UK Population Norms for the Modified Dental Anxiety Scale
UK Population Norms for the Modified Dental Anxiety Scale
A two-stage cluster sample was used for the survey comprising of 253 primary sampling units (PSU) across England and Wales, and a further 15 PSUs in Northern Ireland. Each PSU consisted of two postcode sectors with 25 addresses sampled from each, giving a total sample of 13,400 addresses. Of these 12,054 were eligible for inclusion (1,346 ineligibles were unoccupied households, business addresses, care homes etc.). These procedures were consistent with previous ADHS collections using multi-stage stratified sampling. Postcode sectors were paired together to help reduce the effects of clustering and increase the diversity of the population within each PSU. The pairing of neighbouring postcode sectors also helped reduce the design effect. The standard approach in the Office of National Statistics is to pair off contiguous PSUs into collapsed strata, and to base the variance of the estimator on the squared differences between PSUs within strata, summed over strata. The ONS policy is that it would not be appropriate to mix the postcode sector PSUs from multistage samples with those of single stage samples of households, therefore PSUs in the ADHS are paired. In each of the 10 English Strategic Health Authorities and in Wales, 1,150 addresses were sampled and 750 addresses were sampled in Northern Ireland. The survey took place between October-December 2009, and January-April 2010. All interviewers were trained on survey procedures.
Of the 12,054 eligible households, 7,233 participated (60% household response rate), while the remaining 3,895 households refused to participate or were non-contactable (n = 455) or other non-response (n = 471). Within the 7,233 households there were 13,509 adults who were asked to participate in the survey - of these 11,382 participated (84%). All individuals aged 16 yearsand older were invited to participate. So we had a household response rate of 60% (7,233 HHs) and an individual response rate (from within those households) of 84% (11,382 individuals).
A two stage weighting approach was adopted which ensured that the 1,150 addresses were sampled in each English SHA and in Wales, and a further 750 in Northern Ireland. A consequence of the aim to achieve similar sample size samples at the SHA level is that differential sampling rates were utilized in the SHAs, Wales and Northern Ireland. A survey weight had to be employed to compensate for these differential rates. As well as this weighting to address the sample design deficiencies, weighting was also employed to reduce bias attributed to non-response. Unfortunately, minimal information is available about non-responding households: however geographic information associated with non-responding households is available from the 2001 Census. This Census categorises each PSU based upon key characteristics including typical household type, social-economic status, typical ethnicity etc. Hence household non-response was based on the area a household was in. Details of this are contained in the Technical Report.
The ADHS included a clinical examination and a questionnaire. The content included: major indicators of oral health and function, dental diseases, urgent conditions such as pain and sepsis, complex treatments received, oral health risk factors and behaviour, service considerations and outcomes including access and barriers to care. The breadth of subject areas is too great (p21 Foundation Report) in scope for inclusion in this paper. Hence it is the later aspect of barriers and specifically, dental anxiety, that this paper is focused. Regularity of attendance was established from the questionnaire. The wording was: "In general do you go to the dentist for… (1) a regular check up, (2) an occasional check up, (3) or only when you're having trouble with your teeth/dentures".
To assess dental anxiety we employed the MDAS, which asks participants to rate: how anxious one feels the day before a dental appointment, then when in the waiting room, waiting for the receipt of drilling, scaling and a local anaesthetic injection. Responses range from 'not anxious' (scored 1) to 'extremely anxious' (scored 5). The five items are summed to create a total score, which has a range from a minimum of 5 to a maximum of 25. Total scores of 5 and 25 would denote: no dental anxiety and extreme dental anxiety, respectively. Reliability of the English language version from original investigation of the MDAS is good (internal consistency= 0.89; test-retest= 0.82). The scale can be downloaded:
http://medicine.st-andrews.ac.uk/supplemental/humphris/dentalAnxiety.htm. The item wording is reproduced in Table 1 and the scale layout can be reproduced from the dedicated website download.
The fieldwork procedures may be summarised as follows: the household was contacted initially by letter in advance of a household call. The household was informed that an interviewer would call to discuss the interview within a short period (days). To minimize the number of non-contacts (householders not contactable), all the interviewers were instructed to call at the addresses on different days, and at different times of the day (p17 Foundation Report). Participants were asked about demographic status, and other dental-related issues including the 5 questions of the MDAS by trained interviewers in the household.
A single application was submitted to NHS Research Ethics System (NRES) covering all aspects of the survey in England, Wales and Northern Ireland. Approval was granted in June 2009. All participants gave written consent.
