Ancillary Information Systems (Introduction to Medical Informatics) (http://www.cpmc.columbia.edu/edu/textbook) LAST REVIEWED: 19 October 1996 Laboratory Information System as example GENERAL FACTS clinical laboratory is a microcosm of health care with complex problems and difficult implementation every hospital has a set of clinical labs initially manually at bedside then manually in lab today automated 10-15% of health care dollars ($10s million here) volume - 1950s -> 1970s, # procedures far outstripped # admissions or # technicians [see overhead; why] decreased costs convenience new, better tests malpractice suit avoidance brief history 1957: first automated analyzer initially - manual transciption of output then - machine-readable output then - direct instrument-to-computer interface mid-60s: administrative and mgmt functions ex: ordering supplies; monitoring turnaround 1970: computers incorporated into analyzer (e.g., SMAC) can help validate results before release to ancillary computer DRGs [define] have changed laboratories from a profit center to a cost center try to reduce the number of tests (instead of increase) central admin. has tighter control over the budget CLINICAL LABORATORY doctor and patient's view: why order a test screen for a disease make a diagnosis guide therapy estimate prognosis function of lab acquire information validate information interpret information communicate information lab judges its quality based on accuracy (free from error) precision (sharply defined) timeliness availability organization of labs chemistry - chemical analyses on body fluids hematology - cell counts microbiology - cultures, resistance of micro-organisms cytology - neoplastic cells immunology/serology - presence of antibodies anatomic pathology - microscopic and gross specimens blood bank - blood type, compatibility INFORMATION FLOW data flow view patient physician | ^ extralaboratory V | cycle specimen received report generation | ^ intralaboratory V | cycle work sheet -> procedure -> result calculation process view (importance of detail; # persons involved) order entry specimen labelling specimen collection specimen transport accessioning [why] delivery of report that specimen was received test requesting (within lab) specimen preparation and technician worksheets perform test analysis data collection data review and correction release of initial results (by technician) generation of preliminary reports, charts delivery of preliminary reports (to provider) including calling for dangerous results further data collection, review, and correction release of verified results (by technician) generation and delivery of final reports (to provider) possible need to "undo" erroneous prelim report erroneous data: wrong pt, mistake reading, machine problem effect on decision support what if sent alert on high creatinine send second alert? (beep, email); undo first? LABORATORY INFORMATION SYSTEMS (LISs) 1989: 40% of hospitals have an LIS first area computerized after billing little tradition to hold computerization back (in 1960s automated lab were just starting) used to be profitable revenue center cost of LIS $1500/bed ($2.25 million here) computer has a role in each process mentioned above test ordering and result reporting patient and specimen identification - preprint labels; specimen collection lists data processing and record keeping - track and route specimens; allocate technicians data acquisition - two basic types: quantitative and visual - quant: process raw data (voltage; light conductance) into digital data and report to LIS report generation: allows different "views" of data - trend analysis - summary reports quality control standard samples in production run = calibration absolute limit check: absurdly outside usual range? delta check: absurd change from last result? managerial reporting test requests completion time productivity, e.g., turnaround time CAP standards (1 CAP = 1 clerk minute) procedures assigned CAP values compare ideal time (CAP x # procedures) vs time used by technician in order to assess productivity and discover bottlenecks inventory control observation on evolution of LISs: despite similar data flow and process flow, different areas computerized at different times chemistry, hematology has automated analytical machines generate large amount of data mostly numeric results possible to build instrument-computer interfaces therefore early automation microbiology, blood bank, anatomic path relies on manual labor => data entry problem more complex process (eg, microbiology) multistep diagnostic work over long periods of time for one specimen managing encyclopedic set of bacteria complex nested data many failed naive attempts to move chem into micro therefore late automation physician's history and physical are closer to microbiology lab than to chemistry lab ADVANTAGES analytic results improved QC identify reference population decreased clerical errors result reporting flexible formats timely and available administration accurate workload and inventory optimize workflow (eg, count #tests per hour of day) personnel less tedium decision-support: integration flag abnormal results list possible causes suggest need for f/u detect trends, changes [should this be in lab system or CIS?] reduce costs via decrease labor costs minimize number of tests contribute to decentralization SYSTEMS contemporary LISs very few hospitals have all labs under one LIS few LISs integrated with HIS mostly vendor-supplied systems flexibility was key point in their early success now about 40 vendors systems improved via competition went from 50% success in 1970 to 90% in 1987 differentiation: support, price, special features specimen identification is a major issue CPMC as example of custom built system 10 years ago few vendors, worse products therefore developed custom system staff 4 programmers cost 1/2 that of turnkey system at the time major risk of key staff leaving difficulty of maintenance most attempts at custom systems fail or partial usually found in very small or very large labs CPMC will now replace with "turnkey" (vendor) system cost about $1 million for software [what do you think happened?] major changes in data organization, coding slowly modify to acceptible form related reading: Smith JW, Svirbely JR. Laboratory information systems. M.D. Computing 1988;5(1):38-47.