Scolaris Content Display Scolaris Content Display

درناژ پروفیلاکتیک شکمی در جراحی پانکراس

Contraer todo Desplegar todo

چکیده

پیشینه

استفاده از درن‌های (drain) جراحی یک رویکرد بسیار رایج پس از جراحی پانکراس است. نقش درناژ‌های پروفیلاکتیک شکمی (prophylactic abdominal drainage) در کاهش عوارض پس از جراحی پس از انجام جراحی پانکراس بحث‌برانگیز است. این سومین به‌روزرسانی از مرور کاکرین است که قبلا منتشر شده و به مزایای نامشخص درناژ پیشگیرانه شکم در جراحی پانکراس پرداخت.

اهداف

ارزیابی مزایا و مضرات درناژ روتین شکمی پس از جراحی پانکراس، مقایسه تاثیرات انواع مختلف درن‌های جراحی، و ارزیابی زمان بهینه برای خارج کردن درن.

روش‌های جست‌وجو

در این مرور به‌روز شده، CENTRAL؛ MEDLINE؛ Embase؛ Science Citation Index Expanded، و پایگاه اطلاعاتی منابع علمی بیومدیکال چین (Chinese Biomedical Literature Database; CBM) را در 08 فوریه 2021 دوباره جست‌وجو کردیم.

معیارهای انتخاب

تمامی کارآزمایی‌های تصادفی‌سازی و کنترل شده‌ای را وارد مرور کردیم که به مقایسه درناژ شکمی در برابر عدم‐استفاده از آن در افراد تحت جراحی پانکراس پرداختند. هم‌چنین مطالعات تصادفی‌سازی و کنترل شده‌ای را وارد مرور کردیم که انواع مختلف درن‌ها و برنامه‌های مختلف را برای خارج کردن درن در افراد تحت جراحی پانکراس مقایسه کردند.

گردآوری و تجزیه‌وتحلیل داده‌ها

دو نویسنده مرور به‌طور مستقل از هم به شناسایی مطالعات برای ورود به مرور، گردآوری داده‌ها، و بررسی خطر سوگیری (bias) پرداختند. با استفاده از Review Manager 5 نسبت به انجام متاآنالیز اقدام کردیم. خطر نسبی (RR) را برای پیامدهای دو‐حالتی و تفاوت میانگین (MD) یا تفاوت میانگین استاندارد شده (SMD) را با 95% فواصل اطمینان (CI) برای پیامدهای پیوسته محاسبه کردیم. برای تمامی تجزیه‌و‌تحلیل‌ها، از مدل اثرات‐تصادفی (random‐effects model) استفاده شد. از سیستم درجه‌‏بندی توصیه‏، ارزیابی، توسعه و ارزشیابی (GRADE) برای ارزیابی کیفیت شواهد پیامدهای مهم استفاده کردیم.

نتایج اصلی

در مجموع، نه RCT را با مجموع 1892 شرکت‌کننده شناسایی کردیم.

استفاده از درن در برابر عدم‐استفاده از آن

چهار RCT را شامل 1110 شرکت‌کننده وارد مرور کردیم، که پس از جراحی پانکراس به صورت تصادفی بین گروه درناژ (N = 560) و عدم‐استفاده از درناژ (N = 550) تقسیم شدند. شواهدی با قطعیت پائین نشان می‌دهد که استفاده از درن ممکن است مورتالیتی 90‐روز را کاهش دهد (RR: 0.23؛ 95% CI؛ 0.06 تا 0.90؛ دو مطالعه، 478 شرکت‌کننده). شواهدی با قطعیت پائین حاکی از آن است که استفاده از درن در مقایسه با عدم‐استفاده از آن ممکن است منجر به تفاوتی اندک تا عدم تفاوت در مورتالیتی 30‐روز (RR: 0.78؛ 95% CI؛ 0.31 تا 1.99؛ چهار مطالعه، 1055 شرکت‌کننده)، نرخ عفونت زخم (RR: 0.98؛ 95% CI؛ 0.68 تا 1.41؛ چهار مطالعه، 1055 شرکت‌کننده)، طول مدت بستری در بیمارستان (MD؛ 0.14‐ روز؛ 95% CI؛ 0.79‐ تا 0.51؛ سه مطالعه، 876 شرکت‌کننده)، نیاز به دریافت پروسیجرهای باز بیشتر برای مدیریت عوارض پس از جراحی (RR: 1.33؛ 95% CI؛ 0.79 تا 2.23؛ چهار مطالعه، 1055 شرکت‌کننده) و کیفیت زندگی (105 امتیاز در مقابل 104 امتیاز؛ اندازه‌گیری شده با پرسشنامه کیفیت زندگی ویژه پانکراس (مقیاس 0 تا 144، مقادیر بالاتر نشان‌دهنده کیفیت بهتر زندگی)؛ یک مطالعه، 399 شرکت‌کننده) شود. هیچ گونه عارضه مرتبط با درن در گروه درناژ رخ نداد (0.2%). شواهدی با قطعیت متوسط نشان می‌دهد که استفاده از درن احتمالا منجر به تفاوتی اندک تا عدم تفاوت در موربیدیتی می‌شود (RR: 1.03؛ 95% CI؛ 0.94 تا 1.13؛ چهار مطالعه، 1055 شرکت‌کننده). شواهد پیرامون تاثیر استفاده از درن بر نرخ عفونت داخل‐شکمی (RR: 0.97؛ 95% CI؛ 0.52 تا 1.80؛ چهار مطالعه، 1055 شرکت‌کننده؛ شواهد با کیفیت بسیار پائین)، یا نیاز به دریافت مداخلات رادیولوژیکی بیشتر برای مدیریت عوارض پس از جراحی (RR: 0.87؛ 95% CI؛ 0.40 تا 1.87؛ سه مطالعه، 660 شرکت‌کننده؛ شواهد با کیفیت بسیار پائین) بسیار نامطمئن است.

درن فعال در مقابل غیر‐فعال

دو RCT را شامل 383 شرکت‌کننده وارد مرور کردیم، که پس از جراحی پانکراس به صورت تصادفی در گروه درن فعال (N = 194) و درن غیر‐فعال (N = 189) قرار داده شدند. شواهد در مورد تاثیر درن فعال در مقایسه با درن غیر‐فعال بر مورتالیتی 30‐روز (RR: 1.23؛ 95% CI؛ 0.30 تا 5.06؛ دو مطالعه، 382 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین)، نرخ عفونت داخل‐شکمی (RR: 0.87؛ 95% CI؛ 0.21 تا 3.66؛ دو مطالعه، 321 شرکت‌کننده، شواهد با قطعیت بسیار پائین)، نرخ عفونت زخم (RR: 0.92؛ 95% CI؛ 0.44 تا 1.90؛ دو مطالعه، 321 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین)، موربیدیتی (RR: 0.97؛ 95% CI؛ 0.53 تا 1.77؛ دو مطالعه، 382 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین)، مدت بستری در بیمارستان (MD؛ 0.79‐ روز؛ 95% CI؛ 2.63‐ تا 1.04؛ دو مطالعه، 321 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین)، و نیاز به دریافت پروسیجرهای باز بیشتر برای مدیریت عوارض پس از جراحی (RR: 0.44؛ 95% CI؛ 0.11 تا 1.83؛ دو مطالعه، 321 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین) بسیار نامطمئن بود. هیچ عارضه مربوط به درن در هر دو گروه وجود نداشت.

خارج کردن زودهنگام درن در مقابل دیرهنگام درن

سه RCT شامل 399 شرکت‌کننده با خطر پائین بروز فیستول پانکراس پس از جراحی پانکراس وارد شدند که بیماران را به صورت تصادفی در گروه خروج زودهنگام درن (N =200) و گروه خروج دیرهنگام درن (N = 199) قرار دادند. شواهد در مورد تاثیر برداشتن زودهنگام درن در مقایسه با خروج دیرهنگام آن بر مورتالیتی 30‐روز (RR: 0.99؛ 95% CI؛ 0.06 تا 15.45؛ سه مطالعه، 399 شرکت‌کننده، شواهد با قطعیت بسیار پائین)، نرخ عفونت زخم (RR: 1.32؛ 95% CI؛ 0.45 تا 3.85؛ دو مطالعه، 285 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین)، هزینه‌های بیمارستان (SMD: ‐0.22؛ 95% CI؛ 0.59‐ تا 0.14؛ دو مطالعه، 258 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین)، نیاز به دریافت روش‌های باز بیشتر برای مدیریت عوارض پس از جراحی (RR: 0.77؛ 95% CI؛ 0.28 تا 2.10؛ سه مطالعه، 399 شرکت‌کننده، شواهد با قطعیت بسیار پائین)، و نیاز به دریافت روش‌های رادیولوژیکی بیشتر برای مدیریت عوارض پس از جراحی (RR: 1.00؛ 95% CI؛ 0.21 تا 4.79؛ یک مطالعه، 144 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین) بسیار نامطمئن بود. ما دریافتیم که خروج زودهنگام درن ممکن است نرخ عفونت داخل شکمی (RR: 0.44؛ 95% CI؛ 0.22 تا 0.89؛ دو مطالعه، 285 شرکت‌کننده، شواهد با قطعیت بسیار پائین)، موربیدیتی (RR: 0.49؛ 95% CI؛ 0.30 تا 0.81؛ دو مطالعه، 258 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین)، و طول مدت بستری در بیمارستان (MD؛ 2.20‐ روز؛ 95% CI؛ 3.52‐ تا 0.87‐؛ سه مطالعه، 399 شرکت‌کننده؛ شواهد با قطعیت بسیار پائین) را کاهش دهد، اما شواهد بسیار نامطمئن بود. هیچ یک از مطالعات بروز عوارض مرتبط با درن را گزارش نکردند.

نتیجه‌گیری‌های نویسندگان

مشخص نیست استفاده روتین از درن در مقایسه با عدم‐استفاده از آن، بر مورتالیتی 30 روز یا عوارض پس از جراحی پانکراس تاثیری دارد یا خیر. شواهدی با قطعیت پائین نشان می‌دهد که استفاده روتین از درن در مقایسه با عدم‐استفاده از آن، ممکن است مورتالیتی را در 90 روز کاهش دهد. شواهد در مورد تاثیر استفاده از درن فعال در مقایسه با درن غیر‐فعال، بر مورتالیتی 30 روز یا عوارض پس از جراحی بسیار نامشخص است. درآوردن زودهنگام درن در مقایسه با خارج کردن دیرهنگام آن، ممکن است نرخ عفونت داخل شکمی، موربیدیتی، و طول مدت بستری در بیمارستان را برای افراد با خطر پائین فیستول پانکراس پس از جراحی کاهش دهد، اما شواهد بسیار نامشخص است.

PICO

Population
Intervention
Comparison
Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

خلاصه به زبان ساده

استفاده از درن پس از جراحی پانکراس

سوال مطالعه مروری

آیا استفاده از درن (drain) می‌تواند عوارض پس از جراحی را پس از انجام جراحی پانکراس کاهش دهد؟

پیشینه

استفاده از درن‌های (drain) جراحی یک رویکرد بسیار رایج پس از جراحی پانکراس است. با وجود این، نقش درن در کاهش عوارض پس از جراحی پانکراس (تحت عنوان عوارض پس از جراحی) بحث‌برانگیز است.

تاریخ جست‌وجو

شواهد تا فوریه 2021 به‌روز است.

ویژگی‌های مطالعه

تا فوریه 2021 برای یافتن همه مطالعات مرتبط و با روش انجام خوب، به جست‌وجو پرداختیم. نه مطالعه تصادفی‌سازی و کنترل شده (آزمایشی که در آن شرکت‌کنندگان به صورت تصادفی بین دو یا چند مداخله تخصیص داده می‌شوند، به طوری که احتمالا مداخله کنترل یا عدم‐مداخله را دربرگرفته، و نتایج مقایسه می‌شوند) را وارد مرور کردیم. نه مطالعه، 1892 شرکت‌کننده‌ای را وارد کردند که تحت جراحی پانکراس قرار گرفتند. چهار مطالعه از این نه مطالعه، 1110 شرکت‌کننده را برای استفاده از درن (تعداد شرکت‌کنندگان = 560) یا عدم‐استفاده از آن (N = 550) تصادفی‌سازی کردند. دو مطالعه 383 شرکت‌کننده را برای استفاده از درن فعال (درن‌ها با فشار ساکشن پائین یا بالا؛ N = 194) و درن غیر‐فعال (درن‌های بدون ساکشن؛ N = 189) تصادفی‌سازی کردند. سه مطالعه 399 شرکت‌کننده را با خطر پائین فیستول پانکراس پس از جراحی (ارتباط غیر‐طبیعی میان پانکراس و دیگر اندام‌ها به دلیل نشت شیره پانکراس حاوی آنزیم‌های گوارشی از مجاری آسیب‌دیده پانکراتیک) به خارج کردن زودهنگام درن (N = 200) یا خارج کردن دیرهنگام آن (N = 199) به صورت تصادفی اختصاص دادند.

منابع تامین مالی مطالعه

پنج مورد از نه مطالعه وارد شده با کمک‌های مالی غیر‐تجاری تامین شدند. دو مطالعه کمک مالی دریافت نکردند. دو مطالعه دیگر منابع مالی خود را گزارش ندادند.

نتایج کلیدی

استفاده از درن ممکن است مرگ‌ومیر را در 90 روز کاهش دهد. شواهد نشان می‌دهد که استفاده از درن در مقایسه با عدم‐استفاده از آن، ممکن است منجر به تفاوتی اندک تا عدم تفاوت در مرگ‌ومیر در 30 روز، عفونت زخم، طول مدت بستری در بیمارستان، نیاز به دریافت روش‌های باز بیشتر برای مدیریت عوارض پس از جراحی و کیفیت زندگی شود. یک عارضه مرتبط با درن (آسیب‌دیدگی لوله درناژ) در گروه استفاده از درن رخ داد (0.2%). استفاده از درن احتمالا منجر به تفاوتی اندک تا عدم تفاوت در عوارض کلی شد. مطمئن نیستیم استفاده از درن عفونت‌های شکمی، و نیاز به دریافت مداخلات رادیولوژیکی بیشتر را برای مدیریت عوارض پس از جراحی کاهش می‌دهد یا خیر.

