First Trimester Iron Status and Obstetric and Perinatal Outcomes

Purpose: To assess the following in singleton pregnant women: 1) associations between rst trimester iron deciency and obstetric and perinatal outcomes, 2) overall rst trimester iron status and 3) post-treatment iron status after intensied iron supplementation. Methods: Prospective cohort study with linkage of rst trimester hemoglobin and plasma ferritin with obstetric and perinatal data from a hospital database. Blood sample data were obtained at a Danish University Hospital. The cohort was divided into groups according to ferritin and hemoglobin: (1) iron decient anemic (ferritin <30 ng/mL and Hb <110 g/L), (2) iron decient non-anemic (ferritin <30 ng/mL and Hb ≥ 110 g/L), and (3) iron replete non-anemic (ferritin 30– 200 ng/mL and Hb ≥ 110 g/L). Obstetric and perinatal outcomes in each iron decient group were compared to the iron replete non-anemic group using multivariable logistic regression. The effect of 2–8 weeks intensied iron supplementation on hemoglobin and ferritin were assessed by pairwise comparisons. Results: The cohort comprised 5,763 singleton pregnant women, of which 14.2% had non-anemic iron deciency, and 1.2% had iron deciency anemia. Compared to iron replete non-anemic women, iron decient anemic women had a higher risk of gestational diabetes (aOR 3.8, 95% CI 1.4–9.0), and iron decient non-anemic women had a higher risk of stillbirth (aOR 4.0, 95% CI 1.0–14.3). In group 1 and 2, 78.7% and 67.8% remained iron decient after intensied iron supplementation. Conclusion: First trimester iron deciency may be associated with gestational diabetes and stillbirth. First trimester iron deciency was present in 15.4% and often persisted despite 2–8 weeks intensied iron supplementation. Spanish supplementation from an mean sampling performed at a mean age 17 vs. These differences may explain why we observed ferritin concentrations below 12 ng/mL at follow-up in iron deciency anemia may be explained by high exposure to iron supplements. we did not nd a similar increased risk of GDM in the non-anemic iron decient women (who were also expected to have a high exposure), which this theory. Neither mean nor of were obviously different between the groups, and furthermore, our model adjusted for several relevant factors including BMI >25 kg/m 2 and race-ethnicity. The role of socioeconomic status, however, remains unknown.


Introduction
Iron de ciency and iron de ciency anemia in pregnancy are common health problems. According to the World Health Organization (WHO), about 38% of pregnant women worldwide are anemic, whereas iron de ciency affects around 40%. Iron de ciency is the most common pathological cause of anemia in pregnancy, appraised to account for 50% of all cases [1,2].
Iron de ciency occurs when the level of stored body iron becomes deplete. As iron is prioritized for erythropoiesis, iron de ciency anemia is the end-stage result of iron de ciency [3]. Iron is critical for oxygen delivery and enzyme functions involved in fetal organogenesis during pregnancy [4], and iron requirements increase as gestation proceeds [5]. Both low and elevated hemoglobin (Hb) and iron status in pregnancy have been associated with an increased risk of adverse obstetric and perinatal outcomes, and vulnerability to abnormal concentrations appears to depend on the gestational age at which they occur [6]. The association between low Hb and adverse outcomes is most evident when Hb is measured in early pregnancy [6], and anemia, including iron de ciency anemia in early pregnancy, has been associated with increased risk of preterm delivery (PTB) and low birthweight (LBW) [7,8]. While low Hb is a valid and widely available diagnostic marker for anemia, it cannot detect the stages of iron de ciency that precede anemia [4,9]. Hence, the combination of Hb and ferritin re ects the iron status spectrum more speci cally than Hb alone. However, adverse effects of non-anemic iron de ciency in pregnancy are less explored than those of iron de ciency anemia [6].
In Denmark, the National Health Authority recommends routine supplementation with 40-50 mg iron from 10 weeks' gestation [10,11]. In 2016, the Danish Society of Obstetrics and Gynaecology (DSOG) stated that neither routine iron supplementation nor routine screening with individualized iron dosing is supported by direct evidence [12]. However, DSOG further stated that screening is preferable from an expert point of view, as routine supplementation is more likely to lead to both under-and overtreatment with unnecessary and potentially harmful side effects [12]. Therefore, an individualized screening program for rst trimester iron de ciency and anemia was implemented at Copenhagen University Hospital Hvidovre, and we designed a study in parallel to explore the screening-related data.
The overall rst trimester Hb and iron status in women in Denmark is currently unknown and whether the associations of hematological markers and adverse obstetric and perinatal outcomes found in other populations also exist in women in Denmark remains unexplored. Therefore, the primary aim of this study was to investigate the associations between both anemic and non-anemic iron de ciency and obstetric and perinatal outcomes. Further objectives included to determine rst trimester Hb and iron status in pregnant women and assess how these markers change in iron de cient women who receive intensi ed iron supplementation.

