Abstract
Introduction/objective
A systematic review was conducted to explore whether tooth loss affects dietary intake and nutritional status among adults.
Data
Longitudinal studies of population-based or clinical samples of adults exploring the effect of tooth loss on food/dietary/nutrient intake and/or nutritional status were included for consideration. The risk of bias was assessed using the Newcastle-Ottawa Scale for cohort studies.
Sources
A search strategy was designed to find published studies on MEDLINE, EMBASE and LILACS up to March 2017.
Study selection
Eight longitudinal studies in 4 countries (United States, Japan, Australia and Brazil) were included. Five of the six studies investigating the association between tooth loss and dietary intake showed significant results. The only consistent association, as reported in 2 studies, was for greater (self-reported) tooth loss and smaller reductions in dietary cholesterol. Three of the 4 studies investigating the association between tooth loss and nutritional status showed significant results. However, most results were contradicting. The quality of the evidence was weak.
Conclusion
There is at present no strong evidence on the effect of tooth loss on diet and nutrition, with inconsistent results among the few studies identified. Additional high-quality longitudinal studies should address the limitations of previous studies identified in this review.
1
Introduction
Diet is an important component of leading a healthy life as it has a role in the aetiology, and thus prevention, of many chronic conditions such as obesity, cardiovascular disease, diabetes and cancer among other chronic conditions . Tooth loss and nutritional intake are intricately connected . The oral cavity is not only the entryway for nutrient intake but the primary function of teeth is mastication . Tooth loss reduces masticatory function and chewing ability, which in turn can limit food choices and variety in the diet . For these reasons, dietary intake has been regarded as an intermediate in the pathway between tooth retention and a number of diet-related chronic diseases .
Given these claims, it is not surprising to find a few reviews on the interrelationship between tooth loss, diet and nutritional status . However, they are not without limitations. Earlier reviews did not follow a systematic procedure for the identification and synthesis of studies . Later reviews have been more systematic in their approach to review the available literature but have had a limited scope looking at older adults , free-living older adults or papers published very recently ; missed some important longitudinal studies ; included evidence from cross-sectional studies ; or did not assess the quality of the included studies . The latter point is important since confounding by participants’ socioeconomic status and health status needs to be addressed in observational studies . Without addressing these limitations, robust conclusions on the association between tooth loss and nutrition cannot be reached. The aim of this study was to systematically review longitudinal evidence on whether tooth loss affects dietary intake and nutritional status among adults. Although a poor diet, especially one low in calcium and fibre , may be a risk factor for tooth loss, we are interested in how tooth loss may influence dietary intake, and subsequently, nutritional status, given the increasing interest in tooth loss as a risk factor for various chronic diseases and mortality.
2
Methods
This systematic review followed the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) recommendations . The review protocol was registered in PROSPERO (Registration number CRD42017065361).
2.1
Criteria for considering studies for this review
Broad criteria were predefined to select articles for inclusion, following the PICO format. Only longitudinal/panel studies were included as they provide the strongest observational evidence. Case-control, cross-sectional, case report/series and expert opinions were excluded. Participants were adults aged 18 years or above, irrespective of recruitment setting (community-dwelling, nursing/care homes, hospitals) and health status (generally healthy or with one or more morbidities). The exposure was tooth loss measured at least once during the duration of the study (baseline assessment) through self-reports or clinical examination. The outcome measures were dietary/food/nutrient intake (measured as total energy intake or specific nutrient intake from questionnaires, recalls, blood samples, etc.) and nutritional status (measured as weight loss, body mass index, anthropometric measurements, etc.).
2.2
Study selection and data extraction
Three electronic databases (MEDLINE via PubMed, EMBASE via Ovid and LILACS via BIREME) were searched for published literature up to March 2017 using a combination of Medical Subject Headings (MeSH) terms and text words around three main topics: the exposure (tooth loss) and the outcomes (nutrient intake or nutritional status). These were combined with methodological filters for longitudinal studies specific for each database. Search terms were chosen based on the team expertise and previous related reviews. No language restrictions were applied. Search strategies are shown in Supplemental file 1.
