This is an update to a previously reported issue.In the initial report, it was indicated that one rule to evaluate microalbumin had an issue, and it was created 8 years ago.Further investigation indicates the originally reported rule was created only 2 years ago.Further investigation of the customer's configuration also revealed there were additional rules written with the same issue of the qualifier "is numeric" missing.Rules that needed "is numeric" but did not have them are limited to the architect 1 and architect 2 configurations which have the same rules associated with them.If the rules had any of the "greater thans" first and the "less thans" second for evaluation of a particular test code, it would have set the result to the "less than" value.Summary of test codes from the architect analyzers that should have had "is numeric" added to the rules: hcv, core-m, hbsag, hep a, trig, folate, hiv, b12, upro.The original rule that was reported, and the additional rules that were identified as having the similar issue, are now configured properly.Data innovations is still working with the hospital's lab technologist, lab supervisor and pathologist to determine if there was any adverse patient impact.
|
This is the final report to close mdr 1225673-2017-00002.Investigation by data innovations and (b)(6) has determined that the time period that the misconfigured rules were in place cannot be precisely determined but were in place at least as far back as march 2013, when the available system log begins and could be longer than that.Data innovations has learned from (b)(6) (laboratory information manager) and dr.(b)(6) (acting chief pathologist) at the (b)(6) that certain actions have been taken to correct patient charts where in incorrect result for microalbumin was sent.From a phone call on (b)(6) 2017, according to dr.(b)(6), "charts will be updated with a note indicating the test was invalid.Providers will also be notified of the issue." dr.(b)(6) stated the (b)(6) would not share any additional patient impact information with data innovations.Data innovations received a copy of a report from (b)(4), included for reference here: start of included report- "subject: report of erroneous test results for microalbumin from 2014 through 2017.Date: sep.27, 2017.Problem: patient test results for microalbuminuria erroneously reported as negative when albuminuria was present.Patients: 124 patient specimens at pb-vamc.Time period: 12/11/2014 through 7/25/2017.When identified: july 24, 2017.Underlying problem: conflicting programming rules of laboratory information system (data innovations) action: rules adjusted in data innovations.47 of 124 patient results retrospectively corrected.Provider notified: in process.Description: on or about (b)(6) 2017, it was reported to me by a tech that there was an issue in which a patients' microalbumin profile results did not seem right.I began investigating, and found that, under certain circumstances, two separate "rules" in instrument manager would be triggered at the same time, causing erroneous results to be transmitted to vista.What would happen is this: if the microalbumin result from the chemistry instrument was significantly elevated and was transmitted from the analyzer to im as ">500" (which is the initial cutoff for microalbumin on the analyzer), im would not recognize >500 as a number due to the > sign, and would inadvertently trigger the rule that was supposed to be used when a result was <5 (since it counted a non-numeric result as less than 5).It would then change the microalbumin result from >500, to <5 and add the comment "unable to calculate ratio due to microalbumin being below detectable limits." we would then have released results with an elevated urine protein, and a microalbumin of <5.I contacted data innovations (instrument manager) and requested help in fixing the problem.They had me amend the rule to simply add "{and} ( {result} {on test} "2839" {is numeric}" to keep it from firing when the result contained a > sign.This was done on (b)(6) 2017.I then began communicating with my supervisor (b)(6), and pathologist dr.(b)(6), as well as data innovations representative (b)(4), about the issue to discern if any patient harm occurred.I began the process of gathering data, and by the end of august i concluded that a total of 124 results had erroneous results reported in which the urine protein result was elevated (>30) and a microalbumin was reported as <5.The first erroneous results were reported (b)(6) 2014 (i went back a total of 10 years to ensure i captured all data).On the instructions of dr.(b)(6), i went back in and canceled out the microalbumin results that had an elevated urine protein and removed the conflicting comment.As you can only go back and amend results for one year, i was only able to amend 47 of the 124 results, the remaining 77 were reported prior to (b)(4) 2016.These remaining 77 results were spread between 26 providers.(b)(4).For the other tests effected by the same misconfiguration of rules, dr.(b)(6) forwarded an email from (b)(6) stating the following: start of included email "i conducted a thorough review when di asked me to, as well as again today to re-confirm my findings.For infectious disease tests: (b)(6).Although not impossible, it is highly improbable that there would have been an issue.The linear range for all 5 of these tests is set to the maximum of 9999999.9.For a result to have been affected, the patients result would have had to have been above 10 million.Seeing as a positive result starts at either 1.0 or 1.2 depending on the test above, i don't think a patient can get 10 million.We rarely see anything even in the thousands.Trig - no < in the rule, would not have been affected folate - the rule is set to look for an exact value of "<5.4", so not having {is numeric} would not have affected it.(confirmed by (b)(6), we still do not have {is numeric} on the rule as there is no need the way its written.) b12 - no < in the rule, would not have been affected hemolysis - has no linear range, would not have transmitted a < or > in result icteric - has no linear range, would not have transmitted a < or > in result lipemia - has no linear range, would not have transmitted a < or > in result upro - rule is actually looking for a < sign, so would not have been affected.Based on the above information, i can confidently affirm that no other tests have been affected." - (b)(6), end of included email.Data innovations has determined that no malfunction of the instrument manager software occurred and therefore data innovations will close this report based on the statements of patient impact provided to data innovations by the staff at (b)(6).
|