Study overview
Prospective cohort study based on blood samples from a rst trimester (before 14 weeks' gestation) iron de ciency and anemia screening program in pregnant women, collected from June 2017 through May 2018 at a single tertiary University hospital in Denmark (Department of Obstetrics and Gynecology, Copenhagen University Hospital Hvidovre). Blood sample data were collected prospectively and subsequently cross-linked to demographic, obstetric, and perinatal data from a high-validity hospital-based clinical database -the Obstetric Database. The database has been described in detail by Brixval et al. and is a valid source for assessing the prevalence of obstetric interventions and outcomes [13]. Data were cross-linked using the unique personal identi cation code allocated to all citizens in Denmark at birth.
The anemia and iron de ciency screening program The individualized screening program was implemented locally at Copenhagen University Hospital Hvidovre and adapted to routine antenatal care between 2017 and 2020. The practice was implemented in accordance with the guidelines developed by DSOG (2016) [12] as an alternative to the Danish standard of routine prophylactic iron supplementation (Danish National Health Authority recommendation: 40-50 mg elemental iron starting from 10 weeks' gestation [10,11]). The screening was performed together with the routine screening for trisomy 21 (double test), which approximately 94% of all pregnant women in Denmark attend [14]. The screening included Hb and plasma ferritin measurements from a venous blood sample, as guidance for an individualized iron dosing regimen. Based on the screening results, women were recommended daily iron supplementation dosed as described in Figure 1. Women with ferritin concentrations <30 ng/mL were recommended to intensify iron supplementation and have Hb and ferritin measurements repeated after four weeks.

Population
The cohort comprised all women with a singleton pregnancy who ful lled the following criteria: screened for anemia and iron de ciency before 14 weeks' gestation from June 2017 through May 2018 after referral to standard antenatal care at Copenhagen University Hospital Hvidovre with their data included in the Obstetric Database (Fig. 2). If a woman had more than one screening sample in the same pregnancy, only the screening results from the rst date were included. Further, if a woman had more than one pregnancy during the study period, only the rst pregnancy was included.
The cohort was further categorized into groups to investigate the associations between iron de ciency and obstetric and perinatal outcomes (Fig. 2). Before grouping, we excluded women with non-iron de cient anemia (Hb <110 g/L and ferritin ≥30 ng/mL) (n=91) and women with high ferritin levels (>200 ng/mL) (n=145) as these states can have a range of underlying causes (e.g. thalassemia, chronic disease, hemochromatosis) that per se may negatively affect pregnancy outcomes. Furthermore, women who participated in a randomized controlled trial (RCT) on intravenous and oral iron treatments [15] in their pregnancy were excluded (n=48) since the allocated interventions rather than their Hb and iron status could affect pregnancy outcomes and lead to bias.
Subsequently, the cohort was categorized into three groups as follows: 1. Iron de cient anemic cases (ferritin <30 ng/mL and Hb <110 g/L) 2. Iron de cient non-anemic cases (ferritin <30 ng/mL and Hb ≥110 g/L) 3. Iron replete non-anemic references (ferritin 30-200 ng/mL and Hb ≥110 g/L) Finally, hematological changes were assessed in a subgroup of women with rst trimester iron de ciency (i.e. group 1 and 2 above) who attended a follow-up within 2-8 weeks after their screening sample.

Obstetric and perinatal outcomes
Detailed de nitions of obstetric and perinatal outcomes are included as supplementary information (Online Resource 1). Obstetric outcomes included: hypertensive disorders in pregnancy, gestational diabetes (GDM), induction of labor, oxytocin augmentation of labor, instrumental vaginal birth (vacuum or forceps), emergency cesarean, labor dystocia (emergency cesarean and/or oxytocin augmentation and/or instrumental vaginal birth), postpartum hemorrhage (PPH ≥500 mL), and severe PPH (>1,000 mL).

