and gender-matched normal controls(NC). Medial prefrontal cortex(MPFC)was selected as seed region for identifying DMN and CEN; right anterior insula(rAI) for SN.

Compared with NC, SZ showed increased FC with bilateral dorsolateral prefrontal cortex(DLPFC) and bilateral putamen of the MPFC, and increased FC with left middle frontal cortex and precuneus/ posterior cingulate cortex(Pcu/PCC) of the rAI. SZ also showed enhanced interconnectivity strengths of CEN-DMN, CEN-SN, and DMN-SN(p>0.05). Correlation analyses showed that the increased FC between MPFC and left DLPFC significantly negatively correlated with PANSS-negative symptoms(r=-0.224,p=0.030) and increased FC between rAI and Pcu/PCC significantly correlated with PANSS-positve symptoms (r=0.243,p=0.020).

This study provides evidence for resting state functional abnormalities of DMN, CEN, and SN in schizophrenia patients. These aberrant functional connectivities in some key brain regions of the three network could be responsible for the schizophrenic symptoms.

Planning ability as a critical component of executive function has been used to investigate prefrontal cortex (PFC) function in Schizophrenia patients by several neuroimaging studies. However, the changes of PFC activation after effective antipsychotic treatment are still unclear.

The aim of this study is to explore whether there is any variation in the prefrontal hemodynamic response during Tower of London test after 6 weeks’ antipsychotic treatment in schizophrenia patients, and the relationship between the changes in PFC activation and some demographic factors as well as the severity of the patients’ psychiatric symptoms.

40 patients with first-episode schizophrenia were recruited for the present study. 28-channel NIRS (near infrared spectroscopy) was used to measure changes in hemoglobin concentration in the prefrontal cortical surface area during Tower of London (TOL) test—a classic neuropsychological test for planning abilities. The patients were examined before treatment and after six weeks’ treatment with second-generation antipsychotic medicines.

After the short-term treatment, the patients’ TOL test performance and the activations in PFC during the task period did not differ from baseline (P>0.05), although the psychiatric symptoms of the patients were improved significantly(positive subscale score 18.25±2.86 & 12.75±2.60; general psychopathology 33.67±3.65 & 27.00±3.67; PANSS total score 72.25±7.07 & 55.42±7.53; P<0.001).

It suggests that the impairment of cognitive function and the function of the PFC of schizophrenia patients would not be improved with the improvement of psychiatric symptoms, as further support the hypothesis that PFC damage is a durable impairment for schizophrenia.

Many institutions are attempting to implement patient-reported outcome (PRO) measures. Because PROs often change clinical workflows significantly for patients and providers, implementation choices can have major impact. While various implementation guides exist, a stepwise list of decision points covering the full implementation process and drawing explicitly on a sociotechnical conceptual framework does not exist.

To facilitate real-world implementation of PROs in electronic health records (EHRs) for use in clinical practice, members of the EHR Access to Seamless Integration of Patient-Reported Outcomes Measurement Information System (PROMIS) Consortium developed structured PRO implementation planning tools. Each institution pilot tested the tools. Joint meetings led to the identification of critical sociotechnical success factors.

Three tools were developed and tested: (1) a PRO Planning Guide summarizes the empirical knowledge and guidance about PRO implementation in routine clinical care; (2) a Decision Log allows decision tracking; and (3) an Implementation Plan Template simplifies creation of a sharable implementation plan. Seven lessons learned during implementation underscore the iterative nature of planning and the importance of the clinician champion, as well as the need to understand aims, manage implementation barriers, minimize disruption, provide ample discussion time, and continuously engage key stakeholders.

Highly structured planning tools, informed by a sociotechnical perspective, enabled the construction of clear, clinic-specific plans. By developing and testing three reusable tools (freely available for immediate use), our project addressed the need for consolidated guidance and created new materials for PRO implementation planning. We identified seven important lessons that, while common to technology implementation, are especially critical in PRO implementation.

The diagnosis of major depressive disorder (MDD) is symptom based due to the lack of biological biomarker. p11 protein was recently found to be an important factor mediating depression-like states and antidepressant responses. The aim of the study was to assess whether p11 protein in urine can serve as a potential biomarker for major depression, and the relationship of its levels among urine, serum and cerebrospinal fluid (CSF).

We obtained urine samples from 13 drug-free MDD patients and 13 age- and gender-matched healthy controls. We also collected urine, serum and cerebrospinal fluid samples from 13 of fracture patients or cesarean section patients in the spinal anesthesia. The concentrations of p11 protein were measured using ELISA.

In MDD patients, urine levels of p11 protein were all less than the minimum detectable concentration of the ELISA kit. The urine levels of p11 were detectable only in one healthy control. In the spinal anesthesia patients, we can detect p11 concentrations in both serum and urine in only two patients. Besides, levels of p11 were detectable in the serum of one patient and urine of another patient. We were unable to measure CSF levels of p11 in all patients.

Concentrations of p11 protein in the body fluids are very low and unstable. The sensitivity of the current p11 ELISA kit is currently unsatisfactory, requiring the development of an ELISA kit of higher sensitivity to determine whether p11 in body fluids can serve as biomarker for depression.

The authors have not supplied their declaration of competing interest.