All references retrieved were managed in bibliographic software EndNote X7 (Clarivate Analytics, New York, United States). Duplicated articles were excluded at this stage. Two reviewers (PG and EB) independently and in duplicate screened the titles and abstracts of all identified publications against the eligibility criteria for inclusion. The full-text of publications were sought if at least one of the reviewers considered the study as potentially meeting the inclusion criteria. The final decision about whether a study met the inclusion criteria was made based on the full-text and after discussion between reviewers. The grey literature was searched by looking for relevant material in OpenGrey repository, Google Scholar and searching the internet using the pre-set text words as well as searching all relevant reference lists of identified articles and related reviews.
A master file was created in excel listing all studies retrieved and including their title, authors, journal, publication year and reason for exclusion (Supplemental file 2). For eligible studies, the two reviewers additionally extracted information on study design, participants’ characteristics (sample size, age range and country), length of follow-up, attrition rate, exposure variables, outcome measurements, covariates/confounders, data analysis and main findings. Disagreements were resolved through discussion.
2.3
Risk of bias assessment
Included studies were assessed for risk of bias using the Newcastle-Ottawa Quality Assessment Scale (NOS) . The NOS evaluates three domains: selection (4 items), comparability (1 item) and outcome (3 items). A study could be given one star for each item under selection and outcome and two stars under comparability. For selection, a star was given when the exposed cohort was truly or somewhat representative of exposed adults in the community, when the non-exposed cohort was drawn from the same community as the exposed cohort, when the exposure (tooth loss) was ascertained through clinical examinations, and when the outcome of interest was measured both at baseline and follow-up. For comparability, a star was given when the study controlled for socio-demographic characteristics (sex, age and any socioeconomic position indicator) during the design or analysis, and it was given two stars when it additionally controlled for participants’ health status (chronic conditions, comorbidities, activities of daily living and the like). For outcome, one star was given when the assessment of outcome was independent/blinded or through record linkage, when the follow-up period was long enough for changes in outcomes to occur, and when all participants were accounted for during follow-up or those lost to follow-up were unlikely to introduce bias (<20% attrition rate and description provided of those lost). A good quality scored required 3–4 stars in selection domain AND 1–2 stars in comparability domain AND 2–3 stars in outcome domain; a fair quality study required 2 stars in selection domain AND 1–2 stars in comparability domain AND 2–3 stars in outcome domain; and a poor quality study 0–1 stars in selection domain OR 0 stars in comparability domain OR 0–1 stars in outcome domain .
2.4
Data synthesis
A meta-analysis of the findings (i.e. forest and funnel plots) was not feasible given the high level of heterogeneity found across studies. Instead, we opted for a narrative synthesis of the results . To that end, we created tables summarising the key methodological characteristics of all included studies and the methodological quality assessment of the studies based on NOS.
2
Methods
This systematic review followed the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) recommendations . The review protocol was registered in PROSPERO (Registration number CRD42017065361).
2.1
Criteria for considering studies for this review
Broad criteria were predefined to select articles for inclusion, following the PICO format. Only longitudinal/panel studies were included as they provide the strongest observational evidence. Case-control, cross-sectional, case report/series and expert opinions were excluded. Participants were adults aged 18 years or above, irrespective of recruitment setting (community-dwelling, nursing/care homes, hospitals) and health status (generally healthy or with one or more morbidities). The exposure was tooth loss measured at least once during the duration of the study (baseline assessment) through self-reports or clinical examination. The outcome measures were dietary/food/nutrient intake (measured as total energy intake or specific nutrient intake from questionnaires, recalls, blood samples, etc.) and nutritional status (measured as weight loss, body mass index, anthropometric measurements, etc.).
