Long-term motor outcomes of very preterm and/or very low birth weight individuals without cerebral palsy_ A review of the current evidence

We reviewed literature on long-term motor outcomes of individuals aged five years or older born very preterm (VP: ≤32 weeks of gestation) or with very low birth weight (VLBW: ≤1500g), without cerebral palsy (CP). PubMed produced 2827 articles, whereof 38 were eligible. Assessed by standardised and norm-based motor tests, the Movement Assessment Battery for Children being the most widely used, VP/VLBW individuals showed poorer motor skills compared with term-born controls with differences of approximately 1 SD in magnitude. Some studies assessed subdomains and differences were present in fine motor/manual dexterity, ball skills and gross motor/balance. Prevalence of motor problems varied largely from 8-37% in studies with cut-off at the 5th percentile or -1.5 SD to 12-71% in studies with cut-off at the 15th percentile or -1 SD. This review shows that the degree of motor impairments continues to be substantial among VP/VLBW individuals who do not develop CP.


Introduction
Motor function is one of the very early signs of typical or adverse neurodevelopment in the child's first year of life [1]. Motor impairments that might be seen in children born preterm vary from cerebral palsy (CP) to minor or subtle motor problems. Most children do not develop CP [2], and minor motor problems might not be evident before early school age when being competent in various motor skills becomes increasingly important for everyday life activities [2]. Fine motor skills or manual dexterity are required for writing, typing, drawing and manipulating objects, while gross motor or balance skills are required amongst others for walking, running and jumping. Further, ball skills include both fine and gross motor skills, and are important for many sports and leisure activities as the child grows older.
Motor problems have been consistently reported in children born very preterm (VP: ≤32 weeks) and/or with very low birth weight (VLBW: ≤1500 g). Over the past decades, there has been increased survival of VP/VLBW children, due to advances in prenatal and newborn care. However, rates of adverse neurodevelopmental outcomes have been relatively stable, with a slight decrease in moderate to severe CP [3][4][5] and an increase in non-CP motor impairment among extremely preterm (EP: ≤28 weeks) and/or extremely low birth weight (ELBW: ≤1000g) [6,7]. Motor problems, or developmental coordination disorder (DCD) as used by some authors, are usually defined as a score below a cut-off on a standardised motor test, compared with a normative sample or a control group. There are different recommendations as to cut-offs, ranging from scores −2 SD or < 5th percentile [8] to −1 SD or < 15th percentile [9].
Three systematic reviews on motor outcomes of VP/VLBW survivors https://doi.org/10.1016/j.siny.2020.101116 have been published previously [10][11][12], and in addition, five reviews including any children born preterm or with low birth weight (< 37 weeks/ < 2500 g) [13][14][15][16][17]. De Kieviet et al., 2009 [11] investigated the relationship between VP/VLBW and motor development from birth through adolescence. In comparison with term-born peers, VP/VLBW children obtained significantly lower scores on all the three motor tests included: the Bayley Scales of Infant Development version II, the Movement Assessment Battery for Children (Movement ABC) and the Bruininks-Oseretsky Test of Motor Proficiency (BOTMP). Edwards et al., 2011 [10] concluded that DCD was more prevalent in the VP/VLBW population than full-term/normal birth weight children, with significantly greater odds of developing the disorder (6.29; 95%CI: 4.37-9.05 for scoring < 5th percentile and 8.66; 95%CI: 3.40-22.07 for scoring 5 to 15th percentile). Both reviews included several studies that did not exclude children with mild CP. FitzGerald et al., 2018 [12] concluded that VP children aged 3-6 years had poorer motor outcomes compared with term-born children using the International Classification of Functioning, Disability and Health domains of body structure and function and activity. We aimed to review long-term motor outcomes including domainspecific motor skills, assessed by standardised motor tests, of VP/VLBW children with no manifest CP compared with term-born controls. We also examined the prevalence of motor problems in VP/VLBW children and risk factors for these problems. We focused on individuals ≥5 years of age, which is a commonly used cut-off in health statistics and the age when hospital follow-up programmes generally are terminated.

Methods
A comprehensive literature search was carried out by one author (PH) in MEDLINE Database, using PubMed, between 8 October and 6 November 2019. The search produced 2827 articles (Fig. 1). Search terms are listed in Appendix A.
One author (KAIE) conducted the initial screening of titles and abstracts, using the following inclusion criteria: exposure was VP/VLBW (gestational age ≤32 weeks or birth weight ≤1500g), age of assessment ≥5 years, motor outcome was assessed by a standardised and norm-based motor test, the results were reported as continuous scores (mean/median) compared with a control group or as proportion of children with motor problems according to a defined cut-off. The studies had to be observational cohort studies, exclude participants with all degrees of CP or present results for participants without CP. No study population size or publication year restrictions were applied. Included studies had to be original research articles with full-text available in English. If more than one study reported the same findings on the same cohort assessed at the same age only the publication on the original or larger study sample was included in the review.

