Open access Original research Excess cases of influenza and the coronavirus epidemic in Catalonia: a time-­series analysis of primary-­care electronic medical records covering over 6 million people Ermengol Coma Redon,1,2 Nuria Mora,1,2 Albert Prats-­Uribe ‍ ‍,3 Francesc Fina Avilés,1,2 Daniel Prieto-­Alhambra ‍ ‍,2,3 Manuel Medina1,2 To cite: Coma Redon E, ABSTRACT Mora N, Prats-­Uribe A, et al. Strengths and limitations of this study Objectives There is uncertainty about when the first Excess cases of influenza and cases of COVID-19 appeared in Spain. We aimed to the coronavirus epidemic in ►► We used good quality data covering >6 million peo- determine whether influenza diagnoses masked early Catalonia: a time-­series analysis ple and >85% of the Catalan population, obtained COVID-19 cases and estimate numbers of undetected of primary-­care electronic directly from primary record records. medical records covering over COVID-19 cases. ►► Data had previously been validated against gold-­ 6 million people. BMJ Open Design Time-­series study of influenza and COVID-19 standard influenza sentinel systems. 2020;10:e039369. doi:10.1136/ cases, 2010–2020. ►► We used ecological data and modelled it using data bmjopen-2020-039369 Setting Primary care, Catalonia, Spain. from previous seasons, therefore assuming a di- Participants People registered in primary-­care rect link between excess influenza cases and the ►► Prepublication history and additional material for this practices, covering >6 million people and >85% of the COVID-19 pandemic. paper are available online. To population. ►► Excess influenza cases could also have been due to view these files, please visit Main outcome measures Weekly new cases of influenza a panic effect, where current coronavirus epidemic the journal online (http://​dx.​doi.​ and COVID-19 clinically diagnosed in primary care. had encouraged people to consult healthcare pro- org/​10.​1136/​bmjopen-​2020-​ Analyses Daily counts of both cases were computed fessionals more frequently and for milder symptoms 039369). using the total cases recorded over the previous 7 days to than usual. avoid weekly effects. Epidemic curves were characterised ►► We lack confirmatory tests or antigenic data for the ECR, NM and AP-­U are joint first for the 2010–2011 to 2019–2020 influenza seasons. estimated excess influenza cases, but our results authors. Influenza seasons with a similar epidemic curve and agree with the proportion of influenza samples that Received 13 April 2020 peak case number as the 2019–2020 season were used tested positive for SARS-­CoV-2 in a recent study. Revised 01 July 2020 to model expected case numbers with Auto Regressive Accepted 09 July 2020 Integrated Moving Average models, overall and stratified by age. Daily excess influenza cases were defined as the number of observed minus expected cases. BACKGROUND Results Four influenza season curves (2011–2012, A new infectious disease, now named © Author(s) (or their 2012–2013, 2013–2014 and 2016–2017) were used to COVID-19, was identified by Chinese author- employer(s)) 2020. Re-­use estimate the number of expected cases of influenza in ities on 7 January 2020 as the cause of an permitted under CC BY. 2019–2020. Between 4 February 2020 and 20 March Published by BMJ. 2020, 8017 (95% CI: 1841 to 14 718) excess influenza outbreak of pneumonia in Wuhan.1 Caused 1 Sistemes d’Informació dels cases were identified. This excess was highest in the by SARS-­CoV-2, COVID-19 is asymptomatic Serveis d’Atenció Primària 15–64 age group. or presymptomatic in a high proportion of (SISAP), ICS, Barcelona, Conclusions COVID-19 cases may have been present in patients, with estimates around 15%–30%.2–4 Catalunya, Spain the Catalan population when the first imported case was 2 Most patients present mild influenza-­ like IDIAP Jordi Gol, Universitat reported on 25 February 2020. COVID-19 carriers may Autònoma de Barcelona, symptoms, including fever, dry cough, have been misclassified as influenza diagnoses in primary Barcelona, Catalunya, Spain fatigue, sore throat, dyspnoea, headache and 3 care, boosting community transmission before public Nuffield Department of health measures were taken. The use of clinical codes myalgia.5 6 Around 15%–20% of symptomatic Orthopaedics Rheumatology and Musculoskeletal Science, could misrepresent the true occurrence of the disease. cases present severe forms of disease that University of Oxford, Oxford, UK Serological or PCR testing should be used to confirm these require hospital admission.1 7 Older people, findings. In future, this surveillance of excess influenza men and those with multiple comorbidities Correspondence to could help detect new outbreaks of COVID-19 or other appear more likely to suffer more serious Dr Daniel Prieto-­Alhambra; influenza-­like pathogens, to initiate early public health ​daniel.