Parental Functional Health Literacy Relates to Skip Pattern Questionnaire Error and to Child Oral Health

Abstract

The study's purpose was to determine if parental dental functional-health-literacy (FHL) related to child oral health; a secondary aim was to assess if errors in completing the questionnaire related to FHL and child oral health. Parents of pediatric clinic children (N=101) completed questionnaires; dental caries indices were recorded. Higher FHL was negatively correlated with worse child caries (r=-0.23), but not subjective oral health. Mean FHL seemed to differ by skip pattern (p=0.087), indicating it may be a potential FHL proxy.

Background

According to the 2003 National Assessment of Adult Literacy, 53 percent of adults have intermediate health literacy, while 22 percent have basic and 14 percent have below basic health literacy levels. Thus, 36 percent of the U.S. adult population possesses only enough skills to perform the most simple, concrete everyday literacy activities (e.g. searching text and decoding and recognizing basic words).¹ As reported by the surgeon general's first report on oral health in 2000, families who lack basic health literacy skills have an increased prevalence of dental caries, most of which is untreated.² Oral health research in health literacy has increased in recent years, often following the lead of medicine.³ In dentistry, verbal directions regarding maintenance of oral health and written educational brochures are often provided. Low FHL levels mean that patients are less likely to understand verbal directions or not even read the brochures, resulting in lower oral health knowledge and health status.⁴ In addition, those groups with less education or lower incomes lack the literacy skills to seek out information about preventive health services or treatment options.⁵ Decreased FHL has been shown to be associated with a wide variety of poor health outcomes, including disproportionately high rates of disease and mortality.¹ In a recent preliminary report, FHL has been shown to relate to self-rated oral health (e.g. 5-point ordinal scale from poor to excellent) and dental utilization (i.e annual dental visit).²

The National Institute of Dental and Craniofacial Research Working Group on Functional Health Literacy defined dental FHL as “the degree to which individuals have the capacity to obtain, process, and understand basic oral health information and services needed to make appropriate health decisions.”⁶ However, many dental instructions and brochures are presented at a reading level beyond the 12^th grade, containing dental terminology that is unfamiliar and difficult for patients to understand.^5,7 The short form Test of Functional Health Literacy in Adults (STOFHLA) significantly related to children's caries index.² The 30-item REALD-30 is a validated and reliable instrument that is significantly related to perceived dental health status and to oral health-related quality of life, an abbreviated version of the REALD-99.^6,8

Using the 1993-4 California Oral Health Needs Assessment of Children (COHNAC), Gansky and Wilson determined that children of parents who did not correctly follow the self-administered questionnaire skip patterns (i.e. instructions for completing later questionnaire items conditional on the responses to earlier items) had significantly worse oral health.⁵ Errors in the self-administered parent questionnaire of the 2004-5 COHNAC (California Smile Survey), which differed from the one in 1993-4, also were associated with worse oral health status. For example, incorrectly following the skip pattern (more specific details are given in the Methods section below) was associated with lack of dental sealants.⁹ These questionnaire inconsistencies (errors) might empirically identify parents with low FHL who may have other problems such as completing insurance forms or understanding health care provider instructions for themselves or their children. However, since a validated reliable measure of FHL was not part of the COHNACs, questionnaire inconsistencies have not been proven to be a FHL proxy; a proxy must relate to both the standard predictor (here a measure of FHL) as well as an accepted measure of oral health.¹⁰

Hypotheses

(1) Decreased parental FHL would relate to decreased child's oral health status (decreased Michigan Oral Health-Related Quality of Life (MOHRQOL) and increased caries index); (2) parents with skip pattern inconsistency (error) would have decreased REALD-30 FHL score; (3) mean scores would differ between the modified color-coded MOHRQOL instrument with a sample item and the original reliable, validated version instrument; and (4) parental FHL would moderate the difference in scores between versions of MOHRQOL.

