Javascript is currently disabled in your browser. When javascript is disabled, some functions of this website will not work.

Open access for scientific and medical research

From submission to the first editing decision.

From editor acceptance to publication.

The above percentage of manuscripts have been rejected in the past 12 months.

Open access to peer-reviewed scientific and medical journals.

Dove Medical Press is a member of OAI.

Batch reprints for the pharmaceutical industry.

We provide real benefits for authors, including fast processing of papers.

Register your specific details and specific drugs of interest, and we will match the information you provide with articles in our extensive database and send you a PDF copy via email in a timely manner.

Back to Journal »Clinical Epidemiology» Volume 10

Author Sediq R, van der Schans J, Dotinga A, Alingh RA, Wilffert B, Bos JHJ, Schuiling-Veninga CCM, Hak E 

Published on August 16, 2018, Volume 2018: 10 pages 981-989

DOI https://doi.org/10.2147/CLEP.S163037

Single anonymous peer review

Editor who approved for publication: Dr. Vera Ehrenstein

Rahmat Sediq, 1 Jurjen van der Schans, 1 Aafje Dotinga, 2 Rolinde A Alingh, 2 Bob Wilffert, 1, 3 Jens HJ Bos, 1 Catharina CM Schuiling-Veninga, 1 Eelko Hak 1, 4 1 Drug therapy, epidemiology and economics Faculty, University of Groningen, Groningen Institute of Pharmacy, Groningen, Netherlands; 2Lifelines Cohort Study, Lifelines Databeheer BV, Groningen, Netherlands; 3 Department of Clinical Pharmacy and Pharmacology, Groningen University Medical Center, Groningen, Netherlands; 4 Department of Epidemiology, University of Groningen Medical Center, Groningen, Netherlands. Background: Although self-reported data is often used as a source of drug use for drug epidemiological studies, such information is prone to bias. Several previous studies have shown that various factors such as age, drug type, and data collection method may affect accuracy. We aimed to assess the consistency of self-reported drug use recorded at the entrance of a lifeline cohort study (a three-generation follow-up study that started in the Netherlands in 2006 and included more than 167,000 participants). Materials and methods: As part of the PharmLines Initiative, we collected drug data coded according to the Anatomical Therapeutic Chemistry (ATC) coding scheme from Lifeline participants, and linked the data to universities’ widely used and representative pharmacies through Statistics Netherlands The prescription database Groningen, IADB.nl. The analysis is performed at the second level of the ATC code for all recorded drugs and the top list of the most commonly used drugs at the fifth level of the drug specific. Cohen's kappa statistics are used to measure the consistency of all participants according to gender and age. Result: The level of agreement between the two data sources varies by treatment category. Drugs for the cardiovascular system and diabetes, thyroid therapy, bisphosphonates and antithrombotic drugs showed very good consistency (κ>0.75). Drugs used on-demand or prone to stigma bias showed moderate consistency (κ=0.41–0.60), while short-term drugs showed moderate consistency (κ=0.0–0.4). The consistency of men and women is similar, but the consistency rate of young people is often lower than that of the elderly. Conclusion: The self-report method is effective for capturing chronic drug use that is popular at a certain moment, but it is not effective for drugs used for a short period of time. Gender has no effect on consistency, and the effect of age needs more research. Future drug epidemiological studies are best to combine the two data sources to achieve the highest accuracy of drug exposure rates. Keywords: self-report data, prescription data, pharmacy records, agreement, questionnaire, medication, interview, validity

The lifeline cohort study began in 2006 to study the risk factors for the long-term development of the disease and use self-reported data to conduct "real-world" drug epidemiological studies to determine drug exposure. 1,2 However, self-reported data are prone to recall bias, which may lead to misclassification of large (non-)differences in actual drug exposure, and therefore may lead to underestimation or overestimation of actual drug use and effects. 3-5 Due to its high accuracy, electronic medication records (EMR) are often regarded as the gold standard compared to self-reported medication usage. 6 In contrast, self-reported data contains information about the drugs actually used by the respondent and may also provide information about excessive drug use. -Over the counter (OTC) medicines and herbs. 7 To date, few studies have evaluated the consistency of self-reported drug use compared with EMR, with inconsistent results. 8-15 These studies report d In most cases, there is good agreement between self-reported drug use and electronic records, but the agreement varies by treatment group. For example, in some studies, for long-term drugs, such as cardiovascular drugs, diabetes drugs, anticoagulation-related drugs, and hormone drug treatments, the consistency between pharmacy records and self-reports is very high. 8-11 In contrast, lower coincidence rates were observed for local, required, or short-time medications. 7,12,13 Studies in various environments reported that with age, the level of recall decreases, and gender does not seem to affect recall of drug use. 14,15

