European Psychiatry Quality of care in psychiatry is related to www.cambridge.org/epa research activity J. de Arriba-Enriquez1 , E. Sanz-Casado2 , E. Vieta3 , 4,5 4 M. Rapado-Castro * and C. Arango Research Article 1 CIBERNED, Network Centre for Biomedical Research in Neurodegenerative Diseases, National Institute of Health Carlos Cite this article: de Arriba-Enriquez J, Sanz- III, Madrid, Spain; 2Research Institute for Higher Education and Science (INAECU), University Carlos III of Madrid and Casado E, Vieta E, Rapado-Castro M, Arango C Autonomous University of Madrid, Madrid, Spain; 3Hospital Clinic, Institute of Neuroscience, University of Barcelona, (2021). Quality of care in psychiatry is related IDIBAPS, CIBERSAM, Barcelona, Catalonia, Spain; 4Department of Child and Adolescent Psychiatry, Institute of to research activity. European Psychiatry, Psychiatry and Mental Health, Hospital General Universitario Gregorio Marañón, School of Medicine, Universidad 64(1), e53, 1–6 https://doi.org/10.1192/j.eurpsy.2021.16 Complutense, IiSGM, CIBERSAM, Madrid, Spain and 5Melbourne Neuropsychiatry Centre, Department of Psychiatry, The University of Melbourne & Melbourne Health, Melbourne, VI, Australia Received: 11 November 2020 Revised: 10 February 2021 Abstract Accepted: 19 February 2021 Background. The top biomedical research institutions have traditionally been assumed to Keywords: provide better medical treatment for their patients. However, this may not necessarily be the Bibliometrics; biomedical research; healthcare case. Low-to-moderate negative associations between research activity and the quality-of-care quality; mental health; psychiatry provided by clinical departments have been described. We aimed to examine this relationship in Author for correspondence: the psychiatric units of the largest hospitals in Spain. *Marta Rapado-Castro, Methods. Scientific publications for 50 hospitals were retrieved from the Web of Science (2006– E-mail: mrapado@iisgm.com 2015), and quality of mental healthcare data were gathered from Spanish National Health System records (2008–2014). Spearman-rank correlation analyses (adjusting for number of beds and population) were used to examine the associations between research data and quality-of- care outcomes in psychiatry. Stepwise regression models were built in order to determine the predictive value of research productivity for healthcare outcomes. Results. We found a positive association between research activity indicators (i.e., number of publications, number of citations, cumulative impact factor, and institutional H-index) and better quality-of-care outcomes in psychiatry (i.e., number of readmissions, transfers, and discharges from hospital). In particular, a higher research activity predicted a lower level of readmissions for individuals with psychoses (p = 0.025; R = 0.317), explaining 8.2% of the variance when other factors were accounted for. Conclusions. Higher research activity is associated with better quality of mental healthcare in psychiatry. Our results can inform decision-making in clinical and research management settings in order to determine the most appropriate quality measures of the impact of research on the prognosis of individuals with psychiatric conditions. Introduction It is commonly asserted that leading biomedical research hospitals provide better clinical care for their patients. However, there is no specific data to support this claim. Very few studies have analyzed the relationship between research productivity and quality of the healthcare provided in clinical research centers. Previous studies have assessed quality of care mostly in teaching hospitals [1]. These assessed the reputation of hospitals based on their level of scientific production [2] or explored the relationship between the number of publications and measures of research activity such as number of citations, scholarship selection processes, and investment in research and quality-of-care indicators [3,4]. A global study in England exploring the association between mortality rates in National Health Service (NHS) trusts and academic output (number of citations per admission) attributed © The Author(s), 2021. Published by Cambridge to each NHS trust and constituent hospitals found a significant correlation between number of University Press on behalf of the European Psychiatric Association. This is an Open Access citations and mortality rates [5]. In cardiology, Pons et al. [6] studied the association between article, distributed under the terms of the bibliometric measures of research outputs in hospitals in Spain using hospital mortality from two Creative Commons Attribution licence (http:// common cardiac conditions as an outcome measure. These conditions were congestive heart creativecommons.org/licenses/by/4.0/), which failure and acute myocardial infarction. This cross-sectional study established a low to moderate permits unrestricted re-use, distribution, and negative correlation between the mortality rate and the weighted citations ratio for both heart reproduction in any medium, provided the original work is properly cited. conditions, concluding that it was important to further research the interaction