The OIDP is a self-administered instrument that measures the effect of oral impacts on an individual’s ability to perform eight daily performances: eating and enjoying food; speaking and pronouncing clearly; cleaning teeth; sleeping and relaxing; smiling; laughing and showing teeth without embarrassment; maintaining usual emotional state without being irritable; carrying out major work or social roles; and, enjoying contact with people. The total impact of each performance is calculated by multiplying a frequency score with a severity score. Frequency scores are obtained using the criteria used for the description of both frequency (for people affected on a regular or periodic basis) and the duration (for people affected for a period/spell) Severity scores are obtained by asking respondents to rate each item, ranging from 0 to 5, as an indication of how much it impacted on their daily living. The total score is the sum of all the performance scores for an individual. Then sum is divided by the maximum possible score and multiplying by 100 to give a percentage score [11].
The process of adapting the OIDP for Sri Lankan adolescents and evaluating of its psychometric properties involved two stages, summarized in Figure 1. Both were conducted in Gampaha zone, Gampaha district, Sri Lanka. Stage I was carried out in the 1st quarter of 2015, followed by the stage II in the 2nd and 3rd quarters of 2015. Administrative clearance was obtained from the relevant education and health ministerial personnel and the study protocol was approved by Ethics Committee of Colombo Medical Faculty (Ref No EC 15-171).
Stage I: Modification for Sri Lankan adolescents
Stage I involved three main steps: linguistic translation; cultural adaptation and pretesting; and simplifying the scoring system and shortening the recall period. The process of cross-cultural adaptation including translation, adaptation and pretesting adopted the methods recommended by Guillemin and Beaton et al [12, 13].
As the original version of the OIDP is in English, a Sinhalese translation was produced before any modifications were made. The original version of the OIDP was given to two translators whose first language was Sinhalese. Translation and back translation methods were applied and a third independent expert compared the back translated version with the original version and discrepancies were resolved with the consensus of the two translators [14].
The eight items of the modified OIDP were adapted for relevance to an adolescent population, while keeping the dimensions consistent with the original OIDP tool [11, 15] . A panel of experts including three specialists in community dentistry, two specialists in community medicine, two specialists in restorative dentistry, one specialist in orthodontics, one specialist in oral and maxillo-facial surgery and a sociologist were involved in this process. The experts were selected for their specialized knowledge, experience and unique perspectives on the content of the instrument [16, 17]. Public health expertise, clinical expertise, national representation and experience in research on the phenomenon of interest were used as the criteria when selecting content experts in to the panel of experts of the present study. Public health experts with experience in the designing and validation of measurement tools were prioritized. Adaptations are described in Table 1.
A draft of the modified OIDP scale was pretested by interviewing a convenient sample of 20 adolescents, aged 15-19 years. These participants were native Sinhalese speakers recruited from a secondary school in Gampaha Zone, Gampaha District. The Gampaha zone is located in the central part of the Gampaha district. It is an urbanized area with relatively high socio-economic indicators when compared to other zones in the District. The school was selected from the school list of the Gampaha Zone using a random number table. The interviewer recorded any difficulties that subjects had encountered, along with their comments. All records were reviewed by a study investigator and a discussion session with the interviewer. Six participants were followed up in order to clarify their comments. A series of re-interviews were conducted two weeks following the initial interviews in a subset of 10 participants in order to gain further insights into the scoring system and recall period.
The pre-testing process revealed that several adolescents gave a different set of responses in the re-interviews, unless the impacts were extremely low or extremely high. It was therefore determined by the panel of experts to limit the scoring system to a severity score only, as these responses were more consistent than those given for frequency. This is consistent with findings reported by the authors of the original instrument which suggest that, as the multiplication of both frequency and severity scores did not show any significant improvement over using the frequency or severity score alone, either the frequency or the severity score could be used alone for simplicity [11]. Modified OIDP scores were recorded on a six-point likert scale to reflect how severe the impact of each event was over the recall period, ranging from 0 (indicating no impact), to 5 (indicating a very severe impact). The total modified OIDP scores for individual domains were calculated as a simple sum of the response codes. Total modified OIDP scores could range from 0 to 40, where higher OIDP scores indicate poorer OHRQoL.
The pretesting further revealed that adolescents had poor memory of their oral health impacts over six months, as they gave different answers during the re-interview. The consensus of the panel of experts was therefore to shorten the recall period to three months. This is consistent with previous studies in children conducted in Brazil, France and India that also used the OIDP tool with a three months recall period [18-20], as well as a study that modified the OIDP scale for children in Thailand, without impacting on the validity of the tool [10]
The final modified OIDP consists of 8 self-rated items which ask participants to assess the impact severity of eight daily performances over the past three months. A full list of modifications are presented in additional file 1. The final modified OIDP is included in additional file 2.