شواهد در مورد تاثیر درن فعال در مقایسه با درن غیر‐فعال بر مرگ‌ومیر در 30 روز، عفونت در شکم، عفونت زخم، عوارض کلی، طول مدت بستری در بیمارستان، و نیاز به دریافت پروسیجرهای باز بیشتر برای مدیریت عوارض پس از جراحی بسیار نامشخص بود. هیچ عارضه مربوط به درن در هر دو گروه وجود نداشت.

شواهد در مورد تاثیر خارج کردن زودهنگام درن بر مرگ‌ومیر در 30 روز، عفونت زخم، هزینه‌های بیمارستان، و نیاز به دریافت پروسیجرهای باز یا رادیولوژیکی بیشتر برای مدیریت عوارض پس از جراحی بسیار نامشخص بود. ما دریافتیم که خارج کردن زودهنگام درن ممکن است عفونت‌های شکم، عوارض کلی، و طول مدت بستری را کاهش دهد، اما شواهد بسیار نامشخص بود. هیچ یک از مطالعات بروز عوارض مرتبط با درن را گزارش نکردند.

مشخص نیست که استفاده روتین از درن در مقایسه با عدم‐استفاده از آن تاثیری بر مرگ‌ومیر در 30 روز یا عوارض پس از جراحی داشته یا خیر. استفاده روتین از درن در مقایسه با عدم‐استفاده از آن ممکن است مرگ‌ومیر را در 90 روز کاهش دهد. شواهد در مورد تاثیر درن فعال در مقایسه با درن غیر‐فعال بر مرگ‌ومیر 30 روز یا عوارض پس از جراحی بسیار نامشخص است. به نظر می‌رسد خارج کردن زودهنگام درن در مقایسه با درآوردن دیرهنگام آن، باعث کاهش عفونت در ناحیه شکم، عوارض کلی، طول مدت بستری در بیمارستان برای افرادی با خطر پائین فیستول پانکراس پس از جراحی می‌شود، اما شواهد بسیار نامشخص است.

قطعیت شواهد

هر نه مطالعه دارای نقاط ضعفی بودند که به‌طور بالقوه قابلیت اطمینان نتایج را تحت تاثیر قرار دادند. در مجموع، قطعیت شواهد از بسیار پائین تا متوسط متغیر بود.

Authors' conclusions

Implications for practice

Compared with no drain use, it is unclear whether routine drain use has any effect on mortality at 30 days or on postoperative complications after pancreatic surgery. Low‐certainty evidence suggests that routine abdominal drainage may reduce mortality at 90 days, compared with no drain use. The evidence is very uncertain about the effect of an active drain on mortality at 30 days or postoperative complications compared with the use of a passive drain. Compared with late drain removal, early drain removal may reduce intra‐abdominal infection rate, morbidity, and length of hospital stay for people with a low risk of postoperative pancreatic fistula, but the evidence is very uncertain.

Implications for research

More studies with an adequate sample size are necessary to assess the benefits and harms of abdominal drainage for people with high risk of postoperative pancreatic fistula. A sample size of 870 (435 in each group) would be required to detect an absolute reduction in the intra‐abdominal infection rate of 5% (from 10% to 5%), at 80% power and an alpha‐error set at 0.05 (Conlon 2001).

Future studies should report the rate and grade of the postoperative complication according to the Clavien‐Dindo Classification (Clavien 2009; Dindo 2004). Future studies should analyse the data on an intention‐to‐treat basis in the case of post‐randomization dropouts. More studies are needed to assess drain versus no drain use in laparoscopic pancreatic resections.

Summary of findings

Open in table viewer
Summary of findings 1. Drain use versus no drain use following pancreatic surgery

Drain use versus no drain use following pancreatic surgery

Patient or population: people undergoing elective pancreatic resections
Setting: hospital
Intervention: drain use
Comparison: no drain use

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with no drain use

Risk with drain use

Mortality

Follow‐up: 30 days

23 per 1000

18 per 1000

(7 to 46)

RR 0.78
(0.31 to 1.99)

1055
(4 studies)

⊕⊕⊝⊝
Lowa,b

Low‐certainty evidence suggested that drain use may result in little to no difference in 30‐day mortality compared with no drain use.

Mortality

Follow‐up: 90 days

42 per 1000

10 per 1000

(3 to 38)

RR 0.23
(0.06 to 0.90)

478
(2 studies)

⊕⊕⊝⊝
Lowc,d

Low‐certainty evidence suggested that drain use may result in a slight reduction in 90‐day mortality compared with no drain use.

Intra‐abdominal infection

Follow‐up: 30 days

82 per 1000

80 per 1000

(43 to 148)

RR 0.97
(0.52 to 1.80)

1055
(4 studies)

⊕⊝⊝⊝
Very lowa,b,e

The evidence was very uncertain about the effect of drain use on intra‐abdominal infection rate compared with no drain use.

Wound infection

Follow‐up: 30 days

99 per 1000

97 per 1000

(68 to 140)

RR 0.98
(0.68 to 1.41)

1055
(4 studies)

⊕⊕⊝⊝
Lowa,b

Low‐certainty evidence suggested that drain use may result in little to no difference in wound infection rate compared with no drain use.

Drain‐related complications

Follow‐up: 30 days

See comment

See comment

Not estimable

179
(1 study)

⊕⊕⊝⊝
Lowf,g

Low‐certainty evidence suggested that drain use may result in little to no difference in drain‐related complications compared with no drain use. There was 1 drain‐related complication in the drainage group. The drainage tube was broken.

Morbidity

Follow‐up: 30 days

597 per 1000

614 per 1000

(561 to 674)

RR 1.03
(0.94 to 1.13)

1055
(4 studies)

⊕⊕⊕⊝
Moderate a

Moderate‐certainty evidence suggested that drain use probably resulted in little to no difference in morbidity compared with no drain use.

Length of hospital stay

Follow‐up: 30 days

The mean length of hospital stay in the no drain groups was 11.3 days

The mean length of hospital stay in the drain groups was
0.14 days lower
(0.79 lower to 0.51 higher)

MD ‐0.14 (‐0.79 to 0.51)

876
(3 studies)

⊕⊕⊝⊝
Lowh

One study including 179 participants reported the median and range values which were not suitable for pooling. This study reported no difference in the length of hospital stay between groups. Low‐certainty evidence suggested that drain use may result in little to no difference in length of hospital stay compared with no drain use.

* The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the control group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; MD: mean difference; RR: risk ratio

GRADE Working Group grades of evidence
High certainty. We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty. We are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty. Our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect
Very low certainty. We have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect

aDowngraded one level for serious risk of bias: three studies with high risk of incomplete outcome data
bDowngraded one level for serious imprecision: few events and a confidence interval that includes both potential benefit and potential harm from the intervention
cDowngraded one level for serious risk of bias: two studies with high risk of incomplete outcome data
dDowngraded one level for serious imprecision: few events
eDowngraded one level for serious unexplained inconsistency: substantial heterogeneity I²= 52%
fNot downgraded for risk of bias because this is considered an objective outcome that is unlikely to be affected by selection bias or performance bias
gDowngraded two levels for very serious imprecision: small sample size and very few events
hDowngraded two levels for very serious risk of bias: three studies with high risk of incomplete outcome data, all studies with high risk of performance bias, and this outcome was primarily determined by the surgeons

Open in table viewer
Summary of findings 2. Active drain versus passive drain following pancreatic surgery

Active drain versus passive drain following pancreatic surgery

Patient or population: people undergoing elective pancreatic resections
Setting: hospital
Intervention: active drain
Comparison: passive drain

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with passive drain

Risk with Active drain

Mortality

Follow‐up: 30 days

16 per 1000

20 per 1000
(5 to 80)

RR 1.23
(0.30 to 5.06)

382
(2 studies)

⊕⊝⊝⊝
Very lowa,b

The evidence was very uncertain about the effect of an active drain on 30‐day mortality compared with passive drain.

Mortality

Follow‐up: 90 days

Not reported

Intra‐abdominal infection

Follow‐up: 30 days

101 per 1000

88 per 1000
(21 to 368)

RR 0.87
(0.21 to 3.66)

321
(2 studies)

⊕⊝⊝⊝
Very lowa,b

The evidence was very uncertain about the effect of an active drain on intra‐abdominal infection rate compared with passive drain.

Wound infection

Follow‐up: 30 days

88 per 1000

81 per 1000
(39 to 167)

RR 0.92
(0.44 to 1.90)

321
(2 studies)

⊕⊝⊝⊝
Very lowa,b

The evidence was very uncertain about the effect of an active drain on wound infection rate compared with passive drain.

Drain‐related complications

Follow‐up: 30 days

See comment

See comment

Not estimable

223

(1 study)

⊕⊝⊝⊝
Very Lowa,b

There were no drain‐related complications in either group.

Morbidity

Follow‐up: 30 days

370 per 1000

359 per 1000
(196 to 656)

RR 0.97
(0.53 to 1.77)

382
(2 studies)

⊕⊝⊝⊝
Very lowa,c,d

The evidence was very uncertain about the effect of an active drain on morbidity compared with passive drain.

Length of hospital stay

Follow‐up: 30 days

The mean length of hospital stay in the passive drain group was 14.5 days

The mean length of hospital stay in the active drain group was
0.79 days lower
(2.63 days lower to 1.04 days higher)

MD ‐0.79
(‐2.63 to 1.04)

321
(2 studies)

⊕⊝⊝⊝
Very lowc,e,f

The evidence was very uncertain about the effect of an active drain on length of hospital stay compared with passive drain.

* The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the control group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; RR: risk ratio

GRADE Working Group grades of evidence
High certainty. We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty. We are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty. Our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect
Very low certainty. We have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect

aDowngraded one level for serious risk of bias: one study with unclear risk of incomplete outcome data
bDowngraded two levels for very serious imprecision: small sample size, few events, and wide confidence intervals
cDowngraded one level for serious imprecision: small sample size and a confidence interval that includes both potential benefit and potential harm from the intervention
dDowngraded one level for serious unexplained inconsistency: substantial heterogeneity I²= 76%
eDowngraded two levels for very serious risk of bias: one study with high risk of incomplete outcome data, both studies with high risk of performance bias, and this outcome was primarily determined by the surgeons
fDowngraded one level for serious unexplained inconsistency: substantial heterogeneity I²= 72%

Open in table viewer
Summary of findings 3. Early versus late drain removal following pancreatic surgery

Early versus late drain removal following pancreatic surgery

Patient or population: people undergoing elective pancreatic resections
Setting: hospital
Intervention: early drain removal
Comparison: late drain removal

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with late drain removal

Risk with early drain removal

Mortality

Follow‐up: 30 days

5 per 1000

5 per 1000
(0 to 78)

RR 0.99
(0.06 to 15.45)

399
(3 RCTs)

⊕⊝⊝⊝
Very lowa,b,g

The evidence was very uncertain about the effect of early drain removal on 30‐day mortality compared with late drain removal.

Mortality

Follow‐up: 90 days

Not reported

Intra‐abdominal infection

Follow‐up: 30 days

162 per 1000

71 per 1000
(36 to 144)

RR 0.44
(0.22 to 0.89)

285
(2 RCTs)

⊕⊝⊝⊝
Very lowa,c,g

The evidence suggested that early drain removal may reduce intra‐abdominal infection rate compared with late drain removal but the evidence was very uncertain.

Wound infection

Follow‐up: 30 days

35 per 1000

46 per 1000
(16 to 136)

RR 1.32
(0.45 to 3.85)

285
(2 RCTs)

⊕⊝⊝⊝
Very lowa,b,g

The evidence was very uncertain about the effect of early drain removal on wound infection rate compared with late drain removal.

Drain‐related complications

Follow‐up: 30 days

Not reported

Morbidity

Follow‐up: 30 days

659 per 1000

323 per 1000
(198 to 534)

RR 0.49
(0.30 to 0.81)

258
(2 RCTs)

⊕⊝⊝⊝
Very lowa,d,f,g

The evidence suggested that early drain removal may reduce morbidity compared with late drain removal but the evidence was very uncertain.

Length of hospital stay

Follow‐up: 30 days

The mean length of hospital stay in the late drain removal group was 15.4 days

The mean length of hospital stay in the early drain removal group was 2.2 days lower
(3.52 days lower to 0.87 days lower)

MD ‐2.2
(‐3.52 to ‐0.87)

399
(3 RCTs)

⊕⊝⊝⊝
Very lowe,f,g

The evidence suggested that early drain removal may reduce length of hospital stay compared with late drain removal but the evidence was very uncertain.

* The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the control group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; RR: risk ratio

GRADE Working Group grades of evidence
High certainty. We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty. We are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty. Our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect
Very low certainty. We have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect

aNot downgraded for risk of bias because this is considered an objective outcome that is unlikely to be affected by performance bias or detection bias
bDowngraded two levels for very serious imprecision: small sample size, few events, and wide confidence intervals
cDowngraded two levels for very serious imprecision: small sample size and few events
dDowngraded one level for serious unexplained inconsistency: substantial heterogeneity I²= 67%
eDowngraded two levels for very serious risk of bias: all studies with high risk of performance bias, and this outcome was primarily determined by the surgeons
fDowngraded one level due to serious imprecision: total population size was less than 400
gDowngraded one level for indirectness: different time points for early drain removal, and different definitions of low risk of postoperative pancreatic fistula

Background

Description of the condition

See 'Glossary' for an explanation of terms (Appendix 1).

Pancreatic cancer ranks fourteen of the most common cancers, and seven of the causes of cancer death from a global viewpoint (Bray 2018; Ferlay 2019). Regional differences exist in the number of new cases diagnosed per year (Kamisawa 2016; Maisonneuve 2019; Siegel 2019). The overall incidence of pancreatic cancer is approximately six cases per 100,000 people per year (Bray 2018; Dragovich 2020). The most common cause of pancreatic cancer is heavy tobacco usage (Kamisawa 2016; Maisonneuve 2019).

Although the exact worldwide incidence of chronic pancreatitis is unknown, the estimated incidence of chronic pancreatitis is six cases per 100,000 people per year in Europe, and probably in all western countries (Gupte 2018; Lew 2017; Singh 2019). The prevalence of chronic pancreatitis per 100,000 people is 3 cases in the UK, 4 in Japan, 26 in France, 41 in the USA, and 114 to 200 cases in south India (Bornman 2001; Braganza 2011; Garg 2004; Machicado 2017; Yadav 2011). The most common cause of chronic pancreatitis is alcohol abuse (Kleeff 2017; Lew 2017; Singh 2019).

