A population-based register study was conducted. The study population included all pedestrians aged 16-64 years, living in Sweden 31 December 2009, who in 2010 received in- or specialized outpatient healthcare due to an injury sustained in a new traffic-related accident, including falls.
Data from five nationwide registers from the following three authorities were used and linked at the individual level, using the unique personal identity number assigned to all residents in Sweden [13]:
-
From Statistics Sweden, the Longitudinal Integration Database for Health Insurance and Labour Market Studies (LISA) was used to identify the source population of all 16-64 years old individuals living in Sweden 31 December 2009 (N=5 973 769) and information on sociodemographics (sex, age, educational level, country of birth, type of living area, and marital status).
-
From the National Board of Health and Welfare, the in- and specialized outpatient registers were used to identify the study population as well as for medical information related to the injury. The Cause of Death Register was used to identify those who had died within the first 30 days after the accident.
-
From the Swedish Social Insurance Agency, the register, Micro-data for Analyses of the Social Insurance (MiDAS) was used for information on dates and extent of SA and DP.
In the in- and specialized outpatient healthcare registers, the diagnoses (one main and all secondary diagnoses) and external causes of morbidity are both recorded according to the International Statistical Classification of Diseases and Related Health Problems; ICD-10 [14]. Individuals who received in- or specialized outpatient healthcare in 2010 due to an injury caused by a falls, a collision with another person, or a collision with a transport vehicle were identified by the ICD-10 codes for external causes of morbidity: “Pedestrian injured in transport accident” (V01-V09), “fall accidents, street and highway” (W00.4, W01.4, W02.4, W03.4, W04.4, W05.4, W10.4, W15.4, W17.4, W18.4, W19.4), and “striking against or bumped into by another person, street and highway” (W51.4).
These individuals (n=6299) will hereafter be referred to as the ‘pedestrians’ and the injuries as ‘traffic-related injuries’. The date of the accident, denoted as T0, refers the date of the in- or specialized outpatient healthcare visit/hospitalization as the actual date of their accident/fall is not included in the registers.
To include only new accidents, the pedestrians who had received in- or specialized outpatient healthcare for another traffic-related accident during the three years prior to their date of the accident (T0) were excluded (n=498), leaving 5801 pedestrians. Furthermore, those 225 who did not have an injury diagnosis as a main or secondary diagnosis (ICD10: S00-T88 “Injury, poisoning and certain other consequences of external causes”) were excluded, leaving a study population of 5576 pedestrians. Those who died in immediate connection to the crash, i.e., who did not receive in- or specialized outpatient healthcare, were not included. Whereas the six pedestrians who died within 30 days of the accident were included.
Based on type of accident, the pedestrians were categorized into the following six groups: collision with pedestrian/bicyclist (V01, W03.4, W04.4, W51.4); collision with motor vehicle (V02-V06, V09.0, V09.2); unspecified (V09.1, V09.3, V09.9, W19.4); fall - snow/ice, street and highway (W00.4); fall - slipping, tripping, and stumbling, street and highway (W01.4) (reference group); and fall - other, street and highway (W02.4, W05.4, W10.4, W15.4, W17.4, W18.4) (including: involving ice-skates, skis, roller-skates or skateboards, involving wheelchair, on and from stairs and steps, from cliff, from one level to another, and on same level). Type of healthcare was also categorized into three groups as: only specialized outpatient healthcare (reference group); inpatient healthcare <3 days; and inpatient healthcare ≥3 days (the median duration of the hospital stay among those hospitalized was 3 days).
According to the patient registers, some pedestrians had up to three different healthcare visits registered on T0. Every visit had a main diagnosis and could also have a number of additional secondary diagnoses. The majority had only one injury diagnosis. For categorization purposes, for those with several injury diagnoses, we selected one injury diagnosis per pedestrian, in the following way: The main injury diagnosis was selected over secondary injury diagnoses, the diagnoses for inpatient healthcare over outpatient healthcare, and S00-S99 over T00-T88.
A modified version of the Barell matrix [15] was used to classify the ICD-10 codes into categories of type of injury and injured body region. Similar categorizations were used in recent studies on injuries among car occupants and bicyclists [16,17,18,19]. Type of injury was categorized into the following six groups: fracture; dislocation; sprains and strains; internal (brain, spinal cord, and other internal organs); external (open wounds, contusions, and superficial injuries) (reference group); and “other and unspecified”.
