Welfare Dental Intervention Improves Employment and Quality of Life

Abstract

Chronic, untreated oral disease adversely affects one's systemic health, quality of life, and economic productivity. This study evaluated the effect of rehabilitative dental treatment on the oral-health-related quality of life and employment of welfare recipients. Three hundred and seventy-seven participants in a novel welfare dental program received oral examinations, questionnaires, and rehabilitative dental treatment. Seventy-nine percent of participants exhibited improvement in their oral-health-related quality-of-life scores following dental treatment. Improved OHIP-14 change scores were associated with being Caucasian or African-American, initial poor general health, severity of treatment urgency, worse baseline oral-health-related quality-of-life scores, subsequent patient satisfaction with the Dental Program, and resolution of their chief complaint (all p < 0.04). Those who completed their dental treatment were twice as likely to achieve a favorable/neutral employment outcome (OR = 2.01, 95%CI = 1.12, 3.62). Thus, oral health improved the quality of life and employment outcome for this welfare population.

INTRODUCTION

The Surgeon General's Report on Oral Health revealed oral health disparities in some population groups, as classified by age, sex, income, and race/ethnicity (US Department of Health and Human Services, 2000). However, the oral health and oral-health-related quality of life (OHRQoL) of those marginalized through extended unemployment are unknown. Compromised nutrition, hygiene, and access to care in this underserved population likely result in extensive dental caries, tooth loss, and periodontal diseases. Chronic, untreated oral disease adversely affects one's systemic health, quality of life, and economic productivity (Gift et al., 1992; Hollister and Weintraub, 1993; US Department of Health and Human Services, 2000). Facial attractiveness has been found to affect social attitudes and actions, and is important in employment situations (Oosterhaven et al., 1989; Eli et al., 2001).

The San Francisco Department of Human Services (SFDHS) offers the Personal Assisted Employment Services (PAES) program as a Temporary Assistance to Needy Families benefit directed toward employable, single, indigent adults. As compared with the general population of San Francisco, PAES recipients are almost six times as likely to be African-American (46% vs. 8%), and 19 times as likely to be homeless (38% vs. 1–2%) (San Francisco Department of Human Services, 2003). In contrast to Caucasian adults, African-American adults are twice as likely to have untreated dental caries (48% vs. 24%), more likely to have missing teeth (78% vs. 69%), and less likely to have visited a dentist in the preceding year (54% vs. 66%) (Dental, Oral, and Craniofacial Data Resource Center). Homeless adults are at even higher risk for poor oral health, since 91% have untreated decay, 89% have missing teeth, and only 27% have had a dental visit in the preceding year (Gelberg et al., 1988; Kaste and Bolden, 1995).

The PAES Dental Program began in 1999 as a collaboration between the San Francisco Departments of Human Services and Public Health, and is the only program of its kind in the United States. The goal of the Dental Program is to eliminate severe dental problems that pose a barrier to employment and self-sufficiency. The Program includes an oral health needs assessment, measurement of the OHRQoL, treatment planning, rehabilitative dental treatment, and program evaluation.

Based on data from a cohort of participants in the PAES Dental Program, this study evaluated the intervention effects of rehabilitative dental treatment on the OHRQoL and employment of welfare recipients.

MATERIALS & METHODS

This research project was reviewed and approved by the University of California, San Francisco, and the University of California, Berkeley, Institutional Review Boards. All participants provided signed informed consent prior to entering the study.

Participants

Eligibility criteria for the Dental Program included those welfare recipients who had been working cooperatively with their welfare social worker for at least 3 mos and had either self-identified or been identified by their social worker as needing extraordinary dental services. Dental Program participants could not be eligible for California's Medicaid or for other dental insurance. The minimum age was 21 yrs old, and an interpreter was provided for those participants who did not speak English.

