This study was an analysis of data collected from participants in the New York University Alzheimer’s Disease Research Center (NYU ADRC) longitudinal study, which is approved by the Institutional Review Board of the NYU School of Medicine. All subjects provided informed consent.

Analysis included de-identified data from middle to older age (46–87 years old) participants seen between 2006 and 2019 at the NYU ADRC. Participants were recruited from the community via events, advertisements, and physician referral. ADRC inclusion criteria were having a study partner, fluency of participant and study partner in English or Spanish, and willingness to undergo phlebotomy and neuroimaging. ADRC exclusion criteria were major neurological disease (other than Alzheimer’s disease and related disorders), severe psychiatric illness, HIV/AIDS, organ failure/transplant, severe autoimmune disease, alcohol/recreational drug abuse, and significant malignancy within 5 years. Participants underwent annual evaluation using the National Alzheimer’s Coordinating Center Uniform Data Set (NACC UDS) and additional measures. Consensus diagnoses were determined by a team of physicians and neuropsychologists.

From this resulting dataset, participants were further selected for this analysis if they self-identified as Hispanic or NHW, had their first visit between 2006 and 2019 with a consensus diagnosis of normal cognition with SCD based on the Global Deterioration Scale,³⁶ and had a 2-year follow-up. While NACC UDS underwent modifications during this period, we chose to analyze a subset of surveys and tests that were consistently done at each visit in order to minimize missing values.

Participants completed a comprehensive neuropsychological assessment and a clinical evaluation. SCD characteristics were measured by the Brief Cognitive Rating Scale (BCRS).³⁷ The presence of depressive symptoms was assessed by the total score of the short version of the Geriatric Depression Scale (GDS).³⁸ Objective cognitive performance was evaluated by a neuropsychological battery of tests to assess global cognition, attention, processing speed, executive function, memory, and language. This battery included the Montreal Cognitive Assessment (MoCA),³⁹ Mini-Mental State Examination (MMSE),⁴⁰ and the NACC UDS⁴¹ subtests: Digit Symbol Substitution Test (DSST), Trail Making Test A and B (Trails A and B), semantic verbal fluency (Animals and Vegetables), and the Guild Memory Test’s Paragraph Recall - Immediate and Delayed (PARI, PARD) and geometric design recall (Designs).⁴² (See Supplemental Table 1)

Cross-sectional and longitudinal analyses were performed on participants enrolled. Descriptive statistics were used to summarize the demographic characteristics of the sample, including mean and standard deviation (SD) for continuous variables and counts and percentages for categorical variables. Those characteristics were compared between Hispanics and NHW using chi-square tests or Fisher exact tests for categorical variables and two-sample t-tests for continuous variables.

We conducted preliminary power analyses for detecting potential mean difference between the objective cognitive performance among the study subjects with SCD of Hispanics vs NHW. Our study has comparable sample size with some of the recently published studies such as the cross-sectional study reported by Nakhla M et al (2021),¹² where within the Latino group, self-reported SCD was related to lowered objective cognition measured using MDRS scores in linear regression (r = −.672, P = .001). We have data at both baseline and at 2-year follow-up at NYU ADRC. Preliminary power analyses using a two-sample t test indicate that we have power >85% at 5% significance level to distinguish the baseline as well as follow-up objective cognitive performance scores between 23 Hispanics and 165 NHW in MMSE, DDST, Trails A, Trails B and Guild PARD. Preliminary results (see Table 5 of this manuscript) indicated that Hispanics ethnicity was indeed significant (P ≤ .001) in linear regression models of these objective cognitive performance scores at follow-up. These differential objective cognitive performance scores between Hispanics and NHW SCD subjects motivated our further investigation for the SCD rate of progression to MCI at 2-year, as well as baseline features and longitudinal outcomes of SCD complaints, objective cognitive performance and depressive symptoms among Hispanics compared to NHW.

In that sense, to study whether Hispanic ethnicity was associated with baseline or follow-up scores, slope of score change (changes in the scores using follow-up year’s score minus baseline score divided by duration of its follow-up) or decline status longitudinally (performance decline = 1, no change or better = 0), we applied multiple linear regression or logistic regression models depending on the types of the response variables. To establish a more accurate cross-sectional and longitudinal analysis of our data, regression models were adjusted for age, sex, education level, GDS scores, and/or baseline values of each individual test being examined. All statistical analyses were performed using R, version 4.2.1 (R Core Team, Vienna, Austria), and P values <.05 were considered significant.

