Complications in humeral shaft fractures - non-union, iatrogenic radial nerve palsy, and postoperative infection: a systematic review and meta-analysis.

The aim of this systematic review and meta-analysis was to assess risk for iatrogenic radial nerve palsy (iRNP), non-union, and post-operative infection in humeral shaft fractures. A PubMed search including original articles comparing different treatments for humeral shaft fractures published since January 2000 was performed. Random effect models with relative risks (RR) and 95% CIs were calculated for treatment groups and outcomes. Of the 841 results, 43 studies were included in the meta-analysis (11 level II, 5 level III, 27 level IV). Twenty-seven compared intramedullary nailing (IM) with ORIF, nine conservative with operative treatment, four ORIF with minimally invasive plate osteosynthesis (MIPO), and three anterior/anterolateral with posterior approach. iRNP risk was higher for ORIF vs IM (18 studies; RR: 1.80; P = 0.047), ORIF vs MIPO (4 studies; RR: 5.60; P = 0.011), and posterior vs anterior/anterolateral approach (3 studies; RR: 2.68; P = 0.005). Non-union risk was lower for operative vs conservative therapy (six studies; RR: 0.37; P < 0.001), but not significantly different between ORIF and IM (21 studies; RR: 1.00; P = 0.997), or approaches (two studies; RR: 0.36; P = 0.369). Post-operative infection risk was higher for ORIF vs IM (14 studies; RR: 1.84; P = 0.004) but not different between approaches (2 studies; RR: 0.95; P = 0.960). Surgery appears to be the method of choice when aiming to secure bony union, albeit risk for iRNP has to be considered, particularly in case of ORIF vs IM or MIPO, and posterior approach. Due to the limited number of randomised studies, evidence on the best treatment option remains moderate, though.

Introduction

The incidence of humeral shaft fractures is estimated at 13 per 10 000 patients per year (1), amounting to 1.0–3.0% of all fractures (2, 3, 4). Humeral shaft fractures may be treated by conservative or operative modalities, with specific constellations as polytrauma, vascular injury, instability, or open fracture demanding a surgical approach (5, 6). The frequency of surgical therapy has increased over the years (7), including plate osteosynthesis via a conventional open reduction and internal fixation (ORIF) or minimally invasive plate osteosynthesis (MIPO), and retrograde and antegrade intramedullary nailing (IM) (6, 8, 9). Complications of both conservative and operative treatment approaches are non-union, while post-operative infections and iatrogenic radial nerve palsy (iRNP) do naturally occur following surgical therapy.

As complication profiles are known to differ between conservative and surgical therapy, operative techniques, and implant types, but this evidence is largely based on individual comparative studies, the herein performed systematic review and meta-analysis aim at analysing the outcome for treatment of humeral shaft fractures, focusing on risk for non-union, iRNP, and post-operative infection.

Methods

Search strategy, selection criteria

The current systematic review adhered to the PRISMA guidelines (10). A literature search in PubMed was performed for the current systematic review and meta-analysis, using the following search term: ((humerus) AND (shaft)) AND (fracture). All studies published from 1 January 2020 until 21 December 12 were potentially eligible (Supplementary Table 1 for full search code, see section on supplementary materials given at the end of this article). Notably, studies identified through other sources were not limited to a specific publication date.

Inclusion criteria

Any English or German original cohort study (either randomised controlled trial (RCT), prospective non-randomised study, or retrospective cohort study) dealing with humeral shaft fractures, comparing two treatment approaches, and assessing at least one of the three outcomes namely iRNP, non-union, or infection were potentially eligible.

Exclusion criteria

All other publications, including (systematic) reviews, meta-analyses, Cochrane reviews, case reports, expert opinions, editorials, study protocols, non-comparative studies, those with insufficient information, pathological fractures, and non- English or -German articles were excluded. Data extraction was performed by one of the co-authors (MAS). In case of discrepant findings, decision towards article inclusion or exclusion was reached after thorough consideration and discussion amongst co-authors (SB, MAS).

