Abstract
Background Tuberculosis (TB) in children and adolescents is a sentinel event for ongoing transmission. In the Netherlands, epidemiological characteristics of childhood and adolescent TB have not been fully evaluated. Therefore, we aimed to assess TB epidemiology within this population to provide guidance for TB elimination.
Methods A retrospective time-series analysis using national surveillance data from 1993–2018 was performed in children (aged <15 years) and adolescents (aged 15–19 years) with TB. Poisson regression models offset with log-population size were used to estimate notification rates and rate ratios. Trends in notification rates were estimated using average annual percentage changes (AAPC) based on the segmented linear regression analysis.
Results Among 3899 children and adolescents with TB notified during 1993–2018, 2418 (62%) were foreign-born (725 (41.3%) out of 1755 children and 1693 (78.9%) out of 2144 adolescents). The overall notification rate in children was 2.3 per 100 000 person-years, declining steadily during the study period (AAPC −10.9%, 95% CI −12.6–−9.1). In adolescents, the overall notification rate was 8.4 per 100 000 person-years, strongly increasing during 1993–2001 and 2012–2018. Compared to Dutch-born children and adolescents, substantially higher notification rates were observed among African-born children and adolescents (116.8 and 316.6 per 100 000 person-years, respectively). Additionally, an increasing trend was observed in African-born adolescents (AAPC 18.5%, 95% CI 11.9–25.5). Among the foreign-born population, those from countries in the horn of Africa contributed most to the TB caseload.
Conclusion TB notification rate among children was low and constantly declining across different demographic groups. However, heterogeneities were shown in adolescents, with an increasing trend in the foreign-born, particularly those from Africa.
Abstract
During 1993–2018, TB notification in children in the Netherlands declined steadily, but there was an increase of TB in foreign-born adolescents. Enhancing active case-finding through contact investigation and entry screening is needed to optimise TB control. https://bit.ly/36AjNhm
Introduction
Tuberculosis (TB) remains a major global health challenge, with an estimated 1.1 million new cases and 205 000 deaths in children aged <15 years in 2018 [1]. The risk of death is highest in children aged <5 years and among those with HIV co-infection not receiving antiretroviral therapy [2–5]. While TB is considered treatable and curable, management of childhood TB faces significant challenges, due to the lack of sensitive diagnostic tools, the lack of adaptable treatment approaches for both TB infection and TB disease, and poor access to child-friendly drug formulations with proven efficacy and limited toxicity [6–8]. Similarly, even with a significant burden of the disease [9], adolescents have been neglected in TB surveillance due to the fragmented approach of grouping all persons aged <15 years as children and those aged ≥15 years as adults, leaving little specific knowledge on TB epidemiology, prevention and management for this population [10–12].
In high-income countries, TB is still a concern, and a major cause of morbidity and mortality among migrant populations originating from high TB incidence countries [13]. Importantly, TB in children is a sentinel event reflecting recent transmission of Mycobacterium tuberculosis from adults [12]. In the Netherlands, the burden and trends in childhood TB notification rates during 1993–2012 by migration status (i.e. first-generation immigrant, native Dutch and second-generation immigrant) have been reported [14]. However, a complete picture of epidemiological characteristics in other demographic groups of children is lacking. Furthermore, adolescents with TB have not been considered as a specific group in any epidemiological study in the Netherlands, providing a gap in understanding the TB burden within this population. Our study aimed to assess the epidemiology of childhood and adolescent TB including national estimates of notification rates, rate ratios and trends in notification rates, stratified by several demographic groups such as age, sex, country of birth, migration status and geographical area of residence. Our secondary aims were to describe clinical and bacteriological characteristics, to optimise screening for TB in high-risk groups and provide guidance for TB elimination.