Pancreatic surgery is performed to treat various pancreatic and extra‐pancreatic diseases, including pancreatic cancers, chronic pancreatitis, and biliary and duodenal malignancies (Cheng 2017; Cheng 2019; Deng 2020; Gurusamy 2013a). Although mortality due to pancreatic surgery has been reduced to less than 5%, the overall morbidity is still high, ranging from 30% to 60% (Bassi 2017; Gurusamy 2013a; Wente 2007a; Wente 2007b). The most common complications after pancreatic surgery include delayed gastric emptying (19% to 23%; Wente 2007b), postoperative pancreatic fistula (2% to 30%; Bassi 2017; Deng 2020; Hackert 2011; Wente 2007a; Wente 2007b), intra‐abdominal abscess (9% to 10%; Wente 2007a; Wente 2007b), wound infection (5% to 15%; Andrén‐Sandberg 2011; Halloran 2002), and postoperative bleeding (1% to 8%; Wente 2007a).

Description of the intervention

As a measure to reduce postoperative complications after pancreatic resections, prophylactic drains are traditionally placed in the subhepatic space near both the biliary and pancreatic anastomoses (Fisher 2018). Abdominal drainage has been in use for over 1000 years (Memon 2001).

Surgical drains are artificial tubes used to remove blood, pus, or other body fluids from wounds (Durai 2009). There are two main types of surgical drains: open and closed (Gurusamy 2007a; Li 2018; Wang 2015). An open drain communicates with the atmosphere (e.g. corrugated drain, Penrose drain, sump drain; Gurusamy 2007a; Li 2018; Wang 2015). A closed drain consists of a tube that drains into a collection bag or bottle, where the contents are not exposed to the atmosphere (Gurusamy 2007a; Li 2018; Wang 2015). Closed drains may be either active (suction drains under low or high pressure, e.g. Jackson‐Pratt drain, Redivac) or passive (drains without suction, e.g. Robinson drain, Pigtail drain; Gurusamy 2007a; Li 2018; Wang 2015).

How the intervention might work

Surgeons have routinely used drains after pancreatic surgery because of the possible collection of bile, pancreatic juice, or blood, which may require additional procedures. The primary reasons for placing abdominal drains after pancreatic resections are: (1) drainage of established intra‐abdominal collections (e.g. bile, pancreatic juice, pus); (2) prevention of further fluid accumulation; and (3) identification and monitoring of any fistula or bleeding. Theoretically, abdominal drainage has the potential to prevent or control postoperative complications (e.g. intra‐abdominal abscess, pancreatic or biliary fistula, bleeding; Bassi 2010; Conlon 2001; Van Buren 2014; Van Buren 2017; Adham 2013; Correa‐Gallego 2013; Fisher 2011; Giovinazzo 2011; Heslin 1998; Jeekel 1992; Kawai 2006; Lim 2013; Mehta 2013; Paulus 2012). The use of surgical drains is a very common practice after pancreatic surgery since the mid‐1930s (Allen 2011).

However, some surgeons have argued that abdominal drainage may fail to reduce postoperative complications, because a drain may become sealed and ineffective within a few days after pancreatic surgery (Heslin 1998; Paulus 2012). The drain itself appears to act as a foreign body, and may interfere with wound healing (Correa‐Gallego 2013; Fisher 2011; Paulus 2012). The drainage tube creates a pathway for contamination, and may increase the risk of postoperative infectious complications (Inoue 2011; Jeekel 1992). In addition, the use of a drain may be associated with an increased length of hospital stay (Fisher 2011; Mehta 2013; Paulus 2012; Van Buren 2017). Abdominal drainage may also be associated with rare adverse events, such as bowel perforation, hernia, and bleeding (Cameron‐Strange 1985; Henkus 1999; Makama 2010; Nomura 1998; Reed 1992; Sahu 2008; Srivastava 2007; Van Hee 1983). Studies have suggested that the routine placement of prophylactic abdominal drains may be unnecessary, and may be associated with an increased complication rate (Adham 2013; Correa‐Gallego 2013; Fisher 2011; Giovinazzo 2011; Heslin 1998; Jeekel 1992; Lim 2013; Mehta 2013; Paulus 2012).

Why it is important to do this review

Routine use of prophylactic abdominal drainage in people undergoing pancreatic surgery is controversial. This is an update of a Cochrane Review assessing the role of prophylactic abdominal drainage for pancreatic surgery (Zhang 2018). We conducted this review to explore uncertainty arising from conflicting results from a number of studies in this area.

Objectives

To assess the benefits and harms of routine abdominal drainage after pancreatic surgery, compare the effects of different types of surgical drains, and evaluate the optimal time for drain removal.

Methods

Criteria for considering studies for this review

Types of studies

We included all randomised controlled trials (RCTs), regardless of sample size, language, or publication status, which compared (1) drain use versus no drain use, (2) different types of drains, or (3) different schedules for drain removal in people undergoing pancreatic surgery. We excluded quasi‐randomized studies, in which the allocation was performed on the basis of a pseudo‐random sequence (e.g. odd or even hospital number or date of birth, alternation, and non‐randomized studies) because of the potential for bias (Reeves 2021).

Types of participants

We included people, regardless of age, sex, or race, who underwent elective pancreatic resections (open or laparoscopic) for any pancreatic or extra‐pancreatic disease.

Types of interventions

  1. Drain use versus no drain use

  2. One type of drain versus another

  3. Early versus late drain removal (up to four days versus more than four days)

Types of outcome measures

Primary outcomes

  1. Mortality

    1. 30‐day mortality

    2. 90‐day mortality

  2. Infectious complications

    1. intra‐abdominal infection (measured at 30 days)

    2. wound infection (measured at 30 days)

  3. Drain‐related complications (measured at 30 days)

Secondary outcomes

  1. Morbidity, as defined by study authors (measured at 30 days). We classified morbidity by the Clavien‐Dindo classification of surgical complications (Clavien 2009).

  2. Length of hospital stay

  3. Hospital costs (measured at 30 days)

  4. Additional procedures for postoperative complications

    1. open procedures (measured at 30 days)

    2. radiological interventions (radiological drainage requiring insertion of drain or percutaneous aspiration; measured at 30 days)

  5. Pain (measured at 30 days)

  6. Quality of life (measured at 30 days)

The main reason to justify abdominal drainage was the assumption that it would reduce the infectious complication rate, and subsequent mortality and morbidity rates. We chose other clinical outcomes to assess whether abdominal drainage resulted in earlier discharge from hospital, fewer reoperations, and improvement in health‐related quality of life and cost effectiveness.

Reporting of the outcomes listed here was not an inclusion criterion for the review.

Search methods for identification of studies

Before searching, we designed the search strategies with the help of the Cochrane Gut Group Information Specialist. We placed no restrictions on the language of publication when searching the electronic databases, or reviewing reference lists in identified studies.

Electronic searches

This is the third update of a Cochrane Review originally published in 2015 (Peng 2015), and previously updated in 2016 (Cheng 2016), and 2018 (Zhang 2018).

For this updated review, we updated the search strategies and re‐searched the following electronic databases, with no language or date of publication restrictions:

  1. Cochrane Central Register of Controlled Trials Ovid (CENTRAL; 2021, Issue 2) in the Cochrane Library (searched 08 February 2021; see Appendix 2);

  2. MEDLINE Ovid (1946 to 08 February 2021; see Appendix 3);

  3. Embase Ovid (1974 to 08 February 2021; see Appendix 4);

  4. Science Citation Index Expanded Web of Science (1900 to 08 February 2021; see Appendix 5); and

  5. Chinese Biomedical Literature Database (CBM; 1978 to 08 February 2021; see Appendix 6).

Searching other resources

We checked reference lists of all primary studies and relevant review articles that were identified during the search for RCTs, for additional references. We contacted authors of identified studies, and asked them to identify other published and unpublished studies.

We searched PubMed for errata or retractions of eligible studies, and reported the date this was done (www.ncbi.nlm.nih.gov/pubmed; accessed 08 February 2021). We also searched meeting abstracts of the Society of American Gastrointestinal and Endoscopic Surgeons (SAGES; www.sages.org/; accessed 08 February 2021), and Conference Proceedings Citation Index, to explore further relevant clinical studies.

Clinical trials registers/trial result registers

We searched the following databases to identify ongoing studies (accessed 08 February 2021):

  1. World Health Organization International Clinical Trials Registry Platform search portal (apps.who.int/trialsearch/);

  2. US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov/);

  3. Current Controlled Trials (www.controlled-trials.com/);

  4. EU Clinical Trials Register (www.clinicaltrialsregister.eu/);

  5. Chinese Clinical Trial Register (www.chictr.org/).

Data collection and analysis

We conducted this review according to the Cochrane Handbook for Systematic Reviews of Intervention (Higgins 2021).

Selection of studies

Two review authors (SH, WZ) independently screened the titles and abstracts of all the reports we identified as a result of the search, and coded them as 'retrieve' (eligible, potentially eligible, or unclear) or 'do not retrieve'. We retrieved the full‐text study reports and publications of identified reports, and two review authors (SH, WZ) independently screened the full text, identified studies for inclusion, and identified and recorded reasons to exclude the ineligible studies. We resolved any disagreements through discussion, or if required, we consulted a third review author (YC). We identified and excluded duplicates and collated multiple reports of the same study, so that each study, rather than each report, was the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram and a 'Characteristics of excluded studies' table (Moher 2009).

Data extraction and management

We used a standard data collection form for study characteristics and outcome data, which we piloted on at least one study in the review. Two review authors (ML, ZL) extracted the following study characteristics from included studies:

  1. methods: study design, total duration of study and run in, number of study centres and location, study setting, withdrawals, and date of study;

  2. participants: number (N), mean age, age range, gender, severity of condition, diagnostic criteria, inclusion criteria, and exclusion criteria;

  3. interventions: intervention, comparison;

  4. outcomes: primary and secondary outcomes specified and collected, time points reported;

  5. notes: funding for trial, notable conflicts of interest of trial authors.

Two review authors (ML, ZL) independently extracted outcome data from included studies. We noted in the 'Characteristics of included studies' table if outcome data were not reported in a usable way. We resolved disagreements by consensus, or by involving a third review author (YC). One review author (SH) copied the data from the data collection form into Review Manager 5 (Review Manager 2020). We double‐checked that the data were entered correctly by comparing the study reports with the data in the review. A second review author spot‐checked study characteristics for accuracy against the trial report.

Assessment of risk of bias in included studies

Two review authors (SH, JX) independently assessed risk of bias for each study, using the original Cochrane risk of bias tool for randomised trials (RoB 1), as outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We resolved any disagreements by discussion, or by involving a third review author (YC). We assessed the risk of bias according to the following domains:

  1. random sequence generation;

  2. allocation concealment;

  3. blinding of participants and personnel;

  4. blinding of outcome assessment;

  5. incomplete outcome data;

  6. selective outcome reporting;

  7. other bias.

We graded each potential source of bias as high, low, or unclear risk, and provided a quote from the study report together with a justification for our judgement in the risk of bias table. We summarized the risk of bias judgements across different studies for each of the domains listed. We considered blinding separately for different key outcomes where necessary (e.g. for unblinded outcome assessment, risk of bias for 30‐day mortality may be different than for a participant‐reported pain scale). Where information on risk of bias related to unpublished data or correspondence with a trialist, we noted this in the risk of bias table. We considered a trial to be at low risk of bias if we assessed the trial at low risk of bias across all domains. We considered trials at high risk of bias when one or more domains were at unclear or high risk of bias. We resolved any difference in opinion by discussion.

When considering treatment effects, we took into account the risk of bias for the studies that contributed to that outcome.

Assessment of bias in conducting the review

We conducted the review according to the published protocol, and reported any deviations from it in the 'Differences between protocol and review' section of the review (Cheng 2013).

Measures of treatment effect

We analysed dichotomous data as risk ratios (RR) and continuous data as mean differences (MD) with 95% confidence intervals (CI). For continuous data with different measurement scales in different randomised clinical trials, we calculated standardized mean differences (SMD) with 95% CI. We ensured that higher scores for continuous outcomes had the same meaning for the particular outcome, explained the direction to the reader, and reported where the directions were reversed, if this was necessary.

We undertook meta‐analyses only when this was meaningful, that is, if the treatments, participants, and underlying clinical question were similar enough for pooling to make sense.

A common way that trialists indicate when they have skewed data is by reporting medians and interquartile ranges. When we encountered this, we noted that the data were skewed, and considered the implication of this.

When multiple trial arms were reported in a single trial, we included only the relevant arms. If two comparisons (e.g. drug A versus placebo and drug B versus placebo) were entered into the same meta‐analysis, we halved the control group to avoid double‐counting.

Unit of analysis issues

The unit of analysis was the individual participant. We did not find any cross‐over or cluster‐randomized studies.

Dealing with missing data

We contacted investigators or study sponsors to verify key study characteristics and request missing numerical outcome data, when needed (e.g. when a study was identified as abstract only). We did not get any responses. Thus, we only used the available data in the analyses.

Assessment of heterogeneity

We used the I² statistic to measure heterogeneity among the studies in each analysis (Higgins 2003). When we identified substantial heterogeneity (greater than 50%), we explored it by prespecified subgroup analysis, and we interpreted summary effect measures with caution.

Assessment of reporting biases

We had planed to use funnel plots to assess reporting biases. However, we did not generate any funnel plots because we included fewer than 10 studies (Higgins 2021).

Data synthesis

We performed the meta‐analyses using Review Manager 5 software (Review Manager 2020). For all analyses, we used a random‐effects model.

Subgroup analysis and investigation of heterogeneity

We intended to perform the following subgroup analyses:

  1. RCTs with low risk of bias versus RCTs with high risk of bias;

  2. different etiologies (pancreatic cancer, chronic pancreatitis, and others);

  3. type of operation (proximal, distal, or central pancreatectomy).

We conducted subgroup analyses for the following outcomes.

  1. 30‐day mortality;

  2. 90‐day mortality;

  3. intra‐abdominal infection;

  4. wound infection;

  5. drain‐related complications.

Sensitivity analysis

We intended to perform sensitivity analyses to assess the robustness of our conclusions. The sensitivity analyses included:

  1. using a fixed‐effect model rather than a random‐effects model;

  2. calculating risk differences (RD), and odds ratios (OR) for dichotomous outcomes, as well as RR;

  3. calculating SMD for continuous data with same measurement scales in different RCTs and calculating MD for continuous data with different measurement scales in different RCTs;

  4. conducting worst‐case (the events happened in the experimental group but did not happen in the control group for missing participants) and best‐case (the events happened in the control group but did not happen in the experimental group for missing participants) scenario analyses for missing data.

