Ponjoan et al. Alzheimer's Research & Therapy (2020) 12:60 https://doi.org/10.1186/s13195-020-00625-2 RESEARCH Open Access Is it time to use real-world data from primary care in Alzheimer’s disease? Anna Ponjoan1,2,3, Josep Garre-Olmo2, Jordi Blanch1, Ester Fages1,4, Lia Alves-Cabratosa1, Ruth Martí-Lluch1,2,3, Marc Comas-Cufí1, Dídac Parramon1,4, María Garcia-Gil1 and Rafel Ramos1,5,6* Abstract Background: The analysis of real-world data in clinical research is rising, but its use to study dementia subtypes has been hardly addressed. We hypothesized that real-world data might be a powerful tool to update AD epidemiology at a lower cost than face-to-face studies, to estimate the prevalence and incidence rates of AD in Catalonia (Southern Europe), and to assess the adequacy of real-world data routinely collected in primary care settings for epidemiological research on AD. Methods: We obtained data from the System for the Development of Research in Primary Care (SIDIAP) database, which contains anonymized information of > 80% of the Catalan population. We estimated crude and standardized incidence rates and prevalences (95% confidence intervals (CI)) of AD in people aged at least 65 years living in Catalonia in 2016. Results: Age- and sex-standardized prevalence and incidence rate of AD were 3.1% (95%CI 2.7–3.6) and 4.2 per 1000 person-years (95%CI 3.8–4.6), respectively. Prevalence and incidence were higher in women and in the oldest people. Conclusions: Our incidence and prevalence estimations were slightly lower than the recent face-to-face studies conducted in Spain and higher than other analyses of electronic health data from other European populations. Real-world data routinely collected in primary care settings could be a powerful tool to study the epidemiology of AD. Keywords: Family physician, Electronic medical records, Epidemiology, General practitioner, Dementia Background the next decades [1]. However, some recent studies have Alzheimer’s disease (AD) is the most common form of shown a decline in the incidence or prevalence of AD in dementia, a major worldwide public health concern, and high-income countries, and improved education levels a huge burden to patients, families, and public health and better management of cardiovascular risk factors systems. The lack of any effective treatment to tackle it have been suggested as driving factors for such declining and the aging of the population—age is the main risk pattern [2, 3]. Therefore, updated regional or national factor for AD—may partly explain the forecasted strong epidemiology of AD is urged to adapt public policies to increase in the worldwide prevalence of AD expected for the secular and geographical trends of the disease [4]. Nevertheless, updated estimates of AD prevalence and * Correspondence: rramos.girona.ics@gencat.cat incidence in Europe are scarce [5]; for example, epi- 1 Vascular Health Research Group (ISV-Girona), Jordi Gol Institute for Primary demiological studies in Spain are based on data collected Care Research (IDIAPJGol), Barcelona, Catalonia, Spain 5 Department of Medical Sciences, School of Medicine, Campus Salut, in face-to-face surveys conducted in the 1990s or 2000s University of Girona, Girona, Catalonia, Spain [6–10]. Real-world data refers to observational data Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Ponjoan et al. Alzheimer's Research & Therapy (2020) 12:60 Page 2 of 9 routinely collected in healthcare settings derived from specialist [16, 17]. Prescription of anti-dementia drugs multiple sources: electronic health records (EHR), dis- can be requested by the general practitioner but requires ease registries, claims and billing data, and even data an external approval from a geriatrician, psychiatrist, or gathered through personal devices and health applica- neurologist; before 2014, the approval came from the ad- tions. Real-world data collected in primary care settings visory board that evaluated all patients with dementia might provide a new opportunity to study the epidemi- and prescribed pharmacological treatment. Therefore, all ology of AD; general practitioners, as gatekeepers of the patients prescribed with anti-dementia drugs have been healthcare services, might provide a large number of evaluated by one or several experts. The renewal of anti- cases compared to other clinical databases. However, dementia drug prescription is carried out in the primary epidemiological studies based on real-world data from care services. primary care have hardly addressed specific dementia subtypes, such as AD [11–13]. This lack of evidence may Data sources be related to the high level of under-diagnosis of demen- We obtained data from the Information System for tia or AD cases in primary care [14]. However, recent re- Research in Primary Care (SIDIAP database) [24], which ports stated good accuracy of diagnoses of dementia and gathers information from the primary care services in a AD recorded in primary care databases for research structured way so it can be reliably used for research. purposes [15–17]. Therefore, we hypothesized that real- SIDIAP contains data from about 6 million patients (> world data could provide regular updates on AD 80% of the Catalan population and 15% of the Spanish prevalence and incidence estimates, at reduced econom- population) attended by 1365 general practitioners in ical costs. the primary care settings of the Catalan Health Service We sought to estimate the crude and standardized (Institut Català de la Salut). SIDIAP contains anon- prevalence and incidence rates of AD in Catalonia using ymized longitudinal medical records related to demo- data from a large primary care database, including strati- graphics, symptoms, diagnoses (coded according to the fied results