Methods
Data sources and study population
This was a retrospective observational time-series analysis using data obtained from the Netherlands Tuberculosis Register (NTR; more details are given in the supplementary material). All children and adolescents aged <20 years with TB notified to the NTR between January 1993 and December 2018 were included in this study; patients with missing information for country of birth were excluded. TB diagnosis follows the criteria set by the World Health Organization (WHO) European Region [15], including TB confirmed by culture identifying M. tuberculosis complex, and those clinically diagnosed with TB (without bacteriological confirmation).
Data collection
Anonymised data of TB cases were obtained from the NTR on November 13, 2019, consisting of demographics (year of diagnosis, age, sex, origin, migration status and geographical area of residence), TB background and clinical characteristics (type of case-finding, history of TB contact, travel history in TB-endemic areas, disease sites, previous treatment of TB infection or disease, bacille Calmette–Guérin (BCG) vaccination, TB symptoms (especially persistent cough), HIV status, patient delay and health system delay) and bacteriological characteristics (smear microscopy, mycobacterial culture and drug-susceptibility testing (DST)).
Definitions
Individuals aged 0–14 years at diagnosis were defined as children following the WHO recommendations for reporting childhood TB [16], and those aged 15–19 years were defined as adolescents. Foreign-born patients (first-generation immigrants) included those detected within the first 6 months after arrival in the country (period covered by entry screening), during 6–29 months after arrival (follow-up screening period) and at a later stage after follow-up screening. Dutch-born patients consisted of either native Dutch (born in the Netherlands with both parents born in the Netherlands) or second-generation immigrants (born in the Netherlands and have at least one foreign-born parent). Active case finding (ACF) was defined as the systematic screening for TB disease in predetermined high-risk groups, while passive case-finding (PCF) was defined as patients self-presented to the healthcare system because of TB symptoms. TB in the lungs, isolated tracheal or bronchus TB, laryngeal TB and other specified respiratory TB, were classified as pulmonary TB (PTB). TB in locations other than the lungs, including mediastinal/hilar lymphadenopathy, were classified as extrapulmonary TB (EPTB). Severe forms of TB included cavitary PTB, TB of the central nervous system (CNS; including meningitis) and miliary TB. Patient delay was defined as the time between initial onset of TB symptoms and the patient's first consultation with a healthcare provider. Health system delay was defined as the time from the patient's first consultation with a healthcare provider to treatment initiation. A cut-off of >4 weeks was used to define either prolonged patient delay or health system delay [17]. Both delays were applicable for patients with symptomatic PTB or combined PTB/EPTB. Since 2005, only “persistent cough” was registered in the NTR as TB symptom; other signs and symptoms of TB were not registered. DST results of resistance to at least isoniazid and rifampicin were defined as multidrug-resistant (MDR)-TB. Until 2005, DST results were not systematically registered in the NTR, and were only registered if resistance against isoniazid or rifampicin was detected. Definitions of other variables are presented in supplementary table S1.
Data analysis
Notification rates and rate ratios along with 95% confidence intervals, were estimated using Poisson regression models, offset with log population size. Overall and stratum-specific notification rate estimates from 1993–2018 were reported per 100 000 person-years, unless otherwise stated. Total annual population in different demographic groups from 1993–2018 (refer to the situation on January 1 of the year of observation), were derived from the Central Agency for Statistics (Statistics Netherlands (CBS) https://opendata.cbs.nl/statline/#/CBS/en/). Given that population numbers based on migration status and country of birth have only been available since 1996, notification rate estimates in these particular groups were calculated from 1996–2018. Trends in notification rates were estimated using average annual percent changes (AAPC) based on the segmented linear regression analysis [18]. Data were analysed using R-3.6.0 for Windows; additionally, the R package “segmented” was used for segmented analysis (www.CRAN.R-project.org/package=segmented). Associations of patient characteristics with severe forms of TB, prolonged patient and health system delays, mycobacterial culture results and MDR-TB were evaluated using univariate and multivariate logistic regression analyses. For these secondary outcomes, details are provided in the supplementary material.