If the results did not change, we considered them to be robust. If the results changed, we considered them to be less robust.

We planned sensitivity analyses for the following outcomes:

  1. 30‐day mortality;

  2. 90‐day mortality;

  3. intra‐abdominal infection;

  4. wound infection;

  5. drain‐related complications.

Reaching conclusions

We based our conclusions only on findings from the quantitative or narrative synthesis of included studies for this review. We avoided making recommendations for practice, and tried, in our implications for research, to give the reader a clear sense of where the focus of any future research in the area should be, and the remaining uncertainties.

Summary of findings and assessment of the certainty of the evidence

We included these outcomes in our summary of findings tables: 30‐day mortality, 90‐day mortality, intra‐abdominal infection, wound infection, drain‐related complications, morbidity, and length of hospital stay. We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the certainty of the body of evidence based on the studies that contributed data to the meta‐analyses for each outcome. We used the methods and recommendations described in Chapter 14 of the CochraneHandbook (Schünemann 2021), and GRADEpro GDT software to develop and display the summary of findings tables (GRADEpro GDT). We justified all decisions to downgrade the certainty of the evidence using footnotes, and made comments to aid the reader's understanding of the review, where necessary.

We defined the certainty of the evidence as follows:

  • High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.

  • Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.

  • Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect.

  • Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

Results

Description of studies

See 'Characteristics of included studies', 'Characteristics of excluded studies', and Characteristics of ongoing studies tables.

Results of the search

For this updated review, we identified 7959 records through the electronic searches of Cochrane Central Register of Controlled Trials Ovid (CENTRAL; 765 records), MEDLINE Ovid (1636 records), Embase Ovid (2993 records), Science Citation Index Expanded Web of Science (2080 records), and Chinese Biomedical Literature Database (CBM; 485 records). We did not identify any records by scanning the reference lists of the identified randomised controlled trials (RCTs). We excluded 1812 duplicates and 6127 clearly irrelevant records by screening the titles and abstracts. We retrieved the full text of the remaining twenty records for further assessment. We excluded 16 studies for the reasons listed in the 'Characteristics of excluded studies' table, and identified one ongoing study (Kaiser 2019). Three new RCTs fulfilled the inclusion criteria for this update. We show the study flow diagram in Figure 1.


Study flow diagram: 2021 review update

Study flow diagram: 2021 review update

Included studies

The last published version of this review included six RCTs, published between 2001 and 2016 (Bassi 2010; Conlon 2001; Jiang 2016; Van Buren 2014; Van Buren 2017; Witzigmann 2016). We added three recently published RCTs to this update (Čečka 2018; Dai 2020; Dembinski 2019). Therefore, we included nine RCTs (1892 participants) in this update, all of which provided data for the analyses. We listed details of the studies in the 'Characteristics of included studies' table.

Drain use versus no drain use

Four studies randomised 1110 participants who underwent elective pancreatic resections (604 pancreaticoduodenectomy, 439 distal pancreatectomy, and 67 other pancreatic surgery) to two groups: those who had drainage tubes inserted postoperatively (N = 560), and those who did not (N = 550 (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016)). Two of the RCTs (163 participants) included participants who underwent laparoscopic pancreatic resections (Van Buren 2014; Van Buren 2017). All four RCTs (947 participants) included participants who underwent open pancreatic resections (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016). Three of these studies were conducted in the USA (Conlon 2001; Van Buren 2014; Van Buren 2017), and one was conducted in Germany (Witzigmann 2016). The sample size calculation was described in three of the four RCTs (Van Buren 2014; Van Buren 2017; Witzigmann 2016). All of the studies used multiple surgeons to perform the surgeries, regardless of the number of study centres. Two RCTs were multi‐centre studies (Van Buren 2014; Van Buren 2017), one RCT was a dual‐centre study (Witzigmann 2016), and the other RCT was a single‐centre study (Conlon 2001). One study was terminated after an interim analysis because of excess mortality in the group with no drainage (Van Buren 2014). For all four trials, the mean age was 63.9 years; one or two drainage tubes were placed near the pancreatic anastomoses or pancreatic stumps; and the measured outcomes were mortality, morbidity, wound infection, intra‐abdominal infection, various postoperative complications, reoperation, additional radiological intervention, length of hospital stay, and quality of life.

One type of drain versus another

Two studies randomised 383 participants undergoing elective pancreatic resections (321 pancreaticoduodenectomy and 62 distal pancreatectomy) to the active drain group (N = 194) or the passive drain group (N = 189 (Čečka 2018; Jiang 2016)). All 383 participants in the two RCTs underwent open pancreatic resections. One RCT was conducted in China (Jiang 2016), the other in Czech Republic (Čečka 2018). Both RCTs described the sample size calculation. One RCT was a dual‐centre study involving many surgeons (Čečka 2018), while the other RCT was a single‐centre study involving one surgeon (Jiang 2016). For both trials, the mean age was 62.1 years; two drainage tubes were placed near both the biliary and pancreatic anastomoses and one drainage tube was placed near the pancreatic stump after distal pancreatectomy; and the measured outcomes were mortality, morbidity, wound infection, intra‐abdominal infection, various postoperative complications, reoperation, operation time, and length of hospital stay.

Early versus late drain removal

Three studies randomised 399 participants with low risk of postoperative pancreatic fistula, who were undergoing elective pancreatic resections (320 pancreaticoduodenectomy and 79 distal pancreatectomy) to the early removal group (N = 200), or the late removal group (N = 199 (Bassi 2010; Dai 2020; Dembinski 2019)). The definition of low risk of postoperative pancreatic fistula was not uniform among RCTs. Bassi 2010 defined low risk of postoperative pancreatic fistula as "an amylase value in drains on postoperative day 1 < 5000 IU/L". Dai 2020 defined it as "drain amylase on postoperative day 1 and 3 less than 5000 U/L, and drain output within postoperative day 3 less than 300 mL/d". Dembinski 2019 defined it as "drain amylase on postoperative day 3 less than 3 times the serum amylase activity". Two RCTs included 53 participants who underwent laparoscopic pancreatic resections (Dai 2020; Dembinski 2019). The other 346 participants in the three RCTs underwent open pancreatic resections. One RCT was conducted in Italy (Bassi 2010), one in China (Dai 2020), and the other in France (Dembinski 2019). Two RCTs described the sample size calculation (Bassi 2010; Dembinski 2019). All three RCTs were single‐centre studies involving many surgeons. For all three trials, the mean age was 59.5 years; two drainage tubes were placed near both the biliary and pancreatic anastomoses; and one drainage tube was placed near the pancreatic stump after distal pancreatectomy. In the early removal group, drains were removed on postoperative day 3 in two RCTs (Bassi 2010; Dai 2020), and postoperative day 4 in the third RCT (Dembinski 2019). In the late removal group, drains were removed on postoperative day 5 or later in all three RCTs. The three RCTs reported outcomes of mortality, wound infection, intra‐abdominal infection, postoperative pancreatic fistula, abdominal complications, pulmonary complications, reoperation, length of hospital stay, hospital readmission, morbidity, and hospital costs.

Excluded studies

We listed the details for the 16 excluded studies in the 'Characteristics of excluded studies' table. We excluded one RCT because it focused on pancreatic duct drainage (Lee 2009); the remaining studies were excluded because we identified them as non‐randomized studies (Adham 2013; Behrman 2015; Correa‐Gallego 2013; Fisher 2011; Giovinazzo 2011; Heslin 1998; Jeekel 1992; Kawai 2006; Kawaida 2021; Kunstman 2017; Lemke 2021; Lim 2013; Mehta 2013; Paulus 2012; Zaghal 2020).

Ongoing studies

We identified one ongoing study (Kaiser 2019). They plan to include 252 participants for open or minimally invasive distal pancreatectomy for pancreatic disease, and will randomize them to the drainage group or no drainage group. This trial started in March 2018. It is currently recruiting participants in Germany. The primary outcome is morbidity. The secondary outcomes are mortality, various postoperative complications, reoperation rate, operation time, length of hospital stay, duration of intensive care unit stay, and readmission rate. See the 'Characteristics of ongoing studies' table for more details.

Risk of bias in included studies

Figure 2 and Figure 3 summarize the risk of bias of the included studies. All nine studies were at high risk of bias.


Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies


Risk of bias summary: review authors' judgements about each risk of bias item for each included study

Risk of bias summary: review authors' judgements about each risk of bias item for each included study

Allocation

We judged random sequence generation at low risk of bias in eight studies, in which participants were randomised using computer‐generated numbers (Bassi 2010; Čečka 2018; Dai 2020; Dembinski 2019; Jiang 2016; Van Buren 2014; Van Buren 2017; Witzigmann 2016), and unclear risk of bias in one study, as it did not provide any specific information on the randomization process, and we did not receive a response from the authors (Conlon 2001). We judged allocation concealment at low risk of bias in three studies that used sealed opaque envelopes (Čečka 2018; Witzigmann 2016), or central allocation (Dai 2020), to conceal the allocation, and unclear risk of bias in the remaining six studies, which provided no information.

Blinding

We judged blinding of participants and personnel as unclear risk of bias in one study, which did not describe whether adequate blinding of participants and personnel was achieved, and we did not receive a response from the authors (Jiang 2016). We judged blinding of participants and personnel at high risk of bias in the six studies that stated 'open label' in the protocols (Bassi 2010; Čečka 2018; Dai 2020; Dembinski 2019; Van Buren 2014; Van Buren 2017). For the remaining two studies, the surgeons could not be blinded to group allocation (drain use versus no drain use (Conlon 2001; Witzigmann 2016)). Therefore, we judged both studies at high risk of blinding of participants and personnel.

We judged blinding of outcome assessment at low risk of bias in one study (Čečka 2018), unclear risk of bias in three studies, since they provided no information (Conlon 2001; Jiang 2016; Witzigmann 2016), and high risk of bias in the remaining five studies which stated 'open label' in the protocols (Bassi 2010; Dai 2020; Dembinski 2019; Van Buren 2014; Van Buren 2017).

Incomplete outcome data

There were no post‐randomisation dropouts in four studies (Bassi 2010; Conlon 2001; Dai 2020; Jiang 2016). Although there were seven dropouts (5.0%) in one study, the study authors analysed the data on an intention‐to‐treat basis (Dembinski 2019). We considered these five studies at low risk of attrition bias (Bassi 2010; Conlon 2001; Dai 2020; Dembinski 2019; Jiang 2016). There were some dropouts in three studies, but the data were not analysed on an intention‐to‐treat basis, therefore, we judged these three studies at high risk of bias for incomplete outcome data (Van Buren 2014; Van Buren 2017; Witzigmann 2016). There was one dropout in a study, it was not clear whether there was a substantial departure from the intervention in this study that would cause a per‐protocol analysis to differ significantly from an intention‐to‐treat analysis, therefore, we judged this study at unclear risk of bias for incomplete outcome data (Čečka 2018).

Selective reporting

The study protocols were available for seven studies (Bassi 2010; Čečka 2018; Dai 2020; Dembinski 2019; Van Buren 2014; Van Buren 2017; Witzigmann 2016). All of the studies reported their prespecified outcomes. Thus, we considered these seven studies to be free of selective reporting. The study protocols were not available for two studies, but hey fully reported on all expected outcomes, so we judged both studies to be free of selective reporting (Conlon 2001; Jiang 2016).

Other potential sources of bias

We judged baseline imbalance at low risk of bias in all nine studies.

Five of the nine studies were sponsored by non‐commercial grants (e.g. university grant, national cancer control program, charitable funding (Čečka 2018; Conlon 2001; Dai 2020; Van Buren 2014; Van Buren 2017)). Two studies did not receive any funding (Dembinski 2019; Witzigmann 2016). The other two studies provided no information about funding sources (Bassi 2010; Jiang 2016). Six of the nine studies declared no conflict of interest (Dai 2020; Dembinski 2019; Jiang 2016; Van Buren 2014; Van Buren 2017; Witzigmann 2016). The other three studies did not report on conflicts of interest (Bassi 2010; Čečka 2018; Conlon 2001).

Effects of interventions

See: Summary of findings 1 Drain use versus no drain use following pancreatic surgery; Summary of findings 2 Active drain versus passive drain following pancreatic surgery; Summary of findings 3 Early versus late drain removal following pancreatic surgery

Drain use versus no drain use

Four studies (1110 participants) compared drain use with no drain use (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016). Five hundred and sixty participants were randomised to the drainage group, and 550 participants to the no drainage group. See summary of findings Table 1.

Mortality (30 days)

Four studies (1055 participants: 532 drain use, 523 no drain) reported on 30‐day mortality (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016). There were eight deaths in the drainage group, and 12 deaths in the no drainage group. The evidence suggested that drain use may result in little to no difference in 30‐day mortality compared with no drain use (risk ratio (RR) 0.78, 95% confidence interval (CI) 0.31 to 1.99; four studies, 1055 participants; low‐certainty evidence; Analysis 1.1). We downgraded the certainty of the evidence one level for serious risk of bias because three studies with high risk of incomplete outcome data contributed 77.2% of the weight towards this effect estimate (Van Buren 2014; Van Buren 2017; Witzigmann 2016). We downgraded one level for serious imprecision because there were few deaths in either group, and the confidence interval included both potential benefit and potential harm from the intervention.

Mortality (90 days)

Two studies (478 participants: 242 drain use, 236 no drain) reported on 90‐day mortality (Van Buren 2014; Van Buren 2017). There were two deaths in the drainage group and 10 deaths in the no drainage group. The evidence suggested that drain use may reduce 90‐day mortality compared with no drain use (RR 0.23, 95% CI 0.06 to 0.90; two studies, 478 participants; low‐certainty evidence; Analysis 1.2). We downgraded the certainty of the evidence one level for serious risk of bias because both studies were at high risk of incomplete outcome data; and one level for serious imprecision because there were few deaths in either group.

