An initial literature search, using the broad search terms defined in Box 1, identified over 12,000 studies. These were screened for full manuscript review (n = 499), of which 486 were excluded (Fig. 1). An additional study was identified from secondary searches [15]. Of 14 studies included for final analysis, 4 were excluded due to: being not directly relevant (2); lacking full English translation (1); including only secondary data (1) - see details in Additional File 1. Results from ten finally selected guidelines, published between 2004 and 2017, are summarised below.
Quality Assessment
A summary of quality assessment scores is given in Table 1, with full details of consensus scoring in Additional Files. Two out of ten guidelines exceeded a score of 70% indicating highly rigorous and robust guideline development processes (NICE and Surviving Sepsis Campaign recommendations) which reflects the resources available to develop them [9, 16]. For each of the others, significant deficiency was noted in at least 2 domains, of which the most frequent concern (in 8 out of 10 of the guidelines) was the “Applicability” domain (whether consideration had been given to “the likely barriers and facilitators to implementation, strategies to improve uptake and resource implications of applying the guideline”) [14].
Table 1
Combined AGREE-II scores, by domain
| Cecconi (2014) | Dunser (2012) | Hollenberg (2004) | Misango (2017) | Moller (2016) | NICE (2016) | Perner (2016) | Reinhart (2010) | Rhodes (2016) | WHO (2011) |
Scope and Purpose | 61.1% | 72.2% | 75.0% | 66.7% | 91.7% | 94.4% | 88.9% | 41.7% | 88.9% | 66.7% |
Stakeholder Involvement | 41.7% | 55.6% | 25.0% | 47.2% | 44.4% | 69.4% | 41.7% | 36.1% | 77.8% | 61.1% |
Rigour of Development | 43.8% | 61.5% | 41.7% | 56.3% | 61.5% | 86.5% | 67.7% | 37.5% | 63.5% | 25.0% |
Clarity of Presentation | 86.1% | 86.1% | 91.7% | 91.7% | 88.9% | 88.9% | 91.7% | 69.4% | 94.4% | 80.6% |
Applicability | 8.3% | 43.8% | 12.5% | 27.1% | 6.3% | 68.8% | 12.5% | 12.5% | 54.2% | 37.5% |
Editorial Independence | 33.3% | 75.0% | 8.3% | 95.8% | 91.7% | 91.7% | 66.7% | 4.2% | 75.0% | 58.3% |
TOTAL | 44.2% | 63.4% | 42.4% | 59.4% | 59.8% | 83.0% | 60.5% | 34.8% | 72.1% | 47.5% |
Sepsis definitions
The definitions used for sepsis varied (see Table 2) and were not explicit in three guidelines [15, 17, 18]. One provided a definition which was abstract and not directly clinically applicable [19]. Three defined bespoke criteria for sepsis [16, 20, 21] and three employed international definitions with and without modification [9, 22, 23].
Table 2
Sepsis definitions used by guidelines
Guideline | Definition |
Dunser et al.[22] | Sepsis-2* modified to replace criteria based on white blood cell count with “malaise and/or apathy” |
Hollenberg et al.[20] | Haemodynamic support considered for hypoperfusion (Defined as: Systolic BP < 90 mmHg; MAP < 65 mmHg; fall of systolic BP > 40 mmHg; change in mental status; decrease in urine output; increased lactate) |
NICE[16] | Clinical suspicion of infection, with risk criteria for death (e.g. altered mental status; evidence of microvascular perfusion defect – mottled/delayed capillary refill time; high respiratory rate) |
Reinhart et al.[23] | Sepsis-2* |
Rhodes et al.[9] | Sepsis-3† |
WHO[21] | Severe sepsis/septic shock defined as: suspected infection plus hypotension (systolic BP < 90 mmHg) plus ≥ 1 of pulse > 100 bpm; respiratory rate > 24; temperature < 36 °C or > 38 °C |
Published definitions | |
*Sepsis-2 [28] | Sepsis: Proven or highly suspected infection plus presence of ≥ 2 of the following conditions: heart rate > 90 bpm; respiratory rate ≥ 20/min or PaCO2 < 32 mmHg; temperature < 36 °C or > 38 °C; white blood cell count < 4 × 106 or > 12 × 106 g/L or > 10% immature forms Severe sepsis: sepsis plus confusion; hypoxaemia or elevated lactate |
†Sepsis-3 [1] | Sepsis: Life-threatening organ dysfunction caused by a dysregulated host response to infection. Organ dysfunction: ≥2 points on qSOFA score; with 1 point scored for each of: respiratory rate > 22/min; altered mentation; systolic BP ≤ 100 mmHg |
Recommendations
We identified recommendations in the guidelines relating to indications for: 1) initiating fluid management; 2) choice of type and volume of fluid; 3) criteria for assessing response to fluid administration; 4) criteria for cessation; 5) indications and choice for initial vasopressor treatment.
