|Home | About | Journals | Submit | Contact Us | Français|
This study sought to determine the impact of health literacy and acculturation on oral health status of Somali refugees in Massachusetts.
Survey of 439 adult Somalis who arrived in the U.S. < 10 years ago. Subjects had an oral examination with decayed, missing, and filled teeth (DMFT) counts. STOFHLA was used to measure health literacy. Generalized linear multivariable regression models were used to assess the association between English literacy (STOFHLA) and the oral health parameters.
Participants had means of 1.4 decayed, 2.8 missing, and 1.3 filled teeth. Among subjects in the U.S. 0-4 years, subjects with low STOFHLA scores had lower mean DMFT (Mean Rate Ratio=0.78, p=0.016) compared with subjects with higher STOFHLA scores; however, among subjects in the U.S. 5-10 years, those with low STOFHLA scores had higher mean DMFT (Mean Rate Ratio=1.37, p=0.012) compared with subjects with higher STOFHLA scores. No significant association between STOFHLA and decayed teeth was detected. Participants with low STOFHLA scores had marginally lower risk of periodontal disease compared to those with high scores (OR=0.22, p=0.047)
Overall Somali oral health status is good. Oral health of Somalis with low health literacy worsens over time in the U.S. It is possible that beneficial factors linked to low literacy for newly arrived Somali refugees diminish and may be countered by less access to preventive care and less utilization of beneficial oral hygiene practices.
Among refugees newly arrived in Massachusetts, oral abnormalities are the most common health problem in refugee children and the second most common problem in refugee adults.(1) One major determinant of oral health disparities is access to preventative and restorative dental care.(2) Other determinants include oral hygiene practices and diet.(2) Linguistic and cultural factors may play important roles in determining access to oral health services as well as personal oral hygiene practices, and limited literacy skills have been hypothesized as “one of the many barriers to better oral health outcomes.”(3)
Health literacy is embedded in a variety of determinants of oral health, but the relationship between health literacy and oral health has never been studied in a refugee population. Health literacy is “the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions,” and is thought to play a pervasive role in all aspects of health care and oral health status. (3, 4) Inadequate health literacy has been associated with a long and growing list of adverse health outcomes.(5) Specifically, inadequate literacy has been associated with limited access and utilization of care,(6, 7) worse clinical outcomes,(8) hospitalization,(9) and mortality.(10)
The relationship between literacy and health outcomes has not been characterized in refugee populations in the United States (U.S.). Somalis are one of largest refugee populations to have entered the U.S. in recent years. As a result of civil war over the past 20 years, many Somalis have lived in refugee camps for long stretches of time. More recently arrived Somalis have very low literacy in English.(11) Somalis are almost all practicing Muslims and have relative homogeneity of language, culture, and religion. (12, 13) The Somali community in the U.S. also has been observed to be close-knit.
Past research has indicated a strong role for such social support as a moderator of health literacy and its impact on health status.(14) A refugee with low literacy may be able to effectively access care with the help of the community network. Thus, health literacy may function differently in the context of the Somali community, due to the existence of a particularly strong social support network. The degree to which an individual identifies with the traditional community and social structure or that of the dominant, host community varies and may impact how an individual negotiates competing priorities related to personal oral hygiene, diet, and access to dental care.
