Evidence
This page contains a snapshot of featured content which highlights evidence addressing key clinical questions including areas of uncertainty. Please see the main topic reference list for details of all sources underpinning this topic.
BMJ Best Practice evidence tables
Evidence tables provide easily navigated layers of evidence in the context of specific clinical questions, using GRADE and a BMJ Best Practice Effectiveness rating. Follow the links at the bottom of the table, which go to the related evidence score in the main topic text, providing additional context for the clinical question. Find out more about our evidence tables.
This table is a summary of the analysis reported in a Cochrane Clinical Answer that focuses on the above important clinical question.
Confidence in the evidence is very low or low where GRADE has been performed and the intervention may be less effective or likely to be more harmful than the comparison for key outcomes. However, this is uncertain and new evidence could change this in the future.
Population: Adults undergoing ERCP for pancreaticobiliary diseases
Intervention: Pancreatic duct guidewire or double guidewire technique in whom cannulation of the common bile duct had failed with conventional contrast, or guidewire-assisted cannulation techniques
Comparison: Other endoscopic techniques (persistent attempts with conventional cannulation technique, precut sphincterotomy, or pancreatic duct stent placement)
| Outcome | Effectiveness (BMJ rating)? | Confidence in evidence (GRADE)? |
|---|---|---|
Post‐ERCP pancreatitis (PEP) (24 hours after ERCP) | Favours comparison | Low |
Severity of PEP (24 hours after ERCP): moderate | No statistically significant difference | Low |
Severity of PEP (24 hours after ERCP): severe | No statistically significant difference | Low |
Overall cannulation success | No statistically significant difference | Low |
Post‐ERCP bleeding | No statistically significant difference | Low |
Post‐ERCP perforation | No statistically significant difference | Low |
Post‐ERCP cholangitis | No statistically significant difference | Low |
Mortality | No statistically significant difference | Low |
Quality of life | - | The reviewers did not assess this outcome |
Note The Cochrane review which underpins this Cochrane Clinical Answer (CCA) noted that in contrast to popular belief, pancreatic duct guidewire increases the risk of PEP and does not appear to improve cannulation success when compared with other endoscopic techniques. The CCA states that there doesn’t appear to be any justification for the use of pancreatic duct guidewire placement outside of clinical trials.
This evidence table is related to the following section/s:
This table is a summary of the analysis reported in a Cochrane Clinical Answer that focuses on the above important clinical question.
Confidence in the evidence is moderate or low to moderate where GRADE has been performed and there may be no difference in effectiveness between the intervention and comparison for key outcomes.
Population: Adults with clinical signs/symptoms suggestive of acute gallstone pancreatitis or confirmed acute gallstone pancreatitis ᵃ
Intervention: ERCP combined with conservative management (within 72 hours of admission)
Comparison: Early conservative management (within 30 days of admission)
| Outcome | Effectiveness (BMJ rating)? | Confidence in evidence (GRADE)? |
|---|---|---|
All-cause mortality during hospitalisation (up to 3 months) | No statistically significant difference ᵃ | Low |
Local complications defined by the Atlanta Classification (10 days to 3 months) | No statistically significant difference ᵃ | Moderate |
Systemic complications defined by the Atlanta Classification (10 days to 3 months) | No statistically significant difference ᵃ | Moderate |
ERCP-related complication: post-ERCP bleeding (time period unclear) | No statistically significant difference | GRADE assessment not performed for this outcome |
ERCP-related complications other than bleeding (time period unclear) | See note ᵇ | GRADE assessment not performed for this outcome |
ERCP-related mortality | - | The reviewers did not assess this outcome |
Note The Cochrane review which underpins this Cochrane Clinical Answer (CCA) states that the timing of ERCP (urgent <24 hours versus early <72 hours) should depend on the level of suspicion, the condition of the patient, and response to initial conservative management. The CCA also mentioned a possible benefit of early routine ERCP for patients with biliary obstruction (see the underlying Cochrane review for more details and subgroup analysis). ᵃ The CCA is for unselected patients with gallstone pancreatitis. However, it includes subgroup analyses selecting patients with concurrent cholangitis which found some benefit for early routine ERCP, with reduced mortality, local complications, and systemic complications compared with conservative management. The reviewers did not perform a GRADE assessment (see CCA for more details and subgroup analysis). ᵇ Results reported narratively (see the CCA for more details).