In the PharmLines Initiative, we first conducted a cross-sectional descriptive study to evaluate self-reports recorded when entering the Dutch Lifeline Cohort study by comparing this information with extensive research and representative University of Groningen pharmacy prescriptions Consistency database IADB.nl for drug use. IADB.nl is considered the "gold standard" for these analyses (see the description in the "Materials and Methods" section). In addition, we also aim to describe the effect of drug type, age, and gender on the consistency rate.

In 2017, the Department of Epidemiology and Clinical Pharmacy and the Department of Pharmacology of the University of Groningen Medical Center, the Groningen Institute of Pharmacy and the Lifeline Cohort Study (see https://www.lifelines.nl/researcher/about-lifelines) started the PharmLines plan Data from the Lifeline Cohort Study is linked to the prescription database IADB.nl of the University of Groningen. The Lifeline Cohort study was launched in the northern part of the Netherlands from 2006 to 2013 as an academic resource for researchers worldwide. The study design involves observational follow-up studies with long-term prospective measurements in a large number of populations, aiming to explore the interaction of genetic and environmental factors in the development of multifactorial diseases. 16 In short, since all residents of the Netherlands are working as general practitioners (GP), eligible participants are invited to participate in the lifeline cohort study through their GP. A large number of general practitioners in the three northern provinces of the Netherlands (Friesland, Groningen, and Drenthe) participated and invited all patients between 25 and 50 years old, unless the participating general practitioners deemed the patient ineligible according to the following criteria: Severe mental or physical illness; limited life expectancy (<5 years); or insufficient knowledge of Dutch to complete the Dutch questionnaire. Subsequently, individuals interested in participating received detailed information about the lifeline cohort study and informed consent via email. After Lifelines received the signed informed consent form, participants received a baseline questionnaire and an invitation to conduct a comprehensive health assessment at the Lifelines research site. More than 167,000 people from the three northern provinces of the Netherlands participated in this three-generation study. During the competition, participants completed a number of questionnaires, received medical examinations, and collected and stored biological samples including DNA. The methods used to obtain drug use information are as follows: 1) ask the participants about their drug use through questionnaires and 2) bring the drugs they have used during the interview. Then all medications are recorded by the doctor's assistant. All prescription drugs are classified according to the Anatomical Therapeutic Chemistry (ATC) coding scheme. 17

In order to validate this self-reported lifeline cohort study drug database in the cross-sectional study design, we compared it with the pharmacy prescription database of the University of Groningen IADB.nl, which we regarded as the "gold standard". The IADB.nl database is an evolving pharmacy database. In 2013, it contained more than 60 public pharmacies and contained prescription data from approximately 600,000 patients in the northern part of the Netherlands. Coverage accounts for about 20% of residents in northern provinces. Except for registration at participating pharmacies, no specific inclusion or exclusion criteria are applied. In the Netherlands, all patients are registered in the pharmacy regardless of their health insurance status. It is found that the prevalence of drug use is representative of the entire population of the country. 18 Since 1994, prescription data has been collected for nearly 20 years, and each patient has been uniquely tracked in the system. The prescription record includes information about the drug, such as the date and quantity of the drug dispensed, the dose of the drug administered, and the ATC code. Data is stored anonymously, and all patients receive an anonymous patient code. You can use demographic information, such as date of birth and gender. The database provides almost complete drug records, with the exception of over-the-counter drugs and drugs dispensed during hospitalization. 18