Measures

One investigator (SH), trained and calibrated for the clinical methods and diagnostic criteria used by the National Institute of Dental and Craniofacial Research (US Department of Health and Human Services, 1991), conducted all interviews, clinical examinations, and treatment planning. Demographic information was obtained for age, sex, race/ethnicity, education, employment, and housing. The medical history was limited to (1) an assessment of whether antibiotic prophylaxis would be required for dental treatment and (2) 3 general questions about overall health. The clinical examinations were conducted in the welfare building with portable dental equipment. Universal precautions were used for infection control, and no radiographs were exposed. Clinical measures included the Community Periodontal Index (World Health Organization, 1997), the Decayed Missing Filled Index (Klein et al., 1938), and the American Dental Association's Classification of Treatment Urgency (Council on Dental Health and Bureau of Dental Health Education, 1956). Treatment plans were rehabilitative (e.g., scaling and root planing, restorations, extractions, dentures) rather than involving full-mouth reconstruction. Four private-practice dentists, two university clinics, and one city clinic provided the treatment prescribed by the treatment plan. After completion of the dental treatment, participants were asked to return to the PAES Dental Program for an evaluation of their satisfaction with the Program and a re-evaluation of their OHRQoL. The satisfaction survey included process (e.g., appointment scheduling, telephone interaction) and outcome (e.g., Dental Program satisfaction, chief complaint resolution) measures.

Oral-health-related Quality of Life

The Oral Health Impact Profile (OHIP) was used to measure the OHRQoL (Slade and Spencer, 1994). The OHIP measures the social impact of oral health according to 7 hierarchical subscales of oral health outcomes. A five-point Likert scale captured the responses, in categories of never, hardly ever, occasionally, fairly often, and very often, and with scores of 0 to 4, respectively. The OHIP has been found to have excellent internal reliability (α = 0.90) and validity (p < 0.001) (Locker and Slade, 1993; Locker and Jokovic, 1996; Slade et al., 1996; Allison et al., 1999; Locker et al., 2001). It has been in worldwide use since 1994, and translated into 8 languages (Allison et al., 1999; Wong et al., 2002; Ekanayake and Perera, 2003; Kushnir et al., 2004). The OHIP's responsiveness to detect change in the quality of life over time, or following an intervention, was determined to be sensitive to both improvement and deterioration in scores (Slade, 1998; Awad et al., 2000; Allen et al., 2001). Although item weights were developed for the OHIP, they were not used in the calculations for this study, since Allen et al. (2001) found that the poorest sensitivity to change in quality of life was associated with the weight-standardized scores. A short-form version was subsequently developed (OHIP-14), consisting of 14 rather than 49 questions, which also possessed high internal reliability (α = 0.88) and comparable construct validity (Slade, 1997). Due to the constricted appointment scheduling of the Dental Program, the short-form OHIP-14 was used in this study.

Employment Outcome

Employment outcomes were recorded by the SFDHS as a complex series of codes. Over 50 different codes were used to characterize the disposition of the PAES participants. Favorable outcomes were assigned to participants who gained employment or transferred to other benefit programs, such as Social Security or Veteran's Affairs. Neutral outcomes were given to participants who voluntarily left PAES, continued to remain compliant in the PAES program, or were ineligible to receive benefits due to institutionalization, death, or moving out of the county. Unfavorable outcomes were assigned to participants who became ineligible to receive benefits due to non-compliance or fraud. Differentiating between favorable and neutral outcomes was somewhat ambiguous. Classifications such as "client's request", "program change request", "decline services", and "other" could apply to either favorable or neutral outcomes. Therefore, to minimize misclassification bias, we grouped favorable and neutral outcomes together.

Data Analysis

We calculated change scores by subtracting the OHIP-14 scores at baseline from those at follow-up. We calculated the effect size by dividing the change scores by the OHIP-14 baseline standard deviation. Cohen (1988) has defined an effect size of 0.2 as small, 0.4 as moderate, and 0.8 as large.

