Patient characteristics
From 2003 to 2020, 130 children were admitted with BSI. Sixty-two patients (47.7%) were transferred to our hospital, while 68 (52.3%) arrived from scene. Blunt splenic injury most often occurred during spring (34.6%). The study population had a median age of 12 years [range 3–17], of whom the majority were male (63.1%). As presented in Table 1, falls from height were the most common cause of BSI (16.9%), followed by bicycle-vehicle collisions (16.2%) and individual bicycle injuries (15.4%). Those trauma mechanisms preceding BSI were categorised as high-energetic trauma (HET) or low-energetic trauma (LET) in 68.5% and 31.5% of cases, respectively. The LET group was mostly represented by individual bicycle accidents (26.8%). Additional patient demographics of the entire cohort are listed in Table 1.
Table 1
Characteristics of study population
Total study population (N = 130)
|
Age in years, median [range]
|
12 [3–17]
|
Age in years, N (%)
- < 6 years
- 6–10 years
- > 10 years
|
13 (10.0)
45 (34.6)
72 (55.4)
|
Sex, N (%)
- Male
- Female
|
82 (63.1)
48 (36.9)
|
Trauma mechanism, N (%)
- Accidents involving a motorized vehicle
■ Vehicle-vehicle
■ Bicycle-vehicle
■ Pedestrian-vehicle
■ Single vehicle
- Bicycle accidents
- Falls
■ Falls from height (> 1 m)
■ Falls on same level (< 1 m)
- Stump object in abdomen
■ Kick from a horse
■ Sports/bumping into obstacles
|
59 (45.4)
14 (10.8)
21 (16.2)
10 (7.7)
14 (10.8)
20 (15.4)
30 (23.1)
22 (16.9)
8 (6.2)
21 (16.2)
11 (8.5)
10 (7.7)
|
ISS, mean ± SD
|
21.6 ± 13.7
|
Splenic injury grade, N (%)
- Grade I
- Grade II
- Grade III
- Grade IV
- Grade V
- Undetermined
|
11 (8.5)
23 (17.7)
43 (33.1)
35 (26.9)
4 (3.1)
14 (10.8)
|
Presence of active blush on CT, N (%)
|
21 (16.2)
|
Isolated splenic injury, N (%)
|
47 (36.2)
|
Abbreviations:
CT Computerized Tomography, ISS Injury Severity Score, SD Standard Deviation
Injury characteristics
Nearly 64% of patients were multi-trauma patients with concomitant injuries. Multi-trauma patients needed to be transfused more often (30.1% vs 14.9% in the isolated injury group; p = 0.037), had a higher ISS (26.7 ± 14.3 vs 12.2 ± 4.9; p < 0.001) and were more often hemodynamically unstable (37.3% vs 19.1%; p = 0.031). Clinical injury data including the distribution of concomitant injuries are presented in Supplementary Table 1. The mean ISS of BSI patients was 21.6 ± 13.7. In 20.3% of cases, GCS was lower than 8, indicating a severely injured neurological state. Mean haemoglobin (Hb) concentration during admission was 7.2 ± 1.1 mmol/l. Of all patients, 30.8% were hemodynamically unstable.
They had a significantly higher ISS (32.2 ± 14.6 vs 17.5 ± 11.0 in the stable group; p < 0.001) and a lower serum Hb concentration on admission (6.2 ± 1.1 vs 7.5 ± 0.9; p < 0.001), leading to more blood transfusions (77.5% vs 1.1%; p < 0.001), and they more often showed an active blush on CT-scan (22.5% vs 10.0%; p = 0.006). Fifty-nine percent of patients had a low-grade splenic injury (grade I = III) and 30.5% of patients had a high-grade splenic injury (grade IV-V) based on their CT scan on admission. With a mean splenic injury grade of 3.0 (± 1.0). The distribution in splenic injury grade is presented in Table 1. An active blush was present in 16.2% of patients. Prior to a CT scan, an abdominal ultrasound was performed for 81.5% of patients to detect abdominal trauma or free fluid.