Data were analysed using SPSS™ version 19 and AMOS™ version 19. Internal consistency and confirmatory factor analysis (CFA) was performed to assess, respectively, the internal consistency (Cronbach's alpha) and the level of fit (Chi-square, CFI, TLI & RMSEA) to a unidimensional model of scaling to a continuous latent construct. CFA was estimated using maximum likelihood and distribution free methods for comparison. All five items were described by a single latent variable. The first item was selected to set the factor coefficient to unity for identification purposes. This selection is generally regarded as arbitrary. Frequencies, means and standard deviations were calculated across the major demographic factors and self-reported visiting. A comparison was made between the original data set reported in 2008 and the current data using fixed factor analysis of variance, with and without adjustment for major demographic variables, namely: age, sex and socio-economic status.
A set of percentiles was prepared across gender and major age groups. A threshold of 19 and above was adopted, as the level for which is it likely that a dental practitioner would consider using additional approaches to manage the patient such as relaxation, systematic desensitisation or pharmacological adjunct. Fixed factor analysis of variance was performed utilizing the continuous scale data to inspect variation of dental anxiety across major demographic, behavioural and socioeconomic status factors. Significance level was set at the conventional 5%, two-tailed.
The standard method of obtaining percentile ranks was used. That is,
where m is the number of members of the normative sample obtaining a score lower than the score of interest, k is the number obtaining the score of interest, and N is the overall normative sample size.
As noted, a further aim of the present study was to accompany the point estimates of the percentile ranks corresponding to raw score with interval estimates of these quantities. A percentile rank is simply a proportion multiplied by 100 thus methods of obtaining an interval estimate of a proportion (such as classical methods based on the binomial distribution) can be used to obtain interval estimates of a percentile rank. However, for the present problem there is a complication. Although anxiety scores are discrete (i.e., integer-valued), the underlying dimension they index are generally taken to be continuous, real-valued quantities. Thus, a raw score of, say, 7 is regarded as a point estimate of a real-valued score which could lie anywhere in the interval 6.5 to 7.4999 (plus an infinite number of additional 9s after the 4 decimal place). Put another way, in principle we could distinguish among individuals obtaining the same raw score were we to introduce tie-breaking items. This assumption of a continuous underlying score is ubiquitous in psychological measurement and motivates the standard definition of a percentile rank (formula 1).
Normative data for scales such as the MDAS will always contain a sizeable number of tied scores; that is, a large number of people in the normative sample will obtain the same raw test score. Indeed, if a normative sample is large and the data are skewed (as would usually be the case for anxiety scales as the majority of the general population are not clinically anxious), then there could literally be hundreds of such ties for a given raw score. The present problem therefore differs from those dealt with by standard binomial sampling in which there can be no possibility of multiple ties.
Crawford, Garthwaite and Slick have recently developed Bayesian and classical methods that incorporate the additional uncertainty arising from tied scores. Crawford et al. and Crawford et al. have used these methods to provide interval estimates for self-report mood scales, such as the HADS, DASS, and PANAS; the methods have also been used to provide interval estimates for a variety of neuropsychological test scores. In the present study we apply the Bayesian method to MDAS scores.
To illustrate the issue the methods address: suppose that in a normative sample of 100 people, 89 obtained lower scores than a case and 2 obtained the same score as the case. Then the point estimate of the percentile rank for the case's score (using formula 1) is 90 and applying Crawford et al's. Bayesian method, the interval estimate is from 82.15 to 95.27. Suppose, however, that 85 obtained lower scores and 10 obtained the same score. The point estimate of the percentile rank is the same as in the foregoing example (90) but the interval estimate is from 79.79 to 97.10; the latter interval is wider because of the increased uncertainty introduced by the larger number of ties (10 versus 2). The technical details of these methods are not set out here: see Crawford et al. for their derivation, and for an additional mathematical treatment and evaluation, see Garthwaite and Crawford.
In practice there will be occasions in which a one-sided interval may be preferred over a two-sided interval. For example, a clinician may be interested in whether a patient's score is less extreme than is indicated by the point estimate but not particularly interested in whether the score is even more extreme, or vice-versa. The methods developed by Crawford et al. are easily adapted to provide a one-sided limit. However, without prior knowledge of which limit is of interest (the situation here, as the aim is to provide intervals for use by others) it is more convenient to generate 100(1-[α/2]) two-sided intervals which then provide 100(1-α) one-sided lower and upper limits. For example, if a 95% lower limit on the percentile rank is required then a 90% two-sided interval is generated: The user then simply disregards the upper limit of the two-sided interval and treats the lower limit as the desired one-sided 95% limit.
The point and interval estimates of percentile ranks for MDAS scores can be obtained using the tabled values provided in the present paper. However, we considered that some health professionals might find it more convenient if theses norms were also available via a computer program.