Statistical analysis
Descriptive statistics were used to summarize the distribution of rst trimester Hb and ferritin in the total cohort and present characteristics of the three iron status groups. Among women with iron de ciency at screening (i.e. group 1 and 2), all attending a follow-up blood sampling 2-8 weeks after screening contributed with paired data in analyses of Hb and ferritin changes. In these comparative analyses, McNemar exact test was used for categorical data and paired t-test and Wilcoxon signed rank test as appropriate for continuous data.
For the primary aim, obstetric and perinatal outcomes were compared between the grouped women using the iron replete non-anemic women as a reference group. All comparisons were made using univariate and multivariable logistic regression, and results were presented as odds ratios (OR) and adjusted OR (aOR) with 95% con dence intervals (CI). The multivariable logistic regression analyses adjusted for the following potential confounding variables selected based on previous research: multiparity (parity ≥1); maternal age >35; maternal tobacco-smoking during pregnancy; child's sex; maternal body mass index (BMI) >25 kg/m 2 ; Non-White race-ethnicity. Race-ethnicity data were collected and included in the adjusted model because anemia and adverse pregnancy outcomes may be higher among Non-White women. Race-ethnicity had been classi ed either by a caregiver or self-reporting and was categorized as White, Non-White, or unknown.
Hb concentrations were reported as mmol/L, and we converted them to g/L using conversion factors recommended by The International Council for Standardization in Haematology [17]. The analysis plan was nalized post-hoc. The statistical software program R was used for all statistical analyses and the MASS-package for logistic regression analyses [18,19]. A p-value <0.05 was considered statistically signi cant.

Results
In total, 7,391 women were screened with rst trimester Hb and ferritin, of which approximately 20% were missing outcome data in the Obstetric Database, were multiple pregnancies, or missing rst trimester samples. Accordingly, 5,763 unselected singleton pregnant women were included in the study cohort ( Fig. 2). Screened women with and without obstetric and perinatal data seem to have been comparable in measures of average Hb, ferritin, and age (data not shown).
Of the 5,763 women, 284 were excluded from the regression analyses due to non-iron de cient anemia (n=91), high ferritin levels (n=145), and participation in a RCT on intravenous and oral iron treatments (n=48). The remaining 5,479 women were categorized in the following three groups: (1) iron de cient anemia cases (n=64), (2) iron de cient non-anemic cases (n=778), and (3) iron replete non-anemic references (n=4,637) (Fig. 2). Pre-post treatment assessments included women in group 1 and 2 (i.e. iron de cient women) who attended a new Hb and ferritin sampling 2-8 weeks after their primary sample.

First trimester Hb and iron status
First trimester Hb and ferritin levels of the 5,763 included women are presented in Table 1. A total of 890 (15.4%) were iron de cient: 821 (14.2%) were iron de cient non-anemic and 69 (1.2%) women were iron de cient anemic, respectively. Overall, 91 (1.6%) women had anemia due to other causes than iron de ciency (Hb <110 g/L and ferritin ≥30 ng/mL). No cases of severe anemia (Hb <70 g/L) were observed. A total of 2,376 (41.2%) women had ferritin concentrations ≥70 ng/mL. In groups 1 and 2, a total of 842 women had rst trimester iron de ciency and were recommended oral iron supplementation in daily doses according to Figure 1. Among these women, 557 (66.2%) had a new blood test taken within 2-8 weeks after their primary sample, and their hemoglobin and ferritin changes are presented in Table 2. In 47 women who initially had iron de ciency anemia, the mean Hb concentration increased in the subsequent blood samples (102.6 vs. 107.9 g/L, p<0.001) and anemia (Hb <110 g/L) resolved in 19 (40.4%). Among 510 initially iron de cient non-anemic women, the mean Hb concentration remained within the normal range though it decreased from baseline to follow-up (126.7 vs. 123.6 g/L, p<0.001), and 480 (94.1%) remained non-anemic (Hb ≥110 g/L). Median ferritin concentrations increased both in the initially iron de cient anemic (11.0 vs. 20.0 ng/mL, p<0.001) and iron de cient non-anemic women (23.0 vs. 25.0 ng/mL, p<0.001). The overall prevalence of iron de ciency decreased in both groups, however, most women remained iron de cient: 78.7% of initially iron de cient anemic and 67.8% of initially iron de cient non-anemic women, respectively. No initially iron de cient anemic women developed high Hb (>130 g/L), and the prevalence of high Hb decreased in the initially iron de cient non-anemic women (34.9% vs. 21.6%, p<0.001). Hb, hemoglobin; MD, mean difference; CI, con dence interval; OR, odds ratio; SD, standard deviation; IQR, interquartile range; NA, not applicable.