by age and sex. Additionally, we aimed to International Classification of Diseases, 10th revision examine the suitability of real-world data to study the (ICD-10)), and laboratory tests. SIDIAP is linked with epidemiology of AD, comparing our estimates with those other external databases (and anonymization is pre- from previous face-to-face studies. served); we were able to use information from the regis- ter of mortality and the pharmacy-invoicing database Methods provided by the Catalan Health Service, in which medi- This is a population-based study conducted in Catalonia cations are recorded using the Anatomical Therapeutic using a large EHR database from primary care. Chemical (ATC) codes). SIDIAP data is highly represen- tative of the population of Catalonia in terms of geo- Settings graphical, age, and sex distributions [25]. The quality of The Catalan public health system provides universal these data for research use has been previously docu- health coverage to all citizens and attends patients mented for several diseases [26–28] such as vascular dis- within both primary and secondary care. General practi- eases [29], dementia [16], or Alzheimer’s disease [17]. tioners and primary care nurses, as the gatekeepers of the healthcare services, assess cognition, functional Definition of AD cases ability, and independence of patients with memory com- We identified AD cases using a previously validated plaints reported by themselves and their relatives or algorithm that yielded a positive predictive value of detected by the general practitioner. Cognitive assess- 72.3% (95%CI 70.7–73.9) and a sensitivity of 83.3 ment is also performed in patients without memory (95%CI 81.8–84.6) [17]. This algorithm defined AD cases complaints when they are institutionalized due to as (i) patients with a code of AD diagnosis (ICD-10: F00, chronic complex morbidity, or included in specific G30), excluding those who had their AD diagnosis code healthcare programs (for example, the “at-home care deregistered or changed for another dementia subtype, program”). Assessment mainly consists in the adminis- or (ii) patients who had a prescription or billing of anti- tration of one cognitive test (such as Mini-Mental Status dementia drugs (ATC codes: N06DA, N06DX01) at any Examination (MMSE) [18] or Pfeiffer test [19]), one time, excluding those with a history of cerebrovascular functional ability and independence tests (Barthel Index disease (ICD-10: I60-I69, G45, G46) up to 2 years before [20], Lawton Scale [21], Katz Index [22], or Blessed the AD diagnosis, a code for a specific subtype of de- Dementia Rating Scale [23]), and, in some cases, one de- mentia such as Lewy body dementia and vascular or pression scale. Diagnosis of AD can be made by general frontotemporal dementia (ICD-10: F01, F02), or a code practitioners, but about 80–90% of the dementia and for diagnosis (ICD-10: G20–G22) or treatment (ATC: AD diagnoses recorded in SIDIAP were made by a N04) of Parkinson. Ponjoan et al. Alzheimer's Research & Therapy (2020) 12:60 Page 3 of 9 Prevalence and incidence estimates Table 1 Characteristics of the study population to estimate the The estimation of AD prevalence included persons aged prevalence of Alzheimer’s disease (AD) 65 years or over, who were alive and registered in Characteristics Study Identified Other patients SIDIAP on December 31, 2016—persons who died or population patients with (N = 1,009,508) (N = 1,048,956) AD (N = 39,448) transferred to another health provider before December Age, x̅ (sd) 75.9 (7.9) 83.1 (6.6) 75.6 (7.8) 31, 2016, were excluded. Prevalent cases were defined as patients with AD who fulfilled the algorithm conditions Women, n (%) 599,942 (57.2) 28,360 (71.9) 571,582 (56.6) on December 31, 2016. Rurality, n (%) 192,764 (19.1) 8021 (20.3) 200,785 (19.1) The estimation of the annual incidence rate included Anemia, n (%) 160,524 (15.3) 9852 (25.0) 150,672 (14.9) persons aged 65 years or over who were alive on January Cerebrovascular 107,475 (10.3) 6659 (16.9) 100,816 (10.0) 1, 2016—persons who died, transferred to another health disease, n (%) provider, or fulfilled the algorithm conditions for AD Coronary heart 100,871 (9.6) 3995 (10.1) 96,876 (9.6) before January 1, 2016, were excluded. Incident rate was disease, n (%) estimated considering incident cases in the numerator Diabetes mellitus, 258,214 (24.6) 10,732 (27.2) 247,482 (24.5) and population at risk in the denominator. Incident n (%) cases were defined as patients who fulfilled the algo- Hyperlipidemia, n (%) 567,751 (54.1) 22,144 (56.1) 545,607 (54.1) rithm conditions for AD throughout 2016. The denom- Hypertension, n (%) 673,663 (64.2) 27,490 (69.7) 646,173 (64.0) inator was defined as the time contributed by each Depression, n (%) 175,765 (16.8) 11,222 (28.5) 164,543 (16.3) person registered in SIDIAP throughout 2016, i.e., dur- ing the follow-up. Finally, we described the clinical characteristics of the prevalent cases of AD, 28,242 of which were identified study population using diagnostic records from SIDIAP by their AD diagnosis and 11,206 by their code for anti- database (details of the ICD-10 in Additional file 1); in dementia treatment. The crude prevalence was 3.8% particular, we provided prevalences of the main risk fac- (95%CI 3.7–3.8); the age-standardized prevalence, 3.3% tors for AD: cardiovascular diseases [30], cerebrovascular (95%CI 3.1–3.5); the sex-standardized prevalence, 3.6% diseases [30], anemia [31], cardiovascular risk factors (95%CI 3.5–3.7); and the age- and sex-standardized [32], and depression [33]. prevalence, 3.1% (95%CI 2.7–3.6). Prevalence increased with age and was higher in women (Table 2). The in- Statistical analyses creasing trend of prevalence with age was significantly We used