CVLT-II indicates California Verbal Learning Test–Second Edition; RFS, Role Functioning Scale; and SANS, Scale for the Assessment of Negative Symptoms.
Associations between nodes—observed variables (rectangles) and latent variables (ovals)—are represented by edges (lines) that can be either directed (single-headed arrow) or undirected (double-headed arrow). Coefficients for the completely standardized solution are reported in the figure. Information in italics indicates constrained loadings (eMethods in the Supplement). CVLT-II indicates California Verbal Learning Test–Second Edition; RFS, Role Functioning Scale; and SANS, Scale for the Assessment of Negative Symptoms.
aP < .05.
Associations between nodes—observed variables (rectangles) and latent variables (ovals)—are represented by edges (lines) that can be either directed (single-headed arrow) or undirected (double-headed arrow). Information in italics indicates constrained loadings (eMethods in the Supplement). Coefficients for the completely standardized solution are reported in the figure. CVLT-II indicates California Verbal Learning Test–Second Edition; RFS, Role Functioning Scale; and SANS, Scale for the Assessment of Negative Symptoms.
eAppendix. Estimation Approach
eTable. Descriptive Statistics for Indicators of Early Auditory Information Processing, Cognition, Negative Symptoms, and Functional Outcome
eFigure 1. Specific Deficits Model (M1)
eFigure 2. Nonspecific Cognitive Deficits Model (M2)
eFigure 3. Nonspecific Negative Symptoms Model (M3)
eFigure 4. Direct EAP Model (M4)
eFigure 5. Direct Cognition Model (M5)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Thomas ML, Green MF, Hellemann G, et al. Modeling Deficits From Early Auditory Information Processing to Psychosocial Functioning in Schizophrenia. JAMA Psychiatry. 2017;74(1):37–46. doi:10.1001/jamapsychiatry.2016.2980
Copyright 2016 American Medical Association. All Rights Reserved.
How do deficits in early auditory information processing lead to poor functional outcomes in schizophrenia?
In a cross-sectional study of 1415 outpatients diagnosed with schizophrenia or schizoaffective disorder, early auditory information processing had a direct effect on cognition, cognition had a direct effect on negative symptoms, and both cognition and experiential negative symptoms had direct effects on functional outcome.
The data support a model in which early auditory information processing deficits lead to poor functional outcome via impaired cognition and increased negative symptoms.
Neurophysiologic measures of early auditory information processing (EAP) are used as endophenotypes in genomic studies and biomarkers in clinical intervention studies. Research in schizophrenia has established correlations among measures of EAP, cognition, clinical symptoms, and functional outcome. Clarifying these associations by determining the pathways through which deficits in EAP affect functioning would suggest when and where to therapeutically intervene.
To characterize the pathways from EAP to outcome and to estimate the extent to which enhancement of basic information processing might improve cognition and psychosocial functioning in schizophrenia.
Design, Setting, and Participants
Cross-sectional data were analyzed using structural equation modeling to examine the associations among EAP, cognition, negative symptoms, and functional outcome. Participants were recruited from the community at 5 geographically distributed laboratories as part of the Consortium on the Genetics of Schizophrenia 2 from July 1, 2010, through January 31, 2014. This well-characterized cohort of 1415 patients with schizophrenia underwent EAP, cognitive, and thorough clinical and functional assessment.
Main Outcome and Measures
Mismatch negativity, P3a, and reorienting negativity were used to measure EAP. Cognition was measured by the Letter Number Span test and scales from the California Verbal Learning Test–Second Edition, the Wechsler Memory Scale–Third Edition, and the Penn Computerized Neurocognitive Battery. Negative symptoms were measured by the Scale for the Assessment of Negative Symptoms. Functional outcome was measured by the Role Functioning Scale.
Participants included 1415 unrelated outpatients diagnosed with schizophrenia or schizoaffective disorder (mean [SD] age, 46  years; 979 males [69.2%] and 619 white [43.7%]). Early auditory information processing had a direct effect on cognition (β = 0.37, P < .001), cognition had a direct effect on negative symptoms (β = −0.16, P < .001), and both cognition (β = 0.26, P < .001) and experiential negative symptoms (β = −0.75, P < .001) had direct effects on functional outcome. The indirect effect of EAP on functional outcome was significant as well (β = 0.14, P < .001). Overall, EAP had a fully mediated effect on functional outcome, engaging general rather than modality-specific cognition, with separate pathways that involved or bypassed negative symptoms.
