Hand fracture epidemiology and etiology in children—time trends in Malmö, Sweden, during six decades

The hand is one of the most commonly injured anatomic regions, accounting for 29% of pediatric visits in emergency departments [1, 2]. Around one fourth of those patients have a fracture [3, 4], with the hand being the second most fractured region, surpassed only by the distal forearm [5‐9]. However, there are gender differences in hand fracture occurrence, as well as differences in age incidences, with boys having higher incidences than girls and girls having an earlier peak incidence than boys [6, 8]. As time trends in fracture epidemiology have changed in boys and in girls during recent decades [3, 6, 7, 9‐18], there is a need for updated pediatric fracture epidemiology. This also accounts for etiology data, in order to identify new fracture prone activities in need of prevention and evaluate the effect of already existing fracture prevention strategies.

The aim of this study is to describe hand fracture epidemiology/etiology in children in 2005–2006 and with use of previously published data from the same area [6, 9], evaluate time trends in age- and gender-standardized fracture incidences from 1950/1955 to 2005–2006.

Our city is located in the southern part of the country with a population of 276,244 inhabitants (46,429 < 16 years of age) in year 2006 [19]. One hospital provides trauma care for the entire city and this hospital archives store all medical charts, referrals, radiographs, and reports since a century [20], making it possible to re-evaluate former patients. Radiographs were until the year 2001, sorted according to anatomical region, year of injury, and diagnosis. Researchers have utilized this archive to describe pediatric fracture epidemiology in city residents < 16 years during the years 1950, 1955, 1960, 1965, 1970, 1975–1979 [6], and 1993–1994 [9]. These studies have reported changes in crude hand fracture incidences [6] and crude overall fracture incidences [9], however, without adjusting for changes in demography within the population at risk.

In 2001, the radiographic archive was replaced by a digital archive. This system includes radiographs performed within the healthcare system in the entire southern part of the country, including the Malmo city hospital. The radiographs are now classified according to each specific patient (through a unique 10-digit personal identity number), diagnosis, and anatomical location. Pediatric fracture cases 2005–2006 were therefore identified through this classification in the digital in- and outpatient diagnosis records at the Emergency Department, Departments of Orthopedics, Hand Surgery, and Otorhinolaryngology. We included records that fulfilled the following criteria: (i) ICD-10 fracture diagnosis: S02.3–S02.4, S02.6–S02.9, S12.0–S12.2, S12.7, S22.0, S32.0–S32.8, S42.0–S42.9, S52.0–S52.9, S62.0–S62.8, S72.0–S72.9, S82.0–S82.9, and S92.0–S92.9; (ii) age < 16 years at the time of the incident fracture; and (iii) city resident at the time of the injury.

We identified 4459 visits with a fracture diagnosis during 2005–2006 and 1548 with a hand fracture diagnosis (ICD-10 code S62.0-S62.8). To verify hand fractures, we reviewed medical charts, referrals, radiographic reports, and radiographs of all fracture visits. This enabled us to exclude all follow-up visits after an index fracture visit, thus avoiding double counting of a fracture. Patients who received emergency treatment in other hospitals are in our country, as standard, referred to the home hospital for follow-up evaluation. Such fractures were then captured in the hospital archives.

We collected pediatric fracture data from 2005 to 2006, following the same protocol as in the previous studies [6, 9]. Thus, we included data on patient age and gender, number of fractures, date of the fracture, fractured region/regions, and fracture etiology. During the years 2005–2006, we could determine the etiology in 70% of the cases compared to 40% during the years 1950/1955.

Multiple phalangeal fractures, multiple metacarpal fractures, and combinations of carpal and metacarpal fractures were registered as separate fractures. This was done to provide detailed information about the anatomical distribution of hand fractures and, by this, facilitate comparisons with other single center reports that register multiple metacarpal and phalangeal fractures separately [4, 12, 13, 17, 18, 21‐26]. To compare the current data to the previously collected data from the same city, we recorded fracture events according to the previous classification protocol [6, 9], that is with three different groups: (i) fractures of the phalanges of the digits, (ii) fractures of the metacarpals or carpal bones, and (iii) fractures of the scaphoid. Multiple fractures in phalangeal, metacarpal, and carpal bones sustained at the same event were in this classification regarded as a single fracture, while multiple fractures on the same patient, bilateral fractures, and new fractures of an already fractured bone were classified as independent fractures [6, 9]. As in previous studies [6, 9], we did not include patients with traumatic amputations.

To validate the new fracture ascertainment system, one author (VL) performed a search in the digital radiological archive for all pediatric skeletal radiographs, independent on the reason for the referral, at the radiology department of the hospital from January 1, 2005, to February 28, 2005. The author reviewed all radiographs and then identified 103 pediatric fractures sustained in city residents. The same researcher then conducted a second search by the use of the same search criteria in the digital in- and outpatient diagnosis record archive. This second search also identified 103 fractures. One hundred fractures were identified by both methods; each method alone identified 103 fractures, while the two methods combined identified 106 fractures. Each method thereby missed three fractures, a miscalculation rate of 3%.

