Negative pressure wound therapy for patients with mediastinitis: A meta-analysis.

To evaluate clinical effects between conditional treatment and negative pressure wound therapy for mediastinal infection. Multiple databases were searched to identify relevant studies, and the articles that eventually satisfied the criteria were included. All the meta-analyses were conducted with the Review Manager 5.2. To estimate the quality of each article, risk of bias table was performed. Finally, nine studies including 648 patients met the eligibility criteria. The negative pressure wound therapy (NPWT) group and the control group included 353 and 295 patients, respectively. The meta-analysis showed no significant difference in operative time (RR = -6.13, 95%CI [-50.00, 37.74], P = .78; P for heterogeneity <.000001, I2 = 100%). The length of hospital stay (MD = -3.07, 95%CI [-4.38, -1.77], P < .00001; P for heterogeneity = .99, I2 = 0%), re-infection (RR = 0.18, 100%CI [0.08, 0.40], P < .00001; P for heterogeneity = 0.48, I2 = 0%), and mortality were significantly different between the two groups (RR = 0.27, 95%CI [0.12, 0.63], P of overall effect = .002). NPWT is a better therapy than conventional treatment for mediastinitis.

INTRODUCTION

Mediastinal infection is a serious infection of involving the mediastinal connective tissue. The most common cause of acute mediastinal infection is the perforation of the oesophagus. , , Sometimes infections around the odontogenic or tonsil can cause an oropharyngeal abscess or severe infection of the anterior cervical space. Pus spreads into the mediastinum along the neck space and causes acute mediastinal infection. Because the neck and neck space is connected with the mediastinal neck membrane space, the infection can directly spread down the tracheal posterior space, the anterior tracheal tube, and the carotid sheath gap under the action of gravity, respiration, and intrathoracic negative pressure, resulting in infection in the thorax. , ,

The conventional drainage method is passive drainage, that is, a drainage strip or a drainage tube is placed at the incision drainage port so that the pus or inflammatory exudate naturally flows out of the body. , If the position of the abscess is deep, pain is obvious when changing the dressing, the drainage tube cannot be in full contact with the drainage lumen, and the fluid outflow is limited, resulting in pus accumulation.

Negative pressure wound therapy (NPWT) is active drainage, which uses negative pressure suction to continuously generate negative pressure below the atmospheric pressure in soft tissue so that harmful substances such as exudates and pus that are continuously produced as bacterial propagation medium are continuously removed. , NPWT reduces the biological damage caused by local harmful substances to the body. At the same time, the continuous negative pressure causes the local microcirculation flow rate and microvascular diameter to increase significantly.

There are several reviews of randomised trials comparing NPWT with control in mediastinitis patients. Previous reviews of NPWT have lacked a consistent comparator. In order to address these issues, we conducted a meta‐analysis of randomised trials/case‐control studies to determine the potential benefit of NPWT for mediastinitis patients and analysed multiple outcomes from several aspects.

METHODS

Literature search strategy

Published studies comparing NPWT versus conventional measurement (control) in patients with mediastinitis from January 2000 to May 2019 were retrieved. The searchable databases included PubMed, Springer, EMBASE, OVID, and China National Knowledge, and the following keywords were used: (1) NPWT; (2) control; (3) mediastinitis. All these words were assembled with the Boolean operators “and” in the strategy. No restriction in publication language was used in the literature retrieval. In order to maximise the search specificity and sensitivity, the authors also examined the reference lists of identified studies to seek additional relevant studies not identified through the retrieval strategy.

Study selection

Studies were included if:

They were randomised trials or case‐control studies.

They compared NPWT and control.

They involved patients with mediastinitis.

Studies were excluded if:

They were case studies/meta‐analyses/letter to editors.

NPWT was not compared versus control.

Patients did not have mediastinitis.

Data in research was limited or insufficient.

They were duplicates.