Conclusions and Relevance
The data support a model in which EAP deficits lead to poor functional outcome via impaired cognition and increased negative symptoms. Results can be used to help guide mechanistically informed, personalized treatments and support the strategy of using EAP measures as surrogate end points in early-stage procognitive intervention studies.
Schizophrenia is characterized by widespread deficits that range from abnormalities in basic information registration and processing1,2 to impairments in cognitive3-5 and psychosocial6 domains. Given the current evidence that cognitive impairments are robust predictors of functional outcome,7,8 there is renewed interest in the longstanding challenge of understanding how deficits in information processing contribute to cognitive and psychosocial impairments in schizophrenia.9-12 In this article, we use structural equation modeling13 to refine a theory characterizing the mediating pathways that relate impaired information processing to poor functional outcome.14 Clarifying the associations between information processing and outcome can improve the use of novel procognitive therapeutics, such as pharmacologic enhancement,15 neuroplasticity-based cognitive training,16-19 and neurostimulation,20,21 by informing clinicians and researchers when and where to intervene.
Auditory information processing deficits have been consistently identified in patients with chronic, recent-onset, and unmedicated schizophrenia and in individuals at high clinical risk for developing psychosis.14 Neurophysiologic measures of early auditory information processing (EAP) in schizophrenia correlate with cognition22-24 and functional outcome,25-28 can be feasibly measured across diverse settings,29 and thus have promising applications as endophenotypes in genomic studies and as biomarkers in clinical outcome studies.14,29-36
In one commonly used EAP assessment paradigm, electroencephalographic recordings are collected while participants passively listen to standard stimuli interspersed by deviant oddball sounds. The electroencephalographic recordings are dominated by 3 peaks labeled mismatch negativity (MMN), P3a, and reorienting negativity (RON). The MMN is thought to index preattentive processing of sounds, the P3a is thought to reflect the transition from perception and sensory registration to focal attention or orienting to the stimulus, and the RON is thought to reflect reorienting of attention after distraction.37 Together, these components reflect processes underlying auditory perception, auditory learning and memory, and other complex cognitive functions.14,38,39 The MMN, P3a, and RON arise from broadly distributed patterns of neural activation28,40 and are sensitive to N-methyl-d-aspartate receptor functioning41—a key component of long-term neuroplasticity.
Reduced MMN, P3a, and RON amplitudes, as well as longer peak latencies, are common features of schizophrenia.14,33 A previous analysis of data from the Consortium on the Genetics of Schizophrenia 2 (COGS-2)29 found EAP impairments in schizophrenia with correlations detected between EAP and global cognition, clinical symptoms, and functional outcome, findings that support the increasing literature.25,26,42-46 Although such correlations underscore the functional significance of EAP deficits, they do not disentangle the multivariate associations among domains or allow for modeling the cumulative effects of impaired information processing.
Research has suggested that measures of EAP are sensitive to procognitive behavioral and pharmacologic therapies47,48 and that changes in EAP might predict improvements in more distant treatment outcomes, that is, outcomes (eg, work functioning) that are indirectly affected by EAP. This finding is supported by a bottom-up model that suggests deficits in EAP lead to impairments in auditory attention, language, and memory14,22 and that diminished cognition subsequently contributes to defeatist beliefs and other social cognitive phenomena, negative symptoms, and, ultimately, poor functional outcome.9,49 Social cognition and clinical symptoms—mainly experiential negative symptoms (ie, avolition and anhedonia) more so than expressive negative symptoms (ie, affective blunting or flattening and alogia)50,51—appear to mediate associations between cognition and functional outcome10,52-54 and between visual perception and functional outcome.9,55,56
We sought to further clarify the multidomain pathways relating EAP to functional outcome in schizophrenia using a large sample of patients who participated in the COGS-2 study. In particular, we aimed to examine whether cognition and negative symptoms mediate the EAP-to-outcome association. We also sought to estimate the extent to which enhancement of basic information processing might improve both the cognitive and psychosocial functioning of patients with schizophrenia to provide a benchmark for future intervention studies. We hypothesized that impairments in EAP would affect negative symptoms via task modality–specific deficits in cognition (hypothesis 1). That is, we believed that measures of EAP would be more closely linked to auditory rather than visual modes of stimulus presentation. Consistent with previous results, we also hypothesized that negative symptoms would affect functional outcome via separate pathways, reflecting a dissociation of experiential and expressive negative symptoms (hypothesis 2). Last, given that EAP is thought to indirectly influence distant outcomes, we hypothesized that impairments in EAP would indirectly affect functional outcome via a single pathway running through cognition and negative symptoms (hypothesis 3).
