Results

Institute for Quality and Efficiency in Health Care (IQWiG)

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute for Quality and Efficiency in Health Care (IQWiG). Negative pressure wound therapy for wounds healing by primary intention: IQWiG Reports – Commission No. N17-01B [Internet]. Cologne (Germany): Institute for Quality and Efficiency in Health Care (IQWiG); 2019 Sep 25.

Negative pressure wound therapy for wounds healing by primary intention: IQWiG Reports – Commission No. N17-01B [Internet].

Show details

Contents

< Prev Next >

4Results

4.1. Results of the comprehensive information retrieval

The information retrieval identified 121 randomized controlled trials (266 documents) as being relevant for N17-01A and/or N17-01B.

For N17-01A and N17-01B together, a total of 139 studies without reported results were found.

The search strategies for bibliographic databases and trial registries are found in the appendix. The most recent search was conducted on 8 February 2019.

Studies to be considered for N17-01B

The information retrieval identified 54 randomized controlled trials (135 documents) as being relevant for the research question of this benefit assessment on NPWT in patients with wounds healing by primary intention.

Six studies were included only formally because they fulfilled all inclusion criteria, but failed to supply usable data on any outcome. Three additional studies reported usable data only on outcomes presented for supplementary information. For transparency purposes, these 9 studies were included in the study pool since they met the documented inclusion criteria.

Hence, the study pool included a total of 45 studies which reported usable data on patient-relevant outcomes.

For NPWT in patients with wounds healing by primary intention, 14 planned and 50 ongoing studies were found. Furthermore, 10 studies of unclear status, 4 prematurely terminated studies, and 11 completed studies without reported results were found.

Table 1

Study pool of the benefit assessment – wounds healing by primary intention (n = 54) (multi-page table)

Table 2

Study pool of the benefit assessment – wounds healing by secondary intention (n = 67) (multi-page table)

4.2. Characteristics of the studies included in the evaluation

The 45 studies reporting usable results on patient-relevant outcomes for the benefit assessment provided data on a total of 7376 evaluation units (wounds) from 6981 patients. The individual studies included between 16 and 876 patients and were conducted worldwide in the years 2001 to 2018. All studies had a 2-arm design. Thirty-five studies were monocentric and 10 multicentric. All studies were conducted in inpatient settings, and 1 study also included the home care setting (HIC#1010007535). In 38 studies, randomization and analysis were performed at the patient level. In 5 studies, randomization and analysis were done at the wound level (7179, CE/US/11/01/PIC, INVIPS-Trial, Manoharan 2016, Tanaydin 2018). In the remaining 2 studies, patients were randomized, and patients or at least 1 wound per patient were analysed, depending on the specific outcome (Pleger 2017, VAC 2001-05).

The included studies comprised a wide range of different wounds of varying aetiologies, localization and risk factors for impaired wound healing, which were distributed as follows:

Abdominal surgery wounds (n = 10)
- Open abdominal surgeries (n = 3) (Li 2016, PICO Trial, Shen 2017)
- Panniculectomy (n = 2) (AHS.2012.Customizable.01, HIC# 1010007535)
- Open kidney transplantation (n = 1) (AHS.2012.Prevena.Cooper.01)
- Ileostomy closure (n = 1) (Uchino 2016)
- Colorectal surgery (n = 2) (Cantero 2014, NEPTUNE)
- Open duodenopancreatectomy (n = 1) (IRB00109564)
Caesarean section wounds (n = 7) (AHS.2011.Prevena.Heine) 03, H-20292, NPWTCS, PROVAC, S-20130010, The DRESSING Trial, VACCS)
Vascular surgery wounds (n = 8)
- Exposure of the femoral artery (n = 5) (Engelhardt 2018, IMS Study, INVIPS-Trial, Pleger 2017, RRG-104871)
- Procedure at the lower extremity (n = 3) (7179, KCI VAC Study, Yu 2017)
Wounds after hip/knee endoprosthesis (n = 8)
- Total hip endoprosthesis or knee endoprosthesis (n = 5) (1208434, CCF 14-273, Giannini 2018, Karlakki 2016, Keeney 2018)
- Total hip endoprosthesis (n = 2) (14-1920, Gillespie 2015)
- Knee endoprosthesis (n = 1) (Manoharan 2016)
Traumatic and plastic surgery wounds (n = 5) (Mendame Ehya 2017, Peter Suh 2016, Shim 2018, VAC 2001-04, VAC 2001-05)
Breast surgery wounds (n = 3)
- Breast reduction (n = 2) (CE/US/11/01/PIC, Tanaydin 2018)
- Breast reconstruction (n = 1) (DEPRES)
Cardiac/thoracic surgery wounds (n = 2)
- Sternotomy wounds due to heart surgery (n = 2) (KCI.2013.Prevena.01, Witt-Majchrzak 2014)
Other wounds (n = 2) (Pro00040054, VAC NPWT KCI Dressing Study)

Forty-two studies included predominantly or exclusively patients with risk factors for impaired wound healing; of these studies, 20 included people with at least 1 wound-specific or patient-specific risk factor for impaired wound healing. This included all 7 studies on wound therapy after Caesarean section with the inclusion criterion of obesity. In contrast, only 3 studies reported no risk factors in the study population (1208434, 14-1920 and Tanaydin 2018).

