Results

Descriptive statistics

The study included 609 community-dwelling older adults. Participants reported a mean OHRQoL (GOHAI) score of 37.27 (SD=7.76). Participants reported mean scores of 3.81 for both sleep sufficiency (SD=1.09) and subjective sleep quality (SD=1.08). Daytime napping was relatively common (M=3.63, SD=1.20), while dream recall was reported at a moderate frequency (M=3.00, SD=1.12). Detailed descriptive statistics are presented in Table 1.

Variable | Mean | SD | Minimum | Maximum
Oral Health-Related Quality of Life (GOHAI) | 37.27 | 7.76 | 12 | 60
Emotional Loneliness | 2.81 | 2.19 | 0 | 6
Sleep Sufficiency | 3.81 | 1.09 | 1 | 5
Sleep Quality | 3.81 | 1.08 | 1 | 5
Daytime Napping | 3.63 | 1.2 | 1 | 5
Dream Recall | 3 | 1.12 | 1 | 5

Associations between oral health, loneliness, sleep, and dream recall

Bivariate correlation analyses revealed a consistent pattern linking OHRQoL with both psychosocial and nocturnal indicators (Table 2).

Variable | 1 | 2 | 3 | 4 | 5 | 6
1. GOHAI | — |  |  |  |  | 
2. Emotional Loneliness | -0.329*** | — |  |  |  | 
3. Sleep Sufficiency | 0.146*** | -0.221*** | — |  |  | 
4. Sleep Quality | 0.135*** | -0.215*** | 0.989*** | — |  | 
5. Daytime Napping | -0.105** | -0.021 | 0.321*** | 0.313*** | — | 
6. Dream Recall | 0.104* | -0.036 | 0.178*** | 0.178*** | 0.123** | —
7. Age | -0.229*** | 0.303*** | 0.02 | 0.019 | 0.184*** | -0.027

Better OHRQoL was associated with lower emotional loneliness (r=-0.329, p<0.001), representing the most substantial association among the primary study variables. In addition, higher GOHAI scores were associated with greater sleep sufficiency (r=0.146, p<0.001), better sleep quality (r=0.135, p<0.001), less frequent daytime napping (r=-0.105, p<0.01), and more frequent dream recall (r=0.104, p<0.05).

Emotional loneliness demonstrated a significant negative relationship with both sleep sufficiency (r=-0.221, p<0.001) and sleep quality (r=-0.215, p<0.001), suggesting that older adults experiencing greater emotional disconnection also reported less favorable sleep experiences. Notably, emotional loneliness was not significantly associated with dream recall frequency.

Age was negatively associated with oral health-related quality of life (r=-0.229, p<0.001) and positively associated with emotional loneliness (r=0.303, p<0.001) and daytime napping (r=0.184, p<0.001). No significant relationship emerged between age and dream recall.

Sleep sufficiency and subjective sleep quality were nearly perfectly correlated (r=0.989). Among the 608 participants with complete responses to both items, 597 (98.2%) selected exactly the same response category, while the remaining 11 differed by only one category. The two indicators were therefore not empirically distinguishable in this sample and were interpreted as highly overlapping measures of a single subjective sleep appraisal rather than as separate sleep dimensions. Subjective sleep quality was retained as the sleep outcome in the multivariable analysis because it represented the broader global appraisal of sleep; conducting a parallel regression analysis for sleep sufficiency would have been substantively redundant.

Hierarchical regression analysis predicting sleep quality

A hierarchical multiple regression analysis was conducted to examine whether OHRQoL and emotional loneliness were independently associated with sleep quality after controlling for age and gender.

In the first step, age and gender explained a modest proportion of variance in sleep quality (R²=0.011). Gender emerged as a significant predictor, whereas age was not significantly associated with the outcome.

Following the addition of OHRQoL and emotional loneliness in the second step, the explained variance increased significantly to 6.1% (ΔR²=0.050, p<0.001). Better OHRQoL was independently associated with better subjective sleep quality (β=0.101, p=0.018), whereas greater emotional loneliness was independently associated with poorer subjective sleep quality (β=−0.182, p<0.001). Gender also remained independently associated with sleep quality, with women reporting poorer subjective sleep quality than men, while age was not significantly associated with the outcome. Emotional loneliness showed the strongest standardized association in the final model, whereas the standardized associations of gender and OHRQoL were broadly comparable in magnitude (Table 3).

