Article
Balancing benefits and harms in a randomized clinical trial through modeling expected health over time – a secondary analysis of SPRINT comparing blood pressure targets
Search Medline for
Authors
Published: | February 12, 2020 |
---|
Outline
Text
Background/research question: When the results of randomized clinical trials are reported, there is a focus on the primary outcome, typically a measure of the efficacy of an intervention, i.e. the benefit. Harms or side effects are typically reported as secondary outcomes. An interpretation of the benefit-harm balance is not usually given, and uncertainty remains even after large, well powered trials. The SPRINT trial compared a standard blood pressure target of 140 mmHg vs a low target of 120 mmHg in people with hypertension and without diabetes. The incidence of cardiovascular risk was reduced with the lower target, but the incidence of adverse effects increased. We aimed to model overall health based on individual patient data on events.
Methods: We developed a model of health over time. The model estimates health on a scale from 0 (death) to 100 (perfect health). Based on patient preferences, literature on quality of life changes after events and clinical judgement, we defined drops in health due to clinical events (of different magnitude and duration). We predefined the minimal important difference as half the standard deviation of health at baseline, and assessed the mean difference in health between the two trial arms over time. We received access to individual patient data underlying a publication [1] of SPRINT through the SPRINT Data Analysis Challenge.
Results: Mean health in the two blood pressure target groups was similar over the first two years, after which 120 mmHg was associated with better health. The difference was statistically significant, but not clinically relevant. The difference was larger in men aged 75 or older.
Conclusion: When trials are performed to clarify the benefit-harm balance of an intervention, the current reporting standards do not directly inform patient-centered care. Modeling the benefit-harm balance over time based on individual patient data has clear advantages. The clinical relevance can be determined, time-dependent effects detected, and if the trial is sufficiently large, subgroup differences investigated. If quality of life is not elicited directly, trials ideally would include nested preference surveys, so benefits and harms can be balanced against each other.
Competing interests: The authors declare no conflicts of interests. This abstract was prepared using the SPRINT_POP Research Materials obtained from the NHLBI Biologic Specimen and Data Repository Information Coordinating Center and does not necessarily reflect the opinions or views of the SPRINT_POP or the NHLBI.
References
- 1.
- SPRINT Research Group, et al. A randomized trial of intensive versus standard blood-pressure control. New England Journal of Medicine. 2015;373(22): 2103-2116.