Updating the evidence-base for suctioning adult patients: a systematic review.

OBJECTIVES
To update a previous clinical practice guideline on suctioning in adult patients, published in the Canadian Respiratory Journal in 2001.


METHODS
A primary search of the MEDLINE (from 1998), CINAHL, EMBASE and The Cochrane Library (all from 1996) databases up to November 2007, was conducted. These dates reflect the search limits reached in the previous clinical practice guideline. A secondary search of the reference lists of retrieved articles was also performed. Two reviewers independently appraised each study before meeting to reach consensus. Study quality was evaluated using the Jadad and PEDro scales. When sufficient data were available, a meta-analysis was conducted using a random effects model. Data are reported as ORs, weighted mean differences and 95% CIs. When no comparisons were possible, qualitative analyses of the data were completed.


RESULTS
Eighty-one studies were critically appraised from a pool of 123. A total of 28 randomized controlled trials or randomized crossover studies were accepted for inclusion. Meta-analysis was possible for open versus closed suctioning only. Recommendations from 2001 with respect to hyperoxygenation, hyperinflation, use of a ventilator circuit adaptor and subglottic suctioning were confirmed. New evidence was identified with respect to indications for suctioning, open suction versus closed suction systems, use of medications and infection control.


CONCLUSIONS
While new evidence continues to be varied in strength, and is still lacking in some areas of suctioning practice, the evidence base has improved since 2001. Members of the health care team should incorporate this evidence into their practice.

Given the concerns of previous reviews, and the numerous gaps in the evidence on suctioning, there is a need to systematically review the recent literature regarding this technique. Thus, the purpose of the present article was to systematically review the literature pertaining to suctioning of the airway in both intubated and nonintubated adult patients published since the Brooks et al (4) (4) CPG. Key words included "suction", along with "airway", "trachea", "pharynx" (and variants "oro-" and "nasopharynx"), "subglottic'', "oxygenation" (and variants "hyper-oxygenation" and "preoxygenation"), "insufflation", "endotracheal" and "tracheostomy". These are essentially the same terms that were used by Brooks et al (4) in their CPG. A secondary search of the reference lists in all retrieved articles was also conducted. Papers were identified for potential review if their titles suggested relevance to either of our research questions.

selection of articles
All English language studies conducted with adult patients (18 years or older), published since 1998 (MEDLINE) or 1996 (CINAHL, The Cochrane Library, EMBASE) were selected. The selection converged on the strongest sources of evidence: randomized controlled trials (RCTs) or randomized crossover (RCO) study designs. However, to obtain information on all aspects of suctioning (eg, harms, infection control), we also reviewed a variety of low-level experimental designs (ie, nonrandomized trials, case reports) to provide evidence for topics in which no high-level studies were available.
A title and abstract review of the initial search results was performed independently by two reviewers who then met to reach consensus on which full articles were potentially eligible for review. If abstracts were not available, full articles were selected based on their titles.

Data extraction
The research team was divided into three pairs for critical appraisal of the papers. Two reviewers independently extracted all of the data from each selected trial using a predefined form modified from forms used previously (5,6). The extracted data included general information about the studies (eg, eligibility criteria, study design), quality assessment including randomization, participant dropout and withdrawals, and description of participants, groups and/or interventions, outcome measures and results. The two reviewers were able to reach consensus regarding the data extracted for all studies. Attempts were made to contact four authors to obtain additional or missing data, and randomization information and data (7)(8)(9)(10). Two responses were received (7,8).

Quality assessment
Each pair of reviewers independently assessed the quality of each study before meeting to reach consensus. They considered the sources of bias that have a major influence on the magnitude of effect size in clinical trials such as unconcealed randomization, blinding, dropouts and withdrawals. The quality of each study was rated using formal quality scores according to two scales: a 0 to 5 scale developed by Jadad et al (11), and the 0 to 10 scale used in the Physiotherapy Evidence Database (PEDro) suggested by Maher et al (12). The scores presented in the present report represent the consensus of the two reviewers for each scale. Higher scores indicate higher quality. Agreement between reviewers was measured using the kappa statistic (13,14). SAS software (version 9.1.3, SAS Institute Inc, USA) was used for kappa calculations.