Management
The majority of patients with BSI were treated non-operatively (74.6%), including embolisation of the spleen due to hemodynamic instability and the presence of an arterial blush. Of the patients undergoing OM, 24.2% had a splenectomy, 18.2% had a spleen-preserving procedure, and 57.6% experienced an intervention on another organ besides the spleen. Splenectomy rate for the entire cohort was 6.2%. The main reasons for opting for OM were hemodynamic instability (71.4%), presence of arterial blush (7.1%) and persistence of abdominal pain (7.1%). Age and ISS were higher among the OM group than in the NOM group (12.9 ± 3.9 vs 10.8 ± 4.0; p = 0.009 and 33.9 ± 13.5 vs 17.5 ± 11.2; p < 0.001, respectively). Moreover, patients in the OM group more often needed transfusion (45.5% vs 17.5%; p < 0.001) and were hemodynamically unstable more frequently (66.7% vs 18.6%; p < 0.001), and all were multi-trauma patients due to a HET (p < 0.001). Multivariable logistic regression analysis shows that being hemodynamically unstable is an independent predictor for OM (odds ratio [OR] 0.151; [95% Confidence Interval (CI) 0.05–0.45]; p = 0.001).
The post-splenectomy protocol was followed by 75.0% of patients who underwent splenectomy and 40.0% of those who underwent embolisation. They all received pneumococcal vaccinations and in some cases additional influenza and meningococcal vaccinations. Eighty-nine percent of our cohort received pain medication, consisting most often of paracetamol (84.9%) or opioids (64.2%). Follow-up after BSI mostly consisted of clinical check-ups (25.7%) that were sometimes combined with secondary ultrasound of the spleen (13.3%). Follow-up time varied from one to eight weeks, with an average of three weeks. After discharge, doctors advised a return to activities within seven weeks in 77.0% of cases. This was extended to > 7 weeks for multi-trauma or operated patients.
Treatment outcomes
The main treatment outcomes of BSI are displayed in Table 2. One-hundred-and-three patients (79.2%) were admitted to the ICU during admission. After bedrest, initialisation of mobilisation/ambulation was delayed with a mean of 5.3 ± 3.5 days. The duration of stay in the ward, duration of stay in the ICU and LOS were significantly higher (p < 0.01) in patients that were either hemodynamically unstable, multi-trauma patients or managed operatively. The amount of bedrest was significantly higher (p < 0.03) in both unstable patients and when undergoing OM.
Table 2
Treatment outcomes for blunt splenic injuries
Treatment outcomes:
Mean ± SD [range]
|
Total
|
High-grade
|
Low-grade
|
Length of ward stay (in days)
|
6.8 ± 5.9 [0–24]
|
6.5 ± 4.8 [0–24]
|
6.6 ± 6.4 [0–24]
|
Length of ICU stay (in days)
|
3.5 ± 5.6 [0–30]
|
3.8 ± 5.5 [0–30]
|
3.5 ± 6.0 [0–29]
|
LOS (in days)
|
10.2 ± 9.0 [1–43]
|
10.1 ± 8.0 [1–39]
|
10.2 ± 9.7 [2–43]
|
Bedrest (in days)
|
5.0 ± 3.1 [0–16]
|
5.5 ± 2.6 [2–16]
|
4.8 ± 3.3 [0–14]
|
Splenic complications
|
No splenic complications
|
In-hospital mortality
(% of all patients)
|
3.1
|
2.6
|
1.3
|
Abbreviations: ICU Intensive Care Unit, LOS Length Of Stay, SD Standard Deviation |
According to the multivariate linear regression analysis, independent predictors for an extended LOS are OM (unstandardised regression coefficient [β] 6.98; [95% CI 2.98–10.98]; p = 0.001), unstable hemodynamics ([β] 4.97; [95% CI 1.46–8.48]; p = 0.006) and presence of concomitant injuries ([β] 4.80; [95% CI 8.59–1.01]; p = 0.014). These predictors explain a significant part of the variance in LOS (Adjusted R-squared = 0.311; F(6,109) = 9.664; p < 0.001). Patients had a longer LOS of 7.0 days when managed operatively, 5.0 days when they were hemodynamically unstable and 4.8 days when they were multi-trauma patients.