The independent variables for the regression model explaining the OHIP-14 change scores fell into 5 categories: demographics, clinical measures, treatment needs, baseline OHIP-14 scores, and follow-up patient satisfaction. Bivariate associations were assessed with one-way analysis of variance and linear regression. Those independent variables that were found to be statistically significant in bivariate analysis (α = 0.05) were added to a stepwise multivariate regression model. We assessed interaction by grouping together a cross-product term with its precedent terms, and decided to enter it into the model on the basis of the significance of the group's joint F test (SAS Institute Inc., 2001). Subsequently, we assessed confounding factors by comparing the regression coefficients with and those without the addition of a potential confounder to the model. Covariates with an α ≤ 0.20 were considered for control as potential confounders, and confounding was deemed present if the adjusted coefficients differed by more than 10% of their crude value (Kleinbaum et al., 1998).

We performed the chi-square test and logistic regression using demographic and clinical variables to determine whether any significant differences existed between those who completed their rehabilitative dental treatment and those who did not, as well as for those who were lost to follow-up. Chi-square analysis was also done for rehabilitative dental treatment and employment outcome. All data entry and analyses were conducted with the JMP Version 4 statistical analysis software from SAS Institute Inc. (Cary, NC, USA).

RESULTS

During the 18-month enrollment period (October, 1999–March, 2001), there were 2930 PAES welfare recipients, of whom 379 were referred or sought dental services. Of these, 377 (99.5%) agreed to participate in this study. Two hundred and sixty-five (70%) study participants completed their dental treatment, while 54 (14%) were discontinued from either PAES or the Dental Program prior to completing their treatment. Fifty-five participants (15%) failed to keep their first and all subsequent dental treatment appointments, and three (1%) died due to unrelated causes prior to completion of their treatment. Of the 265 participants who completed their dental treatment, 173 (65%) returned their oral-health-related quality-of-life questionnaires, and 92 (35%) were lost to follow-up. Employment status was ascertained for the entire cohort (n = 377), while changes in the OHRQoL were measured for those 173 participants who returned the questionnaires following completed dental treatment.

Seventy-one percent of the participants were male, ages ranged from 21 to 63 yrs old, 46% were African-American, 22% did not complete high school, 45% lived in a welfare-subsidized hotel room, 11% were homeless, and 26% rated their general health as either fair or poor (Table 1). Thirty-one percent had ≥ 6 mm periodontal pocket depths, 85% were missing one or more teeth, 84% had one or more untreated decayed teeth, and 63% had severe or emergency dental treatment urgency.

Table 1.

Baseline Characteristics of PAES Dental Program Participants, N = 377

Demographic or Clinical Variable	Percent
Sex
Male	71.2
Female	28.8
Age
Range, yrs	21 to 63
Mean yrs (SD)	44.8 (7.9)
Race/Ethnicity
African-American	46.4
Caucasian	32.5
Hispanic	10.9
Asian	5.9
Other	4.3
Education
Less than grade 9	4.5
Less than high school	17.9
High school/GED	30.5
Vocational training	4.3
Some college	34.0
College degree	6.9
Post-college degree	1.9
Type of Dwelling
House/apartment	43.3
Hotel room	45.5
Shelter	5.9
Half-way house	1.3
Car	1.3
Streets	1.1
Other	1.6
Community Periodontal Index: Maximum Sextant Score
0 = healthy	0.0
1 = bleeding	0.0
2 = calculus	15.8
3 = periodontal pocket 4 to 5 mm	53.2
4 = periodontal pocket ≥ 6 mm	31.0
Decayed Missing Filled Index*
Edentulous (missing all teeth)	4.5
Missing any teeth	85.4
Any decayed teeth	84.2
Any filled teeth	75.5
Mean number missing teeth (SD)	7.5 (7.9)
Mean number decayed teeth (SD)	4.5 (4.8)
Mean number filled teeth (SD)	4.8 (4.7)
Treatment Urgency
1 = No obvious problems	0.5
2 = Mild to moderate problems	36.1
3 = Severe problems	60.5
4 = Emergency problems	2.9

^*Third molars and teeth extracted for orthodontic purposes were excluded.

