Safety criteria to start early mobilization in intensive care units. Systematic review

Mobilization of critically ill patients admitted to intensive care units should be performed based on safety criteria. The aim of the present review was to establish which safety criteria are most often used to start early mobilization for patients under mechanical ventilation admitted to intensive care units. Articles were searched in the PubMed, PEDro, LILACS, Cochrane and CINAHL databases; randomized and quasi-randomized clinical trials, cohort studies, comparative studies with or without simultaneous controls, case series with 10 or more consecutive cases and descriptive studies were included. The same was performed regarding prospective, retrospective or cross-sectional studies where safety criteria to start early mobilization should be described in the Methods section. Two reviewers independently selected potentially eligible studies according to the established inclusion criteria, extracted data and assessed the studies' methodological quality. Narrative description was employed in data analysis to summarize the characteristics and results of the included studies; safety criteria were categorized as follows: cardiovascular, respiratory, neurological, orthopedic and other. A total of 37 articles were considered eligible. Cardiovascular safety criteria exhibited the largest number of variables. However, respiratory safety criteria exhibited higher concordance among studies. There was greater divergence among the authors regarding neurological criteria. There is a need to reinforce the recognition of the safety criteria used to start early mobilization for critically ill patients; the parameters and variables found might contribute to inclusion into service routines so as to start, make progress and guide clinical practice.


INTRODUCTION
The survival rates of the critically ill have increased in the past years; consequently, the number of morbidities such patients develop arising from long stays at the intensive care unit (ICU) has also increased. (1)(2)(3) Within this context, early mobilization (EM) performed in a safe manner might reduce such deleterious effects.
Information on safety criteria for EM in adult ICUs were initially published by Stiller and Philips, (4) followed by Stiller. (5) Both studies were based on physiological principles and the authors' clinical experience. Gosselink et al., (1) together with the European Respiratory Society & European Society of Intensive Care Medicine, recommend that patient mobilization ought to be performed under adequate monitoring and with due safety. In turn, Hodgson et al. (6) cited evidence provided by clinical studies and participants' consensus. Finally, Sommers et al. (7) formulated evidence-based recommendations for effective and safe EM in the ICU setting.
Rehabilitation of ICU patients depends on various factors, such as previous physical strength and functioning, level of cooperation, devices connected and the prevalent mobilization culture in each individual service. (8)(9)(10) Some studies have shown that EM is safe and feasible; (11)(12)(13) however, there is not yet a consensus on its outcomes. Some studies (3,6,(13)(14)(15)(16)(17) have described potential benefits, such as reduction of the duration of mechanical ventilation (MV), length of stay in the ICU and the hospital, sedation and duration of delirium and hospital costs, in addition to improvement of the clinical and functional outcomes at hospital discharge. However, these results disagree with those from randomized controlled studies (18)(19)(20) showing that early and intensive mobilization does not change patient functioning and quality of life either at discharge or 6 months after hospital discharge.
For outcomes to be favorable, knowledge of the relationship among potential benefits, eligibility for EM and its related adverse events are relevant. (6,21) Even though the rate of adverse events is equal to or lower than 4%, (14,(22)(23)(24)(25) patients need to be thoroughly assessed based on safety criteria before starting EM. (6) Yet, the safety criteria used vary among different types of ICUs. As a function of this lack of standardization of safety criteria, there is no consensus on which should be used to start EM so as to minimize risk. To provide increasingly more consistent grounds for clinical practice, the aim of the present study was to establish, by means of a systematic review, the most widely used safety criteria to start EM for patients under MV and admitted to the ICU.

METHODS
The present systematic review followed the recommendations formulated in Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). (26)

Inclusion criteria
The following types of studies were included: randomized clinical trials, prospective and retrospective studies, case series with at least 10 consecutive patients and studies with independent or parallel group design. Determination of design followed the classification formulated by the Cochrane Collaboration. (27) Randomized clinical trial protocols and care delivery protocols were also included. Patients had to be over 18 years old, admitted to the ICU and under MV for more than 24 hours. Articles in Portuguese, English, Spanish and French were included. Articles had to contain, in the Methods section, a description of the safety criteria used to start EM.

Exclusion criteria
Articles in which safety criteria to start EM in patients admitted to the ICU and under MV were not described were excluded. In addition, review studies, monographs/ dissertations/theses, annals, chapters from books and experts' points of view or opinions were excluded.