Although no complications related to splenic injury occurred, 20.0% endured complications from another origin. Most often these were neurological or psychiatric complications (34.6% and 50.0%, respectively) due to multi-trauma. Four patients (3.1%) died within 10 days of admission due to non-survivable brain damage. They were multi-trauma patients with an ISS exceeding 50.
A comparison of the clinical data and outcomes of management between low-grade and high-grade splenic injuries is depicted in Table 3.
Differences between OM and NOM in patients with a high-grade BSI
Of the patients with a high-grade splenic injury (n = 39), 79.5% underwent NOM. Five of these patients were treated with embolisation, due to hemodynamic instability or the presence of an active blush on their CT scan. Of those managed operatively (20.5%), three had a splenectomy (37.5%), one underwent a laparotomy (12.5%) and one had a spleen-preserving procedure (12.5%).
A comparison of the clinical data and outcomes of management for the NOM and OM group for patients with a high-grade splenic injury is depicted in Table 4. Patients who were treated operatively were significantly older (15.6 ± 1.8 vs 10.6 ± 3.7; p < 0.001), had a higher ISS (43.7 ± 7.1 vs 20.2 ± 7.9; p < 0.001) and had an extended LOS (16.1 ± 10.9 vs 8.5 ± 6.4; p = 0.013). Similarly, operatively treated patients were all multi-trauma patients due to a HET (p = 0.003 and p = 0.042, respectively). The multivariable logistic regression analysis shows that age is an independent predictor for OM (OR 0.855; [95% CI 0.75–0.97]; p = 0.015).
Table 3
Comparison of characteristics: Low-grade vs high-grade splenic injuries
|
Low-grade
(N = 77)
|
High-grade
(N = 39)
|
P-value
|
Age (in years)
|
11.0 ± 4.0
|
11.6 ± 4.0
|
0.359c
|
Male (%)
|
62.3
|
66.7
|
0.647a
|
Serum Hb (in mmol/l)
|
7.5 ± 1.0
|
6.6 ± 1.2
|
0.005c
|
Need of blood transfusion (%)
|
18.2
|
35.9
|
0.047a
|
ISS (in points)
|
19.4
|
24.4
|
0.022c
|
Trauma mechanism that occurred most (%)
■ LET (%)
■ HET (%)
|
Bicycle vs vehicle accident (19.5)
32.5
67.5
|
Bicycle accident (20.5)
30.8
69.2
|
0.960b
0.853a
0.853a
|
Presence of active blush (%)
|
9.1
|
30.8
|
0.002a
|
Hemodynamic stability
■ Stable (%)
■ Unstable (%)
|
76.6
23.4
|
64.1
35.9
|
0.154a
|
Isolated splenic injury (%)
|
32.5
|
48.7
|
0.088a
|
Management
■ NOM (%)
■ OM (%)
■ Embolization (%)
■ Splenectomy (%)
|
76.6
23.4
0.0
2.6
|
79.5
20.5
12.8
7.7
|
0.727a
0.727a
0.004b
0.333b
|
Stay in ward (in days)
|
6.6 ± 6.4
|
6.5 ± 4.8
|
0.134c
|
Stay in ICU (in days)
|
3.5 ± 6.0
|
3.8 ± 5.5
|
0.160c
|
LOS (in days)
|
10.2 ± 9.7
|
10.1 ± 8.0
|
0.058c
|
Bedrest (in days)
|
4.8 ± 3.3
|
5.5 ± 2.6
|
0.004c
|
Mortality (%)
|
1.3
|
2.6
|
0.561b
|