absolute frequencies (percentages) to describe higher in women than in men (LR = 205.10; p value < categorical variables and means (standard deviation 0.001) (Table 2). Prevalence estimates for different age [SD]) for continuous variables. We calculated prevalence populations are shown in Additional file 2. and incidence estimates with a confidence interval (95%CI) based on binomial and Poisson distributions, re- Incidence spectively. Results were stratified by age groups (5-year To estimate the incidence rate, we observed 1,059,422 intervals) and sex. The likelihood ratio (LR) test was persons registered in SIDIAP during 2016 (1,032,699 used to examine the age-sex interaction. We standard- person-years) and identified 4844 incident cases. The ized the estimates using the direct method, and the 2013 crude incidence rate per 1000 person-years was 4.7 Revision of the European Standard Population, which (95%CI 4.6–4.8); the age-standardized incidence, 4.3 provided the following weights for age groups: 0.28, (95%C,: 4.0–4.6); the sex-standardized incidence, 4.6 0.26, 0.21, 0.13, 0.08, and 0.05 for 65–69, 70–74, 75–79, (95%CI 4.4–4.7); and the sex- and age-standardized inci- 80–84, 85–89, and ≥ 90 years, respectively, and a weight dence, 4.2 (95%CI 3.8–4.6). The incidence rate was of 0.5 for sex groups. Results were replicated in different higher in women and increased with age (Table 3), but age populations, including persons aged ≥ 70, ≥ 75, ≥ 80, dropped in the oldest persons (> 90 years). The age-sex or ≥ 85 years. All analyses were performed using the R interaction was significant, and women showed a higher software v3.5.2 [34]. increment of the incidence rate with age (LR = 53.24; p value < 0.001) (Table 3). Incidence rates in persons aged Results ≥ 70, ≥ 75, ≥ 80, and ≥ 85 years old are provided in Prevalence Additional file 3. To estimate the AD prevalence, we obtained data from 1,048,956 persons registered in SIDIAP on December Discussion 31, 2016. The mean age of the study population was This is the first comprehensive epidemiological study on 75.9 years (SD = 7.9); they mainly lived in urban areas AD to update prevalence and incidence estimates based and were mostly women (Table 1). We detected 39,448 on real-world data from primary care. In this respect, we Ponjoan et al. Alzheimer's Research & Therapy (2020) 12:60 Page 4 of 9 Table 2 Prevalence of Alzheimer’s disease in Catalonia in 2016 by age groups and sex Age group (years) Cases Population Crude prevalence Standardized (95%CI) prevalence (95%CI) All population 65–69 1396 282,815 0.5 (0.5–0.5) 0.5 (0.4-0.5)a 70–74 3131 242,888 1.3 (1.2–1.3) 1.3 (1.2–1.4)a 75–79 5885 179,329 3.3 (3.2–3.4) 3.2 (3.0–3.4)a 80–84 11,402 171,494 6.6 (6.5–6.8) 6.3 (6.1–6.6)a 85–89 11,381 110,530 10.3 (10.1–10.5) 9.6 (9.3–10.0)a ≥ 90 6253 61,900 10.1 (9.9–10.3) 9.1 (8.6–9.6)a Total 39,448 1,048,956 3.8 (3.7–3.8) 3.1 (2.7–3.6)b Women 65–69 850 149,811 0.6 (0.5–0.6) – 70–74 1979 131,252 1.5 (1.4–1.6) – 75–79 3864 100,054 3.9 (3.7–4.0) – 80–84 8093 102,434 7.9 (7.7–8.1) – 85–89 8496 71,533 11.9 (11.6–12.1) – ≥ 90 5078 44,858 11.3 (11.0–11.6) – Total 28,360 599,942 4.7 (4.7–4.8) 3.8 (3.5–4.2)c Men 65–69 546 133,004 0.4 (0.4–0.4) – 70–74 1152 111,636 1.0 (1.0–1.1) – 75–79 2021 79,275 2.5 (2.4–2.7) – 80–84 3309 69,060 4.8 (4.6–5.0) – 85–89 2885 38,997 7.4 (7.1–7.7) – ≥ 90 1175 17,042 6.9 (6.5–7.3) – Total 11,088 449,014 2.5 (2.4–2.5) 2.4 (2.1–2.8)c a Sex-standardized b Age-sex-standardized c Age-standardized contribute to fulfill a concerning gap in the literature the patient has not yet been admitted to hospital. Sur- and aim to delve into the epidemiology of AD using prisingly, AD incidence estimates from previous studies real-world data from primary care. Our findings are based on primary care databases were more similar to strengthened by the previous validation of the algorithm the study based on hospital data than to our estimates of used to identify AD cases [17]. 4.2 per 1000 person-years (95%CI 3.8–4.6). The AD inci- Real-word data are increasingly used in overall demen- dence rate was estimated in 1.5 per 1000 person-years tia research [16, 35, 36], but not in the field of AD, using data from The Health Improvement Network where it is surprisingly incipient. To our knowledge, the database (THIN) [37] and in 1.59 per 1000 person-years literature about AD epidemiology based on EHR is lim- using the General Practice Research Database (GPRD) ited and restricted to the British population. We only [11], two databases that contain EHR from primary care found two studies that used primary care databases to services in the UK. Our estimate of AD incidence rate estimate the incidence rate of AD [11, 37], and one (4.2 per 1000 person-years (95%CI 3.8–4.6)) was higher study based on EHR from hospitals, which estimated AD than those in previous studies based on EHR from pri- prevalence in 5% and incidence rate in 3.7 per 1000 mary care [11, 37, 38]. person-years in people aged at least 75 years [38]. Our Unfortunately, we could not compare our prevalence estimates were higher than the ones reported by the of AD with similar literature because, to the best of our abovementioned study based on EHR from hospitals, knowledge, no previous studies based on EHR from pri- since we obtained an AD prevalence of 6.7% and an inci- mary care have reported such estimates. Other studies dence of 8.1 per 1000 person-years among people aged were based on EHR, but did not assess AD, only all- at least 75 years old. Estimates from primary care data- cause dementia. For example, Jaakkimainen et al. [39] bases are expected to be higher than those from second- estimated the prevalence of overall dementia in 72.0 per ary care, because there is a time frame in which the 1000 persons using