2.2
Study selection and data extraction
Three electronic databases (MEDLINE via PubMed, EMBASE via Ovid and LILACS via BIREME) were searched for published literature up to March 2017 using a combination of Medical Subject Headings (MeSH) terms and text words around three main topics: the exposure (tooth loss) and the outcomes (nutrient intake or nutritional status). These were combined with methodological filters for longitudinal studies specific for each database. Search terms were chosen based on the team expertise and previous related reviews. No language restrictions were applied. Search strategies are shown in Supplemental file 1.
All references retrieved were managed in bibliographic software EndNote X7 (Clarivate Analytics, New York, United States). Duplicated articles were excluded at this stage. Two reviewers (PG and EB) independently and in duplicate screened the titles and abstracts of all identified publications against the eligibility criteria for inclusion. The full-text of publications were sought if at least one of the reviewers considered the study as potentially meeting the inclusion criteria. The final decision about whether a study met the inclusion criteria was made based on the full-text and after discussion between reviewers. The grey literature was searched by looking for relevant material in OpenGrey repository, Google Scholar and searching the internet using the pre-set text words as well as searching all relevant reference lists of identified articles and related reviews.
A master file was created in excel listing all studies retrieved and including their title, authors, journal, publication year and reason for exclusion (Supplemental file 2). For eligible studies, the two reviewers additionally extracted information on study design, participants’ characteristics (sample size, age range and country), length of follow-up, attrition rate, exposure variables, outcome measurements, covariates/confounders, data analysis and main findings. Disagreements were resolved through discussion.
2.3
Risk of bias assessment
Included studies were assessed for risk of bias using the Newcastle-Ottawa Quality Assessment Scale (NOS) . The NOS evaluates three domains: selection (4 items), comparability (1 item) and outcome (3 items). A study could be given one star for each item under selection and outcome and two stars under comparability. For selection, a star was given when the exposed cohort was truly or somewhat representative of exposed adults in the community, when the non-exposed cohort was drawn from the same community as the exposed cohort, when the exposure (tooth loss) was ascertained through clinical examinations, and when the outcome of interest was measured both at baseline and follow-up. For comparability, a star was given when the study controlled for socio-demographic characteristics (sex, age and any socioeconomic position indicator) during the design or analysis, and it was given two stars when it additionally controlled for participants’ health status (chronic conditions, comorbidities, activities of daily living and the like). For outcome, one star was given when the assessment of outcome was independent/blinded or through record linkage, when the follow-up period was long enough for changes in outcomes to occur, and when all participants were accounted for during follow-up or those lost to follow-up were unlikely to introduce bias (<20% attrition rate and description provided of those lost). A good quality scored required 3–4 stars in selection domain AND 1–2 stars in comparability domain AND 2–3 stars in outcome domain; a fair quality study required 2 stars in selection domain AND 1–2 stars in comparability domain AND 2–3 stars in outcome domain; and a poor quality study 0–1 stars in selection domain OR 0 stars in comparability domain OR 0–1 stars in outcome domain .
2.4
Data synthesis
A meta-analysis of the findings (i.e. forest and funnel plots) was not feasible given the high level of heterogeneity found across studies. Instead, we opted for a narrative synthesis of the results . To that end, we created tables summarising the key methodological characteristics of all included studies and the methodological quality assessment of the studies based on NOS.
3
Results
A flow chart of the screening and selection of studies is shown in Fig. 1 . Of the 2232 unique citations retrieved, 2133 articles were excluded after screening titles and abstracts as clearly irrelevant. The full text of 99 articles was retrieved to check eligibility and 89 articles were subsequently removed as not meeting the inclusion criteria. The major cause for exclusion was using a cross-sectional design (n = 43). Therefore, a total of 10 reports in 8 cohorts were included in this systematic review.