Methods
Sample and Procedure
A two-stage cluster sample was used for the survey comprising of 253 primary sampling units (PSU) across England and Wales, and a further 15 PSUs in Northern Ireland. Each PSU consisted of two postcode sectors with 25 addresses sampled from each, giving a total sample of 13,400 addresses. Of these 12,054 were eligible for inclusion (1,346 ineligibles were unoccupied households, business addresses, care homes etc.). These procedures were consistent with previous ADHS collections using multi-stage stratified sampling. Postcode sectors were paired together to help reduce the effects of clustering and increase the diversity of the population within each PSU. The pairing of neighbouring postcode sectors also helped reduce the design effect. The standard approach in the Office of National Statistics is to pair off contiguous PSUs into collapsed strata, and to base the variance of the estimator on the squared differences between PSUs within strata, summed over strata. The ONS policy is that it would not be appropriate to mix the postcode sector PSUs from multistage samples with those of single stage samples of households, therefore PSUs in the ADHS are paired. In each of the 10 English Strategic Health Authorities and in Wales, 1,150 addresses were sampled and 750 addresses were sampled in Northern Ireland. The survey took place between October-December 2009, and January-April 2010. All interviewers were trained on survey procedures.
Of the 12,054 eligible households, 7,233 participated (60% household response rate), while the remaining 3,895 households refused to participate or were non-contactable (n = 455) or other non-response (n = 471). Within the 7,233 households there were 13,509 adults who were asked to participate in the survey - of these 11,382 participated (84%). All individuals aged 16 yearsand older were invited to participate. So we had a household response rate of 60% (7,233 HHs) and an individual response rate (from within those households) of 84% (11,382 individuals).
A two stage weighting approach was adopted which ensured that the 1,150 addresses were sampled in each English SHA and in Wales, and a further 750 in Northern Ireland. A consequence of the aim to achieve similar sample size samples at the SHA level is that differential sampling rates were utilized in the SHAs, Wales and Northern Ireland. A survey weight had to be employed to compensate for these differential rates. As well as this weighting to address the sample design deficiencies, weighting was also employed to reduce bias attributed to non-response. Unfortunately, minimal information is available about non-responding households: however geographic information associated with non-responding households is available from the 2001 Census. This Census categorises each PSU based upon key characteristics including typical household type, social-economic status, typical ethnicity etc. Hence household non-response was based on the area a household was in. Details of this are contained in the Technical Report.
Questionnaire and Measures
The ADHS included a clinical examination and a questionnaire. The content included: major indicators of oral health and function, dental diseases, urgent conditions such as pain and sepsis, complex treatments received, oral health risk factors and behaviour, service considerations and outcomes including access and barriers to care. The breadth of subject areas is too great (p21 Foundation Report) in scope for inclusion in this paper. Hence it is the later aspect of barriers and specifically, dental anxiety, that this paper is focused. Regularity of attendance was established from the questionnaire. The wording was: "In general do you go to the dentist for… (1) a regular check up, (2) an occasional check up, (3) or only when you're having trouble with your teeth/dentures".
To assess dental anxiety we employed the MDAS, which asks participants to rate: how anxious one feels the day before a dental appointment, then when in the waiting room, waiting for the receipt of drilling, scaling and a local anaesthetic injection. Responses range from 'not anxious' (scored 1) to 'extremely anxious' (scored 5). The five items are summed to create a total score, which has a range from a minimum of 5 to a maximum of 25. Total scores of 5 and 25 would denote: no dental anxiety and extreme dental anxiety, respectively. Reliability of the English language version from original investigation of the MDAS is good (internal consistency= 0.89; test-retest= 0.82). The scale can be downloaded:
http://medicine.st-andrews.ac.uk/supplemental/humphris/dentalAnxiety.htm. The item wording is reproduced in Table 1 and the scale layout can be reproduced from the dedicated website download.
Procedure
The fieldwork procedures may be summarised as follows: the household was contacted initially by letter in advance of a household call. The household was informed that an interviewer would call to discuss the interview within a short period (days). To minimize the number of non-contacts (householders not contactable), all the interviewers were instructed to call at the addresses on different days, and at different times of the day (p17 Foundation Report). Participants were asked about demographic status, and other dental-related issues including the 5 questions of the MDAS by trained interviewers in the household.
Ethical Issues
A single application was submitted to NHS Research Ethics System (NRES) covering all aspects of the survey in England, Wales and Northern Ireland. Approval was granted in June 2009. All participants gave written consent.
