Hemorrhage

We constructed all-cause rehospitalization within 30 days of discharge from Medicare claims (34 –36 ). Other variables drawn from Medicare files, included patient age, gender, race, Medicaid status, initial Medicare enrollment due to disability, index hospitalization length of stay and discharge to a skilled nursing facility. Race was categorized into ‘White’, ‘Black’, and ‘Other’ based upon the beneficiary race code. Each patient’s Centers for Medicare and Medicaid Services hierarchical condition category (HCC) score, calculated from all outpatient and inpatient claims over the 12 months prior to the index hospitalization, was included as a risk adjustment measure (38 (link)). Comorbid conditions were identified using Elixhauser methods, incorporating data from the index hospitalization and from all hospitalizations and physician claims during the year prior to the index hospitalization (39 (link)). Of the comorbidities identified using this approach, 17 had frequencies of greater than 5% in the sample and were included as indicators. Comorbidities occurring less often were compiled into an ‘other comorbidity’ indicator and included alcohol/drug abuse, rheumatoid arthritis/collagen vascular disease, chronic blood loss anemia, liver disease, lymphoma, metastatic cancer, solid tumor without metastases, paralysis, psychoses and peptic ulcer disease. We assessed rurality of each patient’s zip code of residence using the US Department of Agriculture’s Rural/Urban Commuting Area (RUCA) Codes, grouped into categories of “urban core areas,” “suburban areas,” “large town areas,” and “small town/isolated rural areas” (40 , 41 ). Index hospital characteristics, including Medicare geographic region, for-profit status and medical school affiliation, were drawn from the Medicare provider of services file corresponding to the patient’s index hospitalization date (42 ). We estimated annual Medicare discharge volume for each hospital by multiplying the number of claims from each hospital in the 5% national sample, by 20. We then grouped hospitals into low, middle and high volume tertiles. About one percent of our sample was missing race data (n=291), and less than 3% were missing hospital medical school affiliation (n=777) and for-profit status (n=777). There were no missing data for other patient-level variables.

The methods and search strategy for Mini-Sentinel systematic reviews are described in the accompanying manuscript by Carnahan.¹² Briefly, PubMed and Iowa Drug Information Service (IDIS) searches of the English language literature were performed to identify studies published between 1990 and 2010 that evaluated the validity of algorithms for identifying CVAs and/or TIAs in administrative data. Search terms related to administrative data are described in detail by Carnahan¹² and were included in all Mini-Sentinel systematic review searches. In addition, the following key words were used as PubMed search terms for the CVA/TIA review: (“Brain Ischemia”[Mesh] OR “Basal Ganglia Cerebrovascular Disease”[Mesh]) OR “Carotid Artery Thrombosis”[Mesh]) OR “Intracranial Embolism and Thrombosis”[Mesh]) OR “Intracranial Hemorrhages”[Mesh]) OR “Stroke”[Mesh]) OR “Vasospasm, Intracranial”[Mesh]. The IDIS search included specification of the following terms: 435. or 432. or 433.1 or 434. or 436. (NOTE: 435. ISCHEMIA, CEREBRAL, TRANSIENT, 432. HEMORRHAGE, INTRACRANIAL NEC, 433.1 EMBOLISM/THROMBOSIS, CAROTID, 434. EMBOLISM/THROMBOSIS, CEREB, 436. DISEASE, CEREBROVASCULAR NEC) for the disease and “ischemi*” or “intracranial” or “stroke” in the abstract.
Two study investigators independently reviewed the abstracts to identify potentially relevant articles for retrieval; articles identified as potentially relevant by either investigator were retrieved. The study investigators independently reviewed the articles with a goal of identifying validation of CVAs or TIAs described in the article itself or from the reference section of the article if it included validation studies. Citations from the article’s references were selected for full-text review if they were cited as a source for the algorithm to identify CVAs or TIAs, or were otherwise deemed likely to be relevant. Discrepancies regarding the inclusion of a study for the review report were resolved by consensus following the independent reviews.
Mini-Sentinel investigators were surveyed to request information on any published or unpublished studies that validated an algorithm to identify CVAs or TIAs in administrative data. These studies were similarly reviewed by two study investigators to determine their relevance.
A single investigator abstracted information on the study design and population, algorithm, and validation statistics for each study. The data were confirmed by a second investigator for accuracy. Based upon the specific outcomes reported, we categorized studies by the following CVA/stroke subtypes: acute events including 1) strokes, 2) TIAs, and 3) intracranial bleeds (intracerebral hemorrhage and subarachnoid hemorrhage), and 4) the composite endpoints of stroke/TIA or cerebrovascular disease (including prevalent disease).

