Status and risk factors of unintentional injuries among Chinese undergraduates: a cross-sectional study

A cross-sectional study was conducted in three main colleges in Wenzhou, China. We used a multi-stage probability sampling method to obtain a representative sample. We used school and grade as stratum, and then selected classes by cluster sampling in each stratum. In total, 2,350 students (aged 17 to 23) were selected from 82 classes and were all surveyed, 2,287 of whom made valid replies, yielding a response rate of 97.3%. The major of students included Medicine, Education, Mathematics, Business, and Management Science.

A college-based cross-sectional study was conducted from January to March, 2009. We used self-administered questionnaires (see additional file 1: A transcript of the questionnaire) to survey these randomly selected undergraduates. The questionnaire solicited socio-demographic information and experience of injuries during the preceding 12 months. For each reported event, the student was asked to specify the location where it happened, activities at the time of injuries, the cause, and the consequences of the event.

Additionally, the scale of type A behaviour pattern (TABP) revised by the Chinese National Collaborative Study Group for TABP & coronary heart disease [31] in China was used for all participants, to explore the relationship between type A behaviour pattern and injury. This scale is a revised version of some foreign scales such as the scale of Jenkins Activity Survey (JAS), and has been commonly used in China since the 1980s. It has a good reliability (Cronbach's α coefficient of reliability is 0.78 in this study), and there are 60 items in the scale of TABP, including three dimensions: TH (time hurry), CH (competitive hostility), and L (lie). If the score for the L dimension is higher than 7, the questionnaire is invalid and should be deleted in the final data analysis. The higher the total score on the scale, the more competitive, the more hostile, and the closer to the type A behaviour pattern; otherwise, the lower the total score, the closer to type B behaviour pattern. According to the mean(= 27) and standard deviation(= 8) of the total score of the scale, we divided the students into five groups, similar to previous researchers: A(≥ 36), mA (28~35), M (27), mB (19~26), and B(≤ 18) [31], that is, if the total score on the scale is larger than the mean + standard deviation (SD), then the student will be grouped into type A, if the total score on the scale is less than mean-SD, then the student is grouped into type B, which was a little different from the standard in some other countries [26].

The questionnaire was pre-tested with 60 students from two classes prior to the formal survey. Eight investigators were college students majoring in preventive medicine who received formal training before the survey. During the survey, participants were not allowed to discuss the survey material and they were requested to check for any missed questions before handing in the questionnaire.

In order to avoid selection and drop-out bias, we surveyed the students during the period before the final examination so that almost all students were present. If students were absent at the time of the survey, the class monitor was instructed to give the questionnaires to them upon their return to college, and the investigator would subsequently retrieve the completed survey.

This research study was done in compliance with the Helsinki Declaration, and was reviewed and approved by the Wenzhou Medical College, School of Environmental Science and Public Health. Principals of selected schools signed written consent forms. Students' verbal consent was obtained before the study. Assurance was given that all questionnaire information about the students would be confidential and only used for research.

A reportable injury in this study was defined as an injury meeting at least one of the following criteria [16]: (1) an injury for which the student received medical treatment at the school nurse's office, or received medical care from a doctor at a hospital or a private medical office, (2) an injury for which the student received first aid from his/her schoolmates, teachers, or parents, or (3) an injury that was not treated but caused the student to miss a half day or more of school or regular activities.

For analyses of injuries by cause, we used the tenth Revision of the International Classification of Diseases (ICD-10) and grouped the codes into categories, such as traffic injury, fall, cut, burn, animal/insect bite, poisoning, drowning, choking, and others. The focus of this study was unintentional non-fatal injuries; therefore, child abuse, suicide, and homicide cases were excluded. Unintentional injuries were defined as any injuries that were not deliberately caused by the student or another person [32].

Analyses were performed with the use of SAS software (version 9.1, SAS Institute) or SPSS statistics package (version 14.0). Considering that there are repeated injuries for some students, besides prevalence, injury incidence was calculated and compared by student gender, age group, family characteristics, including total number of children in the family, marital status, parents' education, and family income. All statistical analyses were conducted by 2-tailed tests with a level of significance set at 0.05.

Logistic regression models were used to control for potential confounding factors, and crude odds ratios and adjusted odds ratios (OR) and 95% confidence intervals (CIs) for OR were calculated. In all regression models, the outcome variable was whether the student had an injury in the past year (if yes, then y = 1; otherwise, y = 0). And in this study, the independent variables included individual level determinants (gender, age, major, behaviour pattern, sports), and family socioeconomic determinants (family annual per-capita income, the number of siblings), we used different independent variables in different models for controlling specific factors in every model.

Among the 2,287 respondents, a total of 428 injuries requiring medical attention were reported for 320 students during the year preceding the interview, leading to an incidence of 18.71 injuries per 100 person-years. The corresponding 95% confidence interval (95%CI) for this incidence was 17.12~20.31 injuries per 100 person-years. However, the percentage of students injured during the year was 13.99% (320/2287), with 95% confidence interval (95%CI) at 12.57%~15.41%. Among this group of 320 students, 23.44% had multiple episodes of injury during the preceding 12 months.

Table 1 presents the incidence of injuries by gender and cause of injuries. As recorded, falls were the leading cause of injury, followed by traffic injuries, bites, and burns. There are some differences between male and female students; the three leading causes for male students are falls, traffic injuries, and bites respectively; but for female students the three leading causes are falls, traffic injuries, and burns, respectively. The incidence for falls of male students was 15.41 injuries per 100 person-years, which was significantly higher than that of female students (8.78 injuries per 100 person-years); and for burns, female students have significantly higher incidence than males (table 1). In addition, for both males and females, sports were the most common activities associated with injury.