aChi-squared test; bFisher’s exact test; cMann Whitney U test. Abbreviations: Hb Haemoglobin, HET High-Energetic Trauma, ICU Intensive Care Unit, ISS Injury Severity Score, LET Low-Energetic Trauma, LOS Length Of Stay, NOM Non-Operative Management, OM Operative Management. Bold parameters are significant (as P-value < 0.05) |
Table 4
Comparison of characteristics: NOM vs OM in patients with high-grade splenic injury
|
Non-operative
(N = 31)
|
Operative
(N = 8)
|
P-value
|
Age (in years)
|
10.6 ± 3.7
|
15.6 ± 1.8
|
0.000b
|
Male (%)
|
64.5
|
75.0
|
0.694a
|
Serum Hb (in mmol/l)
|
6.7 ± 1.2
|
6.3 ± 1.6
|
0.223c
|
Need of blood transfusion (%)
|
32.3
|
50.0
|
0.424a
|
ISS (in points)
|
20.2 ± 7.9
|
43.7 ± 7.1
|
0.000b
|
Trauma mechanism that occurred most (%)
■ LET (%)
■ HET (%)
|
Bicycle accident (22.6)
38.7
61.3
|
Vehicle vs vehicle accident (37.5)
0.0
100.0
|
0.223a
0.042a
|
Presence of active blush (%)
|
29.6
|
50.0
|
0.402a
|
Hemodynamic stability
■ Stable (%)
■ Unstable (%)
|
71.0
29.0
|
37.5
62.5
|
0.109a
0.109a
|
Isolated splenic injury (%)
|
61.3
|
0.0
|
0.003a
|
Stay in ward (in days)
|
5.8 ± 3.8
|
8.9 ± 7.4
|
0.150b
|
Stay in ICU (in days)
|
2.9 ± 3.4
|
7.3 ± 10.0
|
0.184b
|
LOS (in days)
|
8.5 ± 6.4
|
16.1 ± 10.9
|
0.013b
|
Bedrest (in days)
|
5.4 ± 2.3
|
6.5 ± 4.4
|
0.932b
|
Mortality (%)
|
0.0
|
12.5
|
0.205a
|
a Fisher’s exact test; b Mann Whitney U test; c Independent Sample t-test. Abbreviations: Hb Haemoglobin, HET High-Energetic Trauma, ICU Intensive Care Unit, ISS Injury Severity Score, LET Low-Energetic Trauma, LOS Length Of Stay. Bold parameters are significant (as P-value < 0.05) |
Interobserver variability
The difference in injury grades, as determined by the radiologist in the acute trauma setting and reassessed by a paediatric radiologist, is presented in Table 5. Blunt splenic injuries tend to be graded higher in the acute trauma setting, with a mean injury grade of 3.0 ± 1.10, as compared to reassessment by a paediatric radiologist, with a mean injury grade of 2.7 ± 1.37. However, this difference was not significant (p = 0 .519). The Cohens Kappa coefficient was 0.493, indicating that interobserver agreement is moderate.
Furthermore, presence of an active injury on a CT scan in the acute trauma setting was detected significantly more often than during reassessment (16.2% vs 10.0%; p < 0.001).
Table 5
Interobserver variability: Initial vs rescored injury grade
Grade
|
|
Rescoredb
|
Total
|
0
|
I
|
II
|
III
|
IV
|
V
|
|
Initiala
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
0
|
I
|
2
|
6
|
2
|
0
|
0
|
0
|
10
|
II
|
0
|
1
|
15
|
3
|
0
|
0
|
19
|
III
|
1
|
0
|
7
|
26
|
7
|
0
|
41
|
IV
|
0
|
0
|
1
|
13
|
17
|
1
|
32
|
V
|
0
|
0
|
0
|
0
|
1
|
3
|
4
|
Total
|
|
3
|
7
|
25
|
42
|
25
|
4
|
106
|
aInitial grading (based on the AAST score) was performed in the acute trauma setting by the radiologists on call bProspective rescoring of injury grades was done by Karin Kamphuis-van Ulzen, a paediatric radiologist at RadboudUMC Nijmegen