Indications for intravenous fluid treatment
Two papers gave no specific recommendations on indications for intravenous fluid treatment, being practice guidelines for initial vasopressor therapy and choice of fluid respectively [17, 18]. Of the remaining eight guidelines, hypotension was a common indication, with specific criteria in terms of systolic blood pressure (BP) described in four [9, 16, 20, 24]. A systolic BP of 90 mmHg was given as the threshold in all other than the Surviving Sepsis guidelines, where 100 mmHg was used as per the qSOFA scoring system. Suspected hypovolemia was referenced at an unspecified threshold in a further two guidelines [15, 23]. One guideline recommended treatment for patients with shock and explicitly stated that hypotension was not required to make this diagnosis, which should be based on a constellation of clinical findings (not specifically described) and lactate (> 2 mmol/l) [19].
Clinical manifestations of hypoperfusion, representing indications for commencement of fluid therapy, were described in six guidelines [9, 15, 16, 20, 22, 24] either alone [15, 22] or in combination with blood pressure criteria [9, 16, 20, 24]. Of these, altered mental state was cited most frequently [9, 16, 20, 22, 24] with reduced urine output, increased respiratory rate, prolonged capillary refill time, cool peripheries and skin mottling also featuring across multiple guidelines.
Serum lactate was also recommended to identify those at high risk or need for circulatory support using a threshold of > 2 mmol in one study [16] and advocated without specific thresholds in two others [9, 20].
Fluid type, volume and rate
Preference for crystalloids for initial resuscitation was prevalent [9, 15, 16, 18, 22, 23, 25] and only one early guideline promoted colloid as equally or more effective [20]. In one of the guidelines tailored to management of sepsis in resource-limited settings [22], no specific recommendations were made regarding the relative efficacy of crystalloids or colloids, however the authors acknowledged that “considering high costs, the risk of allergies and potential renal and coagulopathic side effects of colloids, crystalloid solutions appear more suitable”. Three guidelines suggest consideration of human albumin solution as a second-line fluid choice in those patients with refractory shock or requiring large volumes of crystalloid solutions [9, 16, 23]. Five guidelines specifically recommend administering fluids using a fluid challenge technique during initial resuscitation, using boluses of between 250 and 1000 ml [16, 19, 20, 22, 24].
Specific recommendations on total initial volume of fluid for resuscitation included 30 ml/kg, “at least 20 ml/kg” and a note to aggressively treat with estimated 24-hour requirement of up to 4 litres [15, 22]. Hollenberg et al suggest 6–10 litres in the first day would be typical, titrated by fluid bolus [20]. WHO guidelines were more liberal: 1 litre as a bolus and up to 60/ml/kg in the first 2 hours [25]. More conservative were NICE who recommended 2 × 500 ml boluses rapidly, followed by senior review if no clinical improvement [16]. Two guidelines made no relevant volume recommendations, although titration to fluid challenges was promoted in one [17, 19].