Analysis and interpretation
Descriptive data were summarized in tables that included information on study characteristics, groups/interventions, outcome measures and a summary of main results.
There were enough high-quality papers regarding open suction (OS) versus closed suction (CS) to combine for metaanalysis (with the Cochrane Collaboration's Review Manager 4.2 software) using a random effects model on certain outcome measures. Meta-analyses were not performed for other outcome measures because of large heterogeneity in the groups or interventions, and the outcome measures. Thus, a qualitative analysis of the literature with summaries based on the new evidence published since the Brooks et al (4) review was also completed.

ResUlTs AnD DisCUssiOn
The primary literature search of the four databases yielded 302 citations. After eliminating duplications, 123 citations were screened for possible review, of which 81 papers received full review. Thirty-two papers were identified from the secondary search; 10 of these were screened and five received full review. In total, 15 RCTs and 13 RCO studies were accepted following review. Information from these studies is presented in Tables 1 to 7. Also included is information from 11 low-level studies to fill in gaps and provide the 'best available evidence' for topics in which there were no randomized studies. It should be noted that almost all of the evidence included relates to the suctioning of intubated patients; there is very little information available regarding the use of this technique in nonintubated patients.
There was good agreement between reviewers on both the Jadad scale (11) (kappa score 0.89; 95% CI 0.74 to 1.00) and the PEDro scale (12) (kappa 0.90; 95% CI 0.80 to 1.00). Mean quality scores for the 28 included articles were 2.04 (out of 5) on the Jadad scale and 6.21 (out of 10) on the PEDro scale. The most frequent scores were 2 (range 1 to 5; n=17) for the Jadad scale, and 6 (range 4 to 9; n=11) for the PEDro scale. The most common internal validity errors for the Jadad scale were reporting an appropriate method of randomization and describing the study as a double-blind design. For the PEDro scale, the most frequently missed items were concealing the treatment allocation and blinding of subjects, caregivers and outcome measure assessors. It is difficult to conduct double-blind studies in which the interventions are generally easily differentiated by the patients and caregivers (eg, OS versus CS).

First research question
What are the clinical indications and precautions for suctioning adult patients? indications for suctioning: Our first research question examined the precautions and clinical indications for suctioning. Only one new study was identified. Predictors of retained secretions were examined in a prospective observational study (15) on 66 patients requiring mechanical ventilation (MV). A sawtooth pattern on a flow-volume loop (seen on the ventilator screen) and respiratory sounds over the trachea had likelihood ratios of a positive test (presence of secretions greater than 0.5 mL) of 2.70 and 2.50, respectively, and likelihood ratios of a negative test of 0.25 and 0.45, respectively. When these two parameters were both present, the likelihood ratio for a positive test was increased to 14.7, while the negative test did not have an appreciable change (0.42) (15). Summary: For patients on MV, the best available evidence suggests the presence of retained secretions if a sawtooth pattern and/or respiratory sounds over the trachea are present.

second research question
What is the evidence for the methods and techniques that optimize benefits and reduce complications or risks when suctioning intubated, nonintubated or tracheotomized adult patients?
The following sections of the Results and Discussion are organized according to how the pool of new literature was categorized. It should be noted that all of the new evidence is derived from papers that either studied patients on MV or in a few cases, used a lung model. No papers that studied nonintubated patients were found.

hyperoxygenation
Hyperoxygenation is the practice of increasing the oxygen concentration (typically to 100%) for a short period before and, in some cases, after suctioning. The 2001 CPG recommended that hyperoxygenation should be used before and after suctioning to prevent oxygen desaturation in mechanically ventilated patients who underwent trauma, or had cardiac or chronic obstructive pulmonary disease (COPD) (4). Only one new study regarding the effect of hyperoxygenation was found (9). This study compared no hyperoxygenation with supplying oxygen for 1 min before and after CS in 30 patients in the intensive care unit (ICU) who required MV. It was not clear whether the order of the interventions was randomized, and the request for information from the authors was unsuccessful. Hyperoxygenation resulted in higher levels of aterial oxygen saturation (SaO 2 ) and pressure of arterial oxygen (PaO 2 ), with no adverse effects. Summary: For patients on MV, the new evidence continues to support the use of hyperoxygenation before and after suctioning to maintain oxygenation levels.