Data extraction and quality assessment

The full texts of the articles were read carefully, and the characteristics were extracted from each study using a predetermined form. The data extracted from these studies included the first author's name, year of publication, country, gender, and sample size (NPWT/Control).

Statistical analysis

The Review Manager 5.2 software (The Cochrane Collaboration, 2011) was used to estimate the overall outcomes between the NPWT and control groups in this study. The risk of bias table was used to assess the quality of the included studies following the Review Manager 5.2 Tutorial. The assessment included the following criteria: (1) random sequence generation; (2) allocation concealment; (3) blinding of participants and personnel; (4) blinding of outcome assessment; (5) incomplete outcome data; (6) selective reporting; (7) other biases.

Means with a standard mean difference (SMD) were calculated and compared for NPWT for patients with mediastinitis. For binary data, related risk (RR) with 95% confidence intervals (CIs) was calculated to estimate the clinical outcomes. Heterogeneity was assessed by the inconsistency index (I 2) statistic in this study, and the value of I 2 statistics also reflected the heterogeneity level. If I 2 was >50%, the trials were considered to be heterogeneous, and the random‐effects model was adopted. Otherwise, the fixed‐effects model was chosen. For all these comparisons, a P value <.05 was considered to be statistically significant.

In addition, bias analyses of the studies were conducted with STATA 10.0 software to examine the quality of articles. Funnel plots were used to estimate publication bias.

RESULTS

Search process

The electronic search ended with a total number of 813 articles. After a thorough reading, 87 papers met the preliminary criteria. In further screening, 81 articles were excluded because of the design of the study, insufficient data, and type of articles. Finally, six papers , , , , , were selected for analysis. Figure 1 shows the flowchart of identification, inclusion and exclusion, reflecting the search process, and the reasons for exclusion.

FIGURE 1

Flow diagram of the study selection

Characteristics of the included studies

Detailed characteristics of the included studies are presented in Table 1. All these studies were published from 2000 to 2019. The sample size ranged from 21 to 157. Totally 353 patients were in the NPWT group, and 295 patients were in the control group.

TABLE 1

Characteristics of studies included in the meta‐analysis

Study	Year	Language	Country	Male/Female	Groups	n	Recruitment time
De Feo 13	2011	English	Italy	120/37	NPWT	74	1995‐2010
					Control	83
Deniz 14	2012	English	Turkey	33/57	NPWT	47	2000‐2011
					Control	43
Petzina 15	2010	English	Germany	90/28	NPWT	69	2004‐2009
					Control	49
Rashed 16	2016	English	Hungary	14/7	NPWT	10	2013‐2015
					Control	11
Risnes 17	2012	English	Norway	115/15	NPWT	64	1997‐2010
					Control	66
Vos 18	2012	English	Netherlands	97/35	NPWT	89	2000‐2011
					Control	43

Results of quality assessment

The quality of the studies was assessed using the risk of bias table in the Review Manager 5.2 Tutorial, and Figures 2 and 3 show the evaluation of study quality in this study. As the results show, the limited risk was observed.

FIGURE 2

Assessment of the quality of the included studies: low risk of bias (green hexagons), unclear risk of bias (yellow hexagons), and high risk of bias (red hexagons)

FIGURE 3

Quality assessment of the included studies

Results of heterogeneity test

Figure 4 presents the forest plot for operative time (min) between the NPWT and control groups. Four studies were involved in this comparison. The results in the studies had no difference. The pooled results suggested that in the articles included, the operative time was comparable between the NPWT and control groups (RR = −6.13, 95%CI [−50.00, 37.74], P = .78; P for Heterogeneity <.000001, I 2 = 100%).

FIGURE 4

Forest plots of the length of operation (min) between NPWT and control groups

The forest plot for the length of hospital stay is shown in Figure 5. Since I 2 < 50%, the fixed effects model was adopted. In the analysis, the results suggested that the length of hospital stay in the NWPT group was shorter than the control group (MD = −3.07, 95%CI [−4.38, −1.77], P < .00001; P for heterogeneity = .99, I 2 = 0%).