Ethics
As this was a retrospective study using anonymous routinely collected data, ethics clearance and individual patient written informed consent were not required under Dutch law (https://wetten.overheid.nl/BWBR0007021/2006-02-01). Research approval was obtained from the research committee of the NTR. All TB prevention and control activities, including those for high-risk groups, follow the guidelines set by the Dutch Government and the WHO European Region [15, 19].
Results
Demographic characteristics
Among 31 376 TB cases notified to the NTR during 1993–2018, 3931 (12.5%) were children and adolescents. After excluding 32 patients with missing information for country of birth, 3899 records (1755 (45.0%) children and 2144 (55%) adolescents) were eligible for analysis. Foreign-born patients accounted for 2418 (62%) of the 3899 study population (725 (41.3%) out of 1755 children and 1693 (78.9%) out of 2144 adolescents). Of these, 692 (28.6%), 787 (32.5%) and 707 (29.2%) had lived in the Netherlands for <6 months, 6–29 months and ≥30 months at the time of diagnosis, respectively. Annual number of TB cases during 1993–2018 in Dutch-born children were relatively higher compared to foreign-born children (figure 1ai). In contrast, the annual number of TB cases in foreign-born adolescents constantly exceeded Dutch-born adolescents over the whole period (figure 1aii). TB cases in Dutch-born children were mostly detected through ACF (685 (66.5%) out of 1030), while the proportions of TB cases detected through ACF versus PCF in foreign-born children were approximately equal (345 (47.6%) versus 362 (49.9%) out of 725, respectively). In adolescents, most of the patients were detected through PCF: 286 (63.4%) of 451 Dutch-born and 1046 (61.8%) of 1693 foreign-born (table 1). Furthermore, children aged <5 years and aged 5–10 years were more likely to be detected through ACF, irrespective of their country of birth (supplementary figure S1).
Number of annual tuberculosis (TB) cases and notification rate estimates in children and adolescents in the Netherlands. a) Number of annual TB cases in Dutch- and foreign-born i) children and ii) adolescents from 1993–2018; b) TB notification rate estimates per 100 000 person-years in i) children and ii) adolescents from 1996–2018, stratified by Dutch- and foreign-born.
Demographic and clinical characteristics of children and adolescents with tuberculosis (TB) in the Netherlands, 1993–2018
Notification rate estimates and trends
During 1993–2018, the overall TB notification rate in the total population aged 0–19 years was 3.8 per 100 000 person-years (95% CI 3.3–4.5). In children, the overall rate was 2.3 per 100 000 person-years. The rate in foreign-born children was 26.3 per 100 000 person-years, 20.0 times higher than in Dutch-born (95% CI 11.9–33.0). Among subgroups of children, the highest rate occurred in African-born children (116.8 per 100 000 person-years), whereas native Dutch children had the lowest rate (0.4 per 100 000 person-years). In adolescents, the overall rate was 8.4 per 100 000 person-years, 3.6 times higher than in children (95% CI 2.6–5.0). With an estimated notification of 93.7 per 100 000 person-years, foreign-born adolescents had 53.4 times higher rate compared to Dutch-born adolescents (95% CI 31.7–95.7). Among adolescent subgroups, the highest rate occurred in African-born adolescents (316.6 per 100 000 person-years), while the lowest rate occurred in native Dutch adolescents (0.7 per 100 000 person-years). Compared to the overall years from 1993–2018, notification rates during 2014–2018 were lower in all demographic groups, except for African-born adolescents: 447.3 per 100 000 person-years (table 2).
