Randomized controlled trial: The reasons for selecting only randomized controlled trials and not observational studies are that:

·         There are enough, well conducted RCTs that have addressed our review question.

·         RCTs by virtue of randomization effectively controls for known and unknown confounders. In our review question, main confounders that we may encounter are those that are associated with bleeding tendencies. Hence, RCTs by controlling for these confounders will help us arrive at an unbiased effect estimate.

All types of RCTs namely parallel group, cross over, N - of 1 etc. will be considered eligible for the review. Only human studies will be included. Only English articles will be included as no one in our group working on the review is familiar with any other language. We will only include articles where the intervention allocation was truly random. Quasi-randomized or any other type of non-random intervention allocation will be criteria for exclusion of the RCT. We will include RCTs conducted in any part of the world. We will include any RCT that has used LMW heparin that has been approved by the FDA. RCTs that have investigated LMW heparins not currently approved by FDA will not be included in the review. We will not restrict RCTs based on when the study was conducted. Any RCT relevant to our topic that can be retrieved using the databases that we have decided to search, irrespective of when it was conducted, will be considered eligible. RCTs will be included irrespective of study duration, sample size, presence or absence of run in period, duration of run in period. RCTs will be included irrespective of whether ethics approval or trial registration was mentioned in the manuscript. However, these points will be considered in the risk of bias assessment.

We will include only ESRD patients receiving chronic, intermittent, out-patient hemodialysis. The reasons for selecting this patient population are the following:

·         Chronic: dialysis is also a treatment modality for acute conditions such as acute renal failure, acute poisoning etc. In this review we will only include patients receiving chronic dialysis for ESRD. Though we are not defining chronicity by any predefined length of time a patient had to have received hemodialysis, it is left to the authors of the included papers to have their own definitions to explain the concept described here.

·         Intermittent: Patients’ receiving continuous dialysis and continuous venovenous hemofiltration are not included as there are not many RCTs conducted in these population of patients that had addressed our question and also to prevent heterogeneity we decided to exclude these patients.

·         Outpatient: we are excluding patients receiving home dialysis and hospitalized patients to maintain homogeneity, and to prevent confounding due to factors that may increase the risk of bleeding.

·         Hemodialysis: We are not including patients receiving peritoneal dialysis as anti-coagulants are generally not administered to these patients before dialysis.

The diagnosis of ESRD should have been physician (primary care physician or a nephrologist) made. All adult patients aged > 18 years, all races, both males and females will be included for the review. We will exclude patients with hyper-coagulable states.  

We will be including all studies that have used any analogue of low molecular weight heparin that is approved for use in the United States by the FDA (Food and Drug Administration). This includes but not limited to Dalteparin, Enoxaparin, and Tinzaparin. Studies will not be considered ineligible based on route of administration, dose, duration of intervention, or frequency of administration. Further, we will not exclude studies based on when the LNM heparin was administered relative to the dialysis session (before or after dialysis or the previous day etc.). We will not exclude studies based on route of administration, dose of administration, duration, frequency of administration of LMW heparin. However, we will exclude studies where LMW heparin was administered to patients not for the indication of anti-coagulation for hemodialysis but for therapy of another condition such as deep vein thrombosis, pulmonary embolism etc. We will also exclude articles that have used LMW heparin as lock solution.

The comparison intervention should be conventional unfractionated heparin only. Since there are a lot of RCTs that have compared LMW heparin with unfractionated heparin we believe we will have good number of included articles in our review and also for a meta-analysis. Hence, by consensus we have decided to exclude all other comparison interventions such as citrate, other analogues of LMW heparin, direct thrombin inhibitors (example: argotroban), vitamin K antagonists (warfarin), anti-platelets (aspirin, clopidogrel) and any other anti-coagulant with any other mechanism of action. We are also not including passive interventions such as no interventions, placebo, and sham treatment. However, we will not exclude studies based on dose, route, frequency, and duration of administration of unfractionated heparin. We will exclude articles that have used unfractionated heparin as lock solution.

For including studies into the review we do not exclude articles based on outcomes. So long as all other criteria mentioned above are satisfied we will include the studies irrespective of the outcomes addressed in the trials.

