This study aims to demonstrate the use of the tree-based machine learning algorithms to predict the 3- and 5-year disease-specific survival of oral and pharyngeal cancers (OPCs) and compare their ...performance with the traditional Cox regression. A total of 21,154 individuals diagnosed with OPCs between 2004 and 2009 were obtained from the Surveillance, Epidemiology, and End Results (SEER) database. Three tree-based machine learning algorithms (survival tree (ST), random forest (RF) and conditional inference forest (CF)), together with a reference technique (Cox proportional hazard models (Cox)), were used to develop the survival prediction models. To handle the missing values in predictors, we applied the substantive model compatible version of the fully conditional specification imputation approach to the Cox model, whereas we used RF to impute missing data for the ST, RF and CF models. For internal validation, we used 10-fold cross-validation with 50 iterations in the model development datasets. Following this, model performance was evaluated using the C-index, integrated Brier score (IBS) and calibration curves in the test datasets. For predicting the 3-year survival of OPCs with the complete cases, the C-index in the development sets were 0.77 (0.77, 0.77), 0.70 (0.70, 0.70), 0.83 (0.83, 0.84) and 0.83 (0.83, 0.86) for Cox, ST, RF and CF, respectively. Similar results were observed in the 5-year survival prediction models, with C-index for Cox, ST, RF and CF being 0.76 (0.76, 0.76), 0.69 (0.69, 0.70), 0.83 (0.83, 0.83) and 0.85 (0.84, 0.86), respectively, in development datasets. The prediction error curves based on IBS showed a similar pattern for these models. The predictive performance remained unchanged in the analyses with imputed data. Additionally, a free web-based calculator was developed for potential clinical use. In conclusion, compared to Cox regression, ST had a lower and RF and CF had a higher predictive accuracy in predicting the 3- and 5-year OPCs survival using SEER data. The RF and CF algorithms provide non-parametric alternatives to Cox regression to be of clinical use for estimating the survival probability of OPCs patients.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, PNG, SAZU, UL, UM, UPUK
'Culturally And Linguistically Diverse (CALD)' populations have diverse languages, ethnic backgrounds, societal structures and religions. CALD populations have not experienced the same oral health ...benefits as non-CALD groups in Australia. However, the socio-demographic profile of Australian CALD populations is changing. This study examined how household income modifies the oral health of CALD and non-CALD adults in Australia.
Data were from two National Surveys of Adult Oral Health (NSAOH) conducted in 2004-06 (NSAOH 2004-06) and 2017-18 (NSAOH 2017-18). The outcome was self-reported number of missing teeth. CALD status was identified based on English not the primary language spoken at home and country of birth not being Australia. Social disadvantage was defined by total annual household income. Effect-measure modification was used to verify differences on effect sizes per strata of CALD status and household income. The presence of modification was indicated by Relative Excess Risk due to Interactions (RERIs).
A total of 14,123 participants took part in NSAOH 2004-06. The proportion identifying as CALD was 11.7% and 56.7% were in the low-income group, and the mean number of missing teeth was 6.9. A total of 15,731 participants took part in NSAOH 2017-18. The proportion identifying as CALD was 18.5% and 38.0% were in the low-income group, and the mean number of missing teeth was 6.2. In multivariable modelling, the mean ratio (MR) for CALD participants with low household income in 2004-06 was 2% lower than the MR among non-CALD participants with high household income, with the RERI being - 0.23. Non-CALD participants from lower income households had a higher risk of having a higher number of missing teeth than low income CALD individuals (MR = 1.66, 95%CI 1.57-1.74 vs. MR = 1.43 95%CI 1.34-1.52, respectively). In 2017-18, the MR for CALD participants with low household income was 3% lower than the MR among non-CALD participants with high household income, with the RERI being - 0.11. Low income CALD participants had a lower risk of missing teeth compared to their non-CALD counterparts (MR = 1.43, 95% CI 1.34-1.52 vs. MR = 1.57, 95% CI 1.50-1.64).
The negative RERI values indicate that the effect-measure modification operates in a negative direction, that is, there is a protective element to being CALD among low income groups with respect to mean number of missing teeth.
Full text
Available for:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Abstract
Background
The ways children spend their time is one of the most valuable inputs for healthy child development. It is unknown which time investment yields the greatest return for children’s ...language and behavioural outcomes at school entry.
