Differences in Functional Outcomes for Adult Patients with Prosthodontically-Treated and -Untreated Shortened Dental Arches: A Systematic Review

This review examined differences in functional outcomes and patient satisfaction when shortened dental arches are left untreated compared to their restoration to complete arch lengths with different prosthodontic interventions. Methods A protocol was developed according to the criteria for a systematic review. All relevant databases were searched to identify appropriate clinical trials regardless of language or publication status. Predetermined eligibility criteria were applied, trial quality assessed and data extracted for each study. Relevant outcomes assessed were: functioning ability, patient satisfaction and harmful effects on oral structures. Results Searches yielded 101 articles: 81 from electronic databases and 20 from reference lists of retrieved articles (PEARLing searches). Sixty-nine citations were assessed for eligibility after removing 32 duplicate records. After reading titles and abstracts, a total of 41 records were excluded and the full-texts of the remaining 28 records were read. Only 21 records were included for the SR because 7 records were excluded after reading the full-text reports. These 21 records report the outcomes of four randomized controlled trials (RCTs) and one non-randomized clinical trial (CT) which were pre-specified and used for this review. No on-going studies were found and no eligible studies were excluded for failure to report the reviewer’s pre-specified outcomes. Outcomes were reported in the retrieved 21 articles. A narrative explanation of the pre-specified outcomes is reported for the 3 comparison groups (which were based on the different interventions used for the individual clinical trials). The shortened dental arch as a treatment option is encouraging in terms of functioning, patient satisfaction and cost-effectiveness. By using only high quality studies it was expected that the results would be more reliable when making conclusions and recommendations, but some of the included studies had to be downgraded due to methodological errors.


Introduction
Prosthodontic treatment planning customarily includes the replacement of all missing teeth with the intention of achieving complete dental arches (CDAs) comprising 28 teeth [1][2][3]. The rationale for this approach includes impaired oral function with a perceived detrimental impact on chewing ability, occlusal stability and temporomandibular joint (TMJ) function due to the loss of the molar teeth [4]. On the other hand, several studies and reviews have indicated that twenty occluding teeth provide sufficient oral functional ability and the need to replace all missing posterior teeth has been questioned [3][4][5][6][7][8][9][10][11].
Several research reports tend to support the view that the underlying objective of the SDA to preserve a functional dental arch can be realized through a functionally-oriented treatment approach [5,[15][16][17]22,24,26]. This entails directing the limited resources towards that part of the dentition that can be successfully preserved and in the most cost-effective manner, rather than on the remaining molar teeth that often have a poorer prognosis [5,7,. The minimum number of teeth or shortness of the arch will also depend on the periodontal condition of the remaining teeth, the age of the patient, occlusal activity, food types and adaptive capacity of the patients' temporomandibular joints [3,7,9].
Following the large body of published research data related to the SDA conducted in different parts of the world, several efforts at collating these data have been made. Thus a number of systematic reviews (SR) focusing on the SDA have been completed [8,[56][57][58]. A SR conducted by Gotfredsen and Walls (2007) focused on studies that reported on the assessment of normative needs only, although it did not include quality of life studies that considered the perceived oral health needs of partially dentate patients [8]. In the SR by Fueki et al (2011), different types of study designs were included, in addition to the randomised controlled trials (RCT) [56]. The quality of evidence from longitudinal studies related to restorative and non-restorative approaches to adult patients with SDAs were assessed by Faggion (2011) using GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach [58]. With this study, even though all the results from the included studies were not reported, it demonstrated how important methodological rigor is and that these need to be reported [58]. In a recent electronic search in the Cochrane database for systematic reviews, Abt, Carr and Worthington (2012) focused on a broad research question to include all different types of interventions for partially dentate patients, including the SDA [57]. No conclusive evidence was found to indicate that any intervention was better for partially dentate patients, irrespective of particular interventions, procedures or materials used [57].
The aim of this systematic review was to identify and analyse existing clinical trials which compare the functional outcomes of prosthodontic interventions used for treating shortened arches versus un-restored shortened arches in partially dentate adult patients.
The following research question addresses the aim and objectives of the study: In adult patients with shortened dental arches, what is the effect of prosthodontics interventions on the functional outcomes compared to having no treatment?