Participants included 1415 unrelated outpatients diagnosed with schizophrenia or schizoaffective disorder who were recruited as part of COGS-2 from July 1, 2010, through January 31, 2014. A total of 1020 participants (72.1%) reported being prescribed atypical antipsychotics, 108 (7.6%) reported being prescribed typical antipsychotics, 135 (9.5%) reported being prescribed both, and 152 (10.7%) reported being prescribed no antipsychotic medication. Test sites included the University of California, San Diego, La Jolla; UCLA (University of California, Los Angeles); University of Washington, Seattle; University of Pennsylvania, Philadelphia; and Mount Sinai School of Medicine, New York, New York. Participants were excluded if they had evidence of neurologic or Axis I psychiatric disorders other than schizophrenia or schizoaffective disorder. Exclusionary factors also included head injury, stroke, and substance abuse (except tobacco). Diagnoses were verified using the patient edition of the Structured Clinical Interview for DSM-IV.57 Urine toxicology screens were used to rule out recent drug use. Additional information on selection criteria are described elsewhere.58 Written consent was obtained from all participants and this study was approved by the local human research protection committees (Human Research Protection Program of University of California, San Diego, VA Greater Los Angeles Healthcare System Institutional Review Board, Mt Sinai School of Medicine Program for the Protection of Human Subjects, University of Pennsylvania Institutional Review Board, and VA Puget Sound Healthcare System Institutional Review Board) at each testing site. All data were deidentified.
Details on deriving MMN, P3a, and RON components are provided elsewhere.29 Briefly, an auditory oddball paradigm that consisted of frequently presented tone standards interspersed with infrequent duration-increment deviants was used following established procedures.25 Waveforms to standard and deviant stimuli were calculated by averaging electroencephalographic responses to each stimulus type. Deviant minus standard difference waveforms were calculated for the MMN, P3a, and RON components.
Measures of cognition with an auditory mode of stimulus presentation included total correct scores from the Total Learning (list A trials 1-5) and Recognition Hits subscales from the California Verbal Learning Test–Second Edition (CVLT-II),59,60 Letter Number Span test,61,62 and Letter Number Sequencing subtest from the Wechsler Memory Scale–Third Edition.62,63 Measures of cognition with a visual mode of stimulus presentation were all from the Penn Computerized Neurocognitive Battery64-67 and included accuracy scores from the Visual Object Learning Test, the Penn Letter N-Back Test, the Penn Face Memory Test, and the Penn Word Memory Test. For all cognitive measures, higher scores indicate better performance.
Negative symptoms were assessed using the Scale for the Assessment of Negative Symptoms (SANS).68 The SANS includes 5 interviewer-rated global ratings: affective flattening or blunting, alogia, avolition or apathy, anhedonia or asociality, and attention. For all SANS items, higher scores indicate more symptoms. Research suggests that negative symptoms often separate into expressive and experiential factors.9,69,70 Attention ratings were not included in the analyses.
Functional outcome was assessed using the Role Functioning Scale (RFS).71 The RFS includes 4 interviewer-rated role functions: working productivity, independent living and self-care, immediate social network relationships, and extended social network relationships. For all RFS items, higher scores indicate better functioning.
Structural equation modeling13,72 was used to summarize associations among measures described in the previous sections using latent variables and then to test the plausibility of causal associations among these constructs. Our measurement model (M0) consisted of EAP, cognition for stimuli presented aurally (hereinafter auditory cognition), cognition for stimuli presented visually (hereinafter visual cognition), expressive negative symptoms, experiential negative symptoms, and functional outcome. To determine whether causal effects were mediated by specific vs nonspecific pathways, path models additionally assumed higher-order cognition (hereinafter cognition) and higher-order negative symptoms (hereinafter negative symptoms) constructs. We fitted a series of path models to the data that were designed to test the plausibility of the 3 causal hypotheses. Models are shown in Figure 1.