Then, we analyzed the proportions of reported medically attended non-fatal injuries by cause for both male and female students. As Figure 1 shows, the proportions of female injuries by cause were very similar to that of male injuries. For both male and female students, falls were the most common cause of injury, as depicted in Figure 1.

We analyzed the time distribution of injuries among college students, and found that there seemed to be two peaks for non-fatal injuries among these students. As Figure 2 shows, one peak was in spring, the other was in autumn.

We found that 66.9% of the 428 injured cases occurred on campus, 12.9% of them on the roads, 10.7% at home, and 9.5% at other locations. All participants in this study came from three different colleges in Wenzhou city. Our survey found that the incidence of injury for students from medical college was 17.5 injuries per 100 person-years (209 injuries/1192 person-years), the incidence for students from vocational college was 15.0 injuries per 100 person-years (32/213), whereas that for students from a comprehensive university was 21.2 injuries per 100 person-years (187/882). Meanwhile, the prevalence for these three colleges was 12.6%, 16.7%, 10.8% respectively, and there was a significant difference among them (χ² = 9.01, P = 0.011).

In addition, among the 1,140 students from rural areas, 207 injuries occurred, resulting in an incidence of 18.2 injuries per 100 person-years; whereas students from urban areas had an incidence of 19.3 injuries per 100 person-years, but the difference of the percentage injured between these two groups(13.2% vs 14.7%) was not significant (χ² = 1.05, P = 0.305).

We analyzed other characteristics of injuries among undergraduates, and found that male students, students majoring in non-medicine, students from a one-child family, students from low-income families, students with type A behaviour, and students liking sports and activities tend to experience the highest prevalence of injuries.

As indicated in table 2, the incidence for male students was 22.1 injuries per 100 person-years (95% CI: 19.5, 24.7), while that for female students was 16.3 injuries per 100 person-years (95% CI: 14.3, 18.3); the difference of percentage of students injured between male and female students was statistically significant (χ² = 4.08, P = 0.043). Medical students had an incidence of 15.9 per 100 person-years (95% CI: 13.7, 18.1), and that for non-medical students was 21.3 (95% CI: 19.0, 23.6), the difference of percentage of students injured between these two groups was also statistically significant. Likewise, students with no sibling had higher injury rates than those with ≥ 1 sibling. The injuries for students from families with an annual per-capita income less then 2,000 Yuan was the highest with an incidence of 28.3 per 100 person-years (95% CI: 22.6, 34.0), however, the difference of percentage of students injured between these five different groups was not significant. The incidence for students with type A behaviour pattern was 52.3 injuries per 100 person-years (95% CI: 37.5, 67.0), which was higher than that of any other type of behaviour pattern, and the percentages of students injured among the difference types of behaviour patterns were also significantly different. Students who liked sports or took part in club activities in college experienced significantly higher rates than those who disliked sports or didn't attend club activities.

Based on the above univariate and bivariate analyses, and in order to control for other confounding factors, we conducted multivariate logistic regression analyses, and found some factors that may be associated with non-fatal injuries among undergraduates. As shown in table 3, non-medical students experienced significantly higher rates than medical students, the odds ratios (ORs) changed very little even after adjusting for gender or other variables, which indicate students' major may be an influencing factor for non-fatal injuries. Consistent with the results from bivariate analyses, students who like sports are more likely to be injured than those that do not like sports (adjusted OR = 1.86, 95%CI: 1.41~2.45).

After adjusting for gender or other variables, Type A behaviour pattern (TABP) still remained statistically significant in the model. Compared with the students with type B behaviour pattern, students with type A behaviour pattern had a higher risk of injury (adjusted OR = 2.99, 95%CI: 1.45~6.14).

From family level, we found that students who had no siblings were more likely to be injured than other students; however, this difference became non-significant after adjusting for gender or other variables (Table 3).

Of all injured students, 3.16% were hospitalized, 56.45% were treated as outpatients, 21.41% were treated by themselves or other persons, and 18.98% were not treated at all. Most (74.94%) were minor injuries, and no disability was found. As for the anatomical location of the injury, 44.1% were wounded at the lower extremity, 19.4% the upper extremity, 12.2% the head and face, and 6.6% were injured at multiple parts.

On average, injuries led to 7.28 days of rest(however, 56.0% of the cases had no rest, 44.0% had a mean of 15.8 days of rest, median = 7 days, the first quartile was 2 days, the third quartile was 16 days), 0.55 days of hospitalization (96.8% of them had no hospitalization, only 3.2% of the cases were hospitalized and had a mean of 15.3 days of hospitalization, median = 7 days, the first quartile was 1.5 days, the third quartile was 21 days), and 0.53 work days off for family members (87.1% of the cases had no work days off for family members, 12.9% had a mean of 6.9 days, median = 2 days, the first quartile was 1 day, the third quartile was 10 days). The modes of the days of rest and hospitalization were both 7 days. Traffic injuries, and falls caused the longest period of both hospitalization and days of rest.

The distribution of the direct costs caused by injuries from treatments was right skewed, these 320 students resulted in the sum of all treatment fees at 115 224.90 yuan (mean = 360.08 yuan); however, 31.5% of the injuries resulted in no treatment fees, and the 5% trimmed mean of direct costs from injuries was 119.52 yuan, with the median of 20.00 yuan, and an interquartile range of 187.50 yuan. The highest costs that were caused by traffic injury amounted to 12,000 yuan.