between research outcomes and clinical practice. In a different study, Lascurain et al. [7] descriptively assessed the impact of scientific production in the most frequently researched medical specialties (measured in terms of publications cited in MEDLINE) on the Spanish health system (in terms of R&D resources allocated to medical science, mortality, morbidity, and drug spending) using National Institute of Health data gathered between 1991 and 2002. Medical specialties were sorted by 2 J. de Arriba-Enríquez et al. frequency of appearance using “The Serials Directory/EBSCO-CD- collected from the National Health System (CMBD) of the Spanish ROM” classification, with neuroscience being among the most Ministry of Health. CMBD data were obtained through an extrac- frequently researched. In general, authors found a disparity tion request form submitted to the Spanish Health Information between the causes of mortality in Spain and the subject matter Institute, which provided data on patients with a main or prede- of research papers published in Medline by Spanish institutions fined secondary diagnosis of a mental health condition. (both universities and healthcare centers). However, they observed Mental health diagnoses were selected on the basis of the most an equivalence between the published research subject and the prevalent psychiatric disorders according to the International Clas- cause of morbidity. A similar pattern was identified in the analysis sification of Diseases ICD-9-CM, as this was the classification used of drug spending, with neuroscience-related disorders, as the item in the national database used. The a priori designated most prev- occasioning the highest drug spending, being the medical area alent diagnoses were grouped into eight different diagnostic groups accounting for the largest share of papers published by Spanish by one of the expert coauthors (CA) based on his clinical and researchers. research experience in psychiatry. Consensus was reached with The lack of psychiatry-specific studies led us to explore the an expert external clinical psychiatrist who independently reviewed association between research and healthcare quality measures in the selected diagnosis. Diagnoses were then grouped and analyzed this field. The current study aims to determine if there is an as separate individual diagnoses as follows: (a) nonaffective psy- association between research activity and quality of care as pro- chosis disorders (schizophrenic disorders, delusional disorders, and vided in psychiatric units in Spanish hospitals. Spain has a public other nonorganic psychoses), (b) bipolar disorders (bipolar I dis- healthcare system, guaranteeing universal coverage for all residents. order, single manic episode, manic disorder recurrent episode, State healthcare is free of charge to anyone living and working in bipolar I disorder, most recent episode (or current) manic, bipolar Spain. Around 90% of Spaniards use the public healthcare system, I disorder, most recent episode (or current) depressed, bipolar I which is called the National Health System. The health system in disorder, most recent episode (or current) mixed, bipolar I disorder, Spain is very decentralized with service delivery organized at the most recent episode (or current) unspecified, other and unspecified regional level (Autonomous Communities). The system is overseen bipolar disorders, episodic mood disorders), (c) depressive disor- by the Spanish Ministry of Health, which develops policy and ders (depressive type psychosis, major depressive disorder single oversees the national health budget [8]. We hypothesized that there episode, major depressive disorder recurrent episode), (d) other would be a positive correlation between research and healthcare affective disorders (cyclothymic disorder, dysthymic disorder, quality. This would have important implications for clinical and depressive disorder, not elsewhere classified), (e) personality dis- research decision-makers, as it may help better determine the orders, (f) alcohol and drug abuse disorders (alcohol dependence quality measures or indicators most appropriate for assessing the syndrome, drug dependence, nondependent abuse of drugs, impact of research on the prognosis of individuals with mental alcohol-induced mental disorders, and drug-induced mental dis- health conditions. orders), (g) other organic mental disorders (persistent mental disorders due to conditions classified elsewhere), and (h) developmental disorders (other specified intellectual disabilities Methods and unspecified intellectual disabilities). This is a national, longitudinal, descriptive study on the relation- Based on the healthcare delivery data provided by each of the ship between scientific production and mental healthcare outcomes participating clinical research institutions, the following measures over time carried out in the top 50 leading hospitals with psychiatric were calculated for the 2008–2014 period: Average hospital stay, units in Spain. Selection, search, retrieval, gathering, and analysis of admissions (planned or unplanned), hospital discharges (voluntary representative data indicators were carried out between the years