Primary outcomes: For outcomes abstracted from included trials we plan to predominantly concentrate on clinically relevant outcomes.

Secondary outcomes

Pubmed search: The two main concepts we will focus for our Pubmed search are: 1) Anticoagulants and 2) dialysis. We will retrieve all MeSH terms related to these two concepts and also all possible key words from title, abstracts and all fields based on clinical knowledge, from articles used in the review by Lim et al., micromedix (looking for various versions of heparin used in the United States) and from expert opinion from hematology and nephrology friends. The third step in our search, we will use the highly sensitive Cochrane search for RCTs. We will then combine steps 1 AND 2 AND 3. From the final list of citations that we retrieve, we will recheck to see if all at least the articles that were used by Lim et al. were retrieved by our search strategy. We expect to see 16 of the 17 articles retrieved by our search. The 1 article that will not retrieved by our search is the one that is not indexed to Pubmed. Our final combined Pubmed Search stragety will be as follows:

1) ("randomized controlled trials as topic"[MeSH Terms] OR ((randomized controlled trial[pt] OR controlled clinical trial[pt] OR randomized[tiab] OR placebo[tiab] OR "drug therapy"[Subheading] OR randomly[tiab] OR trial[tiab] OR groups[tiab]) NOT ("animals"[MeSH Terms] NOT "humans"[MeSH Terms]))) - Highly sensitive cochrane search strategy for RCTs

AND

2) ("Anticoagulants"[Mesh] OR "Anti-coagulants"[All Fields] OR "Anti-coagulant"[All Fields] OR “anticoagulant”[all fields] OR “antithrombin”[all fields] OR "Anticoagulant Agents"[All Fields] OR "Anticoagulant Agents"[All Fields] OR "Anticoagulant Drugs"[All Fields] OR "Antithrombins"[Mesh] OR "Antithrombins"[All Fields] OR "Heparin, Low Molecular Weight"[MESH] OR “Adomiparin”[all fields] OR “antixarin”[all fields] OR “ardeparin”[all fields] OR “bemiparin”[all fields] OR “certoparin”[all fields] OR “cy 222”[all fields] OR “danaparoid”[all fields] OR “deligoparin”[all fields] OR “embolex”[all fields] OR “idrabiotaparinux”[all fields] OR “idraparinux”[all fields] OR “livaparin”[all fields] OR “minolteparin”[all fields] OR “monoembolex”[all fields] OR “parneparin”[all fields] OR “rd 11885”[all fields] OR “reviparin”[all fields] OR “semuloparin”[all fields] OR “tafoxiparin”[all fields] OR “tedelparin”[all fields] OR “logiparin”[all fields] OR “eurodal”[all fields] OR “boxol”[all fields] OR “low liquemine”[all fields] OR “enoparin”[all fields] OR “decipar”[all fields] OR “henoxil”[all fields] OR “hepaclex”[all fields] OR “lomoh”[all fields] OR “nu-mox”[all fields] OR “plaucina”[all fields] OR "Dalteparin"[MESH] OR "Enoxaparin"[MESH] OR "Heparin, Low Molecular Weight"[all fields] OR "LMWH"[all fields] OR "Low Molecular Weight Heparin"[all fields] OR "Low-Molecular Weight Heparin"[all fields] OR "Low-Molecular-Weight Heparin"[all fields] OR "Low Molecular-Weight Heparin"[all fields] OR "LMW Heparin"[all fields] OR "LMW-Heparin"[all fields] OR "Tinzaparin"[all fields] OR "Innohep"[all fields] OR "tinzaparin sodium"[All Fields] OR "innohep"[All Fields] OR "logiparin"[All Fields] OR "Fondaparinux"[all fields] OR "Arixtra"[all fields] OR "Dalteparin"[all fields] OR "Fragmin"[all fields] OR "Dalteparin Sodium"[all fields] OR "Tedelparin"[all fields] OR "Kabi-2165"[all fields] OR “kabi2165”[all fields] OR “kabi 2165”[all fields] OR "Kabi 2165"[all fields] OR "Fragmin"[all fields] OR "Fragmine"[all fields] OR "Dalteparin Sodium"[all fields] OR "Sodium, Dalteparin"[all fields] OR "FR-860"[all fields] OR "FR 860"[all fields] OR “FR860”[all fields] OR "Enoxaparin"[all fields] OR "Lovenox"[all fields] OR "Enoxaparin"[all fields] OR "Enoxaparine"[all fields] OR "PK-10,169"[all fields] OR "PK 10,169"[all fields] OR "PK-10169"[all fields] OR "PK 10169"[all fields] OR "PK10169"[all fields] OR "EMT-967"[all fields] OR "EMT 967"[all fields] OR "EMT-966"[all fields] OR "EMT 966"[all fields] OR “Clexan”[all fields] OR "Clexane"[all fields] OR "Cutenox"[all fields] OR "Dripanina"[all fields] OR "Enoxaparin Sodium"[all fields] OR "Klexane"[all fields]): Search Strategy for anti-coagulants and heparin.