Baseline characteristics
Demographics and baseline characteristics of women in group 1-3 are summarized in Table 3. The proportion women with BMI <18.5 kg/m 2 , parity ≥1, Non-White race-ethnicity, hyperemesis, and previous bariatric surgery appeared to be unevenly distributed between groups with the highest prevalence in iron de cient anemic women and lowest prevalence in iron replete non-anemic women (although mean BMI appeared comparable across groups). In addition, iron de cient anemic women seemed more likely to be non-smokers and deliver a female child than the women in the other two groups. No other obvious differences were observed between the groups. Obstetric and perinatal outcomes The obstetric and perinatal outcomes of women in the three groups and the logistic regression analyses results are shown in Table 4. Women with iron de ciency anemia were more likely to develop GDM (aOR 3.8, 95% CI 1.4-9.0) than the iron replete non-anemic references. Furthermore, the iron de cient nonanemic women had a higher risk of stillbirth (aOR 4.0, 95% CI 1.0-14.3) than the iron replete non-anemic references. The remaining comparisons showed no statistically signi cant between-group differences. Table 4 Multivariable logistic regression analyses: obstetric and perinatal outcomes in iron de cient anemic (n=64), iron de cient non-anemic (n=778), and iron replete non-anemic women (n=4,637) with singleton pregnancies  Obstetric and perinatal outcomes divided by iron status. Frequencies expressed as number (%). Logistic regressions with the iron replete non-anemic group as reference expressed as odds ratios with 95% CI in parenthesis. Bold font indicates statistical signi cance. NA indicate that analysis was not possible due to no cases in one or both groups. f Variables with missing values: postpartum bleeding data in the iron de cient anemic group (two missing), iron de cient non-anemic group ( ve missing), and iron replete non-anemic group (16 missing), respectively; birthweight data in the non-anemic iron de ciency group (one missing).
g Umbilical artery pH <7.05 and/or Apgar Score <7 at 5 minutes h Birth weight <-2 SD from the expected sex-speci c birth weight for the given gestational age.
i Birth weight >+2 SD from the expected sex-speci c birth weight for the given gestational age.

Discussion
In this cohort of 5,763 women screened in early pregnancy, 14.2% had non-anemic iron de ciency while 1.2% had iron de ciency anemia. GDM were more likely to occur in women with iron de ciency anemia than in iron replete non-anemic women (aOR 3.8, 95% CI 1.4-9.0), whereas stillbirth was more common in iron de cient non-anemic women than in iron replete non-anemic women (aOR 4.0, 95% CI 1.0-14.3). All iron de cient anemic and iron de cient non-anemic women were recommended to intensify oral iron supplementation and new Hb and ferritin measurements after approximately four weeks. Among women with initial iron de ciency anemia, overall Hb concentration increased (102.6 vs. 107.9 g/L), and anemia resolved in 40.4%. In initially iron de cient non-anemic women, the overall Hb concentration remained normal, although it decreased (126.7 vs. 123.6 g/L), and 94.1% remained non-anemic. The prevalence of iron de ciency decreased, yet iron de ciency persisted in most women as ferritin remained below 30 ng/mL in 78.7% of the initially iron de cient anemic women and 67.8% of the initially iron de cient non-anemic women, respectively.