EHR recorded by family physicians diagnosis might only be in the primary care records—if from Canada. Perera et al. [35] reported prevalences of Ponjoan et al. Alzheimer's Research & Therapy (2020) 12:60 Page 5 of 9 Table 3 Incidence rate (per 1000 person-years) of Alzheimer’s disease in Catalonia in 2016 by age and sex Age group (years) Cases Person-years Crude incidence Standardized (95%CI) incidence (95%CI) All population 65–69 199 283,590 0.7 (0.6–0.8) 0.7 (0.6–0.9)a 70–74 547 242,267 2.3 (2.1–2.5) 2.2 (2.0–2.5)a 75–79 915 176,418 5.2 (4.9–5.5) 5.1 (4.6–5.6)a 80–84 1555 164,679 9.4 (9.0–9.9) 9.1 (8.5–9.8)a 85–89 1200 104,223 11.5 (10.9–12.2) 11.1 (10.2–12.0)a ≥ 90 428 61,522 7.0 (6.3–7.7) 6.8 (5.9–7.9)a Total 4844 1,032,699 4.7 (4.6–4.8) 4.2 (3.8–4.6)b Women 65–69 118 149,749 0.8 (0.7–0.9) – 70–74 355 130,254 2.7 (2.5–3.0) – 75–79 580 97,468 6.0 (5.5–6.5) – 80–84 1064 96,633 11.0 (10.4–11.7) – 85–89 844 65,875 12.8 (12.0–13.7) – ≥ 90 311 43,700 7.1 (6.4–8.0) – Total 3272 583,680 5.6 (5.4–5.8) 4.9 (4.5–5.3)c Men 65–69 81 133,841 0.6 (0.5–0.8) – 70–74 192 112,012 1.7 (1.5–2.0) – 75–79 335 78,950 4.3 (3.8–4.7) – 80–84 491 68,046 7.2 (6.6–7.9) – 85–89 356 38,348 9.3 (8.3–10.3) – ≥ 90 117 17,822 6.6 (5.4–7.9) – Total 1572 449,019 3.5 (3.3–3.7) 3.5 (3.1–3.9) c a Sex-standardized b Age-sex-standardized c Age-standardized overall dementia in 2012 of 0.4, 1, 2, 5, 8, 10, and 11 for and thus should be interpreted cautiously. In Spain, pre- the following age groups: 65 to 70, 70 to 75, 75 to 80, 80 vious face-to-face studies estimated AD incidence rates to 85, 85 to 90, 90 to 95, and ≥ 95 years, respectively. of 7.4 per 1000 person-years (95%CI 6.0–8.8) [10] and Since AD corresponds to about 50–60% of all dementia 10.8 per 1000 person-years (95%CI 7.8–13.7) in people cases, we expected to find lower prevalences than the aged at least 65 and 75 years, respectively [42], while our ones observed by Jaakkimainen et al. [39] or by Perera crude estimates were 4.7 (95%CI 4.6–4.8) and 8.09 et al. [35], but we obtained estimates similar or slightly (95%CI 7.84–8.34) per 1000 person-years in persons lower than the abovementioned prevalences of overall aged at least 65 and 75 years, respectively. Our preva- dementia. This could suggest that SIDIAP might capture lence estimates were slightly lower than the ones re- a higher number of AD and dementia cases than data- ported by face-to-face studies. Age-standardized AD bases from other healthcare systems [35]. prevalence was estimated in 4.2% worldwide [40], and in The literature on AD epidemiology from face-to-face 4.4% or 5.1% in Europe [41, 43], while our estimate was studies is much more comprehensive than that from 3.3% (95%CI 3.1–3.5). In Spain, age- and sex- EHR studies. Our incidence figures were concordant standardized prevalence was estimated in 4.0% in people with the Delphi consensus study that reported a demen- aged ≥ 65 years and in 5.6% in people aged ≥ 75 in face- tia incidence rate of 8.8 per 1000 person-years in Central to-face studies [6, 7]; these findings were slightly lower Europe [1], about 60% of which corresponded to AD [9]. and similar to our estimates of 3.1% (95%CI 2.7–3.6) in However, our incidence estimates were lower than the the population aged ≥ 65 years and 5.8% (5.3–6.4) in the observed in two meta-analyses that provided estimates population aged ≥ 75 years. Our crude prevalence esti- of 15.8 per 1000 person-years worldwide [40] and 11.1 mate (3.8% (95%CI 3.7–3.8)) fell within the low range of per 1000 person-years in Europe [41]. Notably, both prevalence estimates from previous studies in the meta-analyses reported significantly high heterogeneity Spanish population, which was from 2.9 to 6.9% [44–47]. Ponjoan et al. Alzheimer's Research & Therapy (2020) 12:60 Page 6 of 9 Our estimates of prevalence and incidence of AD prescription of anti-dementia drugs to benefit from were similar or slightly lower than face-to-face stud- economic discounts or for a prescription renewal. ies, which could indicate a decreasing trend in AD in- In SIDIAP, under-registration of AD was concentrated cidence and prevalence. Previous face-to-face studies in the oldest group: we observed increasing trends of were conducted in Spain during the 1990s, and AD prevalence and incidence with age except for this thenceforth, high-income countries have presented a group (≥ 90 years), in line with previous studies [50]. decreasing trend in dementia [3] or AD [48] inci- Some general practitioners might be reluctant to use de- dence. However, other countries only observed a de- mentia codes among the oldest persons because they creasing trend in vascular dementia but not in AD may consider the diagnosis and treatment as ineffectual [49]; the incidence of AD might actually be increas- in this population or may interpret their memory diffi- ing, and the decline observed in the incidence of de- culties as part of the normal aging process rather than as mentia in general could be related to a decline in a disability [16]. Besides, people aged at least 90 years vascular dementia. Further research to clarify the might have higher levels of comorbidity that could affect trend of AD and of vascular dementia would be of the quality of AD diagnosis and register. For example, interest. Our lower estimates could also be explained those persons who have impaired mobility due to by methodological differences, such as population chronic complex diseases might not be able to visit the characteristics, different case definitions, or under- specialist, resulting in a diagnosis of unspecified demen- registration of AD. In our study, under-registration of tia instead of AD; this is in line with previous findings AD cases might have occurred for different reasons. that reported