hyperinflation
Hyperinflation is the practice of adding extra volume to the lungs, either via a ventilator or a manual resuscitation bag, before suctioning. In the 2001 CPG (4), the use of hyperinflation was not recommended as a means to improve oxygenation in preoxygenated patients following coronary artery bypass surgery. One new RCT (16) examined the issue of hyperinflation (Table 1). Barker and Adams (16) reported no effect of hyperinflation on oxygenation, heart rate (HR), blood pressure, mixed venous oxygen saturation (SvO 2 ) and dynamic compliance following OS in 17 preoxygenated patients with acute lung injury who received pressure support ventilation. Hyperinflation also did not reverse the effects of disconnection from the ventilator during endotracheal suctioning. The authors noted that disconnection from MV in patients with acute lung injury may result in significant derecruitment of the lungs.
Two new RCO studies (17) investigated the effect of hyperinflation on other variables (Table 1). Choi and Jones (17) examined the effect of manual hyperinflation on static lung compliance and airway resistance in 15 ICU patients with ventilator-associated pneumonia (VAP). Hyperinflation improved lung compliance and decreased airway resistance for up to 30 min following suction. The second study (18) investigated the effect of manual versus ventilator hyperinflation on sputum wet weight and static compliance in 20 ICU patients who required MV. There was no difference in sputum weight, but both methods of hyperinflation improved static compliance following suction. Summary: For mechanically ventilated patients, the new evidence supports the previous recommendation of not to use hyperinflation to improve oxygenation. However, manual or ventilator hyperinflation may result in increased lung compliance in patients on MV, and manual hyperinflation may decrease airway resistance in patients with VAP. The clinical implications of these physiological changes are unclear.

Use of an adaptor in the ventilator circuit
An adaptor is a device added to a ventilator circuit to allow for suctioning without disconnecting the patient from the ventilator. In the 2001 CPG (4), the use of an adaptor that allowed the continuation of MV for the duration of the suctioning procedure was preferred over disconnection from the ventilator if no access to hyperinflation or hyperoxygenation was available. Suctioning through an adaptor was as effective as disconnecting the ventilator circuit and providing hyperoxygenation before and after suction. One new study (19) has examined the effect of an adaptor with hyperoxygenation (Table 1). Metz et al (19) evaluated the effect of a close-fitted adaptor with intermittent endotracheal tube (ETT) clamping and pressure-controlled hyperinflation after suction in an RCO study of 16 patients requiring MV for respiratory failure. Offventilator suction resulted in a decrease in PaO 2 (from baseline values) for up to 60 min, while suctioning through an adaptor resulted in an increase in PaO 2 . Between-group differences from raw data reported in the paper by Metz et al (19) were calculated by our group. Suctioning through an adaptor resulted in significantly higher PaO 2 values for up to 1 h postsuction compared with off-ventilator suction. Summary: The new evidence continues to support the use of an adaptor for suctioning to maintain oxygenation in mechanically ventilated patients.

Os versus Cs systems
OS refers to the suctioning technique in which patients are disconnected from the ventilator and a single use catheter is inserted into the ETT to suction secretions before the patient is reconnected. CS is the suctioning technique in which the suction catheter is enclosed in a plastic sleeve and can be inserted into the ETT without the need for disconnection from the ventilator. In the previous CPG, there was insufficient evidence on which to base a recommendation regarding the effectiveness of CS or OS on oxygenation, secretion removal or frequency of colonization for pneumonia. No meta-analyses were conducted in the previous CPG. Our group located seven new RCTs and five new RCO studies ( Table 2). There were enough new studies to conduct meta-analyses on the effectiveness of CS versus OS systems on SaO 2 (without preoxygenation), end-expiratory lung volume (without preoxygenation) and VAP (incidence and incidence per 1000 patient days). Qualitative analyses on outcome measures were conducted in cases for which there was either large heterogeneity or too few studies to combine for meta-analysis. Quantitative analyses: Two new RCO studies (8,20) assessed changes in SaO 2 in 19 subjects during OS and CS, both without preoxygenation. A meta-analysis ( Figure 1) indicated that there was no difference between the two systems (weighted mean difference [WMD] = 3.21%; 95% CI −3.65% to 10.07%). The same two studies assessed the effect of OS and      Neither study showed a significant difference between CS and OS methods on immediate (1 min or less) or short-term (3 min to 5 min) postsuctioning PaO 2 levels. Two RCTs (22,23) investigated the cost of OS and CS systems. Lorente et al (22) reported a higher cost for CS (US $11.11±2.25/patient/day versus US $2.50±1.12/patient/day). A subsequent study (23) supported this result, but only in patients who were mechanically ventilated for less than four days. When patients were ventilated for longer periods, CS systems were less expensive (€1.6±2.8/patient/day versus €2.5±0.5/patient/day).
Two new RCTs (21,25) compared OS and CS systems with respect to length of stay (LOS) in the ICU. Neither study reported a significant difference. Topeli et al (25) also found no difference between suctioning systems on mortality.
Two new RCTs (22,25) compared OS and CS systems on the duration of MV in ICU patients. Neither of the studies reported a significant difference between the two procedures.
Only two new RCO studies (20,28) investigated the effect of CS and OS systems on HR and blood pressure (BP). Cereda et al (20) reported no differences between systems. Lee et al (28) found higher HR and BP values following OS; however, the clinical significance of the magnitude of the differences (3% to 6%) between the suctioning systems is difficult to assess. Summary: The new evidence suggests that there is no difference between CS and OS for oxygenation, mortality, LOS in ICU or duration of MV. CS systems are more expensive than OS systems, although the difference may depend on the duration of MV.