Stage II
Stage II involved in exploring the factor structure and assessing the validity of the modified OIDP scale.
Exploring factor structure
The factor analysis and the psychometric properties of the modified OIDP scale were assessed in a sample of 15-19 year school children from a secondary school in Gampaha zone, Dompe Medical officer of health area. Two classes were randomly selected from each grade (Grade 10 to Grade 13) to ensure the minimum sample size was met. The recommended minimum participant-to-item ratio in exploratory factor analysis is 5:1. A widely acceptable rule of thumb is 10:1 [21, 22]. We adopted a conservative 20:1 participant-to-item ratio to derive a minimum sample size of 160. A total of 220 participants from eight classes were recruited for the data collection. Data collection commenced by providing participants with the modified OIDP scale to be completed at the school premises during their class time as a self-administered instrument. All quantitative analyses were performed using the Statistical Package for Social Sciences (SPSS) version 23 by the study investigator. Socio-demographic data of the participants were described in frequency tables as numbers and percentages. No missing data were reported.
Using the approach described in Tabachnick and Fidell (2007), inspection of correlation matrix was performed to assess factorability [23]. Prior to proceeding further with factor extraction, Kaiser-Meyer-Olkin (KMO) a measure of Sampling Adequacy and Bartlett's Test of Sphericity tests were performed. Williams (2010) has suggested that the KMO index should be at least 0.50 and Bartlett's test of Sphericity should be significant (p<0.05) to be considered suitable for factor analysis [24, 25].
Factor extraction is generally applied to reduce a large number of items into common groups or factors [14]. After assessing the factorability of the scale, the factor analysis of the eight items of the modified OIDP scale was conducted using Principal Component Analysis (PCA) and Principal Axis Factoring (PAF), the two most commonly used factoring procedures in published literature [23, 24, 26]. Simultaneous use of multiple decision rules, namely Kaiser’s criteria, Scree test and cumulative percent of variance extracted were recommended and considered [27]. Once the number of factors or components was decided, we adopted PCA with oblimin rotation which demonstrated a clearer and more interpretable structure relative to others methods. We adapted the PCA with oblimin rotation in order to allow factors to correlate, which is a low-risk, high benefit choice when compared to the orthogonal rotations [21]. Tabachnick and Fidell (2007) suggested that factor loading of 0.3 was a good rule of thumb for the minimum factor loading of an item [23]. A factor with a fewer than three items is generally weak and unstable; five or more strongly loading items (0.5) are desirable and indicate a solid factor [28]. Tabachnick and Fidell (2007) further advised that decisions about number of factors and appropriate rotational method should ultimately be based on realistic criteria, over an arbitrary rule of thumb [29]. These criteria were utilized during the selection of factors and relevant items for the modified OIDP scale.
Validation
Psychometric analysis of the Sinhalese version of the modified OIDP involved the assessment of face, content and concurrent validity, as well as internal and test retest reliability assessment. The psychometric properties were assessed among the same sample that participated in the factor analysis. During that process, in addition to the modified OIDP scale, a questionnaire relating to perceived oral treatment need and perceived oral health problems were given to the participants.
Internal reliability was measured by using standardized Cronbach’s alpha coefficient, inter-item correlations and corrected item correlations [30]. It has been reported that Cronbach’s alpha coefficient should be at least 0.7 for early stage of research, 0.8 for basic research and 0.9 for clinical instruments and correlations need to be in moderate range, between 0.2 to 0.8 [14, 31].
In order to assess the test retest reliability, which provides an estimate of the degree to which the results are reproducible [32], a randomly selected subgroup of 20 participants from Stage II were given the modified OIDP scale to recomplete two weeks after their initial response. The total score of the two sets of data were compared to assess the correlation. As the modified OIDP scale presents continuous data which were not normally distributed, the non-parametric spearman rho test was used to calculate the total scores of the sub scales and for the total scale.
Since a gold standard measure cannot be identified to assess oral health related quality of life, criterion validity could not be achieved. Hence, face and content validity were assessed by ascertaining opinions from a second panel of experts [33]. The panel included three consultants in community dentistry, two consultants in community medicine, two consultants in restorative dentistry, one consultant in orthodontics and a sociologist. The panel members were selected based on the previous criteria used in selecting experts for the cross-cultural adaptation (stage I study). Each item in the instrument was checked for its relevance and appropriateness in the local context.
Concurrent validity was assessed by testing the modified OIDP scale against two subjective perceptions [10]; by assessing the self-reported perceived oral treatment need and perceived oral health problems. Due to the skewed nature of the modified OIDP scores, the non-parametric Kruksal-Wallis test was used to assess relationships between the modified OIDP and subjective perceptions.