A total of 188 participants met the inclusion criteria, including Hispanic (n = 23) and NHW (n = 165) groups. Basic participant demographic data are shown in Table 1. Hispanic participants were younger than NHW (63.32 ± 7.27 vs 67.61 ± 7.24; P = .009) and had on average 1.9 fewer years of formal education (15.17 ± 3.05 vs 17.10 ± 2.16, P < .001). The overall cohort was comprised mostly by women. Average follow-up interval was 2.27 ± .31 years, with no statistically significant difference between the two groups (P = .473).

As shown in Table 2, Hispanic ethnicity associated with a higher risk for progression from SCD to MCI during the 2-year follow-up, (Model 1: OR: 6.10, 95% CI 1.09–34.20, P = .040). This effect was attenuated when baseline GDS score (Model 2: OR 4.62, 95% CI .80–26.56, P = .086) or change in MMSE (Model 3: OR 5.98, 95% CI 1.00–35.565, P = .050) was added to the model. This suggests that depressive symptoms and cognitive decline may influence the increased SCD→MCI progression risk in Hispanic participants.

We analyzed possible differences in the presence of depressive symptoms, the characteristics of symptomatology and change over time in Hispanic vs NHW participants, considering previous literature describing a stronger association between depressive symptoms and SCD in the former. Analyses demonstrated that Hispanic ethnicity associated with higher scores on the GDS at baseline (β = .82, SE: .41, P = .048), and at follow-up (β = 1.15, SE: .45, P = .010), as shown in Table 3. Moreover, results showed that Hispanic participants at baseline were more likely to report dropping activities (OR: 3.86, 95% CI 1.01–14.71 P = .048) and boredom (OR: 10.34, 95% CI 2.06–51.80, P = .005). At follow-up, Hispanic participants also were more likely to report memory loss (OR: 5.74, 95% CI 1.28–25.74, P = .023) in addition to dropping of activities (OR: 14.49, 95% CI 3.72–56.52, P < .001) and boredom (OR: 5.30, 95% CI 1.19–23.49, P = .028). Longitudinally, Hispanics had a higher likelihood for the GDS score to increase during the interval period (OR: 3.16, 95% CI 1.18–8.51, P = .023). Across items, Hispanic participants had a higher risk to newly report dropping activities upon follow-up (OR: 13.28, 95% CI 1.80–97.80, P = .011).

To examine the impact of Hispanic ethnicity on SCD, we analyzed the first 5 items of traditional BCRS as well as additional 3 items of the expanded BCRS that rated subjective concerns, as shown in Table 4. Importantly, models were adjusted for depressive symptoms. At baseline, there was no difference in the BCRS 1–5 total score. However, analyzing specific items showed that Hispanic participants reported statistically higher cognitive concerns in concentration (β = .70, SE: .15, P < .001) but less concerns than NHW in recent memory (β = −.16, SE: .07, P = .026). At follow-up, only the increased concern about concentration remained (β = .47, SE: .15, P = .002). Also, Hispanic ethnicity associated with an increase in the total BCRS score (β = .40, SE: .17, P = .023), and in the specific BCRS domains of concentration (β = .13, SE: .07, P = .047), past memory (β = .13, SE: .06, P = .039) and functional abilities (β = .10, SE: .05, P = .037).

As shown in Table 5, the Hispanic cohort reported lower baseline performance on measures of global cognition and domain-specific tests: MMSE (β = −1.17, SE: .32, P = .003), DSST (β = −13.82, SE: 2.78, P < .001), Trails A (β = 11.40, SE: 2.31, P < .001) and Trails B (β = 39.95, SE: 7.64, P < .001), Verbal fluency on Animals (β = −2.98, SE: 1.31, P = .024) and Vegetables (β = −2.61, SE: .93, P = .005), Guild Delayed Paragraph Recall (β = −2.09, SE: .76, P = .006), and Guild Designs (β = −1.18, SE: .54, P = .029). On follow-up, lower performance was also associated with Hispanic ethnicity in these same tests, except Verbal fluency on Animals and Guild Designs, and the inclusion of Guild Immediate Paragraph Recall (β = −1.54, SE: .59, P = .010).

We then analyzed if Hispanic ethnicity was associated with worsening performance in those tests longitudinally, beyond lower performance from baseline. As shown in Table 6, the adjusted score change (or slope of score change) for Hispanic ethnicity was significantly associated with lower performance over time in the MMSE (β = −.27, SE: .13, P = .039), MoCA (β = −.80, SE: .34, P = .032), Trails A (β = 2.75, SE: .89, P = .002), Trails B (β = 9.18, SE: 2.71, P = .001) and Guild Paragraph Recall Delayed (β = −.80, SE: .28, P = .005).