Methods
We used data from the first three waves (2004, 2006, 2008) of the Longitudinal Study of Australian Children (n = 4253). At every wave, parents completed 24-h time-use diaries on one randomly selected week and one weekend day. The amount of time children spent on 11 activities at ages 0–1, 2–3 and 4–5 years was analysed. Receptive vocabulary was assessed using the Peabody Picture Vocabulary Test, and externalizing behaviours were measured by the Strengths and Difficulties Questionnaire, completed by parents and teachers at 4–5 years. To identify which time investment in the first 5 years of life mattered most for children’s outcomes, a new production function was developed. This production function was estimated using a log–log linear regression model.
Results
Relative to other time investments, time spent on educational activities at 2–3 years of age was the most important time investment for receptive vocabulary and behavioural outcomes at school entry. After adjusting for confounding, every 1 h invested in educational activities at 2–3 years was associated with a 0.95% 95% CI (confidence interval): 0.62, 1.28 increase in receptive vocabulary, and a −5.72% (95% CI: −7.71, −3.73) and −9.23% (95% CI: −12.26, −6.20) reduction in parent- and teacher-reported externalizing problem behaviours. Time invested in play was also important to both receptive vocabulary and behaviour. One hour invested in play at 2–3 and 4–5 years was associated with a 0.68% (95% CI: 0.38, 0.98) and 0.71% (95% CI: 0.39, 1.03) increase in children’s receptive vocabulary at school entry. In addition, time invested in play at 2–3 and 4–5 years was associated with reduced problem behaviours at school entry. In contrast, screen time at all ages was associated with poorer parent- and teacher-reported externalizing problem behaviours.
Conclusions
These results suggest that time invested in educational activities at 2–3 years of age yield the greatest return for children’s receptive vocabulary and behaviour at school entry.
IMPORTANCE: Growing up in a language-rich home environment is important for children’s language development in the early years. The concept of “technoference” (technology-based interference) suggests ...that screen time may be interfering with opportunities for talk and interactions between parent and child; however, limited longitudinal evidence exists exploring this association. OBJECTIVE: To investigate the longitudinal association between screen time and 3 measures of parent-child talk (adult words, child vocalizations, and conversational turns) when children are 12 to 36 months of age. DESIGN, SETTING, AND PARTICIPANTS: This Australian prospective cohort study used advanced speech recognition technology to capture young children’s screen time and home language environment, on an average 16-hour day. Data were collected from 220 families once every 6 months in the family home when children were 12, 18, 24, 30, and 36 months of age, from January 1, 2018, to December 31, 2021. Statistical analysis took place from November 1, 2022, to July 31, 2023. EXPOSURE: Language Environment Analysis (LENA) technology provided automated counts of children’s language environment and exposure to electronic noise. The exposure of interest was screen time, which was calculated based on manual coding of LENA electronic noise audio segments. MAIN OUTCOMES AND MEASURES: Three measures of parent-child talk were outcomes of focus: adult words, child vocalizations, and conversational turns. Separate models were run for each of the 3 outcomes and included adjustment for child sex, child age, maternal educational level, number of children at home, number of home activities, and primary caregiver’s psychological distress. RESULTS: The study included 220 families (120 girls 54.6%; mean SD gestational age of children, 39.3 1.5 weeks; mean SD age of mother at childbirth, 31.3 4.8 years). Adjusted linear mixed-effect models demonstrated that increases in screen time were associated with decreases in measures of parent-child talk. The largest decreases were seen at 36 months, when an additional minute of screen time was associated with a reduction of 6.6 (95% CI, −11.7 to −1.5) adult words, 4.9 (95% CI, −6.1 to −3.7) child vocalizations, and 1.1 (95% CI, −1.4 to −0.8) conversational turns. CONCLUSION AND RELEVANCE: Findings of this study support the notion of technoference for Australian families, whereby young children’s exposure to screen time is interfering with opportunities to talk and interact in their home environment. This finding has implications for interventions and supports aimed at promoting a language-rich home environment, with families needing support in understanding the potential association of screen time with opportunities for children and adults to talk and interact in their home environment.
This study investigates the protective role of ethnic-racial identity (ERI) affirmation on the longitudinal association between racism and Aboriginal Australian children's social and emotional ...well-being (SEWB).
