Intended for healthcare professionals

CCBYNC Open access
Practice Rapid Recommendations

Corticosteroid therapy for sepsis: a clinical practice guideline

BMJ 2018; 362 doi: https://doi.org/10.1136/bmj.k3284 (Published 10 August 2018) Cite this as: BMJ 2018;362:k3284

Population

Recommendation applies to: Recommendation does not apply to: Adults and children Patients with pre-existing adrenal insufficiency Non-infectious causes of shock Neonates Pregnant women Any infectious source Patients with and without shock Intra abdominal infections Pneumonia Anaphylactic Cardiogenic Hypovolaemic People with sepsis SOFA score of at least 2

Comparison

or Corticosteroid therapy No corticosteroid therapy Usual care only Intravenous corticosteroids plus usual care Corticosteroids No corticosteroids Usualcare Usualcare + CCS

More details We suggest corticosteroid therapy rather than no corticosteroid therapy. Either option is reasonable. Strong Benefits outweigh harms for almost everyone. All or nearly all informed patients would likely want this option. Weak Benefits outweigh harms for the majority, but not for everyone. The majority of patients would likely want this option. Weak Benefits outweigh harms for the majority, but not for everyone. The majority of patients would likely want this option. Strong Benefits outweigh harms for almost everyone. All or nearly all informed patients would likely want this option.

Comparison of benefits and harms

Favours corticosteroids Favours no corticosteroids Evidence quality Events per 1000 people No important difference The panel found that these differences were not important for most patients, because the intervention effects were negligible and/or very imprecise (such as statistically not significant)

18 fewer Mortality Low More 254 236

Risk of Bias No serious concerns Imprecision Serious Indirectness No serious concerns Inconsistency Serious Publication bias No serious concerns Short term (28-31 days) Corticosteroids may achieve a small or no reduction in short term mortality

Neuromuscular weakness Low More 303 53 fewer 250

Risk of Bias No serious concerns Imprecision Serious Indirectness Serious Inconsistency Borderline Publication bias No serious concerns Corticosteroids may increase the risk of neuromuscular weakness

Quality of Life None More Unknown

Not reported in any of the included studies

Stroke Very low More 5 10 No important difference

Risk of Bias No serious concerns Imprecision Very serious Indirectness Serious Inconsistency No serious concerns Publication bias No serious concerns Whether or not corticosteroids impact the risk of stroke is uncertain

Myocardial infarction Very Low More 30 27 No important difference

Risk of Bias No serious concerns Imprecision Very serious Indirectness Serious Inconsistency No serious concerns Publication bias No serious concerns Whether or not corticosteroids impact the risk of myocardial infarction is uncertain
Mean number of days

Length of ICU stay Moderate More 13.1 0.7 fewer 12.4

Risk of Bias No serious concerns Imprecision Serious Indirectness No serious concerns Inconsistency No serious concerns Publication bias No serious concerns Corticosteroids probably achieve a small reduction in the duration of initial stay in an intensive care unit (ICU)

Length of hospital stay Moderate More 32.0 0.7 fewer 31.3

Risk of Bias No serious concerns Imprecision Serious Indirectness No serious concerns Inconsistency No serious concerns Publication bias No serious concerns Corticosteroids probably achieve a small reduction in the duration of hospitalization
See patient decision aids
See all outcomes
Those who place more value on avoiding functional deterioration and maximising quality of life than on avoiding death may be more likely to choose not to use corticosteroids Preferences and values Patients at greatest risk of death (such as those with shock, high qSOFA/SOFA scores) will probably have the greatest reduction in risk of death with corticosteroids Risk of death There are no clear differences in efficacy or adverse effects between different corticosteroids or corticosteroid combinations. Most studies used hydrocortisone Choice of corticosteroid Key practical issues No key practical issues Infusion or intermittent bolus dosing Monitoring for serum sodium, potassium, and glucose Corticosteroids No corticosteroids

©BMJ Publishing Group Limited.