This evidence table is related to the following section/s:
This table is a summary of the analysis reported in a systematic review that focuses on the above important clinical question.
Confidence in the evidence is high or moderate to high where GRADE has been performed and there is no difference in effectiveness between the intervention and comparison for key outcomes.
Population: Critically ill adults in intensive care units or high dependency units
Intervention: Balanced crystalloid solution
Comparison: Normal saline
| Outcome | Effectiveness (BMJ rating)? | Confidence in evidence (GRADE)? |
|---|---|---|
All-cause mortality at 90 days ᵃ | No statistically significant difference ᵇ | High |
Incidence of acute kidney injury | No statistically significant difference | Moderate |
New treatment with renal replacement therapy | No statistically significant difference | Low |
Ventilator-free days (to day 28) | No statistically significant difference | Moderate |
Vasopressor-free days (to day 28) | No statistically significant difference | High |
Health-related quality of life | See note ᶜ | GRADE assessment not performed for this outcome |
Note The reviewers noted that the limitations of this review are mostly related to the characteristics of the included trials, which reported outcomes at different time points and used different definitions for outcome measures such as acute kidney injury. Furthermore, many of the studies did not include or did not report outcomes in subgroups of interest, resulting in limited power to detect clinically important subgroup effects. None of the included studies reported on longer-term quality of life or functional outcomes. ᵃ Primary outcome as stated in the systematic review underpinning this evidence table. ᵇ The estimated effect of balanced crystalloids versus normal saline ranges from a 9% relative reduction to a 1% relative increase in the risk of death; high probability that the average effect of using balanced crystalloids was to reduce mortality; RR 0.96 (95% CI 0.91 to 1.01). The reviewers noted that this estimate of effect was stable when including low risk-of-bias studies only or when including all studies regardless of risk of bias. The authors of the systematic review also noted that reduction in mortality was significant if a Bayesian approach was taken. ᶜ No data was available to provide a pooled estimate of quality of life.
This evidence table is related to the following section/s:
This table is a summary of the analysis reported in a guideline (underpinned by a systematic review) that focuses on the above important clinical question.
Confidence in the evidence is moderate or low to moderate where GRADE has been performed and the intervention may be more effective/beneficial than the comparison for key outcomes.
Population: People admitted to hospital (secondary care or tertiary care) with acute pancreatitis and infected or suspected infected necrosis
Intervention: Minimally invasive surgery (percutaneous or endoscopic transgastric drainage)
Comparison: Open surgery
| Outcome | Effectiveness (BMJ rating)? | Confidence in evidence (GRADE)? |
|---|---|---|
Mortality (follow-up during admission) | No statistically significant difference | Low |
Enterocutaneous fistula or perforation of a visceral organ requiring intervention (follow-up during admission) | No statistically significant difference | Low |
Intra-abdominal bleeding (follow-up during admission) | No statistically significant difference | Low |
Multiple organ failure (follow-up during admission) | Favours intervention | High |
Multiple systemic complications (follow-up during admission) | No statistically significant difference | Moderate |
New-onset multiple organ failure (follow-up during admission) | Favours intervention | High |
New-onset diabetes (follow-up during admission) | Favours intervention | Moderate |
Use of pancreatic enzymes (follow-up during admission) | Favours intervention | High |
Recommendations as stated in the source guideline Offer people with acute pancreatitis an endoscopic approach for managing infected or suspected infected pancreatic necrosis when anatomically possible. Offer a percutaneous approach when an endoscopic approach is not anatomically possible.