Frequency of changing Cs systems
Two new studies (29,30) investigated the effect of changing CS catheters at 24 h or 48 h intervals ( Table 2). Darvas and Hawkins (29) found no differences in the incidence of VAP, duration of MV or LOS in ICU, or mortality. Quirke (30) also found no differences in LOS in ICU or mortality. There were also no differences in the overall number of colonized suction tips or the number of sputum colonies; however, the study did report a greater number of colonies in colonized tips in the group whose catheters were changed every 48 h. Summary: The new evidence suggests that there is no difference between 24 h and 48 h changes of CS systems with respect to the incidence of VAP, mortality, LOS in ICU, duration of MV or colonization of catheter tips.
The Dutch Working Group on Infection Prevention (26) also investigated the frequency of changing CS systems and recommended a change at 48 h unless there was mechanical failure or soiling of the suction system.

subglottic suctioning
Subglottic suction refers to the technique in which an ETT is modified to allow suction above the cuff. The suction is used to remove secretions that pool above the cuff, which prevents aspiration of these secretions into the lung. The 2001 CPG (4) found some evidence to support the use of such modified ETTs, with capabilities for continuous or intermittent subglottic suctioning to prevent subglottic aspiration. Two new RCTs (31,32) that addressed subglottic aspiration of secretions in subjects expected to receive more than 72 h of MV were found (  Kollef et al (33) investigated whether continuous aspiration of subglottic secretions (CASS) decreased the incidence of VAP in patients receiving MV (mean 1.5 days) following cardiac bypass surgery. Patients received either CASS through a specially designed ETT (n=160) or routine postoperative care without CASS (n=183). Risk factors for VAP were equal between groups. The relative risk for VAP suggested no difference between the two systems (RR=0.61; 96% CI 0.27 to 1.40; P=0.238). Episodes of VAP occurred significantly later among patients receiving CASS, thus suggesting a possible delaying role for this procedure. The inferences that can be made from these results may be limited because subjects were randomly assigned according to their year of birth.
One case report (34) described a patient who had a fatal tracheal-innominate artery fistula formation associated with an ETT tip. The authors believed that the higher rigidity of the ETT with a subglottic suction port, the extended duration of intubation and the complexity of the patient's airway management problems (second-and third-degree burns to the face and neck, ETT sutured in place) may have contributed. Under normal circumstances, ETTs with subglottic suction ports do not impose greater risk.
The optimal pressure used for subglottic drainage is not clear. A laboratory model study (35) found that a pressure of 30 mmHg was more effective than 10 mmHg, 40 mmHg or 50 mmHg in removing all types of secretions; however, the authors did not examine the effect on mucosal damage.

Summary:
The new evidence continues to support the use of subglottic suctioning for the prevention of pneumonia in patients expected to require greater than 72 h of MV.