Notification rate estimates and notification rate ratios of tuberculosis (TB) among children and adolescents in the Netherlands, 1993–2018
Among the foreign-born population, those from countries in the horn of Africa contributed most to the TB caseload. Overall, notification rates among Somali-born children and adolescents during 1996–2018 were 300.4 and 778.5 per 100 000 person-years, respectively. During 2014–2018, Somalian and Eritrean birth accounted for 56 (70.9%) out of 79 TB cases among foreign-born children, and 201 (73.6%) out of 273 TB cases among foreign-born adolescents (figure 2a). Detailed results for notification rates based on 30 selected countries of birth are presented in figure 2b. Stratified by area of residence, Amsterdam, Rotterdam and The Hague accounted for the highest overall notification rates in both children and adolescents, particularly during the earlier period from 1993 to around 2005 (supplementary figure S2).
Number of annual tuberculosis (TB) cases and notification rate estimates in foreign-born children and adolescents in the Netherlands, stratified by 30 selected countries of birth. a) Number of annual TB cases in foreign-born children and adolescents during 1996–2018; b) TB notification rate estimates per 100 000 person-years in foreign-born children and adolescents during 1996–2018. D.R. Congo: Democratic Republic of Congo.
TB notification rates in children declined steadily from 3.8 to 0.7 per 100 000 person-years during 1993–2018 (AAPC −10.9%, 95% CI −12.6–−9.1) (figure 1bi and table 3). Significant decreases were also observed in almost all subgroups of children except for one European-born subgroup (table 3; supplementary figure S3). In adolescents, although the AAPC was not statistically significant, our segmented analysis identified an increasing trend during 1993–2001 and 2012–2018, and a decreasing trend during 2001–2004 (figure 1bii and table 3). Significant average decreases over the whole period were shown in female, Dutch-born, native Dutch, second-generation immigrant, European-born, American-born, Asian-born and urban residing adolescents. In contrast, an increasing trend during 1996–2018 was observed in African-born adolescents (AAPC 18.5%, 95% CI 11.9–25.5). Additionally, our segmented analysis identified recent increasing trends in TB notification rates from 2013–2018 among male adolescents, and from 2012–2018 in both foreign-born and suburban residing adolescents (table 3; supplementary figure S4).
Trends in tuberculosis (TB) notification rates in children and adolescents in the Netherlands, 1993–2018
Clinical features
Out of the 3899 study population, 1716 (44.0%) were diagnosed with PTB, 1745 (44.8%) with EPTB and 438 (11.2%) with combined PTB/EPTB. Among 1716 PTB cases, 317 (18.5%) had pulmonary cavitation, and among 2183 patients with EPTB or combined PTB/EPTB, 113 (5.2%) had CNS or miliary TB. Our multivariate analysis revealed that children aged 10–14 years (adjusted (a)OR 20.28, 95% CI 2.71–151.6), adolescents (aOR 23.02, 95% CI 3.15–168.41), symptomatic patients with prolonged patient delay (aOR 2.23, 95% CI 1.35–3.69) and without prolonged patient delay (aOR 1.74, 95% CI 1.07–2.83), those with unknown history of TB contact (aOR 1.95, 95% CI 1.14–3.33) and those with positive mycobacterial cultures (aOR 6.10, 95% CI 2.18–17.08), were associated with increased odds of having cavitary PTB. In contrast, ACF was associated with lower odds of having cavitary PTB (aOR 0.42, 95% CI 0.29–0.60). Furthermore, several groups had increased odds of having CNS or miliary TB, including children aged <2 years (aOR 4.30, 95% CI 1.76–10.54) and aged 2–4 years (aOR 2.88, 95% CI 1.20–6.90), those with culture-confirmed disease (aOR 2.61, 95% CI 1.19–5.70) and those with HIV co-infection (aOR 7.90, 95% CI 2.89–21.62). In contrast, ACF was associated with lower odds of having CNS or miliary TB (aOR 0.12, 95% CI 0.05–0.28) (table 4). In a subgroup analysis of children aged <5 years, those who previously received BCG vaccination had lower odds of having CNS or miliary TB compared to those who were not vaccinated (aOR 0.26, 95% CI 0.07–0.97), adjusted for case-finding methods, history of TB contact and HIV status.
