Dr Daniel Vincent O'Hara1,2, Mrs Danielle Horne3, Mrs Alison Smyth3, Mr Xiaobing Ma3, Dr Parthasarathy Shanmugasundaram3,4, Mrs Suh Wong2, Dr Kristy Robledo1, Professor Mark Brown1,5
1NHMRC Clinical Trials Centre, Camperdown, Australia, 2Department of Renal Medicine, Royal North Shore Hospital, St Leonards, Australia, 3St George Hospital Department of Nephrology, Kogarah, Australia, 4The Sutherland Hospital, Caringbah, Australia, 5University of New South Wales, Kensington, Australia
Biography:
Dr Daniel O’Hara is a nephrologist and postdoctoral fellow with the University of Sydney and the NHMRC Clinical Trials Centre. Daniel's research interests include initiatives to provide people living with chronic kidney disease and/or diabetes with a more consistent and high-quality standard of care around the world. The pursuit of this goal includes working closely with patients as key stakeholders, developing programs to provide patient education and long-term support, investigating new medications and technologies that could improve long term health and wellbeing outcomes, and advocating for improved healthcare equity.
Aim:
To assess the accuracy of the double surprise questions for predicting mortality among people receiving dialysis.
Background:
Prediction of death for people receiving dialysis can facilitate pro-active support, but the surprise question may over-estimate risk.
Methods:
We conducted a prospective cohort study using clinician surveys and routinely collected patient data. Nephrologists and senior nurses across 4 dialysis units were asked the surprise question of “Would I be surprised if this patient were to die within the next 12 months?” for people receiving hemodialysis and peritoneal dialysis under their care. When death wouldn’t be surprising, they were also asked “Would I be surprised if this patient is still alive after 12 months?” to identify a highest risk group for whom survival would be surprising. We assessed the sensitivity, specificity, positive predictive value (PPV), negative predictive value, and log-rank survival outcomes of the answer risk groups. Sensitivity and specificity were compared with McNemar's test.
Results:
407 patients were included, with 17 nephrologists and 31 nurses providing answers. There were 46 deaths. The high risk group identified through doctors’ answers to the double surprise questions had much lower sensitivity for predicting death than the single surprise question (35.7%, 95% confidence interval [CI] 21.6-52.0% compared to 78.6%, 95%CI 63.2-89.7%, p<0.001 for difference) but a greater PPV (50.0%, 95%CI 31.3%-68.7%, compared to 26.0%, 95%CI 18.6-34.5%). Responses by doctors and nurses were broadly comparable.
Conclusion:
The double surprise question can identify individuals receiving dialysis who are at particularly high risk of death within 12 months, but only identifies a minority of those who will die in this time frame. The single surprise question remains a sensitive tool for predicting death.