PharmLines Initiative: Connecting and researching people

In the PharmLines Initiative, both databases from Lifelines Cohort Study and IADB.nl are linked by the Statistics Netherlands (CBS), which acts as a third-party trusted party at the patient level and uses deterministic links based on uniquely identifiable information. IADB.nl is a longitudinal database. If the information about the patient changes, it will create a new patient record. For example, if the patient moves to another address, the new zip code and all patient information will be stored in the new record. All these patient records are uploaded to CBS for linking. The link between the IADB and the CBS vertical database is completed in four steps. The first linkage was carried out by combining gender and date of birth on all 6 characters of the postal code. In the second link, the patient record with the same IADB number as the linked patient is added to the linked patient. The total connection of these two steps gives 48.3% connections. The third link is a combination of gender and date of birth on the 4 digits of the postal code. Similarly, only the unique combination of CBS personnel number and IADB.nl number was selected. In the fourth link, the patient record with the same IADB.nl number as the linked patient is added to the linked patient. The total link for these last two steps gives 42.8% of links. The total link rate of all uploaded records is 90.1%. After the link was completed, CBS deleted all unique patient identifiers. Then provide each participant with a unique identifier code so that each included ATC code can be compared with the patient-level medication usage. The analysis was done in a password-protected electronic environment provided by CBS.

The study population is limited to all adults (≥18 years old) who have recorded baseline information in the Lifelines Cohort Study database and the IADB.nl database. The participant’s registration date in the IADB.nl database is at least 6 months before the baseline measurement of each participant’s lifeline. This restriction ensures that a different time window can be selected for the IADB.nl data, while the study population will remain the same.

Research parameters and data analysis

Linked information related to this study includes anonymous identifier code, ATC code, date of dispensing, date of entry into the IADB database, gender and age when the cohort was first accessed. The first interview with the lifeline cohort study participants was considered a baseline measurement, and only the baseline measurement (entry phase) data was used to compare the two databases.

All drugs grouped in the second level of the ATC code were examined in the study. In addition to the secondary ATC code, some chemical grade specific drugs were also included in the study. A list of the most commonly used drugs in the Netherlands, including omeprazole, psyllid seeds, polyethylene glycol, calcium, hydrochlorothiazide, metoprolol, enalapril, simvastatin, ketoconazole, triamcinolone, Clobetasol, levothyroxine, oxazepam, temazepam, paroxetine, salbutamol, fluticostatin also checked salmeterol/fluticasone, desloratadine, artificial tears, approximately carbonate, Diclofenac and ibuprofen.

The SQL Server program was used to compare the medication records of each participant. The data obtained is then classified into true positive, true negative, false positive (FP) and false negative (FN). These values ​​are given in the cross table and used to calculate the agreement. Each of the four cells in these crosstabs requires at least 5 participants. If the number of participants in one or more cells is less than 5, the medication group or specific medication is ignored.

To address the possible root causes of low consistency, FN and FP were checked. Overreporting represents the number of FPs. This means that some participants reported using a certain drug group, while the prescription was not registered in the pharmacy record. On the other hand, underreporting represents the number of FNs. This means that the participant did not report the use of a certain drug group, and at the same time registered the prescription in the pharmacy record. If a drug group or a specific drug shows poor consistency and high overreport rate, it may indicate that a database with self-reported data can better record the use of that specific drug (or drug group).

Participants were also stratified according to age and gender to study the influence of these factors on the consistency rate. This only applies to the most commonly used drugs with the lowest level 5 ATC code (chemical substance specific code) selected.

To determine whether participants reported on drugs that they had not yet started using during the interview but started immediately before or after the interview, we defined various overlapping time windows. The effect of overlapping time windows was checked, so a sensitivity analysis was performed on 6 different time windows. In order to determine the impact of different time windows on the database comparison results, time windows of 30 days, 90 days, 105 days, 120 days, and 180 days before the baseline measurement were used. In addition to the previous time window, a time window of 90 days before and 2 weeks after the baseline measurement (called 14-90 days) was also selected.