Note The NICE guideline committee also identified one individual patient data meta-analysis of non-randomised trials that supported the RCT data by finding a clinically important benefit of minimally invasive procedures (either endoscopic or percutaneous) for mortality, especially in patients at high baseline risk of death. The committee also looked at non-randomised studies comparing different surgical approaches. They considered conducting their own meta-analysis but decided against it based on the heterogeneity in both interventions and comparisons. Overall, they concluded that there was enough evidence to recommend the use of minimally invasive approaches for the management of necrosis. The guideline committee discussed that the choice of endoscopic versus percutaneous approach is governed by the anatomy of the necrosis and its relationship to the stomach, with only about one third of patients suitable for both approaches. Therefore, they considered that centres offering minimally invasive procedures for the management of necrosis should offer both procedures. The committee’s decision to recommend the endoscopic approach as the first choice reflected the evidence and patient experience. Endoscopic interventions are associated with a shorter length of hospital stay, fewer complications, and greater patient acceptability, whereas percutaneous techniques involve ongoing drainage which can leak, cause pain, and require regular flushing.
This evidence table is related to the following section/s:
This table is a summary of the analysis reported in a Cochrane Clinical Answer that focuses on the above important clinical question.
Confidence in the evidence is very low or low where GRADE has been performed and there may be no difference in effectiveness between the intervention and comparison for key outcomes. However, this is uncertain and new evidence could change this in the future.
Population: Adults with acute pancreatitis ᵃ
Intervention: Opioids ᵇ
Comparison: Non-opioid analgesics ᵇ
| Outcome | Effectiveness (BMJ rating)? | Confidence in evidence (GRADE)? |
|---|---|---|
Number of participants showing improvements in pain intensity at 2 days | No statistically significant difference | Low |
Number of participants with a supplementary analgesic option offered at 2-4 days | No statistically significant difference ᶜ | Low |
Number of participants with pancreatitis complications at 2-4 days | No statistically significant difference | Low |
Number of participants with drug-related adverse events at 2-3 days | No statistically significant difference | Low |
Number of deaths from any cause | No statistically significant difference | Very Low |
Note The Cochrane Clinical Answer (CCA) notes that the majority of included studies were underpowered. ᵃ Sixty-five percent men. ᵇ The included studies used the following opioids: intravenous buprenorphine, subcutaneous morphine, or intravenous pentazocine. These were compared with intravenous non-opioid analgesics, procaine, or metamizole. ᶜ Although not statistically significant, fewer people required supplementary analgesia with opioid analgesia (30 per 100 people) compared with non-opioid analgesia (73 per 100 people). See CCA for more details.
This evidence table is related to the following section/s:
This table is a summary of the analysis reported in a guideline (underpinned by a systematic review) that focuses on the above important clinical question.
Confidence in the evidence is moderate or low to moderate where GRADE has been performed and the intervention may be more effective/beneficial than the comparison for key outcomes.
Population: People (adults and young people >16 years; children ≤16 years) with severe or moderately severe acute pancreatitis admitted to hospital
Intervention: Early reinstitution of oral feeding (based on hunger)
Comparison: Delayed reinstitution of oral feeding (recommenced once abdominal pain resolved and biochemical markers normalised)
| Outcome | Effectiveness (BMJ rating)? | Confidence in evidence (GRADE)? |
|---|---|---|
Length of hospital stay (follow-up: unclear; better indicated by lower values) | Favours intervention | Low |
Requiring parenteral nutrition (follow-up: unclear) | No statistically significant difference | Moderate |
Adverse events (abdominal pain relapse; follow-up: unclear) | No statistically significant difference | Very Low |
Recommendations as stated in the source guideline Ensure that people with acute pancreatitis are not made ‘nil-by-mouth’ and do not have food withheld unless there is a clear reason for this (e.g., vomiting).
Note The guideline committee noted that managing people with acute pancreatitis as ‘nil-by-mouth’ initially has no consensus in terms of clinical opinion or on an ethical basis, as this is no longer believed to be beneficial.
This evidence table is related to the following section/s:
This table is a summary of the analysis reported in a guideline (underpinned by a systematic review) that focuses on the above important clinical question.
Confidence in the evidence is moderate or low to moderate where GRADE has been performed and there is a trade-off between benefits and harms of the intervention.