Protocol Development
A protocol (Registration No: 11/4/39) was developed (not published) to include all aspects of a SR namely: selection criteria, search strategy, selection methods using predetermined eligibility criteria, data collection, data extraction, assessment of risk of bias using the Cochrane tool, the GRADE tool to grade the evidence of each clinical trial and statistical analysis by calculating risk ratios (RR) for dichotomous outcomes and presented at 95% confidence intervals [59][60].

Criteria for considering studies for this review
Types of studies. Only RCTs and Clinical trials (CTs) are included in the systematic review (SR).
Types of interventions. Interventions included in this study are described as any prosthodontic intervention used to restore and treat the SDA such as RPDPs and FDPs. The control group for this study included patients with the classic SDA.
Types of participants. Participants included in the SR were: 1. Adult male and female participants aged 18 years and older. 2. Study population included patients with posteriorly reduced or shortened dental arches.
Types of outcome measures. Primary and secondary outcomes were pre-specified for the SR and these include: Primary outcomes 1) Functional outcomes (patient-or investigator-reported) as measured by masticatory function, chewing ability, occlusal effects, nutrient intake (using nutritional assessments and haematological markers) and subjective functioning ability. 2) Survival of the interventions (fixed or removable partial denture prostheses) used for the extension of SDAs. Inclusion criteria. Studies that included above interventions and outcomes and addressed the pre-specified outcomes were eligible for this SR.

Secondary outcomes
Exclusion criteria. The following study designs: casecontrol, cross-sectional and cohort studies; case-series and case reports; other SRs; analytical and narrative reviews and different types of animal studies that were not eligible for inclusion, were excluded.
Search strategy. All relevant databases were searched: Medline, Cochrane Central Register of Controlled Trials, EMBASE, CINAHL, Science Direct, ProQuest, Science Journals, Scopus, PsycINFO, ClinicalTrials.gov, WHO ICTRP, TRIP and PACTR. Further hand-searching was conducted including citations from reference lists of retrieved studies (PEARLing searches) for additional references [59]. Where data were missing and full texts unavailable, these unclear reports were clarified by contacting authors or research institutes. Efforts were made to obtain English versions of studies reported in other languages either by requesting English versions from authors or using language experts to translate key findings. Authors were also contacted for unpublished reports or conference proceedings, where it was needed. Where registries were available for on-going studies, these were included as well and experts in the field of research related to the SDA were contacted.
Key terms were combined using Boolean operators and search strategies for each database were developed using the database specific functions [59]. Medical subject headings were applied in databases which allowed this function [59]. A wide search strategy was developed and modified according to the requirements of the different databases to ensure no eligible studies were excluded and an example includes the following: (shortened dental arch OR shortened dental arches) AND (Clinical Trial OR Comparative Study OR Evaluation Studies OR Randomized Controlled Trial OR clinical trial) AND 1980/ 01/01-2014/12/31).
Search limits. Databases were searched for articles of over a period of three decades from 1980 to April 2014. The limits included in the search strategy were: human studies, adult patients and randomized and non-randomized controlled clinical trials.
Selection methods. Two review authors (SK and AM) independently screened titles and abstracts from the electronic searches to select potentially relevant studies using a predetermined eligibility form based on the inclusion criteria [59]. Full text articles of potential studies were then retrieved and re-assessed for eligibility. Each article was scrutinized to ensure that multiple publications from the same study were included only once. Where eligibility was unclear, clarification was sought from the trial authors and the corresponding articles were re-assessed. Differences between the eligibility results were resolved by consulting the other review authors (UMEC and RO). Studies that did not meet the inclusion criteria were excluded and the reasons for exclusion were reported. Data extraction for the selected studies was completed by the principal researcher (SK) using a specially designed pre-piloted data extraction form for this SR [59]. All disagreements regarding this process were resolved through discussion with the other review authors (AM, UMEC and RO).
Qualitative analysis. The quality of the studies included for this SR were evaluated for any risk of bias by researchers (SK and AM) using the Cochrane Risk of Bias tool and as described in the Cochrane Handbook for Systematic Reviews of Interventions [59]. The assessment was done across the following six components: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other bias. Each of these were judged as 'yes', 'no', or 'unclear' corresponding to low, high, or unclear risk of bias respectively. Where information in the articles was insufficient for making the judgements, trial authors were contacted for clarification. Disagreements were resolved through discussion with other review authors. Results of risk of bias were summarised in a risk of bias table. In addition, GRADE assessments were completed by the researchers (SK and AM) for each clinical trial and these were used to grade the evidence and strength of recommendations for clinical intervention (where possible) using the GRADE Profiler system [60]. These are reported in the summary of findings tables.
Data synthesis and management. Results were reported separately for the following three comparisons: 1) FDP versus RPDP; 2) RPDP versus no treatment (SDA); 3) SDA versus CDA. No imputation of missing data was carried out and study authors were requested to provide any missing data. Available case analysis was applied where data were missing. Risk ratios with corresponding 95% confidence intervals were calculated for dichotomous outcomes using Review Manager 5 software. Although a meta-analysis of outcomes across study results had been anticipated, the included studies reported outcomes in different forms that could not be pooled in a meta-analysis. Consequently, results for individual studies were reported separately.

