Head injury is recognised as a major public health problem that is a frequent cause of death and disability in young people and makes considerable demands on health services. Epidemiological data are required to initiate appropriate preventive measures and to plan necessary services. However, reliable statistics are difficult to extract from routinely collected data.

International statistics for accidental deaths and road accident deaths do not identify head injuries, but they do indicate differences in accident rates between countries and over time. For example, road traffic accident (RTA) deaths are more than twice as frequent in France, Australia, and the USA as in the UK or the Netherlands, but in developed countries they are steadily decreasing each year.1 In developing countries accident rates are increasing as traffic increases, and they greatly exceed those of developed countries. Asked about the main health hazard of the next decade a Chinese professor of public health replied “the motorcycle”. Head injuries account for one quarter to one third of all accidental deaths, and for two thirds of trauma deaths in hospital. They are also the main cause of lifelong disability after trauma.

While the codes of the International Classification of Diseases do allow some estimates of the frequency of head injuries from routine statistics on deaths and hospital discharges, the 10 codes that cover head injury are variably applied; moreover, multiple injuries and transfers after first admission make estimates inaccurate. Injury severity scores used for trauma in general have been shown to be very inaccurate when applied to head injuries. Although people who present to hospital but are not admitted are a major part of the head injury problem for hospitals, no routine statistics are published for accident and emergency departments. Data published from clinical series are mostly from neurosurgeons whose selection criteria for transfer vary widely and whose population base can seldom be defined, making these data of limited use for epidemiology.

The best sources are therefore research surveys, of which there have been several from different countries in the past 10–20 years.1 The problem with these surveys is the variation in the definition of head injury and its severity. Most definitions of head injury depend on evidence of a blow to the head, and exclude facial injuries and foreign bodies in the nose and ears. Scalp, skull, and brain can be injured independently of each other, so that only a proportion of patients with head injuries have initial brain injury. About half of all attenders at accident and emergency departments have a scalp laceration and 15% have some evidence of brain damage, but only 2% a skull fracture.2 One prominent US epidemiologist suggested that only patients with brain damage should be counted as head injuries—that is, those with some impairment of consciousness either before reaching hospital or on arrival.3 But that overlooks the fact that patients with mild injuries without obvious brain damage make a considerable impact on the health care system, both as attenders at accident and emergency departments who are sent home and as admissions. The concern is that a small minority of patients with these mild injuries have serious complications: intracranial haematoma, brain swelling, intracranial infection or epilepsy. They are also important in exploring patterns of causation as the basis for prevention strategies, because most mild injuries might have been more severe with only slightly different circumstances. As for defining severity, most countries use the Glasgow coma scale4: the score can range from 3 to 15 and a convention has emerged that patients with a coma score of 8 or less are classed as severe and those with a score of 13 or more as mild, the others being classed as moderate. Paediatric coma scales have been developed for assessing younger children.5

The only systematic data on attenders to accident and emergency departments are from surveys carried out across Scotland in 1974 and 1985, and in one district general hospital in Glasgow in 1984.2 These provide a dataset of 12 000 attenders, which indicates that about 10% of all new attenders at accident and emergency departments present with head injuries. Studies have also been done of all admissions and of neurosurgical unit transfers throughout Scotland.6 Head injury deaths can be difficult to ascertain from hospital surveys because about half of them occur at the scene and because many fatalities have multiple injuries.7

Another source of confusion is the definition of a child. In all our studies we have used under 15 years, but others use under 16 years and some children’s hospitals accept only patients of 12 or under. Some studies in other countries classify by decades and record all those under 20 as children. In regard to cause and injury type, complications, and mortality, 15–20 year olds are much more like young adults than children.

The proportion of children in each of the different categories of head injuries varies greatly,2 from nearly half of attenders to 20% of deaths (table 1). About half of accident and emergency child attenders are younger than 10 years, but only a fifth of the severely injured and a third of fatalities are patients younger than 5. Male patients make up more than 70% of attenders over the age of 5 but only 60% of the younger children are male.

There are 4011 attenders, 400 admissions, and 5.3 deaths per 100 000 children per year.1 Age specific rates for deaths and admissions from head injury in the UK across all ages show the peak incidence in the 15–30 age range, but for attenders at accident and emergency departments, this peak is younger than 10 years.1 Although the case fatality rate for admitted children is only 0.7%, a fifth of the rate for all ages, the large number of admissions together with the number of deaths before admission means that head injury is an important cause of death in childhood. According to Sharples et al,5 head injury is the most common cause of death between 1 and 15 years, accounting for 15% of all deaths, and for 25% of deaths in the 5–15 years age group.