Assessing response/Targets of resuscitation
Two guidelines made no specific recommendations on targets or response assessment [17, 18]. Lactate was promoted for assessment of adequate response in three documents, either alone [19], or in combination with clinical signs [16, 23]. Specific thresholds included a 20% reduction in serum lactate over the first hour [16] and either absolute values of ≤ 1.5 mmol/L or a decrease in non-specified time period [23].
Sequential evaluation of dynamic variables was promoted, including passive leg raise and cardiac ultrasound in ventilated patients [9, 15, 19]. Clinical measures of adequate tissue perfusion (capillary refill, skin temperature and degree of mottling, pulse, blood pressure and conscious level) were advocated in two guidelines specifically tailored to LIC [15, 22]. Earlier guidelines implied additional invasive monitoring (pulmonary artery occlusion pressure, titration to CVP and cardiac output) following a treat-reassess cycle with quarter-hourly boluses of 250–500 ml [20, 23]. SvO2 monitoring was advocated in two guidelines [19, 23]. The most recent guidelines noted the lack of evidence of improved outcomes related to CVP and SvO2 monitoring [9].
Criteria for termination
Three guidelines gave no specific indication of stop criteria.[16–18] Others suggested care in continuing therapy, but were not specific about objective criteria to guide the decision, for example “Fluid resuscitation should be stopped or interrupted when no improvement of tissue perfusion occurs in response to volume loading.”[22] and “possible repeat volume restitution is guided by the effects” [23]. Three guidelines specifically warned of the dangers of fluid overload or pulmonary oedema, with varying degrees of caution: “… even in the context of fluid-responsive patients, fluid management should be titrated carefully” [19], stressing the need for arterial oxygenation monitoring [20] and explicitly cautioning liberal fluids where there was no or limited access to vasopressors and mechanical ventilation [15].
Vasopressors
Three studies made no specific recommendations on choice of vasopressor [16, 18, 19]. Norepinephrine was identified as the preferred first line vasopressor therapy in 5 studies [9, 15, 17, 20, 23]. Two studies recommended dopamine or epinephrine [21, 22].
Starting criteria were not specified in one study [17]. In those guidelines that gave them, thresholds of arterial hypotension were the main indicators on which to base commencement of pressor support [9, 15, 21, 23], to be initiated after initial fluid management. Two other studies gave non-specific indications as “persistent tissue hypoperfusion” [22] and inadequate “arterial pressure and organ perfusion” [20]. Only the WHO guidelines specified a fluid volume trigger for consideration of pressors; 60 ml/kg within first 2 hours [21]. Once commenced, the most common target was arterial pressure of MAP 65 mmHg [9, 15, 23], with one guideline suggesting higher targets in chronic pre-existing hypertension [19]. Two studies specifically recommended administration via a central venous line using a syringe or infusion pump when available [9, 15].
Clinical applicability
Table 3 describes the applicability of each guideline against the pre-determined assessment scenarios (summarised in Fig. 2, full scenarios in Additional Files). Scenario A describes the initial resuscitation of a previously healthy adult patient with suspected infection and evidence of possible hypovolemia (tachycardia), poor peripheral perfusion (cool peripheries, prolonged capillary refill time) and end-organ dysfunction (altered mental status). In this scenario there is no serum lactate result available.
Table 3: Specific fluid therapy recommended in pre-described clinical case scenarios
All guidelines adopt a universal initial approach to fluids (do not take into consideration presenting co-morbidities).