​prietoalhambra@​ndorms.​ types of COVID-19.5 6 8–10 Conversely, chil- responses. ox.​ac.​uk dren seem to have a similar probability of Coma Redon E, et al. BMJ Open 2020;10:e039369. doi:10.1136/bmjopen-2020-039369 1 Open access infection, but milder and often asymptomatic forms of la Salut (ICS). ICS is the main primary-­care health service the disease.11 provider in Catalonia and covers about 85% of practices Cases of COVID-19 have grown exponentially and have in the region, who all use the same electronic medical been reported all over the world. The first three cases in record software, ECAP.17 Diagnosticat includes all clinical Europe were reported in France on 24 January 2020.12 influenza diagnosis codes (International Classification of The first imported COVID-19 case in Spain was dated 31 Diseases-10 codes in online supplementary table 1) and January 2020 in La Gomera, and the first in Catalonia is updated from ECAP daily (since 2010). It presents the reported a month after, on 25 February 2020. The total frequency of daily influenza cases and the weekly inci- number of confirmed cases in Catalonia then increased dence rates per 105 population, a unit that allows diag- exponentially, with 715 cumulative cases reported by 14 noses to be compared between territories independently March 2020 and a striking 4203 on 20 March 2020. Despite of the number of inhabitants. Influenza data on Diag- these official figures, it is uncertain whether SARS-­CoV-2 nosticat has been shown to accurately represent that in was circulating in the community before the first offi- a gold-­standard source, the sentinel network of influenza cial cases. It is difficult to believe, for example, that this infection reports data set.15 airborne infection did not cross the uncontrolled borders The number of COVID-19 clinical diagnoses were between Catalonia and France for a whole month. Some extracted and aggregated using the same data source and have thus speculated that undetected COVID-19 cases methods as for influenza diagnoses. Clinical diagnoses of may have been categorised as influenza before the first COVID-19 have been recorded in ECAP since 27 February official case was reported in Spain.13 2020, when bespoke codes were introduced (online Catalonia is fortunate to have a reliable system for supplementary table 1). Since 15 March 2020, Catalan influenza surveillance in place. A network of 60 sentinel policies have advocated for cases to be defined based on general practitioners (GPs) covering 1% of the total symptoms alone, with serological or PCR confirmation population report daily cases of influenza-­like illness (ILI) only required when patients are admitted to hospital or and take samples for differential diagnosis and confirma- are healthcare staff.18 tion of influenza infections in the region.14 A specialised hospital-­based system takes samples from severe hospital- Statistical analysis ised influenza cases.14 A community-­based surveillance Daily counts of influenza and COVID-19 cases were system called Diagnosticat also extracts counts of ILI diag- computed based on the frequency of cases recorded noses from a network of GP health records in real-­time, in the previous 7-­day period to avoid weekly effects on covering 85% of the population.15 This last approach recording practice. All influenza seasons in the study allows us to examine trends with granularity and to stratify period (2010–2011 to 2019–2020) were analysed sepa- analyses by age and other factors. rately to characterise annual epidemic curves for seasonal As the first cases of SARS-­CoV-2 appeared in Catalonia influenza. during the influenza epidemic season and the disease Influenza seasons with a visually similar epidemic curve shares some symptomatology with influenza, we hypothe- and similar peak case number to that of the 2019–2020 sised that SARS-­CoV-2 could have been circulating