Common Core Assessments in follow-up studies of adults born preterm—Recommendation of the Adults Born Preterm International Collaboration

Background: Of all newborns, 1%-2% are born very preterm (VP; < 32 weeks) or with very low birthweight (VLBW; ≤1500 g). Advances in prenatal and neonatal care have substantially improved their survival, and the first generations who have benefited from these advances are now entering middle age. While most lead healthy lives, on average these adults are characterised by a number of adversities. These include cardiometabolic risk factors, airway obstruction, less physical activity, poorer visual function, lower cognitive performance, and a behavioural phenotype that includes inattention and internalising and socially withdrawn behaviour that may affect life chances and quality of life. Outcomes in later adulthood are largely unknown, and identifying trajectories of risk or resilience is essential in developing targeted inter-ventions. Joint analyses of data and maintenance of follow-up of cohorts entering adulthood are essential. Such analyses are ongoing within the Adults Born Preterm International Collaboration (APIC; www.apic-prete rm.org). Joint analyses require data harmonisation, highlighting the importance of consistent assessment methodologies. Objective:

approaches one million adults in the European Union. 6 With improving survival, 1 their proportion is also increasing in middle-and low-income countries.

| Adults born VP/VLBW-what is currently known
While most VP/VLBW young adults lead healthy lives 7 and many case histories point to remarkable resilience, 8 VP/VLBW adults on average have more physical, cognitive, behavioural, and social challenges than their peers born at term. These challenges include increased levels of cardiometabolic risk factors, 9,10 airflow obstruction, 11 lower bone mineral density, lower levels of physical activity, 9 lower scores on tests of cognition 12 and educational attainment, 7,13 higher rates of mental health problems, 14,15 difficulties in establishing social relationships, and being less likely to partner and reproduce. 16,17 Studies have thus far extended to young adult life, around the peak of most physiological functions, but late-life outcomes are largely unknown. Even with average rates of physiological decline, individuals at lower peak levels are expected to attain symptomatic levels of impairment earlier.
Moreover, adaptations to pre-and neonatal adversities could come at the cost of more rapid rates of decline in cardiovascular, pulmonary, and cognitive function, or "accelerated ageing." Not all findings are negative, including less externalising and rule-breaking behaviours, 14 and in some studies lower risks of substance abuse, 15 and atopic allergy. 18

| Adults born preterm international collaboration and common core assessments recommendation
Most of existing literature comes from register studies or a small number of longitudinal clinically studied case-control cohorts in high-income countries. Clinical cohorts include much more detail on physical and behavioural traits, but typically only range from tens to a few hundred VP/VLBW cases. This prompted researchers in the field to form the APIC Adults Born Preterm International Collaboration (www.apic-prete rm.org) to address study questions that cannot be assessed in one study alone. One aim of the collaboration focuses on individual-participant meta-analyses that have been performed for blood pressure, 10 adult behaviour problems, 14 and lung function. 11 Studies are currently underway addressing adolescent psychiatric symptoms, psychiatric diagnoses, adult body size and pubertal growth, handgrip strength, and effects of childhood cognitive and mathematic abilities on adult wealth. Such meta-analyses and crossvalidations are only possible if similar measures of adult outcomes have been used. Therefore, the other aim of the APIC collaboration is to agree on common core measures for key outcomes relevant in studying adults born preterm. Many APIC partners also participate in other similar initiatives such as the EU-funded RECAP Research on Children and Adults Born Preterm (www.recap-prete rm.eu) that is developing a technical platform for secure distributed analysis of harmonised joint data sets for RECAP partners and non-partners alike.

| Common core assessments-process, rationale, and criteria
The process was started by discussing principles of Common Core Assessments at APIC meetings. Experts of specific outcome domains (Appendix 1) wrote the initial draft on assessments pertaining to that domain. These drafts were combined and revised based on review by all contributors. Consensus was reached by discussion at APIC meetings.
The most important benefits of having harmonised measures in VP/VLBW follow-up studies include the following: • Assessment of hypotheses for which single studies have insufficient power • Assessment of whether findings are universal versus country-/ culture-specific (cross-validation) • Comparison of adult outcomes of infants born during different periods, which also serves as a "quality control" for prenatal and neonatal treatments during a specific period.
Criteria for selecting core measurements were decided as follows: • Measures an outcome relevant to study in VP/VLBW adults (ie, the "preterm phenotype") • Measures an outcome relevant for health, community participation, and quality of life, or is predictive of such an outcome • Adheres to standard administration and test performance (validity and reliability) criteria • Is cross-culturally comparable • Has population-specific norms, if the aim is to use cut-offs (eg to identify impairment) • Is feasible to administer (in terms of time and training required) • Measures that are widely used have an advantage (well known, likely to have been used before in VP/VLBW cohorts).
We have focused on methods that are relevant in the follow-up of VP/VLBW adults, because this group carries the highest risks and the largest investments in peri-and neonatal health care resources.
Many of the recommended measurements are also relevant in following up adults born at any degree of preterm birth or in general population health examinations. Similarly, many of these adult measures have versions that can be used in childhood and adolescence, and can be useful in assessing continuity across ages.
Another important consideration is appropriate equipment, training of staff, and quality control to allow standardised performance and comparability of measurements. For the core measurements, we describe key methodological aspects, pitfalls, and references to more detailed guidelines.