Intra‐abdominal infection

Four studies (1055 participants: 532 drain use, 523 no drain) reported on intra‐abdominal infection (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016). There were 42 intra‐abdominal infections in the drainage group, and 43 intra‐abdominal infections in the no drainage group. The evidence was very uncertain about the effect of drain use on intra‐abdominal infection rate compared with no drain use (RR 0.97, 95% CI 0.52 to 1.80; four studies, 1055 participants; very low‐certainty evidence; Analysis 1.3). We downgraded the evidence one level for serious risk of bias because three studies with high risk of incomplete outcome data contributed 80.6% of the weight towards this effect estimate (Van Buren 2014; Van Buren 2017; Witzigmann 2016). We downgraded the certainty of the evidence one level for serious imprecision, because there were few events in either group and confidence intervals included both potential benefit and potential harm from the intervention; and one level for serious inconsistency (unexplained statistical heterogeneity I² = 52%).

Wound infection

Four studies (1055 participants: 532 drain use, 523 no drain) reported on wound infection. There were 52 wound infections in the drainage group, and 52 wound infections in the no drainage group (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016). The evidence suggested that drain use may result in little to no difference in wound infection rate compared with no drain use (RR 0.98, 95% CI 0.68 to 1.41; four studies, 1055 participants; low‐certainty evidence; Analysis 1.4). We downgraded the certainty of the evidence one level for serious risk of bias because three studies with high risk of incomplete outcome data contributed 80.8% of the weight towards this effect estimate (Van Buren 2014; Van Buren 2017; Witzigmann 2016). We downgraded one level for serious imprecision, because there were few events in either group and the confidence interval included both potential benefit and potential harm from the intervention.

Drain‐related complications

One study (179 participants: 91 drain use, 88 no drain) reported this outcome (Conlon 2001). There was one drain‐related complication (broken drain) in the drainage group, and no drain‐related complications in the no drainage group. The evidence suggested that drain use may result in little to no difference in drain‐related complications compared with no drain use (one study, 179 participants; low‐certainty evidence). We did not downgrade the certainty of the evidence for risk of bias because this is considered an objective outcome, which is unlikely to be affected by selection bias or performance bias in this study. We downgraded two levels for very serious imprecision because of the small sample size and very few events.

Morbidity

Four studies (1055 participants: 532 drain use, 523 no drain) reported on morbidity (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016). There were 328 participants with complications in the drainage group, and 312 participants with complications in the no drainage group. The evidence suggested that drain use probably resulted in little to no difference in morbidity compared with no drain use (RR 1.03, 95% CI 0.94 to 1.13; four studies, 1055 participants; moderate‐certainty evidence; Analysis 1.5). We downgraded the certainty of evidence one level for serious risk of bias because three studies with high risk of incomplete outcome data contributed 84.4% of the weight towards this effect estimate (Van Buren 2014; Van Buren 2017; Witzigmann 2016).

Length of hospital stay

Four studies (1055 participants: 532 drain use, 523 no drain) reported on length of hospital stay (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016). One study reported the mean and standard deviation (SD (Witzigmann 2016)). Two studies reported the median and interquartile range (IQR) values (Van Buren 2014; Van Buren 2017). Assuming a normal distribution of the data, we imputed the mean and SD values using the formula (1) mean = median, and (2) SD = IQR / 1.35; according to the Cochrane Handbook (Higgins 2021). The evidence suggested that drain use may result in little to no difference in length of hospital stay compared with no drain use (MD ‐0.14 days, 95% CI ‐0.79 to 0.51; three studies, 876 participants; low‐certainty evidence; Analysis 1.6). One study including 179 participants reported the median and range values which were not suitable for pooling (Conlon 2001). Conlon 2001 reported no difference in the length of hospital stay between groups (Analysis 1.7). We downgraded the certainty of evidence two levels for very serious risk of bias, because three studies were at high risk of incomplete outcome data (Van Buren 2014; Van Buren 2017; Witzigmann 2016). In addition, all of the studies were at high risk of performance bias, and length of hospital stay was primarily determined by the surgeons.

Hospital costs

None of the studies reported this outcome.

Additional open procedures for postoperative complications

Four studies (1055 participants: 532 drain use, 523 no drain) reported on this outcome (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016). There were 50 participants who needed additional open procedures for postoperative complications in the drainage group, and 37 participants in the no drainage group. The evidence suggested that drain use may result in little to no difference in the need for additional open procedures for postoperative complications compared with no drain use (RR 1.33, 95% CI 0.79 to 2.23; four studies, 1055 participants; low‐certainty evidence; Analysis 1.8). We downgraded the certainty of evidence one level for serious risk of bias, because three studies with high risk of incomplete outcome data contributed 83.1% of the weight towards this effect estimate (Van Buren 2014; Van Buren 2017; Witzigmann 2016). We downgraded the certainty of the evidence one level for serious imprecision, because there were few events in either group and the confidence interval included both potential benefit and potential harm from the intervention.

Additional radiological interventions for postoperative complications

Three studies (660 participants (330 drain use, 330 no drain) reported on this outcome (Conlon 2001; Van Buren 2014; Van Buren 2017). There were 35 participants who needed additional radiological interventions for postoperative complications in the drainage group, and 40 participants in the no drainage group. The evidence was very uncertain about the effect of drain use on the need for additional radiological interventions for postoperative complications compared with no drain use (RR 0.87, 95% CI 0.40 to 1.87; three studies, 660 participants; very low‐certainty evidence; Analysis 1.9). We downgraded the certainty of evidence one level for serious risk of bias, because two studies with high risk of incomplete outcome data contributed 69.7% of the weight towards this effect estimate (Van Buren 2014; Van Buren 2017). We downgraded the certainty of the evidence one level for serious imprecision because there were few events in either group, and the confidence interval included both potential benefit and potential harm from the intervention; and one level for serious inconsistency (unexplained statistical heterogeneity I² = 64%).

Pain

None of the studies reported on pain.

Quality of life

One study (399 participants: 202 drain use, 197 no drain) reported this outcome (Van Buren 2017). Van Buren 2017 used the pancreas‐specific quality of life questionnaire (FACT‐PA) to assess quality of life. The FACT‐PA is a scale of 0 to 144, with higher values indicating better quality of life. The mean quality of life score at 30 days after pancreatic surgery was 105 points in the drainage and 104 points in the no drainage group. The study reported the mean quality of life score, without providing the standard deviation. The evidence suggested that drain use may result in little to no difference in quality of life compared with no drain use; low‐certainty evidence. We downgraded the certainty of the evidence one level for serious risk of bias, because the study was at high risk of incomplete outcome data; and one level for serious imprecision, because of the small sample size.

Active drain versus passive drain

Two studies (383 participants) compared active drain versus passive drain (Čečka 2018; Jiang 2016). One hundred and ninety‐four participants were randomised to the active drain group and 189 participants to the passive drain group. See summary of findings Table 2.

Mortality (30 days)

Both studies (382 participants: 193 active drain, 189 passive drain) reported on 30‐day mortality. There were four deaths in the active drain group, and three deaths in the passive drain group. The evidence was very uncertain about the effect of an active drain on 30‐day mortality compared with a passive drain (RR 1.23, 95% CI 0.30 to 5.06; two studies, 382 participants; very low‐certainty evidence; Analysis 2.1). We downgraded the certainty of the evidence one level for serious risk of bias, because one study with unclear risk of incomplete outcome data contributed 80.3% of the weight towards this effect estimate (Čečka 2018). We downgraded two levels for very serious imprecision, because of the small sample size, few events, and wide confidence interval.

Mortality (90 days)

Neither study reported this outcome.

Intra‐abdominal infection

Both studies (321 participants: 162 active drain, 159 passive drain) reported on intra‐abdominal infection. There were 17 intra‐abdominal infections in the active drain group, and 16 intra‐abdominal infections in the passive drain group. The evidence was very uncertain about the effect of an active drain on intra‐abdominal infection rate compared with a passive drain (RR 0.87, 95% CI 0.21 to 3.66; two studies, 321 participants; very low‐certainty evidence; Analysis 2.2). We downgraded the certainty of the evidence one level for serious risk of bias, because one study with unclear risk of incomplete outcome data contributed 81.8% of the weight towards this effect estimate (Čečka 2018). We downgraded two levels for very serious imprecision, because of the small sample size, few events, and wide confidence interval.

Wound infection

Both studies (321 participants: 162 active drain, 159 passive drain) reported on wound infection. There were 13 wound infections in the active drain group, and 14 wound infections in the passive drain group. The evidence was very uncertain about the effect of an active drain on wound infection rate compared with a passive drain (RR 0.92, 95% CI 0.44 to 1.90; two studies, 321 participants; very low‐certainty evidence; Analysis 2.3). We downgraded the certainty of the evidence one level for serious risk of bias, because one study with unclear risk of incomplete outcome data contributed 55.6% of the weight towards this effect estimate (Čečka 2018). We downgraded two levels for very serious imprecision because of the small sample size, few events, and wide confidence interval.

Drain‐related complications

One study (223 participants: 112 active drain, 111 passive drain) reported this outcome (Čečka 2018). There was no drain‐related complication in either group. We downgraded the certainty of the evidence one level for serious risk of bias, and two levels for very serious imprecision because of the small sample size and few events.

Morbidity

Both studies (382 participants: 193 active drain, 189 passive drain) reported on morbidity. There were 75 participants with complications in the active drain group, and 70 participants with complications in the passive drain group. The evidence was very uncertain about the effect of an active drain on morbidity compared with a passive drain (RR 0.97, 95% CI 0.53 to 1.77; two studies, 382 participants; very low‐certainty evidence; Analysis 2.4). We downgraded the certainty of the evidence one level for serious risk of bias because one study with unclear risk of incomplete outcome data contributed 56.3% of the weight towards this effect estimate (Čečka 2018). We downgraded one level for serious imprecision, because of the small sample size, and a confidence interval that included both potential benefit and potential harm from the intervention; and one level for serious inconsistency (unexplained statistical heterogeneity I² = 76%).

Length of hospital stay

Both studies (321 participants: 162 active drain, 159 passive drain) reported on length of hospital stay. One study reported the mean and SD (Jiang 2016). The other study reported the median and interquartile range (IQR (Čečka 2018)). Assuming a normal distribution of the data, we imputed the mean and SD values using the formula (1) mean = median, and (2) SD = IQR / 1.35; according to the Cochrane Handbook (Higgins 2021). The evidence was very uncertain about the effect of an active drain on length of hospital stay compared with a passive drain (MD ‐0.79 days, 95% CI ‐2.63 to 1.04; two studies, 321 participants; very low‐certainty evidence; Analysis 2.5). We downgraded the certainty of the evidence two levels for very serious risk of bias, because one study was at unclear risk of incomplete outcome data (Čečka 2018) and both studies were at high risk of performance bias, and length of hospital stay was primarily determined by the surgeons. We downgraded one level for serious imprecision, because of the small sample size, and a confidence interval that included both potential benefit and potential harm from the intervention; and one level for serious inconsistency (unexplained statistical heterogeneity I² = 72%).

Hospital costs

Neither study reported this outcome.

Additional open procedures for postoperative complications

Both studies (321 participants: 162 active drain, 159 passive drain) reported on this outcome. There were six participants who needed additional open procedures for postoperative complications in the active drain group, and 13 participants in the passive drain group. The evidence was very uncertain about the effect of an active drain on the need for additional open procedures for postoperative complications compared with a passive drain (RR 0.44, 95% CI 0.11 to 1.83; two studies, 321 participants; very low‐certainty evidence; Analysis 2.6). We downgraded the certainty of the evidence one level for serious risk of bias because one study with unclear risk of incomplete outcome data contributed 67.2% of the weight towards this effect estimate (Čečka 2018). We downgraded two levels for very serious imprecision, because of the small sample size, few events, and wide confidence interval; and one level for serious inconsistency (unexplained statistical heterogeneity I² = 39%).

Additional radiological interventions for postoperative complications

Neither study reported this outcome.

Pain

Neither study reported this outcome.

Quality of life

Neither study reported this outcome.

Early versus late drain removal

Three studies (399 participants with low risk of postoperative pancreatic fistula: 200 early drain removal, 199 late drain removal) compared early versus late drain removal (Bassi 2010; Dai 2020; Dembinski 2019). See summary of findings Table 3.

Mortality (30 days)

Three studies reported on 30‐day mortality. There was one death in the early removal group, and one death in the late removal group. The evidence was very uncertain about the effect of early drain removal on 30‐day mortality compared with late drain removal (RR 0.99, 95% CI 0.06 to 15.45; three studies, 399 participants; very low‐certainty evidence; Analysis 3.1). We downgraded the certainty of the evidence two levels for very serious imprecision, because of the small sample size, few events, and wide confidence interval; and one level for indirectness, because of the different time points for early drain removal and different definitions of low risk of postoperative pancreatic fistula.

We did not downgraded the certainty of the evidence for risk of bias, because we considered this an objective outcome, which was unlikely to be affected by performance bias or detection bias.

Mortality (90 days)

None of the studies reported this outcome.

Intra‐abdominal infection

Two studies (285 participants: 143 early removal, 142 late removal) reported on intra‐abdominal infection (Dai 2020; Dembinski 2019). There were 10 intra‐abdominal infections in the early removal group, and 23 intra‐abdominal infections in the late removal group. The evidence suggested that early drain removal may reduce intra‐abdominal infection rate compared with late drain removal, but the evidence was very uncertain (RR 0.44, 95% CI 0.22 to 0.89; two studies, 285 participants; very low‐certainty evidence; Analysis 3.2). We downgraded the certainty of the evidence two levels for very serious imprecision, because of the small sample size and few events; and one level for indirectness, because of the different time points for early drain removal and different definitions of low risk of postoperative pancreatic fistula.

We did not downgraded the certainty of the evidence for risk of bias, because we considered this an objective outcome, which was unlikely to be affected by performance bias or detection bias.

Wound infection

Two studies (285 participants: 143 early removal, 142 late removal) reported on wound infection (Dai 2020; Dembinski 2019). There were 13 intra‐abdominal infections in the early removal group, and 14 intra‐abdominal infections in the late removal group. The evidence was very uncertain about the effect of early drain removal on wound infection rate compared with late drain removal (RR 1.32, 95% CI 0.45 to 3.85; two studies, 285 participants; very low‐certainty evidence; Analysis 3.3). We downgraded the certainty of the evidence two levels for very serious imprecision, because of the small sample size, few events, and wide confidence interval; and one level for indirectness because of the different time points for early drain removal and different definitions of low risk of postoperative pancreatic fistula.

We did not downgraded the certainty of the evidence for risk of bias, because we considered this an objective outcome, which was unlikely to be affected by performance bias or detection bias.

Drain‐related complications

None of the studies reported this outcome.