Table 1 summarises the characteristics of the included studies. Two Japanese studies and two Unites States (US) studies used data from the same cohorts, the Niigata Study and the Health Professionals’ Follow-up Study, respectively. They were considered as different analyses of their respective cohorts. Thus, we summarised findings based on 8 original studies; 4 in the US, 2 in Japan, 1 in Australia and 1 in Brazil. Five studies were subsets of population-based cohorts whereas the three remaining studies recruited male health professionals , female nurses and patients admitted to hospital . The follow-up times of all studies varied from a few days to 10 years. Sample sizes ranged from 134 to 59,467 participants. Participants’ age varied from 30 to 65+ years.
| Authors | Study design | Study sample | Exposure | Outcomes | Covariates | Main findings |
|---|---|---|---|---|---|---|
| Sato et al. ; Iwasaki et al. | Panel study with 10-year follow-up (1998 to 2008) | 600 community-dwelling adults aged 70 years (Niigata Study, Japan); 41.8% attrition (n = 251) | Change in dentition status: 10+ occlusal supports (Zone A), 5–9 occlusal supports (Zone B), <4 occlusal supports or 11+ remaining teeth (Zone C), and ≤10 remaining teeth (Zone C) | Number of chewable items (0–8): peanuts, pickled daikon radish, hard-baked rice crackers, beefsteak, vinegar octopus, scallions, dried squid and raw squid at baseline and follow-up | None | The number of food items that could be chewed decreased in subjects who remained in Zone A, those who changed from Zone B to Zone D, and those who changed from Zone B to Zone C. |
| Longitudinal study with 5-year follow-up (2003 to 2008) | 370 free-living adults aged 75 years (Niigata Study, Japan); 21% attrition (n = 79) | FTU (pair of opposing natural or prosthetic teeth excluding third molars) in 2003 | Total energy, protein, carbohydrates, fat, sodium, potassium, calcium, vitamins A, D, E, B6, B12, folate and dietary fibre from a 1-month brief-type diet history questionnaire at baseline and follow-up | Sex, education, income, smoking status, ADL, BMI, comorbidities | Greater decline in protein, sodium, potassium, calcium, dietary fibre and vitamins A and E intake as well as in vegetable and meat intake in those with impaired dentition than in those without impaired dentition. | |
| de Andrade et al. | Longitudinal study with 4-year follow up (2006–2010) | 1413 community-dwelling adults aged 60+ years (Survey on Health, Well-being and Aging, Brazil); 30% attrition rate (n = 423) | Edentulism (no/yes) and need for dental prostheses (no/yes) in 2006 | Changes in weight and WC from 2006 to 2010: stable (within 5% of 2006 values), loss (decrease of 5% or more) and gain (increase of 5% or more) | Sex, age, education, number of self-reported chronic diseases, baseline weight and WC, smoking status and physical activity | The odds of weight and WC loss were higher among edentate than dentate adults. Edentulism was not associated with weight or WC gain. The need for dental prostheses was not associated with change in weight or WC |
| Mudge et al. | Longitudinal study with 2-to-6-day follow-up | 134 patients aged 65+ years admitted to general medical wards in 2007/08 (Australia); no attrition rate | Poor dentition defined as missing teeth or ill-fitting or absent dentures by a dietitian | BMI, MNA, and inadequate energy (energy intake less than REE) from visual estimation of plate waste on a single day between days 3 and 7 of admission | Sex, age, residence, diagnosis, co-morbidities, medications and hospital ward | There was no association between poor dentition or nutritional status and inadequacy energy intake |
| Kwon et al. | Panel study with 8-year follow-up (1992–2000) | 738 free-living adults aged 65+ years (Longitudinal Interdisciplinary Study on Aging, Japan); 43.5% attrition rate (n = 321) | Self-reported chewing ability in 1992 (good/poor) and changes in chewing ability from 1992 to 2000 (always good, deteriorating, improving and always poor) | DVS (0–10, counting items eaten: meat, eggs, fish and shellfish, milk, dark-coloured vegetables, soybean products, potatoes, fruits, seaweeds, and fats and oils) from a 1-week FFQ. Decline was defined as a change in DVS < = −2 points from 1992 to 2000 | Sex, age, education, baseline functional capacity and DVS, change in spouse status and new chronic diseases during the study period | Adults with deteriorating self-perceived chewing ability had greater odds of experiencing a decline in dietary variety |