Population: People admitted to hospital (secondary care or tertiary care) with acute pancreatitis and infected or suspected infected necrosis
Intervention: Endoscopic approach ᵃ
Comparison: Percutaneous approach ᵃ
| Outcome | Effectiveness (BMJ rating)? | Confidence in evidence (GRADE)? |
|---|---|---|
Mortality (follow-up: 6 months) | No statistically significant difference ᵇ | Low |
Length of hospital stay (follow-up: 6 months; better indicated by lower values) | No statistically significant difference ᶜ | Moderate |
Complications: bleeding requiring reintervention (follow-up: 6 months) | No statistically significant difference | Low |
Complications: new-onset multiple organ failure (follow-up: 6 months) | No statistically significant difference | Low |
Complications: new-onset single organ failure (follow-up: 6 months) | No statistically significant difference | Moderate |
Complications: pancreatic fistula (follow-up: 6 months) | Favours intervention | High |
Complications: perforation of visceral organ or enterocutaneous fistula requiring intervention (follow-up: 6 months) | No statistically significant difference | Low |
Pancreatic function: endocrine insufficiency (follow-up: 6 months) | No statistically significant difference | Low |
Pancreatic function: exocrine insufficiency (follow-up: 6 months) | No statistically significant difference | Low |
Recommendations as stated in the source guideline Offer people with acute pancreatitis an endoscopic approach for managing infected or suspected infected pancreatic necrosis when anatomically possible. Offer a percutaneous approach when an endoscopic approach is not anatomically possible.
Note The NICE guideline committee noted that the single randomised study evaluating these interventions included pigtail stents and nasocystic drain, which have been superseded by self-expanding metal stents in UK practice. Therefore, the current UK endoscopic approach is likely to be more effective than the endoscopic approach included in the evaluated RCT, albeit more expensive. The guideline committee discussed that the choice of endoscopic versus percutaneous approach is governed by the anatomy of the necrosis and its relationship to the stomach, with only about one third of patients suitable for both approaches. Therefore, they considered that centres offering minimally invasive procedures for the management of necrosis should offer both procedures. The committee’s decision to recommend the endoscopic approach as the first choice reflected the evidence and patient experience. Endoscopic interventions are associated with a shorter length of hospital stay, fewer complications, and greater patient acceptability, whereas percutaneous techniques involve ongoing drainage which can leak, cause pain, and require regular flushing. ᵃ NICE identified one RCT comparing an endoscopic approach with a percutaneous approach. In the study, the intervention was endoscopic ultrasound-guided transluminal drainage followed by, if necessary, endoscopic necrosectomy; the comparison was percutaneous catheter drainage followed by, if necessary, video-assisted retroperitoneal debridement. ᵇ The NICE guideline committee noted that the endoscopic step-up approach was apparently associated with a non-statistically significant increase in mortality compared with the percutaneous step-up approach (absolute difference 49 more per 1000 [from 60 fewer to 332 more]). However, they did not feel this translated to a true clinical difference due to uncertainty caused by the small sample size and low event rate. ᶜ The mean length of hospital stay in the intervention group was 16 days lower (mean difference 32.86 days lower to 0.86 days higher).
This evidence table is related to the following section/s:
This table is a summary of the analysis reported in a guideline (underpinned by a systematic review) that focuses on the above important clinical question.
Confidence in the evidence is moderate or low to moderate where GRADE has been performed and there may be no difference in effectiveness between the intervention and comparison for key outcomes.