Analysis of data

Items assessed from the individual studies included: outcome measure (i.e. iRNP, non-union, infection), comparison group (e.g. conservative vs operative, MIPO vs ORIF, IM vs ORIF), patient number, and those experiencing and non-experiencing the outcome parameter. Furthermore, level of evidence as defined by the Oxford Centre for Evidence-Based Medicine (OCEBM)was ascertained (11).

The Newcastle–Ottawa scale (NOS) for non-randomised case–control and cohort studies was used to assess quality of all publications included (12). Notably, also the quality of those eight studies reporting on randomised cohorts was assessed with the NOS to allow comparability to publications with non-randomised cohorts. The NOS itself is based on the three items selection (maximum 4 points), comparability (maximum 2 points), and outcome (for cohort studies; maximum 3 points) or exposure (for case-control studies; maximum 3 points). A total score ranging from 0 to 9 results from adding the scores of individual items, with higher values indicating higher study quality (12).

Statistical analysis

All statistical analyses were performed with Stata/SE 16.1 for Mac (StataCorp LLC, College Station, Texas, United States). For each outcome parameter and comparison group, random effect models using restricted maximum likelihood methods were calculated and relative risks (RR) with corresponding 95% CI provided. RCTs, prospective case–control studies, and retrospective case–control studies were pooled, thus increasing the sample size, as previously described (13, 14, 15). Heterogeneity between studies pooled in each model was assessed with Higgins I2 (16). Small, medium, and large heterogeneity was defined as I2 amounting to ≤25%, >25–<50%, and >75%, respectively (16). Harbord test was used to assess presence of publication bias (17). In case heterogeneity was medium to large, and/or publication bias was present, additional sensitivity analyses to assess the robustness of findings were performed. As recommended by the Cochrane Handbook for Systematic Reviews of Interventions, studies with zero events in both arms were excluded from the meta-analysis itself (18).

Results

Literature search retrieved 841 results. Additional 57 studies were identified through cross-reference checking of literature in previous original articles and (systematic) reviews on humeral shaft fractures, resulting in 898 articles. After removal of 5 duplicates, 893 entries were screened, of whom 641 were subsequently excluded. The remaining 252 full-text articles were assessed for eligibility. After excluding further 193 full-text articles not meeting the inclusion criteria, 59 studies remained. Further 16 studies were excluded as comparing similar approaches or implant types (n  = 10), dealing with outcome following persisting non-union (n  = 4), involving pathological humeral fracture (n  = 1), or analysing outcome between primary and delayed surgery (n  = 1). Thus, 43 studies were included in the final quantitative meta-analysis (Fig. 1).

Figure 1

Flow chart showing study selection.

Study characteristics

Altogether, the 43 studies investigated comprised 7688 patients with humeral shaft fractures. Eleven studies were randomised controlled trials (level II; 25.6%) (19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29), 5 were prospective non-randomised case-control studies (III; 11.6%) (30, 31, 32, 33, 34), and 27 were retrospective cohort studies (IV; 62.8%) (6, 8, 9, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58). Twenty-seven studies compared IM with ORIF (6, 19, 20, 22, 24, 25, 26, 27, 28, 29, 30, 31, 33, 36, 39, 40, 41, 42, 47, 48, 52, 53, 54, 55, 56, 57, 58) (including two studies with external fixator (39) or MIPO (42) as third comparator that was not considered upon meta-analysis), nine studies compared conservative with operative treatment attempt (23, 34, 35, 38, 43, 45, 49, 50, 51) (including one study on paediatric humeral shaft fractures (43)), four studies compared MIPO with ORIF (21, 32, 37, 46), and three studies compared anterior or anterolateral with posterior approach to the humeral shaft (8, 9, 44). Mean NOS score over all 43 studies included was 6.0 ± 0.8 points (Table 1).

Table 1

Description of studies included together with Newcastle–Ottawa scale (NOS).