Disclaimer: This infographic is not a validated clinical decision aid. This information is provided without any representations, conditions or warranties that it is accurate or up to date. BMJ and its licensors assume no responsibility for any aspect of treatment administered with the aid of this information. Any reliance placed on this information is strictly at the user's own risk. For the full disclaimer wording see BMJ's terms and conditions: https://www.bmj.com/company/legal-information/

Find recommendations, evidence summaries and consultation decision aids for use in your practice
  1. Francois Lamontagne, chair, critical care clinician1 2,
  2. Bram Rochwerg, critical care clinician3 4,
  3. Lyubov Lytvyn, patient partnership liaison4,
  4. Gordon H Guyatt, general internist4,
  5. Morten Hylander Møller, anesthesiologist and critical care clinician5,
  6. Djillali Annane, critical care clinician6,
  7. Michelle E Kho, physiotherapist7,
  8. Neill K J Adhikari, critical care clinician8 9,
  9. Flavia Machado, critical care clinician10 11,
  10. Per O Vandvik, general internist12 13,
  11. Peter Dodek, critical care clinician14,
  12. Rebecca Leboeuf, endocrinologist15,
  13. Matthias Briel, corticosteroid trialist4 16,
  14. Madiha Hashmi, critical care clinician17,
  15. Julie Camsooksai, critical care nurse18,
  16. Manu Shankar-Hari, critical care clinician19 20,
  17. Mahder Kinfe Baraki, anaesthesiologist and critical care clinician21,
  18. Karie Fugate, family caregiver of patient22,
  19. Shunjie Chua, family caregiver of patient23,
  20. Christophe Marti, critical care clinician24,
  21. Dian Cohen, patient25,
  22. Edouard Botton, patient26,
  23. Thomas Agoritsas, general internist4 27,
  24. Reed A C Siemieniuk, methods co-chair, general internist4
  1. 1Department of Medicine, Université de Sherbrooke, Sherbrooke, Canada
  2. 2Centre de recherche du CHU de Sherbrooke, Centre intégré universitaire de santé et de services sociaux - Estrie, Sherbrooke, Canada
  3. 3Department of Medicine, McMaster University, Hamilton, Canada
  4. 4Department of Health Research Methods, Evidence and Impact, McMaster University, Hamilton, Ontario, Canada
  5. 5Department of Intensive Care, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark
  6. 6Service de Médecine Intensive et Réanimation, Hôpital Raymond Poincaré, Garches, France
  7. 7School of Rehabilitation Science, McMaster University, Hamilton, Canada
  8. 8Department of Critical Care Medicine, Sunnybrook Health Sciences Centre, Toronto Canada
  9. 9Interdepartmental Division of Critical Care Medicine, University of Toronto, Toronto, Canada
  10. 10Federal University of Sao Paulo, Sao Paulo, Brazil
  11. 11Latin America Sepsis Institute, Sao Paulo, Brazil
  12. 12Institute of Health and Society, Faculty of Medicine, University of Oslo, Oslo, Norway
  13. 13Department of Medicine, Innlandet Hospital Trust-division, Gjøvik, Norway
  14. 14Center for Health Evaluation and Outcome Sciences and Division of Critical Care Medicine, St Paul's Hospital and University of British Columbia, Vancouver, Canada
  15. 15Department of Medicine, Centre Hospitalier de l’Université de Montréal, Montréal, Canada
  16. 16Basel Institute for Clinical Epidemiology and Biostatistics, Department of Clinical Research, University Hospital Basel, University of Basel, Basel, Switzerland
  17. 17Department of Anaesthesiology, Aga Khan University, Karachi, Pakistan
  18. 18Poole Hospital NHS Foundation Trust, Dorset, United Kingdom
  19. 19Guy’s and St Thomas’ NHS Foundation Trust, London, United Kingdom
  20. 20NIHR Clinician Scientist, School of Immunology & Microbial Sciences, Kings College London, United Kingdom
  21. 21Department of Anesthesiology, Addis Ababa University, Addis Ababa, Ethiopia
  22. 22Renton, USA
  23. 23Carevise Mexico, Mexico City, Mexico
  24. 24Division of General Internal Medicine, Rehabilitation and Geriatrics, University Hospitals of Geneva, Geneva, Switzerland
  25. 25Centre de santé de la vallée Massawippi, Ayer’s Cliff, Canada
  26. 26Comité stratégique patient-partenaire, Centre de recherche du CHU de Sherbrooke, Centre intégré universitaire de santé et de services sociaux - Estrie, Sherbrooke, Canada
  27. 27Division General Internal Medicine & Division of Clinical Epidemiology, University Hospitals of Geneva, Geneva, Switzerland
  1. Correspondence to: R A C Siemieniuk reed.siemieniuk{at}medportal.ca