in the season were selected to model predictions for 2019–2020. community before the first confirmed case, resulting in We selected these specific seasons after an assessment an excess of influenza diagnoses. We aimed to estimate of the number of cases at the peak to maximise compa- the number of excess influenza cases in Catalonia, glob- rability with the current influenza season before the ally and by age, and to examine its relationship with the COVID-19 outbreak. Auto Regressive Integrated Moving number of clinically diagnosed COVID-19 cases. Average (ARIMA) models19 were fitted to the seasons included in the analysis for the whole population and for three age groups, paediatric patients (under 15), adults METHODS (15–64) and elderly (over 64 years old). We used a time-­series study of influenza and COVID-19 From the fitted time series, the expected speed of cases. We extracted data from primary-­ care electronic decrease in the number of weekly influenza cases for medical records covering about 85% of the population the 2019–2020 influenza season was calculated for each of Catalonia, around 6 million people. The study period day after the peak. The expected speed of decrease was included all influenza seasons from autumn–winter 2010– defined as the difference between the number of influ- 2011 to autumn–winter 2019–2020. enza diagnoses predicted between the current day t and The key study outcomes were diagnoses of influ- the previous day t−1, divided by the number of diagnoses enza and COVID-19. Daily frequency of influenza cases predicted for the previous day t−1 ((casest(casest−1)−1). recorded in primary-­ care records were obtained from Expected influenza cases were calculated using the electronic medical records, as is routinely done for the sequence Gt = G0 *∏tk=1 Vk, where Gt was the expected Diagnosticat database.16 influenza cases in the period t, G0 the number of cases at Diagnosticat is a website that reports in real-­time all the peak and Vk the speed of decrease at day k. influenza diagnoses recorded by all GPs working at any The expected influenza cases for each day on the 2019– of the primary-­care centres run by the Institut Català de 2020 season were calculated from the day of the season 2 Coma Redon E, et al. BMJ Open 2020;10:e039369. doi:10.1136/bmjopen-2020-039369 Open access Figure 1 Epidemic curves showing the weekly number Figure 2 Epidemic curves for the 2019–2020 Catalonia of new influenza cases during the influenza seasons from influenza season (solid line) and the four seasons in the past autumn–winter 2010–2011 to autumn–winter 2019–2020 in decade with a similar peak number of cases (dotted lines: Catalonia, Spain. Curves in solid lines were similar to the 2011–2012, 2012–2013, 2013–2014 and 2016–2017), centred 2019–2020 season and included in further modelling. Curves on the day of the peak number of cases in each curve. in dashed lines were not similar to the 2019–2020 season and were excluded from further modelling. 2019–2020 influenza epidemic description In Catalonia, the 2019–2020 influenza epidemic peak to 20 March 2020, the day of the data extraction. reached its peak on 4 February 2020, with 12 066 cases Excess influenza cases were defined as the number of in the previous 7 days. Figure 2 shows the evolution of observed minus expected cases, estimated daily as above. the season compared with past seasons, centred on the We calculated 95% CIs for each estimate. All analyses day of the peak. By eye, the downwards trend after the were performed in R V.3.5.1.20 peak initially looks very similar to the previous seasons. We further tested our method, as a sensitivity analysis, However, 20 days after the peak, the curve starts to flatten, with data from the most recent (2018–2019) season as a and the slope slows down. This abnormal pattern in the negative control. We checked whether the method was descending part of the curve differs from the pattern in able to identify the season as a ‘regular’ influenza season the previous seasons. not detecting excess influenza cases. Expected versus observed cases Patient and public involvement Figure 3 shows the observed and estimated numbers This research was done without patient involvement. of weekly new influenza cases (with 95% CI) after the Patients were not invited to comment on the study peak of the 2019–2020 influenza season. The estimated