Results
The search strategy identified a total of 101 citations ( Figure 1): electronic databases yielded 81 and 20 were from reference lists of retrieved articles (that is, through PEARLing searches). A total of 32 duplicate records were removed, leaving 69 citations which were assessed for eligibility. After reading titles and abstracts, a total of 41 records were excluded and the full-text of the remaining 28 records was retrieved. A further 7 records were excluded after reading the full-text reports, leaving the remaining 21 records as included studies for the SR (Figure 1). Only four RCTs and one CT were used for this review, but outcomes were reported in the retrieved 21 records [14,[31][32][33][34][35][36][37][38][40][41][42][43][44][45][46][47][48][53][54][55]. No on-going studies were found and no eligible studies were excluded for failure to report the reviewer's pre-specified outcomes.
Qualitative analysis. Table 2 specifies the quality assessment of the included studies and these are summarized in the 'risk of bias table' and 'risk of bias graph' where judgements are categorized to indicate a low, high, or unclear risk of bias ( Figure 2) following the Cochrane guidelines [59]. Below we give a detailed explanation of these results: Sequence Generation: Three of the five trials were reported as having been randomised. For sequence generation, two clinical trials used computer-generated numbers and a third trial used randomly permuted block randomisation for generating the allocation sequence, which we judged as having a low risk of bias [34][35][36][37][38][40][41][42][43][44][45][46][47][48]. The Witter et al (2001) clinical trial invited subjects to join the department for a study, and no attempt was made to randomise patients, thus it is judged as having a high risk of bias [14,[53][54][55]. The Budtz-Jorgensen and Isidor (1987) trial did not mention how the sequence was generated and provided insufficient information to enable us to judge whether there was a high or low risk of bias, and we thus rated it as having an unclear risk of bias [31][32][33].
Blinding: The Moynihan et al (2000) study was referred to as a double blinded study with the clinician blinded to allocation of intervention and statistician being blinded to treatment and thus it is judged as having a low risk of bias [34][35][36][37][38]. The Witter et al (2001) study can be considered as a single blinded study because evaluation of outcomes was completed by a calibrated observer at all intervals, but it was not stated as such, thus it is judged as having an unclear risk of bias [14,[53][54][55]. Mc Kenna (2012) indicated that the researcher was not involved in the intervention allocation, making it a single-blinded study, thus it is judged as having a low risk of bias [47][48]. The Wolfart et al (2005) study indicated that it was impossible to blind the dentist and patient due to discrepancies of the treatments; thus it was judged as having a high risk of bias, whereas Budtz-Jorgensen and Isidor (1987) provided insufficient information related to blinding and it was regarded as having an unclear risk of bias [31][32][33][40][41][42][43][44][45][46].  (2012) studies were conducted on the ''intention-to-treat'' (ITT) principle; and the studies reported proportionate numbers of losses to follow-up (which were small) and some having no losses between the intervention and control [34][35][36][37][38][40][41][42][43][44][45][46][47][48]. Witter et al (2001) indicated that regression models accounted for the subjects lost during the study [53]. Thus, all 4 studies above were judged as having a low risk of bias [34][35][36][37][38][40][41][42][43][44][45][46][53][54][55]. On the other hand, Budtz-Jorgensen and Isidor (1987) did not indicate and specify how the analysis was completed, but all pre-specified outcomes were reported, and the number of losses to follow-up was small, thus it was judged as having a low risk of bias [31][32][33].
Other potential sources of bias: No other sources of bias were detected with four of the five included studies. The Budtz-Jorgensen and Isidor (1987) study was judged as having high risk of bias because there were six patients who did not wear the RPDP at all during the study [32][33].