What you need to know

  • Sepsis is a syndrome of life threatening infection with organ dysfunction, and most guidelines do not advise use of corticosteroids to treat it in the absence of refractory shock

  • Two new trials of corticosteroid treatment for sepsis came to differing conclusions

  • Corticosteroids may reduce the risk of death by a small amount and increase neuromuscular weakness by a small amount, but the evidence is not definitive

  • This guideline makes a weak recommendation for corticosteroids in patients with sepsis; both steroids and no steroids are reasonable management options

  • Fully informed patients who value avoiding death over quality of life and function would likely choose corticosteroids

Do corticosteroids reduce death or improve recovery in people with sepsis or septic shock? Our panel make a weak recommendation to give corticosteroids to people with all types and severity of sepsis, based on new evidence. Because we are not certain that they are beneficial, it is also reasonable not to prescribe them. Patients’ values and preferences may guide this decision-making process.

This rapid recommendation was triggered by two trials, with differing conclusions whose results might change practice:

  • ADRENAL (3658 patients who had septic shock) found no statistically significant difference in 90 day mortality between the hydrocortisone and placebo groups.1

  • APROCCHSS (1241 patients who had septic shock) found that hydrocortisone plus fludrocortisone reduced 90 day mortality.2

The trials are incorporated into a linked systematic review comparing corticosteroids with placebo.3 This BMJ Rapid Recommendation promptly and transparently translates this evidence using GRADE methodology for trustworthy guidelines. Sepsis is a life threatening organ dysfunction from infection. Currently most guidelines advise against giving corticosteroids in sepsis in the absence of refractory shock, but these guidelines have not taken into account the new evidence. We do not anticipate that new clinical trials will substantively alter the evidence suggesting a small but uncertain mortality reduction. Box 1 shows publications linked in this Rapid Recommendation package. The main infographic provides an overview of the absolute benefits and harms. The table at the end of the article shows any evidence that has emerged since the publication of this guideline.

Box 1

Linked articles in the BMJ Rapid Recommendation cluster

  • Lamontagne F, Rochwerg B, Lytvyn L, et al. Corticosteroid therapy for sepsis: a clinical practice guideline. BMJ 2018;362:k3284

    • Summary of the results from the Rapid Recommendation process

  • Rochwerg B, Oczkowski SJ, Siemieniuk RAC, et al. Corticosteroids in sepsis: an updated systematic review and meta-analysis. Crit Care Med 2018. doi:10.1097/CCM.00000000000032623

    • Review and meta-analysis of all available randomised trials that assessed corticosteroid therapy for sepsis

  • MAGICapp (https://app.magicapp.org/public/guideline/EZ1w8n)

    • Expanded version of the results with multilayered recommendations, evidence summaries, and decision aids for use on all devices

RETURN TO TEXT

Current understanding

Sepsis is life threatening organ dysfunction caused by a dysregulated host response to infection.4 In practice, a sepsis-related organ failure assessment (SOFA) score of ≥2 in patients with infections is sepsis (table 1).45 Worldwide, about 30 million people are hospitalised with sepsis every year and up to six million of them die.6

Table 1

Sepsis-related organ failure assessment (SOFA) score to help diagnose sepsis (adapted from Vincent et al5)*

View this table:

Clinicians typically manage sepsis with early, broad spectrum antibiotics. They may provide supportive treatment such as vasoactive drugs and mechanical ventilation. They track and adjust treatment based on clinical signs and laboratory data.7 Septic shock is the most severe form of sepsis. These patients experience profound circulatory, metabolic, and cellular abnormalities.48 They require vasopressors to maintain perfusion pressure and have elevated serum lactate concentrations despite adequate fluid repletion.

It is possible that corticosteroids help improve the dysregulated immune response caused by sepsis9 and increase blood pressure if it is low.10 Some clinicians have found this biological rationale, and results of early studies, compelling. Others disagree and do not use corticosteroids.11

Most professional organisations recommend against corticosteroid use in the absence of refractory shock.12Table 2 summarises current professional society guidelines.