Study	Groups	Patients, n	Outcome factors	LOE	Newcastle–Ottawa scale
					Selection	Comparability‡	Outcome	Total
Conservative vs operative treatment
Westrick et al. (35)	ConservativeOperative	69227	NU	IV	****		**	6
Harkin et al. (38)	ConservativeOperative	9630	NU	IV	****		*	5
Matsunaga et al. (23)	ConservativeOperative	4650	NU, infection	II	***	*	**	6
O’Shaughnessy et al. (43)	ConservativeOperative	6515	Iatrogenic RNP, NU	IV	***		**	5
Jawa et al. (45)	ConservativeOperative	2119	NU	IV	****		*	5
Denard et al. (49)	ConservativeOperative	63150	NU	IV	***		**	5
Broadbent et al. (34)	ConservativeOperative	8921	NU	III	****		**	6
Dielwart et al. (50)	ConservativeOperative	3140	NU	IV	****		*	5
Ekholm et al. (51)	ConservativeOperative	207	NU	IV	****		**	6
Intramedullary nailing vs ORIF
Schwab et al. (36)	IMORIF	6289	Iatrogenic RNP	IV	***		**	5
Bisaccia et al. (39)	IMORIF(Fixateur Externe†)	2632(21†)	NU, infection	IV	****	*	**	7
Changulani et al. (19)	IMORIF	2324	Iatrogenic RNP, NU, infection	II	***		**	5
Li et al. (20)	IMORIF	3233	Iatrogenic RNP, NU, infection	II	***	*	***	7
Denies et al. (40)	IMORIF	4942	Iatrogenic RNP, NU, infection	IV	****		**	6
Putnam et al. (41)	IMORIF	5911418	Infection	IV	***		**	5
Singisetti & Ambedkar (30)	IMORIF	2016	Iatrogenic RNP, NU, Infection	III	****		**	6
McCormack et al. (22)	IMORIF	2123	Iatrogenic RNP, NU, infection	II	***		**	5
Kulkarni et al. (42)	IMORIF(MIPO†)	4434(34†)	NU, infection	IV	****		*	5
Kumar et al. (31)	IMORIF	1515	Iatrogenic RNP, NU, infection	III	****		*	5
Wali et al. (24)	IMORIF	2525	Iatrogenic RNP, infection	II	***	*	**	6
Akalın et al. (25)	IMORIF	3033	Iatrogenic RNP, NU, infection	II	****	*	***	8
Fan et al. (26)	IMORIF	3030	Iatrogenic RNP, NU, infection	II	****		***	7
Khan et al. (47)	IMORIF	3030	Iatrogenic RNP, NU, infection	IV	***		**	5
Raghavendra & Bhalodia (33)	IMORIF	1818	Iatrogenic RNP, NU, infection	III	***		***	6
Hashmi et al. (48)	IMORIF	2239	Iatrogenic RNP, NU, infection	IV	***		***	6
Lian et al. (27)	IMORIF	2324	NU	II	****		***	7
Chao et al. (52)	IMORIF	5636	Iatrogenic RNP, NU	IV	***		***	6
Chapman et al. (28)	IMORIF	3846	NU, infection	II	****		***	7
Goncalves et al. (53)	IMORIF	541	NU	IV	****		**	6
Gottschalk et al. (6)	IMORIF	8702560	NU, infection	IV	****		**	6
Kulkarni et al. (54)	IMORIF	3125	Iatrogenic RNP, NU	IV	***		**	5
Putti et al. (54)	IMORIF	1618	Iatrogenic RNP, NU, infection	IV	***	*	***	7
Radulescu et al. (29)	IMORIF	10282	Iatrogenic RNP, NU	II	***	*	***	7
Wang et al. (56)	IMORIF	2630	NU	IV	****		***	7
Zhang et al. (57)	IMORIF	3446	Iatrogenic RNP, NU	IV	***	*	***	7
Benegas et al. (58)	IMORIF	1114	NU	IV	****	*	***	8
MIPO vs ORIF
Lee and Yoon (37)	MIPOORIF	2428	Iatrogenic RNP	IV	****	*	**	7
Kim et al. (21)	MIPOORIF	3632	Iatrogenic RNP (infection*)	II	****		***	7
Esmailiejah et al. (32)	MIPOORIF	3233	Iatrogenic RNP, NU (infection*)	III	****	*	**	7
An et al. (46)	MIPOORIF	1716	Iatrogenic RNP (infection*)	IV	****		*	5
Surgical Approaches
Lotzien et al. (9)	A/AAPA	3325	Iatrogenic RNP, NU, infection	IV	***		**	5
Streufert et al. (44)	A/AAPA(PS†)	9794(68†)	Iatrogenic RNP, infection	IV	****	*	**	7
Li et al. (8)	A/AAPA	5057	Iatrogenic RNP, NU, infection	IV	***		***	6