Methods

Methods/Design

At the suggestion of the director of the pediatric dentistry clinic, the lead investigator (Garrett) provided an information session to faculty, staff, residents, and students of the pediatric clinic. This facilitated participant recruitment and enrollment. The pediatric dentistry clinic sees approximately 40 patients per day. This planned study sample size of 100 over a period of 2.5 months was logistically feasible.

Parents of patients in the UCSF Pediatric Dental clinic were told about the study and invited to participate. Inclusion criteria consisted of the ability to read and write English, and the ability to give consent for one's self and child. Exclusion criteria included having participated already with another child, having hearing, visual, or cognitive impairment, and parents/guardians younger than the age of 18. Thus, this was a convenience sample of consenting parents and their children in the UCSF Pediatric Dental clinic August-September, 2007.

After the study was explained and participants consented, parents completed a 20-item questionnaire including their perception of their child's oral health status, the Michigan Oral Health-Related Quality-of-Life (MOHRQOL) scale in regard to their child's health, and the California Smile Survey. Parents then were asked to read the REALD-30 to the investigator.⁷ Each parent received a small incentive ($5 coffee retailer gift card) for participation. Clinical data were abstracted by ten attending pediatric dentistry residents onto a study dental chart form that had the participants’ study identification number as the only unique identifier. This allowed determining the number of decayed, missing and filled primary and permanent teeth (dmft+DMFT) and tooth surfaces (dmfs+DMFS) indices.

Instruments

MOHRQOL scale

This is a 10-item validated reliable measure of parent report of child's oral health.⁷ Half of the parent participants were randomly selected to receive the standard black and white MOHRQOL, while the other half were randomly selected to receive a revised MOHRQOL in a red and green shaded color format with an example question at the top of the page. The random ordering was determined using randomizer.org and questionnaire packets were compiled in advance with three digit study ID numbers on the outside of packets. The one MOHRQOL item phrased in the negative was reverse scored and all 10 items were summed to determine the score, ranging from 10 to 50 with larger scores indicating worse QOL. Measuring MOHRQOL allowed comparing the inter-relationships among FHL, MOHRQOL and clinically determined dmft+DMFT and dmfs+DMFS.

Parent-reported child oral health status

The valid and reliable single ordinal scaled item of parent-reported child oral health status (“How would you describe the condition of your child's teeth?” with responses of excellent, very good, good, fair, or poor) was assessed.

California Smile Survey

The six-item survey items about child dental care, containing a skip pattern associated with poor dental care postulated as a low FHL indicator, was administered.

Skip Pattern Inconsistency (Error)

In the one-page, six-question, self-administered written California Smile Survey given to parents/guardians, Question 3 was “During the past year, was there a time when you wanted dental care for your child but could not get it?” with response options “Yes (go to Question 4),”No (go to Question 5)” and “Don't know (go to Question 5).” Respondents who answered “No” or “Don't know” for Question 3, but answered Question 4 instead of skipping to Question 5 were classified as having “answered anyway,” while those who answered “Yes” to Question 3, did not answer Question 4 but did answer Question 5 were classified as having “omitted” the next question to be answered. Both situations were classified as a skip pattern inconsistency (error).

REALD-30

This contains 30 commonly used dental words arranged in order of increasing difficulty and tests patient's ability to correctly pronounce each word.⁴ It is short and easy to administer, taking less than five minutes. The interviewer (Garrett) consulted the American Heritage Steadman Medical Dictionary to establish the correct pronunciation of each dental term in the REALD-30. The REALD-30 is scored on a scale of 0-30, tabulating the number of correctly pronounced terms.

Confidentiality and data security were preserved in accordance with UCSF Investigational Review Board, the Committee for Human Research, procedures.

Sample Size

The investigators planned the study to have a sample size of 100 which was calculated using an alpha=0.05, two-sided Fisher's z test of the null hypothesis that the Pearson correlation coefficient is rho=0.09 and 90 percent power to detect an observed rho of 0.40. If the sample size dropped to 75, the power was estimated at 80 percent. Adequate power was estimated for other relevant projected values of rho.