discharge or discharge to home, discharges by psychiatric diagno- 2008 and 2014 for healthcare activity and between 2006 and 2015 ses, referrals, and exitus), and readmissions (readmission rate and for research activity. nonreadmissions by psychiatric diagnoses within 30 days after discharge). Quality-of-care outcomes Research activity outputs The clinical research hospitals from which the data was retrieved were selected according to the following inclusion criteria: (a) being Research outcome data were obtained by selecting psychiatry pub- one of the following types of institution: hospital consortiums, lications from the Web of Science-WoS (Clarivate Analytics) teaching hospitals, or general hospitals, (b) being a public and/or through the institutional affiliation of each of the participating private institution with a welfare approach, (c) having a psychiatric clinical research centers. Psychology publications were excluded unit, (d) having data included in the Spanish National Health from this search due to the representativeness of the psychiatry System records, (e) having more than 500 beds, and (f) serving a area, the higher predominance of psychiatric professionals in the population of more than 50,000 inhabitants. These criteria were psychiatric units participating in this study, and the difficulty of established based on available representative data gathered from distinguishing between publications on clinical and nonclinical the Spanish National Catalogue of Hospitals and the National psychology. Statistics Institute. We included hospitals with more than The scientific publication data retrieval was carried out through 500 beds/serving a population of more than 50,000 inhabitants. a personalized search strategy that included all possible affiliations These are healthcare institutions with specialized units that receive or institutional names that the authors may have included in their referrals from other (smaller) clinical centers in the same area and publications (see supplementary material). All plausible affiliations are thus a representative sample of the population in their corre- were included in Spanish, English, and the official languages in sponding autonomous community in Spain. Healthcare outcome Spain other than Spanish (i.e., Basque, Galician, and Catalonian) of data were obtained from the Minimum Hospital Data Set (CMBD) the Autonomous Communities of each clinical center. In order to for hospital discharge records and specialized outpatient care data make the scientific activity retrieval process more precise, an European Psychiatry 3 additional search strategy (see supplementary material) was used in (dependent variables), stepwise regression models were built for order to further compile the production of clinical research groups those measures that showed significant associations in the previ- that are part of the Spanish Network for Biomedical Research in ous Spearman’s Rho correlation tests (Rs), adjusting for the Mental Health (CIBERSAM) [9]. The search, filtering, and retriev- number of beds and population (i.e., number of inhabitants) as ing of bibliographic information was conducted by a data manage- variables of noninterest. All statistical analyses were performed ment and quality-of-care specialist. using IBM SPSS Statistics for Windows, Version 19.0 (IBM Corp, A total of 3,888 records were obtained. Data was exported to Released 2010). The level of bilateral significance for all statistical EndNote bibliographic citation manager for editing. Automatic tests was set at α = 0.05. elimination of duplicates and a manual review of the records were peer-conducted by the data management specialist and a clinical Results psychiatrist. Research publications that did not belong to the par- ticipating centers or to the current study’s area of interest were A total of 2.263 scientific documents were analyzed, accounting for excluded. A total of 2,263 valid documents were analyzed. a total of 27,432 citations, which indicated a ratio of citations per Data were exported to Microsoft Excel (Microsoft Corp., Red- document of 12,12. For the sake of clarity and conciseness, only the mond, WA) spreadsheets where indicators of scientific activity main significant results are provided in Tables 2–3. Tables showing were calculated. Research indicators of the scientific activity of all comparisons can be provided upon request. the clinical centers included: (a) Research production (proportion of publications produced in psychiatry during the 2006–2015 Associations between scientific productivity and quality of period by each center selected. Only original articles, reviews, and mental healthcare editorials as a typology were included), (b) Number of citations (number of citations during the 2006–2015 period, including self- When examining Spearman’s rank-order correlation, significant citations), (c) Ratio of citations per documents (average number of associations were found between higher scientific production and citations per document including self-citations), (d) National col- higher quality of mental healthcare. In particular, higher research laborations (proportion of documents including only national CAs production (i.e., number of publications p = 0.022; Rs = 0.324, or institutions), (e) International collaborations (proportion of greater number of citations p = 0.036; Rs = 0.297, greater cumulative documents including at least one foreign CA or institution), (f) impact factor p = 0.023; Rs = 0.321, and higher Institutional Impact factor (average and cumulative impact factor of publica- H-Index p = 0.035; Rs = 0.298) were positively correlated with a tions produced by the analyzed institutions during the 2006–2015 higher proportion of nonreadmission discharges (i.e., individuals period including self-citations), and (g) Institutional H-Index (the who were not subsequently readmitted to hospital after an initial largest number h such that at least h articles from that institution discharge). In addition, higher research production (p = 0.021; were cited at least h times each). Rs = 0.325), greater number of citations (p = 0.033; Rs = 0.302), greater cumulative impact factor (p = 0.022; Rs = 0.324), and higher H-Index (p = 0.034; Rs = 0.301) were also negatively Statistical analysis associated with a lower readmission rate (see Table 1). Normal distribution of variables was assessed by means of the More specifically, higher scientific productivity in terms of Kolmogorov–Smirnov test. Means and standard deviations were higher research production (i.e., number of publications used to describe continuous variables. Frequencies were used to p = 0.026; Rs = 0.315, higher number of citations p = 0.041; Rs = 0. describe discrete variables. All variables were transformed into 290, higher cumulative impact factor p = 0.025; Rs = 0.317, and ranges to adjust to a normal distribution. The proportion (relative higher Institutional H-Index p = 0.036; Rs = 0.297) was positively frequency) expressed as a percentage was calculated for all variables correlated with a higher proportion of nonreadmission discharges of interest (i.e., research activity and quality of care variables) in in patients with schizophrenia. In addition, a significant positive order to establish their relative frequency. Spearman Rank Corre- association was found between higher research production lation analyses (controlling for number of beds and population, as (p = 0.006; Rs = 0.387), higher number of citations (p = 0.010; appropriate) were used to explore the associations between research Rs = 0.363), higher cumulative impact factor (p = 0.010; Rs = 0.359), activity and quality-of-care variables. higher Institutional H-Index (p = 0.008; Rs = 0.373), and higher In order to analyze the predictive value of research produc- number of nonreadmission discharges in patients with alcohol tivity (independent variables) for quality-of-care outcomes dependence syndrome (see Table 2). Table 1. Associations between indicators of scientific activity and measures of quality of mental healthcare. Research production Number of citations Cumulative impact factor Institutional H-index Rs p Rs p Rs p Rs p Nonreadmission discharges 0.324a 0.022 0.297a 0.036 0.321a 0.023 0.298a 0.035 a a a a Readmission rate 0.325 0.021 0.302 0.033 0.324 0.022 0.301 0.034 Proportion of referrals 0.290a 0.041 0.263 0.065 0.296a 0.037 0.259 0.069 a Significance was set at α = 0.05. Only significant results are presented (see results). Quality of care variables were: nonreadmission discharges, length of stay (hospitalization), readmission rate, planned admissions, voluntary admissions, discharges to home, referrals, exitus, and nonreadmission discharges by diagnosis. Research activity variables were: research production (number of publications), number of citations, ratio of citations per document, national collaborations, international collaborations, average impact factor, cumulative impact factor, and institutional H-index. 4 J. de Arriba-Enríquez et al. Table 2. Associations between indicators of scientific activity and nonreadmission discharges by diagnosis. Research Number of Ratio of citations per Cumulative impact Institutional Production citations document factor H-index Rs p Rs p Rs p Rs p Rs p Schizophrenic disorders 0.315a 0.026 0.290a 0.041 0.041 0.776 0.317a 0.025 0.297a 0.036 Other nonorganic psychoses 0.269 0.058 0.268 0.060 0.137 0.343 0.279a 0.050 0.241 0.092 Cyclothymic disorder 0.236 0.100 0.273 0.055 0.201 0.161 0.266 0.062 0.305a 0.031 a a a a Alcohol dependence syndrome 0.387 0.006 0.363 0.010 0.174 0.227 0.359 0.010 0.373 0.008 Generalized developmental disorders 0.182 0.206 0.219 0.126 0.304a 0.032 0.189 0.190 0.176 0.223 a Significance was set at α = 0.05. Only significant results are presented (see results). Quality of care variables were: nonreadmission discharges, length of stay (hospitalization), readmission rate, planned admissions, voluntary admissions, discharges to home, referrals, exitus, and nonreadmission discharges by diagnosis. Research activity variables were: Research production (number of publications), number of citations, ratio of citations per document, national collaborations, international collaborations, average impact factor, cumulative impact factor, and institutional H-index. Table 3. Associations between indicators of scientific activity and nonreadmission discharges by diagnostic