Total motor test scores of VP/VLBW individuals without CP compared with term-born controls
Eighteen articles reported continuous total test scores compared with a term-born control group (Table 2). In studies with more than 20 participants, the VP/VLBW children had mean Movement ABC scores 4.0 to 8.5 points lower compared with controls at 5-6 years of age [28][29][30], corresponding to differences ranging from 0.93 [28] to 1.84 SD [29]. Mean Movement ABC-2 scores were 6.1 [31] to 13.3 [32] points lower, corresponding to a 0.70 SD difference in young adults [31]. The difference in median TOMI, Movement ABC or Movement ABC-2 scores ranged from 1.5 to 7.5 points in children aged 6-14 years [33][34][35][36][37][38][39]. Assessed by the BOTMP, mean differences were 10.7 and 4.9 points in VLBW children with and without BPD [40], and 8.0 points in ELBW children [41], corresponding to 0.41 to 0.89 SD difference from their respective control groups at 8 years of age. Measured by the KTK, differences in mean scores between VLBW and control groups ranged from 3 [22] to 7 points [21] and was as large as 12 points (1.50 SD) in ELBW children compared with controls at 6 years [22]. The only adult study [31] reported a difference of 3 points (0.88 SD) between the VLBW and the control group at 22.5 years of age, assessed by the HiMAT. VLBW adults used 5.6-6.3 s more than controls to complete the GP with the non-dominant and dominant hand, a difference of 0.53 and 0.66 SD, respectively [30], similar to the 0.50 SD difference in performance on the Purdue Pegboard in 9-year-old children for all three conditions [42] (Table 2).

Domain-specific test scores of VP/VLBW individuals without CP compared with term-born controls
Eleven articles reported continuous test scores for subdomains of manual dexterity/fine motor function, ball skills and balance/gross motor function (data not shown). Mean manual dexterity scores on Movement ABC were 1.9 points lower [30], and on Movement ABC-2 2.1 [31] to 7.7 [32] points lower, in VP/VLBW individuals compared with controls, corresponding to differences of 0.39 [31] to 0.91 SD [30]. Differences in median TOMI or Movement ABC manual dexterity scores ranged from 0.5 to 2.0 points [33,34,[37][38][39], and was 1.5 points lower for posting coins in ELBW children [43]. VLBW children had BOTMP scores 1.3 to 5.2 points lower for fine motor and upper limb tasks, and differences ranged from 0.34 to 0.63 SD for the various subtests [44].
For ball skills, VP/VLBW individuals had Movement ABC scores 1.5 points lower [30], and Movement ABC-2 scores 1.1 [32] to 1.4 [31] points lower compared with controls, corresponding to differences of 0.30 [31] to 1.66 SD [30]. The difference in median TOMI and j Odds ratio adjusted for mother's education, mother's country of birth, complex sample analysis to adjust for clustering effects caused by presence of twins.    [30]. Differences in median Movement ABC and TOMI balance scores ranged from 0.5 to 2.0 points [33,34,[37][38][39], and were 1.0-2.0 points for heel-toe walking and one-leg balance, respectively [43]. Differences in balance, laterality and jumping task scores of the KTK ranged from 0.50 to 1.00 SD for ELBW children and from 0.13 to 0.62 SD for VLBW children versus controls [22].
The prevalence of ZNA pure motor scores < 10th percentile was similar (24% and 25%) at 6 [61] and 10 years [62] in the same VLBW cohort. Poor fine motor scores occurred in 38% at 6 years, and 17% at 10 years, while poor gross motor scores occurred in 26% at 6 years and 34% at 10 years. Rates of poor static balance were rather similar at 6 (20%) and 10 years (18%) [62]. Holsti et al., 2002 [59] reported that of the 9-year-old ELBW children that were classified as having DCD based on BOTMP scores < -1 SD on the gross motor, fine motor or battery composite, 57% had a low gross motor composite score, 16% had a low fine motor composite score and 27% had low scores in both gross and fine motor composite areas.

Risk factors for poor motor outcomes in VP/VLBW individuals without CP
Twenty-five articles reported on risk factors for poor motor outcomes. Several studies found that boys performed worse than girls [33,43,45,48,53,56,57], but one study reported poorer outcome for girls [35] and some found no effect of sex [47,52,59,60].