design and were not consulted to develop patient rele- expected number of cases were predicted using the vant outcomes or interpret the results. Patients were not selected previous influenza seasons. invited to contribute to the writing or editing of this docu- In the whole population, observed cases were greater ment for readability or accuracy. than expected after the seasonal influenza peak, but not always significant during the whole study period. The difference was statistically significant for 23 days between 4 February 2020 and 20 March 2020. Most of RESULTS these days fell after 8 March 2020, when the difference Previous influenza epidemic curves between observed and expected increased significantly Four of the previous nine influenza season curves (2011– and observed cases remained above the 95% CI band for 2012, 2012–2013, 2013–2014 and 2016–2017) had an expected cases for 2 weeks. epidemic curve and number of influenza cases during There was a greater difference between observed and the peak similar to the 2019–2020 season, as shown in expected cases among people aged 15–64 years than in figure 1. These four curves were used to estimate the both the total population and other age groups, with 25 number of expected cases of influenza in 2019–2020. The total days of significant difference. The observed and mean peak number of cases in the included and excluded expected cases diverged earlier than for the total popula- seasons was 12 762 and 14 680, respectively. The peak tion, separating around 26 February 2020 and remaining number of cases in 2019–2020 was 12 066. significantly different for the rest of the study period. ARIMA models were fitted using the included seasons. Observed and expected cases were generally similar in Online supplementary table 2 shows the full modelling those older than 64 years, until 6 March 2020. Observed process and the fitted parameters. cases then quickly rose above expected cases, with the Coma Redon E, et al. BMJ Open 2020;10:e039369. doi:10.1136/bmjopen-2020-039369 3 Open access Figure 3 Observed and expected (with 95% CI) weekly new influenza cases each day after the peak of the 2019–2020 Catalonia influenza season, in the full population and in each age group. difference becoming significant on 11 March 2020 and 2019–2020 seasonal influenza epidemic. Excess influenza remaining so for 9 days, until 19 March 2020. cases increased rapidly from 24 February 2020, peaking The shape of the observed cases curve for people on 7 March 2020. They steeply declined from 15 March younger than 15 years was similar to that for people aged 2020, coinciding with an increase in the number of 15–64 years. However, the difference between observed COVID-19 diagnoses. and expected cases was only significantly different for 11 There were 4347 excess influenza cases and 1497 clin- days, between 6 March 2020 and 16 March 2020. ical diagnoses of COVID-19 on 14 March 2020, comparing We estimated 8017 excess influenza cases (95% CI: with just 2575 excess influenza cases (40% less) and 1841 to 14 718) between 4 February 2020 and 20 March a striking 16 547 (539% increase) clinical diagnoses of 2020. This excess is presented stratified by age in table 1. COVID-19 on 20 March 2020. Results for our negative control influenza season are shown in online supplementary figure 1. We found no excess influenza in the previous (2018–2019) influenza season using the same method. DISCUSSION In mid-­February 2020, we observed an unusually high, Excess influenza cases compared with COVID-19 diagnoses larger than expected number of influenza cases in the Figure 4 depicts the number of excess influenza cases daily published data. In Catalonia, the 2019–2020 seasonal and COVID-19 diagnoses each day after the peak of the influenza epidemic reached its peak on 4 February 2020. Table 1 Number of excess influenza cases in Catalonia from 4 February 2020 to 20 March 2020, after the peak of the seasonal influenza epidemic, and the percentage of all influenza cases in that period that they make up, overall and by age group Estimated number of excess Percentage of all influenza cases in this age group made influenza cases up by the estimated excess cases Age group (95% CI) (95% CI) Younger than 15 2078 (160 to 4078) 13.6 (1.0 to 26.7) Between 15 and 64 4670 (2387 to 7124) 20.9 (10.7 to 31.8) Older than 64 142 (33 to 260) 8.9 (2.1 to 16.3) Total 8017 (1841 to 14 718) 20.4 (4.7 to 37.5) 4 Coma Redon E, et al. BMJ Open 2020;10:e039369. doi:10.1136/bmjopen-2020-039369 Open access Figure 4 Excess influenza cases and clinically diagnosed COVID-19 cases in Catalonia, Spain, as number of cases in the previous 7-­day period, from the peak of the 2019–2020 seasonal influenza epidemic (4 February 2020). Based on previous years’ data, influenza diagnoses were Our study has several limitations. We used ecological expected to decrease rapidly over the following weeks. data and modelled it using data from previous seasons, However, the number of influenza diagnoses instead therefore assuming a direct causal link between excess remained stable, which was counterintuitive and inconsis- influenza cases and the COVID-19 pandemic. As our tent with data from past influenza seasons. This increase method is based on crude count of influenza cases, major in observed influenza diagnoses over those expected, changes in denominator and population structure could here named ‘excess influenza’, correlates over time with limit the use of the proposed method, but population age the observed number of COVID-19 cases. Excess influ- and gender has remained relatively stable in the study enza cases could be used in future for the early detection period.21 Our main limitation is the possible misclassifi- of competing outbreaks. cation of disease status due to limitations related to the Using four of the previous nine influenza seasons as use of clinical codes. We lack serological tests or anti- a benchmark, we detected 8017 excess influenza cases genic data for confirmation, and this should be investi- between 4 February 2020 and 20 March 2020. This excess gated to confirm our findings. Our results agree with a was higher in people aged 15–64 years, with over 20% study that tested all influenza samples in Los Angeles for more cases than expected. The excess started to decrease SARS-­CoV-2, finding 2.2%–10.7% of the tested samples after 15 March 2020. Worryingly, these results suggest positive for the pathogen, and a centre for disease control that SARS-­CoV-2 could have already been circulating report that times the start of limited community transmis- in the Catalan population when the first imported case sion round mid-­January to February.22 23 was reported on 25 February 2020. People infected with The observed excess influenza cases could have been COVID-19 may have been masked under ILI diagnoses due to a panic effect, in which the current coronavirus in primary care, allowing continuing community trans- infodemic, a rapid spread of misinformation, has encour- mission of COVID-19 before public health measures were aged people to consult healthcare professionals more taken. frequently and for milder symptoms than usual. However, To our knowledge, this is the first study attempting to our data showed that the number of influenza diagnoses quantify the start of the COVID-19 epidemic in Spain by dropped drastically and COVID-19 diagnoses increased comparing the number of reported ILI cases with the after 15 March 2020. New COVID-19 guidelines were expected figures based on previous influenza seasons. released on 15 March 2020 in Spain that recommended The excess influenza cases metric could be useful for only testing hospital-­ admitted patients and healthcare monitoring future outbreaks of COVID-19 and other staff and encouraging GPs to diagnose COVID-19 clin- competing viral epidemics. ically without PCR confirmation.18 At least some of the Coma Redon E, et al. BMJ Open 2020;10:e039369. doi:10.1136/bmjopen-2020-039369 5 Open access excess ILI cases were thus likely to have actually been ORCID iDs COVID-19 cases. Albert Prats-­Uribe http://​orcid.​org/​0000-​0003-​1202-​9153 Daniel Prieto-­Alhambra http://​orcid.​org/​0000-​0002-​3950-​6346 Our study also has strengths. The data used were good quality, as demonstrated in many previous publi- cations,24–30 were obtained directly from primary-­ care records,and have been validated against gold-­ standard sentinel systems. This existing database covers over 85% REFERENCES 1 Ecdc. Novel coronavirus disease 2019 (COVID-19) pandemic: of the population of Catalonia, which allowed us to rapidly increased transmission in the EU/EEA and the UK – sixth update, detect excess influenza cases across the whole population 2020. and in different age groups. 