AND

3) ("Renal Dialyses"[all fields] OR "Renal Dialysis"[MeSH] OR "hemodialysis"[all fields] OR "Hemodialyses"[all fields] OR "Extracorporeal Dialyses"[all fields] OR "Haemodialysis"[all fields] OR "Dialysis"[MeSH] OR "Dialysis"[all fields] OR "Microdialysis"[MeSH] OR "Microdialysis"[all fields] OR "Hemodiafiltration"[mesh] OR "hemodiafiltration"[all fields] OR "hemo-dialysis"[all fields] OR "hemo-dialyses"[all fields] OR "haemodialyses"[All Fields] OR “renal replacement therapy”[all fields] OR "RRT"[all fields]): Search strategy for dialysis.

Final Search: #1 AND #2 AND # 3

Cochrane CENTRAL Search: Our Cochrane search strategy will be as follows:

#1 MeSH descriptor Anticoagulants explode all trees     

#2 MeSH descriptor Antithrombins explode all trees

#3 MeSH descriptor Heparin, Low-Molecular-Weight explode all trees

#4 (anticoagulant agent) OR heparin OR (low molecular weight heparin) OR Adomiparin OR (antixarin) OR (ardeparin) OR (bemiparin) OR (certoparin) OR (cy 222) OR (dalteparin) OR (danaparoid) OR (deligoparin) OR (embolex) OR (enoxaparin) OR (fondaparinux) OR (idrabiotaparinux) OR (idraparinux) OR (livaraparin calcium) OR (minolteparin) OR (monoembolex) OR (nadroparin) OR (parnaparin) OR (rd 11885) OR (reviparin) OR (semuloparin) OR (tafoxiparin) OR (tedelparin) OR (tinzaparin) OR (heparin derivative)/exp OR (anticoagulant agent)/exp OR (Anticoagulants) OR (Anti-coagulants) OR (Anti-coagulant) OR (Anticoagulant) OR (Antithrombins) OR (Low Molecular Weight Heparin) OR (Low Molecular Weight Heparins) OR (LMWH) OR (Low-Molecular Weight Heparin) OR (Low-Molecular Weight Heparins) OR (Low Molecular-Weight Heparin) OR (Low Molecular-Weight Heparins) OR (LMW Heparin) OR (LMW-Heparin) OR (tinzaparin) OR (innohep) OR (logiparin) OR (Arixtra) OR (Fragmin) OR (Tedelparin) OR (Kabi-2165) OR (Kabi 2165) OR (Kabi2165) OR (Fragmine) OR (FR-860) OR (FR 860) OR (FR860) OR (Eurodal) OR (Boxol) OR (Low Liquemine) OR (Lovenox ) OR (Enoxaparine) OR (PK-10,169)  OR (PK 10,169) OR (PK10,169) OR (PK-10169) OR (PK 10169) OR (PK10169) OR (EMT-967) OR (EMT 967) OR (EMT967) OR (EMT-966) OR (EMT 966) OR (EMT966) OR (Clexa) OR (Clexane) OR (Cutenox) OR (Decipar) OR (Dripanina) OR (Enoparin) OR (Flenox) OR (Henoxil) OR (Hepaclex) OR (Klexane) OR (Lomoh) OR (Nu-Rox) OR (Plaucina) OR (Trombenox)

#5 MeSH descriptor Dialysis explode all trees

#6 MeSH descriptor Hemodiafiltration explode all trees

#7 MeSH descriptor Hemodialysis, Home explode all trees

#8 (Renal Dialyses) OR Hemodialysis OR Hemodialyses OR (Extracorporeal Dialyses) OR Dialysis OR Microdialysis OR Hemodiafiltration OR hemodiafiltration OR haemodialyses OR dialyses OR haemodialysis OR (renal replacement therapy)

#9 ((#1 OR #2 OR #3 OR #4) AND (#5 OR #6 OR #7 OR #8)): will be the final combined search strategy for Cochrane central.