First trimester Hb and ferritin
First trimester iron status has previously been reported, including a Belgian study (Vandevijvere et al.) that found ferritin concentrations <15 ng/mL in 6.1% and the combination of ferritin <15 ng/mL and Hb <110 g/L in 1.9% [20]. In our cohort, the corresponding prevalences were 2.8% and 0.7%, respectively. Among pregnant women in Australia (Khambalia et al.) and the US (Mei at al.), 19.6% and 7.3% have rst trimester ferritin <12 ng/mL, respectively [21,22]. However, the prevalence of ferritin concentrations <30 ng/mL in non-anemic women in our cohort (14.2%) was similar to that in a recent US report including rst trimester pregnant women (14% reported by Auerbach et al.) [23]. First trimester ferritin concentrations ≥70 ng/mL, which has been suggested to re ect the amount of body iron required to complete pregnancy without developing iron de ciency [21,24], were present in 27.5% of the Belgian study population [20] versus 9.7% of women in Australia [21] and 41.2% of our Danish cohort. Risk factors for iron de ciency in pregnancy identi ed in some of these previous studies included multiparity, Non-White race-ethnicity, younger age and low socioeconomic status [20][21][22]. The proportion of multiparous women and women of Non-White race-ethnicity were lower in our population compared to the Belgian study and the US study by Mei et al. [20,22]. In addition, approximately 14% of participants were 12-19 years old in the study by Mei et al. [22] whereas only six (0.01%) women in our study were aged 19 or younger. Furthermore, the rst trimester samplings in studies conducted in countries with iron supplemental policies (i.e. Denmark and US) [10,25] may more frequently have been in uenced by iron supplement use than samplings in study settings without such guidelines (i.e. Belgium and Australia). Hence, parity, race-ethnicity, age, and probably also iron intake and socioeconomic status in our cohort was not comparable with that of previous studies [20][21][22][23], which may explain why some of our ndings differ.

Hb and Ferritin changes after intensi ed iron supplementation
Guidelines suggest that Hb should increase by at least 10 or 20 g/L two or four weeks after initiation of oral iron supplementation, respectively [26][27][28][29]. Similar to our results, a previous comparative Danish trial has reported that concentrations of Hb and ferritin decrease in early pregnancy despite oral supplementation [30], although the average ferritin at baseline in this population was higher than our iron de cient subpopulation.
A Spanish study have previously explored how Hb and ferritin change in rst trimester non-anemic women taking iron supplements [31]. All 41 initially iron depleted non-anemic women (de ned as Hb ≥110 g/L and ferritin <12 ng/mL determined at 8-12 weeks' gestation) were recommended iron supplements.
Among women attending two blood samplings in our study, 30 were initially non-anemic with ferritin <12 ng/mL, of whom only 20% still had ferritin below 12 ng/mL and 13% had developed anemia at follow-up (data not shown). Compared to the Spanish study, the 30 women in our cohort were recommended iron supplementation from an earlier mean gestational age (from 11 vs. 17 weeks' gestation) in higher daily dosage (100 vs. 40 mg) and had the follow-up sampling performed at a lower mean gestational age (at 17 vs. 24 weeks' gestation). These differences may explain why we observed ferritin concentrations below 12 ng/mL at follow-up in only 20%.
First trimester Hb and ferritin and obstetric and perinatal outcomes Our study examined the rst trimester iron status and showed signi cant associations between non-anemic iron de ciency and stillbirth and between iron de ciency anemia and GDM. However, these ndings should be interpreted with caution due to few events, wide CIs, and the fact that information on oral iron supplementation was limited to the recommended doses and not what was actually taken.
We are not aware of previous studies investigating the relationship between non-anemic iron de ciency and stillbirth, whereas low Hb in early pregnancy has been associated with an increased risk of perinatal death [32]. In our study, no stillbirths occurred among the iron de cient anemic women, and the overall incidence of stillbirth was low (12 of 5,479 babies). Hence, our stillbirth risk estimate is therefore imprecise.
The higher risk of GDM in women with iron de ciency anemia compared to iron replete non-anemic women was unexpected, as a previous review and metaanalysis conversely concluded that iron de ciency anemia lowers the risk of GDM [33]. In a previous Danish observational study, pregnant women who developed GDM had higher concentrations of rst trimester ferritin than healthy pregnant controls, and ferritin was positively associated with GDM risk [34]. In the previously mentioned Australian cohort, rst trimester iron de ciency (ferritin <12 ng/mL) was associated with a reduced risk of GDM [21]. Further previous reviews and meta-analyses have shown that elevated ferritin and Hb are associated with GDM, and that the average concentrations of ferritin and Hb are higher in women with GDM [35][36][37]. However, it should be noted that Hb and iron status were measured at heterogenous times during gestation in the reviewed studies [35][36][37], and that two of the reviews assessed but did not nd an association between prenatal iron supplementation and GDM [35,36]. The underlying pathophysiological mechanisms of abnormal iron status and GDM is not fully understood, and it remains unknown if the association can be explained by the iron status per se, oral iron exposure, or other factors. Our nding of increased risk of GDM with iron de ciency anemia may be explained by high exposure to iron supplements. However, we did not nd a similar increased risk of GDM in the non-anemic iron de cient women (who were also expected to have a high exposure), which does not support this theory. Neither mean BMI nor prevalence of overweight were obviously different between the groups, and furthermore, our model adjusted for several relevant factors including BMI >25 kg/m 2 and race-ethnicity. The role of socioeconomic status, however, remains unknown.