Minimally invasive suctioning
Minimally invasive suction is a technique designed to have less impact on the patient. No saline is instilled, no hyperinflation or hyperoxygenation is applied, and the suction catheter (open) does not pass beyond the tip of the ETT. Two new RCTs (36,37) investigated minimally invasive suction (Table 4). Leur et al (36) compared standard OS (49 cm catheter; n=197) with on-demand minimally invasive OS (29 cm catheter; n=186). There were no differences between groups for duration of MV, incidence of pulmonary infection, mortality or LOS in the ICU. Patients who received minimally invasive OS reported fewer adverse effects (decreased SaO 2 and arrhythmias, increased BP and pulse pressure, and hemoptysis). However, this study was flawed because 163 members of the minimally invasive suction group received standard OS (total of 10% of all suction interventions) when nurses believed that the minimally invasive treatment was not working well. Treatment was also withheld from 63 standard OS patients because nurses believed treatment was not indicated.
Leur et al (37) assessed the recollection of suctioning and discomfort with suctioning in a subset of these patients, and found a lower prevalence of recollection in the group receiving minimally invasive suctioning. However, among subjects who recalled the suctioning experience (OS n=46; minimally invasive OS n=19), there was no difference in the reported level of discomfort between the two suctioning techniques. The issue of minimally invasive suctioning has emerged only since the most recent CPG (4), and these two studies (36,37) are too weak to enable any conclusions.

saline instillation
Saline instillation is the practice of instilling aliquots of saline (generally 2.5 mL to 5 mL) into the ETT just before inserting the suction catheter. The controversy over the use of saline has continued since the completion of the 2001 CPG (4), which reported insufficient evidence to make a recommendation. We found one new RCT and two new RCO studies in mechanically ventilated patients (Table 5). Outcome measures included saturation and dyspnea (by visual analogue scale).  (38) demonstrated that the SpO 2 took longer to return to baseline following OS with increasing amounts of saline (0 mL, 2 mL, 5 mL). However, it is unlikely that these differences would be clinically significant because the changes were very small (less than 2% difference in SpO 2 ). Similarly, in a small RCT, Ackerman and Mick (39) found that SaO 2 decreased to a greater extent with the use of saline at 4 min, 5 min and 10 min postsuctioning. Again, the changes were very small and likely clinically insignificant. In a lower-level study, O'Neal et al (40) reported no difference in the level of dyspnea with saline instillation in mechanically ventilated patients. In a subanalysis, the authors found that older patients (older than 60 years of age) reported higher levels of dyspnea with saline. However, these results must be interpreted with caution due to the very small sample size (n=5).
The distribution of saline in patients receiving MV was examined in a small crossover pilot study (41). In the supine position, greater amounts of instilled saline reached the posterior portion of the right lower lobe compared with the left. Klockare et al (41) also reported a more uniform distribution of saline between and within the lungs with nebulization versus instillation.

Summary:
The new evidence suggests that the use of saline with suctioning may cause a decrease in oxygen saturation; however, this change may not be clinically significant. There is also some suggestion that it may be possible to manipulate the distribution of saline in the lung by the mode of delivery.

Medications administered during suctioning
The 2001 CPG (4) made recommendations regarding the use of medications for the prevention of increased intracranial pressure or cerebral perfusion pressure, and minimizing or decreasing bradycardia during suctioning. We found two new RCTs and one RCO that examined a variety of medications to assist in secretion clearance, suppressing cough and preventing deleterious changes in HR and mean arterial pressure (MAP) ( Table 6).
Leone et al (42) studied 81 patients undergoing maxillo-facial surgery who received propofol (target concentration 3.5 µg/mL) as well as three randomly assigned target concentrations of remifentanil. Patients who were given the highest concentration of remifentanil (15 ng/mL) had less coughing with suctioning than subjects receiving 5 ng/mL or 10 ng/mL. MAP (74 mmHg versus 68 mmHg versus 62 mmHg for the three  Studies have shown that pain stimulation results in an increase in the bispectral index (BIS) of an electroencephalograph if the level of analgesia is weak. In a small case series (44), the effect on BIS during suctioning of an intravenous bolus of alfentanil (14 µg/kg/h) in addition to a consistent sedation protocol (midazolam 0.1 mg/kg/hr and fentanyl at 4 µg/kg), found that BIS values were significantly lower during the alfentanil period compared with the control period. Leone et al (43) reported that BIS increased significantly when higher doses of remifentanil were used alone without sedation. It should be noted, however, that a recent large RCT (45) failed to support routine BIS monitoring as part of standard practice to reduce anesthesia awareness; consequently, the value of BIS to assess responses to suction is unclear.
Frass et al (46) showed that compared with placebo, twice daily administration of the homeopathic mucolytic potassium dichromate (five spherical saccharose globules impregnated with a 30:1 dilution of potassium dichromate), reduced secretions, facilitated earlier extubation and decreased ICU LOS in patients with COPD who had failed extubation due to profuse, tenacious, stringy secretions. Summary: The new evidence suggests that while remifentanil may blunt the cough reflex, it may also be associated with a decrease in HR and MAP. Potassium dichromate may help facilitate weaning in COPD patients with profuse, tenacious, stringy secretions.