the highest prevalence and incidence rate First, under-registration is expected to be higher in estimates of overall dementia among the oldest elderly studies based on EHR than in face-to-face studies, es- in Catalonia [16]. Patients institutionalized in private ge- pecially in the early stages of AD. In face-to-face riatrics or residences might lose contact with their gen- studies, patients affected by subclinical symptoms or eral practitioner, increasing under-registration of EHR. early stages of AD might likely be identified as AD Moreover, patients with vascular comorbidities who had cases because they are evaluated by specialists during a diagnosis of AD and were diagnosed with another de- the study. But in studies based on EHR, it is possible mentia subtype (such as mixed dementia or vascular de- that many of these patients have not been cognitively mentia) were excluded from our study, because the assessed by the end of the study period, and there- algorithm we used only defined dementia cases as AD if fore, they are not identified as AD cases. Second, the this was the last diagnosis. Finally, certain aspects other analysis of free text has been reported to enhance the than under-registration could partially explain our lower identification of dementia cases [15]. But we used an estimates in the oldest elderly: some level of survival bias algorithm that identified AD cases exclusively based that could not be rejected and an attenuation of the rela- on codes. Therefore, those AD diagnoses recorded in tionship between AD diagnosis and clinical expression the clinical history using the free text instead of the of dementia in the oldest old, which has been pointed ICD-10 codes enhanced under-registration in our out in recent studies [49]. study. Third, patients diagnosed by a specialist from Our estimates were clearly higher than those in previ- private healthcare might not have the AD diagnosis ous studies based on EHR, suggesting that under- recorded in their clinical history, unless they (or their diagnosis in primary care settings probably varies relatives) inform the general practitioner about their strongly depending on the clinical practice. In the condition. Despite the universal coverage of the Cata- Catalan primary care, underestimation appeared to be lan public health system, about 25% of persons aged less frequent than in other primary care settings, which at least 60 years have contracted a private health in- reported that only half of the expected number of pa- surance. However, not all of them might have con- tients with dementia were recognized by the general tributed to under-registration in our study. Most of practitioner [14]. This could be partly explained because the private healthcare users also consult public the accuracy of AD diagnosis routinely collected in pri- healthcare professionals, especially among the elderly. mary care might improve in active dementia diagnosis The public health system provides free access or low centers; in our study settings, about 25% of patients copayment for pharmacological treatment prescribed without dementia and 75% of patients with dementia by general practitioners to all Catalan pensioners. In aged 65 years or over have taken a cognitive test [16]. our study, we identified AD cases combining data re- Additionally, about 90% of the AD prevalent cases re- corded in the primary care settings with data from corded in primary care had been evaluated by a special- the pharmacy-invoicing database. Therefore, we could ist, which suggests that general practitioners are identify as AD cases those patients attended by a pri- coordinated with the secondary care settings [17]. This vate specialist who asked their general practitioner for coordination is promoted by initiatives like an Ponjoan et al. Alzheimer's Research & Therapy (2020) 12:60 Page 7 of 9 asynchronous telemedicine program that establishes Conclusions protocols for the screening and diagnosis of dementia, The prevalence and incidence rates of AD based on thereby enabling primary care and specialized care pro- EHR from primary care were higher than other estimates fessionals to reach shared diagnoses of dementia [51]. Fi- based on primary care databases from other European nally, under-registration in our study was probably countries and slightly lower than the most recent face- moderate because we might have captured a high num- to-face studies. Despite some limitations, real-world data ber of AD cases among treated patients. In Catalonia, routinely collected in primary care settings could be about 68% of dementia patients receive anti-dementia applied to epidemiological studies to complement face- drugs [52], and this percentage may be higher than in to-face studies and could provide regular updates of other countries: in France, it is about 38% [53], and in prevalence and incidence estimates of AD at a reason- Germany, it is about 52% [54]. Furthermore, our defin- able cost. Better monitoring of the AD trends might help ition of AD case might have contributed to increase the adjust the national dementia plans and the allocation of number of AD cases identified by anti-dementia treat- resources for families and caregivers to the current num- ment. We included patients treated with anti-dementia ber of cases occurring in the community. Additionally, drugs at any time—that is, patients who were treated be- our results contribute to raise awareness about the im- fore the study period were also included—whereas other portance of the accuracy of EHR and might encourage studies accounted for currently treated patients—pa- general practitioners, nurses, and specialists to keep im- tients who were medicated in the past were excluded. proving the quality of the data they routinely