In this paper, acculturation is framed primarily in terms of behavioral acculturation: with whom one spends time, to what types of media they are exposed, in which language they feel most comfortable conversing and reading, and with whom they identify. The effects of acculturation on oral health have been studied in Haitian immigrants in the U.S. who had a low baseline rate of caries. Acculturation was found to be associated with lower rates of development of caries.(15) In Australia, a study of Vietnamese refugees also revealed associations between acculturation and dental health status.(16) The Vietnamese also had very good oral health status, and those with high acculturation had even better oral health status. This finding suggested that more acculturation led to protective practices and care that were additive to those of the refugees’ traditional culture. Of interest, though, the findings documented a non-linear relationship in which those refugees with moderate levels of acculturation had worse oral health status. In this case, the researchers hypothesized, based on the “cultural marginality model” and its previous application to oral health research, (17, 18) that refugees with moderate levels of acculturation were alienated from their traditional culture without adequate integration into the dominant culture. In this manner, moderately acculturated refugees might adopt behaviors deleterious to oral health, such as Western dietary habits, without adopting preventive aspects of Western oral hygiene and related behaviors. (16) In contrast, individuals with a low level of acculturation may have continued beneficial traditional practices and not adopted a cariogenic Western diet. In the Somali community, such a practice may include use of a stick brush. Studies have found the use of the stick brush (“miswak” or “aday”) to be effective in removing plaques.(11),(19) In addition, such brushes have an inhibitory effect on oral cariogenic streptococci (20),(21) and periodontal pathogens.(21)
Given the importance of health literacy to health in the general population, this study sought to determine the relationship of English health literacy with oral health clinical outcomes of Somali refugees in Massachusetts. The investigators hypothesized that when controlling for acculturation, subjects with high health literacy would be more likely than others to have: 1) less lifetime history of decay, untreated dental decay, and periodontal disease; 2) a higher rate of traditional or Western personal hygiene practices and behaviors known to be associated with better oral health outcomes; and 3) more utilization of professional dental care for preventive services. This paper reports on the relationship of health literacy with oral health clinical outcomes, which was the primary objective of the study that also included assessment of functional and mental health outcomes and a qualitative ethnographic assessment of a variety of social and cultural domains relevant to the roles of literacy, acculturation, and oral health care and personal hygiene practices in determining oral health status in the Somali community.
This cross-sectional survey entailed structured interviews of a convenience sample of Somali adults living in Massachusetts. All contacts with subjects, informed consent, interviews, and examinations were conducted in Somali by a Somali research assistant and dental examiner. With the exception of English language measures (BEST Plus, STOFHLA, and REALD), all written materials were translated into Somali to enhance standardization of the interviews. The translation process used standard procedures for group, consensus translation.(22) The study was approved and monitored by the Institutional Review Board of the Massachusetts Department of Public Health (MDPH).
Individual refugees were eligible for the study if they were 18 years of age or older, had arrived from overseas no more than 10 years prior to enrollment and were of Somali nationality. Exclusion criteria included functional visual impairment sufficient to prevent reading of test materials; medical diagnoses known to interfere with speech articulation; known cognitive impairment, learning disabilities or traumatic brain injury; and medical diagnoses requiring antibiotic prophylaxis for oral examination. Recruitment techniques were by word of mouth and then continued via snowball technique. The study used purposive sampling to ensure that the sample was representative of the general population of Somali refugees in Massachusetts with respect to age and gender. Specifically, age and gender groups were targeted for enrollment to match the distribution that would be expected in 2009 based on the cohort of Somali refugees for whom the Refugee and Immigrant Health Program (RIHP), MDPH, received formal notification of arrival in Massachusetts since 1999. This aged population had the following distribution among age groups by gender: 25% of the female Somali population was expected to be between 18-24 years of age in 2009; 51%, 25- 44 years; 16%, 45-64 years; and 8%, 65 years or older. Among the male Somali population, 27% were expected to be 18-24 years of age; 51%, 25-44 years; 15%, 45-64 years; and 6%, 65 years or older. While interviews were primarily conducted during the typical workweek, individuals were also interviewed outside of routine daytime business hours to facilitate enrollment of working subjects. Participants received a $50 gift card on completion of the interview. Twenty percent of subjects also participated in an extended qualitative interview, to be presented separately.
The research interview included questions on demographics, cultural practices, educational experiences, oral health practices, and experience with dental care services. Assessment of dental and oral health care experiences was conducted through the Access to Care Questionnaire of the Basic Screening Survey (BSS).(23) Questions on access to medical care came from the National Health Interview Survey; these were supplemented with questions regarding personal care and lifestyle practices relevant to oral health.