Effects of interventions
See: Summary of findings for the main comparisons of functional outcomes and patient satisfaction with FDPs compared to RPDPs in treating patients with SDAs (  [35]. Similarly, Isidor and Budtz-Jorgensen (1990) observed 22 dental carious lesions in the RPDP group compared with only two lesions in the FDP group; however we could not calculate a treatment effect since the respective number of patients was not reported. Our unit of analysis was individual patients and not individual teeth [33].
The following effects were only reported for the Budtz-Jorgensen and Isidor study (33): TMJ dysfunction: Isidor and Budtz-Jorgensen (1990) found no significant difference in the number of patients showing TMJ dysfunction between the FDP and RPDP groups (RR 0.64, 95%CI: 0.36 to 1.16, 53 participants) [33].
Tooth Loss: In the Isidor and Budtz-Jorgensen (1990) study, 11 teeth were extracted in the RPDP group compared with only one tooth in the FDP group during the five years of observation. However, no treatment effect could be calculated because the respective numbers of patients were not reported [33].
Plaque Index: Isidor and Budtz-Jorgensen (1990) reported the mean plaque index ranging from 0.4 to 0.7 in the FDP group and from 0.7 to 1.0 in the RPDP group; the difference between the two groups was significant (p,0.05) during the first two years of examination as reported by study authors [33].
Gingival Index: Isidor and Budtz-Jorgensen (1990) indicated that the mean gingival index was always higher in the RPDP than in the FDP group, the difference being significant (p,0.05) at the 12-, 18-, 36-, and 48-month examinations [33].  Table 6 [47]. The values in the table were used to calculate a treatment effect which showed no significant difference in the change in MNA score between the RPDP and SDA treatment groups (MD 20.03, 95%CI: -1.35 to 1.29, 42 participants:  [44]. The respective Kaplan-Meier survival rates at 38 months were 0.83 (95%CI: 0.74 to 0.91) in the RPDP group and 0.86 (95%CI: 0.78 to 0.95) in the SDA group, the difference is not significant (as reported by study authors) [44].
Occlusal tooth wear:  reported the mean occlusal tooth wear scores using transformed values for subjects of 40 years of age [55]. However, no significant differences between the SDA subgroups [means (SD) ranging from 1.1(0.1) to 1.6(0. . They also reported that spacing remained the same in all regions over time in the SDA group [55]. No treatment effect could be calculated because the results were given per region and also because the respective number of patients were not specified in the results.
Overbite:  stated this outcome only for some subgroups but did not compare their results between the SDA and CDA groups [55]. Therefore we could not calculate a treatment effect.
Periodontal support: Witter et al (1994) described the mean relative bone heights using transformed values for subjects of 40 years of age [55]. The authors reported that maxillary premolars and mandibular second premolars in the SDA subgroups showed significantly lower mean bone height scores than those in the CDA group, whereas mandibular first premolars did not differ. The values reported were not sufficient for the calculation of a treatment effect.
TMJ problems: The Witter et al study (2007) indicated that patients with SDAs (65-79%) had similar prevalence, severity and changes in signs and symptoms related to the TMJ as patients with CDAs (70-75%) [54].
Excluded study characteristics: All non-RCTs and reviews were excluded from this SR. Other SRs and summary articles were viewed as potentially included studies, but these were however later not considered for inclusion (Table 8).
Survival of fixed bridges 5 years post study was similar to other trials [30][31][32][37][38]. RPDP patients chose not to wear RPDPs which was similar to other studies [31][32][37][38]. For patient satisfaction, the small sample size does not allow us to generalize our results to other settings, thus it is advised to conduct these studies amongst different populations.
For the Wolfart et al study (2010): Post hoc power calculations implied that the pilot sample size was too small to generalize results and for comparison to other studies [40][41][42][43]. The larger study results are free of bias with a large enough sample due to it being a multi-centre study. While it reduced the bias, it still could not be generalized to patients that are different to the study  [44]. Tooth loss as a primary outcome is questioned due to extended time periods, thus it was advised to use caries and periodontal attachment loss as outcomes instead [44]. The Mc Kenna study (2012), which is the most recently conducted RCT; the results are similar to other RCTs completed in the past, where small sample sizes would not necessarily show a significant difference between interventions given the follow-up period [47][48]. In this case, follow-up after only one month of treatment was too short to show any difference between interventions [47][48]. But the cost-effectiveness reported with this RCT has been noted as researchers and clinicians are under the impression that the cost for FDPs far outweighs that of RPDP treatment [22,39,48]. And this has been in line with the findings of the Danish study published some years ago [32][33].