| Hung et al. | Longitudinal with 4-year follow-up (1990–1994) | 59,467 female nurses aged 46–71 years, with 11+ remaining teeth and who completed FFQ in 1990 and 1994 were analysed; (Nurses’ Health Study, USA); attrition was not reported | Self-reported number of teeth lost in the past 2 years (0, 1–4, 5 + ) in 1992 | Changes in total energy, saturated, trans, mono- and poly-unsaturated fats, cholesterol, fibre, carotene, beta-carotene, vitamins C, E, B6, B12, folate, potassium, flavonoids, fruits and vegetables from 1-year semi-quantitative FFQ over 4 years | Total energy intake, age, physical activity and smoking status | Women who lost 5+ teeth had smaller reduction in intake of monounsaturated fats whereas women with 1–4 teeth lost had smaller reductions in saturated fat, trans fat and cholesterol, and smaller increases in fibre, carotene, vitamin C, and potassium, and greater reduction in folate than women with no tooth loss |
| Lee et al. | Longitudinal with 1-year follow-up (1997–1998) | 3068 free-living adults aged 70–79 years (Health ABC Study, USA); 11.7% attrition rate (n = 362) | Self-reported edentulism in 1997 | Weight change from baseline to follow-up (loss: loss of >5%, stable: ±5% weight change, gain: gain of >5% of baseline body weight) | Sex, age, race, education, living alone, study site and family income | Edentate adults were more likely to have weight gains than were dentate adults, even after controlling for confounders |
| Hung et al. ; Joshipura et al. | Panel study with 8-year follow-up (1986–1994) | 31,813 male health professionals aged 40–75 years with 11+ teeth at baseline and answered FFQ (Health Professionals’ Follow-up Study, USA); attrition was not reported | Self-reported number of teeth lost (0, 1–4 and 5 + ) during the 8-year period | Total calories, carbohydrate, fats, dietary fibre, vitamin, fruits and vegetables from 1-year semi-quantitative FFQ in 1986 and 1994 | Total energy intake and baseline dietary intake, age, number of teeth, smoking, physical activity and profession | Adults who lost 5+ teeth had smaller reduction in consumption of dietary cholesterol and vitamin B12, greater reduction in consumption of polyunsaturated fat and smaller increase in consumption of dietary fibre and whole fruit than those who had lost no teeth |
| Panel study with 4-year follow-up (1986–1990) | 49,501 male health professionals aged 40–80 years with 17+ teeth at baseline and who answered FFQ (Health Professionals’ Follow-up Study, USA); 37.9% attrition (n = 18,763) | Self-reported number of teeth lost (no tooth loss versus 5+ teeth lost over 4 years | Dietary fibre, crude fibre, carotene, cholesterol, saturated fat, fruits servings (excluding juices) and vegetables (servings) from 1-year semi-quantitative FFQ in 1986 and 1990 | Baseline intake of nutrient, age, health profession, smoking status and exercise | Participants who lost 5+ teeth reduced their cholesterol intake by 11 milligrams compared to those with no tooth loss | |
| Ritchie et al. | Longitudinal study with 1-year follow-up | 979 community-dwelling adults aged 70–96 years (New England Elders Dental Study, USA); 25.5% attrition (n = 250). | Self-reported chewing difficulty and clinical measures: dentate status, number of teeth, functional units and chewing surfaces (number of adjacent functional units) | Weight loss using 2 thresholds (≥4% and ≥10% of baseline weight) | Sex, income, advanced age (≥80 years), >2 chronic conditions, dependence in 1+ daily activities and baseline weight | Edentulousness was an independent risk factor for weight loss after adjusting for sex, income, age and baseline weight |
Stay updated, free dental videos. Join our Telegram channel
VIDEdental - Online dental courses
