Characteristics of included RCTs
Eventually, a total of 10 RCTs involving 2780 patients, comprising of 1394 patients managed with NPWT and 1386 patients managed with traditional gauze dressings, were included in this meta-analysis [3, 7, 9, 12, 17, 21–25]. The baseline data of included RCTs was presented in Table 1. Each included study was prospective randomized controlled trial, and the publication years ranged from 2009 to 2021. Most studies were completed in Asia (seven studies), two in Europe and another one in America. The follow-up period ranged from 1 to 36 months and the mean age ranged from 31.9 to 49.8 years. Overall, there was no significant difference in patient number, mean age, BMI, follow-up period and sex proportions between the two groups. The clinical characters of included RCTs were summarized in Table 2. All patients were Type II or III fracture according to the Gustilo and Anderson classification. Similarly, there were no significant difference between the NPWT and control groups in terms of fracture type, diagnosis of diabetics, smoking and method of wound closure. Figure. 2 showed the quality evaluation of included RCTs. As a whole, the risk of bias of included studies was minor, but three studies were rated to high risk in blinding of participants and personnel [3, 12, 17]. Most studies were judged as unclear risk because the authors did not state the random sequence generation, allocation concealment, blinding of outcome assessment and incomplete outcome data clearly.
Table 1
Baseline data of included studies
Included Studies | Year | Design | Country | Patients | Fracture | Follow up (Month) | Mean Age | BMI | Sex (Female/Male) |
NPWT | ST | NPWT | ST | NPWT | ST | NPWT | ST | NPWT | ST |
Stannard, JP [7] | 2009 | RCT | USA | 35 | 23 | 37 | 25 | 28 | NA | NA | NA | NA | 9/26 | 10/13 |
Jayakumar M [22] | 2013 | RCT | India | 20 | 20 | 20 | 20 | 36 | 32.0 | NA | NA | NA | NA |
Sinha K [24] | 2013 | RCT | India | 15 | 15 | 15 | 15 | 12 | 39 | NA | NA | NA | NA |
Ketan G [23] | 2013 | RCT | India | 15 | 15 | 15 | 15 | 1.5 | 39 | NA | NA | 4/11 | 3/12 |
Ghulam R [21] | 2013 | RCT | Pakistan | 25 | 25 | 25 | 25 | 1.5 | NA | NA | NA | NA | 8/17 | 7/18 |
Arti H [9] | 2016 | RCT | Iran | 45 | 45 | 45 | 45 | 1 | 31.9 | NA | NA | 22/68 |
Virani SR [25] | 2016 | RCT | India | 43 | 50 | 43 | 50 | 5.75 | 34.8 | 37.4 | NA | NA | 15/28 | 18/32 |
Costa, ML [12] | 2018 | RCT | England | 220 | 230 | 220 | 230 | 12 | 46.1 | 44.5 | 26.72 | 27.3 | 48/178 | 70/164 |
Tahir M [3] | 2020 | RCT | Pakistan | 206 | 214 | 206 | 214 | 12 | 37 | 34 | NA | NA | 65/141 | 80/134 |
Costa ML [17] | 2020 | RCT | England | 770 | 749 | 770 | 749 | 6 | 49.8 | 26.4 | 26.7 | 320/482 | 281/482 |
NPWT: Negative pressure wound therapy, ST: Standard treatment, BMI: Body Mass Index, RCT: Randomized controlled trial, NA: Not available data |
Table 2
Clinical characteristic of included studies
Included Studies | GA classification | Diabetics (n/%) | Smoking (n/%) | Method of wound closure |
Type II | Type III | NPWT | ST | NPWT | ST | DPC | Flap (R/F) | Skin graft | Amputation/closure |
NPWT | ST | NPWT | ST | NPWT | ST | NPWT | ST | NPWT | ST | NPWT | ST |
Stannard JP [7] | 2 | 2 | 33 | 23 | 1/2.8 | 0/0 | 18/51 | 7/30 | 21 | 18 | 7 | 3 | 7 | 4 | 2 | 0 |
Jayakumar M [22] | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA |
Sinha K [24] | 4 | 26 | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA |
Ketan G [23] | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA |
Ghulam R [21] | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA |
Arti H [9] | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA |
Virani SR [25] | 5 | 8 | 38 | 42 | 3/6.9 | 5/10 | 11/25.5 | 14/28 | 34 | 40 | 1 | 2 | 8 | 8 | 0 | 0 |
Costa ML [12] | 34 | 30 | 192 | 204 | 14/6.2 | 13/5.6 | 70/31.0 | 79/33.8 | NA | NA | NA | NA | NA | NA | NA | NA |
Tahir M [3] | 17 | 21 | 189 | 193 | 25/12.1 | 27/12.6 | 77/37.4 | 81/37.9 | 169 | 172 | 15 | 22 | 19 | 23 | 0 | 0 |
Costa ML [17] | NA | NA | NA | NA | 63/8.1 | 85/11.3 | 218/28.6 | 216/29.2 | NA | NA | NA | NA | NA | NA | NA | NA |
NPWT: Negative pressure wound therapy, ST: Standard treatment, GA classification: Gustilo and Anderson classification, DPC: Delayed primary closure, R/F: Rotational/Free, NA: Not available data, |
Meta-analysis
Anti-infection function evaluation of the two dressings
For accurate comparison of infection rate between the two groups, we conducted data analysis in terms of overall infection rate, deep infection rate and acute wound infection rate. Totally, 8 RCTs [3, 7, 9, 12, 17, 22, 23, 25] involving 2414 patients reported the data of overall infection rate. There was no significant heterogeneity (P = 0.06, I2 = 48%) detected, and a fixed-effect model was applied for data analysis. The pooled results showed that NPWT reduced the overall infection rate significantly (MD = 0.70, 95% CI: 0.54 − 0.90, P = 0.005) (Figure. 3a). However, with respect to rate of deep infection, data was available in 6 RCTs comprising 2344 patients [3, 7, 9, 12, 17, 25]. As illustrated in Figure. 3b, a fixed-effect model was applied for analysis due to the absence of significant heterogeneity (P = 0.45, I2 = 0%), and no significant difference was detected between the two groups (MD = 0.76, 95% CI: 0.58 − 1.00, P = 0.05). Regarding the rate of acute wound infection, 5 RCTs containing 315 patients were included for meta-analysis [7, 9, 22, 23, 25]. Data analysis was performed with a fixed-effect model because the heterogeneity was not significant (P = 0.29, I2 = 20%), and the pooled results suggested that the rate of acute wound infection in the NPWT group was significantly reduced (MD = 0.35, 95% CI: 0.16 − 0.77, P = 0.009) (Figure. 3c). The positive rate for culturing was provided in 3 studies (185 patients) [7, 24, 25]. As shown in Figure. 3d, without significant heterogeneity (P = 0.67, I2 = 0%), we conducted the data analysis using a fixed-effect model. The merged result showed that NPWT was effective in reducing the positive rate for culturing (MD = 0.19, 95% CI: 0.08 − 0.45, P = 0.0001) compared with traditional gauze dressings. Likewise, we collected data about the rate of clinical infection from positive culturing in 2 RCTs [7, 25]. After merging with a fixed-effect model, no significant difference was identified between the two groups (MD = 1.21, 95% CI: 0.29 − 5.06, P = 0.80) (Figure. 3e).