†Not included in meta-analysis; ‡patients with primary nerve injury excluded; *Asterisks depict allocated points per item assessed.

A/AA, anterior/anterolateral approach; IM, intramedullary nailing; LOE, level of evidence; MIPO, minimally invasive plate osteosynthesis; NU, non-union; ORIF, open reduction internal fixation; PA, posterior approach; PS, posterior split; RNP, radial nerve palsy.

Subsequently, the three outcomes, iRNP, non-union, and post-operative infection, will be outlined for individual comparator groups.

Iatrogenic radial nerve palsy risk

Of 34 studies involving patients treated surgically by different methods, 25 reported outcomes on post-operative radial nerve palsy (8, 9, 19, 20, 21, 22, 24, 25, 26, 29, 30, 31, 32, 33, 36, 37, 40, 44, 46, 47, 48, 52, 54, 55, 57). Eighteen of these studies compared IM with ORIF (19, 20, 22, 24, 25, 26, 29, 30, 31, 33, 36, 40, 47, 48, 52, 54, 55, 57), four MIPO with ORIF (21, 32, 37, 46), and another three anterior/anterolateral with posterior surgical approaches to the humeral shaft (8, 9, 44).

IM vs ORIF

Of the 18 studies reporting on radial nerve palsy risk following IM vs ORIF, 7 were level II (19, 20, 22, 24, 25, 26, 29), 3 level III (30, 31, 33), and 8 level IV (36, 40, 47, 48, 52, 55, 55, 57).

Whilst the subgroup analyses for retrospective (RR: 1.59; 95% CI: 0.76–3.33; P  = 0.089) and prospective studies (RR: 2.19; 95%CI: 0.86–5.57; P  = 0.100) revealed a non-significantly higher iRNP risk for ORIF in comparison to IM, the overall meta-analysis showed a significantly increased iRNP risk for ORIF (RR: 1.80; 95%CI: 1.01–3.21; P  = 0.047; Fig. 2). Between-study heterogeneity was small (I2 = 0.0%), and no significant publication bias was present (Harbord’s P  = 0.927).

Figure 2

Forest plot of studies analysing post-operative radial nerve palsy risk depending on intramedullary nailing vs ORIF. Amber diamond depicts effect size for subgroups prospective (top) and retrospective (bottom) studies. Red diamond shows overall effect size, solid red line overall effect size value, and dashed black line the no-effect line.

MIPO vs ORIF

According to the meta-analysis of four studies on MIPO vs ORIF (21, 32, 37, 46) (one level II (21), one level III (32), two level IV (37, 46)), risk for iRNP was significantly higher for ORIF in comparison to MIPO (RR for ORIF: 5.60; 95% CI: 1.48–21.18; P  = 0.011; Fig. 3A). Heterogeneity between studies was small (I2 = 0.0%). No significant publication bias was found (Harbord’s P  = 0.839).

Figure 3

Forest plot of studies analysing post-operative radial nerve palsy risk depending on MIPO vs ORIF (A) and anterior/anterolateral vs posterior approach (B). Red diamond shows overall effect size, solid red line overall effect size value, and dashed black line the no-effect line.