| Cardiometabolic and related biomarkers and blood and urine samples
Many of the recommended measurements (Table 1)  The guidelines include recommendations on avoiding vigorous exercise, smoking, and use of alcohol. 22 Menstrual cycle phase, pregnancy, lactation, and menopause should be recorded; some physiological measurements may not be informative and some may be contraindicated during pregnancy.

| Height, weight, and waist circumference
While the measurement of height and weight may seem trivial, rigorous protocols, appropriate training, and regular calibration of scales are essential. Waist circumference provides a simple measure of abdominal, metabolically harmful obesity, but is particularly prone to error and highlights the importance of training and quality control. 22 Other potentially relevant body proportion measures are listed in Table 1.

5-10
Sleep is a key determinant of health and has been little studied in VP/VLBW adults. Some studies report more morningness, which is a predictor of good health

Food and nutrient intake
Questions on single items embedded in national surveys can be used. 23 More comprehensive analysis requires a food frequency questionnaire or food diary 60

Variable
Little studied in VP/VLBW adults. Calculation of nutrient intake from data from data on food intake requires a (national) nutrition database, which includes country-/culture-specific average nutrient contents of food items/meals Substance use Drug Abuse Screening Test (DAST-10) 27 is a widely used tool to assess the use of substances other than alcohol or tobacco  Table 4 Body Morning sample after overnight fast. The use of a non-fasting sample is possible but will limit the utility of a number of metabolic markers.
c Time for drawing blood samples and ingesting glucose/infusing inulin. Assessments that do not require significant mental of physical exertion can be performed between the samples. d A spot urine sample may be random or standardised. A frequently used standardisation is to use "day's 2nd urine": the participant, who attends a clinic for a fasting blood sample, voids urine when getting up at home, comes to the clinic and gives a urine sample at the clinic.

TA B L E 1 (Continued)
In studies using only questionnaires, we recommend inclusion of self-reported height and weight, which, however, include well-documented sources of error. 26

| Blood pressure
Blood pressure has the highest global public health burden of all risk factors and is elevated in VLBW 10 and other preterm adults. We propose blood pressure and heart rate as one of the core assessments.
A list of validated blood pressure devices and validation standards is available at the British Hypertension Society website (http://www. bhsoc.org/bp-monit ors/bp-monit ors/). National/regional standards on the arm of measurement and number of recordings vary; the local standard should be used and reported. Sources of measurement variation, which are well characterised, 22 should be minimised. In particular, it is important to use a cuff appropriate to the participant's arm circumference.

| Blood sample
Although venipuncture can be a concern for some adults, feedback on analyses made from a blood sample can also serve as motivation provides a reasonable compromise. In any case, time of day, fasting time, and any irregularities in sampling should be recorded.
Usually, some key analyses are performed immediately or within a few weeks, in part to provide feedback to the participants, while the remaining plasma and serum are stored for later analysis. These analyses often include plasma total, high-, and low-density lipoprotein cholesterol and glucose and may also include haemoglobin A1c (HbA1c), analysed from whole EDTA blood. A blood cell count needs to be analysed from fresh samples. For these analyses, the interassay variation in routine analysers is small enough not to cause concern if the same analyses are run by the same method in the same laboratory along the clinical visits, whereas for some other analytes (eg, insulin), analysis of all samples at the same time may be important to reduce interassay variation.
As for storing samples, most relevant biochemical analyses can be done from serum and plasma, which are stored at −80°C or in liquid nitrogen in small (1 or 1.5 22 mL) aliquots. For DNA extraction, adult studies usually collect one or two ~5-to 10-mL samples of EDTA blood, which can be stored at −20°C. If the pre-extraction storage time is expected to exceed 2 years, storage at −80°C is preferred. If a blood sample is not feasible, DNA can also be obtained from saliva, which can, with adequate preservatives, be stored for years in room temperature. DNA yield is higher in salivary than in buccal swab samples.
When obtaining biological samples for storage, it is important that the consent forms adequately cover the use of samples for future analyses, and for international sharing of pseudonymised data.
Consent forms for modern population health surveys and biobanks may be helpful.