collect in We acknowledge some limitations. First, our results the electronic health histories. It is time to use real- could be underestimated because we did not use free world data from the primary care settings to update AD text recorded in the clinical history, an accredited epidemiology complementing face-to-face studies, which method to identify persons with AD [15]. However, we might contribute to improve dementia policies and re- used a validated algorithm to identify AD cases [17] that source management for patients and their families. combined diagnosis with prescription or dispensing codes, involving not only data from general practitioners Supplementary information but also from a pharmacy-invoicing database. In Catalo- Supplementary information accompanies this paper at https://doi.org/10. nia, the prescription of anti-dementia drugs is requested 1186/s13195-020-00625-2. by the general practitioner but requires approval by cer- tain other specialists (geriatricians, psychiatrists, or neu- Additional file 1. Diagnostic and pharmacological codes. Description of data: Definition of comorbidities based on the International Classification rologists). The use of two different data sources of Diseases, 10th revision (ICD-10). minimized the possible under-recording of dementia Additional file 2. Prevalence of AD. Prevalence of Alzheimer’s disease in diagnoses in primary care. Second, the use of certain the Catalan population in 2016 in different age-specific populations (≥ additional clinical parameters to diagnose AD could have 70, ≥ 75, ≥ 80 and ≥ 85 years old), by age groups and sex. increased the number of identified AD cases. However, Additional file 3. Incidence of AD. Description of data: incidence rate of Alzheimer’s disease in the Catalan population in 2016 in different age- we prioritized the use of a simple algorithm that was specific populations (≥ 70, ≥ 75, ≥ 80 and ≥ 85 years old), by age groups validated and had a good positive predictive value and sex. (72.3%, 95%CI 70.7–73.9), and sensitivity (83.3, 95%CI 81.8–84.6) [17]. Third, we did not account for educa- Acknowledgements tional level, a risk factor for AD, because such informa- We thank Elí Carlomagno León Paz for sharing with us his knowledge about tion was not available in SIDIAP. Fourth, AD is one of the electronic medical history. the many dementia disorders, and epidemiological trends might vary between dementia subtypes, for ex- Authors’ contributions ample, decreasing trends of overall dementia have been AP, RR, MGG, and JGO participated in the conceptualization and study reported due to the reduction of vascular dementia but design. RML and LAC participated in the data acquisition. JB and MCC conducted the data management and statistical analyses. AP and LAC wrote not of AD [49]. Thus, further details on the epidemi- the first manuscript draft EF and DP lead the interpretation of the results. All ology of other dementia subtypes are crucial to plan ef- authors revised and contributed to improve several drafts. Finally, all authors fective preventive measures against dementia. read and approved the final manuscript. The main strength of our study is our large sample size (including more than 1 million persons aged at least Funding This work was supported by the Real world Outcomes across the AD 65 years old, and 40,000 dementia cases), which en- spectrum for better care: Multi-modal data Access Platform (ROADMAP) from hanced external validity; SIDIAP contains high-quality the Innovative Medicines Initiative (Grant Agreement number 116020); by data from about 80% of the Catalan population and good clinical research grants from Carlos III Health Institute, within the Net for Re- search in Preventive Activities and Health Enhancement (RedIAPP RD16/ representativeness in terms of geographical, age, and sex 0007/0004) framework; and by the Agency for Management of University distributions [25]. and Research Grants (2017 SGR 1146). Ponjoan et al. Alzheimer's Research & Therapy (2020) 12:60 Page 8 of 9 Availability of data and materials 13. Happich M, Kirson NY, Desai U, King S, Birnbaum HG, Reed C, et al. Excess The datasets generated during and/or analyzed during the current study are costs associated with possible misdiagnosis of Alzheimer’s disease among not publicly available due to legal reasons related to data privacy protection. patients with vascular dementia in a UK CPRD population. J Alzheimers Dis. 2016;53:171–83. Ethics approval and consent to participate 14. Connolly A, Gaehl E, Martin H, Morris J, Purandare N. Underdiagnosis of This is study was approved by the Clinical Research Ethics Committee from dementia in primary care: variations in the observed prevalence and the Primary Care Research Institute Jordi Gol (IDIAPJGol). Written informed comparisons to the expected prevalence. Aging Ment Health. 2011;15:978–84. consent was not required because patients were not involved. All data was 15. Wilkinson T, Ly A, Schnier C, Rannikmäe K, Bush K, Brayne C, et al. obtained from the SIDIAP database which contains anonymized longitudinal Identifying dementia cases with routinely collected health data: a medical records routinely collected in primary care settings. systematic review. Alzheimers Dement. 2018;14:1038–51. 16. Ponjoan A, Garre-Olmo J, Blanch J, Fages E, Alves-Cabratosa L, Martí-Lluch R, et al. Epidemiology of dementia: prevalence and incidence estimates using validated Consent for publication electronic health records from primary care. Clin Epidemiol. 2019;4:217–28. Not applicable. 17. Ponjoan A, Garre-Olmo J, Blanch J, Fages E, Alves-Cabratosa L, Martí-Lluch R, et al. How well can electronic health records from primary care identify Competing interests Alzheimer’s disease cases? Clin Epidemiol. 