Population: People admitted to hospital with acute pancreatitis
Intervention: Any antimicrobial therapy administered prophylactically, including antifungals
Comparison: No prophylactic antimicrobial therapy or placebo
| Outcome | Effectiveness (BMJ rating)? | Confidence in evidence (GRADE)? |
|---|---|---|
Antimicrobial prophylaxis versus no therapy | ||
Mortality (follow-up: 1–6 weeks) | Favours intervention | High |
Mortality (selective decontamination) | No statistically significant difference | Moderate |
Length of hospital stay (follow-up: 10 days; better indicated by lower values) | No statistically significant difference | Very Low |
Infected necrosis (follow-up: 1–6 weeks) | Favours intervention | Moderate |
Infected necrosis (selective decontamination) | Favours intervention | Moderate |
Infected necrosis (peri-pancreatic infection; follow-up: 5–14 days) | No statistically significant difference | Very Low |
Extra-pancreatic infection (follow-up: 1–6 weeks) | No statistically significant difference | Very Low |
Extra-pancreatic infection (blood-culture-positive sepsis; follow-up: 14 days) | No statistically significant difference | Very Low |
Extra-pancreatic infection (pneumonia/ARDS; follow-up: 14 days) | No statistically significant difference | Low |
Extra-pancreatic infection (urinary tract infection; follow-up: 14 days) | Favours intervention | Low |
Serious adverse events (multiorgan failure; follow-up: 1–6 weeks) | No statistically significant difference | Moderate |
Serious adverse events (major organ complications) <6 months | No statistically significant difference | Very Low |
Antimicrobial prophylaxis versus placebo | ||
Mortality (follow-up: 10–42 days) | No statistically significant difference | Very Low |
Infected necrosis (follow-up: 10–42 days) | No statistically significant difference | Very Low |
Extra-pancreatic infection (follow-up: 10–42 days) | No statistically significant difference | Moderate |
Serious adverse events <6 months (follow-up: 42 days) | No statistically significant difference | Very Low |
Serious adverse events (pulmonary insufficiency; follow-up: 21 days) | No statistically significant difference | Very Low |
Serious adverse events (renal insufficiency; follow-up: 21 days) | No statistically significant difference | Very Low |
Serious adverse events (shock; follow-up: 21 days) | No statistically significant difference | Very Low |
Serious adverse events (SIRS; follow-up: 21 days) | No statistically significant difference | Very Low |
Serious adverse event (multiorgan failure) | No statistically significant difference | Very Low |
Colonisation by resistant organism <6 months (follow-up: 42 days) | No statistically significant difference | Very Low |
Recommendations as stated in the source guideline Do not offer prophylactic antimicrobials to people with acute pancreatitis.
Note The NICE guideline committee noted that the evidence for prophylactic antimicrobial therapy in people with acute pancreatitis was mixed, and that there was no clear demonstration of benefit or harm. The committee also noted inconsistencies in the evidence, in particular that the decrease in mortality and in infected necrosis seen with antimicrobial therapy compared with no treatment was not confirmed when it was compared with placebo. The committee also agreed that it was important to align practice recommendations with the general principle of antimicrobial stewardship. The guideline committee considered the following outcomes as being critical for decision-making: quality of life, mortality, length of stay (in CCU or hospital), and infected necrosis. Quality of life is not included in this table as no evidence was identified for this outcome. The guideline committee also evaluated the effectiveness of prophylactic antimicrobial therapy versus other prophylactic antimicrobial therapy (both for same class and different class).
This evidence table is related to the following section/s:
This table is a summary of the analysis reported in a guideline (underpinned by a systematic review) that focuses on the above important clinical question.
Confidence in the evidence is moderate or low to moderate where GRADE has been performed and the intervention may be more effective/beneficial than the comparison for key outcomes.