Data Analysis

Summary statistics (mean, median, standard deviation, standard error) were compiled for REALD-30, MOHRQOL, dmft+DMFT and dmfs+DMFS index. Counts and frequencies were tabulated for dichotomous skip pattern inconsistency and for ordinal categories of parent report of child health. Pearson correlation coefficients (r) with 95 percent Fisher confidence intervals were estimated to assess the relationships of REALD-30 with MOHRQOL and caries indices (dmft+DMFT and dmfs+DMFS). Spearman nonparametric correlation coefficients (r_s) with 95 percent Fisher confidence intervals were estimated to assess the relationships among skip pattern inconsistency, REALD-30 and parent-rated ordinal child oral health, as well as skip pattern inconsistency with MOHRQOL and caries indices (dmft+DMFT and dmfs+DMFS).

Since variables might not have been normally distributed, appropriate normalizing transformations were utilized (natural logarithm of one plus the dmf+DMF index and the Box-Cox transformation for REALD-30). To assess the validity of the skip pattern inconsistency as a proxy for FHL, a two sample (unpaired) t-test was used to compare the validated, reliable REALD-30 score between those parents with a questionnaire skip pattern inconsistency (error) and those with a consistent questionnaire.

Results

The study consisted of 101 children, mostly white, black or multiracial including one of those races (76 percent); 17 percent were Hispanic. The children were a mean age of 6.6 years (standard deviation of 3.5 and median of 5.5) with slightly more males (53 percent) than females (47 percent). Of these children, 75 percent were on MediCal (California's Medicaid) or Healthy Families alone or with another program; 62 percent were in the free or reduced-cost lunch program, and 84 percent reportedly visited the dentist in the prior year, but 10 percent had never been to a dentist.

The overall REALD-30 score of the 100 parents who were tested had a mean of 21.8 with a standard deviation (SD) of 5.0 and a range of 3 to 30. Eighteen (18 percent) of the 100 parents improperly (inconsistently) followed the questionnaire skip pattern. The overall mean MOHRQOL score was 19.7 with a SD of 7.7. For the caries indices, pediatric dentist residents did not transfer clinical chart data for two children, so the sample size with exam data was 99. The overall mean dmft+DMFT index was 6.3 with a SD of 4.6 and a range of 0 to 18. The overall mean dmfs+DMFS index was 14.3 with a SD of 14.2 and a range of 0 to 68.

Results testing Hypothesis 1 that decreased parental FHL would relate to decreased child's oral health status (decreased Michigan Oral Health-Related Quality of Life (MOHRQOL) and increased caries index) are shown in Tables 1 and 2. Results indicated that the better the parent's dental health literacy, the lower the child's clinically determined (objective) dental disease. Table 1 shows that parental dental functional health literacy (REALD-30) significantly but modestly related inversely to dental examination determined caries experience, but did not relate to subjective oral health measures. Table 2 shows that there were no differences in children's oral health status between parents who correctly followed the skip pattern and those who did not. Additionally, both groups of parents reported the median (and mode) ordinal oral health level of “good” for their children (1 df chi-square test, P=0.874).

Table 1.

Relationship of parental functional health literary with child oral health measures; N=99, Aug-Sept 2007, UCSF Pediatric Dentistry Clinic; Pearson product correlation, [Fisher 95% confidence interval], P-value.

	Children's Tooth Surfaces log(dmfs+DMFS+1)	Children's Teeth log(dmft+DMFT+1)	Parent-reported children's QOL log(MOHRQOL)	Parent-reported children's oral health (ordinal ranking)
Parental functional health literacy (REALD-30)2	-0.23	-0.24	-0.04	-0.11
	[-0.41, -0.04]	[-0.41, -0.04]	[-0.23, 0.16]	[-0.30,-0.09]
	P=0.020	P=0.018	P=0.727	P=0.278

REALD-30 = Rapid Examination of Adult Literacy in Dentistry

dmfs+DMFS = number of primary and adult decayed, missing or filled tooth surfaces

dmft+DMFT = number of primary and adult decayed, missing or filled teeth

MOHRQOL = Michigan Oral Health-Related Quality of Life scale

ordinal ranking = excellent, very good, good, fair, poor

Table 2.