408 children from the K-Cohort of the Longitudinal Study of Indigenous Children were included in the analysis. Data were collected through questionnaire-guided interviews at 7-10 and 9-12 years of age. Children's racism experience, SEWB (Strengths and Difficulties Questionnaire), and confounding were reported by caregivers. ERI was reported by children and dichotomized into high versus low. Generalized linear models with log-Poisson links and robust errors were used to estimate adjusted Risk Ratios (RRa) for the effect of racism on SEWB domains. Effect-measure modification analysis was used to verify differences on effect sizes per strata of ERI affirmation. The presence of modification was indicated by the Relative Excess Risk due to Interaction (RERI).
Slightly above half (51.4%) of the children presented high ERI affirmation. Children exposed to racism and with low ERI affirmation were at increased risk of hyperactive behavior (RRa 2.53, 95% CI 1.17, 5.48), conduct problems (RRa 2.35, 95% CI 1.07, 5.15), and total difficulties (RRa 1.73, 95% CI 0.84, 3.55). Positive RERIs indicated the joint effects of racism and low ERI affirmation surpassed the sum of their separate effects in these domains. Children with high ERI affirmation were at increased risk of peer problems (RRa 1.66, 95% CI 0.78, 3.52).
These findings suggest that ERI may mitigate the risk of poor SEWB due to racism. Fostering affirmative ERI can be an important strategy in promoting resilience in Aboriginal Australian children.
Full text
Available for:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Background
There is a dearth of studies on the extent to which perceived racial discrimination shapes oral health. Following an intersectional perspective, we estimated the prevalence of perceived ...racial discrimination in Australia, its association with oral health impairment, and examined whether this association was more severe among low socioeconomic status (SES) groups.
Methods
Data came from the 2013 National Dental Telephone Interview Survey (N = 2798), a population‐based study of Australian adults. Multivariable Poisson regression models were estimated to test the relationship between perceived racial discrimination and self‐reported oral health impairment, as well as to investigate whether the magnitude of this association was greater among low‐SES respondents. Relative Excess Risks due to Interaction (RERI) were used to indicate the presence of potentially large discrimination effects within low‐SES strata.
Results
Racial discrimination in the past 12 months was reported by 11.5% of all participants. Australians reporting racial discrimination had 1.4 (95% CI 1.1, 1.7) times the prevalence of impaired oral health. The association between perceived racial discrimination and oral health impairment was stronger among low‐SES groups. The RERI was 0.55, indicating a super‐additive Effect Measure Modification (EMM) by income on the additive scale. Similar results were observed with the EMM analyses by educational attainment.
Conclusion
Our findings indicate that perceived racial discrimination, as a specific form of widespread inequality, is associated with higher frequencies of oral health impairment among Australian adults. We also suggest that socially marginalized groups bear a greater burden of the oral health effects of racial discrimination.
Full text
Available for:
BFBNIB, CMK, FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBMB, UL, UM, UPUK
Abstract
Knowledge of and practice around health inequities have been limited by scarce investigations on intersecting forms of structural oppression, including the extent to which their effects are ...more severe among multiply marginalized groups. We addressed these insufficiencies by adopting a structural intersectionality approach to the study of edentulism (i.e., complete tooth loss), the dental equivalent of mortality. While individual information was gathered from approximately 200,000 adult (ages 18–64 years) respondents to the 2010 US Behavioral Risk Factor Surveillance System survey, state-level data for 2000 and 2010 were obtained from a 2021 study by Homan et al. (J Health Soc Behav. 2021;62(3):350–370) and the US Census. These 3 sources provided information on edentulism, race, sex, structural racism, structural sexism, and income inequality, in addition to multiple covariates. Analyses showed that the intersections between structural sexism and state-level income inequality and structural racism were associated with 1.4 (95% confidence interval: 1.1, 1.9) and 1.5 (95% confidence interval: 1.1, 2.2) times’ increased odds of complete tooth loss, respectively. The frequency of edentulism was highest among non-Hispanic Black men residing in states with high structural racism, high structural sexism, and high economic inequality. Based on these and other findings, we highlight the importance of a structural intersectionality approach to research and policy related to health inequities in the United States and elsewhere.
