“…The choice of the current matching variables was based on their empirical link with mortality (e.g. Grundy & Kravdal, 2007;Lundin, Lundberg, Hallsten, Ottosson, & Hemmingsson, 2010;Park, Oh, Roh, & Moon, 2017) but was also contingent upon their availability throughout the entire study period. Relying on a broader set of mortality-related factors, including but not limited to medical conditions, doctor visits, and social participation, would likely have optimized the prediction of [partner] death within the study.…”