Efficacy evaluation of promoting wound healing
Time for wound ready for closure
Regarding the time for wound ready for closure, data from 152 patients in 2 RCTs was available for data analysis [7, 9]. As presented in Figure. 4a, because no significant heterogeneity was found (P = 0.42, I2 = 0%), a fixed-effect model was utilized to finish the meta-analysis, demonstrating that NPWT significantly shortening the time for wound ready for closure (MD =-1.19, 95% CI: -2.03 − 0.35, P = 0.006).
Time for the wound coverage
Regarding the Time for the wound coverage, data was provided in 2 RCTs involving 70 patients [22, 23]. Because of the absence of significant heterogeneity (P = 0.92, I2 = 0%), we used a fixed-effect model to perform the data synthesis and detected that more wound coverage was completed within 3 weeks in the NPWT group (MD = 24.00, 95% CI: 6.82–84.46, P < 0.00001), as depicted in Figure. 4b. However, more wound coverage was completed in the ST group beyond 3 weeks (MD = 0.04, 95% CI: 0.01–0.15, P < 0.00001) (Figure. 4c). In summary, NPWT shortened the time for the wound coverage.
Time for wound complete healing
4 RCTs with 475 patients described the information of time for wound complete healing [12, 21–23]. Data analysis was conducted utilizing a random-effect model on account of significant heterogeneity (P < 0.0001, I2 = 91%). As presented in Figure. 4d, even though more patients in the NPWT group reported that the wound completely healed within 6 weeks, the difference between the two groups was not statically significant (MD = 7.48, 95% CI: 0.59–94.69, P = 0.12). Likewise, we found no significant difference between NPWT and ST with respect to wound complete healing beyond 6 weeks (MD = 0.13, 95% CI: 0.01–1.69, P = 0.12) (Figure. 4e). To sum up, NPWT made no significant effect on the time for wound complete healing.
Efficacy evaluation of function restoration
DRI
DRI information was available in 2 RCTs with over 1200 patients at different time point [12, 17]. We performed subgroup analysis at postoperative 3 and 6 months. Similarly, the data of DRI was not heterogeneous at postoperative 3 months (P = 0.48, I2 = 0%) and 6 months (P = 0.32, I2 = 0%), a fixed-effect model was applied to merge results. And the pooled results suggested that NPWT could improve the DRI at postoperative 3 months (MD = 0.49, 95% CI: 0.23 − 0.76, P = 0.0003) and 6 months (MD = 0.41, 95% CI: 0.10–0.72, P = 0.01). Overall, there was a significant difference between the two groups regarding to DRI (MD = 0.46, 95% CI: 0.25–0.66, P < 0.0001) (Figure. 5a).
EQ-5D
EQ-5D was reported in 2 RCTs (1296 patients)[12, 17]. As shown in Figure. 5b, because no significant heterogeneity between the two groups was detected (P = 0.77, I2 = 0%), data analysis was conducted using a fixed-effect model, and the difference between NPWT and ST was not significant (MD =-0.00, 95% CI: -0.00–0.00, P = 0.99).
Length of hospital stay
In terms of length of hospital stay, 2 RCTs were included for data analysis [22, 23]. For length of hospital stay < 1 month, as shown in Figure. 5c, the heterogeneity was not significant (P = 0.92, I2 = 0%) and a fixed-effect model was used for data analysis, showing that more patients in the NPWT group discharged within 1 month postoperatively (OR = 24.00, 95% CI: 6.82–84.46, P < 0.00001). On the contrary, more patients in the ST group discharged beyond 1 month postoperatively (OR = 0.04, 95% CI: 0.01–0.15, P < 0.00001) (Figure. 5d). In short, NPWT shortened the length of hospital stay significantly.
Complications
Due to the data availability, we selected 2 complications containing amputation and bone nonunion for meta-analysis, which were reported in 2 RCTs at least. We merged each complication with a fixed-effect model because the heterogeneity was not significant. The pooled results were summarized in Figure. 6. The difference between NPWT and ST regarding each complication was not statistically significant.