Table 2

Summary of findings

Pulse oximeters versus no pulse oximeters to inform diagnosis and treatment (excluding operative surgical care)
Population: newborns, children and adolescents aged up to 19 years
Intervention: pulse oximeter readings
Control: populations with no pulse oximeter readings
Outcomes: mortality rates, morbidity, length of hospital stay
Outcomes Overall outcome difference between control and intervention groupNumber of participants by outcome (studies) Relative effect (with 95% CI) Absolute effect (with 95% CI)Quality of the evidence—Grade
Mortality ratesThe introduction of pulse oximeters alone may lead to a reduction in mortality rates.2711 29127RR: 0.648 (0.533 to 0.788)Reduction of 1.75% (1.101 to 2.398) or 17 fewer deaths per 1000 patientsVery low*
Morbidity:When pulse oximeter results are obtained in the ED, the assessed degree of illness and the diagnosis for children may be different than if pulse oximeter results are not obtained. This is especially the case for children who do not have a diagnosis of ‘well’, ‘minor orthopaedic injuries’ or ‘minor surgical injuries’, and/or is more likely in children who have low SaO2 values.25 29256425 29n/an/aVery low†
Length of hospital stayThe introduction of pulse oximetry into triage may decrease the average time children spend in triage and may increase the proportion of hypoxic children who are admitted.26 28 2962226 28 29Time spent in triage: Mean difference: 50 min (5.405 to 94.595)
Proportion of hypoxic children admitted: n/a
Time spent in triage: 17 fewer minutes spent in triage per 100 min
Proportion of hypoxic children admitted: n/a
Very low‡
Secondary research question: treatment and managementWhen pulse oximeter results are obtained in the ED, the management plans for children may be different than if pulse oximeter results are not obtained. This is especially the case for children who do not have a diagnosis of ‘well’, ‘minor orthopaedic injuries’ or ‘minor surgical injuries’, and/or is more likely in children who have low SaO2 values, particularly if these are unexpectedly low.25 28 29263325 28 29n/an/aVery Low§
  • See online supplementary appendix IV for a more detailed summary of findings table.

  • *Non-controlled before-after study: Study limitations—there is a high risk of bias as the Duke et al27 study had a serious risk of bias, due mainly to the fact that oxygen concentrators and training were introduced into the study hospitals concurrently with pulse oximeters so it is not possible to determine how much of the change in mortality rates shown in the study was due specifically to pulse oximeter use; indirectness—the study was looking at the impact of the introduction of pulse oximeters and oxygen concentrators on mortality rates, rather than just the introduction of pulse oximeters alone; imprecision—only one study (and it did not report CIs for the measure of interest); this outcome has therefore been downgraded from Low to Very Low.

  • †Non-controlled before-after studies: Study limitations—there is a high risk of bias as both of these studies had a serious risk of bias, because the physicians in both studies were aware of the intervention status of the participants and so may have been more likely to take the pulse oximeter results into account than had they received the pulse oximeter results during their initial evaluations; in addition the authors of Mower et al29 excluded 20% of children who could have been included in the study, potentially affecting the results, and the authors of Anderson et al25 excluded a subgroup of children from the analyses when it became evident that pulse oximeter results did not impact their management, so the study's results of pulse oximeter impact were exaggerated; indirectness—the changes in degree of illness and diagnosis shown in these studies are not actual changes in morbidity, they are changes in physicians’ perceptions of morbidity; also both studies were looking at different suboutcomes and different subgroups from each other, most of which were not directly relevant to, or only partially relevant to, the review; imprecision—only two studies (neither of which reported any CIs); this outcome has therefore been downgraded from Low to Very Low.

  • ‡Non-controlled before-after studies: Study limitations—there is a high risk of bias as two of the studies had a serious risk of bias, because the physicians in both studies were aware of the intervention status of the participants and so may have been more likely to take the pulse oximeter results into account than had they received the pulse oximeter results during their initial evaluations; in addition 20% and 32% of potential participants were not included in the Mower et al29 and Maneker et al28 studies, respectively, potentially affecting the results; indirectness—the outcomes investigated in the three studies (length of stay in emergency department (ED) triage, and % admitted) are indirectly related to but not exactly the same as, the outcome of length of hospital stay; imprecision—only three studies (none of which reported any CIs); this outcome has therefore been downgraded from Low to Very Low.

  • §Non-controlled before-after studies: Study limitations—there is a high risk of bias as all three of these studies had a serious risk of bias, because the physicians in all three studies were aware of the intervention status of the participants and so may have been more likely to take the pulse oximeter results into account than had they received the pulse oximeter results during their initial evaluations; in addition 20% and 32% of potential participants were not included in the Mower et al29 and Maneker et al28 studies, respectively, potentially affecting the results; also the authors of Anderson et al25 excluded a subgroup of children from the analyses when it became evident that pulse oximeter results did not impact their management, so the study's results of pulse oximeter impact were exaggerated; indirectness—the secondary research question considered the impact of pulse oximeter use on the proportion of children receiving oxygen therapy—only one of the studies actually reported the number of children in both groups who received oxygen therapy while the other two studies only reported results on outcomes that are related to oxygen therapy, by, like oxygen therapy, being examples of treatment and management; also all three studies were looking at different suboutcomes and different subgroups from each other, most of which were not directly relevant to, or only partially relevant to, the review; imprecision—only three studies (none of which reported any CIs); this outcome has therefore been downgraded from Low to Very Low.