4.3. Studies without reported results / calculation of data gap

On the basis of the procedure described in Section 3 for determining potential publication bias for the assessment of NPWT for wounds healing by primary intention, a total of 17 studies must be considered (7 completed, 4 prematurely terminated, and 6 of unclear status). Table 3 lists the studies with the respective sample size used for calculating the data gap.

Table 3

Studies considered for calculating the data gap (multi-page table)

In total, data on at least 2058 patients remain unpublished. Usable data are available on 6981 patients (see Section 4.2). In total, the data of at least 23% [2058 / (2058 + 6981)] of patients included in the studies on NPWT with wounds healing by primary intention are therefore inaccessible.

Furthermore, the remaining 72 studies which neither reported results nor fell under the 12-month rule yet at the time the search was conducted (4 completed, 4 of unclear status, 14 planned, and 50 ongoing studies) are expected to supply data on a total of 29 748 patients with wounds healing by primary intention on the basis of their planned sample size or any alternative information which is already available.

4.4. Overview of assessment-relevant outcomes

Data on patient-relevant outcomes were extracted from 45 studies. Table 4 presents an overview of the available data on patient-relevant outcomes from the included studies. No studies reported usable data regarding the outcomes for either amputation or the need of third-party help or need long-term care.

Table 4

Matrix of patient-relevant outcomes (multi-page table)

4.5. Assessment of the risk of bias at study level and at outcome level

The risk of bias at study level was rated as low for 8 studies (14-1920, AHS.2011.Prevena. Heine.03, CE/US/11/01/PIC, Gillespie 2015, IMS Study, NEPTUNE, S-20130010, The DRESSING Trial). For the remaining 37 studies, the risk of bias at study level was rated as high. The criterion of allocation concealment alone already put 33 studies at a high risk of bias at study level. For the CCF 14-273 study, the risk of bias at study level was rated as high due to missing information on the generation of the randomization sequence and lack of blinding. For the Giannini 2018 study, the risk of bias at study level was rated as high particularly due to the disproportionate exclusion of patients from groups after randomization. For the RRG-104871 and Tanaydin 2018 studies, the risk of bias at study level was rated as high primarily due to reporting bias.

The risk of bias at outcome level was assessed for the 8 studies with low risk of bias at study level. In the remaining 37 studies, the high risk of bias at study level translated directly into the risk of bias at outcome level.

For 14-1920 and the IMS Study, a low risk of bias was found for all outcomes.

For AHS.2011.Prevena.Heine.03, the risk of bias was rated as high for all surveyed outcomes (mortality, adverse events, and pain), particularly since the application of the intention-to-treat (ITT) principle was inadequate or unclear.

For the CE/US/11/01/PIC study, the risk of bias for the outcome wound closure was high due to a violation of the ITT principle. The risk of bias was low with respect to the remaining outcomes (re-intervention, infection, total rate of SAEs, dehiscence, and rehospitalization).

For the studies Gillespie 2015, NEPTUNE and The DRESSING Trial, there was a high risk of bias regarding the outcome for bleeding due to a lack of blinding and absence of a recognizable system. The risk of bias was low for the remaining outcomes (mortality, reintervention, infection, dehiscence, length of hospital stay and rehospitalization).

For the S-20130010 study, a low risk of bias was found with respect to the outcomes for infection and rehospitalization. Due to subjective data collection and lack of blinding, the risk of bias regarding the outcomes for reintervention, dehiscence, and quality of life was rated as high.

4.6. Results on patient-relevant outcomes

4.6.1. Results on mortality

For the outcome “mortality”, usable results were available from 17 studies. No hint of an effect can be inferred from the results of the studies which showed high qualitative certainty of conclusions (14-1920, NEPTUNE). Similarly, the collective analysis of studies with moderate and high qualitative certainty of conclusions failed to show a statistically significant difference between the two treatment groups (OR 0.99; 95% CI [0.46; 2.11]). This analysis is not contradicted by the result of the beta-binomial model, which was used to better account for studies containing at least 1 treatment arm without events (OR 1.09; 95% CI [0.38; 3.14]). Regarding the outcome for mortality, there is consequently no hint of benefit or harm of NPWT in comparison with SWT.

4.6.2. Results on wound closure

For the outcome “wound closure”, usable results from 6 studies were available.