Guideline | Scenario A | Scenario B | Scenario C |
| Shock and altered mental status | Non-response to initial management, high lactate | High lactate and likely congestive cardiac failure |
Cecconi (2014) [19] | Guideline on haemodynamic monitoring in circulatory shock, not specific to sepsis. Recommendations are given in general terms and are not directly applicable to the clinical scenarios. |
Dunser (2012)[22] | > 4L crystalloid in first 24 h. | No additional specific guidance. | Warning given regarding fluid overload. No fluid if not clinically hypo-perfused. |
Hollenberg (2004) [20] | 250–500 ml boluses over 15 min titrated to clinical endpoints and cardiac measures of fluid responsiveness. No ceiling given (liberal). | No additional specific guidance. | No additional specific guidance. |
Misango (2017)[15] | 30 ml/kg crystalloid over 3 hours, continue if fluid responsive. | Peripheral perfusion guided therapy. | Peripheral perfusion guided therapy. Clinical examination to detect overload. |
Moller (2016) [17] | Guideline on choice of first-line vasopressor, no specific recommendations relevant to the clinical scenarios. |
NICE (2016)[16] | No definitive guide without lactate. | 500 ml crystalloid over < 15 minutes. Seek senior help at 2L. | 500 ml bolus in response to high lactate, as in Scenario B. No specific guidance regarding fluid overload. |
Perner (2014) [18] | Guideline on choice of resuscitation fluid, general recommendation for use of crystalloid over other fluid types; no other specific recommendations relevant to the clinical scenarios. |
Reinhart (2010) [23] | 500–1000 ml crystalloid over 30 min | Repeat bolus according to response, central monitoring. Target lactate. | Continue and monitor central pressures |
Rhodes (2016)[9] | 30 ml/kg crystalloid over 3 hours | Repeat bolus according to response, including invasive and non-invasive monitoring. Target lactate. No volume ceiling given. | Clinical reassessment to detect pulmonary oedema |
WHO (2011) [24] | 1000 mL crystalloid immediately, continued at 20 ml/kg/hr (max 60 ml/kg in first 2 hours). | Between 2–6 hours, fluid at 5–10 ml/kg/hr if SBP < 90 and signs of poor perfusion continue. | Alert for signs of fluid overload (increased JVP, increasing crackles/rales): reduce rate if present. |
Scenario B describes the ongoing resuscitation of a previously healthy adult patient, presenting with symptoms of pneumonia and displaying definite evidence of hypovolemia (tachycardia, systolic hypotension < 90 mmHg) and tissue hypoperfusion (serum lactate > 4 mmol/L), demonstrating refractory haemodynamic instability following intravenous administration of 2500 ml of crystalloid.
Scenario C describes the initial resuscitation, as well as criteria for terminating fluid therapy, in an elderly patient with suspected infection and evidence of possible hypovolaemia (tachycardia, systolic blood pressure < 100 mmHg in the context of known hypertension), raised serum lactate and presenting comorbidities suggesting likely fluid intolerance (history of congestive cardiac failure).
Three guidelines gave recommendations that were insufficiently specific to apply to any of the pre-described clinical scenarios [17–19]. Of the remaining seven, all except the NICE guidelines recommended fluid resuscitation with crystalloid in Scenario A. The NICE guidelines do not provide a definitive guide to initiating therapy without a lactate measurement [16]. In Scenario B, following non-response to initial fluid resuscitation, two guidelines do not provide additional specific guidance, beyond their initial recommendations to resuscitate liberally with crystalloid [20, 22]. Of the five that do make recommendations, three recommend repeat boluses [9, 16, 23] whilst the WHO guideline recommends continuing infusion at 5–10 ml/kg/hr [24]. The Misango et al. guidelines for resource-limited settings provide a more general recommendation to continue fluid resuscitation to target clinical surrogate markers of peripheral perfusion [15]. Considered against Scenario C, designed to assess recommendations in the management of a patient with likely fluid intolerance, two guidelines do not provide any specific guidance regarding termination of fluid therapy in the context of clinical volume overload [16, 20]. Four of the remaining five guidelines recommend clinical re-assessment to detect fluid overload and/or pulmonary oedema [9, 15, 22, 24], with the WHO guideline recommending a reduction in the rate of fluid infusion if clinical signs of overload are present. The Reinhart et al. guideline recommends monitoring of central venous pressure alone to detect volume overload [23].
None of the guidelines suggested any refinement for patients with malnutrition.