group. Research production Number of citations Cumulative impact factor Institutional H-index Rs p Rs p Rs p Rs p Nonaffective psychosis disorders 0.322a 0.022 0.292a 0.040 0.325a 0.021 0.301a 0.034 a a Developmental disorders 0.302 0.033 0.253 0.076 0.309 0.029 0.255 0.074 a Significance was set at α = 0.05. Only significant results are presented (see results). Quality of care variables were: nonreadmission discharges, length of stay (hospitalization), readmission rate, planned admissions, voluntary admissions, discharges to home, referrals, exitus, and nonreadmission discharges by diagnosis. Research activity variables were: research production (number of publications), number of citations, ratio of citations per document, national collaborations, international collaborations, average impact factor, cumulative impact factor, and institutional H-index. When the outcome variables where explored by diagnostic predicted 6.9% of the variance of the increase in referrals (p = 0.036; group, higher scientific activity, as measured by increased research R = 0.297) when accounting for other variables in the model. production (p = 0.022; Rs = 0.322), higher number of citations In the third model, the cumulative impact factor was entered as (p = 0.040; Rs = 0.292), higher cumulative impact factor (p = .021; the independent variable and explained a significant proportion of Rs = 0.325), and higher Institutional H-Index (p = 0.034; Rs = 0.301), change in nonreadmission discharges in patients with schizo- were significantly associated with a higher proportion of nonread- phrenic disorders (p = 0.025; R = 0.317). A higher cumulative mission discharges in patients with nonaffective psychosis impact factor predicted an 8.2% increase in the proportion of (Table 3). nonreadmission discharges in patients with schizophrenic disorder over time when accounting for the effect of the covariates. In a fourth model, the proportion of number of documents Scientific production predictors of quality of mental healthcare (research production) and Institutional H-index were entered as Backward stepwise multiple regression models (controlling for predictive variables for the change over time in the proportion of number of beds and population, when appropriate) were used to nonreadmission discharges (dependent variable) in patients with determine the scientific production predictors of quality of mental cyclothymic disorder. Both production and Institutional H-index healthcare delivered in the selected psychiatric units of the clinical predicted a significant amount (17.4%) of the variance of change in research centers of interest. A total of 7 predictive models were built nonreadmission discharges (increase) (p = 0.012, R = 0.456; from the significant results obtained based on the reported associ- (,) = p = 0.004, R = 0.456) when other variables in the model were ation between variables of scientific activity and mental healthcare accounted for. outcomes. For the fifth model, the proportion of number of documents For the first model, proportion of number of documents (research production) was entered as the independent variable and (research production) was entered as the independent variable explained a significant proportion of nonreadmission discharges in and explained a significant proportion of the change in readmission patients with alcohol dependence syndrome (p = 0.006; rate (p = 0.022; R = 0.324). Specifically, higher research produc- R = 0.387). Specifically, an increase in research production over tion over time predicted a significant amount (8.6%) of the variance time predicted a significant amount (13.2%) of the variance of of change in readmission rate (decrease) when other variables were change in nonreadmission discharges (increase) when controlling controlled for. for other variables in the model. In the second model, a backward multiple regression was run to In a sixth model, the ratio of citations per document was entered explore the effect of a higher cumulative impact factor (indepen- as the independent variable and explained a significant proportion dent variable) on the prediction of an increase in the number of of nonreadmission discharges in patients with generalized devel- hospital referrals, controlling for the number of beds and popula- opmental disorders (p = 0.032; R = 0.304). An increase in this rate tion in the study. As a result, this independent variable statistically over time predicted a significant amount (7.3%) of the variance of European Psychiatry 5 change in nonreadmission discharges (increase) when controlling which can result in selection bias. Future research on this matter for covariates. is warranted. In the last model, a higher cumulative impact factor over time As for the results obtained in the regression analyses, these point predicted a significant amount (8.7%) of the variance of change in the direction of the main hypothesis of the study. Although the (increase) in the proportion of nonreadmission discharges in predictive value found between the measures of scientific activity patients with nonaffective psychosis disorders (p = 0.021; and quality of care is generally low, increased research measures R = 0.325) when