Table 2

Current recommendations for corticosteroid therapy in patients with sepsis

View this table:

How this recommendation was created

Our international panel included sepsis survivors, family caregivers of patients who had sepsis, intensivists, internists, nurses, an endocrinologist, physiotherapists, trialists, and methodologists (see appendix 1 on bmj.com). They decided on the scope of the recommendation and the outcomes that are most important to patients. The panel judged death and quality of life to be the most important outcomes. Myocardial infarction, stroke, duration of stay in hospital and in the intensive care unit (ICU), superinfections, and neuromuscular weakness (such as ICU-acquired weakness) were also identified as important outcomes for patients.

Surrogate outcomes such as time to shock reversal, organ dysfunction measured by the sepsis-related organ failure assessment (SOFA) score, hyperglycaemia, and hypernatraemia were less important to the panel. This view is consistent with GRADE recommendations to focus on patient-important outcomes rather than surrogates.29

Subgroups of interest—The panel wanted to know whether the effect of corticosteroids differed in people with sepsis, compared with people who had septic shock, pneumonia, acute respiratory distress syndrome, or were at higher risk of death.330 They also wanted to know whether the type of corticosteroid or its dose influenced outcomes.

The panel met by videoconference to discuss the evidence and formulate a recommendation. No panel member had financial conflicts of interest; intellectual and professional conflicts were minimised and managed (see appendix 2 on bmj.com).

The panel requested a systematic review of randomised controlled trials on the impact of corticosteroid therapy for patients who have sepsis, including those who have septic shock.3 This review examines the two latest, as well as previous studies, on corticosteroids in sepsis. The aim was to resolve apparently conflicting evidence.

The panel followed the BMJ Rapid Recommendations procedures for creating a trustworthy recommendation,31 including using the GRADE approach to critically appraise the evidence and create recommendations (see appendix 3 on bmj.com).32 The panel considered the balance of benefits, harms, and burdens of corticosteroids, the quality of the evidence for each outcome, expected variations in patient values and preferences, and acceptability of corticosteroids.33 Determining patient values and preferences occurred before the panel received the results of the meta-analysis to reduce the risk that opinions regarding outcome importance will be data driven. According to the GRADE approach, recommendations can be strong or weak and for or against a course of action.33 High quality evidence of an effect on surrogate outcomes do not trigger strong recommendations.

The evidence

The linked systematic review identified 42 randomised controlled trials (RCTs) comparing corticosteroids with no corticosteroids (typically placebo).3 Figure 2 provides an overview of the trials and participants.

Fig 2
Fig 2

Characteristics of patients and trials included in systematic review of the use of corticosteroids for treating sepsis3 CAP = community acquired pneumonia. ARDS = acute respiratory distress syndrome.

The systematic review includes total of 10 194 patients who had sepsis. Of the 42 trials included, 24 restricted enrolment to patients who had septic shock. The typical patient was critically ill—a median of 32% of participants died within the first month. The most common sources of sepsis were pulmonary infections (median 44%) and abdominal infections (median 17%). Most of the RCTs used hydrocortisone alone (n=26), others used hydrocortisone plus fludrocortisone (n=2), methylprednisolone (n=6), prednisolone (n=3), or dexamethasone (n=3) (see fig 2). Although most of the clinical trials included patients who had septic shock, many included patients who did not (16 trials, 2241 patients). The linked systematic review provides detailed trial descriptions, including risk of bias assessments and patient characteristics.3

Subgroups of patients

Corticosteroids did not seem to be more or less effective in particular clinical subgroups, for example:

  • Septic shock

  • Pneumonia

  • Acute respiratory distress syndrome (ARDS)

  • Higher baseline risk of death

  • Different corticosteroid drugs (such as hydrocortisone, methylprednisolone)

  • Different corticosteroid doses

  • Different corticosteroid regimens (such as single agents or corticosteroid combinations such as hydrocortisone plus fludrocortisone)

  • More recent v older trials

  • Trials with higher v lower risk of bias.

Older studies tended to use much higher doses of corticosteroids for a shorter time than are typically used now; the pooled evidence from these older studies is imprecise (few events), and the linked meta-analysis was underpowered to detect important subgroup differences such as by dose. All tests for relative subgroup effects may be underpowered to detect true differences because the effect sizes are small, especially for mortality. Therefore, we cannot be certain that a true subgroup effect does not exist. Future meta-analyses of individual patient data may help to identify populations that benefit more or less from corticosteroids. Until such time, we can only conclude that the evidence applies to all subgroups.

