STRIDE Methodology and Discovery
A groundbreaking research investigation at Mount Sinai, entitled “STRIDE” contains several key elements. First we devised a method of performing a computer survey of all the diagnoses of patients admitted to Mount Sinai over time, to identify individual patients with medical diagnoses suggestive of immune deficiency, but who do not have another known reason for these illnesses (such as HIV cancer chemotherapy, debility, aging etc). These diagnoses (complied in the International Classification of Disease codes, Ninth Revision, ICD-9) are used to categorize every medical encounter, and are uniformly used by all physicians. Because the immune system plays an essential role in protecting against infections, many of the conditions that are associated with immune defects, involve infections of the nose, throat, skin, lungs or other organs. In addition, since the immune system is involved in the regulation of normal immunity, in some cases, the conditions that arise are autoimmune, or inflammatory.
Using these diagnostic codes for all these conditions, we developed a scoring algorithm to identify patients who had been given a diagnosis of two or more of the ICD- 9 coded complications associated with immunodeficiency. Using the same ICD codes, we also excluded all subjects with diagnoses that might lead to similar conditions, such as HIV, organ transplant, chemotherapy, etc. Our data showed that for every 1,000 patients admitted to Mount Sinai Hospital, one or two have a medical history suggestive of immunodeficiency.
Study Targeting Recognition of Immune Deficiency and its Evaluation
INCLUSION CRITERIA: 2 or more of the following as examples:
|Fever of unknown origin
Abnormal loss of weight
Failure to thrive
|Otitis media: chronic
Deep abscess: liver/spleen/rectal
|Lymphadenitis/lymph node enlargement
|Autoimmune hemolytic anemia
Primary thrombocytopenia, ITP
These ICD9 codes are also given relative rankings, to roughly indicate the severity of the conditions. These “score” or “report” cards give us an idea of the severity of the other conditions and suggest what kind of immune defect might be present. Using this system, we found that an ICD-9 based scoring algorithm identified patients who have had multiple illnesses suggestive of immunodeficiency; on testing these subjects, we found that this group contained undiagnosed and numerous minority patients with immunodeficiency. These results have been published.
A second study was then performed to record the kinds of illnesses that patients who were referred by other doctors to Mount Sinai for suspicion or confirmation of Primary Immune Deficiency. Over a two-year period, 237 patients were referred by internists, pediatricians, or allergists for suspected immunodeficiency. immunodeficiency was diagnosed 48%. Patients with immunodeficiency had a median score of 8 (interquartile range, 5 to 13), which was significantly higher (P 0.004) than the median score for those who did not have an immune deficiency (median score, 6; interquartile range, 3 to 10). Using age-adjusted comparisons, patients with immune deficiency also had a median score of 10 (interquartile range, 5 to 17.5), which was higher (P 0.025) than the median score of those without immune deficiency (median score, 8; interquartile range, 4 to 14). This study verified the use of the scoring parameters for locating patients with primary immune deficiency, and has been published.
In ongoing studies we are also using the Statewide Planning and Research Cooperative System (SPARCS) dataset which collects ICD-9 codes for every hospital in and out patient encounter, complete demographics, home zip code, care locations, all services and charges in New York State. Patients are given a stable unique identifier so each can be identified through his/her multiple hospitalizations and all in and outpatient medical encounters. Included in each hospitalization are primary and up to 13 secondary ICD codes applied for the disease all complications experienced in that admission. This study was approved by the SPARCS Department of Health Review Board in January 2011 and will be used to further refine the ICD codes that are in used in the computer program. Our further plans will be to continue to refine the computer algorithm so that it may become a validated instrument for surveying electronic medical records.
Charlotte Cunningham-Rundles, MD, PhD
1425 Madison Avenue,1st Floor
New York, NY 10029