Final model for factors associated with severe forms of tuberculosis (TB) in children and adolescents in the Netherlands, 1993–2018
Among 2154 patients with PTB or combined PTB/EPTB, 1392 (64.6%) reported cough. Of these symptomatic patients, 1000 (71.8%) and 1007 (72.3%) had known information on patient and health system delays, respectively. Our multivariate analysis identified that adolescents had a higher odds of prolonged patient delay (aOR 1.81, 95% CI 1.03–3.20) (table 5). Additionally, our subgroup analysis among children showed that unknown history of TB contact was associated with higher odds of prolonged patient delay (aOR 2.17, 95% CI 1.11–4.25), adjusted for year of diagnosis, age, country of birth and area of residence. Several groups including children aged 2–4 years (aOR 2.39, 95% CI 1.07–5.32), females (aOR 1.46, 95% CI 1.07–1.99) and those residing in suburban areas (aOR 1.81, 95% CI 1.25–2.63) had an increased odds of prolonged health system delay. In contrast, patients detected through ACF (aOR 0.51, 95% CI 0.34–0.77) and foreign-born patients with duration of stay <6 months (aOR 0.46, 95% CI 0.26–0.82) were associated with lower odds of prolonged health system delay (table 5).
Final model for factors associated with prolonged patient delay and health system delay in children and adolescents with tuberculosis (TB) in the Netherlands, 1993–2018
Bacterial results and drug resistance
Mycobacterial culture results were known in 2688 (68.9%) of 3899 eligible patients, of which 2313 (59.3%) were culture-positive and 375 (9.6%) were culture-negative. Mycobacterial culture was more likely to be positive in adolescents (aOR 1.93, 95% CI 1.25–2.98), among foreign-born (aOR 1.40, 95% CI 1.05–1.88), those reporting cough (aOR 1.79, 95% CI 1.29–2.45), those detected through PCF (aOR 4.86, 95% CI 3.54–6.66), those with PTB (aOR 8.42, 95% CI 5.90–12.02) and with combined PTB/EPTB (aOR 4.13, 95% CI 2.65–6.43) (supplementary table S2). DST results were available in 976 (42.2%) of 2313 patients with culture-confirmed TB. Of these, isoniazid- and rifampicin-resistant, MDR and extensively drug-resistant isolates were observed in 158 (16.2%), two (0.2%), 37 (3.8%) and one (0.1%), respectively. Additionally, 1337 (57.8%) of 2313 culture-confirmed TB cases had missing information on DST results; of these, 1321 (98.8%) were notified before 2005 (supplementary figure S5). Detection of MDR-TB was more likely to be higher in patients detected through ACF (aOR 2.26, 95% CI 1.11–4.61) and in re-treated TB cases (aOR 8.54, 95% CI 2.55–28.55) (supplementary table S3).
Discussion
The overall TB notification rate among children in our study was relatively low, and is comparable to those reported in other low TB incidence countries such as the UK, Australia, Canada and the USA, with overall rates ranging from 1.0 to 4.3 per 100 000 person-years [20–24]. Almost all subgroups of children in our study had declining trends in TB notification during the study period. Although a declining trend was also shown in foreign-born children, disproportionately higher rates compared to Dutch-born children were observed in this population, particularly those from Africa. Previous reports from the UK and Denmark support that childhood TB burden among the African immigrant population is higher compared to other immigrant groups [22, 25]. In adolescents, the annual number of foreign-born TB cases has consistently exceeded Dutch-born patients since 1993; this mirrors the trends in the adult population with TB [26]. Although Dutch-born adolescents had declining trends of TB notification, an increasing trend was observed in the foreign-born group, especially those from Africa. This might be explained by the increased number of (unaccompanied) minor asylum seekers coming to the Netherlands in recent years.