Understanding the recommendation

The main infographic provides an overview including the benefits and harms, and our certainty in the evidence for each outcome.

Absolute benefits and harms

There was better survival in the group taking corticosteroids, but this was not certain. This drives the weak rather than strong recommendation.

Mortality

Corticosteroids may reduce mortality in the first month after admission to an intensive care unit (ICU) by approximately 2%. However, the panel had low certainty that this is true. The confidence interval crosses the line of no difference, and the results were inconsistent, with some RCTs showing a mortality reduction and others showing none.

The effect on longer term mortality (from 60 days to 1 year) was similar. Fewer studies reported this outcome, so, although the results were consistent in the RCTs that did report this outcome, the panel also had low certainty that corticosteroids reduce longer term mortality.

Quality of life

No RCT reported quality of life outcomes at any time point. The ADRENAL study investigators are collecting quality of life data at six months, but these data have not been published.17

Outcomes of some interest

Corticosteroids may reduce the length of ICU and hospital stay by less than a day each (moderate quality evidence). The impact of corticosteroids on other patient-important outcomes such as stroke and myocardial infarction was extremely uncertain. They may increase the risk of neuromuscular weakness by a small amount (low quality evidence from seven RCTs). Possible explanations include the toxic effects on nerve and muscle cells, and hyperglycaemia from corticosteroid use.18 Weakness may compromise patients’ ability to function independently19 and delay recovery.20

In two of the seven RCTs evaluating weakness, it was prospectively evaluated one month after enrolment.221 Evaluations of neuromuscular weakness, especially in RCTs that relied on investigator identification, were unreliable. The panel therefore believed that the RCTs probably underestimated the risk of neuromuscular weakness.

Outcomes of less importance

Corticosteroids probably increase the risk of hyperglycaemia and hypernatraemia. Corticosteroids probably improve organ function at day 7 and the chance of shock reversal at day 7.

Patient subgroups

Our recommendation applies to all patients with sepsis. There was no meaningful difference in the efficacy of corticosteroids in different groups of patients including those with septic shock, pneumonia, acute respiratory distress syndrome, or other sources of sepsis, or those who were sicker. However, the absolute reduction in mortality from corticosteroids will be greater in patients with a higher risk of death. The absolute harm (such as neuromuscular weakness) will also be greater in sicker patients.

The analysis of a subgroup effect showed no convincing evidence of such an effect. Based on published criteria for credible subgroup effects,22 in the absence of a subgroup effect, the interpretation of the effect of corticosteroids in any particular subgroup should be guided by the effect in the overall population of septic patients.

Several trials of corticosteroids for pneumonia or acute respiratory distress syndrome have enrolled patients who did not have sepsis; we did not consider these trials. Therefore, clinicians treating these conditions should also consider evidence2324 and guidelines12 applicable to patients who have pneumonia and acute respiratory distress syndrome.

Patient values and preferences

Fully informed patients who place a higher value on avoiding death than on quality of life and function would be more likely to choose to receive corticosteroids. We heard from our patient partners that most patients will want to reduce their risk of death, even if this reduction is small and uncertain. This view is consistent with the experiences of the rest of the panel. Most patients will likely be willing to accept a small increased risk of weakness.

Patients (or their care givers and surrogate decision-makers) will probably vary in how they would weigh the balance of expected desirable and undesirable consequences from corticosteroids. We assume that most patients want to avoid death and will value even a small, uncertain reduction in mortality. We judge that they will be less concerned about the possible increase in weakness among survivors. There is also likely to be a sizeable minority of patients who would place a large value on avoiding a very uncertain but possible decline in quality of life and functional abilities even at the cost of a small increase in risk of death.25 Shared decision making conversations about specific interventions in patients with sepsis may not always be feasible, and could delay care. However, clinicians should do their best to elicit each patient's values and preferences. For example, they could talk about the patient’s goals of care with the patient, their family, and friends.

Practical considerations

Figure 3 outlines the key practical issues for patients and clinicians discussing corticosteroid treatment for sepsis.