Population: Patients with severe or moderately severe acute pancreatitis admitted to hospital
Intervention: Enteral nutrition
Comparison: Parenteral nutrition
| Outcome | Effectiveness (BMJ rating)? | Confidence in evidence (GRADE)? |
|---|---|---|
Enteral (jejunal or duodenal) versus parenteral nutrition | ||
Mortality (follow-up during hospitalisation) | Favours intervention | Moderate |
Length of hospital stay: overall (follow-up during hospitalisation; better indicated by lower values) | No statistically significant difference | Low |
Length of hospital stay: patients with severe acute pancreatitis (Ranson's criteria >3; follow-up during hospitalisation; better indicated by lower values) | Favours intervention | Moderate |
Achieving nutrition: kcal/kg/day (day 5; follow-up during hospitalisation; better indicated by lower values) | No statistically significant difference | Low |
Achieving nutrition: days to goal (follow-up during hospitalisation; better indicated by lower values) | No statistically significant difference | Moderate |
Infections: pancreatic (e.g., infected necrosis, abscess; follow-up during hospitalisation) | Favours intervention | Moderate |
Infections: extra-pancreatic (e.g., UTI, pneumonia; follow-up during hospitalisation) | No statistically significant difference | Very Low |
Infections: systemic (e.g., central line infection, blood culture; follow-up during hospitalisation) | Favours intervention | Moderate |
Infections: type not specified (follow-up during hospitalisation) | No statistically significant difference | Very Low |
Serious adverse events (follow-up during hospitalisation) | Occurs more commonly with parenteral nutrition compared with enteral nutrition (favours intervention) | Very Low |
Adverse events: operative intervention (follow-up during hospitalisation) | Occurs more commonly with parenteral nutrition compared with enteral nutrition (favours intervention) | Low |
Adverse events: non-infective pancreatic complications (e.g., necrosis, pseudocyst, fistulae; follow-up during hospitalisation) | No statistically significant difference | Very Low |
Adverse events: feeding complications (e.g., tube displacement, hyperglycaemia, diabetes; follow-up during hospitalisation) | No statistically significant difference | Very Low |
Enteral (gastric) versus parenteral nutrition ᵃ | ||
Mortality (follow-up: 3 months) | No statistically significant difference | Very Low |
Achieving nutrition (25 kcal/kg/day; follow-up: 10 days) | No statistically significant difference | Very Low |
Infections: pancreatic (e.g., infected necrosis, abscess; follow-up: 3 months) | No statistically significant difference | Very Low |
Infections: systemic (e.g., central line infection, blood culture; follow-up: 3 months) | No statistically significant difference | Very Low |
Serious adverse events: multiple- or single-organ failure (follow-up: 3 months) | No statistically significant difference | Very Low |
Adverse events: general (e.g., pleural effusion; follow-up: 3 months) | No statistically significant difference | Very Low |
Adverse events: non-infective pancreatic complications (e.g., necrosis, pseudocyst, fistulae; follow-up: 3 months) | No statistically significant difference | Very Low |
Adverse events: surgical intervention (follow-up: 3 months) | No statistically significant difference | Very Low |
Recommendations as stated in the source guideline Offer enteral nutrition to anyone with severe or moderately severe acute pancreatitis. Start within 72 hours of presentation and aim to meet their nutritional requirements as soon as possible. Offer anyone with severe or moderately severe acute pancreatitis parenteral nutrition only if enteral nutrition has failed or is contraindicated.
Note The overall evidence rating in this table is based upon evidence underpinning critical outcomes as listed in the main guideline (mortality; length of hospital stay; achieving nutrition). The remaining outcomes in this table have been classified as important by the guideline reviewers. The NICE guideline committee also evaluated the effectiveness of different routes of enteral nutrition (gastric versus jejunal or duodenal nutrition) and found no statistically significant difference between treatment groups for all outcomes. ᵃ Only one small study (50 participants) was identified for this comparison. This is reflected in the very low confidence in the evidence for all outcomes, and the overall rating in this table which is therefore based upon the results for the comparison of enteral (jejunal or duodenal) versus parenteral nutrition (as direct comparison found no statistically significant difference between different routes of enteral nutrition).
This evidence table is related to the following section/s:
Cochrane Clinical Answers
Cochrane Clinical Answers (CCAs) provide a readable, digestible, clinically focused entry point to rigorous research from Cochrane systematic reviews. They are designed to be actionable and to inform decision making at the point of care and have been added to relevant sections of the main Best Practice text.
- How does early routine endoscopic retrograde cholangiopancreatography compare with early conservative management in people with acute gallstone pancreatitis?
- How do opioids compare with non-opioid analgesics for the management of acute pancreatitis pain?
- Can the pancreatic duct guidewire technique for biliary cannulation help to prevent ERCP pancreatitis?
Use of this content is subject to our disclaimer