Article Text


Risk of upper limb injury in left handed children: a study in Greece
  1. Alkistis Skalkidou1,
  2. Eleni Petridou2,
  3. Nick Dessypris1,
  4. Elias Karanikas3,
  5. George Pistevos4,
  6. Dimitrios Trichopoulos2
  1. 1Department of Hygiene and Epidemiology, Athens University Medical School, Athens, Greece
  2. 2Department of Hygiene and Epidemiology, Athens University Medical School, Athens, Greece and Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
  3. 3First Department of Orthopaedics, Aglaia Kyriakou Children's Hospital, Athens, Greece
  4. 4Second Department of Orthopaedics, Aglaia Kyriakou Children's Hospital, Athens, Greece
  1. Correspondence to:
 Dr Eleni Petridou, 75 M Asias St, Athens, 11527, Greece
 (e-mail: epetrid{at}


Objectives—To investigate whether left handed children are at increased risk for injuries, particularly upper limb injuries.

Setting—Athens, Greece, during a six month period in 1995–96.

Methods—Cases were 129 children 4–14 years old with unintentional upper limb injuries from a population based injury database. Two control children matched for gender and age were selected from among those seen at the same medical institution for minor, non-injury ailments. On the basis of information provided by the children and their guardians, sociodemographic variables were recorded, hand preference was assessed, and each child's activity score was calculated through an abbreviated version of Achenbach's scale.

Results—Left handed children have a moderately increased upper limb injury risk with a tendency of recurrence of this injury. The risk of upper limb injury is also raised among children of young fathers, whereas it appears to be inversely related to crowding index and activity score—three variables that were controlled for as potential confounders.

Conclusions—This study provides limited support for the hypothesis that left handed children are at increased risk for injury. The excess risk, if genuine, is likely to be limited to cultural settings in which right handedness is perceived as the norm.

Statistics from

Accident proneness is a concept that was popular in early injury research1,2 but gradually fell out of favour, because it came to be identified with the discredited “blaming the victim” philosophy. Second thoughts, however, have been recently expressed,3,4 albeit in a different context. An inherently increased injury hazard is now being linked, not to psychological predisposition, but to personal traits acting in conjunction with environmental conditions.

Recent reports pointed out that left handed children may be at increased injury risk.5–7 This could be attributed either to psychological reasons, reflecting developmental processes, or more likely, to structures and functions in our daily lives that have a built-in bias in favour of the right handed. Studies of upper limb injuries, rather than of injuries in general, could be more powerful in documenting an increased injury hazard of left handed persons, because any hand dexterity-related functional irregularity would be more likely reflected in injuries of the upper limbs. We have, thus, undertaken an epidemiological investigation of risk factors, including handedness, for upper limb injuries.


Data were derived from the Emergency Department Injury Surveillance System (EDISS) database developed by the Centre for Research and Prevention of Injuries among the Young in Greece. In EDISS, data are recorded for individuals who seek medical attention at any of a network of hospitals for an injury of any nature.

In this study, children who were over 4 years old and, therefore, had already developed hand preference, were enrolled if they were seen at one of the two children's teaching hospitals in Athens for an injury of any kind. These two hospitals are located in the same area and, together, cover more than 80% of all childhood admissions.8 The two hospitals accept emergency pediatric cases on alternative days throughout the year so that children admitted to either one are an essentially unselected sample of all injured children in the underlying population.

During the six month period, November 1995 to April 1996, 129 children aged 4–14, with unintentional upper limb injuries were registered in the EDISS database. In all but seven instances, a successful interview was conducted. The interviewers were not blinded but were unaware of the hypothesis under investigation. For every participating child with an upper limb injury, two controls matched for age (±1 year) and gender were randomly selected among those who contacted the hospital with minor, non-injury ailments during the same period. An interview could not be arranged in 17 instances, mostly for administrative reasons, and other children were substituted on the basis of the same criteria.

The interviews, that lasted on average 35 minutes, were conducted by four trained professionals (health visitors), the same one for each matched triplet. Recorded information covered sociodemographic characteristics of the children and their families, history of previous upper limb or other injury, and conditions and aspects of the upper limb injury (cases only). In addition, an activity score was calculated using a 20 item version of Achenbach's child behaviour checklist.9–12

Hand preference was assessed on the basis of the response of the child and the child's guardian to seven questions13: which hand was used in throwing a ball; teeth brushing; writing; drawing a picture; cutting a piece of paper with scissors; unlocking a door; and card dealing. When both the child and the guardian indicated consistent use of the right hand for a particular task, he/she was assigned a value of 1. When there was agreement on consistent use of the left hand, the child was considered left handed and was assigned a value of 0. When different answers were given, or the use of either hand was indicated for a particular task, the child was considered ambidextrous and was assigned a value of 0.5. A composite score of hand dexterity was generated by adding the seven task specific values. A child with a composite score higher than 4 was considered right handed, a child with a composite score less than 3, left handed, whereas those who scored from 3–4, inclusive, were considered ambidextrous.