Anterior/anterolateral vs. posterior approach

All three studies comparing anterior or anterolateral with posterior approach to the humeral shaft and reporting on post-operative radial nerve palsy risk were level IV (8, 9, 44). Following a posterior approach, risk of developing iRNP was significantly higher in comparison to an anterior or anterolateral approach (RR: 2.68; 95% CI: 1.34–5.36; P  = 0.005; Fig. 3B). Heterogeneity was small (I2 = 0.0%), and no publication bias was found (Harbord’s P  = 0.903).

Non-union risk

Thirty-five studies reported on non-union risk following humeral shaft fractures, with 24 studies comparing IMs with ORIF (6, 19, 20, 22, 25, 26, 27, 28, 29, 30, 31, 33, 39, 40, 42, 47, 48, 52, 53, 54, 55, 56, 57, 58), 9 studies conservative with operative treatment (23, 34, 35, 38, 43, 45, 49, 50, 51), and 2 anterior/anterolateral with posterior approach (8, 9). Of these 35 articles, 9 reported on level II studies (19, 20, 22, 23, 25, 26, 27, 28, 29), 4 on level III studies (30, 31, 33, 34), and 22 on level IV studies (6, 8, 9, 23, 35, 38, 39, 40, 42, 43, 45, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58).

Conservative vs operative treatment

Nine studies (one level II (23), one level III (34), seven level IV (35, 38, 43, 45, 49, 50, 51)) reported on non-union rate following conservative vs operative treatment. Three (43, 45, 51) (one including paediatric patients (43)) reported on no events in either arm, wherefore they were excluded from the final meta-analysis.

According to the meta-analysis of remaining six studies, risk for non-union was significantly lower after operative treatment (RR: 0.37; 95% CI: 0.25–0.56; P  < 0.001; Fig. 4A). Heterogeneity was small (I2 = 0.0%). No significant publication bias was present (Harbord’s P  = 0.908).

Figure 4

Forest plot of studies analysing non-union risk after humeral shaft fractures depending on conservative vs operative approach (A) and anterior/anterolateral vs posterior approach (B). Red diamond shows overall effect size, solid red line overall effect size value, and dashed black line the no-effect line.

Anterior/anterolateral vs posterior approach

Meta-analysis of the two studies (both level IV (8, 9)) comparing surgical approaches to the humeral shaft revealed no significantly altered risk to develop non-union (RR for posterior approach: 0.36; 95% CI: 0.04–3.36; P  = 0.369; Fig. 4B), with small heterogeneity (I2 = 0.0%). Publication bias could not be assessed.

Twenty-four studies (8 level II (19, 20, 22, 25, 26, 27, 28, 29), 3 level III (30, 31, 33), 13 level IV (6, 39, 40, 42, 47, 48, 52, 53, 54, 55, 56, 57, 58)) reported on non-union risk for IM vs ORIF. In three (47, 48, 58) of these studies, no event was reported in either arm, wherefore these were excluded from the meta-analysis.

Meta-analysis of the remaining 21 studies revealed no significant difference in non-union risk between IM and ORIF (RR for ORIF: 1.00; 95% CI: 0.71–1.41; P  = 0.997; Fig. 5). Heterogeneity was small (I2 = 0.0%), and no significant publication bias was encountered (Harbord’s P  = 0.882). Subgroup analysis for retrospective (RR: 1.08; 95% CI: 0.70–1.66; P  = 0.719) and prospective studies (RR: 0.86; 95% CI: 0.48–1.55; P  = 0.621) revealed likewise no significantly differing non-union risk for IM vs ORIF.

Figure 5

Forest plot of studies analysing non-union risk after humeral shaft fractures depending intramedullary nailing vs ORIF. Amber diamond depicts effect size for subgroups prospective (top) and retrospective (bottom) studies. Red diamond shows overall effect size, solid red line overall effect size value, and dashed black line the no-effect line.

