Linking patient-reported oral and general health-related quality of life

Introduction

Oral health plays a pivotal role in daily life, influencing various aspects of well-being. From basic functions like eating and speaking to social interactions and self-esteem, the condition of one’s oral health significantly impacts life. Tooth pain can impair the ability to chew food properly, leading to dietary limitations and nutritional deficiencies. Similarly, oral conditions such as gum disease or missing teeth can affect speech, appearance, and confidence. The appearance of one’s face and smile can greatly influence self-perception and interpersonal relationships. Beyond these immediate effects, untreated oral disease can contribute to systemic health problems, including cardiovascular diseases and diabetes, further underscoring the importance of maintaining good oral health for overall well-being (Nazir, 2017; Dörfer et al., 2020). Oral health is interconnected to various facets of daily life, underscoring the significance of understanding this link.

Examining the intricate relationship between oral health-related quality of life (OHRQoL) and health-related quality of life (HRQoL) is vital for complete health assessments. While the influence of poor oral health on HRQoL and vice versa is acknowledged, the magnitude of this connection remains unknown. This gap is particularly significant as it represents a comprehensive summary of the broader health-oral health relationship.

Oral diseases impact patients in one or more of the four OHRQoL dimensions; Oral Function, Orofacial Pain, Orofacial Appearance, and Psychosocial Impact which represent the elemental building blocks of OHRQoL (John et al., 2014). The same applies for diseases which impact HRQoL in its two dimensions, Physical Health and Mental Health. Therefore, the magnitude of the OHRQoL-HRQoL correlation could be different depending on the oral and general diseases being experienced by the patient, and the population being studied.

Studies focused on general adult populations consistently highlight a positive and statistically significant correlation between OHRQoL and HRQoL, although there are considerable variations (Zimmer et al., 2010; Reissmann et al., 2013; Sekulić et al., 2020). For example, German dental patients and the German general population demonstrated correlations between OHRQoL and HRQoL of r = 0.24 and r = 0.28 (per Cohen’s correlation benchmarks, this indicates a moderate correlation), respectively, using the 49-item Oral Health Impact Profile (OHIP-49) and Short-Form 23 (SF-35) with a 1-month recall period and structural equation models (SEMs) (Cohen, 1988, 1992; Reissmann et al., 2013; Naik et al., 2016). Conversely, among adult dental patients of a nonprofit health care provider in Minnesota (USA), OHIP-49 and Patient-Reported Outcomes Measurement Information System (PROMIS) v.1.1 Global Health Instrument using a 1-month recall period produced a SEM-derived correlation of r = 0.56 (Sekulić et al., 2020). Further dividing the OHRQoL-HRQoL relationship into a mental health-OHRQoL component and a physical health-OHRQoL component, and using bivariable correlations with Pearson correlation coefficients, identified correlations of r = 0.45/0.45 (Reissmann et al., 2013), r = 0.47/0.52 (Sekulić et al., 2020), and r = 0.31/0.32 (Zimmer et al., 2010) among a German general population, American dental patients, and German dental patients, respectively (Zimmer et al., 2010; Reissmann et al., 2013; Sekulić et al., 2020).

Disease-specific research in this domain has also consistently emphasized the substantial correlation between OHRQoL and HRQoL. Studies involving temporomandibular disorder (TMD) patients, head and neck cancer patients, and individuals with dentofacial deformities all demonstrate positive correlations between OHRQoL and HRQoL (Öhrn et al., 2001; Balik, Peker & Ozdemir-Karatas, 2021; Duarte et al., 2022). Further substantiating this connection, another study among community-dwelling elders unveiled negative correlations between OHIP-14 scores assessing OHRQoL and all four domains of the World Health Organization Quality of Life Assessment (WHOQoL-Bref). Statistical significance emerged particularly within the physical and mental HRQoL domains (Kuo et al., 2018). Collectively, these studies emphasize the considerable magnitude of the correlation between OHRQoL and HRQoL, underscoring the pivotal intersection of oral health and overall well-being.

Existing studies use different instruments to measure the constructs which include the OHIP-5, -14 and -49 for OHRQoL measurement and the Short Form -12, and -36, as well as the PROMIS General Health instrument v.1.1 and v.1.2 to measure HRQoL. These variations are influenced by the measurement instruments used for OHRQoL and HRQoL, including recall period lengths and instrument length. For example, several OHRQoL instruments exist, and even with respect to the most widely used instrument, the OHIP, there is considerable variation in terms of the length of the recall period, and the length of the instrument, i.e., the number of items included, across its different versions (Slade & Spencer, 1994; Waller et al., 2016; John et al., 2022; Ingleshwar & John, 2023). Different analytical approaches have also been taken, as some studies report correlations of the two constructs, whereas others report SEM models. Additionally, the populations differ; from community settings to general dental, and even oral-disease specific populations. Because the concepts OHRQoL and HRQoL are tied to a time period when individuals experience the oral health impact, instrument recall periods affect the correlation. Longer recall periods are especially interesting for characterization of population health compared to shorter recall periods which are more relevant for clinical applications. Studies using a 12-month recall period are not available. Analytical approaches also shape correlation magnitudes, with bivariable correlations using instrument summary scores, and SEMs taking the latent variable nature of HRQoL and OHRQoL into account. Finally, while chance, i.e., sampling variability, adds to the complexity of correlation findings, the OHRQoL-HRQoL correlation can be truly different in varying populations.

