The Social Vulnerability Index: Assessing Replicability in Predicting Mortality Over 27 Years

To the Editor: With advancing age, the impact of social and living conditions is still greater.1, 2 A decade ago Andrew et al operationalized social vulnerability according to a deficit accumulation approach, and they showed that greater social vulnerability is strongly associated with increased mortality in older adults.3 Beyond the Canadian context where the Social Vulnerability Index (SVI) has been computed, its predictive value has been replicated in only two studies.4, 5 The present study takes advantage of the PAQUID survey involving 27 years of follow-up to assess the replicability of the SVI by confirming its association with mortality in the context of a French population-based study. The PAQUID study is an epidemiological study based on a sample of 3777 community-dwelling individuals aged 65 years and older, randomly selected from the electoral rolls in 75 different sites in southwestern France. The methodology was described elsewhere.6 The study received the approval of the Ethics Committee of the Bordeaux University Hospital, and all participants gave their written informed consent to participate. The SVI and the frailty index (FI) were computed at baseline according to previously published methods.3, 7 Overall, 28 self-reported items covering social domains were included in the SVI. A total of 36 self-reported health-related items were considered for the FI. Both scores range from 0 to 1, with a higher score indicating a higher social vulnerability for the SVI and a higher level of frailty for the FI. In accordance with previous work, the two did not overlap (ie, SVI and FI items were mutually exclusive).3 All-cause mortality was obtained for all participants not seen at follow-up with a systematic request to civil registration, family, and/or the general practitioner until 27 years after the baseline visit. The study sample consisted of 3695 participants. First, we used linear regression modeling and the t test to verify the association between SVI and both age and sex. Two Cox regression analyses assessing the association between SVI and mortality with progressive adjustment were performed. Exact dates of death were available for participants who died during the follow-up. For those remaining alive, age was censored at the last visit date. All statistical analyses were performed using R software, v.3.3.2. Throughout the 27 years of follow-up, 93.6% of the 3695 participants died. At baseline, the mean age was 75.0 years (standard deviation [SD] = 6.8), and men accounted for 41.9% of the sample. The whole sample mean SVI was .38 (SD = .12). The mean FI was .22 (SD = .13). As found in the original study by Andrew et al,3 SVI increased with age (p < .001), and men had a lower SVI than women (.34 vs .40; p < .001). In the same vein, FI increased with age (p < .001), and men had a lower FI than women (.25 vs .27; p < .001). Finally, some participants reported no level of frailty (n = 25), whereas no individual was entirely exempt from social vulnerability (minimal SVI score = .06). Adjusted for age and sex, higher SVI was associated with increased risk of mortality (hazard ratio [HR] = 8.03; 95% confidence interval [CI] = 5.95-10.83; p < .001). After additional control for frailty, higher SVI remained associated with increased risk of mortality (HR = 2.50; 95% CI = 1.75-3.58; p < .001) (Table 1). Based on a large sample selected from the general population followed up for 27 years, our study confirms the strong predictive value of the SVI on mortality in older adults because a higher SVI score was associated with greater mortality after controlling for sociodemographics and frailty. Our findings are consistent with the original results published3, 8, 9 and provide a new replication of the SVI supporting the robustness of this social deficits accumulation approach. This work will potentially contribute to expanding such a measurement instrument in gerontology and geriatrics research because it encompasses many interests. First, integrating an operational measure of social vulnerability in the studies addressing health outcomes may substantially lead to a better understanding of the enormous heterogeneity of aging because social factors are a major cause of it.10 Second, the SVI could be easily implemented in clinical settings as a reliable screening tool for older persons at risk of negative evolution because each segment of information is highly accessible and easy to collect. Most importantly, the number of deficits rather than the nature of deficits is the key point. Indeed, the deficits explored can differ according to individuals and contexts as long as the principle of considering multiple sources or factors of social vulnerability is respected. Finally, the SVI could help in establishing prevention strategies in the socially vulnerable older adult population. Indeed, among the numerous social deficits, some of them are modifiable and could be targeted in future prevention programs aiming, for instance, at promoting social interaction or social activities/engagement that could, in turn, reduce mortality. Conflict of Interest: The authors have no conflicts. Author Contributions: Camille Ouvrard conducted the statistical analysis, interpreted the data, and wrote the manuscript. Hélène Amieva and Maturin Tabue-Teguo designed and supervised the study. Jean-François Dartigues is the principal investigator of the PAQUID study. José Alberto Avila-Funes, Jean-François Dartigues, Hélène Amieva, and Maturin Tabue-Teguo provided critical revisions of the manuscript. Sponsor's Role: This work was supported by AGRICA, ARMA, Caisse Nationale d'Assurance Maladie des Travailleurs Salariés, Caisse Nationale de Solidarité pour l'Autonomie, Conseil Général de la Dordogne, Conseil Général de la Gironde, Conseil Régional d'Aquitaine, Fondation de France, France Alzheimer, GIS Longévité, Institut National de la Santé et de la Recherche Médicale, Ipsen, Mutuelle Générale de l'Education Nationale, Mutualité Sociale Agricole, Novartis Pharma, Roche, and SCORInsurance. The funders had no role in the study design, methods, subject recruitment, data collections, analysis, and preparation of the article.