Functional health literacy was assessed with the Short Test of Functional Health Literacy in Adults (STOFHLA).(24) The STOFHLA is a 36-item test of reading comprehension that uses a set of sentences from medical scenarios with key words missing. Subjects select words to complete the sentences from a list provided. The test has high internal consistency and correlates well with other health literacy tests. STOFHLA scoring was dichotomized as low (0-22 = inadequate) and higher (23-36 = marginal or adequate).
Additional instruments assessed functional health with the Medical Outcomes Study Short-Form 12-Item Survey (SF-12),(25) oral health quality of life using the Oral Health Quality of Life Instrument (OHQOL),(26) and acculturation using a revised Haitian Acculturation Scale (r-HAS).(15) In addition, given concerns about the impact of mental illness on functional and oral health status, the study included mental health screening with the Posttraumatic Stress Disorder Checklist – Civilian Version (PCL-C) and the Patient Health Questionnaire (PHQ) for depression.(27, 28) For the PHQ, the two-item version (PHQ-2) was used for screening with the remaining 7 items used if a subject gave a positive response to the first two items.(29)
All subjects were examined by a trained examiner using a portable light source, dental mirror, explorer and periodontal probe. For the caries assessment, each tooth present in the mouth was classified as sound, decayed, missing, or filled using the Modified Basic Screening Survey criteria.(23) Filled teeth were presumed to have had decay and, throughout this paper, are noted as “DFT” for decayed and filled teeth. Decayed, but unfilled, teeth are noted as “DT.” Periodontal assessment was based upon World Health Organization (WHO) criteria using the Community Periodontal Index of Treatment Needs (CPITN) probe.(18) Probing was performed on all teeth in each sextant and each tooth was classified based on CPITN criteria: no need for care (=0), bleeding gingiva on probing (=1), presence of calculus and other plaque-retentive factors (=2), 4 or 5mm periodontal pockets (=3) or pockets that are 6mm or more deep (=4). The highest scoring tooth (worst condition) was recorded as the overall score for each sextant. The overall dental condition was then classified into four levels to reflect the amount of treatment needed and the urgency of need.
Two medical anthropologists conducted detailed, semi-structured interviews of 84 subjects selected from the quantitative sample. These interviews explored domains of the Eraker theoretical model of health decision making. (30) Transcripts of the interviews were coded thematically by four members of the study team and then analyzed using HyperRESEARCH (HyperRESEARCH computer software, version 2.8.3. ResearchWare, Inc., Randolph, MA: 2009). While not reported in this paper in detail, qualitative findings were used to guide analyses and interpretation of the quantitative data.
Analyses focused on three main oral health outcome measures that are derived from oral exams: 1) the mean number of decayed, missing and filled teeth - DMFT (representing the lifetime history of decay); 2) the ratio of the number of decayed teeth to the total number of decayed and filled teeth – DT/(DT+DFT) (representing the proportion of untreated carious teeth); and 3) CPTIN score. For analysis we dichotomized the DT/(DT+DFT) ratio and CPTIN. Those with the ratio greater than 0.5 were considered to have high unaddressed decay, and those with CPTIN score of 3 or 4 in any sextant were considered to have periodontal disease. We also examined several secondary measures – DT, DFT and MT. For analyses, several main domains of predictor variables that were thought likely to affect oral health status were selected. These included: acculturation, socio-demographic factors, oral health practices, and general mental and physical health scores. Acculturation was defined by the r-HAS score and categorized into three levels – low, medium and high - empirically based on the distribution. (15) Socio-demographic factors included gender, age, ethnicity, education, years lived in U.S. and income. Oral health practices included type of dental insurance, preventive care within the past year, restorative care within the past year, regular tooth brushing and use of the stick brush. General health scores included OHQOL, PCL-C, and SF-12 physical and mental component scores, and language measures included the STOFHLA and BEST.
Analyses included bivariate comparisons of health literacy (STOFHLA) with the dental health outcome measures first. Multivariate analyses then included the language measures, selected confounders, and an interaction term between the risk variable (literacy) and years lived in U.S. The interaction term was added after preliminary analyses determined that the relationship between oral health status and literacy was affected by time lived in US. In order to build more parsimonious models, the backward elimination procedure was used with a 0.2 alpha level for variables to stay in the model.