infection control issues
The 2001 CPG (4) identified only one study that examined whether suctioning technique should be clean or sterile; there was insufficient evidence to make a recommendation. Since that time, there have been two small case series (47,48) (both n=20) investigating the contamination of oral suction devices. One study (47) examined the contamination of endotracheal suctioning equipment, the other (48) was a case series that examined droplet spread with suctioning. Contamination of the oral suction device was common in both studies, yielding positive cultures in 94% and 80% of the devices, respectively. Both studies also reported a concern with the storage of suctioning equipment. Oral suction devices were found in beds, on medical equipment and even on the floor. Furthermore, Sole et al (48) found slightly lower contamination rates on the proximal end of the suction tubing (83%) and connection port of the inline suction catheter (61%). Ng et al (49) reported that visible droplets were scattered over a distance of 60±39 cm from the ETT and 30% of the bacteria were similar to those found in the trachea of OS patients. Summary: Bedside storage of oral suctioning devices in a clean holder or other set up should be considered. Health care professionals must be cognizant of regional guidelines and standards for high-risk respiratory procedures in outbreak and nonoutbreak situations. This will generally include the use of personal protective equipment such as N95 masks for suctioning (

Miscellaneous issues
Frengley et al (51) studied airway pressures during CS suctioning in 16 patients using three different ventilator modes (pressure control, volume control and continuous positive airway pressure/pressure support) ( Table 7). End-expiratory pressure increased to a greater extent with the insertion of the catheter, and subatmospheric airway pressures were detected during suctioning with volume control ventilation. These results were supported by a bench study (52).
Two new papers have studied the size of suction systems. In a very small RCO study, Vandenberg et al (53) demonstrated that a larger oral suction tip and tubing removed thicker substances from the oropharynx more quickly (Table 7). In a bench study, Stenqvist et al (52) found that a larger catheter (14 Fr versus 12 Fr) inserted into a smaller endotracheal tube (7 mm versus 8 mm) generated a much lower subatmospheric pressure (OS system). Finally, the question of continuous versus intermittent suctioning has also been investigated. Stenqvist et al (52) evaluated the effect of suctioning on 'alveolar' pressure using a lung model with 12 Fr and 14 Fr suction catheters. They demonstrated that suction flow was very high for 2 s to 4 s when initial suctioning was applied. Thus, it appears that intermittent suctioning may result in higher suction flows each time suctioning is initiated, compared with continuous suctioning. Summary: The best available new evidence suggests that health care providers should use caution with CS during volume-controlled ventilation. Catheter sizes should be matched appropriately with ETT sizes, and continuous suction may prevent the very high initial flows generated with intermittent suctioning.

COnClUsiOn
Suctioning is a technique used by many members of the health care team along the continuum from critical care to patients at home. Controversy remains over the optimum technique for this procedure, which is not without its risks to patients. The present systematic review has provided an update of the evidence related to suctioning adult patients since the publication of the most current CPG (4). We have been able to provide a quantitative summary of the evidence on OS versus CS systems for specific outcome measures including VAP, endexpiratory lung volume and SaO 2 . We have also presented a qualitative summary of the new evidence relating to many other areas of suctioning practice. The information we have provided in our review is important to guide the health care team in the provision of best practice for our patients.
It is necessary to recognize, however, that sample sizes in most of the new literature are small, and there continues to be gaps with respect to indications for suctioning, optimal suction pressures, clean versus sterile technique and the duration of the suction pass. In addition, there is very limited information regarding suctioning technique in nonintubated patients. Thus, there remains a need for further work in many areas related to suctioning practice.