Discussion
We identified 38 original publications fulfilling the inclusion criteria. Of them, 23 were not part of the three previous reviews of VP/ VLBW children [10][11][12], as we included more recent publications and articles reporting both continuous motor scores and/or prevalence of motor problems assessed by a wider variety of tests. Age at assessment varied from our lower limit of 5 years and up to 14 years, with the exception of the Norwegian NTNU cohort also assessed at 22.5 years [31]. All studies came from high-income countries except two studies from a middle-income country [30,54].
There were substantial differences between VP/VLBW children and controls across all ages. Typical mean differences in continuous scores were in the order of magnitude of 1 SD. In studies that reported subdomain scores, differences were present across all motor domains, as reported by de Kieviet et al., 2009 [10], with increased risk of poor manual dexterity/fine motor function, ball skills and balance/gross motor function. In accordance with Edwards et al., 2011 [9], the prevalences of motor problems were several-fold larger than those in controls, but with some uncertainty, partly because of low number of cases among term-born controls. There was no evidence that younger children had higher prevalence of motor problems, neither did it seem related to years of birth. Potential reasons for the wide variation in difference in mean scores and prevalence of problems may be related to differences in study designs and settings, selection and exclusion criteria, follow-up rates and applied test norms.
These differences compare with differences in other key outcomes in VP/VLBW children and adults. Cognitive abilities have been extensively studied and remain 0.70 to 0.86 SD lower in VP/VLBW children than in controls [63]. Other examples include pulmonary airflow, showing a similar difference, 0.80 SD, in 1-s forced expiratory flow in VLBW adults [64], whereas the difference in blood pressure is smaller, corresponding to 0.3 SD [65,66]. Together these numbers indicate that poorer motor skills, even without the presence of manifest CP, are an essential part of the "VP/VLBW phenotype" at least in childhood.
There is little evidence on whether and to what extent poorer motor skills persist to adulthood. The only published study extending to adulthood is the Norwegian NTNU cohort [31], which suggests that the deficits in fine and gross motor skills in VLBW children persist to adulthood. While the mean differences with controls were relatively small in comparison with studies in childhood, the proportion of those with motor problems was similar to what was shown in the same cohort at 14 years of age [67]. This paucity of evidence contrasts with the increasing research activity on other outcomes in adult VP/VLBW cohorts and may in part be due to a lack of assessment tools validated for young adults; we argue for incorporating motor outcomes in follow-up studies in these cohorts.
Even though some other reviews have included all degrees of prematurity [13][14][15][16][17], the majority of the included studies focused on VP/ VLBW or more immature or smaller subgroups. Much of the improvement in VP/VLBW survival has been related to improvements in resource-intensive neonatal care units in high-resource settings. These typically also have follow-up and medical, social and educational support systems in place, which is reflected in the settings of the included studies. With improving prenatal and newborn care, survival of VP/ VLBW infants is increasing also in lower-resource settings [68]. The published results may therefore reflect a conservative estimate in relation to these settings which may also have lower resources for followup and support.
According to Spittle & Orton 2014 [69], all forms of motor impairment are associated with comorbidities. Thus, there is reason to believe that the motor problems seen in VP/VLBW children are part of more widespread brain pathology [70], and may act as an early biomarker for cognitive and behavioural problems later in childhood [70]. Motor and coordination problems in childhood and adolescence have been found to be associated with overall psychiatric problems in VLBW adults [71], and elevated levels of inattention and symptoms of anxiety and depression in both ELBW and control adults [72]. These co-occurrent problems may have a greater effect on quality of life, academic achievement and participation in extracurricular activities than the motor impairment itself [69]. Several studies have reported lower physical activity, sports participation and recreational activities among young adults born VP/VLBW [73][74][75][76][77] or less than 34 weeks of gestation [78]. While these differences have usually not been seen in studies measuring physical activity by accelerometry [79], they are accompanied by findings of lower lean body mass [80] and physical fitness [74,81]. They may be a result of a vicious cycle where poor motor coordination have been suggested to play a key role, perhaps together with lower pulmonary airflow [64] and poor visual acuity [39]. This may lead to lower physical self-confidence and perceived physical ability [82], together making physical activity less rewarding, leading to lower degrees of physical activity and may also aggravate the lower exercise capacity and lower lean body mass. Therefore, we believe that promotion of health-enhancing physical activity, with special reference to challenges in motor skills, has potential in improving the health of children and adults born VP/VLBW. Spittle & Orton 2014 [69] argue that a reliable diagnosis of motor impairment requires follow-up into childhood and is important to ensure that the appropriate intervention is implemented. Interventions that are activity-oriented and involve environmental support by family and significant others to facilitate participation across contexts are recommended [9]. However, evidence regarding which interventions are most effective in improving motor outcomes is sparse and few studies have assessed long-term effects [83,84].
Several studies reported neonatal risk and male sex as factors explaining part of the motor problems. These findings are supported by the review of Bos et al., 2013 [17], who reported intrauterine growth restriction, inflammatory conditions and BPD as risk factors for fine motor skill impairments in preterm children. Furthermore, Linsell et al., 2016 [85] reported IVH and/or PVL to be a prominent feature in the medical history of children not only developing CP, but also minor neurological dysfunctions.
This review was based on a systematic literature search. We focused on children aged five years or older, whereas previous reviews also have included younger children. We excluded articles that included even mild degrees of CP, and we present both continuous motor test scores and prevalence of motor problems, as well as domain-specific test scores and problems. One reason for not performing a meta-analysis was the different methods used to assess motor skills. Although most studies used either the Movement ABC or Movement ABC-2, some reported mean (SD), whereas other reported median (IQR or range) scores, and different cut-offs were used to identify motor problems.