2 Mizumoto K, Kagaya K, Zarebski A, et al. Estimating the asymptomatic proportion of coronavirus disease 2019 (COVID-19) In conclusion, the full extent of the COVID-19 cases on board the diamond Princess cruise SHIP, Yokohama, pandemic is still unknown. The confirmed number of Japan, 2020. Euro Surveill 2020;25. 3 Kimball A, Hatfield KM, Arons M, et al. Asymptomatic and cases may be just the tip of the iceberg, due to the lack of presymptomatic SARS-­CoV-2 infections in residents of a long-­term testing of patients presenting mild COVID-19 symptoms. care skilled nursing facility — King County, Washington, March 2020. MMWR Morb Mortal Wkly Rep;69:377–81. We need comprehensive, well-­designed, seroprevalence 4 Bianco A, Buckley AB, Overbey J, et al. Testing of patients and studies to know how many people have been infected. support persons for coronavirus disease 2019 (COVID-19) infection This novel analysis approach could offer a quantitative before scheduled deliveries. Obstet Gynecol 202010.1097/ AOG.0000000000003985. [Epub ahead of print: 19 May 2020]. approach to population surveillance that may be useful 5 Aylward B, Liang W. Report of the WHO-­China joint mission on for other institutions/regions/countries and could be coronavirus disease 2019 (COVID-19). WHO, 2020. 6 Young BE, Ong SWX, Kalimuddin S, et al. Epidemiologic features and easily integrated into current information systems. This clinical course of patients infected with SARS-­CoV-2 in Singapore. surveillance of excess influenza cases using widely avail- JAMA 2020. doi:10.1001/jama.2020.3204. [Epub ahead of print: 03 Mar 2020]. able primary-­care electronic medical records could help 7 PRIETO-­ALHAMBRA D, Ballo E, Coma-­Redon E, et al. detect new outbreaks of COVID-19 and other ILI-­causing Hospitalization and 30-­day fatality in 121,263 COVID-19 outpatient pathogens, supporting early testing and public health cases. medRxiv 2020. 8 Zhou F, Yu T, Du R, et al. Clinical course and risk factors for mortality responses. of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. Lancet 2020;395:1054–62. Acknowledgements The authors acknowledge English language editing by Dr 9 Huang C, Wang Y, Li X, et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet 2020. Jennifer A de Beyer of the Centre for Statistics in Medicine, University of Oxford. 10 Chen N, Zhou M, Dong X, et al. Epidemiological and clinical Contributors ECR, NM, AP-­U, FFA, MM and DP-­A contributed to the design of the characteristics of 99 cases of 2019 novel coronavirus pneumonia in study, the interpretation of the results, and reviewed the manuscript. EC and NM Wuhan, China: a descriptive study. Lancet 2020;395:507–13. had access to the data, performed the statistical analysis, and acted as guarantors. 11 Bi Q, Wu Y, Mei S, et al. Epidemiology and transmission of COVID-19 in Shenzhen China: analysis of 391 cases and 1,286 of their close ECR, NM, AP-­U, and DP-­A are joint first authors and wrote the first draft of the contacts. medRxiv : 2020. manuscript. MM and DP-­A are joint senior authors. 12 Bernard Stoecklin S, Rolland P, Silue Y, et al. First cases of Funding The research was partially supported by the National Institute for coronavirus disease 2019 (COVID-19) in France: surveillance, Health Research (NIHR) Oxford Biomedical Research Centre (BRC). DP-­A is investigations and control measures, January 2020. Euro Surveill 2020;25:2000094. funded through an NIHR Senior Research Fellowship (Grant number SRF-2018- 13 Boelle P-­Y. Sentinelles syndromic and viral surveillance group D. 11-­ST2-004). The views expressed in this publication are those of the author(s) excess cases of influenza like illnesses in France synchronous with and not necessarily those of the NHS, the National Institute for Health Research COVID19 invasion. medRxiv 2020;2020. or the Department of Health. AP-­U is supported by Fundacion Alfonso Martin 14 Torner N, Basile L, Martínez A, et al. Assessment of two Escudero and the Medical Research Council (grant numbers MR/K501256/1, MR/ complementary influenza surveillance systems: sentinel primary care N013468/1). influenza-­like illness versus severe hospitalized laboratory-­confirmed influenza using the moving epidemic method. BMC Public Health Competing interests DP-­A reports grants and other from AMGEN; grants, non-­ 2019;19:1089. financial support and other from UCB Biopharma; grants from Les Laboratoires 15 Closas P, Coma E, Méndez L. Sequential detection of influenza Servier, outside the submitted work; and Janssen, on behalf of IMI-­funded EHDEN epidemics by the Kolmogorov-­Smirnov test. BMC Med Inform Decis and EMIF consortiums, and Synapse Management Partners have supported training Mak 2012;12:112. programmes organised by DP-­A's department and open for external participants. 