Subgroup analysis of demographic and clinical variables determined whether any significant differences existed between those who completed their treatment and those who did not, as well as for those who were lost to follow-up. Those who completed their dental treatment had more missing teeth (p = 0.0108) and fewer decayed teeth (p = 0.0109) at baseline than those who did not complete their treatment, which may reflect that missing teeth are associated with a higher perceived need for dental treatment. African-Americans were less likely to return their follow-up questionnaires than were Caucasians and members of other races (p = 0.0100). Otherwise, the baseline demographic and clinical profiles showed no significant differences (all p > 0.05) between those who completed their treatment and those who did not, and for those who were lost to follow-up.

Oral-health-related Quality of Life

After receiving rehabilitative dental treatment, 79% of participants exhibited improvement in their OHIP-14 change scores, 18% showed deterioration in their scores, and 3% exhibited no change. Large effect sizes were found for the change scores of the psychological discomfort (1.09), psychological disability (1.00), and handicap (0.74) subscales (Table 2). Moderate effect sizes were seen for the change scores of the physical pain (0.63), social disability (0.62), and physical disability (0.41) subscales. Only functional limitation exhibited a small effect size (0.26). The OHIP-14 total score had a large effect size (0.87) for the change score.

Table 2.

Oral-health-related Quality-of-Life (OHIP-14) Change Scores, N = 173

OHIP-14 Subscales and Questions	Baseline Meana (SD)	Follow-up Meana (SD)	Change Scoreb (SD)	Effect Size
Psychological discomfort	4.5 (2.7)	1.6 (2.3)	−2.9 (3.0)	1.1
Felt self-conscious
Felt tense
Psychological disability	3.8 (2.5)	1.3 (2.1)	−2.5 (3.0)	1.0
Difficulty relaxing
Felt embarrassed
Handicap	2.4 (2.2)	0.8 (1.5)	−1.6 (2.4)	0.7
Felt life less satisfying
Totally unable to function
Physical pain	4.1 (2.5)	2.5 (2.3)	−1.5 (2.9)	0.6
Painful aching in mouth
Uncomfortable to eat
Social disability	2.3 (2.4)	0.8 (1.6)	−1.5 (2.5)	0.6
Irritable with other people
Difficulty doing usual jobs
Physical disability	2.5 (2.4)	1.5 (2.2)	−1.0 (2.7)	0.4
Diet unsatisfactory
Meals interrupted
Functional limitation	2.0 (2.1)	1.5 (1.9)	−0.5 (2.4)	0.3
Trouble pronouncing words
Sense of taste worsened
OHIP-14 Total Score	21.5 (13.4)	10.1 (10.9)	−11.6 (14.3)	0.9

^aMaximum score: 8.0 per subscale, 56.0 for total score.

^bLower score denotes more improvement.

In the regression model, Caucasians and African-Americans, as compared with members of other races, showed greater improvement (p = 0.0066), as did those who initially reported poor general health vs. those who were in good health (p = 0.0375) (Table 3). The greater the severity of treatment urgency (p = 0.0133), and the higher the baseline OHIP-14 score (p < 0.0001), the greater the improvement in the change scores. Patient satisfaction with the PAES Dental Program (p = 0.0010) and the resolution of their chief complaint (p = 0.0028) resulted in improved change scores. These variables explained 62% of the variance associated with the OHIP-14 change scores. No significant interaction or confounding was detected for the model.

Table 3.

Multivariate Regression Model for OHIP-14 Change Score R² = 0.616, N = 173

Variable	Estimate	Standard Error	p-value
Race: Caucasian + African-Americans vs. Others	− 2.5	0.9	0.0066
Overall Health: Poor vs. Excellent to Fair	− 5.2	2.5	0.0375
Treatment Urgency: Severe vs. Moderate	− 3.4	1.4	0.0133
Satisfaction w/PAES Dental: Satisfied vs. Not + Sometimes	− 15.1	4.5	0.0010
Satisfaction w/Chief Complaint: Satisfied vs. Not + Sometimes	− 8.4	2.8	0.0028
Baseline OHIP-14 Score: Per 10 Points	− 8.0	0.6	< 0.0001

Employment Outcome

There was a significant difference between the level of dental treatment received and the employment outcome (p = 0.0412) (Table 4). Participants who completed their dental treatment had twice the proportion of favorable/neutral outcomes (67.5%) to unfavorable outcomes (32.4%), and were twice as likely to obtain favorable/neutral employment outcomes as were those who did not start their treatment (OR = 2.01, 95%CI = 1.12, 3.62). Those who started treatment but did not finish it were slightly more likely to have favorable/neutral outcomes (57.4%) than unfavorable outcomes (42.6%). Participants who never started their dental treatment were equally as likely to have a favorable/neutral outcome (50.9%) as an unfavorable outcome (49.1%).