Because of the distribution of the data and low numbers in some categories, zero-inflated Poisson models (SAS Proc GENMOD) were used for DMFT, DT, DFT and MT. Logistic regressions (SAS Proc GENMOD) were used for high unaddressed decay and presence of periodontal disease. All statistical analyses were performed using SAS software, version 9.2 (SAS Institute Inc, Cary, NC, USA).
A total of 439 refugees aged 18 years and older participated in the study. The sample was 58.1% female. In terms of age and gender distributions, the sample population roughly approximated the expected distribution of age and gender in the larger Somali community of Massachusetts. Among females, older adults were somewhat over-represented (45-64 years of age, 22% vs. 16%) while younger adults were somewhat under-represented (25-44 years of age, 47% vs. 51%). Among males, older adolescents were over-represented (18-24 years of age, 39% vs. 27%) while younger adults were under-represented (25-44 years of age, 36% vs. 51%). These patterns reflect the greater difficulty entailed in enrolling employed, young adults. Eighty-seven percent of study participants were ethnic Somalis, while 13% were Somali Bantu. All had arrived in the U.S. within the last ten years; only 2% arrived directly from Somalia; others arrived from Kenya (78%), Ethiopia (9%) and fifteen other countries. Seventy-five percent had less than a high school education and most participants reported low or very low income.
While 63% had seen a dentist at some point during their time in the U.S., 53.6% had done so in the past year. Nearly all participants (98%) reported brushing their teeth at least daily, and dental floss was reported by 40%. The use of the traditional miswak stick brush was reported by 43%. In bivariate analyses of descriptive demographic variables, a number of demographic variables varied significantly between STOFHLA literacy levels. In summary, those with lower STOFHLA literacy level were more likely to be less acculturated, female, older, Somali Bantu ethnicity, without any formal education, insured by Medicaid, and use a traditional miswak stick brush while being less likely to report brushing their teeth more than once daily as compared to the group with higher STOFHLA level (all P<.05). (TABLE 1)
On examination, participants had an average of 1.39 decayed, 2.76 missing, and 1.34 filled teeth. Periodontal disease was noted in 6.5%. The distribution of DMFT and periodontal disease prevalence by demographic variables is presented in Table 2. The prevalence of these varied over time in the U.S. and with age. In addition, they tended to be more prevalent among those subjects with no education, Somali ethnicity, and lower income levels. For DMFT, the prevalence was higher among women while that of periodontal disease was higher among men. (TABLE 2)
Of the total sample, 326 (74.3%) had low health literacy. Bivariate analysis showed a trend of higher DMFT count in the low health literacy group. Mean DMFT in low STOFHLA group was 5.8 comparing to 4.7 in high STOFHLA group (p=0.085), which is about 20% higher. In the multivariate analysis the number of years a participant had lived in the U.S. at the time of interview was identified as an effect modifier, so the relationship between STOFHLA and DMFT was stratified by years in the U.S. (0-4, 5-10 years). In the adjusted analysis, among participants living in the U.S. for 0-4 years, those with low STOFHLA scores had on average 1.3 times fewer decayed, missed or filled teeth, representing lifetime history of disease comparing to those with higher scores (Mean rate ratio 0.78, p=0.02). In contrast, among participants living in the U.S. for 5-10 years, those with low STOFHLA scores had 1.4 times more lifetime history of dental decay than those with higher scores (Mean rate ratio 1.37, p=0.01). Those with medium acculturation had 20% less lifetime history of decay than those with high levels of acculturation (p=0.02). Somalis had 1.59 times more lifetime history of disease than Somali Bantus (p<0.0001). Those who did not use the stick brush had 1.23 times more lifetime history of disease than those who did (p=0.0001). (TABLE 3)
In the unadjusted analysis the high DT/(DT+DFT) ratio was highly associated with low health literacy. The proportion of those with high ratio in the low health literacy group was almost twice as high as in the high literary group (19% vs. 38% correspondingly, p<0.0001). However, in the adjusted model designed to reflect the relationship between STOFHLA and untreated decay (DT/DT+DFT), no statistically significant association between them was detected; although after 4 years in the U.S., the relationship showed the opposite pattern as that noted between STOFHLA and DMFT. Participants with low acculturation had 2.75 times the odds of untreated decay as compared to those with high acculturation (p=0.01). Those with medium acculturation were not significantly different from those with high acculturation (TABLE 4).