all variables in the model were accounted for. may be associated with better health outcomes. Limitations of the current study include the exclusion of psy- chology and the sources of information used (only data from the Web of Science). This was an exploratory study. Thus, corrections Discussion for multiple comparisons were not done, which could be perceived In the current study, we found an association between higher as a limitation. Since multiple testing adjustments control false research productivity/outputs in leading clinical research centers positives at the potential expense of false negatives, we chose to and better quality of mental healthcare delivered in their psychiatric report all comparisons (p-values) and regard our findings as tenta- units. Specifically, our results suggest that the higher the scientific tive [10,11]. These limitations notwithstanding, the current study production, the lower the level of readmissions in certain diagnoses suggests a relationship between research outputs and clinical out- and the readmission rates, reflecting maintained improvement in comes. Future studies should pursue this line of research by expand- the patient condition, obviating the need for further hospital ing it to other scientific areas, complementing quality-of-care data admission. These findings may indicate that, in centers with high with indices of users’ subjective satisfaction, increasing the number research activity, patients have better healthcare and, therefore, less of indicators of both quality of care and scientific activity and chance of readmission once discharged. including more sources of information such as PubMed, Scopus, Although a few studies have found significant correlations and/or Google Scholar and/or including data from other countries. between research data and quality-of-care outcomes [5,6], the Our results may inform health institutions and public and results of the present study systematically denote fewer readmis- private funding agencies of relevant research measures that could sions for schizophrenic disorders, neurodevelopmental disorders, facilitate the decision-making involved in clinical and research and nonaffective psychoses in centers with high research activity. management. This would allow appropriate distribution of funding Furthermore, research activity measures had a predictive value for to health managers, resulting in more effective hiring of specialized nonreadmission discharges in patients with these disorders, which clinical research staff, establishing the best indicators for career might indicate that it is possible to determine which centers will advancement, promoting research training among residents, devel- provide more stable recovery over time, once patients with these opment of more accurate research and clinical policies and quality- conditions are discharged from inpatient units. Similarly, centers of-care assessment processes within psychiatric units and a result- with higher scientific activity have a lower level of readmissions in ing improvement in a patient-based perspective, and for the prog- patients with alcohol dependence syndrome, and research produc- nosis of individuals with psychiatric conditions. tion also has a predictive value for nonreadmission discharges with Acknowledgments. We thank Dr. Ines Palanca for her assistance with selec- this diagnosis. The results obtained in patients with developmental tion of the healthcare and quality indicators. Thanks to the Health Information disorders are particularly striking, as they might show that higher Institute from The Spanish Ministry of Health for his assistance with providing research activity predicts a lower number of readmissions in access to the healthcare data from Spanish National Health System. patients with this diagnosis. There is little training in hospitaliza- tion units for this type of neurodevelopmental disorder, which also Financial Support. MR-C was supported by a Ramon y Cajal Research includes autism, so high research activity in these conditions sug- Fellowship (RYC-2017-23144), Spanish Ministry of Science, Innovation and gests that lower readmission rates may be related to a higher level of Universities, an Alicia Koplowitz Research Grant and Short-Term Visiting Fel- clinical specialization in this type of disorder. lowship from the Fundación Alicia Koplowitz, a Fellowship Award from the In contrast, patients with cyclothymic disorder tend to have a Health Research Institute, Hospital Gregorio Marañon (Madrid, Spain), and a higher proportion of readmissions to centers with more research NARSAD independent investigator grant (24628) from the Brain & Behavior Research Foundation. MR-C and CA were partially supported by the Spanish activity. The predictive value associated with a higher proportion of Ministry of Science and Innovation and the Instituto de Salud Carlos III nonreadmission discharges in patients with this condition is related (SAM16PE07CP1, PI16/02012, PI18/00753, and PI19/024), co-financed by ERDF to higher productivity and less impact on research. Contrary to our Funds from the European Commission, “A way of making Europe,” CIBERSAM, main hypothesis, this could be related