16 SISAP. Grip. Seguiment de la grip a Catalunya. Available: https:// www.​ics.​gencat.​cat/​sisap/​grip/​faq?​lang=​ct [Accessed 30 Mar 2020]. AP-­U reports grants from Fundacion Alfonso Martin Escudero and the Medical 17 Bolíbar B, Fina Avilés F, Morros R, et al. Base de datos SIDIAP: La Research Council. No other relationships or activities that could appear to have historia clínica informatizada de Atención Primaria como Fuente de influenced the submitted work. información para La investigación epidemiológica. Medicina Clínica Patient and public involvement Patients and/or the public were not involved in 2012;138:617–21. 18 Guia d’actuació enfront de casos d’infecció pel nou coronavirus the design, or conduct, or reporting or dissemination plans of this research. SARS-­CoV-2. Patient consent for publication Not required. 19 GEP B, Jenkins GM, Reinsel GC. Time series analysis: Forecasting and control. 4th edn, 2013. Provenance and peer review Not commissioned; externally peer reviewed. 20 R Core Team. R software: version 3.5.1. R Found Stat Comput, 2018. Data availability statement Data are available upon reasonable request. Data 21 IDESCAT. Territorial and demographic indicators. Estructura per edats, envelliment I dependència, 2020. Available: https://www.​ on flu cases are publicly available. All other aggregated data are available upon idescat.​cat/​pub/?​id=​inddt&​n=​915&​lang=​en [Accessed 24 Jun 2020]. reasonable request. 22 Spellberg B, Haddix M, Lee R, et al. Community prevalence of SARS-­ Open access This is an open access article distributed in accordance with the CoV-2 among patients with Influenzalike illnesses presenting to a Los Angeles medical center in March 2020. JAMA 2020. doi:10.1001/ Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits jama.2020.4958. [Epub ahead of print: 31 Mar 2020]. others to copy, redistribute, remix, transform and build upon this work for any 23 , Jorden MA, Rudman SL, et al, CDC COVID-19 Response Team. purpose, provided the original work is properly cited, a link to the licence is given, Evidence for limited early spread of COVID-19 within the United and indication of whether changes were made. See: https://​creativecommons.​org/​ States, January-­February 2020. MMWR Morb Mortal Wkly Rep licenses/​by/​4.​0/. 2020;69:680–4. 6 Coma Redon E, et al. BMJ Open 2020;10:e039369. doi:10.1136/bmjopen-2020-039369 Open access 24 Recalde M, Manzano-­Salgado CB, Díaz Y, et al. Validation of cancer development of research in primary care): the EMMA study. Rev Esp diagnoses in electronic health records: results from the information Cardiol 2012;65:29–37. system for research in primary care (SIDIAP) in northeast Spain. Clin 28 Fina-­Aviles F, Medina-­Peralta M, Mendez-­Boo L, et al. The Epidemiol 2019;11:1015–24. descriptive epidemiology of rheumatoid arthritis in Catalonia: a 25 Ponjoan A, Garre-­Olmo J, Blanch J, et al. Epidemiology of retrospective study using routinely collected data. Clin Rheumatol dementia: prevalence and incidence estimates using validated 2016;35:751–7. electronic health records from primary care. Clin Epidemiol 29 Muñoz-­Ortego J, Vestergaard P, Rubio JB, et al. Ankylosing 2019;11:217–28. spondylitis is associated with an increased risk of vertebral and 26 Ponjoan A, Garre-­Olmo J, Blanch J, et al. How well can electronic nonvertebral clinical fractures: a population-­based cohort study. J health records from primary care identify Alzheimer's disease cases? Bone Miner Res 2014;29:1770–6. Clin Epidemiol 2019;11:509–18. 30 Pagès-­Castellà A, Carbonell-­Abella C, Avilés FF, et al. "Burden of 27 Ramos R, Balló E, Marrugat J, et al. Validity for use in research osteoporotic fractures in primary health care in Catalonia (Spain): a on vascular diseases of the SIDIAP (information system for the population-­based study". BMC Musculoskelet Disord 2012;13:79. Coma Redon E, et al. BMJ Open 2020;10:e039369. doi:10.1136/bmjopen-2020-039369 7