Bivariate analysis showed no significant association between health literacy and periodontal disease. In high literacy group the periodontal disease was found in about 6.3% of subjects, and in low literacy group the periodontal disease was found in about 6.5% of subjects (p=0.999). In adjusted analyses, those with low STOFHLA scores had 0.22 times the odds of periodontal disease as those with higher STOFHLA scores (p=0.047). Within this model, those with low levels of acculturation had 11.2 times greater probability of periodontal disease as those with high acculturation (p=0.01), and those with medium acculturation had 6.3 times the odds of periodontal disease as those with high acculturation (p=0.02) (TABLE 5).
Overall the study’s Somali sample had a very positive oral health profile as highlighted by the mean number of DMFT of 5.5. For comparison, the mean number of DMFT in the general U.S. adult population is nearly double that of the Somali sample at 10.3. (31) For periodontal disease, the U.S. prevalence is 8.5%, over 30% higher than the 6.5% prevalence of the Somali sample. (32) Only for decayed teeth is the Somali status notably worse with a mean of 1.4 compared to 0.8 in the U.S. population, a finding that most likely reflects untreated decay near to arrival in the U.S. In addition to the favorable exam findings, nearly the entire sample, 98%, reported daily tooth brushing, and almost 75% reported brushing at least twice daily.
With respect to the study’s main hypothesis, that Somalis with higher health literacy levels would have better clinical oral exam findings independent of acculturation levels, the study findings were mixed. In one respect, the study demonstrated the opposite: for individuals who had lived in the U.S. less than five years, low health literacy as measured by STOFHLA was associated with better oral exam findings (i.e., lower DMFT). This association was independent of a range of potential confounders including acculturation level, as listed in Table 3.
Given that few people saw dentists while in Somalia, Kenya or Ethiopia, lower lifetime history of caries disease among those with limited or no English ability (as well as low acculturation level) shortly after arrival in the U.S. may indicate that those individuals may have benefited from relatively less exposure to - and active avoidance of - refined sugar overseas. Past research has shown that non-Western diets have protective influence on oral health. (33) Individuals who have good health literacy (and general literacy in English) when they arrive in the U.S. were more likely to have lived in cities overseas, particularly Nairobi, and qualitative interviews suggested that those refugees with urban origins consumed more refined sugar overseas than did refugees from rural areas or refugee camps. Lastly, high levels of fluoride have been documented as naturally occurring in the water supply of parts of East Africa (34, 35).
After four years in the U.S., the relationship between STOFHLA scores and lifetime history of disease reversed, with lower literacy associated with increased mean DMFT score as predicted in the study’s main hypothesis. At five or more years in the U.S., those with higher literacy may have been more likely to pursue and gain access to preventive dental care and also to discern between foods that are good for oral health and those which are not. Alternatively, those with low health literacy may have been more likely to adopt dietary patterns that are deleterious to oral health while not pursuing or accessing preventive care. Qualitative findings confirmed that those with low literacy and acculturation were not as connected to care or simply may have reacted differently to symptoms of decay. Also in qualitative interviews, subjects in the U.S. for less than five years were much less likely to be able to access care, so if they had acute decay on arrival in the U.S. or soon after, they may have been more likely to forego treatment. In contrast to DMFT, the proportion of untreated carious teeth was not significantly different between those with low and those with high literacy when controlling for other factors; however, the trend in the relationship between DT/(DT+DFT) and STOFHLA supported the study’s hypothesis for those who had been in the U.S. less than five years.