2019;11:509–18. The authors declare that they have no competing interests. 18. Lobo A, Saz P, Marcos G, Día JL, de la Cámara C, Ventura T, et al. Revalidation and standardization of the cognition mini-exam (first Spanish Author details version of the Mini-Mental Status Examination) in the general geriatric 1 Vascular Health Research Group (ISV-Girona), Jordi Gol Institute for Primary population. Med Clin (Barc). 1999;112:767–74. Care Research (IDIAPJGol), Barcelona, Catalonia, Spain. 2Girona Biomedical 19. Martínez de la Iglesia J, Dueñas Herrero R, Onís Vilches MC, Aguado Research Institute (IDIBGi), Girona, Catalonia, Spain. 3Autonomous University Taberné C, Albert Colomer C, Luque Luque R. Spanish language of Barcelona, Bellaterra (Cerdanyola del Vallès), Catalonia, Spain. 4Primary Care adaptation and validation of the Pfeiffer’s questionnaire (SPMSQ) to Services, Catalan Institute of Health (ICS), Girona, Catalonia, Spain. detect cognitive deterioration in people over 65 years of age. Med Clin 5 Department of Medical Sciences, School of Medicine, Campus Salut, (Barc). 2001;117:129–34. University of Girona, Girona, Catalonia, Spain. 6IDIAPJGol, c/ Maluquer 20. Baztán J, Pérez Del Molino J, Alarcon T, San Cristóbal E, Izquierdo G, Salvador, 11 baixos, 17002 Girona, Catalonia, Spain. Manzarbeitia I. Indice de Barthel: instrumento válido para la valoración funcional de pacientes con enfermedad cerebrovascular. Rev Esp Geriatr Received: 8 November 2019 Accepted: 1 May 2020 Gerontol. 1993;28:32–40. 21. Lawton M, Brody E. Assessment of older people: self-maintaining and instrumental activities of daily living. Gerontologist. 1969;9:179–86. References 22. Alvarez Solar M, de Alaiz Rojo AT, Brun Gurpegui E, Cabañeros Vicente JJ, 1. Ferri CP, Prince M, Brayne C, Brodaty H, Fratiglioni L, Ganguli M, et al. Global Calzón Frechoso M, Cossío Rodríguez I, García López P, García-Cañedo prevalence of dementia: a Delphi consensus study. Lancet. 2005;366:2112–7. Fernández R, Pardo González I, et al. Capacidad funcional de pacientes 2. Lobo A, Saz P, Marcos G, Dia JL, De-la-Camara C, Ventura T, et al. Prevalence mayores de 65 años, según el Indice de Katz. Fiabilidad del método of dementia in a southern European population in two different time Atención Primaria. 1992;10:812–5. periods: the ZARADEMP Project. Acta Psychiatr Scand. 2007;116:299–307. 23. Blessed G, Tomilson B, Roth M. The association between quantitative 3. Roehr S, Pabst A, Luck T, Riedel-Heller SG. Is dementia incidence declining measures of dementia and of senile change in the cerebral grey matter of in high-income countries? A systematic review and meta-analysis. Clin elderly subjects. Br J Psychiatry. 1968;114:797–811. Epidemiol. 2018;10:1233–47. 24. Bolíbar B, Fina Avilés F, Morros R, Garcia-Gil MM, Hermosilla E, Ramos R, 4. Wu Y-T, Fratiglioni L, Matthews FE, Lobo A, Breteler MMB, Skoog I, et al. et al. SIDIAP database: electronic clinical records in primary care as a source Dementia in western Europe: epidemiological evidence and implications for of information for epidemiologic research. Med Clin (Barc). 2012;138:617–21. policy making. Lancet Neurol. 2016;15:116–24. 25. Del Mar G-GM, Hermosilla E, Prieto-Alhambra D, Fina F, Rosell M, Ramos R, 5. Martin Prince A, Wimo A, Guerchet M, Gemma-Claire Ali M, Wu Y-T, et al. Construction and validation of a scoring system for the selection of Prina M, et al. World Alzheimer Report 2015 the global impact of high-quality data in a Spanish population primary care database (SIDIAP). dementia an analysis of prevalence, incidence, cost and trends. 2015 Inform Prim Care. 2011;19:135–45. [cited 2018 Oct 14]; Available from: https://www.alz.co.uk/research/ 26. Recalde M, Manzano-Salgado C, Díaz Y, Puente D, Garcia-Gil del Mar M, WorldAlzheimerReport2015.pdf. Marcos-Gragera R, et al. Validation of cancer diagnoses in electronic health 6. Tola-Arribas MA, Yugueros MI, Garea MJ, Ortega-Valín F, Cerón-Fernández A, records: results from the Information System For Research In Primary Care Fernández-Malvido B, et al. Prevalence of dementia and subtypes in Valladolid, (SIDIAP) in northeast Spain. Clin Epidemiol. 2019;11:1015–24. Northwestern Spain: the DEMINVALL study. PLoS One. 2013;8:e77688. 27. Forns J, Cainzos-Achirica M, Hellfritzsch M, Morros R, Poblador-Plou B, Hallas 7. Virués-Ortega J, de Pedro-Cuesta J, Vega S, Seijo-Martínez M, Saz P, Rodríguez J, et al. Validity of ICD-9 and ICD-10 codes used to identify acute liver injury: F, et al. Prevalence and European comparison of dementia in a ≥75-year-old a study in three European data sources. Pharmacoepidemiol Drug Saf. 2019; composite population in Spain. Acta Neurol Scand. 2011;123:316–24. 28:965–75. 8. de Pedro-Cuesta J, Virués-Ortega J, Vega S, Seijo-Martínez M, Saz P, 28. Martinez-Laguna D, Soria-Castro A, Carbonell-Abella C, Orozco-López P, Estrada- Rodríguez F, et al. Prevalence of dementia and major dementia subtypes in Laza P, Nogues X, et al. Validation of fragility fractures in primary care electronic Spanish populations: a reanalysis of dementia prevalence surveys, 1990- medical records: a population-based study. Reumatol Clínica. 2019;15:e1–4. 2008. BMC Neurol. 2009;9:55. 29. Ramos R, Balló E, Marrugat J, Elosua R, Sala J, Grau M, et al. Validity for use 9. Lobo A, Lopez-Anton R, Santabárbara J, De-la-Cámara C, Ventura T, in research on vascular diseases of the SIDIAP (Information System for the Quintanilla MA, et al. Incidence and lifetime risk of dementia and Development of Research in Primary Care): the EMMA study. Rev Española Alzheimer’s disease in a Southern European population. Acta Psychiatr Cardiol. 2012;65:29–37. Scand. 2011;124:372–83. 