Search strategy
The search was independently performed by two investigators in the PubMed, Physiotherapy Evidence Database (PEDro), Literatura Latino-Americana e do Caribe em Ciências da Saúde (LILACS; in English: Latin American and Caribbean Health Sciences Literature), Cochrane and Cumulative Index to Nursing and Allied Health Literature (CINAHL) electronic databases from the time the databases were launched to May 2015. As per the review aims, the search followed PRISMA recommendations (26) and considered the concepts of target patient and intervention of the PICO strategy, i.e., concepts control and outcome were not included in the search strategy. Outcomes were not defined as search criteria.
Based on Medical Subject Heading (MeSH) terms and adequate descriptors and Boolean operators, the initial search was performed in the PubMed database as follows: [(intensive care units/or intensive care.tw or critical illness/) and (early ambulation/or early mobilization. tw or passive mobilization or active mobilization)]. The search strategy for the other databases was modified as per individual specificities; these details can be requested from the authors. To complement the electronic search, a manual search was performed based on the references cited in the included articles.

Study selection
Two investigators independently conducted a search for potentially eligible studies. Articles were first categorized according to title. Next, their abstracts were analyzed, and only potentially eligible articles were selected. Cases of disagreement were solved by a third examiner, who made the final decision on the eligibility of such articles.

Methodological quality
Randomized clinical trials were assessed according to the PEDro scale, (28) which consists of 11 items to evaluate a study's methodological quality (internal validity and statistical information). With the exception of the first, each item with an affirmative answer was attributed a score of 1 in the final overall classification (score: 0 to 10). Studies with scores of 7 to 10 were considered as high quality, 5 to 6 as intermediate quality and 0 to 4 as low quality. (29) It should be noted that the PEDro score was not used as an inclusion or exclusion criterion but as an indicator of the quality of the scientific evidence provided in the included articles.

Data extraction and variable selection
Data relative to safety criteria were independently extracted from each eligible study by two examiners and recorded on a standardized data extraction form. The safety criteria were categorized as cardiovascular, respiratory, neurological, orthopedic and other; the corresponding variables and parameters were entered in a specific form. Regarding the variables relative to each safety criteria, only the ones cited in at least three articles were considered.

RESULTS
A total of 1,943 articles were located, and 1,462 were selected for triage. A total of 1,223 articles were excluded based on their titles and 96 additional studies based on their abstracts. A total of 143 articles were selected for full-text analysis. Finally, 37 studies were included for systematic review, as they met the inclusion and exclusion criteria ( Figure 1).
The sample size varied from 11 to 2,176 participants, for a total of 6,641 patients from both genders, with an age range of 45.2 to 75.2 years old, and admitted to clinical, surgical or general ICU. Table 1 describes the methodological quality of the randomized clinical trials. (9,13,23,(30)(31)(32) Three out of six studies were registered in PEDro, (9,13,30) and the corresponding score was available in the database. The other three studies (23,31,32) were scored based on full-text analysis and examiner consensus. Scores varied from 4 to 8.
No study was scored on the items related to patient and therapist blinding; in one single study, assessors were blinded. (9) Two studies exhibited the minimum score, 4, (30,32) and only Schweickert et al.'s (9) study had a score of 8.
The safety criteria to start EM are described in table 2. As is shown, the cardiovascular criteria exhibited the largest number of variables (9 total), among which absence of myocardial ischemia, absence of arrhythmia and hemodynamic stability stood out. None of the selected studies reported parameters for tolerated dose of vasoactive drugs or drug combination to attain hemodynamic stability; therefore, these variables could not be quantified.
Relative to the respiratory criteria, variables related with MV -fraction of inspired oxygen (FiO 2 ) < 0.6 and/ or positive end-expiratory pressure (PEEP) < 10cmH 2 O -were the ones with highest concordance, being cited by 14 authors.
As concerns the neurological criteria, the patients' level of consciousness was subjectively assessed. Therefore, this variable exhibited greater variation. Table 3 describes information on study design, sample characteristics, ICU type, mobilization protocols and occurrence of adverse events. Most were general ICUs (14) followed by 8 clinical ICUs. The mobilization protocols were similar regarding the treatment offered; a large part of the studies followed a same order of progression: mobilization in bed, sitting on the edge of bed, standing and walking. The safety of these interventions was assessed based on the occurrence of adverse events. Although 15 studies did not report on this outcome, the rate of adverse events was low. When mentioned, the most frequent adverse events were desaturation, tachypnea, heart rate changes, loss of devices (such as tubes and catheters) and postural hypotension.

