IBD-Associated Thromboembolism Case Search Criteria and Organization
We searched PubMed for letters, case reports, case series, and original studies of IBD-associated TE in the English, Spanish, Portuguese, French, and German language literature from January 1997 to January 2007 using the following search criteria: “Colitis, Ulcerative” [MeSH] AND “Thrombosis” [MeSH], “Crohn Disease” [MeSH] AND “Thrombosis” [MeSH], “Inflammatory Bowel Diseases” [MeSH] AND “Thrombosis” [MeSH], “Colitis, Ulcerative” [MeSH] AND “Thromboembolism”[MeSH], “Crohn Disease” [MeSH] AND “Thromboembolism” [MeSH], “Inflammatory Bowel Diseases” [MeSH] AND “Thromboembolism”[MeSH], “Colitis, Ulcerative”[Mesh] AND “Thrombolytic Therapy” [MeSH], “Crohn Disease” [Mesh] AND “Thrombolytic Therapy” [MeSH], “Inflammatory Bowel Diseases” [MeSH] AND “Thrombolytic Therapy” [MeSH], “Ulcerative Colitis Thrombolysis,” “Crohn's Disease Thrombolysis,” and “Inflammatory Bowel Disease Thrombolysis.” We supplemented this electronic search with a hand-search of references from all of the publications identified by these search criteria. Due to the large number of cases treated with AC and the small number of cases treated with CDT, we expanded the search for CDT cases to include all published cases.
Using a standardized data collection form, demographic and clinical data were collected including TE-treatment regimen, radiologic and symptomatic outcomes, treatment-related HCs, new or progressive TE, and death. Cases were excluded if they: 1) did not have information on the treatment regimen, 2) were not treated with CDT or AC, 3) did not have information on the radiologic and/or symptomatic outcome, 4) had concomitant surgical treatment or systemic thrombolysis, or 5) were due to septic thrombophlebitis. Cases were not excluded if: 1) there was no mention of HCs, new TE, or death, as it was assumed that these discrete, treatment-related outcomes would be reported if they had occurred, and 2) an unspecified dose of anticoagulant (most commonly unfractionated heparin [UFH]) was used, as it was assumed that therapeutic, not prophylactic dose, AC would be employed in the treatment of TE.
Two treatment groups were defined a priori: 1) CDT (with or without AC), and 2) AC only cases. If initial treatment consisted of AC only but rescue CDT was later used, then the modality-specific case outcomes would be included in both the AC only and CDT group, each with its respective outcomes. For example, if AC was used initially and CDT was subsequently required, AC would be deemed a failure and CDT would be assigned the ultimate outcome.
If all the retrieved literature cases pertaining to a particular anatomic area were treated with only one treatment modality (e.g., thoracocervical venous, splanchnic arterial, and coronary arterial TE), all the cases for that anatomic area were excluded since comparative outcomes for each of the treatment modalities would not be available. Cases conforming to the inclusion and exclusion criteria were classified according to treatment modality and stratified by the anatomic site of TE.
Outcomes and Definitions
Demographic and clinical variables consisting of age, sex, IBD type, IBD activity, presence of pre-TE treatment hematochezia, presence of prothrombotic risk factors, and the site of TE were recorded. Treatment-related variables and outcomes were also recorded, as explained below.
“Radiologic outcome” was divided into three categories: complete (grade III lysis), partial (grade II lysis), and no (grade I lysis) resolution. While complete resolution and no resolution were quantified objectively by the respective authors, we chose to define “partial” (grade II) resolution as >50% (but not complete) radiologic resolution, and in cases where a percent value was not provided, “significant,” “substantial,” “good,” or “fair” radiologic resolution were considered surrogates for “partial.”
“Symptomatic outcome” was similarly divided into three categories: complete resolution, partial resolution, and no resolution. Partial resolution was subjectively defined as cases with “significant,” “substantial,” “good,” or “fair” symptomatic resolution.
If data on either the radiologic or symptomatic outcome were not provided, this outcome was designated as not available (“NA”) in the evidence tables. If neither of these two outcomes was reported, the case was excluded from this study, as described previously.
“Treatment-related bleeding” (HCs) was separated into GI versus non-GI sites. These events were recorded as discrete (yes or no), as were “new or progressive TE” and “death.” Treatment-related non-GI bleeding was defined, based on the International Society of Haemostasis and Thrombosis criteria for major bleeding, as blood loss that resulted in a decline in the hemoglobin of 2 g/dL or more or required transfusion of two or more units of blood, or intracranial, retroperitoneal, or intraocular bleeding.7 Treatment-related GI bleeding was defined as new or subjectively worsened hematochezia. A new TE event was defined as an objectively documented TE that occurred in a noncontiguous vascular location, and progressive TE was defined as an objectively documented extension of the existing TE. If the source case did not mention bleeding, TE complications (such as new or distal TE), or death, it was assumed that these outcomes did not occur.
Chi-square test or Fisher's exact test, as appropriate based on group and cell size, was used to identify baseline differences in discrete variables between treatment groups. Analysis of variance was used to assess for differences in age.
Treatment-related outcomes were compared between the two treatment groups with respect to 1) the specific anatomic areas of involvement (for detailed information) and 2) the aggregate of anatomic areas (for better overall clinical impression). P < 0.05 was considered statistically significant.