If there is no drug recorded, the drug data fields from the Lifelines Cohort Study database and IADB.nl are blank, if there is an ATC code, it exists. Therefore, if the record is blank, we cannot be sure that it is missing and assume that there is no drug. Cohen's kappa statistic is used to measure the agreement rate, and pharmacy prescription data is regarded as the gold standard. Cohen's kappa's 95% CI is calculated using standard errors. 19 Cohen's kappa value is interpreted according to the guidelines proposed by Altman et al.: poor (<0.20), fair (0.20-0.40), moderate (0.41-0.60), good (0.61-0.80), very good (0.81-1.00). 20 Since the statistical comparison of kappa measures has some limitations, we decided to explain the 95% CI of Cohen's kappa value. If these do not overlap for each comparison group, then the comparison is considered to indicate a significant difference. Use IBM SPSS software, version 25, 2017 for statistical analysis.

After the two databases are connected, the total overlap includes 45,000 adult lifeline participants. For current comparisons that apply the strict eligibility criteria described above, 16,367 Lifeline Participants who meet the eligibility criteria can be checked. Of the 16,367 people, 64% were women, and the average age of the included population was 44 years old.

In the second level of ATC coding, the agreement between the two data sources ranges from poor to very good for different drug groups (see Table 1). The lowest agreement was found between the two data sets of gynecological anti-infectives (kappa [k] = 0.09; 95% CI: 0.04–0.15), while the best agreement was found in levothyroxine (H03AA01) Sex (k = 0.84; 95% CI: 0.81–0.86, Table 2). In general, long-term drugs show good to very good agreement between the two data sources, including antidiabetic drugs, cardiovascular system drugs, antithrombotic drugs, psychostimulants, obstructive airway disease drugs, Thyroid treatment, bisphosphonates and anti-gout agents. Surprisingly, some drug groups that may also be used for a long time also show moderate consistency, such as immunosuppressants, antitumor drugs, and cardiac therapies, although acute short-term use (<3 months) may be among participants very common.

Table 1 Overview of all secondary ATC codes, including their respective kappa values ​​and 95% CI

Note: In order to gain insight into the reasons for the respective high/low kappa values, the number of people who overreported or underreported drug use will also be displayed. The data displayed is based on a 90-day time window. Bold numbers indicate that the drug shows good and very good consistency (0.61 and more).

Abbreviation: ATC, anatomical treatment chemical.

Table 2 An overview of all included fifth-level ATC codes, including their respective kappa values ​​and 95% CI

Note: In order to gain insight into the reasons for the respective high/low kappa values, the number of people who overreported or underreported drug use will also be displayed. The data displayed is based on a 90-day time window. Bold numbers indicate that the drug shows good and very good consistency (0.61 and more).

Abbreviation: ATC, anatomical treatment chemical.

Moderate agreement (k = 0.41–0.60) is mainly seen in the drugs most commonly used for short-term medication. This has been found to be used in anti-anemia, antihistamines, analgesics and nasal preparations. Short-term or topical medications, such as anti-acne drugs, antifungal drugs, antibacterial drugs, triamcinolone acetonide, antifungal drugs, and antibiotics, showed poor consistency (k = 0-0.40).

It is worth noting that, except for cardiac therapy (C01) and vascular protection (C05), which showed consistency of k = 0.60 and k = 0.21, respectively, all included cardiovascular ATC groups showed consistency of k = 0.61 or higher.

In Figure 1, the results of different time windows for the IADB database are shown. For each secondary ATC code, the 30-day time window resulted in a significant reduction in consistency between the two data sources. The other time windows showed no significant differences. A detailed overview of the results of different time windows based on all secondary ATC codes can be found in Appendix 1-6.

Figure 1 Kappa values ​​for different time windows (30 days, 90 days, 14-90 days, 105 days, 120 days, and 180 days) and the 95% CI corresponding to each secondary ATC code.

Abbreviation: ATC, anatomical treatment chemical.

In Figure 2, the effect of age on the consistency between the two data sources is shown. The agreement rate tends to increase with age, but in most cases, the 95% confidence interval shows overlap. The only significant difference in kappa values ​​was found in the age groups 18-35 and over 65. This difference can only be observed in levothyroxine, calcium mesylate, diclofenac and omeprazole. The topical drugs did not show a significant difference. Detailed results of the effect of participant age on all consistent medications can be found in Appendix 9-12.

Figure 2 shows the Kappa value and the corresponding 95% CI of the selected drug.

Note: Participants are divided into four age groups; 18–35, 36–45, 46–64, and 65 years and older. For 9 types of drugs, there are less than 5 data per cell to calculate kappa (see Materials and Methods).