DISCUSSION
The present study stands out for having systematically assessed the safety criteria most widely employed to start EM for critically ill patients under MV and admitted to the ICU according to their individual clinical condition and the invasive devices connected to them.
The respiratory criteria exhibited higher concordance among the included studies. In this regard, we emphasize peripheral oxygen saturation (SpO 2 ), mentioned in 14 studies, 11 of which consider SpO 2 > 88% safe to start mobilization. According to Stiller and Philips (4) and Amidei et al., (59) SpO 2 is a safe and individualized monitoring parameter to incorporate into clinical practice. This finding is similar to the ones reported by Stiller et al. (5) and Gosselink et al., (1) according to whom SpO 2 > 90% with 4% oscillation is indicative of satisfactory respiratory reserve to tolerate mobilization.
As a function of the need for MV in critically ill patients, they are benefited by advances in intensive care and new approaches to MV. (39) The feasibility and safety of mobilization of patients with artificial airways have already been demonstrated, provided the latter are secured and in their proper place. (12) Twenty studies mentioned ventilation parameters; 14 of them cited FiO 2 < 0.6 and/ or PEEP < 10 cmH 2 O. FiO 2 < 0.6 was also selected by Gosselink et al. (60) as a criterion to start their mobilization protocol. Similar parameters are recommended by Hodgdon et al. (6) and Sommers et al., (7) who consider FiO 2 ≤ 0.6 and PEEP ≤ 10 cmH 2 O to be safe for mobilization of the critically ill.
Among the neurological criteria, assessments of intracranial pressure (ICP) and level of consciousness stood out. Witcher et al. (52) considered that patients with elevated IPC and in whom deep sedation is combined with neuromuscular blockers are not candidates for participation in EM protocols and daily sedation interruption. Other reasons hindering EM are paralysis or paresis, cognitive dysfunction and abnormal brain perfusion, in addition to the use of devices for continuous brain monitoring. (17,52) Regarding continuous monitoring of the patients' level of consciousness, daily interruption of sedation or maintenance of the minimally required levels are recommended to enable a more trustworthy assessment, in addition to reducing the severity of complications associated with stays in the ICU. (9,46) The present systematic review found that the patients' level of consciousness was not assessed in an objective manner, with the help of scales, but subjectively, resulting in a wide variation of parameters. This finding might be explained by the various aims and methods of the studies; some of them required patients to be awake and cooperate with the treatment suggested, while in others, patients were under deep sedation.
Adverse events are usually associated with respiratory or cardiovascular complications and with the devices connected to patients. (25) Collings et al. (13) asserted that such events are a reflection of the limited individual reserve of patients and might manifest the physiological changes expectably induced by exercise. (2) Adverse events do not increase hospital costs or length of stay at the hospital. (13) Some findings do not reflect the situation in clinical practice. Patients under palliative care are often not included in study populations due to their extreme frailty and lack of chances for a cure with treatment and their consequent higher odds for treatment not to modify their functioning. (61) Therefore, one might infer that authors intend to avoid bias in their studies. However, when one considers that the standard physical therapy practices in the ICU setting are similar to the ones reported in studies, the aforementioned assertions differ from Marcucci's view, (62) according to whom physical therapy is complementary to palliative care, has a preventive nature, affords symptom relief and, whenever possible, provides patients an opportunity to develop and maintain their functional independence.
Safety criteria might go beyond the clinical and physiological ones, as shown in the present study. Restrictions in human and material resources might result in limitations to the mobilization of the critically ill, in addition to the particularities of each individual patient, which should always be emphasized. For EM to become essential and indispensable in the rehabilitation of the critically ill, professionals, physical therapists in particular, should be able to assess and suggest a safe treatment, adequate to the patient and duly monitored, so that the potential benefits of mobilization result in patient gains.
For outcomes to be systematically favorable to patients, multidisciplinary staff members should have the required knowledge and be in continuous harmony. (32)

Study limitations
To the best of our knowledge, the present is the first systematic review that analyzed the safety criteria used to start EM. However, as the study was based on the methods used in the analyzed studies, some limitations must be pointed out. First, as in any systematic review, there was potential for bias selection; however, we employed a broad-scoped search strategy so as to include the largest possible number of articles, analysis was independently performed by two reviewers and the exclusion criteria were clearly documented. Second, in some articles, the information was considerably limited (or provided substantially limited information on the methods used). Third, comparisons between studies were difficult due to the heterogeneity between samples and divergence in methods; the diversity of results, derived from the aims of each individual study, posed a true challenge to the present review. In addition, we should observe that the articles provided little information as to the occurrence of adverse events, which could have contributed to the interpretation of some data and helped readers in the choice of measures to adopt in clinical practice. These shortcomings stress the need for articles to include good descriptions of methods and information in general to facilitate reproducibility and the consolidation of the scientific evidence in this field.

CONCLUSION
Cardiovascular criteria were the most frequently cited in the analyzed studies, exhibiting the largest number of variables. For respiratory criteria, the variables related to mechanical ventilation exhibited the highest concordance among authors. The authors considerably diverged in relation to neurological criteria, with lack of consensus mainly for assessment of the level of consciousness.
The present study reinforces findings reported in other studies on the criteria frequently used to ensure safety in the early mobilization of the critically ill, an approach currently growing in the intensive care setting in Brazil and abroad. The parameters and variables located in the present systematic review might be included in service routines so as to start, make progress and guide clinical practice.