to the specific population Madrid Regional Government (B2017/BMD-3740 AGES-CM-2), European enrolled in our current study and the existence of specific clinical Union Structural Funds, European Union Seventh Framework Program and programs for mood disorders in hospitals with higher scientific European Union H2020 Program, Fundación Familia Alonso, Fundación Alicia productivity (e.g., the Bipolar Program at Hospital Clinic in Bar- Koplowitz, and Fundación Mutua Madrileña. EV was supported by the Spanish celona) to which patients are referred from a particular clinical Ministry of Science and Innovation (PI15/00283, PI18/00805) integrated into the catchment area. Thus, patients with particularly severe and recur- Plan Nacional de I + D + I and co-financed by the ISCIII-Subdirección General de rent mental illness are admitted to these programs at the largest/ Evaluación and the Fondo Europeo de Desarrollo Regional (FEDER), the Instituto de Salud Carlos III; the CIBER of Mental Health (CIBERSAM), the Secretaria most productive hospitals (the ones analyzed in the current study), d’Universitats i Recerca del Departament d’Economia i Coneixement (2017 SGR referred by community services, which in turn handle the less 1365), the CERCA Program, and the Departament de Salut de la Generalitat de severe clinical cases. That is to say, it is plausible to find that most Catalunya for the PERIS grant SLT006/17/00357. productive centers provide better healthcare if only their specific catchment area/referral population is considered. However, if these Conflicts of Interest. Dr. Arango has been a consultant to or has received clinical centers are referral hospitals for a specific mental health honoraria or grants from Acadia, Angelini, Gedeon Richter, Janssen Cilag, condition (e.g., cyclothymic disorder), they would probably be Lundbeck, Minerva, Otsuka, Roche, Sage, Servier, Shire, Schering Plow, Sumi- receiving referrals for the most severe cases in the population, tomo Dainippon Pharma, Sunovion, and Takeda. 6 J. de Arriba-Enríquez et al. Dr. Vieta has received grants and served as consultant, advisor, or CME [3] Koren G, Barker G, Mitchell V, Abramowitch L, Strofolino M, Buchwald speaker for the following entities (unrelated to the present work): AB-Biotics, M. Patient-based research in a tertiary pediatric center: a pilot study of Abbott, Allergan, Angelini, AstraZeneca, Bristol-Myers Squibb, Dainippon markers of scientific activity and productivity. Clin Invest Med. 1997; Sumitomo Pharma, Farmindustria, Ferrer, Forest Research Institute, Galenica, 20(5):354–8. Gedeon Richter, Glaxo-Smith-Kline, Janssen, Lundbeck, Otsuka, Pfizer, Roche, [4] Bornmann L, Wallon G, Ledin A. Does the committee peer review select Sage, Sanofi-Aventis, Servier, Shire, Sunovion, Takeda, the Brain and Behavior the best applicants for funding? An investigation of the selection process Foundation, the Generalitat de Catalunya (PERIS), the Spanish Ministry of for two European molecular biology organization programmes. PLoS One. Science, Innovation and Universities (CIBERSAM), EU Horizon 2020, and the 2008;3(10):e3480. Stanley Medical Research Institute. [5] Bennett WO, Bird JH, Burrows SA, Counter PR, Reddy VM. Does aca- All the other authors declare that they have no conflicts of interest concern- demic output correlate with better mortality rates in NHS trusts in ing this article. England? Public Health. 2012;126:S40–3. [6] Pons J, Sais C, Illa C, Mendez R, Sunen E, Casas M, et al. Is there an Data Availability Statement. The data that support the findings of this association between the quality of hospitals’ research and their quality of study are available from Health Information Institute from The Spanish Min- care? J Health Serv Res Policy. 2010;15(4):204–9. istry of Health. Restrictions apply to the availability of these data, which were [7] Lascurain-Sánchez ML, García-Zorita C, Martín-Moreno C, Suárez- used under request for this study. Data are available from the authors with the Balseiro C, Sanz-Casado E. Impact of health science research on the permission of Health Information Institute from The Spanish Ministry of Spanish health system, based on bibliometric and healthcare indicators. Health. Scientometrics. 2008;77(1):131–46. doi: 10.1007/s11192-007-1922-6. [8] Spanish Ministry of Health. 2021. Madrid, Spain. Available from: https:// www.mscbs.gob.es/en/home.htm. [9] Salagre E, Arango C, Artigas F, Ayuso-Mateos JL, Bernardo M, Castro- References Fornieles J, et al. CIBERSAM: Ten years of collaborative translational [1] Kupersmith J. Quality of care in teaching hospitals: A literature review. research in mental disorders. Rev Psiquiatr Salud Ment. 2019;12(1):1–8. Acad Med. 2005;80(5):458–66. [10] Althouse AD. Adjust for multiple comparisons? It’s not that simple. Ann [2] Asenjo MA, Bertran MJ, Guinovart C, Llach M, Prat A, Trilla A. Analysis Thorac Surg. 2016;101(5):1644–5. of Spanish hospital’s reputation: Relationship with their scientific produc- [11] Rothman K, Greenland S, Lash TL. Modern epidemiology third, mid-cycle tion in different subspecialities. Med Clin (Barc). 2006;126(20):768–70. revision edition. Philadelphia: Lippincott, Williams and Wilkins; 2012.