Difference between skip pattern inconsistency (error) and consistency in child oral health measures; N=100, Aug-Sept 2007, UCSF Pediatric Dentistry Clinic; Mean ± Standard Error, P-value.

Skip Pattern	Children's Tooth Surfaces dmfs+DMFS	Children's Teeth dmft+DMFT	Parent-Reported Children's Quality of Life MOHRQOL
Inconsistent (Error) (n=18)	14.2 +/- 3.9	6.3 +/-1.2	20.8 +/- 2.2
Consistent (No Error) (n=82)†	14.4 +/- 1.5	6.3 +/- 0.5	19.4 +/- 0.8
Unpaired t-test P-value*	0.887	0.947	0.682

dmfs+DMFS = number of primary and adult decayed, missing or filled tooth surfaces

dmft+DMFT = number of primary and adult decayed, missing or filled teeth

MOHRQOL = Michigan Oral Health-Related Quality of Life scale

^†n=80 for dental examinations

^*unpaired t-tests used log(dmfs+DMFS+1), log(dmft+DMFT+1), and log(MOHRQOL)

Results from Hypothesis 2, that parents with skip pattern inconsistency (error) would have decreased REALD-30 FHL score, have been shown in Table 3. Mean REALD-30 was suggestive of differing by questionnaire skip pattern as those 18 parents with an inconsistency (error) had a mean REALD-30 score of 19.9 compared to those 82 with a consistent questionnaire who had a mean REALD-30 score of 22.2 (unpaired t-test of REALD-30², P=0.087).

Table 3.

Difference between skip pattern inconsistency (error) and consistency in parent functional health literacy; N=100, Aug-Sept 2007, UCSF Pediatric Dentistry Clinic; Mean ± Standard Error, P-value.

Skip Pattern	Parent Functional Health Literacy (REALD-30)
Inconsistent (Error) (n=18)	19.9 +/-1.4
Consistent (No Error) (n=82)	22.3 +/- 0.5
Unpaired t-test P-value*	0.087

REALD-30 = Rapid Examination of Adult Literacy in Dentistry

^*unpaired t-tests used (REALD)²

Hypothesis 3 was that mean scores would differ between the modified color-coded MOHRQOL instrument with a sample item and the original reliable, validated version instrument. Mean MOHRQOL scores did not differ between the two versions (19.3 and SE=1.2 vs 19.9 and SE=1.0; unpaired t-test of log(MOHRQOL) P=0.490). Thus, the revised form did not appear to provide any changes in scores.

Hypothesis 4 was that FHL would moderate the difference between MOHRQOL versions — this means, one would see if the difference in MOHRQOL scores between versions only occurred in those with low FHL (because only people with low FHL would have benefited from the new color-coded version with an example). Since there was no difference, the next steps in an effect modification (moderation) analysis were not performed. If there had been a significant difference, a statistical interaction effect to determine moderation in the linear regression model would have been used.

In addition, subjective oral health measures were significantly correlated with objective ones: parent ordinal rating with dental examiner caries indices (log(dmfs+DMFS+1), r_s=0.46, P<0.001; log(dmft+DMFT+1), r_s =0.43, P<0.001) and log(MOHRQOL) with caries indices (log(dmfs+DMFS+1), r=0.31, P=0.002; log(dmft+DMFT+1), r=0.29, P=0.004)).

Participating in the free/reduced-cost lunch program was not significantly related to dental caries indices (dmfs+DMFTS or dmft+DMFT). Children having government insurance (e.g. MediCal) had significantly higher dental disease (dmfs+DMFS and dmft+DMFT) than those without government insurance, but REALD-30 was still significantly negatively related to dental disease(r=-0.22, P<0.032) even after adjusting for government insurance status.