Statistical Analysis
Data were analysed using SPSS™ version 19 and AMOS™ version 19. Internal consistency and confirmatory factor analysis (CFA) was performed to assess, respectively, the internal consistency (Cronbach's alpha) and the level of fit (Chi-square, CFI, TLI & RMSEA) to a unidimensional model of scaling to a continuous latent construct. CFA was estimated using maximum likelihood and distribution free methods for comparison. All five items were described by a single latent variable. The first item was selected to set the factor coefficient to unity for identification purposes. This selection is generally regarded as arbitrary. Frequencies, means and standard deviations were calculated across the major demographic factors and self-reported visiting. A comparison was made between the original data set reported in 2008 and the current data using fixed factor analysis of variance, with and without adjustment for major demographic variables, namely: age, sex and socio-economic status.
A set of percentiles was prepared across gender and major age groups. A threshold of 19 and above was adopted, as the level for which is it likely that a dental practitioner would consider using additional approaches to manage the patient such as relaxation, systematic desensitisation or pharmacological adjunct. Fixed factor analysis of variance was performed utilizing the continuous scale data to inspect variation of dental anxiety across major demographic, behavioural and socioeconomic status factors. Significance level was set at the conventional 5%, two-tailed.
Point Estimates of Percentile Ranks
The standard method of obtaining percentile ranks was used. That is,
where m is the number of members of the normative sample obtaining a score lower than the score of interest, k is the number obtaining the score of interest, and N is the overall normative sample size.
Interval Estimates of Percentile Ranks
As noted, a further aim of the present study was to accompany the point estimates of the percentile ranks corresponding to raw score with interval estimates of these quantities. A percentile rank is simply a proportion multiplied by 100 thus methods of obtaining an interval estimate of a proportion (such as classical methods based on the binomial distribution) can be used to obtain interval estimates of a percentile rank. However, for the present problem there is a complication. Although anxiety scores are discrete (i.e., integer-valued), the underlying dimension they index are generally taken to be continuous, real-valued quantities. Thus, a raw score of, say, 7 is regarded as a point estimate of a real-valued score which could lie anywhere in the interval 6.5 to 7.4999 (plus an infinite number of additional 9s after the 4 decimal place). Put another way, in principle we could distinguish among individuals obtaining the same raw score were we to introduce tie-breaking items. This assumption of a continuous underlying score is ubiquitous in psychological measurement and motivates the standard definition of a percentile rank (formula 1).
Normative data for scales such as the MDAS will always contain a sizeable number of tied scores; that is, a large number of people in the normative sample will obtain the same raw test score. Indeed, if a normative sample is large and the data are skewed (as would usually be the case for anxiety scales as the majority of the general population are not clinically anxious), then there could literally be hundreds of such ties for a given raw score. The present problem therefore differs from those dealt with by standard binomial sampling in which there can be no possibility of multiple ties.
Crawford, Garthwaite and Slick have recently developed Bayesian and classical methods that incorporate the additional uncertainty arising from tied scores. Crawford et al. and Crawford et al. have used these methods to provide interval estimates for self-report mood scales, such as the HADS, DASS, and PANAS; the methods have also been used to provide interval estimates for a variety of neuropsychological test scores. In the present study we apply the Bayesian method to MDAS scores.
To illustrate the issue the methods address: suppose that in a normative sample of 100 people, 89 obtained lower scores than a case and 2 obtained the same score as the case. Then the point estimate of the percentile rank for the case's score (using formula 1) is 90 and applying Crawford et al's. Bayesian method, the interval estimate is from 82.15 to 95.27. Suppose, however, that 85 obtained lower scores and 10 obtained the same score. The point estimate of the percentile rank is the same as in the foregoing example (90) but the interval estimate is from 79.79 to 97.10; the latter interval is wider because of the increased uncertainty introduced by the larger number of ties (10 versus 2). The technical details of these methods are not set out here: see Crawford et al. for their derivation, and for an additional mathematical treatment and evaluation, see Garthwaite and Crawford.
One-sided versus Two-sided Intervals
In practice there will be occasions in which a one-sided interval may be preferred over a two-sided interval. For example, a clinician may be interested in whether a patient's score is less extreme than is indicated by the point estimate but not particularly interested in whether the score is even more extreme, or vice-versa. The methods developed by Crawford et al. are easily adapted to provide a one-sided limit. However, without prior knowledge of which limit is of interest (the situation here, as the aim is to provide intervals for use by others) it is more convenient to generate 100(1-[α/2]) two-sided intervals which then provide 100(1-α) one-sided lower and upper limits. For example, if a 95% lower limit on the percentile rank is required then a 90% two-sided interval is generated: The user then simply disregards the upper limit of the two-sided interval and treats the lower limit as the desired one-sided 95% limit.
Computer Program for Obtaining Point and Interval Estimates of Percentile Ranks for Raw Scores on the MDAS
The point and interval estimates of percentile ranks for MDAS scores can be obtained using the tabled values provided in the present paper. However, we considered that some health professionals might find it more convenient if theses norms were also available via a computer program.
Source...