Figure 3 shows the results of the top list of the most commonly used drugs in the population. The Kappa values ​​for different drug classes show that the coincidence rates between men and women overlap, which indicates that there is no significant difference. Detailed results of the effect of participant gender on the consistency rate of all study types of drugs can be found in Appendices 7 and 8.

Figure 3 Kappa values ​​stratified by gender.

Note: The effect of gender on the consistency between data sources for the selection of drugs has been clarified. The error bar shows 95% CI. For 8 types of drugs, there are less than 5 data per cell to calculate Kappa (see Materials and Methods).

This involves a descriptive study whose main purpose is to describe the consistency between self-reports and prescription drug data of the unselected population participating in the lifeline cohort study, and to explore new relationships in the data. In general, we found that the agreement rate between the two data sources varies with treatment category and age. For example, drugs used in the cardiovascular system and diabetes, thyroid therapy, antithrombotic drugs, and obstructive airway disease drugs show good to very good consistency (k>0.60). This indicates that the drugs used for long-term use are reliably self-reported by the patients when entering the cohort, and this finding is consistent with earlier studies. 8-11

Contrary to our expectations, some long-term used drugs, such as anti-tumor drugs and immunosuppressants, only show moderate kappa values. Since these drugs are used to treat serious diseases and are usually used for a long time, it is expected that there will be a high degree of consistency between the two databases. 21 One possible explanation for the moderate kappa value is that anti-tumor drugs and immunosuppressants are usually started and dispensed in hospitals, and the use of such drugs may not be accurately registered in the IADB.nl database. The fact that these drugs are largely overestimated also supports this. Therefore, self-reported data is needed next to the EMR to obtain more accurate information about the exposure rate of such drugs.

Drugs designed for on-demand use generally showed moderate (k=0.41–0.60) consistency, while short-term drugs generally showed poor to general consistency (k=0–0.40). These drugs are taken less frequently or irregularly, so participants may underreport. 7,13

Dermatology showed poor to average (k=0–0.40) agreement overall. The reason for the low consistency is mainly due to insufficient reporting of drug use. This may be because patients do not regard dermatological drugs as drugs and therefore underreport the use of such drugs. 8

On the other hand, there are too many reports of drugs such as sex hormones, analgesics, antihistamines, and drugs for the treatment of acid-related diseases. These drugs are also over-the-counter drugs. Therefore, a significant portion of their use is not recorded in the pharmacy database, which suggests that self-reported data may be a better way to capture the use of such drugs. Unexpectedly, most cardiovascular drugs have also seen a certain degree of overreporting. We can only speculate on the reason, and more research is needed to find the reason for this discovery.

It is worth noting that the consistency of vasoprotective drugs and gynecological anti-infectives is poor. These medical drugs are used to treat diseases such as rectal fissures, hemorrhoids, and genital infections. We may just speculate that patients do not like to disclose information about such situations. Therefore, stigma bias may be the reason for the low consistency between the two data sources. The fact that these drugs are severely underestimated supports this theory.

We studied the effects of time window, gender, and age on the consistency rate. Methodologically, the time window is important for the consistency study to reach effective conclusions. In the Netherlands, long-term medications are prescribed every 3 months. Therefore, the 30-day time window resulted in a significant reduction in the consistency of long-term drug use. However, for short-term drugs, such as antifungals and antifungals, the results in the 30-day time window are not significantly different from other time windows. In fact, the 30-day time window results in a higher consistency of antibiotics and antiprotozoal drugs. The rest of the time windows are not significantly different from each other. The current use of the drug depends on the amount of drug prescribed at a certain point in time. An excessively large time window is not suitable to represent the participants’ current drug use. Most studies completed to date have used a time window of 90 days. Therefore, it is recommended to use a 90-day time window in future studies. Consistent with earlier studies, no effect of gender on consistency was found. However, contrary to earlier findings, the increase in age seems to lead to an increase in consistency. 14,15 Although the increase in age seems to go hand in hand with the increase in consistency, in most cases the 95% confidence interval overlaps. The included study population consisted of a relatively young population with an average age of 44 years, which may be due to age vs. drug Recall the reason for the difference. Compared with young people, older people generally use more drugs and have lower cognitive abilities, so the consistency between the two databases may be much lower for the older population. Larger population sizes may lead to smaller CIs and may help to better understand the impact of participant’s age on drug recalls.