For statistical calculations, we organized hand fracture data in six periods (1950/1955, 1960/1965, 1970/1975, 1976–1979, 1993–1994, and 2005–2006). Data on the population at risk (i.e., city residents < 16 years) during each period was retrieved through official records [19]. Age- and gender-standardized rates were calculated through direct standardization, with the average pediatric city population (in 1-year classes) during the study period as reference. Results are presented as number of fractures, mean fracture incidences per 100,000 person years, and as proportions (%) of all fractures. To estimate differences in rates between evaluated periods, we calculated rate ratios (RR) by use of the chi-square distribution, including 95% confidence intervals (95% CI) to describe uncertainty. We considered p < 0.05 as a statistically significant difference.

One fourth of all pediatric fractures 2005–2006 affected the hand. Hand fractures (any) were 2.5 times and metacarpal/carpal fractures 8 times more common in boys than in girls. Hand fracture incidences increased with age in both genders until puberty. Girls reached peak incidence rates 2 years ahead of boys, that is, our data support the notion that hand fracture incidence rate is higher in boys than in girls and that there is a higher incidence with increasing age until a peak around puberty [4, 6, 9, 12, 13, 17, 23, 25, 27, 28].

The pediatric crude hand fracture incidence in 2005–2006 (448/100,000 person years) was in our city higher than 2005 in Helsinki Finland (344/100,000) [8], higher than 2006–2007 in the northern part of our country (389/100,000) [29], and 18 times higher than 1996–2001 in British Columbia (Canada) (24/100,000 person years) [24]. The discrepancies could be the result of different study methodology. For example, in our study, only one hospital serves the entire region, while several hospitals served the population at risk in the Canadian study. However, there could also be actual differences in fracture incidence rates, supported by the UK data, reporting that fracture incidences in county regions varied compared to London with a incidence rate ratio ranging from 1.0 to 1.7 [5]. Since ethnicity influences the fracture risk [15, 30, 31], a different mixture of ethnicity in different regions and countries could also explain regional fracture incidence differences.

The age-adjusted hand fracture incidence rate was in our city more than double 1976–1979 compared to 1950/1955 and 20% higher than 2005–2006 (not reaching statistical significance). This suggests that there could have been a trend break in pediatric fracture incidence at the end of the 1970s. Some of the time trends could be attributed to changes in demographics, since after adjustment for age, time trend changes were no longer statistically significant in boys. Finnish data however support a time trend change in pediatric fracture rates, reporting a higher pediatric hand fracture incidence in 2005 than in 1967 and around 50% lower incidence in 2005 than in 1983 [8]. However, in contrast to our study, the Finnish study only compared absolute incidences, but without conducting statistical comparisons.

Etiology data indicate that safety requirements for toy manufacturing, playground construction, protective sports gear, and accident prevention in traffic and home environment have positively influenced time trends in pediatric fracture incidence [32, 33]. Time trends in life style, such as more violent fights and less physical activity among children, may also influence time trends. Swedish children have reduced their everyday physical activity during recent decades, instead spending more time in front of monitors [34], and low levels of physical activity is a risk factor for fractures [35, 36]. The proportion of children with foreign background in our city was, in 2005–2006, also higher than during previous decades [37, 38], another difference that could have influenced time trends [15, 30, 31].

Study strengths include the epidemiology and etiology data from a well-defined cohort during six decades. Inclusion of only objectively verified fractures, without double counting due to multiple visits, is another strength. Weaknesses include the modification of the data collection method, due to the change in archiving method in the year 2001. However, our validation found a 3% miscalculation rate, the same as with the previous method [6, 9]. Another weakness is the risk of missing fractures exclusively treated outside the catchment area. People may be more mobile today than previously, sustaining more fractures that receive treatment at other hospitals. Since the standard treatment in Sweden still is to refer these cases to follow-up visits at the home hospitals, there is a minor risk to miss these fractures. Another confounder could be that individuals, decades ago, were less prone to seek medical advice and that doctors at that time were more hesitant to send patients to X-ray exams than today, thereby missing actual fractures. If this actually was the case is today impossible to clarify. A larger sample size would have been advantageous, especially for subgroup analyses making time trend analyses of rarer fracture types possible. Finally, the larger proportion of missing fracture etiology historically than at the last follow-up is another limitation and the reason to why we refrained from statistical analysis of time trend changes in etiology.

In conclusion, the incidence rate of hand fractures is higher in older than younger children, phalangeal fractures are more common than metacarpal fractures, and hand fractures are 2.5 times more common in boys than girls. The age-adjusted incidence rate of hand fractures in children was more than doubled in 1976–1979 compared to that in 1950/1955, with obvious differences also in fracture etiology. There may have been a trend break in 1976–1979, since we found a 20% lower fracture incidence in 2005–2006 than in 1976–1979, a time trend change not reaching statistical significance but inferring the need to continue to follow pediatric fracture incidences.