Strengths and limitations
It is a strength that we assessed baseline Hb and ferritin in the rst trimester, and that follow-up samples were measured with a narrow time window. Plasma volume expansion is a normal process in a healthy pregnancy and exceeds the increase in erythrocyte mass [5,[38][39][40][41], making differentiation between physiological and pathological anemia di cult as gestation proceeds. This may explain why previous studies have found the association between hematological markers and adverse outcomes more evident in early pregnancy[6], when the impact of the changing plasma dynamics on Hb and ferritin concentrations is minimal. Thus, determining Hb and iron status in early pregnancy is presumably more valid than later in pregnancy. Further strengths include the large sample size and that we were able to adjust for important potential confounders. However, with the observational design, we cannot investigate causality but only generate hypotheses of such. The lack of information about socioeconomic status is a limitation, as this factor may in uence both hematological indices and several pregnancy outcomes. Furthermore, women were recommended individualized doses of oral iron supplementation, and no information on compliance was available. However, as compliance to iron polices among pregnant women in Denmark has been reported to be high [42], it is likely that most women in our cohort took iron supplements and that the exposure was higher in the iron de cient than in the iron replete women.
Subsequently, associations may potentially be explained by the high exposure to iron supplementation in pregnancy rather than the rst trimester iron status per se. On the other hand, the intensi ed supplementation in iron de cient women might improve their iron status and thereby dilute the true adverse effects of anemic and non-anemic iron de ciency. Lastly, data for the subgroup assessments of Hb and ferritin changes were short-term and affected by missing, as only 557 of the 890 eligible women attended a follow-up sampling.

Conclusions
In conclusion, we have assessed rst trimester iron status in an unselected sample of singleton pregnant women. In comparison to iron replete non-anemic women, we found that women with iron de ciency anemia had a higher risk of GDM, and iron de cient non-anemic women had a higher risk of stillbirth, although risk estimates were imprecise due to few events, especially for stillbirth. We found a low prevalence of both anemic and non-anemic iron de ciency, and a high prevalence of ferritin levels estimated as su cient to complete pregnancy without developing iron de ciency [21,24]. With this new knowledge of the iron de ciency and repletion prevalence in pregnant women in Denmark we question if routine iron supplementation is the most bene cial approach to maternal and infant health and call for trials that compare short-and long-term effects of routine and individualized iron supplementation Declarations Authors' contributions RH, ALS, VMS, CH, LK and AP were involved in the conception and design of the study. RH and VMS carried out data collection. RH and ALS performed data analyses. RH drafted the manuscript. All authors were involved in the interpretation of the data and reviewed and approved the nal manuscript.

Funding
This study has been covered solely by internal funding.

Con icts of interest / competing interests
When the study was conducted, Rebecka Hansen and Veronika Markova Sommer were investigators and Charlotte Holm primary investigator for a clinical trial sponsored by Pharmacosmos A/S. As investigators, Rebecka Hansen and Veronika Markova Sommer had salary costs funded by Pharmacosmos A/S. Rebecka Hansen and Charlotte Holm and have served on advisory boards for Pharmacosmos A/S. However, Pharmacosmos A/S did not play any role or had any in uence on this cohort study. Only the named authors had in uence on the study design, conduct, analysis, interpretation and the manuscript. The authors have no additional con icts of interest to declare.

Availability of data and material
Requests on data access can be made through contact with the corresponding author. Data may be accessed upon reasonable request and after review and approval by the Danish Data Protection Agency and the Danish Patient Safety Authority.

Code availability
The code (R) will be available from the corresponding author upon reasonable request.

Ethics approval
The study was approved by the Danish Data Protection Agency (J. No.: AHH-2017-031, I-suite number: 05349) and the Danish Patient Safety Authority (J. No.: 3-3013-2410/1). According to Danish legislation, informed written consent from patients is not required for this speci c type of study. The reporting of the study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Consent to participate
Not applicable.

Consent for publication
Not applicable. Figure 1 Overview of local standard care screening procedure