Morbidity

Two studies (258 participants: 129 early removal, 129 late removal) reported on morbidity (Bassi 2010; Dai 2020). There were 41 participants with complications in the early removal group, and 85 participants with complications in the late removal group. The evidence suggested that early drain removal may reduce morbidity, but the evidence was very uncertain (RR 0.49, 95% CI 0.30 to 0.81; two studies, 258 participants; very low‐certainty evidence; Analysis 3.4). We downgraded the certainty of the evidence two levels for very serious imprecision, because of the small sample size and few events; one level for serious inconsistency (unexplained statistical heterogeneity I² = 67%); and one level for indirectness, because of the different time points for early drain removal and different definitions of low risk of postoperative pancreatic fistula.

We did not downgraded the certainty of the evidence for risk of bias, because we considered this an objective outcome, which was unlikely to be affected by performance bias or detection bias.

Length of hospital stay

All three studies (399 participants: 200 early removal, 199 late removal) reported on length of hospital stay, providing the mean and standard deviation (SD). The evidence suggested that early drain removal may reduce length of hospital stay, but the evidence was very uncertain (MD ‐2.20 days, 95% CI ‐3.52 to ‐0.87; three studies, 399 participants; very low‐certainty evidence; Analysis 3.5). We downgraded the certainty of the evidence two levels for very serious risk of bias, because all three studies were at high risk of performance bias, and length of hospital stay was primarily determined by the surgeons; one level for serious imprecision, because of the small sample size; and one level for indirectness, because of the different time points for early drain removal and different definitions of low risk of postoperative pancreatic fistula.

Hospital costs

Two studies (258 participants: 129 early removal, 129 late removal) reported on hospital costs, but reported them using different measurement scales (Bassi 2010; Dai 2020). The evidence was very uncertain about the effect of early drain removal on hospital costs compared with late drain removal (SMD ‐0.22, 95% CI ‐0.59 to 0.14; two studies, 258 participants; very low‐certainty evidence; Analysis 3.6). We downgraded the certainty of the evidence one level for serious risk of bias, because both studies were at high risk of performance bias; one level for serious imprecision, because of the small sample size; one level for serious inconsistency (unexplained statistical heterogeneity I² = 54%); and one level for indirectness, because of the different time points for early drain removal and different definitions of low risk of postoperative pancreatic fistula.

Additional open procedures for postoperative complications

All three studies (399 participants: 200 early removal, 199 late removal) reported this outcome. There were six participants who needed additional open procedures for postoperative complications in the early removal group, and eight participants in the late removal group. The evidence was very uncertain about the effect of early drain removal on the need of additional open procedures for postoperative complications compared with late drain removal (RR 0.77, 95% CI 0.28 to 2.10; three studies, 399 participants; very low‐certainty evidence; Analysis 3.7). We downgraded the certainty of the evidence two levels for very serious imprecision, because of the small sample size, few events, and wide confidence interval; and one level for indirectness, because of the different time points for early drain removal and different definitions of low risk of postoperative pancreatic fistula.

We did not downgraded the certainty of the evidence for risk of bias, because we considered this an objective outcome, which was unlikely to be affected by performance bias or detection bias.

Additional radiological interventions for postoperative complications

One study (144 participants: 72 early removal, 72 late removal) reported this outcome (Dai 2020). There were three participants who needed additional radiological interventions for postoperative complications in the early removal group, and three participants in the late removal group. The evidence was very uncertain about the effect of early drain removal on the need for additional radiological procedures for postoperative complications compared with late drain removal (RR 1.00, 95% CI 0.21 to 4.79; one study, 144 participants; very low‐certainty evidence; Analysis 3.8). We downgraded the certainty of the evidence two levels for very serious imprecision, because of the small sample size, few events, and wide confidence interval; and one level for indirectness, because of the different time points for early drain removal and different definitions of low risk of postoperative pancreatic fistula.

We did not downgraded the certainty of the evidence for risk of bias, because we considered this an objective outcome, which was unlikely to be affected by performance bias or detection bias.

Pain

None of the studies reported this outcome.

Quality of life

None of the studies reported this outcome.

Subgroup analysis

We did not perform any of the planned subgroup analyses because there too few studies included in each comparison.

Sensitivity analysis

We performed the following planned sensitivity analyses:

  1. used a fixed‐effect model rather than a random‐effects model;

  2. calculated RD and OR, as well as RR for dichotomous outcomes;

  3. conducted worst‐case and best‐case scenario analyses for missing data.

We observed no change in the direction of the results between a fixed‐effect and a random‐effects model, by calculating the risk difference (RD) or odds ratio (OR) for dichotomous outcomes (Table 1).

Open in table viewer
Table 1. Sensitivity analyses

Comparisons

Outcomes

Main analysis (random‐effects model)

Fixed‐effect model

RD for dichotomous outcomes

OR for dichotomous outcomes

worst‐case scenario and best‐case scenario analysis for missing data

worst/best‐case

best/worst‐case

Drain use versus no drain use

Mortality (30 days)

RR 0.78 (0.31 to 1.99)

RR 0.67 (0.28 to 1.58)

RD ‐0.01 (‐0.03 to 0.01)

OR 0.77 (0.29 to 2.03)

RR 2.58 (0.41 to 16.09)

RR 0.17 (0.04 to 0.79)

Drain use versus no drain use

Mortality (90 days)

RR 0.23 (0.06 to 0.90)

RR 0.23 (0.06 to 0.90)

RD ‐0.05 (‐0.16 to 0.07)

OR 0.21 (0.05 to 0.87)

RR 2.27 (0.06 to 84.61)

RR 0.07 (0.00 to 1.23)

Drain use versus no drain use

Intra‐abdominal infection

RR 0.97 (0.52 to 1.80)

RR 0.97 (0.64 to 1.45)

RD ‐0.00 (‐0.06 to 0.05)

OR 0.96 (0.48 to 1.92)

RR 1.72 (0.59 to 4.96)

RR 0.45 (0.31 to 0.63)

Drain use versus no drain use

Wound infection

RR 0.98 (0.68 to 1.41)

RR 0.98 (0.68 to 1.41)

RD 0.01 (‐0.03 to 0.04)

OR 0.98 (0.65 to 1.48)

RR 1.79 (0.81 to 3.95)

RR 0.53 (0.29 to 0.95)

Drain use versus no drain use

Drain‐related complications

Active drain versus passive drain

Mortality (30 days)

RR 1.23 (0.30 to 5.06)

RR 1.27 (0.32 to 5.07)

RD 0.01 (‐0.02 to 0.03)

OR 1.24 (0.29 to 5.28)

RR 1.51 (0.40 to 5.77)

RR 1.22 (0.30 to 5.02)

Active drain versus passive drain

Mortality (90 days)

Active drain versus passive drain

Intra‐abdominal infection

RR 0.87 (0.21 to 3.66)

RR 1.07 (0.58 to 1.96)

RD ‐0.00 (‐0.10 to 0.10)

OR 0.87 (0.19 to 4.07)

RR 0.88 (0.20 to 3.99)

RR 0.87 (0.21 to 3.57)

Active drain versus passive drain

Wound infection

RR 0.92 (0.44 to 1.90)

RR 0.91 (0.44 to 1.88)

RD ‐0.01 (‐0.07 to 0.05)

OR 0.91 (0.41 to 2.01)

RR 0.98 (0.48 to 2.01)

RR 0.91 (0.44 to 1.89)

Active drain versus passive drain

Drain‐related complications

Early versus late drain removal

Mortality (30 days)

RR 0.99 (0.06 to 15.45)

RR 0.99 (0.06 to 15.45)

RD ‐0.00 (‐0.02 to 0.02)

OR 0.99 (0.06 to 16.08)

RR 0.99 (0.06 to 15.45)

RR 0.99 (0.06 to15.45)

Early versus late drain removal

Mortality (90 days)

Early versus late drain removal

Intra‐abdominal infection

RR 0.44 (0.22 to 0.89)

RR 0.43 (0.21 to 0.87)

RD ‐0.09 (‐0.17 to ‐0.02)

OR 0.39 (0.18 to 0.87)

RR 0.44 (0.22 to 0.89)

RR 0.44 (0.22 to 0.89)

Early versus late drain removal

Wound infection

RR 1.32 (0.45 to 3.85)

RR 1.35 (0.47 to 3.90)

RD 0.01 (‐0.02 to 0.05)

OR 1.35 (0.43 to 4.29)

RR 1.32 (0.45 to 3.85)

RR 1.32 (0.45 to 3.85)

Early versus late drain removal

Drain‐related complications

‐: not available; OR: odds ratios; RR: risk ratio; RD: risk differences

There were 102 post‐randomization dropouts in four studies (Čečka 2018; Van Buren 2014; Van Buren 2017; Witzigmann 2016). We observed no change in the direction of the results by changing between worst‐case and best‐case scenario analyses for missing data, except for the following outcomes for the comparison drain use versus no drain use: mortality (30 days), mortality (90 days), intra‐abdominal infection, and wound infection (Table 1). We considered these outcomes to be less robust.

In the worst‐case scenario analyses for missing data in the comparison drain use versus no drain use, the results became consistent with both significant benefit and significant harm for 90‐day mortality (RR 2.27, 95% CI 0.06 to 84.61).

In the best‐case scenario analyses for missing data in the comparison drain use versus no drain use, the results became consistent with significant benefit in drain use for 30‐day mortality (RR 0.17, 95% CI 0.04 to 0.79), intra‐abdominal infection (RR 0.45, 95% CI 0.31 to 0.63), and wound infection (RR 0.53, 95% CI 0.29 to 0.95); the results became consistent with both significant benefit and significant harm for 90‐day mortality (RR 0.07, 95% CI 0.00 to 1.23).

We did not perform the planned sensitivity analysis by calculating SMD for continuous data with same measurement scales in different RCTs or calculating MD for continuous data with different measurement scales in different RCTs, because we did not include any of the continuous outcomes in the sensitivity analyses.

Discussion

Summary of main results

Evidence from nine studies, with 1892 people undergoing pancreatic surgery, contributed data to the outcomes of interest for this review.

For the comparison of drain use versus no drain use (four studies, 1110 participants), we found low‐certainty evidence that drain use may reduce 90‐day mortality compared with no drain use. There was low‐certainty evidence suggesting that drain use may result in little to no difference in 30‐day mortality, wound infection rate, length of hospital stay, the need for additional open procedures for postoperative complications, and quality of life compared with no drain use. There was moderate‐certainty evidence that drain use probably resulted in little to no difference in morbidity compared with no drain use. There was very uncertain evidence about the effect of drain use on intra‐abdominal infection rate and the need for additional radiological interventions for postoperative complications compared with no drain use. One study reported one drain‐related complication in the drainage group compared to no events in the no drainage group.

Two studies (383 participants) compared the use of an active drain versus a passive drain. There was very uncertain evidence about the effect of an active drain on 30‐day mortality, intra‐abdominal infection rate, wound infection rate, morbidity, length of hospital stay, and the need for additional open procedures for postoperative complications compared with a passive drain. One study reported there were no drain‐related complications in either group.

Data comparing early versus late drain removal were available from three studies (399 participants). There was very uncertain evidence about the effect of early drain removal on 30‐day mortality, wound infection rate, hospital costs, and the need for additional or radiological open procedures for postoperative complications compared with late drain removal. We found that early drain removal may reduce intra‐abdominal infection rate, morbidity, and length of hospital stay compared with late drain removal, but the evidence was very uncertain. None of the studies reported on drain‐related complications.

A previous version of this review included six studies with 1384 people undergoing pancreatic surgery. However, most of the studies compared drain use versus no drain use (four studies,1110 participants). Only one small study (160 participants) compared the use of an active drain versus a passive drain and suggested the efficacy of an active drain on earlier discharge from hospital. With the addition of three recent studies to this review (Čečka 2018; Dai 2020; Dembinski 2019), the results changed from the previous publication (Zhang 2018).

Overall completeness and applicability of evidence

All of the studies included people who were undergoing elective pancreaticoduodenectomy (N = 1245, 65.8%), distal pancreatectomy (N = 580, 30.7%), or other pancreatic surgery (N = 67, 3.5%) for various pancreatic and extra‐pancreatic diseases, including pancreatic cancers, ampullary cancers, chronic pancreatitis, biliary and duodenal malignancy. The majority (N = 1113, 58.9%) of the participants had either pancreatic cancers (N = 936, 49.5%) or chronic pancreatitis (N = 177, 9.4%). Only four of the studies included participants (N = 216, 11.4%) who underwent laparoscopic pancreatic resections (Dai 2020; Dembinski 2019; Van Buren 2014; Van Buren 2017); therefore, the results of this review may not be applicable to people undergoing laparoscopic pancreatic resections. Therefore, the results of this review are mostly applicable to people undergoing elective open pancreaticoduodenectomy, distal pancreatectomy, or other pancreatic surgery for various pancreatic and extra‐pancreatic diseases, especially pancreatic cancers or chronic pancreatitis.

Quality of the evidence

For drain use versus no drain use, we rated the certainty of evidence for seven important outcomes. We found low‐certainty evidence for 30‐day mortality, 90‐day mortality, wound infection, drain‐related complications, and length of hospital stay; moderate‐certainty evidence for morbidity; and very low‐certainty evidence for intra‐abdominal infection. We did not downgrade the certainty of the evidence for drain‐related complications for risk of bias, because it was unlikely to be affected by selection bias or performance bias. We downgraded the certainty of the evidence for 30‐day mortality, 90‐day mortality, intra‐abdominal infection, and wound infection by one level for serious risk of bias, because there were incomplete outcome data in some studies. We downgraded the certainty of the evidence for length of hospital stay by two levels for very serious risk of bias, because there were incomplete outcome data and lack of blinding of participants and personnel, since length of hospital stay was primarily determined by the surgeons. We downgraded the certainty of the evidence for intra‐abdominal infection by one level for serious unexplained inconsistency (substantial heterogeneity). We downgraded the certainty of the evidence for 30‐day mortality, 90‐day mortality, intra‐abdominal infection, and wound infection by one level for serious imprecision (few events or the confidence intervals of pooled data were either very close to or crossed the line of no effect). We downgraded the certainty of the evidence for drain‐related complications two levels for very serious imprecision, because of the small sample size and very few events. There was no indirectness of evidence, as the included studies assessed the appropriate population, intervention, comparisons, and outcomes. We included too few studies in this comparison to assess publication bias (summary of findings Table 1).