Our study showed a high number of TB cases notified among Somali-born children and adolescents over the whole period, with an additional high TB caseload among Eritrean-born adolescents in recent years. TB incidence among Eritrean and Somalian asylum seekers is high, not only at entry screening, but also within the first 5 years after arrival in the Netherlands [27]. Irrespective of the year of diagnosis, our study showed that approximately one-third of the foreign-born patients were detected within 6 months after arrival, while approximately two-thirds of the foreign-born patients were detected after 6 months of arrival. The latter is probably attributable to reactivation of TB infection in their country of origin [28], or exposure to TB during travel to the Netherlands [29]. This may also be explained by recent transmission within specific communities [28, 30], such as clustered patients coming from sub-Saharan Africa or living in an urban area [30].
In the Netherlands, entry screening for TB by chest radiography is currently mandatory for new immigrants from countries with TB incidence >100 per 100 000 population and for asylum seekers from countries with TB incidence >50 per 100 000 population. Additionally, immigrants and asylum seekers from countries with TB incidence ≥200 per 100 000 population are offered voluntary biannual follow-up screening for a 2-year period [31]. Evidently, the current screening approach is not sufficient to reduce TB incidence in these migrant populations. Introducing universal screening for latent TB infection (LTBI) and preventive treatment for immigrants and asylum seekers from high TB incidence countries might prevent TB disease [32, 33]. This practice has been implemented in the Netherlands since 2016 as entry screening for immigrants aged <18 years, but not yet for all asylum seekers, due to operational challenges [31, 34]. Implementation of LTBI screening and treatment in both immigrants and asylum seekers on arrival in the Netherlands is feasible and effective [35, 36]. Adequate resources as well as community engagement involving well-connected and trusted community members as facilitators are needed to reach and motivate this population for LTBI screening and treatment programmes [37]. Radiographic screening for asylum seekers aged <12 years is not effective [38]; therefore, LTBI screening and treatment would be a preferable intervention in this group.
Children with TB infection not receiving preventive therapy have an increased risk of developing TB disease within 2 years after exposure, and the risk is highest among children aged <5 years [39]. This reinforces the need for immediate contact investigations. In our study, most of the TB cases in children aged <10 years and in Dutch-born children were detected through ACF using contact investigations. However, many of the TB cases among Dutch- and foreign-born adolescents were detected through PCF. In a study among children from 2005–2012, foreign-born children who had lived in the Netherlands for >6 months were significantly more often detected through PCF than by ACF [14]. This association is no longer significant in our subgroup analysis using recent data from 2012–2018 (results not shown). The improvement of eligibility for LTBI screening among child and adolescent TB contacts since 2012, may have resulted in more foreign-born TB contacts being screened and treated for LTBI [40]. It can be assumed that there has been less transmission within these population groups in recent years, and more TB disease being prevented through LTBI treatment of TB contacts.
Beyond benefits of ACF to enhance TB case detection and reduce transmission, our results provide evidence for additional advantages of ACF to reduce severe forms of TB and decrease health system delay. In low- and middle-income countries, ACF among household contacts of patients with TB is effective in enhancing TB case detection [41, 42], and reducing all-cause mortality [41]. In the Netherlands, ACF interventions also benefit children and adolescents in improving TB treatment outcomes, particularly in reducing all-cause mortality [4]. However, it should be acknowledged that even in a setting where ACF is well implemented like the Netherlands, missed opportunities for prevention persist. Erkens et al. [14] reported that among 37% childhood TB cases, there was at least one missed opportunity for prevention, and of children eligible for BCG vaccination, 39% were not vaccinated.