The specific deficits model (M1) assumed that impairments in EAP affect task modality–specific deficits in cognition—which in turn increase negative symptoms—and that expressive and experiential negative symptoms directly affect functional outcome through separate pathways. We then fitted competing models to the data to test specific hypotheses. In the nonspecific cognitive deficits model (M2), which was designed to test hypothesis 1, we restricted the causal pathway between EAP and negative symptoms to run through the cognition factor alone. In the nonspecific negative symptoms model (M3), which was designed to test hypothesis 2, we restricted the causal pathway between cognition and functional outcome to run through the negative symptoms factor alone. In the direct EAP model (M4), which was designed to test hypothesis 3, we allowed EAP to have a direct (ie, nonmediated) effect on functional outcome.
Model parameters were estimated using the latent variable analysis (lavaan) package for R.73 Models were compared using distinguishability (ω2) and closeness (z) test statistics,74 comparative fit index (CFI), root mean square error of approximation (RMSEA), Akaike information criterion (AIC), and Bayesian information criterion (BIC). Comparative fit index values in the range of 0.90 to 0.95 or greater and RMSEA values in the range of 0.08 to 0.06 and lower are typically considered acceptable.75 Smaller AIC and BIC values indicate better fit. Inferential tests were 2-tailed with the significance level set to .05. For additional information on the estimation approach, see the eAppendix in the Supplement.
Participants included 1415 unrelated outpatients diagnosed with schizophrenia or schizoaffective disorder who were recruited as part of COGS-2. The mean (SD) age of participants was 46 (11) years, 979 (69.2%) were male, and 619 (43.7%) self-identified as white, 554 (39.2%) as African American, 196 (13.9%) as Hispanic, 168 (11.9%) as more than one race, 45 (3.2%) as Asian, 14 (1.0%) as Pacific Islander, and 9 (0.6%) as Native American. The mean (SD) years of education was 13 (2), and the mean (SD) age at illness onset was 22 (7) years. Correlations among the indicators are reported in Figure 2 (for descriptive statistics, see the eTable in the Supplement). Model fit statistics are reported in the Table. The M0 model provided acceptable fit. The M0 parameter estimates are shown in Figure 3. Our initial path model (M1) (eFigure 1 in the Supplement) also provided an acceptable fit. Therefore, we continued with tests of the primary hypotheses.
No significant difference was found in model fit comparing M1 and M2 (eFigure 2 in the Supplement), and the latter produced better or equal CFI, RMSEA, AIC, and BIC statistics. Thus, the data do not support the hypothesis that impairments in EAP affect negative symptoms via task modality–specific deficits in cognition.
The M1 model was distinguishable from and fit the data significantly better than M3 (eFigure 3 in the Supplement). The former also produced better CFI, RMSEA, AIC, and BIC statistics. Thus, the data support the hypothesis that negative symptoms affect functional outcome via separate experiential and expressive negative symptoms pathways.
Although distinguishable, no significant difference was found in model fit comparing M1 and M4 (eFigure 4 in the Supplement); however, the latter produced better CFI, RMSEA, AIC, and BIC statistics. Thus, there is evidence to suggest that, counter to hypothesis 3, EAP has a direct effect on functional outcome. To explore this further, we noted that cognition was more strongly associated with functional outcome than with negative symptoms (Figure 3), which is not consistent with the assumption that negative symptoms fully mediate the association between cognition and functional outcome. This finding suggests that an omitted pathway between cognition and functional outcome could be the cause of the nearly significant pathway between EAP and functional outcome. To determine this, a fifth model, the direct-cognition model (M5) (eFigure 5 in the Supplement), was tested with a direct path from cognition to functional outcome estimated. M5 was distinguishable from and fit the data significantly better than M1. M5 also produced better CFI, RMSEA, AIC, and BIC statistics when compared with M1 and M4, suggesting that cognition rather than EAP has a direct effect on functional outcome.
We next fitted the final model that combined all previously accepted hypotheses and ad hoc findings (M6). The parameter estimates for this model are reported in Figure 4. Early auditory information processing has a direct effect on cognition that is not specific to auditory or visual task modalities. In turn, poor cognition has direct and indirect effects on functional outcome. The indirect effect is mediated by negative symptoms in general and experiential negative symptoms in particular. The M6 model was distinguishable from and fit the data significantly better than M1 and produced the best CFI, RMSEA, AIC, and BIC values of all models fitted to the data.