For active drain versus passive drain, we found very low‐certainty evidence for all outcomes. We downgraded the certainty of the evidence for 30‐day mortality, 90‐day mortality, intra‐abdominal infection, wound infection, drain‐related complications, and morbidity by one level for serious risk of bias, because of incomplete outcome data in one study (Čečka 2018). We downgraded the certainty of the evidence by two levels for length of hospital stay for very serious risk of bias, because of incomplete outcome data and lack of blinding of participants and personnel, since length of hospital stay was primarily determined by the surgeons. We downgraded the certainty of the evidence for morbidity and length of hospital stay by one level for serious unexplained inconsistency (substantial heterogeneity). We downgraded the certainty of the evidence for 30‐day mortality, 90‐day mortality, intra‐abdominal infection, wound infection, and drain‐related complications by two levels for very serious imprecision, because of the small sample sizes, few events, and wide confidence intervals. We downgraded the certainty of the evidence for morbidity and length of hospital stay by one level for serious imprecision, because of the small sample sizes, and because the confidence intervals of pooled data were either very close to or crossed the line of no effect. There was no indirectness of evidence, as the included studies assessed the appropriate population, intervention, comparisons, and outcomes. We were unable to assess publication bias as we included too few studies in this comparison (summary of findings Table 2).

For the comparison between early and late drain removal, we found very low‐certainty evidence for all outcomes. We did not downgrade the certainty of the evidence for 30‐day mortality, intra‐abdominal infection, wound infection, and morbidity for risk of bias, because these outcomes were unlikely to be affected by selection bias or performance bias. We downgraded the certainty of the evidence for length of hospital stay by two levels for very serious risk of bias, because of lack of blinding of participants and personnel, as length of hospital stay was primarily determined by the surgeons. We downgraded the certainty of the evidence for morbidity by one level for serious unexplained inconsistency (substantial heterogeneity). We downgraded the certainty of the evidence for 30‐day mortality, intra‐abdominal infection, and wound infection by two levels for very serious imprecision, because of the small sample sizes and few events. We downgraded the certainty of the evidence for morbidity and length of hospital stay by one level for serious imprecision, because of the small sample sizes. We downgraded the certainty of the evidence for all outcomes assessed by one level for indirectness (different time points for early drain removal and different definitions of low risk of postoperative pancreatic fistula). We were unable to assess publication bias, since we included too few studies in this comparison (summary of findings Table 3).

Potential biases in the review process

There were some unavoidable, potential biases of note in the review process. First, when we contacted some investigators to request further information, we did not get a reply. The missing data may introduce bias to this review. In addition, we could not explore publication bias, because there were too few studies included in each comparison, and we did not have access to protocols for two studies (Conlon 2001; Jiang 2016).

Agreements and disagreements with other studies or reviews

There is increasing evidence from Cochrane Reviews that routine abdominal drainage after various abdominal operations is not mandatory (Gurusamy 2007a; Gurusamy 2007b; Gurusamy 2013b; Li 2018; Rolph 2004; Wang 2015). The routine use of surgical drains has also been questioned in other areas, including thyroid, gynaecological, and orthopaedic surgeries (Charoenkwan 2017; Gates 2013; Parker 2007; Samraj 2007).

One systematic review that compared drain use with no drain use in people undergoing pancreatic resections concluded that the routine use of abdominal drains after pancreatic resection may result in a higher risk of major complications. Van der Wilt 2013 included three studies that we had considered for this review (Conlon 2001; Fisher 2011; Heslin 1998). Two of these studies were non‐randomized, so we did not include them in this review. Another systematic review also compared drain use with no drain use after pancreatic resections. Hüttner 2017 included three randomised controlled trials (RCTs), and concluded that drain use and no drain had similar results for mortality, morbidity, and re‐intervention (Conlon 2001; Van Buren 2014; Witzigmann 2016). One recent systematic review on this topic included six non‐randomized studies, and four RCTs (Conlon 2001; Van Buren 2014; Van Buren 2017; Witzigmann 2016). Lyu 2020 showed comparable outcomes for pancreatic resections with or without drainage. The results of this review update disagree with the findings of previous systematic reviews about mortality (Hüttner 2017; Lyu 2020; Van der Wilt 2013). We included the same four RCTs and used GRADE to assess the certainty of the evidence in this review update. We found that drain use reduce 90‐day mortality. With regard to other outcomes, this review update is consistent with the findings of Hüttner 2017 and Lyu 2020.

One recent systematic review compared the use of active drain with passive drain in people undergoing pancreatic resection, and included three RCTs (Čečka 2018; Jiang 2016; Lee 2009). One of these studies was a RCT about pancreatic duct drainage, so we did not include it in this review (Lee 2009). Gachabayov 2019 concluded that there was no difference in clinical outcomes between active drain and passive drain. For this review update, we included the two RCTs for this comparison and used GRADE to assess the certainty of the evidence. Our results agree with the findings of Gachabayov 2019.

Another systematic review compared early drain removal with late drain removal in people undergoing pancreatic resection and included two studies that we had considered for this review (Bassi 2010; Kawai 2006). One of the studies was not randomised, so we did not include it (Kawai 2006). Diener 2011 concluded that early drain removal seemed to be superior to late drain removal. For this review update, we included three RCTs for this comparison and used GRADE to assess the certainty of the evidence (Bassi 2010; Dai 2020; Dembinski 2019). Our results agree with the findings of Diener 2011.

Study flow diagram: 2021 review update

Figuras y tablas -
Figure 1

Study flow diagram: 2021 review update

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies

Figuras y tablas -
Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies

Risk of bias summary: review authors' judgements about each risk of bias item for each included study

Figuras y tablas -
Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study

Comparison 1: Drain use versus no drain use, Outcome 1: Mortality (30 days)

Figuras y tablas -
Analysis 1.1

Comparison 1: Drain use versus no drain use, Outcome 1: Mortality (30 days)

Comparison 1: Drain use versus no drain use, Outcome 2: Mortality (90 days)

Figuras y tablas -
Analysis 1.2

Comparison 1: Drain use versus no drain use, Outcome 2: Mortality (90 days)

Comparison 1: Drain use versus no drain use, Outcome 3: Intra‐abdominal infection

Figuras y tablas -
Analysis 1.3

Comparison 1: Drain use versus no drain use, Outcome 3: Intra‐abdominal infection

Comparison 1: Drain use versus no drain use, Outcome 4: Wound infection

Figuras y tablas -
Analysis 1.4

Comparison 1: Drain use versus no drain use, Outcome 4: Wound infection

Comparison 1: Drain use versus no drain use, Outcome 5: Morbidity

Figuras y tablas -
Analysis 1.5

Comparison 1: Drain use versus no drain use, Outcome 5: Morbidity

Comparison 1: Drain use versus no drain use, Outcome 6: Length of hospital stay (days)

Figuras y tablas -
Analysis 1.6

Comparison 1: Drain use versus no drain use, Outcome 6: Length of hospital stay (days)

Length of hospital stay

Study

Number in study

Comparison

Results

Comment

Conlon 2001

179 (91 versus 88)

Drain use versus no drain use

Drain use versus no drain use: median (range): 9 (3‐34) versus 9 (5‐44), without reporting the p value

Authors reported that there was no difference in the length of hospital stay between groups

Figuras y tablas -
Analysis 1.7

Comparison 1: Drain use versus no drain use, Outcome 7: Length of hospital stay

Comparison 1: Drain use versus no drain use, Outcome 8: Additional open procedures for postoperative complications

Figuras y tablas -
Analysis 1.8

Comparison 1: Drain use versus no drain use, Outcome 8: Additional open procedures for postoperative complications

Comparison 1: Drain use versus no drain use, Outcome 9: Additional radiological interventions for postoperative complications

Figuras y tablas -
Analysis 1.9

Comparison 1: Drain use versus no drain use, Outcome 9: Additional radiological interventions for postoperative complications

Comparison 2: Active drain versus passive drain, Outcome 1: Mortality (30 days)

Figuras y tablas -
Analysis 2.1

Comparison 2: Active drain versus passive drain, Outcome 1: Mortality (30 days)

Comparison 2: Active drain versus passive drain, Outcome 2: Intra‐abdominal infection

Figuras y tablas -
Analysis 2.2

Comparison 2: Active drain versus passive drain, Outcome 2: Intra‐abdominal infection

Comparison 2: Active drain versus passive drain, Outcome 3: Wound infection

Figuras y tablas -
Analysis 2.3

Comparison 2: Active drain versus passive drain, Outcome 3: Wound infection

Comparison 2: Active drain versus passive drain, Outcome 4: Morbidity

Figuras y tablas -
Analysis 2.4

Comparison 2: Active drain versus passive drain, Outcome 4: Morbidity

Comparison 2: Active drain versus passive drain, Outcome 5: Length of hospital stay (days)

Figuras y tablas -
Analysis 2.5

Comparison 2: Active drain versus passive drain, Outcome 5: Length of hospital stay (days)

Comparison 2: Active drain versus passive drain, Outcome 6: Additional open procedures for postoperative complications

Figuras y tablas -
Analysis 2.6

Comparison 2: Active drain versus passive drain, Outcome 6: Additional open procedures for postoperative complications

Comparison 3: Early versus late drain removal, Outcome 1: Mortality (30 days)

Figuras y tablas -
Analysis 3.1

Comparison 3: Early versus late drain removal, Outcome 1: Mortality (30 days)

Comparison 3: Early versus late drain removal, Outcome 2: Intra‐abdominal infection

Figuras y tablas -
Analysis 3.2

Comparison 3: Early versus late drain removal, Outcome 2: Intra‐abdominal infection

Comparison 3: Early versus late drain removal, Outcome 3: Wound infection

Figuras y tablas -
Analysis 3.3

Comparison 3: Early versus late drain removal, Outcome 3: Wound infection

Comparison 3: Early versus late drain removal, Outcome 4: Morbidity

Figuras y tablas -
Analysis 3.4

Comparison 3: Early versus late drain removal, Outcome 4: Morbidity

Comparison 3: Early versus late drain removal, Outcome 5: Length of hospital stay (days)

Figuras y tablas -
Analysis 3.5

Comparison 3: Early versus late drain removal, Outcome 5: Length of hospital stay (days)

Comparison 3: Early versus late drain removal, Outcome 6: Hospital costs

Figuras y tablas -
Analysis 3.6

Comparison 3: Early versus late drain removal, Outcome 6: Hospital costs

Comparison 3: Early versus late drain removal, Outcome 7: Additional open procedures for postoperative complications

Figuras y tablas -
Analysis 3.7

Comparison 3: Early versus late drain removal, Outcome 7: Additional open procedures for postoperative complications

Comparison 3: Early versus late drain removal, Outcome 8: Additional radiological interventions for postoperative complications

Figuras y tablas -
Analysis 3.8

Comparison 3: Early versus late drain removal, Outcome 8: Additional radiological interventions for postoperative complications

Summary of findings 1. Drain use versus no drain use following pancreatic surgery

Drain use versus no drain use following pancreatic surgery

Patient or population: people undergoing elective pancreatic resections
Setting: hospital
Intervention: drain use
Comparison: no drain use

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with no drain use

Risk with drain use

Mortality

Follow‐up: 30 days

23 per 1000

18 per 1000

(7 to 46)

RR 0.78
(0.31 to 1.99)

1055
(4 studies)

⊕⊕⊝⊝
Lowa,b

Low‐certainty evidence suggested that drain use may result in little to no difference in 30‐day mortality compared with no drain use.

Mortality

Follow‐up: 90 days

42 per 1000

10 per 1000

(3 to 38)

RR 0.23
(0.06 to 0.90)

478
(2 studies)

⊕⊕⊝⊝
Lowc,d

Low‐certainty evidence suggested that drain use may result in a slight reduction in 90‐day mortality compared with no drain use.

Intra‐abdominal infection

Follow‐up: 30 days

82 per 1000

80 per 1000

(43 to 148)

RR 0.97
(0.52 to 1.80)

1055
(4 studies)

⊕⊝⊝⊝
Very lowa,b,e

The evidence was very uncertain about the effect of drain use on intra‐abdominal infection rate compared with no drain use.

Wound infection

Follow‐up: 30 days

99 per 1000

97 per 1000

(68 to 140)

RR 0.98
(0.68 to 1.41)

1055
(4 studies)

⊕⊕⊝⊝
Lowa,b

Low‐certainty evidence suggested that drain use may result in little to no difference in wound infection rate compared with no drain use.

Drain‐related complications

Follow‐up: 30 days

See comment

See comment

Not estimable

179
(1 study)

⊕⊕⊝⊝
Lowf,g

Low‐certainty evidence suggested that drain use may result in little to no difference in drain‐related complications compared with no drain use. There was 1 drain‐related complication in the drainage group. The drainage tube was broken.

Morbidity

Follow‐up: 30 days

597 per 1000

614 per 1000

(561 to 674)

RR 1.03
(0.94 to 1.13)

1055
(4 studies)

⊕⊕⊕⊝
Moderate a

Moderate‐certainty evidence suggested that drain use probably resulted in little to no difference in morbidity compared with no drain use.

Length of hospital stay

Follow‐up: 30 days

The mean length of hospital stay in the no drain groups was 11.3 days

The mean length of hospital stay in the drain groups was
0.14 days lower
(0.79 lower to 0.51 higher)

MD ‐0.14 (‐0.79 to 0.51)

876
(3 studies)

⊕⊕⊝⊝
Lowh

One study including 179 participants reported the median and range values which were not suitable for pooling. This study reported no difference in the length of hospital stay between groups. Low‐certainty evidence suggested that drain use may result in little to no difference in length of hospital stay compared with no drain use.

