We conducted an observational cross-sectional study performing a secondary database analysis using national data from the “Socio-Emotional Assessment Form” applied to older adults from September to November 2020 by the Management of the Elderly Person and Social Benefits (GPAMYPS) of the Peruvian Social Security (EsSalud). In Peru, EsSalud serves as a key healthcare provider and insurance entity for employees and their dependents who are part of the formal labor market, covering approximately 35% of the country’s total population [9]. Meanwhile, the Ministry of Health (MoH) of Peru oversees public healthcare services for the remaining population, offering coverage through the Integral Health Insurance program, which is specifically designed to assist uninsured citizens, particularly those who are economically disadvantaged or otherwise vulnerable.

In our study, we included information on older adults (≥ 60 years) assessed through the Socio-Emotional Assessment Form that have reported the death of a family member or a close person during the last six months since the realization of this assessment.

We included information from the Socio-Emotional Assessment Form for older adults, including mental health problems. The evaluation of depression was performed using the PHQ-9 scale validated in Peru [10, 11]. We categorized this variable according to the presence of depression (5 points or more on the PHQ-9 scale) and according to its severity level as “mild” (5 to 9 points), “moderate” (10 to 14 points), “severe” (15 to 19 points), or “very severe” (20 to 27 points). This tool has shown good reliability among Peruvian population (Cronbach’s alpha = 0.89; McDonald’s omega = 0.86), and had a sensitivity of 76.0% and a specificity of 72.1% for major depression when the score is ≥ 7 points [12].

The evaluation of anxiety was assessed through the GAD-7 scale validated in Spanish [13]. We categorized this variable according to the presence of anxiety (5 points or more on the GAD-7 scale) and according to its severity level as “mild” (5 to 9 points), “moderate” (10 to 14 points), or “severe” (15 to 21 points). This tool also has shown good reliability among Peruvian population (Cronbach’s alpha = 0.85; McDonald’s omega = 0.81), and had a sensitivity of 53.6% and a specificity of 78.8% for anxiety when the score is ≥ 8 points [12].

The evaluation of suicidal ideation was performed through questions related to the presence of suicidal thoughts or attitudes in the last two weeks based on the recommendation of Park et al. [14]. Those who affirmed presenting any of these conditions were categorized as “With suicidal ideation”, and the rest as “Not suicidal ideation”. This tool has shown good validity and reliability scores (content validity index = 0.89, content validity ratio = 0.85, Cronbach’s alpha = 0.76) among older adults [15].

The complicated grief was evaluated using the Brief Complicated Grief Scale, which includes attitudes or states of grief following the loss of a close person or family member in the last two weeks. This variable was categorized as “with CG” and “without CG” Each question is scored from 0 to 2 points, where 0 = not at all, 1 = somewhat and 2 = very much. A score of 4 or more indicates risk of complicated grief. This tool has shown good a discriminant validity and reliability scores (Average Variance Extracted = 0.39, Cronbach’s alpha = 0.75) [16].

As intervening variables, we considered the sociodemographic characteristics of the population studied, including sex categorized as “female” and “male,“ age evaluated in years numerically and through categories: “60–69,“ “70–79,“ “80–89,“ and “90–99,“ and the region where the older adult was evaluated categorized as “Lima” and “other regions.“ Additionally, we included information on older adult’s cohabitation, categorized as “lives alone,“ “lives with one person,“ and “lives with two or more people.“

The Socio-Emotional Assessment for older adults was an initiative deployed by the GPAMYPS to assess prevalent mental health problems among older adults from the Peruvian Social Security (EsSalud) during the COVID-19 pandemic. The primary objective was to gather screening data on the mental health of older adults, with a specific focus on those who had exhibited negative mental health symptoms (e.g., sadness, anxiety, stress, fear) during remote monitoring sessions amidst the pandemic in line with the GPAMYPS’s Institutional Operational Plan.

Given logistical constraints, out of the 11,430 older adults with negative mental health symptoms monitored between September and November 2020, EsSalud evaluated a sample of 1,457 individuals (Fig. 1). Of these, 840 agreed to complete the “Socioemotional Evaluation Form”. For our study, we initially considered 255 participants who had experienced the death of a close family member or friend within the two weeks leading up to this evaluation. We then excluded six individuals below 60, resulting in a final sample of 249 older adults for analysis.

Flowchart of older adults selected to the study

The Socio-Emotional Assessment’s data was collected via telephone by trained EsSalud personnel, who collected the information digitally using previously validated questionnaires. A de-identified version of this database was subsequently provided to our research team.

We performed a descriptive analysis of the older adults evaluated. We used the median and interquartile range to describe the numerical variable age, given its skewed distribution, and we used the frequencies and proportions to describe the remaining categorical variables. Additionally, we performed a bivariate analysis where we explored the relationship between sociodemographic characteristics, the presence of CG and anxiety, depression, and suicidal ideation in older adults evaluated through Fisher’s exact test or Chi-square test according to each case.

To evaluate the association between CG and the presence of anxiety, depression, and suicidal ideation, multivariate analyses were performed using Poisson regressions with robust variance, obtaining crude and adjusted prevalence ratios (PR) with 95% confidence intervals (95% CI). Confounding variables for each model included the sociodemographic variables (gender, age, place of origin, and cohabitation) and the coexistence of mental health problems according to each case based on the theoretical framework. All statistical analyses were performed using Stata v17.0 software [17].