The relationship between OHRQoL and HRQoL forms a critical foundation for comprehensive healthcare evaluations. The existing body of research, while acknowledging the shared influence of oral health and overall well-being, has demonstrated variations in the magnitude of this connection across diverse populations and measurement methodologies. This study aimed to determine the OHRQoL-HRQoL relationship over a longer period, employing a 12-month recall period, the OHIP-5, and the 10-item PROMIS 1.2—Global Health Instrument within an adult community sample to provide a deeper understanding of the OHRQoL-HRQoL relationship within an adult community sample. Using an extended recall period, we endeavor to fill current gaps in understanding, providing updated evidence on the OHRQoL-HRQoL relationship and informing modern healthcare practices. Despite the advantages of the practical, valid, and reliable 5-item version of the OHIP, to our knowledge, there is a gap using this practical instrument to assess the relationship between OHRQoL and HRQoL over a longer period. We aim to provide valuable insights and raise awareness among healthcare professionals regarding the significant impact of oral health on individuals’ overall well-being. By emphasizing the interconnectedness of oral health with general health, we strive to facilitate the development of more comprehensive healthcare strategies.

Materials and Methods

Data analysis

Structural equation models (SEMs)

A series of SEMs were generated to investigate the relationship between OHRQoL and HRQoL. We fit four models to these data by treating polytomous OHRQoL and HRQoL items as categorical, using unweighted least squares estimation. In the first two models, we separately considered the Pearson correlation between the latent variables of mental health and OHRQoL and the Pearson correlation between the latent variables of physical health and OHRQoL. The primary model (Model 1) only considered OHRQoL and HRQoL whereas the sensitivity model (Model 2) evaluated the residual correlation between OHRQoL and HRQoL after adjusting for the influence of gender, age, and teeth count (Fig. 1). Adjusting for these sociodemographic traits aligns with common practice in similar studies (Reissmann et al., 2013; Baron et al., 2015; Barrios et al., 2015; Zucoloto, Maroco & Campos, 2016; Pakpour et al., 2016; Sekulić et al., 2020).

In the third and the fourth model (Fig. 2), we specified a higher-order HRQoL factor and considered the Pearson correlation between the latent variables of OHRQoL and HRQoL. Again, we first evaluated only OHRQoL and HRQoL without considering additional variables (Model 3). Then, the influence of gender, age, and teeth count was investigated by adjusting the model for these covariates (Model 4).

Model fit for all four models was evaluated using the root mean square error of approximation (RMSEA), standard root mean squared residual (SRMR), comparative fit index (CFI) and Tucker-Lewis index (TLI). According to previous research, acceptable model fit guidelines include RMSEA ≤ 0.06, SRMR < 0.08 and TLI and CFI ≥ 0.95 (Hu & Bentler, 1999).

We used the listwise method to deal with the missing data. Data was excluded for participants who had one or more missing OHIP responses, in addition to missing any of the four responses for each of the two HRQoL domains, physical and mental health. Data loss could mostly be attributed to missing demographic information (ethnicity), rather than missing OHRQoL or HRQoL indicators. The data was initially collected using REDCap and was analyzed using Stata 17 and the lavaan package for R in later steps (Harris et al., 2009; Rosseel, 2012; StataCorp, 2021; R Core Team, 2023).

Results

Functional relationship between HRQoL and OHRQoL

The best flexible fit and the best linear fit for the relationships between mental HRQoL and OHRQoL, as well as between physical HRQoL and OHRQoL scores, were very close. Figure 3 indicates that these variables had a linear functional form, allowing the use of SEM to correlate the latent variables and the Pearson correlation coefficient to correlate summary scores.

Discussion

This cross-sectional study showed the magnitude of the association between OHRQoL and HRQoL to be r = 0.52 (95% CI [0.50–0.55]), indicating that the two constructs shared 27% of their information. According to guidelines for the interpretation of a correlation’s magnitude, this study provides evidence of a “large” correlation between OHRQoL and HRQoL in a USA-based general adult population (Cohen, 1988, 1992). Our study results are the first to indicate the relationship between the two constructs over longer periods of time.

Although the populations and instruments used varied, we made efforts to compare our results to the existing literature in a standardized manner to further understand and contextualize the magnitude of the correlation between OHRQoL and HRQoL (Table 3; Persson et al., 2009; Zimmer et al., 2010; Inoue et al., 2011; Östberg & Hall-Lord, 2011; Reissmann et al., 2013; Wickert et al., 2014; Baron et al., 2015; Barrios et al., 2015; Zucoloto, Maroco & Campos, 2016; Pakpour et al., 2016; Kuo et al., 2018; Sekulić et al., 2020; Rojas-Alcayaga et al., 2022; Purisinsith et al., 2022). The reported correlations among studies sometimes have a negative sign. While higher HRQoL scores, such as the SF-36 and the SF-12, indicate better HRQoL, higher OHRQoL scores, such as the OHIP-based score, represent worse OHRQoL. In the present study, we recoded the OHIP-score so that a higher score indicated better OHRQoL. Hence, when we compare literature findings with our results, we interpret the absolute magnitude of the correlation.