Transformations (squaring REALD-30 and logging dmfs+DMFS, dmft+DMFT and MOHRQOL) were needed to make the distributions more Normal (bell-shaped). Squaring REALD-30 was found to be appropriate by the Box-Cox transformation determination method. Logarithmic transformations of dmf+DMF indices (plus an offset of one) were also found to be appropriate in other studies.

Discussion

Several implications have resulted from this study, including a new approach for evaluating FHL in other studies. The results of from the California Smile Survey questionnaire items were consistent with the hypothesis that persons with skip pattern inconsistencies have a lower FHL. The child's parent/guardian assessed the child's oral health and included questions with specific instructions to skip to another question, depending on the answer chosen. In analyses of the association between parent skip pattern inconsistency and preschooler dental health in the 1993-4 COHNAC, the questionnaire differed with multiple skip patterns. For example, Question 5 asked, “Does this child take PRESCRIPTION vitamins with FLUORIDE or another kind of FLUORIDE drops or tablets, at school or at home?” with the options of “Yes,”No” or “Don't Know” and “ (IF NO, SKIP TO QUESTION 7).” Question 6 began with “IF YES” and asked how often the child used fluoride. Parents who responded “No” to Question 5, should have skipped to Question 7. Those who answered “No” to Question 5 but answered Question 6 were classified as incorrectly following the skip pattern since they answered anyway when they should have skipped. Those who answered “Yes” to Question 5 and skipped Question 6 to Question 7 were classified as incorrectly following the skip pattern since they improperly skipped or omitted.

Based on comparison to the REALD-30 assessments, improper questionnaire skip pattern is a potential tool in identifying low FHL.

As with all studies, this study has several limitations; first and foremost non-English speakers and readers were excluded. The REALD-30 has only been tested and shown to be reliable and valid in English speaking populations and could not simply be translated into another language. Additionally, the REALD-30 only tests word recognition and pronunciation, not actual comprehension. Another instrument developed by the same group that developed the REALD-30, called the Test of Functional Health Literacy in Dentistry (TOFHLiD), assesses reading and numeracy abilities, but takes longer to administer. It would be possible to develop a reliable, valid version of the TOFHLiD for Spanish-speaking populations. To reduce burden on participants, only the REALD-30 was used. Additionally, the REALD-99 has been shown to correlate highly with the TOFHLiD (r=0.82).^8,11 However, in that report, and in a subsequent systematic review, TOFHLiD was not related to parent or child's oral health status,¹² while this current study showed REALD-30 was significantly related to objectively determined oral health.

This study used a cross-sectional study design assessing parental dental health literacy and child oral health status at the same time. A longitudinal design would be needed to determine if parental health literacy relates to their child's future dental health. Due to the design of the study, reverse causation was possible: some participants may have recognized certain dental words because of prior treatment for their child in the study; only a longitudinal study can give insight on the temporal relationship. Nevertheless, there is a moderate correlation of the skip pattern inconsistencies to parents with low FHL, who were more likely to have children with poor oral health status.

Children in his study had a wide age range (3-18 years), a narrow socioeconomic status range, and access to care. However, not all children had caries (16 percent were caries-free, i.e., dmf+DMF=0). As a sensitivity analysis, the authors analyzed the subset of data from the 57 children aged 3-6, finding very similar results. With the total sample size of 101 and about 18 percent of parents improperly following the skip pattern, power was limited. Still, this illustrates the potential of an empirical FHL measure such as questionnaire error. Thus, the strengths of the study outweigh its limitations to contribute new important information.

As some parents with low FHL may experience a degree of shame in conjunction with being tested for low FHL, it is important for the clinic staff to ensure that the respondent feels as comfortable as possible when administering these assessments. In similar studies, participants whose FHL was assessed were later surveyed on their comfort during the assessment. Protecting patients from feeling shame or embarrassment involves key factors including the minimal use of words such as “help” or “how often” that may imply they had regular difficultly with literacy. The participants in this study were all English speakers which may reduce generalizability because potential participants may have been too embarrassed to participate in the assessment.¹³ Causing the participant to feel uncomfortable during a dental visit may stimulate fear and the patient feeling stigmatized once the assessment is complete.¹⁴ A skip pattern like the one in the California Smile Survey offers benefits in the assessment of low FHL since it virtually eliminates patient discomfort of FHL-focused tests, making it possible for clinicians to identify patients who need more targeted communication.