| Health, medical history, and medication
We recommend the Minimum European Health Module that comprises three questions on self-perceived health, long-standing (6 months or more) illness, and functional limitations. 22

| Respiratory outcomes
Adults born preterm have impaired airflow. 11 Table 2 lists recom-

| Ophthalmic outcomes
There are few comprehensive data on visual outcomes in VP/VLBW adults. In early childhood, there is an increased risk of problems including myopia, strabismus, and amblyopia as well as cerebral visual impairment associated with white matter damage. Table 3 lists common core assessments recommended for visual outcomes, which will be impacted by the presence of retinopathy of prematurity (ROP) and its severity, treatment, and other morbidities, 30 all of which should ideally be prospectively documented. Current populationbased studies suggest around 22% of VLBW infants will have some acute ROP. 31

| Cognitive and motor function
Cognitive impairment is one of the primary concerns for parents of very preterm infants. Extensive research has shown that children and adolescents born VP have on average lower general intelligence (IQ) than full-term peers, 12,32 performing approximately two-thirds of SD lower than expectations. 33 34 Studies in adulthood are few, and we recommend a motor performance test as a core assessment (Table 4).

| Mental health
VP/VLBW birth is associated with a specific cluster of mental health problems that manifests in a high prevalence of ADHD, predominantly the inattentive subtype, anxiety, and autism spectrum traits and disorders across the life course, with some evidence for increased rates of depression. 14,35,36 Accordingly, VP/ VLBW adults have been described as easily worried, rigid in communication, and socially withdrawn. 14,[35][36][37] This preterm behavioural phenotype is evident in studies using both screening tools and diagnostic interviews.
Diagnostic studies represent the gold standard, and we recommend a structured clinical diagnostic interview when psychiatric diagnoses are of interest. While these assessments of dichotomous outcomes may lack statistical power in individual studies, 37 the data can be relatively easily harmonised and adequate power achieved through data pooling with other cohorts.
However, many mental health problems may be best described as a range of symptoms along a continuum rather than a categorical diagnosis. Accordingly, VP/VLBW individuals' daily functioning may be significantly impaired by such subthreshold symptoms and/ or not otherwise specified (NOS) disorders. As an economical and ecologically valid alternative, researchers have used standardised, cross-culturally reliable screening questionnaires 14,38,39 to assess symptoms across multiple domains including behaviour, attention, anxiety, mood, social, and communication problems (Table 5).
Administering these symptom scales to different informants (eg, parents or partners, and adults themselves) may increase objectivity, validity, and reliability of information as self-reports may underestimate symptom severity. have not been shown to differ from each other in psychoticism, 40 and findings with regard to conscientiousness and openness to experience are mixed. [41][42][43] One study has additionally reported that VLBW adults scored lower on impulsivity and hostility. 41   Another widely used indicator is income, which is more vari-

2-6
Having no experience on sexual intercourse more common. 51,52 Sexual orientation and gender identity Self-identification, grading ranging from exclusively heterosexual to exclusively homosexual and exclusively female to exclusively male 1 Non-heterosexual self-identification more common. 52

| Health-related quality of life
It is important to consider an individual's subjective evaluation of their health status, which refers to functional issues and limitations.
Health-related quality of life (HRQoL) implies a subjective appraisal of an individual's health status, preferably by the individuals themselves. Such an appraisal is related to, but not directly determined by, Both approaches have advantages and drawbacks. Therefore, in long-term follow-up research of preterm birth, we propose to combine the two approaches by using the SF-12. 47 The SF-12 is a 12-item subsample of SF-36 items yielding an eight-dimension health status profile. In addition, SF-12 data may be directly mapped into a six-dimension health status classification system (SF-6D) that comes with a health utility function yielding a single preference-based health utility score. 48 Psychometric performance of SF-12 has been documented and validated in many language versions worldwide.

| Social relationships, economic independence, and reproduction
Differences in physical, mental, and cognitive health and childhood social experiences 49 may affect life chances such as independent living, income and social relationships with peers, partnering, and having own children. 13 Fewer women and men born VP have children compared with their peers born at term. 50 Whether this is a voluntary choice or due to the lack of a partner, 51 sexual orientation, 52 or biological fertility problems is not known. Fertility remains a substantial gap of knowledge. Moreover, women and possibly men born preterm have an increased risk of having a preterm child or other pregnancy complications themselves. 50 We recommend a life course interview with focus on transition to adulthood and pregnancy history; depending on resources and study focus, the information may in part be collected by questionnaire. The core items are shown in Table 6.
It is also important to realise that preterm birth may have a substantial impact on the birth family. Giving preterm birth is an important predictor of at least maternal health including cardiovascular disease.
Families that give birth to a preterm child have fewer subsequent children, 53 and preterm birth may have a deep lasting impact on the siblings. 54 While this paper focuses on measures in adults born preterm, researchers may wish to consider incorporating elements of follow-up of parents and siblings.

| CON CLUS IONS
The suggested measures are intended to help current and future researchers or clinicians involved in follow-up of adults born preterm to compile an assessment programme. Adopting these recommendations of well-tested measures will allow for consistent comparisons of outcomes across cohorts and countries, often hampered by heterogeneity of measures or missing measures on key areas of functioning.