EMBASE Search:

Here again we are using the two concepts namely: 1) anti-coagulants and 2) dialysis. We will retrieve all EMTREE terms and key words associated with these concepts. We will further use commercial drug names from micromedix and expert opinion from hematologists and nephrologists to retrieve more keywords. Also, we will use the trials mentioned in the review by Lim et al. to retrieve more keywords.

#1 Anti-coagulant part -
‘anticoagulant agent’/exp OR ‘heparin’/exp OR ‘low molecular weight heparin’/exp OR ‘Adomiparin’ OR ‘antixarin’ OR ‘ardeparin’ OR ‘bemiparin’ OR ‘certoparin’ OR ‘cy 222’ OR ‘dalteparin’ OR ‘danaparoid’ OR ‘deligoparin’ OR ‘embolex’ OR ‘enoxaparin’ OR ‘fondaparinux’ OR ‘idrabiotaparinux’ OR ‘idraparinux’ OR ‘livaraparin calcium’ OR ‘minolteparin’ OR ‘monoembolex’ OR ‘nadroparin’ OR ‘parnaparin’ OR ‘rd 11885’ OR ‘reviparin’ OR ‘semuloparin’ OR ‘tafoxiparin’ OR ‘tedelparin’ OR ‘tinzaparin’ OR ‘heparin derivative’/exp OR ‘anticoagulant agent’/exp OR “Anticoagulants” OR “Anti-coagulants” OR “Anti-coagulant” OR “Anticoagulant” OR “Antithrombins” OR “Low Molecular Weight Heparin” OR “Low Molecular Weight Heparins” OR “LMWH” OR “Low-Molecular Weight Heparin” OR “Low-Molecular Weight Heparins” OR “Low Molecular-Weight Heparin” OR “Low Molecular-Weight Heparins” OR “LMW Heparin” OR “LMW-Heparin” OR “tinzaparin” OR “innohep” OR “logiparin” OR “Arixtra” OR “Fragmin” OR “Tedelparin” OR “Kabi-2165” OR “Kabi 2165” OR “Kabi2165” OR “Fragmine” OR “FR-860” OR “FR 860” OR “FR860” OR “Eurodal” OR “Boxol” OR “Low Liquemine” OR “Lovenox ” OR “Enoxaparine” OR “PK-10,169”  OR “PK 10,169” OR “PK10,169” OR “PK-10169” OR “PK 10169” OR “PK10169” OR “EMT-967” OR “EMT 967” OR “EMT967” OR “EMT-966” OR “EMT 966” OR “EMT966” OR “Clexa” OR “Clexane” OR “Cutenox” OR “Decipar” OR “Dripanina” OR “Enoparin” OR “Flenox” OR “Henoxil” OR “Hepaclex” OR “Klexane” OR “Lomoh” OR “Nu-Rox” OR “Plaucina” OR “Trombenox”

#2 Dialysis part -
‘dialysis’/exp OR dialysis OR ‘hemodialysis’/exp OR hemodialysis OR ‘haemodialysis’/exp OR haemodialysis OR ‘extended daily dialysis’/exp OR ‘home dialysis’/exp OR ‘renal replacement therapy’/exp OR ‘renal replacement therapies’/exp OR ‘hemofiltration’/exp OR hemofiltration or ‘hemodiafiltration’/exp OR hemodiafiltration OR ‘hemo-dia-filtration’/exp OR hemo-dia-filtration OR ‘haemofiltration’/exp OR haemofiltration OR haemodiafiltration OR ‘haemodiafiltration’/exp OR ‘intermittent dialysis’/exp OR ‘outpatient dialysis’/exp OR ‘out-patient dialysis’/exp OR ‘outpatient hemodialysis’/exp OR ‘out-patient hemodialysis’/exp OR ‘outpatient haemodialysis’/exp OR ‘out-patient haemodialysis’/exp OR ‘outpatient hemofiltration’/exp OR ‘out-patient hemofiltration’/exp OR ‘out-patient hemodiafiltration’/exp OR ‘outpatient hemodiafiltration’/exp OR ‘renal dialysis’/exp OR ‘extracorporeal dialysis’/exp OR ‘extra-corporeal dialysis’/exp OR microdialysis OR ‘microdialysis’/exp OR micro-dialysis OR ‘micro-dialysis’/exp