Fig 3
Fig 3

Practical issues about use of corticosteroids for treatment of sepsis

The optimal corticosteroid drug, dose, and duration of treatment are uncertain. Hydrocortisone was the most commonly used corticosteroid in the RCTs and is therefore a reasonable choice. Differences among corticosteroids, if they do exist, are probably small; dexamethasone, methylprednisolone, and prednisolone were also studied and produced similar results. Adding an agent that has additional mineralocorticoid activity, such as fludrocortisone, could be helpful, but that is highly speculative.

The typical hydrocortisone dose for an adult in the RCTs was 200-300 mg/day, given either as an infusion or as boluses every six hours.26 If an infusion is chosen, a bolus of 50-100 mg can be given before the infusion. In the RCTs the duration of treatment was typically 7-14 days, or less for those who were rapidly improving.

Inflammation may recur after discontinuing corticosteroid therapy,27 especially when it is stopped abruptly.28 Clinicians should carefully monitor all patients after discontinuing corticosteroids. In patients who deteriorate after stopping corticosteroids (such as development of shock or need for mechanical ventilation), reinitiating corticosteroid therapy could be helpful, although this is highly speculative. Whether corticosteroids should be tapered rather than stopped abruptly is unclear. Corticosteroid induced adrenal suppression is probably duration dependent, and so patients who receive longer courses of corticosteroids (such as >14 days) might be particularly likely to benefit from a taper before discontinuing and an evaluation of hypothalamo-pituitary-adrenal axis function if in doubt.12

Costs

Corticosteroids are typically inexpensive and widely available. The impact of corticosteroids on the overall costs to patients and to health systems is uncertain and would be driven mostly by ICU and hospital lengths of stay or prolonged periods of rehabilitation.

Future research

With the exception of the awaited analysis of quality of life in the ADRENAL trial, there are currently no planned or ongoing RCTs in patients who have sepsis that are likely to substantively change the overall effect estimates for the key outcomes. Given remaining uncertainty regarding the effect of corticosteroids in different subgroups, additional analyses of existing data to explore heterogeneity of treatment effects are logical next steps before more patients are enrolled in similar trials. Such work mandates individual patient-data meta-analyses that rely on investigators sharing the data from their RCTs and cooperation among research networks.

It is possible that additional adaptive RCTs could help to resolve remaining uncertainty. Key research questions to inform decision makers and future guidelines are:

  • What is the impact of corticosteroid therapy on quality of life in the short and long term?

  • What is the impact of corticosteroid therapy on functional recovery?

  • What is the impact of corticosteroid therapy on healthcare costs?

  • Are there subgroups of patients with sepsis who benefit more or less from corticosteroid therapy?

  • Are there differences between bolus and infusion dosing?

  • Does the addition of fludrocortisone improve outcomes?

Updates to this article

The final table shows evidence that has emerged since the publication of this article. As new evidence is published, a group will assess the new evidence and make a judgment on to what extent it is expected to alter the recommendation.

New evidence which has emerged after initial publication

View this table:

Footnotes

  • This BMJ Rapid Recommendation article is one of a series that provides clinicians with trustworthy recommendations for potentially practice changing evidence. BMJ Rapid Recommendations represent a collaborative effort between the MAGIC group (http://magicproject.org/) and The BMJ. A summary is offered here and the full version including decision aids is on the MAGICapp (https://app.magicapp.org), for all devices in multilayered formats. Those reading and using these recommendations should consider individual patient circumstances, and their values and preferences and may want to use consultation decision aids in MAGICapp to facilitate shared decision making with patients. We encourage adaptation and contextualisation of our recommendations to local or other contexts. Those considering use or adaptation of content may go to MAGICapp to link or extract its content or contact The BMJ for permission to reuse content in this article.

  • Competing interests: All authors have completed the BMJ Rapid Recommendations interests disclosure form, and a detailed description of all disclosures is reported in appendix 2 on bmj.com. As with all BMJ Rapid Recommendations, the executive team and The BMJ judged that no panel member had any financial conflict of interest. Professional and academic interests are minimised as much as possible, while maintaining necessary expertise on the panel to make fully informed decisions.

  • Funding: This guideline was not funded.

  • Transparency: R A C Siemieniuk affirms that the manuscript is an honest, accurate, and transparent account of the recommendation being reported; that no important aspects of the recommendation have been omitted; and that any discrepancies from the recommendation as planned (and, if relevant, registered) have been explained.

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

References