Post-operative infection risk

Altogether, 24 of 34 studies reporting on surgically treated patients with humeral shaft fractures provided information on post-operative infections. These included 18 studies comparing IM with ORIF (6, 19, 20, 22, 24, 25, 26, 28, 30, 31, 33, 39, 40, 41, 42, 47, 48, 55), 3 comparing anterior/anterolateral with posterior approach (8, 9, 44), and 3 comparing MIPO with ORIF (21, 32, 46). However, the latter group was not analysed as two of the three articles reported on zero-event outcomes (21, 46). Of the remaining 21 studies, 7 were level II (19, 20, 22, 24, 25, 26, 28), 3 level III (30, 31, 33), and 11 level IV (6, 8, 9, 39, 40, 41, 42, 44, 47, 48, 55).

Of the 18 studies reporting on post-operative infection risk following IM vs ORIF, 7 were level II (19, 20, 22, 24, 25, 26, 28), 3 level III (30, 31, 33), and 8 level IV studies (6, 39, 40, 41, 42, 47, 48, 55). Four of these (20, 26, 33, 39) reported on zero events in both arms, wherefore they were excluded from the meta-analysis.

The final meta-analysis with 14 studies revealed a significantly higher post-operative infection risk for ORIF in comparison to IM (RR for ORIF: 1.84; 95% CI: 1.22–2.79; P  = 0.004; Fig. 6A). Heterogeneity was small (I2 = 0.0%), and no significant publication bias was encountered (Harbord’s P  = 0.884). Notably, subgroup analysis of retrospective (P  = 0.021) studies revealed a significantly higher infection risk for ORIF, while the effect was only marginally significant for prospective studies (P  = 0.059).

Figure 6

Forest plot of studies evaluating post-operative infection risk in surgically treated humeral shaft fractures depending on intramedullary nailing vs ORIF (A) and anterior/anterolateral vs posterior approach (B). Amber diamond depicts effect size for subgroups prospective (top) and retrospective (bottom) studies on IM vs ORIF (A). Red diamond shows overall effect size, solid red line overall effect size value, and dashed black line the no-effect line (A and B).

Three studies reported on infection risk depending on surgical approach chosen (all level IV (8, 9, 44)). One study (9) reported on zero events in both arms and was thus excluded.

According to the meta-analysis including the remaining two studies, RR for post-operative infection following a posterior approach was 0.95 (95% CI: 0.14–6.66; P  = 0.960; Fig. 6B) in comparison to anterior/anterolateral approach. Heterogeneity was small (I2 = 0.0%). Publication bias could not be assessed.

Discussion

According to this systematic review and meta-analysis involving 7688 patients with humeral shaft fractures, there appears to be an elevated risk for iRNP following ORIF in comparison to IM, ORIF in comparison to MIPO, as well as after posterior approach vs anterior or anterolateral approach. Also, ORIF is associated with higher post-operative infection risk as compared with IM. While non-union risk is comparable between ORIF and IM, and between anterior/anterolateral and posterior approach, conservative treatment is associated with a significantly higher non-union risk when compared with any surgical therapy for humeral shaft fractures.

Similar to a large analysis of the American Board of Orthopaedic Surgery (ABOS) database by Gottschalk et al., we observed an overall significantly higher risk for iRNP after ORIF in comparison to IM (6). Furthermore, our results are in line with those made in previous meta-analyses involving 18 (59), 10 (60), 10 (5) and 3 studies (61), respectively.

Additionally, for MIPO vs ORIF, a five-fold increased risk for intraoperative radial nerve injury was found, corroborating the results published by Zhao et al. in a network meta-analysis involving two RCTs (62). Of note, the estimated risk – as reflected by the odds ratio – reported by Zhao et al. was relatively higher than the herein observed risk – represented by the RR – which may be due to the different methodological approaches chosen, as well as a potential over-estimation of the OR due to an event rate exceeding 10% (63, 64). However, as in our meta-analysis, 8 of 18 and 2 of 4 studies comparing IM with ORIF as well as MIPO with ORIF, respectively, were non-randomised in their nature, it cannot be ruled out that patients undergoing open reduction had injuries contraindicating any closed reduction (e.g. extensive soft tissue involvement, multi-fragmented fracture (65, 66)) that would by themselves have increased the likelihood of the radial nerve being in danger. Interestingly, we also observed a significantly higher risk of iRNP in case a posterior rather than anterior or anterolateral surgical approach to the humeral shaft had been chosen, which could as well be caused by a selection bias as surgeons may have attempted to explore an already endangered radial nerve via a direct posterior exposure of the humeral shaft.