On the basis of the M6 parameter estimates, the effect (in standardized units) of EAP on cognition was estimated to be β = 0.37 (P < .001), and the total effect of EAP on functional outcome was estimated to be β = 0.14 (P < .001). That is, in the current sample, there was an approximately one-third SD difference in cognition and a one-seventh SD difference in functional outcome for every SD difference in EAP.
The present study clarified the multivariate associations among measures of EAP, cognition, negative symptoms, and functional outcome in schizophrenia. Results supported the hypothesized information processing, bottom-up model whereby EAP deficits contribute to cognitive impairments,14,22 which are followed by negative symptoms and reduced functional outcome.9,49 Consistent with our hypothesis, experiential negative symptoms exerted a much stronger effect on functional outcome than did expressive negative symptoms (hypothesis 2). We also found that the effect of EAP on functional outcome was fully mediated by cognition and negative symptoms (hypothesis 3), but results suggest separate pathways that involve or bypass negative symptoms. Impairments in EAP appear to be comparably associated with auditory and visual domains of cognitive functioning. Because EAP arises from a broadly distributed network,28,40 results lend support to the view that measures of EAP generally reflect impaired brain functioning rather than a specific deficit in auditory information processing.
This pattern of results has potential implications for biomarker guided treatment development. Until recently, the prevailing view of schizophrenia has been that cognitive impairments are largely immutable to rehabilitative efforts and serve as a bottleneck to optimal psychosocial functioning. Findings that cognitive impairments are not fixed but can be enhanced via pharmacologic,15 neuroplasticity-based cognitive training,16-19 neurostimulation,20,21 or combined treatment approaches (eg, pharmacologic augmentation of cognitive therapies)76 offers the hope of at least some functional recovery, even for patients with chronic illness. The parameter estimates obtained in this study would conservatively predict that an intervention producing a 1-SD improvement in EAP—or, approximately, a 1-µV change in the mean amplitude of MMN, P3a, or RON—would produce Cohen d improvements of 0.78 for cognition and 0.28 for psychosocial functioning. The number needed to treat77 with an EAP intervention to have 1 additional success would be 2.38 for cognition and 6.37 for functional outcome. Although the time course is unclear, research has revealed that event-related potentials, such as MMN, are modifiable.47 Given that treatment responses vary,78 neurophysiologic biomarkers of EAP might contribute to treatment algorithms designed to predict the likelihood of patient benefit.79 Additional research is needed to characterize the neural substrates that engender a positive treatment response and can be leveraged to guide a mechanistically informed, personalized intervention approach.80
The results of this study should be interpreted in light of key limitations. First, we used statistical modeling to explore the plausibility of causal associations. Experimental studies are needed to support causal associations among the constructs. Second, EAP data were obtained from a 2-channel recording system. Although this is generally sufficient for detecting large effect size deficits in patients with schizophrenia, it is inadequate for reliably quantifying the magnitude of subtle treatment-related effects. Higher-density recordings offer substantial improvements in artifact reduction and for decomposing cortical source dynamics attributable to a drug or cognitive training intervention.28 Third, as in most studies of patients with schizophrenia, medications were not experimentally controlled. The variable and complex medication regimens of patients cannot be convincingly disentangled via cross-sectional analyses.25 Fourth, the omission of constructs such as social cognition and defeatist beliefs limits the completeness of our EAP-to-outcome interpretive framework.9,45,53,81 Unfortunately, we lacked a sufficient number of social cognitive measures to be used in structural equation modeling.
Overall, the present findings point to a key role for EAP in terms of developing comprehensive diagnostic and treatment approaches. Recovery from mental illness may be most feasible when conventional treatments targeting symptoms, motivation, self-efficacy, and socioenvironmental barriers82,83 are combined with intensive remediation of basic information processing. Future research in this area will advance personalized and data-driven treatment by indicating when and where to intervene to affect various end points between EAP and outcome.
Corresponding Author: Michael L. Thomas, PhD, Department of Psychiatry, University of California, San Diego, 9500 Gilman Dr, MC 0738, La Jolla, CA 92093-0738 (email@example.com).
Accepted for Publication: September 18, 2016.
Published Online: December 7, 2016. doi:10.1001/jamapsychiatry.2016.2980
Author Contributions: Dr Thomas had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Thomas, Green, Sugar, Calkins, Greenwood, R. E. Gur, R. C. Gur, Lazzeroni, Siever, Stone, Swerdlow, Light.