MRI lesions for acute or subacute volumes were measured on DWI or MTT images. Subacute infarcts on CT images were measured using windows/levels of 80/20 or 30/30 for Houndsfield units2 (link). Absolute infarct were measured by E.S.R. using Alice software (Parexel Corp.). DWI and MTT lesions volumes were measured by P.W.A. using Analyze 7.0 software (Analyze Direct, KS). The ischemic ROIs were visually segmented to determine the volume. Stroke volumes ranged over 3 orders of magnitude from 0.25 to 403 cm³ by computerized planimetry. Observers (J.R.S. and L.R.G) blinded to planimetric data measured lesions in three perpendicular axes. The slice with the largest lesion was first selected by eye. The longest lesion axis on this slice was measured with the ruler tool on an AGFA R4 Workstation with Impax Select software (v5205.0.0.1). A second line was drawn perpendicular to the first at the widest dimension. These two measurements were called the x (A) and y (B) axes. A third axis, the z (C) axis, was computed by multiplying the number of slices by slice thickness (Fig. 1). The scan slice for CT was 5mm. MRI thickness ranged from 6-7 mm. Time to perform these three measurements was less than 1 minute.
For analysis of DWI and MTT mismatch, mismatch was defined as, (MTT volume/DWI volume ≥ 1.2. A parameter of 20% mismatch was chosen based on trials using “eyeball estimate” of 20% mismatch and may not be the optimal mismatch1 (link), 25 (link). Absolute volumes measured by planimetry were compared to estimates of ellipsoid ABC/2 (see below) for DWI volume and MTT volume. Euclidean Shapes (Fig. 1)
We tested the ellipsoid model both unadjusted and the adjusted model used for ICH20 (link), as well as sphere, cylinder and bicone. For the hemorrhage-adjusted ellipsoid model according to Kothari et al.20 (link), all slices with lesion volume less than 25% of the slice with the maximum lesion volume were not counted in the z axis. For slices in which the lesion volume was between 25-75%, the slice thickness was multiplied by 0.5, and for slices where the lesion volume was >75%, the slice thickness was multiplied by 1. For all geometric models, π was simplified to 3, for ease of clinical assessment. Formula simplifications for A, B, and C axes are below:
Ellipsoid model: $$\mathrm{V}=4∕3\pi {\mathrm{r}}_{\mathrm{A}}{\mathrm{r}}_{\mathrm{B}}{\mathrm{r}}_{\mathrm{C}}=4∕3\times 3\times (\mathrm{A}∕2)\times (\mathrm{B}∕2)\times (\mathrm{C}∕2)\times =\mathrm{ABC}∕2$$
Where
A= longest dimension in axis x
B= longest perpendicular dimension to axis x (y)
C= total length in z dimension
Sphere model: $$\mathrm{V}=4∕3\pi {\mathrm{r}}^3=4∕3\times 3\times (\mathrm{D}∕2{)}^3=(\mathrm{D}{)}^3∕2={\mathrm{D}}^3∕2$$
Where
D= longest measurement of A, B or C
Cylinder model: $$\mathrm{V}=\mathrm{h}\pi {\mathrm{r}}^2=\mathrm{h}\times 3\times (\mathrm{D}∕2{)}^2=3∕4(\mathrm{D}{)}^2\mathrm{h}=3∕4{\mathrm{D}}^2\mathrm{C}$$
Where
D= longest measurement of A or B
h= C
Bicone model: $$\mathrm{V}=(1∕3\mathrm{h}\pi {\mathrm{r}}^2)2=(1∕3\times 3\times \mathrm{h}\times (\mathrm{D}∕2{)}^2)2=(\mathrm{h}{\mathrm{D}}^2∕4)2\text{or}{\mathrm{D}}^2\mathrm{C}∕4$$
Where
D= longest measurement of A or B
h= C/2