An unresolved question about breastfeeding is its effect on caries, in particular, early childhood caries (ECC). In secondary analyses of data from an ECC intervention, we describe breastfeeding ...among Aboriginal children and associations between breastfeeding and ECC. Breastfeeding (duration and exclusivity to six months) was grouped into mutually exclusive categories. ECC was observed by a calibrated dental professional. Outcomes were prevalence of ECC (% decayed, missing, and filled teeth in the primary dentition (% dmft>0)) and caries severity (mean number of decayed, missing, and filled surfaces (mean dmfs)) in children aged three years. Analyses were adjusted for confounding. Multiple imputation was undertaken for missing information. Of 307 participants, 29.3% were never breastfed, 17.9% exclusively breastfed to six months, and 9.3% breastfed >24 months. Breastfeeding >24 months was associated with higher caries prevalence (adjusted prevalence ratio (PR
) 2.06 (95%CI 1.35, 3.13,
-value = 0.001) and mean dmfs (5.22 (95% CI 2.06, 8.38,
-value = 0.001), compared with children never breastfed. Exclusive breastfeeding to six months with breastfeeding <24 months was associated with 1.45 higher caries prevalence (95% CI -0.92, 2.30,
-value = 0.114) and mean dmfs 2.04 (-0.62, 4.71,
-value = 0.132), compared with never breastfeeding. The findings are similar to observational studies on breastfeeding and caries but not with randomized controlled trials of breastfeeding interventions. Despite attending to potential biases, inconsistencies with trial evidence raises concerns about the ability to identify causal effects of breastfeeding in observational research.
Background
This paper examines oral health impacts of toothache experience, self‐rated oral health (SROH), being uncomfortable with dental appearance, and avoiding foods in the Australian adult ...population.
Methods
The explanatory variables include age, sex, region, income, area‐based SES, dental insurance and visiting pattern. The data were collected in the interview in NSAOH 2017–18.
Results
There were lower percentages with: toothache in the highest (14.8%) than middle (21.2%) and lower income tertiles (25.2%); fair/poor SROH in the highest (15.8%) than middle (24.1%) and lower tertiles (34.8%); uncomfortable with appearance in the highest (29.1%) than middle (35.3%) and lower tertiles (42.2%); and food avoidance in the higher (15.3%) than middle (22.9%) and lower tertiles (34.4%). There were higher percentages with: toothache in the unfavourable (32.2%) than intermediate (23.1%) and favourable (11.7%) visiting groups; fair/poor SROH in the unfavourable (44.0%) than intermediate (27.4%) or favourable (10.2%) groups; being uncomfortable about appearance in the unfavourable (47.6%) than intermediate (39.5%) or favourable (25.8%) groups; and avoiding foods in the unfavourable (34.8%) than intermediate (26.0%) or favourable (14.5%) groups.
Conclusions
Socioeconomic status and dental visiting were associated with oral health impacts. Oral health impacts were worse for those with lower income and unfavourable visiting patterns.
Full text
Available for:
BFBNIB, CMK, FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBMB, UL, UM, UPUK
Objectives
The understanding of how subjective socioeconomic status (SSS) relates to objective socioeconomic status (OSS), and how both conditions act together in oral health outcomes is still ...unclear. This study aims to test the independent and joint association between OSS and SSS with oral health, to assess the role of socioeconomic status discrepancies, and to evaluate the role of SSS in the association between OSS and oral health.
Methods
Data from 1140 adults from a population‐based study in Southern Brazil were used. We applied diagonal reference models DRM to disentangle the effects of OSS (education) and SSS (MacArthur Scale) to oral health outcomes. The outcomes were functional dentition clinically evaluated (FD ≥20 teeth) and self‐reported oral health SROH. We also examined the discrepancy between OSS and SSS to oral health indicators and the effect measure modification (EMM) of SSS on the association between OSS and oral health.
Results
Subjective socioeconomic status and OSS contributed equally to SROH, while OSS explained a substantially higher amount of FD than SSS (0.85 vs 0.15). An EMM of SSS was found on the association between OSS and fair/poor SROH, with a relative excess risk due to interaction (RERI) of 1.08. Less evidence of EMM was found for FD (RERI = 0.14). Individuals with lower SSS and OSS had four times the risk of the outcomes than the reference group.
Conclusions
Adults with concordant lower SSS and OSS have a worse oral health than those with concordant higher status. There was evidence that the association between OSS and SROH is modified by SSS.
Full text
Available for:
BFBNIB, CMK, FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBMB, UL, UM, UPUK