The injured body region was categorized into ten groups: head, face, and neck (reference group); vertebral column and spinal cord; torso; shoulder and upper arm; forearm and elbow; wrist, hand, and other; hip, upper leg, and thigh; knee; lower leg, ankle, foot, and other; and “other and unspecified”.
The sociodemographic covariates were categorized as: sex (women; men (reference group)), age group (16-24; 25-34 (reference group); 35-44; 45-54; 55-64 years), level of education (elementary school (≤9 years including missing); high school (9-12 years); university/college (>12 years) (reference group)), country of birth (Sweden (reference group); not Sweden), type of living area (big cities (reference group); medium-sized cities; small cities/villages), marital status (married (reference group); not married). Reference groups were chosen based on size of the groups and expected proportions with new SA, with larger groups or groups expected to have lower proportions of new SA being used as the reference.
All individuals living in Sweden, ≥16 years old, and with income from work, unemployment, or parental-leave benefits can apply for SA benefits from the Social Insurance Agency if having a disease or injury that leads to reduced work capacity [20]. The first day of a SA spell is an unreimbursed qualifying day (more days for self-employed). A physician’s certificate is required after day 7. For employees, day 2-14 are reimbursed by the employer [20]. For others, e.g., unemployed, the Social Insurance Agency administrates the benefits from the second day of SA, thus information on shorter SA spells was available for these individuals. In order not to introduce a bias, only information on SA spells >14 days was used. All individuals aged 19-64 can be granted DP if disease or injury leads to long-term or permanent work incapacity. Both SA and DP can be granted for full- or part-time (100, 75, 50, 25%) of ordinary work hours. That is, someone on part-time DP can at the same time have part-time SA. For the calculation of mean and median net days of SA (for SA spells >14 days), gross SA days were summed to whole days (e.g., two days of 50% part-time SA was counted as 1 net day).
Pedestrians were categorized into four groups regarding SA/DP at T0 as follows: already ongoing full-time DP; already ongoing SA; new SA; and no new SA. To be defined as already having ongoing SA, the SA spell had to have started at least five days before T0 and still be ongoing at T0. When assessing SA, any SA spells >14 days, regardless of extent (full-time, part-time) were included. Considering DP, only full-time DP was categorized as already being on DP. As it is possible to be on part-time DP and receive a new SA, the group “no new SA” includes not only pedestrians without any SA or DP but also pedestrians with ongoing part-time DP but no new SA spell >14 days. Being on part-time DP at the date of the accident was instead included as a covariate in the analyses. The pedestrians might not have received in- or specialized outpatient healthcare on the actual date of the accident, e.g., they could have sought primary healthcare first, and they might not have applied for SA benefits the first day, due to e.g., holidays, thus a window of starting days for SA in relation to T0 was allowed. A new SA spell in relation to the accident was defined as an SA spell that had started between 4 days before and 4 days after T0. The choice of the timespan of ±4 days was based on distribution of start dates of SA in relation to the date of the in- or specialized outpatient healthcare visit/hospitalization (T0), with significantly more SA spells starting on T0, but also during the days immediately before or after.
Statistical analyses
The pedestrian’s sex, age, level of education, country of birth, type of living area, marital status, part-time DP, type of accident, type and duration of healthcare, type of injury, and injured body region were shown by SA and DP status at T0, using descriptive statistics.
Odds ratios (ORs) and 95% confidence intervals (CIs) for new SA were estimated by logistic regression. In these analyses, pedestrians with already ongoing SA or full-time DP were excluded (n=1022) as those pedestrians not were at risk of a new SA spell, leaving 4554 pedestrians. First, the ORs for new SA by the sociodemographic factors were calculated, in univariate models (crude), adjusted for sex (model 1), adjested for age (model 2), then, mutually adjusted (model 3), as well as adjusted by the accident type and injury characteristics (model 4). Then, the ORs for new SA for the characteristics of the accident were estimated, first adjusted for sociodemographic factors (model 1, 2, and 3) and then for type of accident, type of injury, and injured body region (model 4 and 5), and in model 6, 7 and 8 both the sociodemographic factors and the accident and injury characteristics were taken into consideration. These analyses were also stratified by sex. Moreover, sensitivity analyses excluding pedestrians on part-time DP were performed.
The statistical analyses were performed using SPSS (version 26) and STATA (version 14).