For the Witter et al study (2001), results were similar to other studies with regards to outcomes reported and the effect of outcomes on the dentition in the SDA group (tooth wear, TMJ effects) (Aukes, 1988;Mohl, 1988;Eliasson, 1997, cited in Witter et al (2001 [53].
The quality of the evidence is indicative of the integrity of the study and the research conducted. With reference to the quality assessment of the included studies, this has been described in detail above. More importantly, this quality is determined by the study designs. Study designs are graded according to the quality of evidence that they provide. Systematic reviews and RCTs are considered to be designs of the highest quality [59][60]. Within the different design groups, certain concessions can be made for those designs that do not follow the exact guidelines [59][60]. For instance RCTs can be downgraded if their risk of bias is high [59][60].
Only RCTs and CTs were however included in this systematic review which provides stronger evidence and increases the strength of the recommendations [59][60]. After completing the quality assessment (using the GRADE approach) of the included studies, it clearly showed that some of the studies had not followed the exact guidelines for RCTs, but nevertheless had the features thereof [59][60]. These can be regarded as downgraded RCTs (Tables 3-4). These downgraded RCTs did not use randomization, allocation concealment or blinding, and failed to specify the outcomes as primary or secondary. These downgraded RCTs could thus affect the quality of evidence only slightly [59][60]. For example, the Budtz-Jorgensen (1987,1990) and Witter et al (2001) studies could be regarded as downgraded CTs [14,[31][32][33][53][54][55]60].
A meta-analysis could not be completed for this SR for the following reasons: Some of the outcomes for the SR (for example survival of intervention) were not reported by all the included studies; sufficient RCTs were not found related to SDAs; the outcomes were reported in so many different ways for each of the studies that a narrative approach for this review had to be adopted and not all outcomes are reported for the Wolfart et al (2005) study (and no correspondence was received when the authors were contacted). In addition, there was insufficient information reported by studies to allow us to combine continuous data using the mean difference (MD). The outcomes from the studies were thus grouped for this review.
For this SR, a systematic approach to the evaluation of the evidence obtained from the studies was adopted by the researchers and disagreements were resolved by discussion. The researchers highlighted the areas where bias could have been expected (Table 2). Study samples, settings, age categories, interventions and outcomes for the included studies were mostly similar, creating strong evidence (Table 1). Comparison between the groups of the different studies could be systematically recorded in the stipulated groups. And again, for this SR all potential sources were searched and reported. Most studies followed guidelines to protect against bias (some without making reference to the method followed) [14,[31][32][33][53][54][55]. And this was assessed using the Cochrane's risk of bias tool [59]. Since all the included studies in this SR were conducted in developed countries, our findings cannot be generalized to patients in all countries because cultural and socio-economic differences that exist between countries and within communities can influence patients' reactions.
Other SRs were also conducted in the past ten years [8,[56][57][58], where researchers included studies with different study designs and not only RCTs. For the most current SR [57], the research question was so broad that the focus on the SDA was minimal, thus many of the data related specifically to SDAs were not even included in the analysis [57]. For this SR, only the British and Table 5. Summary satisfaction scores for the UK-based study at 1 year (a lower score indicates more satisfaction). German RCTs were mentioned and only the results of the pilot study for the German RCT was reported [57]. The authors concluded citing insufficient evidence to report a difference between RPDP and FDPs in the treatment of SDAs [57]. In addition, when evaluating the quality of the evidence of a systematic review, it is recommended that the GRADE approach should be used [60]. It is a method of evaluating the quality of evidence and strength of recommendations in healthcare, and thus provides the needed rigor and transparency when making specific recommendations [60].