The strength of this study is the inclusion of a relatively long duration of data series, making it possible to perform a robust time-series analysis and provide a complete picture of TB epidemiology in different demographic groups. Our study has some limitations. All patients are assumed to have been notified to the NTR, but the possibility of undernotification cannot be ruled out, particularly during the earlier period of registration. In 1998, the adjusted undernotification of TB was 7.3%, and those with non-culture-confirmed TB were less likely to be notified [43]. In 2005, a central web-based registration system was introduced and laboratory results were matched with the NTR in real time. After these improvements, we expect that the completeness of TB notification has increased over time. Although some EPTB cases such as abdominal TB, osteoarticular TB and urogenital TB could be life-threatening, these forms of EPTB were assumed to be less severe than CNS and miliary TB in our study because EPTB cases other than CNS and miliary TB were not associated with higher risk of mortality in our previous analysis [4]. In addition, this study is limited by its retrospective nature with partly incomplete records for some of the variables. As an example, DST results were not systematically registered until 2005, causing >98% of the culture-confirmed TB cases before 2005 to have missing information on drug-susceptibility status. Therefore, to minimise selection bias in analysing factors associated with MDR-TB, year of diagnosis was included as one of the explanatory variables in the multivariate model.
In conclusion, TB notification rates in children were relatively low across different demographic groups and have decreased to pre-elimination level among native Dutch children. However, substantially higher rates were found among foreign-born adolescents, particularly those from Africa. Optimising current TB control programmes by enhancing ACF, including contact investigations and screening for TB and LTBI in foreign-born children and adolescents, may be the best way forward to reduce the TB burden in these high-risk groups.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material ERJ-01086-2020.SUPPLEMENT
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-01086-2020.Shareable
Acknowledgements
We would like to thank the Dutch National Institute for Public Health and the Environment, Centre for Infectious Disease Control (RIVM-CIb) for providing data through the Netherlands Tuberculosis Register (NTR). We would also like to thank all staff of the TB departments at Municipal Public Health Services for reporting data to the NTR. Special thanks to Henrieke Schimmel (RIVM, Bilthoven, the Netherlands) for her assistance during research approval from the research committee of the NTR, and for providing additional information of the NTR data.
Footnotes
This article has supplementary material available from erj.ersjournals.com
Author contributions: F. Gafar, N. van't Boveneind-Vrubleuskaya, B. Wilffert and J-W.C. Alffenaar contributed to conception and design of the study. F. Gafar obtained data of TB cases from the Netherlands Tuberculosis Register. T. Ochi undertook data extraction of population numbers from the Central Agency for Statistics. F. Gafar performed data analysis and created tables and figures. F. Gafar, T. Ochi, N. van't Boveneind-Vrubleuskaya, O.W. Akkerman, C. Erkens, S. van den Hof, T.S. van der Werf, J-W.C. Alffenaar and B. Wilffert interpreted the results. F. Gafar and T. Ochi drafted the manuscript. J.W.C. Alffenaar and B. Wilffert supervised the entire project. All authors critically revised the manuscript for important intellectual content and approved the final version of the manuscript.
Conflict of interest: F. Gafar has nothing to disclose.
Conflict of interest: T. Ochi is an employee of BaseClear B.V. (commercial, non-R&D).
Conflict of interest: N. van't Boveneind-Vrubleuskaya has nothing to disclose.
Conflict of interest: O.W. Akkerman has nothing to disclose.
Conflict of interest: C. Erkens has nothing to disclose.
Conflict of interest: S. van den Hof has nothing to disclose.
Conflict of interest: T.S. van der Werf has nothing to disclose.
Conflict of interest: J-W.C. Alffenaar has nothing to disclose.
Conflict of interest: B. Wilffert has nothing to disclose.
Support statement: This study was supported by the University of Groningen and the Indonesia Endowment Fund for Education (LPDP; 201711220412046) through a PhD scholarship awarded to F. Gafar. The funding bodies had no role in the study conception and design, data collection and analysis, writing and reviewing of the manuscript, or decision to publish. Funding information for this article has been deposited with the Crossref Funder Registry.
- Received April 8, 2020.
- Accepted May 22, 2020.
- Copyright ©ERS 2020
References
