{
  "abstract": "Background Pulse oximetry measurements of oxygen saturation (SpO 2) may be subject to even number bias and boundary effects similar to other physiological measurements.1 The extent and potential impact of this phenomenon are poorly documented. If present, such bias could have significant impact on the efferent limb of clinical escalation protocols and on future oxygen therapy guidelines. One would expect this phenomenon to be less prominent if the SpO2 was collected automatically compared with manual monitoring.Methods We have reviewed three large SpO 2 datasets from two countries: 11,553 automated SpO2 entries in Denmark (DK-Auto), 7,602,352 manual SpO2 entries in Denmark (DK-Manual) and 949,718 manual SpO2 entries in the UK (UK-Manual).Results 55.6% of DK-Auto entries, 57.5% of DK-Manual entries and 59.7% of UK-Manual entries were even numbers: (DK-Manual vs. DK-Auto, RR: 1.03,95%CI: 1.02–1.05, p<0.0001; UK-Manual vs. DK-Auto RR 1.07, 95%CI: 1.06–1.09, p<0.0001). The even-number bias was more pronounced in UK medical patients for whom 64.9% of SpO2 entries had even numbers (UK-Manual medical entries vs. DK-Auto medical entries RR 1.21, 95%CI: 1.17–1.25, p<0.0001). The proportion of even SpO2 numbers was particularly high (67.0%) for UK medical patients who were using supplemental oxygen. The overall even-number bias is illustrated in figure 1. This bias was more pronounced for manual SpO2 entries. The steps up from even to odd numbers are blunted in both manual datasets, whilst the steps down from even to odd numbers are exaggerated.Abstract P262 Figure 1Conclusions We found significant even number bias in routine pulse oximetry entries in the UK and Denmark. The bias was greater for manually collected data than for automated data. Even number bias in the automated data could arise from rounding bias in the oximeters that were used and/or the presence of SpO 2 peaks at 96% and 100%. Even number bias in the UK manual data was most marked for medical patients and for patients receiving supplemental oxygen where the SpO2 might be rounded to an even number to meet a target saturation range or to avoid mandatory clinical escalation related to Early Warning Score boundaries. This bias could influence patient management and clinical outcomes.Reference Kleinig O, To MS, Ovenden CD, et al. Vital sign measurements demonstrate terminal digit bias and boundary effects. Emergency Medicine Australasia 2024;36: 543–546.",
  "authors": [
    {
      "affiliations": [
        "Northern Care Alliance NHS Foundation Trust, Salford, UK"
      ],
      "name": "BR O’Driscoll"
    },
    {
      "affiliations": [
        "Northern Care Alliance NHS Foundation Trust, Salford, UK"
      ],
      "name": "ND Bakerly"
    },
    {
      "affiliations": [
        "Copenhagen University Hospital – Rigshospitalet, Copenhagen, Denmark"
      ],
      "name": "MH Olsen"
    },
    {
      "affiliations": [
        "Copenhagen University Hospital – Rigshospitalet, Copenhagen, Denmark"
      ],
      "name": "J Molgaard"
    },
    {
      "affiliations": [
        "Copenhagen University Hospital – Rigshospitalet, Copenhagen, Denmark"
      ],
      "name": "EK Aasvang"
    },
    {
      "affiliations": [
        "Copenhagen University Hospital – Bispebjerg and Frederiksberg, Copenhagen, Denmark"
      ],
      "name": "APH Karlsen"
    },
    {
      "affiliations": [
        "Copenhagen University Hospital – Bispebjerg and Frederiksberg, Copenhagen, Denmark"
      ],
      "name": "CS Meyhoff"
    }
  ],
  "title": "P262 Even number bias in pulse oximetry measurements in hospitals in Denmark and the UK",
  "uid": "0591c097-5f23-5a97-b1e6-aa076563ce39"
}