Bearing these complications in mind, conservative treatment for humeral shaft fractures is a historically well-established alternative (66, 67). Nevertheless, the rate of operative treatment has steadily increased over the years, with today roughly one-third of patients – based on data from 2011 – still treated by conservative measures (7). Albeit a conservative approach naturally lacks surgery-associated risks, a longer immobilisation period, eventually reduced quality of life, and prolonged return to work can be seen disadvantageous (15).

According to the current meta-analysis, a conservative treatment attempt seems to be associated with a significantly higher risk for non-union. This is in line with a previous meta-analysis by van de Wall et al. (15) who, apart from nine observational studies, also pooled results from two RCTs. Of note, effect estimates of observational studies and RCTs in that study were comparable, wherefore the authors considered the potential for confounding due to the retrospective non-randomised studies rather minor (15). Furthermore, taking into account that surgery is often indicated in case of open injuries, multi-fragmentary fractures, and considerable displacement – all of which may increase the risk for non-union – the overall higher risk for non-union in case a conservative treatment is chosen appears even more striking.

Regarding different surgical exposures, the risk for non-union was not significantly different between anterior/anterolateral and posterior approaches to the humeral shaft. Furthermore, we observed a non-significantly different non-union risk when comparing IM with ORIF, being in line with previous meta-analyses (5, 59, 60, 61). This contradicts common knowledge that open reduction leads to significant exposure of the fracture site, which could potentially – together with required periosteal stripping to allow anatomical reduction – limit blood supply and thus impair fracture healing. For the same reasons, ORIF may be associated with an increased post-operative infection risk in comparison to IM, an assumption supported by the results of the current meta-analysis. In fact, a nearly two-fold increased risk for post-operative infection was observed for ORIF in comparison to IM. Our observations are in line with a recent met-analysis by Beeres et al. (59), albeit contradicting findings in previous ones (5, 60, 61). Notably, the type of surgical exposure was non-significantly associated with altered risk for post-operative infection in our meta-analysis, although one should bear in mind that the two studies included had been retrospective in nature.

The current systematic review and meta-analysis have some limitations that have to be considered. First, the majority of articles included described retrospective studies, wherefore non-random assignment to one or the other treatment arm has to be assumed. Secondly, several studies had to be excluded as not providing sufficient information on the outcome variables of interest. Thirdly, different surgical exposures chosen for ORIF (when compared with MIPO or IM) were not considered in the analyses as the main goal was to identify differences between surgical techniques exposing and not exposing the fracture itself. Fourthly, we did not perform adjusted meta-analyses for underlying differences between groups due to small sample sizes. Fifthly, time to union and restitution from iRNP were not analysed, as both outcomes depend on frequency of examinations scheduled that may be more prone to bias than steady outcomes as infection, post-operative radial nerve palsy, and non-union.

Conclusions

According to the current meta-analysis involving 43 studies and 7688 patients with humeral shaft fractures reporting on outcomes non-union, post-operative radial nerve palsy, and infection risk, the method of choice when aiming at secure bony union appears to be surgery, albeit risk for radial nerve palsy has to be considered, especially in case a posterior rather than anterior/anterolateral approach, and ORIF rather than MIPO or IM is chosen. Also, higher infection risk following ORIF in comparison to IM should be considered upon treatment planning. Yet, evidence on best treatment option for humeral shaft fractures remains moderate, taking into account the limited number of randomised studies available, as well as the multiple factors contributing to decision making.

ICMJE Conflict of Interest Statement

A L has received Institutional Educational grants from Johnson & Johnson, Alphamed, Implantec, Amgen and Medacta not related to this publication. F J S declares receiving travel grants from AOTrauma and ÖGU. All other authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of this study.

Funding Statement

This work did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.