﻿

==============================================================================

R1.1  The 'complete ascertainment' set wrongly drops event-positive rows

==============================================================================

risk set operations                         : 1898

  observed incident AKI                     : 207

  no observed incident AKI                  : 1691

    of which without day-6 creatinine       : 742

  events WITHOUT day-6 creatinine after them: 46

[REPRODUCED] 46 event-positive operations lack day-6 creatinine and were wrongly excluded

             computed 46

[REPRODUCED] 742 event-free operations lack day-6 creatinine coverage

             computed 742



old 'complete ascertainment' set            : n=1109, events=161

corrected 'ascertainable' set               : n=1155, events=207

[REPRODUCED] corrected ascertainable set is about n=1,155 with 207 events

             computed n=1155, events=207



adjusted top-stratum OR, old set       : 1.52 (0.75-3.07)

adjusted top-stratum OR, corrected set : 1.67 (0.89-3.12)

[NOT REPRODUCED] top-stratum adjusted OR in the corrected set is about 1.89, not 1.52

             computed 1.67 (old set gave 1.52)



==============================================================================

R1.2  Repeat operations share laboratory records and deaths

==============================================================================

eligible operations 2468 from 2365 patients

patients with >1 eligible operation : 98

extra operations beyond one each    : 103



adjacent-operation gaps (days), n=103:

  minimum          : 0.19 days (4.5 hours)

  pairs <= 2 days  : 18

  pairs <= 7 days  : 22

[REPRODUCED] 18 adjacent pairs within 2 days, 22 within 7 days, minimum about 4.5 hours

             computed 18 within 2 days, 22 within 7 days, minimum 4.5 hours



90-day death outcome rows           : 105

distinct patients contributing them : 100

deaths counted more than once       : 5

[REPRODUCED] 105 death rows arise from 100 patients, i.e. 5 deaths double-counted

             computed 105 rows from 100 patients



first-operation-only cohort         : n=2365, deaths=90

  top-stratum crude mortality       : 10.2% (22/215)

[REPRODUCED] first-operation-only reduces deaths from 105 to 90 and top-stratum mortality to 10.2%

             computed deaths=90, top-stratum mortality 10.2%



==============================================================================

R1.4  Timing of incident AKI, and which KDIGO limb produced it

==============================================================================

incident AKI events                 : 207

  first met criteria <=24 h after landmark : 140

  24-48 h                                  : 28

  >48 h                                    : 39

[REPRODUCED] 140 events within 24 h of the landmark, 28 at 24-48 h, 39 after 48 h

             computed 140, 28, 39



events later than 24 h after the landmark that involve the absolute limb: 0 of 67

[REPRODUCED] every event beyond 24 h after the landmark is ratio-limb only

             computed 0 absolute-limb events among 67 later events



top lactate stratum events          : 17

  <=24 h : 15

  24-48 h: 2

  >48 h  : 0

  involving the absolute limb: 15

[REPRODUCED] top stratum: 15 events within 24 h, 2 at 24-48 h, none later, 15 involving the absolute limb

             computed 15 within 24 h, 15 absolute-limb



==============================================================================

R1.7 / R3.4  Baseline creatinine P values, and what blood_gas_count counts

==============================================================================

baseline creatinine across strata, eligible cohort (n=2468) : P=0.039

baseline creatinine across strata, landmark risk set (n=1898) : P=0.625

[REPRODUCED] P=0.039 in the eligible cohort but P=0.63 in the risk set; different populations

             computed P=0.039 (eligible) and P=0.625 (risk set)



What does blood_gas_count contain? Traced from gate 1 source:

  landmark_audit._aggregate_lactate aggregates rows already filtered to

  item_name == 'lactate', with blood_gas_count = ('value', 'size').

  It therefore counts LACTATE measurements, not all arterial blood gases.

[REPRODUCED] Table 1's 'arterial blood gas measurements' row actually counts lactate rows

             confirmed by reading landmark_audit._aggregate_lactate; the label must change



==============================================================================

R3.5  Operations versus patients in the reported denominators

==============================================================================

landmark risk set : 1898 operations from 1852 patients

any AKI           : 777 operations from 755 patients

concurrent AKI    : 570 operations

incident AKI      : 207 operations from 207 patients

[REPRODUCED] the 1,898 risk units are operations from about 1,852 patients

             computed 1852 distinct patients



==============================================================================

SUMMARY

==============================================================================

                                                                                       claim  reproduced                                                                        detail

                46 event-positive operations lack day-6 creatinine and were wrongly excluded        True                                                                   computed 46

                                    742 event-free operations lack day-6 creatinine coverage        True                                                                  computed 742

                                corrected ascertainable set is about n=1,155 with 207 events        True                                                   computed n=1155, events=207

                        top-stratum adjusted OR in the corrected set is about 1.89, not 1.52       False                                             computed 1.67 (old set gave 1.52)

                  18 adjacent pairs within 2 days, 22 within 7 days, minimum about 4.5 hours        True                computed 18 within 2 days, 22 within 7 days, minimum 4.5 hours

                        105 death rows arise from 100 patients, i.e. 5 deaths double-counted        True                                           computed 105 rows from 100 patients

       first-operation-only reduces deaths from 105 to 90 and top-stratum mortality to 10.2%        True                               computed deaths=90, top-stratum mortality 10.2%

                        140 events within 24 h of the landmark, 28 at 24-48 h, 39 after 48 h        True                                                          computed 140, 28, 39

                               every event beyond 24 h after the landmark is ratio-limb only        True                         computed 0 absolute-limb events among 67 later events

top stratum: 15 events within 24 h, 2 at 24-48 h, none later, 15 involving the absolute limb        True                                     computed 15 within 24 h, 15 absolute-limb

            P=0.039 in the eligible cohort but P=0.63 in the risk set; different populations        True                            computed P=0.039 (eligible) and P=0.625 (risk set)

                Table 1's 'arterial blood gas measurements' row actually counts lactate rows        True confirmed by reading landmark_audit._aggregate_lactate; the label must change

                               the 1,898 risk units are operations from about 1,852 patients        True                                               computed 1852 distinct patients



12 of 13 reviewer claims reproduced exactly.



Outputs: <local path>\analysis\results\gate6