4.6.2.1. Wound healing and time to wound healing

Wound healing

Usable results on wound healing were reported in 4 studies with moderate certainty of conclusions. If data were reported for multiple time points, those from the 6-week point (42 days) were used. A statistically significant difference between treatment groups was found in favour of NPWT (OR 2.54; 95% CI [1.35; 4.79]). Consequently, there is an indication of an effect on wound healing in favour of NPWT.

Time to wound healing

Usable results on time to wound healing were reported in 2 studies with moderate qualitative certainty of conclusions. The meta-analysis using Hedges’ g showed a statistically significant difference in favour of NPWT. However, since the range of the 95% CI was not fully below the irrelevance threshold of –0.2, the effect was rated clinically irrelevant (Hedges’ g –0.53; 95% CI [–0.94; –0.13]). Consequently, there is no hint of an effect regarding time to wound healing.

4.6.2.2. Conclusion on benefit regarding wound closure

Overall, an indication of benefit of NPWT in comparison with SWT was initially found with respect to the outcome for wound closure. In view of the potential publication bias due to the calculated total data gap of 23%, this indication of greater benefit must be downgraded. In terms of the outcome for wound closure, this results in a hint of greater benefit of NPWT in comparison with SWT.

4.6.3. Results on adverse events: Wound complications and treatment complications (AEs)

Regarding the outcome for AEs, usable results from 42 studies were available. Since these studies used different operationalizations, the data were first analysed according to operationalization and then aggregated for a conclusion on any benefit regarding AEs.

4.6.3.1. AEs: Re-intervention

Usable results on the reintervention rate were reported in 23 studies. No hint of an effect can be inferred from the results of the studies which showed high qualitative certainty of conclusions (CE/US/11/01/PIC, IMS Study and NEPTUNE). The collective analysis of studies with moderate and high qualitative certainty of conclusions showed a statistically significant difference in favour of NPWT. The odds ratio is 0.71 (95% CI [0.51; 0.99]). This analysis is contradicted by the result of the beta-binomial model, which was used to better account for the studies with at least 1 treatment arm without an event (OR 0.70; 95% CI [0.42; 1.18]); the identified effect is therefore downgraded. Consequently, there is a hint of an effect in favour of NPWT for the AE regarding re-intervention.

4.6.3.2. AEs: Bleeding

Usable results on bleeding were reported in 7 studies which showed moderate qualitative certainty of conclusions. No statistically significant difference between treatment groups was found (OR 1.73; 95% CI [0.83; 3.64]). Consequently, there is no hint of an effect for the AE regarding bleeding.

4.6.3.3. AEs: Infection

Usable results on infection were reported in 36 studies. The meta-analysis revealed a statistically significant difference in favour of NPWT both for the studies with high qualitative certainty of conclusions (CE/US/11/01 /PIC, Gillespie 2015, IMS Study, NEPTUNE, S-20130010 and The DRESSING Trial; OR 0.59; 95% CI [0.37; 0.93]) and for the totality of the studies (OR 0.62; 95% CI [0.52; 0.74]). Consequently, there is proof of an effect on infections in favour of NPWT.

4.6.3.4. Total rate of SAEs

Usable results on the total rate of SAEs were reported in 14 studies. From the results of the studies with high qualitative certainty of conclusions (14-1920 and CE/US/11/01/PIC), no hint of an effect can be inferred. Similarly, the collective analysis of studies with moderate and high qualitative certainty of conclusions failed to show a statistically significant difference between the two treatment groups (OR 0.86; 95% CI [0.54; 1.37]). This analysis is not contradicted by the result of the beta-binomial model, which was used to better account for studies containing at least 1 treatment arm without events (OR 1.21; 95% CI [0.46; 3.14]). Consequently, there is no hint of an effect on the total rate of SAEs.

4.6.3.5. Separately identified SAEs

Usable results on separately identified SAEs were reported in 23 studies. All 23 studies reported usable data regarding the outcome for dehiscence, and 4 studies did so with respect to the outcome for discontinuation due to AEs.

For the separately reported SAE of dehiscence, no hint of an effect can be inferred from the results of the studies which showed high qualitative certainty of conclusions (CE/US/11/01/ PIC, Gillespie 2015, The DRESSING Trial). The collective analysis of studies with moderate and high qualitative certainty of conclusions showed a statistically significant difference in favour of NPWT. The odds ratio is 0.76 (95% CI [0.59; 0.98]). This analysis is contradicted by the result of the beta-binomial model, which was used to better account for the studies with at least 1 treatment arm without an event (OR 0.77; 95% CI [0.44; 1.35]); the identified effect is therefore downgraded. Consequently, there is a hint of an effect on the separately reported SAE of dehiscence.

There is no hint of an effect on the separately reported SAE of discontinuation due to AEs.

4.6.3.6. Conclusion on benefit regarding adverse events: Wound complications and treatment complications (AEs)

The hint of an effect on the AE due to re-intervention in favour of NPWT must be downgraded due to the potential publication bias resulting from the total calculated data gap of 23%. Consequently, there is no hint of an effect on the AE due to re-intervention.

The proof of an effect on the AE due to infection in favour of NPWT must be downgraded to an indication due to the potential publication bias resulting from the total calculated data gap of 23%.

The hint of an effect on the separately reported SAE due to dehiscence in favour of NPWT must be downgraded due to the potential publication bias resulting from the total calculated data gap of 23%. Consequently, there is no hint of an effect regarding the separately reported SAE due to dehiscence.

Overall, for the outcome adverse events: wound complications and treatment complications (AEs), there is no hint of benefit or harm of NPWT in comparison with SWT. This result is primarily based on the total rate of SAEs, which revealed no difference between NPWT and SWT. With respect to the outcome for infection, there is consequently an indication of an effect in favour of NPWT.

4.6.4. Results on amputation

No usable results were available on this outcome.

4.6.5. Results on pain

Regarding the outcome for pain, usable results from 11 studies were available. Since the studies used different operationalizations, the data were first analysed according to operationalization and later aggregated for a conclusion on the benefit regarding the outcome for pain.

4.6.5.1. Pain – continuous

Usable results on pain in the form of continuous data were reported by 5 studies which showed moderate qualitative certainty of conclusions.

The meta-analytical summary revealed heterogeneous results; therefore, no combined effect was presented. The heterogeneity cannot be explained by the studies being based on wounds of different aetiologies.

The prediction interval overlaps the null, and the studies with statistically significant effects made up less than 50% of the total weight of all studies. The effect direction differs. Consequently, there is no hint of an effect on pain as operationalized in the form of continuous data.

4.6.5.2. Pain – dichotomous

Usable results on pain in the form of dichotomous data were reported by 6 studies which showed moderate qualitative certainty of conclusions.

4.6.5.3. Conclusion on benefit regarding pain

In summary, regarding the outcome for pain, the data are heterogeneous and there is no hint of benefit or harm of NPWT in comparison with SWT.

4.6.6. Results on length of hospital stay and (re-)hospitalization

Regarding the outcome for length of hospital stay and (re-)hospitalization, usable results from 20 studies were available.

4.6.6.1. Length of hospital stay

Usable results on length of hospital stay were reported by 17 studies.

No hint of an effect can be inferred from the results of the studies which showed high qualitative certainty of conclusions (Gillespie 2015, IMS Study, NEPTUNE and The DRESSING Trial). The combined analysis of studies with moderate and high qualitative certainty of conclusions revealed heterogeneity; therefore, no combined effect was presented. The heterogeneity cannot be explained by the studies being based on wounds of different aetiologies.

4.6.6.2. Length of stay in intensive care unit (ICU)

Usable results on ICU length of stay were reported in 2 studies. The IRB00109564 study with moderate qualitative certainty of conclusions reported the results in a continuous format. The NEPTUNE study with high qualitative certainty of conclusions used a dichotomous format. No study showed a statistically significant difference between treatment groups. Consequently, there is no hint of an effect on ICU length of stay.

4.6.6.3. Re-hospitalization

Usable results on re-hospitalization were reported in 11 studies. No hint of an effect can be inferred from the results of the studies which showed high qualitative certainty of conclusions (CE/US/11/01/PIC, Gillespie 2015, S-20130010 and The DRESSING Trial). Similarly, the collective analysis of studies with moderate and high qualitative certainty of conclusions failed to show a statistically significant difference between the two treatment groups (OR 0.87; 95% CI [0.62; 1.22]). Consequently, there is no hint of an effect on re-hospitalization.

4.6.6.4. Conclusion on benefit regarding length of hospital stay and (re-)hospitalization

Summarizing the sub-outcomes for both length of hospital stay and (re-)hospitalization, the data are heterogeneous, and there is no hint of benefit or harm of NPWT in comparison with SWT.

4.6.7. Results on health-related quality of life

Usable data on health-related quality of life were reported in 2 studies which each showed moderate qualitative certainty of conclusions. Both studies reported data on the EuroQol Visual Analogue Scale (EQ VAS). No statistically significant difference between the two treatment groups was found in the meta-analysis (Hedges’ g 0.09; 95% CI [–0.04; 0.22]). Regarding the outcome for health-related quality of life, there is consequently no hint of benefit.

4.6.8. Results on functioning

Regarding the outcome for functioning, usable results from 1 study of moderate qualitative certainty of conclusions were available. The study showed no statistically significant difference between the two treatment groups. In terms of the outcome for functioning, there is consequently no hint of benefit.

4.6.9. Results on need of third-party help or need of long-term care