Neuropathologists in Glasgow have compared detailed necropsy findings in 87 children aged 2–15 years with those for 360 adults.8 The frequency of contusions, diffuse axonal injury, ischaemic brain damage, and intracranial haematoma was similar in adults and children. However, bilateral cerebral swelling was three to four times more common in children. In 40% of those the swelling was not accompanied by contusions, ischaemic brain damage or intracranial haematoma—one or more of which are usual in adults with brain swelling.

The vast majority of patients with head injuries coming to accident and emergency departments, and of those admitted, have mild injuries. In the Scottish studies, 80% of admissions of all ages were for mild injuries, two thirds having neither a skull fracture nor evidence of brain damage, nor an extracranial injury requiring admission in its own right, and two thirds were discharged in less than 48 hours. These admissions reflect extreme caution before guidelines based on statistically calculated risk factors were published (see later).

Of accident and emergency attenders in Scotland, only 1% of children had impaired consciousness on arrival at hospital, while 6% had recovered from brief impairment of consciousness. The incidence of brain damage in adults was more than three times that for children, with 5% impaired on arrival and 18% recovered from impaired consciousness. In children brain damage was much more common after an RTA than a fall, but no such difference was found in adults (table 2).

The distribution of causes of head injury in children varies greatly according to severity, with falls predominant for accident and emergency attenders and admissions, and RTA the major cause only for neurosurgical unit transfers, severe injuries, and deaths (table3).3 9 For those attending accident and emergency departments, assaults and RTA are more common in adults than in children (table 4). Moreover the distribution of victims of RTA with head injuries are different for children, with fewer car occupants and more pedestrians and cyclists. Among fatal RTAs concerning children, pedestrians are even more common, 69% in the Newcastle series.10 Most bicycle injuries in children are sustained during play off the road and do not involve collision with another vehicle. In some reports these are classified as recreational rather than road accidents. However, most serious injuries are from the minority of bicycle accidents that do involve collisions with other vehicles.

The high incidence of unsafe behaviour by children as a cause of both pedestrian injuries and those caused by bicycle collisions was noted by the Newcastle team.10 In that study most child occupants in cars who were fatally injured were not wearing seat belts, and few bicycle fatalities had been wearing helmets. The value of bicycle helmets for children has been studied in Australia11 where use is now mandatory in several states. It has been suggested that built-in safety features for restraints in cars should become the rule.12 However, most of the recent reduction in child death rates from road accidents in England and Wales appears to be because there are fewer child pedestrians and cyclists as more children travel by car.13

Assaults form a small minority of childhood injuries in the UK and are often sustained in play. In the United States, firearms are a frequent cause of injury, mostly the result of children playing with their parents’ guns. Even the Newcastle series had four fatal shotgun injuries, all from accidents while playing with parents’ guns.10 In that study fewer than 5% of fatal injuries were non-accidental.

A matter of some concern and controversy is the need to admit children with mild injuries to hospital for observation. If unnecessary, this adds unjustifiable stress to the injured child. The main reason for keeping patients under observation is the fear of complications, particularly acute intracranial haematoma. The Newcastle study found one third of avoidable deaths after head injury were because of missed or late diagnosis of this complication.5 In 1984 guidelines for the admission of adults with head injuries emphasised the importance of skull fracture and impaired consciousness as risk factors for intracranial haematoma,14 and suggested that these criteria should reduce the number of mildly injured patients unnecessarily admitted. More recently, additional data have been published, including those for children, based on the epidemiological studies in Scotland.15 These showed that the risk of intracranial haematoma in children is about one sixth of that in adults. However, the risk factors are exactly the same in children, skull fracture being a more potent factor than impaired consciousness. These data indicate the low risk of haematoma in children without skull fracture, only 1 in 500 develop a haematoma even if there is impaired consciousness. Yet in a series of 900 patients in Glasgow operated because of intracranial haematomas there were more children than adults who had been fully conscious, some of them without a skull fracture. Others have questioned the value of skull x ray in assessing head injuries in children.16

In Scotland between 1974 and 1985 the admission rate for patients with head injuries for all ages fell from 23% to 15%, mostly because fewer children were admitted—10% compared with 20% of adults in 1985.2 Of those with impaired consciousness on arrival at the accident and emergency department, 25% of children and 16% of adults were sent home—most often because no skull fracture was found. A skull x ray was done in 58% of all children, but in 88% of those with brain damage. Reduction in unnecessary admissions, particularly of children for whom this can add psychological trauma, is to be welcomed. The decision to send a child home will depend on full consciousness having been regained and there being no reason to suspect a skull fracture, but may also be influenced by home circumstances. Moreover, it needs to be associated with adequate counselling of parents about the need for rapid return to hospital in the event of new symptoms, supported by a card listing warning symptoms and giving the telephone numbers for contacting the hospital.