#3 RCT part -
We combined (using OR) the ideas of the most sensitive and middle sensitive approaches from the Wong paper on EMBASE searching strategies for RCTs
random*:de,lnk,ab,ti,au OR random:de,lnk,ab,ti,au OR 'clinical trial':de,lnk,ab,ti,au OR 'clinical trials':de,lnk,ab,ti,au OR 'health care quality'/exp OR 'health care quality' OR 'treatment outcome'/exp OR ‘treatment outcome'

                                     FINAL COMBO  #1 and #2 and #3 = final Embase search.

For this review we will be limited to searching only these above mentioned 3 databases. We will not attempt any grey literature search or hand searching or searching conference proceedings or contacting individual authors to retrieve more citations due to time constraints in this course.

Once we retrieve all citations as mentioned above from the three databases as mentioned above we will import them to EndNote directly from these three databases. Once we import them into EndNote we will search for duplicates by title, author, date, volume, and journal. Thus identified duplicates will be removed and then the final list of citations after removal of duplicates will be finalized. Then, these final citations will be exported to Microsoft excel work book. Three reviewer pairs will be created. Reviewer pairs will be created such that one will be a methodology expert and the other will be a clinical expert. The total included citations after excluding duplicates will be randomly divided among the three pairs of reviewers such that each pair gets equal number of citations to screen. Screening will be done using Microsoft excel. Now, within each pair, reviewers will independently screen abstracts and titles for eligibility to retrieve full text articles. At the stage of abstract and title screening we will screen articles as 'yes' or 'no'. Full texts of all articles marked as 'yes' by both reviewers within a pair will be retrieved. Articles marked 'no' at this stage by both reviewers will be excluded. Articles marked 'yes' by one and 'no' by the other will show up as conflict. Conflicts at this stage will be resolved by consensus. Citations that conflict even after discussion will be considered as 'yes' and full text will be retrieved for these articles. Full text retrieval will be done by re-exporting the 'yes' citations to EndNote. Once full texts of all 'yes' articles from title and abstract screening has been retrieved they will again be randomly sorted and allotted to three reviewer pairs to determine eligibility to include into the final review. Within each pair reviewers will screen full text articles independently and mark them as 'yes' or 'no'. Articles marked 'yes' by both reviewers at this stage will be included for the review. Articles marked 'no' by both reviewers will be excluded. Articles marked 'yes' by one and 'no' by the other will show up as conflict. Conflicts will be resolved by consensus. If resolution does not happen after discussion then it will be resolved by discussing as a group with the entire team and consensus will be reached regarding inclusion into the study.

Included studies will be randomly divided into three groups. Three pairs of reviewers will be abstracting data and one set of included articles will be assigned to each pair. Within each pair, both reviewers will independently abstract data from the included trials and registering the abstracted data into data abstraction forms which will be created in Google docs. Each reviewer will be blinded with regards to the entry of the other reviewer. After independent data abstraction is complete the two reviewers in each pair will compare their responses. Discrepancies at this stage will be resolved by consensus. Conflicts at this stage, beyond discussion, will be resolved by group discussion with the entire team. After confirming completeness and accuracy of data abstraction, data will be independently entered into software 'Revman 5.0' using double data entry by two reviewers.