Predictor | B | SE | β | t | 95% CI for B | p
Model 1 (Covariates)
Constant | 4.166 | 0.431 | – | 9.669 | 3.320, 5.012 | <0.001
Age | 0.001 | 0.005 | 0.004 | 0.097 | −0.010, 0.011 | 0.922
Gender | −0.241 | 0.091 | −0.107 | −2.645 | −0.420, −0.062 | 0.008
Model 1 statistics: N = 607; R² =0.011; Adjusted R² =0.008 F(2, 604) = 3.511; p=0.031
Model 2 (Adding GOHAI and Emotional Loneliness)
Constant | 3.082 | 0.513 | – | 6.001 | 2.073, 4.090 | <0.001
Age | 0.01 | 0.005 | 0.082 | 1.96 | −0.000, 0.021 | 0.05
Gender | −0.215 | 0.09 | −0.095 | −2.398 | −0.391, −0.039 | 0.017
GOHAI | 0.014 | 0.006 | 0.101 | 2.374 | 0.002, 0.026 | 0.018
Emotional Loneliness | −0.090 | 0.022 | −0.182 | −4.190 | −0.133, −0.048 | <0.001
Model 2 statistics: N = 607; R²=0.061 Adjusted R²=0.055; ΔR²=0.050 F(4, 602) =9.756; p<0.001

These findings suggest that OHRQoL contributes to sleep quality beyond basic demographic characteristics and that emotional loneliness represents an important psychosocial correlate of nocturnal well-being.

Hierarchical regression analysis predicting dream recall

A second hierarchical regression analysis was conducted with dream recall frequency as the dependent variable. At the first step, age and gender accounted for 1.2% of the variance in dream recall frequency (R²=0.012). Gender was significantly associated with dream recall frequency, whereas age was not.

After OHRQoL and emotional loneliness were entered at Step 2, the final model was statistically significant overall (R²=0.021, F(4, 602) = 3.150, p=0.014). However, the model explained only 2.1% of the total variance in dream recall frequency, and the variables added at Step 2 jointly accounted for an additional 0.9% (ΔR²=0.009). This incremental contribution was very small and of uncertain practical relevance. OHRQoL showed a small positive association with dream recall frequency (β=0.091, unadjusted p=0.032), whereas emotional loneliness was not significantly associated with the outcome (β=−0.004, p=0.934) (Table 4).

Predictor | B | SE | β | t | 95% CI for B | p
Model 1 (Covariates)
Constant | 3.058 | 0.444 | – | 6.888 | 2.186, 3.930 | <0.001
Age | −0.006 | 0.005 | −0.043 | −1.068 | −0.016, 0.005 | 0.286
Gender | 0.232 | 0.094 | 0.101 | 2.477 | 0.048, 0.417 | 0.014
Model 1 statistics: N=607; R²=0.012; Adjusted R²=0.009 F(2, 604)=3.704; p=0.025
Model 2 (Adding GOHAI and Emotional Loneliness)
Constant | 2.355 | 0.541 | – | 4.356 | 1.293, 3.417 | <0.001
Age | −0.003 | 0.006 | −0.021 | −0.499 | −0.014, 0.008 | 0.618
Gender | 0.223 | 0.094 | 0.096 | 2.363 | 0.038, 0.408 | 0.018
GOHAI | 0.014 | 0.006 | 0.091 | 2.146 | 0.001, 0.026 | 0.032
Emotional Loneliness | −0.002 | 0.023 | −0.004 | −0.083 | −0.047, 0.043 | 0.934
Model 2 statistics: N=607; R²=0.021; Adjusted R²=0.014; ΔR²=0.009 F(4, 602)=3.150; p=0.014

Because no adjustment for multiple comparisons was applied, the OHRQoL finding met only the conventional nominal significance threshold and should be regarded as fragile and exploratory rather than as statistically robust evidence of an independent association.

Gender remained independently associated with dream recall frequency, with women reporting more frequent dream recall than men. Its standardized association was slightly larger than that of OHRQoL, although both associations were small. Thus, OHRQoL was neither the predominant nor a practically substantial correlate in this model.

The pattern of findings differed from that observed for sleep quality. Whereas emotional loneliness was closely linked to subjective sleep experiences, dream recall appeared to be associated specifically with OHRQoL rather than with feelings of emotional loneliness. The pattern of findings differed from that observed for sleep quality. Dream recall frequency showed a weak association with OHRQoL but was not associated with emotional loneliness. Given the small effect size, the limited variance explained, the use of a single nonvalidated item, and the absence of adjustment for multiple testing, this finding should be regarded as exploratory. It does not support conclusions about the broader experience, content, or psychological function of dreaming.