* The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the control group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; MD: mean difference; RR: risk ratio

GRADE Working Group grades of evidence
High certainty. We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty. We are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty. Our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect
Very low certainty. We have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect

aDowngraded one level for serious risk of bias: three studies with high risk of incomplete outcome data
bDowngraded one level for serious imprecision: few events and a confidence interval that includes both potential benefit and potential harm from the intervention
cDowngraded one level for serious risk of bias: two studies with high risk of incomplete outcome data
dDowngraded one level for serious imprecision: few events
eDowngraded one level for serious unexplained inconsistency: substantial heterogeneity I²= 52%
fNot downgraded for risk of bias because this is considered an objective outcome that is unlikely to be affected by selection bias or performance bias
gDowngraded two levels for very serious imprecision: small sample size and very few events
hDowngraded two levels for very serious risk of bias: three studies with high risk of incomplete outcome data, all studies with high risk of performance bias, and this outcome was primarily determined by the surgeons

Figuras y tablas -
Summary of findings 1. Drain use versus no drain use following pancreatic surgery
Summary of findings 2. Active drain versus passive drain following pancreatic surgery

Active drain versus passive drain following pancreatic surgery

Patient or population: people undergoing elective pancreatic resections
Setting: hospital
Intervention: active drain
Comparison: passive drain

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with passive drain

Risk with Active drain

Mortality

Follow‐up: 30 days

16 per 1000

20 per 1000
(5 to 80)

RR 1.23
(0.30 to 5.06)

382
(2 studies)

⊕⊝⊝⊝
Very lowa,b

The evidence was very uncertain about the effect of an active drain on 30‐day mortality compared with passive drain.

Mortality

Follow‐up: 90 days

Not reported

Intra‐abdominal infection

Follow‐up: 30 days

101 per 1000

88 per 1000
(21 to 368)

RR 0.87
(0.21 to 3.66)

321
(2 studies)

⊕⊝⊝⊝
Very lowa,b

The evidence was very uncertain about the effect of an active drain on intra‐abdominal infection rate compared with passive drain.

Wound infection

Follow‐up: 30 days

88 per 1000

81 per 1000
(39 to 167)

RR 0.92
(0.44 to 1.90)

321
(2 studies)

⊕⊝⊝⊝
Very lowa,b

The evidence was very uncertain about the effect of an active drain on wound infection rate compared with passive drain.

Drain‐related complications

Follow‐up: 30 days

See comment

See comment

Not estimable

223

(1 study)

⊕⊝⊝⊝
Very Lowa,b

There were no drain‐related complications in either group.

Morbidity

Follow‐up: 30 days

370 per 1000

359 per 1000
(196 to 656)

RR 0.97
(0.53 to 1.77)

382
(2 studies)

⊕⊝⊝⊝
Very lowa,c,d

The evidence was very uncertain about the effect of an active drain on morbidity compared with passive drain.

Length of hospital stay

Follow‐up: 30 days

The mean length of hospital stay in the passive drain group was 14.5 days

The mean length of hospital stay in the active drain group was
0.79 days lower
(2.63 days lower to 1.04 days higher)

MD ‐0.79
(‐2.63 to 1.04)

321
(2 studies)

⊕⊝⊝⊝
Very lowc,e,f

The evidence was very uncertain about the effect of an active drain on length of hospital stay compared with passive drain.

* The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the control group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; RR: risk ratio

GRADE Working Group grades of evidence
High certainty. We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty. We are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty. Our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect
Very low certainty. We have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect

aDowngraded one level for serious risk of bias: one study with unclear risk of incomplete outcome data
bDowngraded two levels for very serious imprecision: small sample size, few events, and wide confidence intervals
cDowngraded one level for serious imprecision: small sample size and a confidence interval that includes both potential benefit and potential harm from the intervention
dDowngraded one level for serious unexplained inconsistency: substantial heterogeneity I²= 76%
eDowngraded two levels for very serious risk of bias: one study with high risk of incomplete outcome data, both studies with high risk of performance bias, and this outcome was primarily determined by the surgeons
fDowngraded one level for serious unexplained inconsistency: substantial heterogeneity I²= 72%

Figuras y tablas -
Summary of findings 2. Active drain versus passive drain following pancreatic surgery
Summary of findings 3. Early versus late drain removal following pancreatic surgery

Early versus late drain removal following pancreatic surgery

Patient or population: people undergoing elective pancreatic resections
Setting: hospital
Intervention: early drain removal
Comparison: late drain removal

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with late drain removal

Risk with early drain removal

Mortality

Follow‐up: 30 days

5 per 1000

5 per 1000
(0 to 78)

RR 0.99
(0.06 to 15.45)

399
(3 RCTs)

⊕⊝⊝⊝
Very lowa,b,g

The evidence was very uncertain about the effect of early drain removal on 30‐day mortality compared with late drain removal.

Mortality

Follow‐up: 90 days

Not reported

Intra‐abdominal infection

Follow‐up: 30 days

162 per 1000

71 per 1000
(36 to 144)

RR 0.44
(0.22 to 0.89)

285
(2 RCTs)

⊕⊝⊝⊝
Very lowa,c,g

The evidence suggested that early drain removal may reduce intra‐abdominal infection rate compared with late drain removal but the evidence was very uncertain.

Wound infection

Follow‐up: 30 days

35 per 1000

46 per 1000
(16 to 136)

RR 1.32
(0.45 to 3.85)

285
(2 RCTs)

⊕⊝⊝⊝
Very lowa,b,g

The evidence was very uncertain about the effect of early drain removal on wound infection rate compared with late drain removal.

Drain‐related complications

Follow‐up: 30 days

Not reported

Morbidity

Follow‐up: 30 days

659 per 1000

323 per 1000
(198 to 534)

RR 0.49
(0.30 to 0.81)

258
(2 RCTs)

⊕⊝⊝⊝
Very lowa,d,f,g

The evidence suggested that early drain removal may reduce morbidity compared with late drain removal but the evidence was very uncertain.

Length of hospital stay

Follow‐up: 30 days

The mean length of hospital stay in the late drain removal group was 15.4 days

The mean length of hospital stay in the early drain removal group was 2.2 days lower
(3.52 days lower to 0.87 days lower)

MD ‐2.2
(‐3.52 to ‐0.87)

399
(3 RCTs)

⊕⊝⊝⊝
Very lowe,f,g

The evidence suggested that early drain removal may reduce length of hospital stay compared with late drain removal but the evidence was very uncertain.

* The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the control group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; RR: risk ratio

GRADE Working Group grades of evidence
High certainty. We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty. We are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty. Our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect
Very low certainty. We have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect

aNot downgraded for risk of bias because this is considered an objective outcome that is unlikely to be affected by performance bias or detection bias
bDowngraded two levels for very serious imprecision: small sample size, few events, and wide confidence intervals
cDowngraded two levels for very serious imprecision: small sample size and few events
dDowngraded one level for serious unexplained inconsistency: substantial heterogeneity I²= 67%
eDowngraded two levels for very serious risk of bias: all studies with high risk of performance bias, and this outcome was primarily determined by the surgeons
fDowngraded one level due to serious imprecision: total population size was less than 400
gDowngraded one level for indirectness: different time points for early drain removal, and different definitions of low risk of postoperative pancreatic fistula

Figuras y tablas -
Summary of findings 3. Early versus late drain removal following pancreatic surgery
Table 1. Sensitivity analyses

Comparisons

Outcomes

Main analysis (random‐effects model)

Fixed‐effect model

RD for dichotomous outcomes

OR for dichotomous outcomes

worst‐case scenario and best‐case scenario analysis for missing data

worst/best‐case

best/worst‐case

Drain use versus no drain use

Mortality (30 days)

RR 0.78 (0.31 to 1.99)

RR 0.67 (0.28 to 1.58)

RD ‐0.01 (‐0.03 to 0.01)

OR 0.77 (0.29 to 2.03)

RR 2.58 (0.41 to 16.09)

RR 0.17 (0.04 to 0.79)

Drain use versus no drain use

Mortality (90 days)

RR 0.23 (0.06 to 0.90)

RR 0.23 (0.06 to 0.90)

RD ‐0.05 (‐0.16 to 0.07)

OR 0.21 (0.05 to 0.87)

RR 2.27 (0.06 to 84.61)

RR 0.07 (0.00 to 1.23)

Drain use versus no drain use

Intra‐abdominal infection

RR 0.97 (0.52 to 1.80)

RR 0.97 (0.64 to 1.45)

RD ‐0.00 (‐0.06 to 0.05)

OR 0.96 (0.48 to 1.92)

RR 1.72 (0.59 to 4.96)

RR 0.45 (0.31 to 0.63)

Drain use versus no drain use

Wound infection

RR 0.98 (0.68 to 1.41)

RR 0.98 (0.68 to 1.41)

RD 0.01 (‐0.03 to 0.04)

OR 0.98 (0.65 to 1.48)

RR 1.79 (0.81 to 3.95)

RR 0.53 (0.29 to 0.95)

Drain use versus no drain use

Drain‐related complications

Active drain versus passive drain

Mortality (30 days)

RR 1.23 (0.30 to 5.06)

RR 1.27 (0.32 to 5.07)

RD 0.01 (‐0.02 to 0.03)

OR 1.24 (0.29 to 5.28)

RR 1.51 (0.40 to 5.77)

RR 1.22 (0.30 to 5.02)

Active drain versus passive drain

Mortality (90 days)

Active drain versus passive drain

Intra‐abdominal infection

RR 0.87 (0.21 to 3.66)

RR 1.07 (0.58 to 1.96)

RD ‐0.00 (‐0.10 to 0.10)

OR 0.87 (0.19 to 4.07)

RR 0.88 (0.20 to 3.99)

RR 0.87 (0.21 to 3.57)

Active drain versus passive drain

Wound infection

RR 0.92 (0.44 to 1.90)

RR 0.91 (0.44 to 1.88)

RD ‐0.01 (‐0.07 to 0.05)

OR 0.91 (0.41 to 2.01)

RR 0.98 (0.48 to 2.01)

RR 0.91 (0.44 to 1.89)

Active drain versus passive drain

Drain‐related complications

Early versus late drain removal

Mortality (30 days)

RR 0.99 (0.06 to 15.45)

RR 0.99 (0.06 to 15.45)

RD ‐0.00 (‐0.02 to 0.02)

OR 0.99 (0.06 to 16.08)

RR 0.99 (0.06 to 15.45)

RR 0.99 (0.06 to15.45)

Early versus late drain removal

Mortality (90 days)

Early versus late drain removal

Intra‐abdominal infection

RR 0.44 (0.22 to 0.89)

RR 0.43 (0.21 to 0.87)

RD ‐0.09 (‐0.17 to ‐0.02)

OR 0.39 (0.18 to 0.87)

RR 0.44 (0.22 to 0.89)

RR 0.44 (0.22 to 0.89)

Early versus late drain removal

Wound infection

RR 1.32 (0.45 to 3.85)

RR 1.35 (0.47 to 3.90)

RD 0.01 (‐0.02 to 0.05)

OR 1.35 (0.43 to 4.29)

RR 1.32 (0.45 to 3.85)

RR 1.32 (0.45 to 3.85)

Early versus late drain removal

Drain‐related complications

‐: not available; OR: odds ratios; RR: risk ratio; RD: risk differences

Figuras y tablas -
Table 1. Sensitivity analyses
Comparison 1. Drain use versus no drain use

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1.1 Mortality (30 days) Show forest plot

4

1055

Risk Ratio (M‐H, Random, 95% CI)

0.78 [0.31, 1.99]

1.2 Mortality (90 days) Show forest plot

2

478

Risk Ratio (M‐H, Random, 95% CI)

0.23 [0.06, 0.90]

1.3 Intra‐abdominal infection Show forest plot

4

1055

Risk Ratio (M‐H, Random, 95% CI)

0.97 [0.52, 1.80]

1.4 Wound infection Show forest plot

4

1055

Risk Ratio (M‐H, Random, 95% CI)

0.98 [0.68, 1.41]

1.5 Morbidity Show forest plot

4

1055

Risk Ratio (M‐H, Random, 95% CI)

1.03 [0.94, 1.13]

1.6 Length of hospital stay (days) Show forest plot

3

876

Mean Difference (IV, Random, 95% CI)

‐0.14 [‐0.79, 0.51]

1.7 Length of hospital stay Show forest plot

1

Other data

No numeric data

1.8 Additional open procedures for postoperative complications Show forest plot

4

1055

Risk Ratio (M‐H, Random, 95% CI)

1.33 [0.79, 2.23]

1.9 Additional radiological interventions for postoperative complications Show forest plot

3

660

Risk Ratio (M‐H, Random, 95% CI)

0.87 [0.40, 1.87]

Figuras y tablas -
Comparison 1. Drain use versus no drain use
Comparison 2. Active drain versus passive drain

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

2.1 Mortality (30 days) Show forest plot

2

382

Risk Ratio (M‐H, Random, 95% CI)

1.23 [0.30, 5.06]

2.2 Intra‐abdominal infection Show forest plot

2

321

Risk Ratio (M‐H, Random, 95% CI)

0.87 [0.21, 3.66]

2.3 Wound infection Show forest plot

2

321

Risk Ratio (M‐H, Random, 95% CI)

0.92 [0.44, 1.90]

2.4 Morbidity Show forest plot

2

382

Risk Ratio (M‐H, Random, 95% CI)

0.97 [0.53, 1.77]

2.5 Length of hospital stay (days) Show forest plot

2

321

Mean Difference (IV, Random, 95% CI)

‐0.79 [‐2.63, 1.04]

2.6 Additional open procedures for postoperative complications Show forest plot

2

321

Risk Ratio (M‐H, Random, 95% CI)

0.44 [0.11, 1.83]

Figuras y tablas -
Comparison 2. Active drain versus passive drain
Comparison 3. Early versus late drain removal

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

3.1 Mortality (30 days) Show forest plot

3

399

Risk Ratio (M‐H, Random, 95% CI)

0.99 [0.06, 15.45]

3.2 Intra‐abdominal infection Show forest plot

2

285

Risk Ratio (M‐H, Random, 95% CI)

0.44 [0.22, 0.89]

3.3 Wound infection Show forest plot

2

285

Risk Ratio (M‐H, Random, 95% CI)

1.32 [0.45, 3.85]

3.4 Morbidity Show forest plot

2

258

Risk Ratio (M‐H, Random, 95% CI)

0.49 [0.30, 0.81]

3.5 Length of hospital stay (days) Show forest plot

3

399

Mean Difference (IV, Random, 95% CI)

‐2.20 [‐3.52, ‐0.87]

3.6 Hospital costs Show forest plot

2

258

Std. Mean Difference (IV, Random, 95% CI)

‐0.22 [‐0.59, 0.14]

3.7 Additional open procedures for postoperative complications Show forest plot

3

399

Risk Ratio (M‐H, Random, 95% CI)

0.77 [0.28, 2.10]

3.8 Additional radiological interventions for postoperative complications Show forest plot

1

144

Risk Ratio (M‐H, Random, 95% CI)

1.00 [0.21, 4.79]

Figuras y tablas -
Comparison 3. Early versus late drain removal