30. STAMPFER MJ. Cardiovascular disease and Alzheimer’s disease: common 10. Bermejo-Pareja F, Benito-León J, Vega S, Medrano MJ, Román GC. Incidence links. J Intern Med. 2006;260:211–23. and subtypes of dementia in three elderly populations of central Spain. J 31. Jeong S-M, Shin DW, Lee JE, Hyeon JH, Lee J, Kim S. Anemia is associated Neurol Sci. 2008;264:63–72. with incidence of dementia: a national health screening study in Korea 11. Imfeld P, Brauchli Pernus YB, Jick SS, Meier CR. Epidemiology, co-morbidities, involving 37,900 persons. Alzheimers Res Ther. 2017;9:94. and medication use of patients with Alzheimer’s disease or vascular 32. Alzheimer’s Association. 2014 Alzheimer’s disease facts and figures. dementia in the UK. J Alzheimers Dis. 2013;35:565–73. Alzheimers Dement. 2014;10:e47–92. 12. Imfeld P, Bodmer M, Jick SS, Meier CR. Proton pump inhibitor use and risk 33. Diniz BS, Butters MA, Albert SM, Dew MA, Reynolds CF. Late-life depression and of developing Alzheimer’s disease or vascular dementia: a case–control risk of vascular dementia and Alzheimer’s disease: systematic review and meta- analysis. Drug Saf. 2018;41:1387–96. analysis of community-based cohort studies. Br J Psychiatry. 2013;202:329–35. Ponjoan et al. Alzheimer's Research & Therapy (2020) 12:60 Page 9 of 9 34. R Development Core Team. R: a language and environment for statistical computing [Internet]. Found. Stat. Comput. Vienna, Austria. 2018 [cited 2018 Sep 8]. Available from: https://www.r-project.org/. 35. Perera G, Pedersen L, Ansel D, Alexander M, Arrighi HM, Avillach P, et al. Dementia prevalence and incidence in a federation of European Electronic Health Record databases: the European Medical Informatics Framework resource. Alzheimers Dement. 2018;14:130–9. 36. van Bussel EF, Richard E, Arts DL, Nooyens ACJ, Coloma PM, de Waal MWM, et al. Dementia incidence trend over 1992-2014 in the Netherlands: analysis of primary care data. PLOS Med. 2017;14:e1002235. 37. Lu N, Dubreuil M, Zhang Y, Neogi T, Rai SK, Ascherio A, et al. Gout and the risk of Alzheimer’s disease: a population-based, BMI-matched cohort study. Ann Rheum Dis. 2016;75:547–51. 38. Abdulrahman GOJ Jr. Alzheimer’s disease: current trends in Wales. Oman Med J. 2014;29:280–4. 39. Jaakkimainen RL, Bronskill SE, Tierney MC, Herrmann N, Green D, Young J, et al. Identification of physician-diagnosed Alzheimer’s disease and related dementias in population-based administrative data: a validation study using family physicians’ electronic medical records. J Alzheimer’s Dis. 2016;54:337–49. 40. Fiest KM, Roberts JI, Maxwell CJ, Hogan DB, Smith EE, Frolkis A, et al. The prevalence and incidence of dementia due to Alzheimer’s disease: a systematic review and meta-analysis. Can J Neurol Sci. 2016;43:S51–82. 41. Niu H, Álvarez-Álvarez I, Guillén-Grima F, Aguinaga-Ontoso I. Prevalence and incidence of Alzheimer’s disease in Europe: a meta-analysis. Neurol (English Ed). 2017;32:523–32. 42. López-Pousa S, Vilalta-Franch J, Llinàs-Regla J, Garre-Olmo J, Román GC. Incidence of dementia in a rural community in Spain: the Girona Cohort Study. Neuroepidemiology. 2004;23:170–7. 43. Lobo A, Launer LJ, Fratiglioni L, Andersen K, Di Carlo A, Breteler MM, et al. Prevalence of dementia and major subtypes in Europe: a collaborative study of population-based cohorts. Neurology. 2000;54:S4–9. 44. Fernández M, Castro-Flores J, Perez-de las Heras S, Mandaluniz-Lekumberri A, Gordejuela M, Zarranz J. Prevalence of dementia in the elderly aged above 65 in a district in the Basque Country. Rev Neurol. 2008;46:89–96. 45. Gavrila D, Antúnez C, Tormo MJ, Carles R, García Santos JM, Parrilla G, et al. Prevalence of dementia and cognitive impairment in southeastern Spain: the Ariadna study. Acta Neurol Scand. 2009;120:300–7. 46. Gascón-Bayarri J, Reñé R, Del Barrio JL, De Pedro-Cuesta J, Ramón JM, Manubens JM, et al. Prevalence of dementia subtypes in El Prat de Llobregat, Catalonia, Spain: the PRATICON study. Neuroepidemiology. 2007;28:224–34. 47. García García FJ, Sánchez Ayala MI, Pérez Martín A, Martín Correa E, Marsal Alonso C, Rodríguez Ferrer G, et al. The prevalence of dementia and its main subtypes in subjects older than 65 years: impact of occupation and education. The Toledo Study. Med Clin (Barc). 2001;116:401–7. 48. Park JE, Kim B-S, Kim KW, Hahm BJ, Sohn JH, Suk HW, et al. Decline in the incidence of all-cause and Alzheimer’s disease dementia: a 12-year-later rural cohort study in Korea. J Korean Med Sci. 2019;34:e293. 49. Farfel JM, Yu L, Boyle PA, Leurgans S, Shah RC, Schneider JA, et al. Alzheimer’s disease frequency peaks in the tenth decade and is lower afterwards. Acta Neuropathol Commun. 2019;7:104. 50. Skinner TR, Scott IA, Martin JH. Diagnostic errors in older patients: a systematic review of incidence and potential causes in seven prevalent diseases. Int J Gen Med. 2016;9:137–46. 51. Vilalta-franch J, Garre-olmo J, López-pousa S, Llinàs-reglà J, Calvó-perxas L, Cubí-montfort R. Viabilidad de un sistema de telemedicina de soporte para el diagnóstico de demencia en atención primaria. Revista de Neurologia. 2012;55:263–9. 52. Turró-Garriga O, Calvó-Perxas L, Albaladejo R, Alsina E, Cuy JM, Llinàs-Reglà J, et al. Pharmaceutical consumption and cost in patients with dementia: a longitudinal study by the Registry of Dementias of Girona (ReDeGi) in Catalonia (Spain). Arch Gerontol Geriatr. 2015;60:448–52. 53. François M, Sicsic J, Pelletier-Fleury N. Determinants of antidementia drug prescription in patients older than 65: a latent class analysis. Pharmacoepidemiol Drug Saf. 2020;pds.4921. 54. Wübbeler M, Wucherer D, Hertel J, Michalowsky B, Heinrich S, Meyer S, et al. Antidementia drug treatment in dementia networks in Germany: use rates and factors associated with treatment use. BMC Health Serv Res. 2015;15:1–8. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.