Bivariable correlations, such as Pearson and Spearman rank correlations are attenuated compared to SEM derived correlations. The attenuation is not small. We observed correlation differences of 0.11 and 0.16. Comparing bivariable correlations with SEM-derived correlations showed a similar magnitude (Reissmann et al., 2013).

The absolute magnitude of reported bivariable correlations were similar to our bivariable correlations, regardless of population or recall period pointing to a similarity if SEM would have been used (Zimmer et al., 2010; Wickert et al., 2014; Kuo et al., 2018; Rojas-Alcayaga et al., 2022). Therefore, it can be expected that the deattenuated correlation would reach a similar magnitude as reported in this study and in the only three other studies that used SEM (Reissmann et al., 2013; Zucoloto, Maroco & Campos, 2016; Sekulić et al., 2020).

Adjustment for other factors, such as depression, should lower the HRQoL-OHRQoL correlation; however, results were mixed. While the present study found a minimal decrease when depression was accounted for, the only other study that adjusted an SEM—derived correlation showed a substantial decrease that was similar for dental patients and general population subjects.

The influence of instrument length was difficult to investigate. A study using OHIP-49 instead of OHIP-5 showed similar results. However, OHIP-5 and OHIP-49 summary scores correlated with r = 0.93, indicating that, at least on the summary score level, OHIP-5 measured OHRQoL well (John, 2022; John et al., 2022). For SF-12 and SF-36, correlations are also very large (Müller-Nordhorn, Roll & Willich, 2004).

For the investigation of influence of the recall period, available data was limited. Many studies did not report the recall period at all. However, our study’s finding using the longest recall period (1 year) showed very similar findings with studies using a 1-month recall period (Reissmann et al., 2013; Sekulić et al., 2020). This may be due to the fact that the recall period for OHIP does not have a very large influence. Previous studies have shown that a 1-month recall period is similar to a 12-month period, and a 1-week recall period is similar to a 1-month period (Waller et al., 2016).

Overall, these results indicate that the OHRQoL-HRQoL relationship is “medium” to “large” according to Cohen’s guidelines for magnitude interpretation (Cohen, 1988, 1992). That the correlation is small, is unlikely. A meta-analysis based on a systematic review is necessary to summarize the evidence and to provide more insight about factors influencing the correlation, for example, the homogeneity of the correlation across cultures.

Strengths and limitations

This study used a practical method to measure OHRQoL and HRQoL with widely used instruments. PROMIS’ Global Health instrument comes from a family of measures that have greater precision than most conventional measures. Many PROMIS instruments, such the Global Health instrument, have short forms and computer-adaptive testing forms, increasing the applicability of such instruments. OHIP-5 is the shortest instrument of a family of instruments with different lengths that are widely used (Yu et al., 2023). These instruments are globally available, Ingleshwar & John (2023) from low-resource setting (Lawal & Omara, 2023) to randomized trials, Reuter-Selbach, Su & Faggion (2023), Tao et al. (2023) and used by different oral health care providers (Chanthavisouk et al., 2022) in adults (Mittal et al., 2019; Rener-Sitar et al., 2021) and children (Shayestehpour et al., 2022).

Our findings are precise. With fewer OHIP variables being present in the OHIP-5, attenuation of the correlation, and greater error, was expected in comparison to the studies that used the OHIP-49 for OHRQoL measurement. This was not necessarily the case, as our study found a similar, or slightly greater correlation between the constructs when compared to those that used the more comprehensive OHIP-49 (Reissmann et al., 2013; Sekulić et al., 2020).

Our results fit well with literature findings even if the constructs HRQoL and OHRQoL can both be captured with a number of methodological options which include a variety of available instruments, version, recall periods to select from, and analytic approaches.

It is a strength of our study that this is the first study to utilize OHIP-5 to assess the association with longer recall periods that are often used to assess public oral health impact which is in contrast to previous studies that utilized longer versions of the OHIP (Zimmer et al., 2010; Reissmann et al., 2013; Sekulić et al., 2020). Nevertheless, per Cohen’s (1988, 1992), the magnitude of the correlation remains positive between the constructs, regardless of the OHIP version used.

It is important to acknowledge a limitation related to the voluntary nature of participation in our study. It is possible that individuals with better oral and/or general health were more inclined to participate. This self-selection bias may introduce a potential source of selection bias in our results, as those who chose to take part may not be entirely representative of the broader population.

Conclusions

Utilizing the briefest version of the OHIP questionnaire, OHIP-5, and a 12-month recall period, the association between the constructs OHRQoL and HRQoL was substantial. We found that the relationship between the two constructs goes beyond a minor, or “small” correlation. Future research, such as a meta-analysis, that collectively examines the results of the existing literature could help provide a much-needed, pooled estimate of the magnitude of the association between OHRQoL and HRQoL, two crucial aspects of patient’s well-being.

Supplemental Information