Acquisition, analysis, or interpretation of data: Thomas, Green, Hellemann, Sugar, Tarasenko, Calkins, Greenwood, R. E. Gur, R. C. Gur, Lazzeroni, Nuechterlein, Radant, Seidman, Shiluk, Siever, Silverman, Sprock, Stone, D. W. Tsuang, M. T. Tsuang, Turetsky, Braff, Light.
Drafting of the manuscript: Thomas, Green, Hellemann, Tarasenko, R. E. Gur, Swerdlow, Braff, Light.
Critical revision of the manuscript for important intellectual content: Thomas, Green, Sugar, Calkins, Greenwood, R. E. Gur, R. C. Gur, Lazzeroni, Nuechterlein, Radant, Seidman, Shiluk, Siever, Silverman, Sprock, Stone, D. W. Tsuang, M. T. Tsuang, Turetsky, Light.
Statistical analysis: Thomas, Green, Sugar, R. C. Gur, Lazzeroni, Light.
Obtained funding: Green, Greenwood, R. E. Gur, R. C. Gur, Nuechterlein, Seidman, Siever, Silverman, Swerdlow, D. W. Tsuang, Turetsky, Braff, Light.
Administrative, technical, or material support: Sugar, Calkins, R. E. Gur, Radant, Shiluk, Sprock, Stone, Swerdlow, Light.
Study supervision: Calkins, R. E. Gur, R. C. Gur, M. T. Tsuang, Light.
Conflict of Interest Disclosures: Dr Green reported working as a consultant to AbbVie, ACADIA, DSP, FORUM, Lundbeck, and Takeda, serving on the scientific board of Luc, and receiving research support from Amgen and Forum. Dr R. E. Gur reported receiving royalties from the Brain Resource Center and working as a consultant for Mindprint Learning. Drs R. C. Gur and Turetsky reported receiving unrelated research support for investigator-initiated grants from Pfizer and AstraZeneca. Dr Nuechterlein reported receiving unrelated research support from Ortho-McNeil Janssen Scientific Affairs and consulting for Wyeth/Pfizer. Dr Swerdlow reported being a paid consultant for Neurocine Inc. Dr Light reported having been a consultant to Astellas, Boehringer-Ingelheim, Heptares, Merck, and NeuroSig. No other disclosures were reported.
Funding/Support: This work was supported in part by grants R01-MH065571 (Dr Braff), R01-MH042228 (Dr Braff), R01-MH079777 (Dr Light), K01-MH087889 (Dr Greenwood), and K23-MH102420 (Dr Thomas) from the University of California, San Diego; grant R01-MH065554 (Dr Siever) from Mount Sinai School of Medicine; grant R01-MH65707 (Dr Green) from University of California, Los Angeles; grant R01-MH65578 (Dr R. C. Gur) from the University of Pennsylvania; and grant R01-MH65558 (Dr Tsuang) from the University of Washington.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.
Additional Contributions: We thank all of the participants and nonauthor support staff who made this study possible, including the following key personnel: University of California, San Diego: Barbara Haugeland, PhD, Lauren Belleville, BA, Stacy Langton, BA, Daniel Mathias, BA, Natalie McCarthy, MA, Marlena Pela, MS, MA, Erich Riesen, BA, Maria Bongiovanni; Mount Sinai School of Medicine: Rui Ferreira, MA, Carolyn Khanian, PhD, Denise Poche-Jetter, MA, Rebecca West, MA; University of California, Los Angeles: William Horan, PhD, Amanda Bender, MS, Heidi Kuppinger, PhD, Mark McGee, BS, Ana Ceci Myers, MS, Felice Reddy, PhD, Amber Tidwell, MS, Christen Chapman, MSW; University of Pennsylvania: Amy Cassidy, MS, Erich Dress, BS, Colin Gallagher, MS, Mary March, MS, Alison Port, BA, Kosha Ruparel, MSE, Chandni Singh; University of Washington: Kate B. Alvey, Andrew C. Shutes-David, BA, Sean P. Meichle, BS, Denise O. Pritzl, ACSW, Sandra Perry, MSW, Annelise Sullivan, MS, Jane Whetstone, BA, Jake Wolf-Saxon, BA. None of these people were financially compensated for their work outside their regular salaries.
Create a personal account or sign in to: