Introduction

Postoperative complications commonly occur after abdominal surgery and lead to an increase in morbidity, mortality, postoperative hospital stay, and contribute to increased hospital costs [1,2,3,4]. Pulmonary complications occur in up to 30% of the patients [5, 6]. Several factors predispose to increased risk of pulmonary complications such as disruptions of the normal breathing activity with shallow rapid breathing, prolonged supine position, impaired mucociliary clearance, postoperative pain, and anesthesia-induced diaphragmatic dysfunction [7].

The prevention of pulmonary complications has been thoroughly investigated and several different respiratory interventions (e.g. structured breathing exercises, incentive spirometry, breathing with the assisted inspiratory flow, and different kinds of positive airway pressure breathing) have been suggested as possible preventive solutions [8,9,10,11,12,13]. Furthermore, several studies have focused on preoperative physiotherapy [9, 14,15,16]. However, far from all patients have the opportunity to undergo preoperative interventions as a large proportion of patients undergoing abdominal surgery, undergo emergency surgery, which emphasizes the need for clarifying the benefit of strictly performing postoperative care pathways [17].

The purpose of this systematic review and meta-analysis was to investigate whether postoperative interventions directed towards improved respiratory function and compared with usual care can prevent postoperative complications following abdominal surgery with a focus on postoperative pulmonary complications.

Methods

The present systematic review and meta-analysis were planned, conducted, and reported in line with the PRISMA guidelines [18]. Before the study start, a detailed protocol in line with PRISMA-P guidelines [19] describing the review was registered at PROSPERO (identifier: CRD42019133629).

Study eligibility

This study was performed according to the following PICO(S):

The participants (P) of interest were patients (≥ 18 years of age) undergoing intraabdominal gastrointestinal surgery. Patients with preexisting pulmonary/respiratory conditions were also included. The intervention (I) of interest was all postoperative respiratory or mobilization interventions initiated from the end of surgery until hospital discharge. We only included interventions performed after the extubation of the patient. If the patient underwent any instructions about the intervention prior to surgery the study was excluded even though the intervention was only performed postoperatively. The interventions included respiratory interventions with and without adjacent breathing devices such as CPAP (Continuous positive airway pressure), EPAP (Expiratory positive airway pressure), BiPAP (Bilevel positive airway pressure), NIV (Non-invasive ventilation), IPPB (intermittent positive pressure breathing), IPAP (inspiratory positive airway pressure), spirometry, and PEP (positive expiratory pressure), muscle training, or structured breathing exercises. A comparison (C) was made between patients who underwent a respiratory or mobilization intervention postoperatively and patients who were treated with usual care. The primary outcome (O) was postoperative complications including all pulmonary complications (pneumonia, atelectasis, pleural effusion, bronchitis), surgical complications (reoperations, wound infections, reoperations etc.), and medical complications (urinary tract infections, cardiovascular complications, sepsis, etc.). Secondary outcome measures were the length of hospital stay, mortality, and possible side effects to physiotherapeutic interventions. The study types (S) of interest were observational studies (retrospective and prospective) and randomized clinical trials.

Exclusion criteria were patients < 18 years old and patients undergoing non-gastrointestinal surgery. Furthermore, any kind of preoperative instruction of the patient or respiratory or mobilization interventions or initiatives excluded the study. Only English, Danish, Swedish, and Norwegian published literature was included, and data from conference abstracts as well as unpublished data were excluded. No restrictions were made regarding the time of publication.

Search and study selection

A detailed and systematic literature search in MEDLINE, Embase, and PEDRO was conducted on 29 April 2019. The literature search strategy was developed by the first and second author along with a professional medical research librarian and was deliberately made wide to avoid excluding relevant studies. The detailed search strategy from MEDLINE is presented in Table 1. No limits were applied. The search strategy was adapted to Embase and PEDRO. The literature search was supplemented with a hand search of the reference lists of the included studies (snowball-search) [20]. Furthermore, a search for potentially relevant trials at the WHO trial registration website was performed and if a relevant trial was registered as finished, the record of the study was sought out. The potential eligible records were imported to Endnote where duplicates and non-English records were removed. Title and abstract screening were performed independently by authors DK and AB in reference to the eligibility criteria in the online platform Covidence [21]. For papers eligible for inclusion, full-text articles were retrieved and detailed evaluation was performed independently by DK and AB. Disagreements were settled by discussion.

Table 1 Search strategy for Pubmed
Full size table

Data collection and data items

The following study data were independently extracted:

Study data: title, author, year, study design, number of participants.

Patient demographics: age, sex, operation type, duration of surgery, BMI (body mass index), American Society of Anesthesiologist classification (ASA), comorbidities, smoking.

Intervention data: type of respiratory or mobilization intervention, frequency of intervention.

Outcome data: type of postoperative complication, frequency of postoperative complication, length of hospital stay, mortality.

Bias assessment

Randomized controlled trials were evaluated with the Cochrane risk of bias tool [22] and observational studies were evaluated with Cochrane’s ROBINS-I tool for non-randomized studies [23].

Data synthesis

All outcomes were narratively summarized. A meta-analysis was conducted on pulmonary complications as a composite outcome including all types of pulmonary complications [24]. Because of the heterogeneity between the interventions, the meta-analysis was stratified on interventions, which were grouped as follows:

  • Interventions with high expiratory resistance: CPAP, EPAP, BiPAP, NIV including a subgroup analysis only including CPAP

  • Interventions with the isolated assisted inspiratory flow: IPPB, IPAP

  • Interventions with patient-operated ventilation devices: Spirometry, PEP

  • Interventions based only on structured breathing exercises without the use of any respiratory devices

A posthoc subgroup analysis on CPAP was conducted as CPAP has the highest expiratory resistance and differs from BiPAP and NIV in that both BiPAP and NIV also have a high inspiratory flow. CPAP can be administered at surgical wards without respirators and is thereby easier to apply. A subgroup analysis on upper abdominal surgery was performed, as it was suspected that patients with upper incisions were more challenged breathing postoperatively [25]. Furthermore, a subgroup analysis of open and laparoscopic surgery was performed.

The data-synthesis of the meta-analysis was performed with the review manager (Revman version 5.1, Cochrane Collaboration, 2011). The generic inverse variance method was used, and, furthermore, the random-effects model was applied, as it was not assumed that the outcome variables were identically defined or collected between the studies. Results of meta-analyses were only reported if heterogeneity was not considerable (I2 more than 75%) and if data on the outcome was sufficient (more than 100 patients and at least 3 trials). The effect measure of the meta-analyses was odds ratio. As the results in the included studies were presented as frequencies, odds ratios were calculated as crude (unadjusted) odds ratios by 2 × 2 tables. Funnel plots were used to explore the existence of publication bias and small sample size bias [26].

The GRADE approach was used to assess the quality of the body of evidence associated with the outcome of the meta-analysis. The GRADE approach appraises the quality of a body of evidence to assess the certainty in the effect estimates. The quality measures of a body of evidence consider the within-study risk of bias, the inconsistency of the results, indirectness of evidence, imprecision, and reporting bias, and based on that along with the design of the study the quality of the meta-analysis is appointed.

A trial sequential analysis on the outcome of the meta-analysis using TSA software v0.9.5.10 Beta (Copenhagen Trial Unit) was conducted. In the trial sequential analysis, a meta-analytic sample-size calculation [required information size (RIS)] was calculated based on the expected or observed event rate within the control population, the expected clinically relevant relative risk reduction inflicted by the intervention, the chosen type 1 error (alpha-level) and the power (1-beta). In addition, this RIS is adjusted by the observed heterogeneity in the meta-analysis. The heterogeneity-adjusted RIS was calculated for all meta-analyses based on an a priori defined clinically relevant relative risk reduction of 20% for postoperative intervention compared with non-intervention and a control event proportion based on the pooled event proportion in the control group. This proposed clinically relevant relative risk reduction is in nature arbitrary but is proposed in the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) Handbook [27].

Results

The selection process is illustrated in Fig. 1. The literature search yielded 3306 potentially relevant studies, and, furthermore, one record was identified through the WHO trial registration website. After removing duplicates and studies not in English, Danish, Swedish, and Norwegian, a total of 2678 potentially relevant studies were screened. From these, 105 records were sought out in full-text of which 28 studies were included in the systematic review and 23 studies were included in the meta-analysis (Fig. 1).

Fig. 1
figure 1

Flow chart of included studies

Full size image

A total of 2279 patients (38% male, 62% female) were included. The mean age ranged from 32 to 73 years. The patients predominantly underwent open upper abdominal surgery (1635 patients, 72%), followed by a group of both upper and lower abdominal surgery (239 patients, 11%), laparoscopic bariatric surgery (224 patients, 10%), laparotomy (150 patients, 6%), and colorectal surgery (31 patients, 1%). Only one study including 150 patients reported data on emergency surgery [28]. The rest of the included studies reported data on elective procedures. None of the included studies reported whether patients were already using CPAP/BiPAP for obstructive sleep apnea at home. The designs of the included studies varied with 25 randomized clinical trials, two prospective cohort studies, and one retrospective cohort studies (Table 2). The majority of the studies focused on respiratory interventions with and without the use of supporting ventilation devices (26 studies). Seven of included studies investigated combined interventions with both mobilization therapy and pulmonary interventions, however, in all of these studies the control group underwent mobilization therapy and the intervention group underwent both mobilization therapy and pulmonary interventions. None of the included studies investigated no intervention vs. mobilization and pulmonary interventions (Table 2). The most commonly reported outcome was pulmonary complications (n = 25 studies), followed by length of stay (n = 8 studies), surgical complications (n = 3 studies), mortality (n = 2 studies), and medical complications (n = 1). None of the studies reported data on the adverse effects of physiotherapy. Only two studies reported data on intensive care unit length of stay, and no studies reported data on respirator days or reintubation rates. The majority of the studies had a low to unclear risk of bias (n = 23), followed by a critical risk of bias (n = 2), a high risk of bias (n = 2), and moderate risk of bias (n = 1) (Tables 3 and 4).

Table 2 Overview of included studies
Full size table
Table 3 Risk of bias assessment for the included randomized controlled trials
Full size table
Table 4 Risk of bias assessment for the included non-randomized controlled trials
Full size table

Pulmonary complications

Twenty-five studies with a total of 2068 patients reported on pulmonary complications [2, 8, 10,11,12, 29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48]. Of these, 23 studies were included in the meta-analyses [2, 8, 10,11,12, 29,30,31,32, 34,35,36, 38,39,40,41,42,43,44,45,46,47,48]. The intervention was high expiratory resistance (CPAP, EPAP, BiPAP, NIV) in five studies [8, 12, 32, 44, 45], isolated assisted inspiratory flow (IPPB, IPAP) in four studies [10, 29, 31, 46], patient-operated ventilation devices in nine studies [2, 10, 34, 38, 39, 41,42,43, 48], and structured breathing exercises in seven studies [11, 30, 35, 36, 38, 40, 47].

High expiratory resistance (CPAP, EPAP, BiPAP, NIV)

The studies reporting on high expiratory resistance (CPAP, EPAP, BiPAP, NIV) included a total of 192 patients undergoing open elective upper abdominal surgery (n = 97 patients) [32, 44, 45] and either open upper or lower abdominal elective surgery (n = 95 patients) [8, 12]. The high expiratory resistance intervention was compared with patients undergoing usual care with mobilization physiotherapy or breathing exercises. The risk of bias for the studies was low/unclear risk (n = 4) [12, 32, 44, 45] and high risk (n = 1) [8] (Table 3).

The meta-analysis found that ventilation with high expiratory resistance (CPAP, EPAP, BiPAP, NIV) reduced the risk of postoperative pulmonary complications with OR 0.42 (95% CI 0.18–0.97, p = 0.04, I2 = 0%) (Fig. 2). The TSA analysis revealed a RIS of 1,658 patients and the present meta-analysis is, therefore, insufficient to make final conclusions (Fig. 3). The GRADE quality assessment found the level of quality to be moderate as the risk of bias in the individual studies was mostly unclear. A separate subgroup analysis on CPAP including four studies [8, 12, 32, 44] with 132 patients found that CPAP reduced the risk of postoperative pulmonary complications even further with OR 0.34 (95% CI 0.13–0.88, p = 0.03, I2 = 0%) (Fig. 4). The TSA analysis revealed a RIS of 1100 patients (Online Resource 1). The GRADE quality assessment found the level of quality to be moderate as the risk of bias in the individual studies was mostly unclear.

Fig. 2
figure 2

Meta-analysis of high expiratory resistance (CPAP, EPAP, BiPAP, NIV) vs usual care on pulmonary complications with a forest plot and b funnel plot

Full size image
Fig. 3
figure 3

TSA analysis of required information size for high expiratory resistance (CPAP, EPAP, BiPAP, NIV)

Full size image
Fig. 4
figure 4

Meta-analysis of CPAP vs usual care on pulmonary complications with) forest plot and b funnel plot

Full size image

Assisted inspiratory flow only (IPPB, IPAP)

Four studies including 278 patients investigated assisted inspiratory flow only (IPPB, IPAP) [10, 29, 31, 46]. All patients underwent elective open upper abdominal surgery. The intervention was compared with patients undergoing usual care or controls undergoing breathing exercises. All of the included studies had low/unclear risk of bias (Table 3). The meta-analysis showed no significant impact of assisted inspiratory flow only (IPPB, IPAP) on the risk of pulmonary complications OR 1.17 (95% CI 0.70–1.96, p = 0.55, I2 = 0%) (Fig. 5). The TSA analysis revealed a RIS of 602 patients (Online Resource 2). The GRADE quality assessment found the level of quality to be moderate as the risk of bias in the individual studies was mostly unclear.

Fig. 5
figure 5

Meta-analysis of assisted inspiratory flow only (IPPB, IPAP) vs usual care on pulmonary complications with a forest plot and b funnel plot

Full size image

Patient-operated ventilation devices (spirometry, PEP)

Nine studies including 863 patients investigated patient-operated ventilation devices (spirometry, PEP) [2, 10, 34, 38, 39, 41,42,43, 48]. The majority of the patients underwent elective open upper abdominal surgery (n = 575 patients) [2, 10, 38, 39, 41, 43, 48], followed by either open upper or lower elective abdominal surgery (n = 64 patients) [34], and laparoscopic bariatric surgery (n = 22 patients) [42]. The intervention was compared with patients undergoing usual care or controls. The risk of bias score for the studies was low/unclear risk (n = 7) [2, 10, 34, 38, 39, 41, 42], high risk (n = 1) [48], and critical risk (n = 1) [43] (Table 3 and 4). The meta-analysis showed no significant impact of patient-operated ventilation devices (spirometry, PEP) on the risk of pulmonary complications OR 1.08 (95% CI 0.75–2.10, p = 0.83, I2 = 56%) (Fig. 6). The TSA analysis revealed a RIS of 6831 patients (Online Resource 3). The GRADE quality assessment found the level of quality to be moderate as the risk of bias in the individual studies was mostly unclear and, furthermore, one of the included studies was an observational study [43].

Fig. 6
figure 6

Meta-analysis of patient operated ventilation devices (spirometry, PEP) vs usual care on pulmonary complications with a forest plot and b funnel plot

Full size image

A subgroup analysis on patients who underwent upper abdominal surgery [2, 10, 38, 39, 41, 43, 48] was performed with similar results OR 1.34 (95% CI 0.64–2.67, p = 0.46, I2 = 44%) (Online Resource 4). A subgroup analysis excluding patients undergoing laparoscopic surgery (n = 22) [42] and including all patients undergoing open surgery (n = 841) [2, 10, 34, 38, 39, 41, 43, 48] was furthermore performed and found similar results with OR 1.00 (95% CI 0.46–2.15, p = 0.01, I2 = 61%) (Online Resource 5).

Respiratory intervention with structured breathing exercises

A total of seven studies including 554 patients reported data on respiratory interventions with structured breathing exercises and without the use of ventilation devices [11, 30, 35, 36, 38, 40, 47]. The patients underwent elective open upper abdominal surgery (n = 504 patients) [30, 35, 36, 38, 40, 47] and either open upper or lower elective abdominal surgery (n = 50 patients) [11]. The intervention was compared with patients undergoing usual care or controls. The risk of bias score was low/unclear for all the studies (Table 3). The meta-analysis showed no significant impact of respiratory intervention with breathing exercises on the risk of pulmonary complications OR 1.12 (95% CI 0.75–1.65, p = 0.58, I2 = 5%) (Fig. 7). The TSA analysis revealed a RIS of 1647 patients (Online Resource 6). The GRADE quality assessment found the level of quality to be moderate as the risk of bias in the individual studies was mostly unclear.

Fig. 7
figure 7

Meta-analysis of respiratory interventions with breathing exercises vs usual care on pulmonary complications with a forest plot and b funnel plot

Full size image

A subgroup analysis on patients who underwent upper abdominal surgery [30, 35, 36, 38, 40, 47] was conducted with similar results OR 1.13 (95% CI 0.72–1.77, p = 0.60, I2 = 20%) (Online Resource 7).

Surgical complications

A total of three studies including 111 patients reported data on surgical complications [11, 49, 50]. The patients underwent elective colorectal surgery (n = 31) [49], and either open upper or lower elective abdominal surgery (n = 80) [11, 50]. In two of the studies the intervention was mobilization and exercise, and those studies found that 17% of the patients in the control group and 19% of the patients in the intervention group developed a surgical complication [49, 50]. The final study comparing mobilization only with mobilization and structured breathing exercises reported that 15% of the patients had a surgical complication (wound infection) in the mobilization only group and 0% had a complication in the group with mobilization plus breathing exercises, p < 0.05 [11]. The risk of bias in these studies was low/unclear (n = 2) [11, 49], and critical (n = 1) [50] (Tables 3 and 4).

Medical complications

Only one study including 50 patients reported data on postoperative medical complications [11]. The patients in this study underwent either open upper or lower elective abdominal surgery. The study compared mobilization only with mobilization plus structured breathing exercises and reported medical complications such as confusion, atrial fibrillation, cardiac failure, inadequate pain relief, AMI, angina, diarrhea, and vomiting. The study found that 31% of the patients in the mobilization group and 47% of the patients in the mobilization plus breathing exercises group had a medical complication, p > 0.05. The risk of bias for this study was unclear (Table 3).

Length of stay

Eight studies including 599 patients reported on length of hospital stay (LOS) [11, 12, 28, 43, 47,48,49,50]. The patients in the studies underwent elective colorectal surgery (n = 31) [49], either open upper or lower elective abdominal surgery (n = 125) [11, 12, 49, 50], elective open upper abdominal surgery (n = 293) [43, 47, 48], or emergency laparotomy (n = 150) [28]. The intervention of interest was mobilization physiotherapy in two of the studies [49, 50] and different extents of respiratory interventions in the rest of the studies [11, 12, 28, 43, 47, 48]. The studies concerned with mobilization physiotherapy found a mean LOS ranging from 5 to 10 days in the control groups and 5–8 days in the intervention groups [49, 50]. For the studies concerned with respiratory interventions the mean LOS ranged from 4 to 13 days in the control groups and 3–16 days in the intervention groups [11, 12, 28, 43, 47, 48]. The risk of bias of the included studies was low/unclear (n = 5) [11, 12, 28, 47, 49], high risk (n = 1) [48], and critical risk (n = 2) [43, 50] (Tables 3 and 4).

Mortality

Two studies including a total of 200 patients reported data on mortality [11, 28]. The patients underwent elective upper or lower abdominal surgery (n = 50) [11] or emergency laparotomy (n = 150) [28]. Deep breathing exercises with spirometry were the intervention in one of the studies [28] and deep breathing exercises without spirometry were the intervention in the other study [11]. One study reported zero mortality in the control group and 3% mortality in the intervention group (p > 0.05) [11] and the other study reported 10.7% mortality in the control group and 1.3% mortality in the intervention group (p = 0.02) [28]. The risk of bias in both studies was low/unclear (Table 3).

Discussion

This review and meta-analysis found that ventilation with high expiratory resistance (CPAP, EPAP, BiPAP, NIV) might reduce the risk of postoperative pulmonary complications and that this risk reduction was largely driven by CPAP. However, the trial sequential analyses revealed that the required information size was insufficient. Postoperative assisted inspiratory flow therapy (IPPB, IPAP), patient-operated ventilation devices (spirometry, PEP), and respiratory intervention with structured breathing exercises did not reduce the risk of postoperative pulmonary complications. No conclusions on the effect of postoperative physiotherapy on postoperative surgical complications could be made from this review as the studies included reported different interventions, had small sample sizes, and varying degrees of bias. Furthermore, no conclusions on the effect of postoperative physiotherapy on the risk of mortality could be made because of limited data.

The ability of CPAP to reduce pulmonary complications is in accordance with previous publications. A meta-analysis from 2008 found that CPAP reduced the risk of postoperative pulmonary complications with a risk ratio of 0.66 favoring CPAP and a number needed to treat of 14.2 patients [51]. This meta-analysis differs from the previous study by only including patients undergoing abdominal gastrointestinal surgery and not vascular surgery and by only including studies where the intervention was applied postoperatively and not preoperatively or intraoperatively. The beneficial effect of CPAP might be explained by the way CPAP affects the lung tissue. The constant pressure over a longer period facilitates alveolar recruitment opening the basal lung sections and increasing functional residual capacity [52, 53]. Furthermore, it has been suggested that cardiac function might be improved with CPAP through reduced left ventricular afterload [54]. These effects cannot be achieved with assisted inspiratory flow therapy, spirometry, PEP, or breathing exercises, which might explain why these modalities did not point towards reducing the incidence of postoperative pulmonary complications. Prior meta-analyses focusing on postoperative spirometry are in accordance with our results as they did not find an effect of spirometry with respect to prevention of postoperative pulmonary complications [55, 56].

This meta-analysis found a scarcity of published data on postoperative physiotherapy to reduce postoperative surgical complications, medical complications, and mortality why no conclusion in regard to these subjects can be made. Furthermore, all trial sequential analyses revealed that the required information sizes were not met. The length of hospital stay varies from country to country because of cultural aspects and reimbursement, why no final conclusions on the effect of postoperative physiotherapy on LOS were drawn in this study. However, it seems that postoperative physiotherapy has the potential to reduce the postoperative length of hospital stay, which is in accordance with prior investigations in ERAS (Enhanced Recovery After Surgery) settings [57,58,59]. In the funnel plots for high expiratory resistance and CPAP, we would have expected a greater number of both small positive and negative studies. The lack of small negative studies could implicate some degree of publication bias. However, the funnel plots should be interpreted with caution because of the small number of included studies, which can cause funnel plots to appear asymmetric.

There are several limitations to this study. The review is limited by the included studies, which largely had an unclear risk of bias, however, this limitation is already taken into account in the meta-analysis as the body of evidence is graded in accordance with this. Furthermore, some of the included studies are published more than 30 years ago were both surgeries and the perioperative setups were quite different from today’s clinical practice, and some of the breathing modalities investigated in the studies are not used in clinical practice anymore. In this study, a large part of the effect of CPAP is driven by the studies by Carlsson [32] and Ricksten [44] both from the 1980s. Furthermore, some of the drivers of postoperative respiratory complications are pain, opioids, and residual anesthetics. Unfortunately, the majority of the included studies did not report on pain management. We might suspect that older studies were using long-acting anesthetics and no regional or non-opioid pain management. It has previously been shown that epidural analgesia reduces postoperative pulmonary complications [60]. Therefore, the older studies might be difficult to compare with studies using modern perioperative management. This study is strengthened by its concise and transparent methodology. By only including data on patients undergoing postoperative respiratory and mobilization interventions the findings of this study are applicable in not only elective, but also emergency surgery settings were no preoperative physiotherapy or patient instruction can be planned and executed.

This meta-analysis points in favor of CPAP as a postoperative intervention to reduce the risk of postoperative pulmonary complications, however, the TSA analysis found that RIS was not sufficiently met to draw any final conclusions. Compared to other respiratory modalities with high expiratory resistance (EPAP, BiPAP, NIV) CPAP has superior feasibility since it can be administered at surgical wards without respirators. A predictive tool like ARISCAT [61] can help identify the patients at risk for postoperative pulmonary complications and thereby help optimize postoperative care pathways and aim resources.

Conclusions

This meta-analysis indicates that postoperative breathing with high expiratory resistance such as CPAP, EPAP, BiPAP, and NIV might be able to prevent postoperative pulmonary complications and that these preventive abilities are largely driven by ventilation with CPAP. This result should, however, be interpreted with caution, as the trial sequential analyses revealed that the required information size was insufficient.