Quality of evidence
As stated above, the quality of evidence was assessed using the GRADE methodology for this SR (Tables 3 and 4). With the assessment, the small sample sizes seriously affected the imprecision, and the risk of bias was very serious with studies where no blinding and selective reporting was observed (Tables 3 and 4).
From the combined effects, the overall quality of the assessment is regarded as being low (Tables 3 and 4). This implies that further research (as in conducting more RCTs) is likely to have an important impact on our confidence in the estimate of effect, and may change the estimate.

Implications for practice
The SDA concept has been researched and used in industrialized countries and this review aimed to highlight its appropriateness and relevance for a developing country such as South Africa. A change in paradigm or thinking should be encouraged, even though results of clinical trials conducted in other countries may not necessarily be generalizable to South African populations. By regarding the research related to SDAs in a positive light (patient satisfaction, caries incidence, TMJ effects and tooth loss), this SR specifies that policy-makers and/or institutions should be encouraged and recommend its teaching and clinical implemen- tation by students and clinicians. These are considered as instances where low-quality evidence can still make a strong recommendation due to the body of available evidence on SDAs.

Implications for research
Sufficient RCTs related to SDAs were not found, and thus it would be advisable to conduct more randomized clinical trials. The RCTs were also conducted in European and Nordic countries and these results may not be generalizable to other context, due to substantial cross-cultural and socio-economic differences between countries. External validity or generalizability of studies conducted in other countries depends on: settings where studies were conducted; participants' characteristics; interventions researched across studies; relevance of the endpoints achieved with each study; results obtained and their comparison to one another and the indirect/direct costs when conducting each study.

Conclusions
The results from this SR related to SDAs as a treatment option were encouraging in terms of functioning, patient satisfaction and cost-effectiveness. However, only the Moynihan et al (2000) study reported on the primary outcome of survival of the SDA, and had this been determined by the other studies, it would have strengthened the recommendation of the SDA as a treatment option even further [34].

Recommendations
The stronger the evidence, the stronger the recommendation for the implementation of the SDA as a treatment option for partially dentate patients. By using only high quality studies such as RCTs and CTs for this SR, it was expected that the results would be more reliable when making conclusions and recommendations. Nevertheless, any conclusion/s from such a SR can still be regarded in a positive light, even though the included studies had to be downgraded due to methodological errors [60]. It is also recommended that when conducting clinical trials, strict protocols need to be prepared and the reporting of the RCT should follow the CONSORT guidelines [61]. This could then be of great benefit to other researchers when critically appraising these clinical trials. More importantly, outcomes for the RCT have to be prespecified and all should be reported so that future systematic reviews may be conducted with the inclusion of a meta-analysis, instead of a narrative report as needed to be done for this SR. Thus further research (as in conducting clinical trials) should be encouraged and for the different settings and contexts (for example developing countries) to create a comprehensive database related to SDAs.