FIGURE 5

Forest plots of the length of hospital days between NPWT and control groups

Three of the six included articles studied re‐infection between the NPWT and control groups of patients with mediastinitis, and the forest plot is presented in Figure 6. These three articles showed difference between two groups, the pooled results suggested that the re‐infection rate in the NPWT group was lower than that of the control group (RR = 0.18, 100%CI [0.08, 0.40], P < .00001; P for heterogeneity = .48, I 2 = 0%) (Figure 6).

FIGURE 6

Forest plots of re‐infection between NPWT and control groups

Three papers analysed mortality. The results of heterogeneity showed that the fixed effects model should be used to analyse the data since I 2 = 0%. In the results of the overall effect, there was a significant difference in mortality between the NPWT and control groups, and the mortality rate of the NPWT group was lower than that of the control group (RR = 0.27, 95%CI [0.12, 0.63], P of overall effect = .002) (Figure 7).

FIGURE 7

Forest plots of mortality between NPWT and control groups

Results of sensitivity analysis and publication bias

To examine the stability of the outcome, a sensitivity analysis was needed. A relative outlier from Vos in 2012 was excluded, and the result demonstrated that in the heterogeneity part, I 2 changed from 100% to 94% and overall P value changed from 0.78 to 0.001. It indicates that the heterogeneity was mainly due to the study by Vos in 2012. The forest plot without Vos's article is shown in Figure 8.

FIGURE 8

Sensitivity analysis forest plots of the length of operation (min) between NPWT and control groups

Funnel plots of death rate in the NPWT group and the control group were performed. All studies were included in the plot. The results showed that the funnel plot had limited symmetry and some publication bias (Figure 9). The result of Begg's test suggested that no significant evidence of potential publication bias was present (z = 1.43, P = .136). The result of Egger's test suggested that no significant evidence of potential publication bias was present (t = 1.23, P = .211).

FIGURE 9

Funnel plot of publication bias

DISCUSSION

The main clinical manifestations of mediastinal infection are fever, leukocytosis, chest pain, difficulty swallowing, and respiratory distress, accompanied by neck infections. Acute mediastinal infections have a higher death rate. Despite broad‐spectrum antibiotic treatment and CT imaging, the mortality rate of patients is as high as 25% to 50%. , Delayed diagnosis and inappropriate mediastinal drainage are the main causes of high mortality. Complications include adult respiratory distress syndrome, acute renal failure, and pneumonia. ,

The treatment of mediastinal inflammation is generally open drainage, but there are certain problems with this type of treatment, such as inadequate drainage of pus and long treatment time. The advantage of continuous negative pressure drainage is that it generates a sub‐atmospheric negative pressure in the soft tissue by continuous uninterrupted vacuum suction so that the continuously generated harmful substances such as exudate and pus as a bacterial propagation medium are continuously removed. ,

Four of the six studies have shown no difference in operative time in the NPWT group and the control group. This study result is consistent with previous reports. In the analysis of length of hospital stay, the NPWT group had a shorter length of hospital stay than the control group. Ingemansson reported that NWPT is significantly better than the control group using conventional open drainage surgery in terms of average wound healing time. This conclusion is consistent with ours.

In the analysis of re‐infection, the re‐infection rate of the control group was much higher than that of the NPWT group. Meanwhile, the mortality of the NPWT group was lower than that of the control group. Petzina reported that continuous negative pressure drainage is more conducive to control of infection, which can lead to the development of the disease. This statement is consistent with ours. In conclusion, NPWT is a more effective and safer therapy for mediastinitis than conventional therapy.

There are some limitations in this study. First, more indicators evaluating other aspects between NPWT and conventional treatment should be included, and this should be conducted in the future. Second, the comparisons in different subgroups could be analysed in future research.