Table 4.

PAES Dental Program Participants Employment Outcome, N = 377

Dental Treatment Status	Favorable + Neutral Outcome	Unfavorable Outcome	Odds Ratio (95% CI)
Complete	67.5%	32.4%	2.0 (1.1, 3.6)
Incomplete	57.4%	42.6%	1.3 (0.6, 2.8)
Never started	50.9%	49.1%	referent

X² p-value = 0.0412.

DISCUSSION

Floor effects, or having a total score of zero at baseline, would result in the OHIP-14 being unable to detect improvements following an intervention. Similarly, ceiling effects, or scoring the maximum possible at baseline, would render the OHIP-14 unable to detect deterioration in the OHRQoL. There were minimal floor effects and no ceiling effects in this study, since only 3.6% of the cohort had a total score of zero on their baseline OHIP-14, and none scored the maximum of 56. Therefore, the OHIP-14 was not impaired in detecting change in the OHRQoL following the intervention.

Since 57% of the cohort were either homeless or provisionally housed, the loss to follow-up of participants who completed their dental treatment was high (35%). Since the OHRQoL was re-assessed only for those who completed their dental treatment, future research could assess changes in OHRQoL more fully by re-surveying those who neither completed nor started their dental treatment.

In their analysis of the effect of literacy on health survey measurements, Al-Tayyib et al. (2002) found that self-administered questionnaires required not only literacy, but also 'forms-literacy', or the ability to implement survey instructions and select consistent responses. Due to the difficulty in scheduling the participants who had completed their dental treatment for a follow-up examination, the assistance of the welfare social workers had to be enlisted to obtain the follow-up questionnaires. It was not possible to train and calibrate the social workers to administer the questionnaires, and no tracking was done regarding which questionnaires were self-administered vs. which received help from the social worker. However, since only 22% of this study population had less than a high school education, and less than 5% of the follow-up questionnaires contained errors such as circling more than one answer or omitting a question, it is not likely that information bias was introduced by the follow-up questionnaires being self-administered.

As discussed in MATERIALS & METHODS, the SFDHS employment outcome codes were complex and contained some ambiguity. Additional analysis of a subgroup of codes that most closely represented gaining employment (n = 56) was compared with the 136 unfavorable outcome codes that reflected non-compliance or fraud. Other than wider confidence intervals, this subgroup analysis yielded results similar to those found in Table 4. Therefore, to minimize misclassification bias associated with the ambiguous employment outcome codes, and to maximize the number of subjects included in the analysis, the authors reported the results in Table 4 for the entire cohort (n = 377), using a combination of outcome codes.

The employment outcome codes reflected only the most recent disposition of the participant, and were overwritten when the outcome status changed. A previous study of the PAES program found that participants flowed back and forth between unfavorable and favorable employment status (San Francisco Department of Human Services, 2003). Ideally, future research could allow the employment outcomes to capture employment stability by providing a running tally of the participant's case disposition, rather than overwriting the employment codes whenever the status changed. Future research could also assess the financial sustainability provided by employment by recording the wage information.

The oral health and OHRQoL have not been previously assessed in a welfare population, nor has a Dental Program ever been offered as a welfare intervention. The improved OHRQoL and employment outcomes found for the participants who completed their dental treatment indicate that, for some welfare recipients, oral disease poses as significant a barrier to employment and self-sufficiency as do problems with general or mental health. Thus, oral health treatment can contribute to the goals of the Federal Personal Responsibility and Work Opportunity Reconciliation Act by eliminating barriers to employment and improving OHRQoL.