Our research has some limitations. Although this is one of the largest studies to date on the consistency of self-reported data on drug use, the number of participants using certain drug categories is still small. Due to insufficient data, some ATC codes cannot be included. As the size of the IADB.nl database increases over time, the overlap between the two databases also increases. This makes it possible to include more participants in future research. In addition, if the record is blank, we cannot be sure that it is missing and assume that there is no drug. If anything, this will lead to potentially random misclassifications. Finally, a previous study showed that the adult lifeline cohort is representative of the Dutch adult population. However, drug use is closely related to culture, and the results may not be directly applicable to countries with other health care systems. twenty two

In summary, the consistency between the self-report database and the prescription database varies from treatment group to treatment group. The self-reporting method applied by Lifelines Cohort Study is suitable for capturing most drugs for long-term use and over-the-counter drugs. However, it is not suitable for recording on-demand medications, short-term medications, and/or topical medications. Stigma bias can have a significant impact on self-reporting of certain drug classes (such as vasoprotective drugs and gynecological anti-infectives). Gender does not affect consistency, and increasing age tends to lead to slightly better memories. However, more research on the effect of age on consistency is needed. Future drug epidemiological studies are best to combine the two data sources to achieve the highest accuracy of drug exposure rates.

Lifelines has been funded by many public resources, especially the Dutch government, the Dutch scientific research organization NWO (appropriation 175.010.2007.006), the Northern Province Cooperation Organization (SNN), the European Regional Development Fund, the Ministry of Economic Affairs of the Netherlands, and the German Delta Picken , Groningen and Drenthe provinces, the University of Groningen and the University of Groningen Medical Center in the Netherlands. Current research within IADB.nl and the PharmLines Initiative is funded by the Groningen Institute of Pharmacy, University of Groningen. The author wishes to thank Lifelines Cohort Study and all study participants for their services, as well as participating IADB.nl pharmacies for providing data for the study.

All authors contributed to data analysis, drafting and revision of the paper, and agreed to be responsible for all aspects of the work.

The authors report no conflicts of interest in this work.

Balder JW, de Vries JK, Nolte IM, Lansberg PJ, Kuivenhoven JA, Kamphuisen PW. Lipid and lipoprotein reference values ​​from 133,450 Dutch Lifeline participants: age and gender-specific baseline lipid values ​​and percentiles. J Clin Lipidol. 2017; 11(4): 1055-1064.

Nass KJ, van den Berg EH, Faber KN, Schreuder TCMA, Blokzijl H, Dullaart RPF. The high prevalence of apolipoprotein B dyslipoproteinemia in nonalcoholic fatty liver: a lifeline cohort study. metabolism. 2017; 72: 37-46.

West SL, Strom BL, Poole C, Kimmel SE. Validity of pharmacoepidemiology drugs and diagnostic data. In: Strom BL, editor. Drug epidemiology. Chichester: John Wiley & Sons Ltd; 2000:661-705.

Groves RM, Fowler FJ, Couper MP, etc. investigation method. Hoboken: John Wiley & Sons; 2004.

Jurek AM, Greenland S, Maldonado G, Church TR. Correctly explain the effects of non-differential misclassification: expectations and observations. International Journal of Epidemiology. 2005;34:680-687.

Schneeweiss S, Avorn J. A review of the use of health care utilization databases in therapeutic epidemiological research. J Clinical Epidemiology. 2005;58(4):323-337.

Nelson MW, Søndergaard B, Kjøller M, Hansen EH. The consistency between self-reported drug use data and prescription records varies by analysis method and treatment group. J Clinical Epidemiology. 2008; 61(9): 919-924.

Richardson K, Kenny RA, Peklar J, Bennett K. The agreement between prescription drug use for patients over 50 years of age and patient interview data recorded by the pharmacy varies by treatment group and reports indicating health conditions. J Clinical Epidemiology. 2013;66(11):1308-1316.