Abstracted data will include:
Methods: type of randomized trial (including: year(s) of conduct, total sample size, study duration, date study commenced, place or region of study), study methodology (including eligibility criteria, methods of randomization, type of randomization sequence followed, allocation sequence concealment, and masking, washout period), Participant characteristics: total number, setting (hospital based or free-standing), age, sex, country, race, comorbidities (diabetes, hypertension, bleeding disorders, autoimmune disorders), frequency of dialysis, type of dialysis access (graft, AV fistula, catheter [e.g. Davol, Hickman, Groshong]) Intervention:  Low Molecular Weight Heparin and unfractionated heparin (dose, name of drug, route of drug, timing relative to hemodialysis, frequency of administration), Outcomes: definition of outcome, for all outcomes we will tabulate the number of patients and number of hemodialysis sessions with an event, if available.  When binary outcomes are extracted, the number of patients with the event will be recorded, timing of assessment, method of assessment. For continuous outcomes mean and standard deviation or standard error mentioned in the manuscript will be recorded. Results (sample size completed the study, summary data from each intervention group, loss to follow-up, compliance to intervention, missing participants), ethical issues (consent, institutional review board approval, registration of trial).

The quality of included studies will be assessed by two independent reviewers. When there is a discrepancy, it will be resolved by consensus. Conflicts that remain even after discussion will be resolved by group discussion with the entire team. The studies will be evaluated for the following criteria:

a) Allocation:

·         Sequence generation: will be termed 'adequate' (low risk) if the random number generation is by using a computer generated table, random number sequence table, coin tossing or any other form that is obviously random and 'inadequate' if the sequence generation has a non-random component (birth data, visit date or day to the clinic or alphabetical order or self selection). If the method of randomization is not specified then it will still be termed "inadequate" (high risk).

·         Concealment: Will be termed 'adequate' (low risk) if concealment was due to a central allocation, opaque envelopes etc and 'inadequate' (high risk) if any other method where there is a possibility that the investigator or the participant may know in advance or during allocation the allocation sequence.

b) Masking of investigators and participants with regards to the intervention: will be termed 'adequate' (low risk) if both investigator and the participants were masked and 'inadequate' (high risk) if either or both were unmasked with regards to the intervention identity.

c) Masking of outcome assessment: The study investigators did not obviously control when the outcome was determined eg: review of case records. Then it is considered ‘low risk’. Otherwise, considered ‘high risk’.

d) Incomplete outcome data: the reasons for incompleteness of the outcome data is explained then ‘low risk’. Not explained ‘high risk’

e) Selective reporting: When methods section of the manuscript details other assessment scales only one is reported that is significant (high risk). All scales are reported (low risk).

f) Loss to follow-up and intention to treat analysis: we will assess studies to see the reasons for loss to follow-up and if they have adequately reported all losses to follow-up and if loss to follow-up was random and not due to a specific reason. Intention to treat analysis is said to have been conducted if participants lost to follow-up were still included in the group they were randomized at the study beginning, when data results were analyzed.

All the components will be assessed before deciding the study quality. We will not follow any scoring system to assess quality of the included studies but will determine quality based on the subjective assessment of the reviewers from the three subheadings discussed above. We will have a risk of bias form created in google docs where 2 independent reviewers will assess and record their assessment of the study quality.

We will report relative risks and 95% confidence intervals of the relative risks for all dichotomous outcomes. The reason is that we are using only RCTs and hence relative risks will be mentioned in the RCTs and they will be free of bias due to confounding. For continuous outcomes such as lipid profile levels and compression times we will calculate means and standard deviations as they are commonly used summary statistics for continuous variables.

The unit of analysis will be individual participants.

Clinical heterogeneity will be determined based on clinical knowledge as assessed by the clinical experts in the team based on patient characteristics, intervention characteristics, co-morbid conditions etc.

Methodological heterogeneity: will be assessed by methodology experts in the team based on run in period, duration of study, adequacy of randomization etc.

Statistical heterogeneity: we plan to use Q, I² and τ² to assess statistical heterogeneity. Q statistics with level of significance less than 0.1 and the degree of overlap between confidence intervals. I² of 25%, 50% and 75% will be considered low, moderate and high heterogeneity respectively. We will assess all three components to decide statistical heterogeneity.

All team members will meet to discuss all these points and by consensus decide about the heterogeneity of included studies.

Studies will be pooled with the random effects model as we expect significant heterogeneity based on clinical variability and methodological variability in the studies that become eligible and will be included in the analysis. However, we will make this decision only after final heterogeneity assessment and if we believe there is not much heterogeneity then we may consider fixed effects model.