Because this survey is limited to English-speaking populations, generalizability is impacted. The surgeon general reported that Mexican-American children living below the poverty level have the highest incidence of untreated, decayed teeth. Many studies in dental FHL have been plagued by this language barrier. The REALD-30 is limited to English speakers and this tool cannot assess persons whose native language is not English. For example, pronunciation rules in Spanish are firm and any person fluent in the language could pronounce the word without knowing its actual meaning; thus the results would not be accurate. The TOFHLiD assessment tool translated to other languages could be used to increase generalizability.¹⁵

Due to low FHL, some parents may not fully understand the instructions for preventative care nor know the extent of their child's oral health. Once the patient has been identified as having potentially low FHL, additional teaching methods can be used so the parent fully understands clinician instructions. Until an assessment for all patients is established clinicians can utilize the model of universal precautions in regard to FHL. A universal precaution is the notion to treat all patients with the same caution.¹⁶ Each patient then would be given information in the same clear manner, regardless of their FHL. Techniques include having patients demonstrate and repeat instructions back to the clinician to ensure understanding.

Studies that incorporate using techniques like the “teach back” method could give further insight on appropriate responses to low FHL.¹ As concluded in the current study, parents with low FHL are more likely to have children with a higher prevalence of dental caries. It is therefore important to present information to patients in a well-designed manner that can appeal to patients of all backgrounds. Similar studies used methods that tested the readability of customary dental educational materials and found that most were rated at or above a ninth-grade reading level. The more common and widespread reading materials, such as newspapers, are written at a sixth-grade reading level. ¹⁷

There is a link between parental FHL with the overall oral health of the child. The noteworthy method of using complex instructions to evaluate patient FHL is a potential breakthrough in gathering more data regarding FHL from patients in which their primary language is not English. From further investigation into this new assessment, it may also be possible to reduce the shame associated with these assessments and to employ universal precautions with every patient regarding their FHL.

Conclusions

Results indicated that children of parents with lower functional dental health literacy had significantly lower objective oral health status, albeit to a modest extent. Dental functional health literacy was not correlated with subjective measures of oral health; however, subjective oral health measures were correlated with objective ones, indicating that parents have an accurate grasp of their child's oral health status. Additionally, FHL, although not statistically significant, as evidenced by the two sample t-test, was suggestive of being lower in those with incorrect skip pattern behavior. Finally, the mean MOHRQOL scores did not differ between the original version and the modified color-coded version.

Parental FHL appears moderately negatively related to children's oral health status. Moreover, skip pattern inconsistency appears as a potential proxy or substitute for longer reliable, validated FHL scales (e.g. REALD-30), but further assessment is needed. Skip pattern inconsistency (error) could be particularly useful in self-administered surveys conducted before FHL scales were developed and widely utilized. If further studies demonstrate a relationship of parent skip pattern inconsistency (error) with child oral health, that would help establish skip pattern inconsistency as a proxy which practitioners and researchers could use as a screen to identify patients with whom to spend additional time reviewing health care instructions.

A health practitioner can have the best training, manual dexterity, and equipment but it is nearly futile if patients do not understand health care instructions for themselves or their children. In the future, a form in a new patient packet could assess dental health literacy or current new patient forms could be scanned for inconsistencies. Patient's oral health status is dependent upon his or her ability to understand and follow directions, so providers should be more careful to explain directions to those with obvious errors in their forms. Furthermore, medical brochures are presented at a third-grade reading level. In the future, dental brochures should be rewritten and this study could be repeated to see if the same results occur. Future longitudinal studies of approaches to overcome low functional dental health literacy should be considered.