Statistical analysis was done through simple bivariate classifications and, subsequently, through conditional logistic regression that took account of the matched design.14


Table 1 shows the distribution of cases by anatomic location, mechanism, and type of injury. This table demonstrates the typical range of upper limb injuries in children.

Table 1

Distribution of 122 cases of upper limb injuries by anatomic location, mechanism, and type of injury

Table 2 shows the distribution of cases and matched controls by sociodemographic and behavioural variables that could have confounding potential and thus, should be controlled for. These data are not directly interpretable, because of mutual confounding, but suggest that paternal education, crowding index, previous upper limb injury, and handedness may be significant correlates of upper limb injuries.

Table 2

Distribution of 122 cases with upper limb injuries and 244 age and gender, matched controls by study variables

Table 3 shows multiple logistic regression derived, mutually adjusted odds ratios for upper limb injuries by the study variables. Left handed children appear to be at higher injury risk, although the excess is not statistically significant. A previous upper limb injury carries a significantly increased risk for a similar injury, whereas there is no association between past history of injury at other body sites and subsequent upper limb injury (data not shown). The risk of injury is, however, significantly raised among children of young fathers, and unexpectedly, the crowding index is inversely related to injury risk. Lastly, in comparison with average activity level, both hypoactive and hyperactive children are at reduced upper limb injury risk, although the risk reduction is significant only among hyperactive children.

Table 3

Multiple logistic regression derived, mutually adjusted odds ratio (OR) and 95% confidence interval (CI) for upper limb injury by study variables


In our study, left handed children, who represent about 10% of the total,13 appear to be at increased risk for upper limb injury. This result did not reach statistical significance, but the likelihood of it being genuine is increased by the fact that upper limb injuries had a significant tendency for recurrence, whereas there was no association between current upper limb trauma and history of injuries in other body sites. Other studies have reached a similar conclusion.5–7,15 There have been reports, however, indicating that there is no association between hand preference and hand injury risk but the risk is higher only in ambidextrous individuals.16 Moreover, Hicks et al believe that people tend to sustain more injuries in the more frequently used hand, be that right or left.17

The use of hospital controls in case-control investigations has been widely debated in the epidemiologic literature.14 Although, population controls are generally considered preferable, they have a non-response proportion exceeding 50% in most settings.14 The biases that may result from non-response are of unknown magnitude and direction. In contrast, hospital controls with a wide range of diagnoses are unlikely to generate converging bias, because different diseases and conditions have different psychosocial correlates. Moreover, the problem of selection of cases as a source of concern in this study was minimized, because the participating hospital has near complete population coverage of the city of Athens and there were few refusals. Information bias is also an issue in case-control investigations and it can never be excluded with absolute confidence when information relies on responses that are subjectively conditioned. The similarity of interview conditions in cases and controls, and use of the same interviewer for each matched triplet, however, substantially reduces the likelihood of this bias. Moreover, the matched design, and the subsequent modelling of the data through conditional logistic regression, facilitates control of mutual confounding among the studied variables.14 Lastly, the study is reasonably large and sufficiently powerful to identify genuine associations in the twofold range.

Several other results have emerged from the study of potential confounding variables, that were controlled for, and these are compatible with injury profile in general, as well as in the Greek population.18,19 Thus, children of young fathers are at increased injury risk, and those of more educated fathers appear to be at reduced risk.20 After controlling for socioeconomic status through paternal education, the number of persons per room are inversely associated with upper limb injury risk. This finding, although unexpected, is not implausible,21 because frequent presence of adults with children in the same space implies more intense supervision with substantial injury prevention potential.

A surprising finding was that hyperactive children, that is those in the upper quartile in this dataset, were at reduced risk for upper limb injury. This result can not be explained in terms of random misclassification linked to administration of the Achenbach instrument under suboptimal conditions, because this type of misclassification can only attenuate associations.14 The present study obviously needs to be replicated. It is noteworthy, however, that a similar association was found in a study of burns.22

Implications for prevention

This investigation provides limited support to the hypothesis that left handed children are at increased risk for upper limb injury and the collective evidence remains equivocal. It is possible that the increased risk for upper limb injuries of left handed children, if indeed it exists, is limited to cultural settings in which right handedness is presumed to represent the norm. It is worth noting, however, that most negative studies have been reported from countries with widespread and long term tolerance for the liberal development of hand preference. If left handedness were established as a risk factor for injuries, or even in the absence of conclusive evidence, preventive measures could be considered along three lines: (1) parental alertness that goes beyond acceptance of left handedness and incorporates guidance and advice on both behavioural and technical issues; (2) notification of teachers and other care givers so that they will provide a supportive environment both during educational and sports activities; and (3) ergonomic adjustment of instruments, when feasible, and changes in the layout of furniture and other equipment to increase compatibility of the macroenvironment with the needs and preferences of the left handed.


The authors would like to thank the health visitors M Diamadopoulou, E Maragaki, E Mera, E Simou, and E Tzemanaki for their contribution in the collection of the data for this study.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.