So L, Morgan SG, Quan H. Does the consistency between survey responses and administrative records vary depending on the ethnicity of prescription drugs? J Popul Ther Clin Pharmacol. 2012;19(2):e248-e258.

Wu C, Lai M, Shur-Fen Gau S, Wang S, Tsai H. Consistency between Taiwanese patient self-reports and claims data used by clinical diagnosis, drug use, and health systems. Public Science Library One. 2014; 9(12): e112257.

Haapea M, Miettunen J, Lindeman S, Joukamaa M, Koponen H. Agreement between self-reports on drug use and pharmacy data in the birth cohort in northern Finland in 1966. Int J method of psychiatric research. 2010;19(2):88-96.

Monster TB, Janssen WM, de Jong PE, de Jong-van den Berg LT; PREVEND research group to prevent kidney and vascular ENT stage diseases. Pharmaceutical data in epidemiological research: an easy-to-obtain and reliable tool. Pharmacoepidemiol drug Saf. 2002;11(5):379–384.

Gama H, Correia S, Lunet N. Questionnaire design and drug therapy recall: a systematic review. Pharmacoepidemiol drug Saf. 2009;18(3):175-187.

Van den Brandt PA, Petri H, Dorant E, Goldbohm RA, van de Crommert S. Comparison of questionnaire information on drug use and pharmacy data. Pharmacy Weekly. 1991;13(2):91-96.

West SL, Savitz DA, Koch G, Strom BL, Guess HA, Hartzema A. Recall accuracy of prescription drugs: comparison of self-report and database information. I am an epidemiological magazine. 1995;142:1103-1112.

Scholtens S, Smidt N, Swertz MA, etc. Cohort overview: LifeLines, a three-generation cohort study and biobank. International Journal of Epidemiology. 2015;44(4):1172–1180.

Who. ATC/DDD Index 2018. It can be obtained from the following website: https://www.whocc.no/atc_ddd_index/. Visited on September 13, 2017.

Visser ST, Schuling-Veninga CCM, Bos HJ, etc. Population-based prescription database IADB.nl: its development, usefulness and challenges in research results. Research on the results of expert Rev Pharmacoecon. 2013;13(3):285-292.

Reichenheim ME. The confidence interval for the kappa statistic. Stata J. 2004;4(4):421-428.

Ultraman DG. Practical statistics for medical research. London: Chapman and Hall; 1991

Caskie GI, Willis SL, Warner Schaie K, Zanjani FA. Consistency of drug information in brown bag data collection and pharmacy records: results of the Seattle longitudinal study. Exp Aging research. 2006;32(1):79-103.

Representatives of cohort studies of Klijs B, Scholtens S, Mandemakers JJ, Snieder H, Stolk RP, Smidt N. LifeLines. Public Science Library One. 2015;10(9):1-12.

This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and include the Creative Commons Attribution-Non-commercial (unported, v3.0) license. By accessing the work, you hereby accept the terms. The use of the work for non-commercial purposes is permitted without any further permission from Dove Medical Press Limited, provided that the work has an appropriate attribution. For permission to use this work for commercial purposes, please refer to paragraphs 4.2 and 5 of our terms.

Contact Us• Privacy Policy• Associations and Partners• Testimonials• Terms and Conditions• Recommend this site• Top

Contact Us• Privacy Policy

© Copyright 2021 • Dove Press Ltd • Software development of maffey.com • Web design of Adhesion

The views expressed in all articles published here are those of specific authors and do not necessarily reflect the views of Dove Medical Press Ltd or any of its employees.

Dove Medical Press is part of Taylor & Francis Group, the academic publishing department of Informa PLC. Copyright 2017 Informa PLC. all rights reserved. This website is owned and operated by Informa PLC ("Informa"), and its registered office address is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 3099067. UK VAT group: GB 365 4626 36

In order to provide our website visitors and registered users with services that suit their personal preferences, we use cookies to analyze visitor traffic and personalize content. You can understand our use of cookies by reading our privacy policy. We also retain data about visitors and registered users for internal purposes and to share information with our business partners. By reading our privacy policy, you can understand which of your data we retain, how to process it, with whom to share it, and your right to delete data.

If you agree to our use of cookies and the content of our privacy policy, please click "Accept".