Conclusion
Including recent studies and excluding all degrees of manifest CP, we found that VP/VLBW individuals had poorer motor skills in terms of total test scores, domain-specific test scores and increased prevalence of total motor and domain-specific problems. There was a wide variability in results between studies, however it seemed not related to age or years of birth. The motor problems are present when the children generally no longer attend preterm-specific follow-up programs. Healthcare and education professionals should be aware of increased risk of such problems, to provide appropriate intervention, including environmental adjustments and understanding of their motor problems, to promote participation and reduce the impact on everyday activities and future health problems.

Practice points
• Children and young adults born very preterm or with very low birth weight have increased risk of poor motor skills in terms of total test scores and domain-specific test scores.
• Healthcare and education professionals should be aware that these problems are manifest when the children generally no longer attend preterm-specific follow-up programs and provide appropriate interventions to reduce the impact on everyday life.

Research agenda
• Motor problems in the much larger groups of children born moderate and late preterm in high and low resource settings.
• Persistence of motor problems and their significance in later adult life.
• The role of motor problems in developing physical activity habits and consequently physical fitness.
• Interventions and their effectiveness long term.

Funding sources
The work of Drs. Evensen 4-12 y Eight items in three subscores; manual dexterity (three items), ball skills (two items), static/dynamic balance (three items).
Raw scores converted to standard scores on a 6-point scale (0-5; 0 being optimal score) for each item and summarised to a total impairment score (maximum 40). Scores < 5th and < 15th percentile indicate definite and borderline motor problems. Movement Assessment Battery for Children-2 Second edition (Movement ABC-2) [8] 3-16 y Eight items in three subscores; manual dexterity (three items), aiming and catching (two items), balance (three items).
Raw scores converted to standard scores 1-19 for each item (higher indicating better skills) and summarised to a total score (maximum 152). Scores ≤5th and ≤15th percentile denote a significant movement difficulty or suggest the child is at risk of having a movement difficulty. Test of Motor Impairment (TO-MI) [20] 5-12 y Eight items in three subscores; manual dexterity (three items), ball skills (two items), static and dynamic balance (three items).
Scoring system varies with each item, ranging from a 2-point (pass/fail) to a 16-point scale. Raw scores converted to a numerical point score and composite scores expressed as centile rank, z-score, T score (mean 50; SD 10), stanine or age-equivalent. Körperkoordinationstest für Kinder (KTK) [21,22] 5-14 y Four tasks; walking backward on beams of decreasing width, jumping with each leg separately over an increasing height (number of foam plates), jumping sideways with both feet and moving across the floor by stepping from one plate on a second plate, then relocating the first plate before taking the next step and so on.
Performances converted to a standard score, or motor quotient (MQ). A MQ between 90 and 110 represents normal coordination, while scores < 80 indicate deficiencies in motor ability.
Raw scores summarised to standard components and expressed as z-scores. Scores < 10th percentile indicate abnormal performance. High-Level Mobility Assessment Tool (HiMAT) [25] > 13 y 13 items; walk, walk backwards, walk on toes, walk over obstacle, run, skip, hop forward, bound and walk up/down stairs.
Raw scores are converted to a score on a 5-point scale (0-4), except stair items on a 6-point scale (0-5; higher score indicates better function). Maximal score of 54.
Grooved Pegboard (GP) [26] > 5 y Inserting 25 pegs, one at a time, into keyhole-shaped holes with various orientations in a 5 × 5 matrix. Dominant and non-dominant hand are tested.
Raw scores consist of time in seconds to complete the board and number of drops for each hand. Age-specific norms.
Purdue Pegboard [27]] 5-89 y Placing as many pins as possible in the holes, within a 30 s time period. Preferred hand, non-preferred hand and both hands together are tested.
Raw scores consist of number of pins inserted within the 30 s time period for each hand. Score for the bimanual condition consists of the total number of pairs of pins inserted.