While it was expected that those with low and high acculturation would have better overall oral health compared to those with medium acculturation, our results were more complex and did not match this U-shaped relationship with oral health as previously reported in the Haitian refugee population.(15) Those Somalis with relatively higher acculturation levels reported in qualitative interviews that they attributed a relatively lower level of importance to regular preventive dental visits as compared to other, more pressing concerns. However, higher levels of acculturation were also associated with better access to care. Consequently, those subjects with greater acculturation had less likelihood of having untreated active decay as compared to the less acculturated group. In reality, though, we found much less overall acculturation than expected.
Overall, participants identified themselves culturally and linguistically as Somali and had very low levels of English literacy, spoken English proficiency, and formal education. The population was also very low economic status; yet, they had relatively low rates of lifetime history of oral disease that may reflect both good oral hygiene practices, such as use of the stick brush, and at least for the first few years after arrival in the United States, diets that were low in refined sugar. These practices could be proxies for a myriad of other habits which ultimately may derive from devout practice of Islamic faith in this community, as was suggested in qualitative interviews.
The difference between Somalis and Somali Bantus may highlight the importance of diet in lifetime history of disease in this population as well as the role of traditional cultural factors. Somali Bantus are a minority ethnic group in Somalia who lived in the South of Somalia and typically followed a very traditional, agrarian lifestyle. Caught in the crossfire of the civil war, many displaced Somali Bantus remained in refugee camps for extended periods and seldom had formal education. Even in the U.S., this population is often on the margins of the Somali community such that Somali Bantus are likely to acculturate at an even slower rate than ethnic Somalis. As might be expected, then, Somali Bantus consumed virtually no refined sugar before they arrived in the U.S. and adhered to traditional dietary and cultural practices. The qualitative research findings suggest that Somali Bantus continue to be much less likely to consume refined sugar in the U.S. than ethnic Somalis.
While the sample was robust, as a convenience sample the participating subjects may not necessarily have been representative of the larger community. With respect to age and gender, the sample appeared similar to the overall cohort of Somalis in Massachusetts since 1995. Some effort was made to recruit participants on weekends, but Somalis who worked during the day were presumably underrepresented in the sample as compared to those who worked nights, weekends, or were unemployed.
English literacy and spoken proficiency may have been underestimated. Several participants suspected to have limited English literacy and spoken ability declined the language assessments and insisted that they could not read any English at all. In reality, those subjects likely would have scored higher than zero. In contrast, we suspected general reporting bias towards responses that would be viewed more favorably. For example, virtually no subjects reported cigarette smoking or symptoms of depression, findings that do not match the researchers’ experiences with the larger Somali community and do not seem to be due to sampling bias. The lack of depressive symptoms reported by our sample also is contradictory with previously published research with Somalis living in Minnesota (36).
All of the results must be viewed cautiously given the significant cultural and linguistic challenges faced by the investigative team. When possible, we used validated instruments with strict instructions to research staff on their non-biased administration. These instruments were not necessarily ideally transferable to the Somali community; however, development and validation of instruments solely for this study were not feasible. In particular, it should be noted that the r-HAS instrument was not designed or validated for the Somali population, and the Somali community may manifest acculturation in domains not measured by an instrument that was developed for Southeast Asian and Haitian populations. Piloting of r-HAS provided some assurance to the investigators, though, that the instrument would be appropriate for the Somali community.
The oral health status of Somali refugees is very good. English health literacy and acculturation had unexpected and complex associations with oral health outcomes. Specifically, those with lower health literacy had a reduced risk of DMFT initially but then had an increased risk at five years after arrival in the U.S. In contrast, low health literacy had the opposite relationship with untreated decay over time in the U.S. Participants’ reports of family resources and knowledge of oral hygiene highlight the significant health assets present in the community; such factors may mitigate the potential impact of health literacy and acculturation on health status. This was illustrated by the overwhelmingly favorable reception of the study by Somalis and their great interest in study findings presented at community meetings. However, relatively higher levels of acute decay may indicate a future problem, as decay was not necessarily identified early by subjects or treated. As Somalis adopt less favorable dietary practices over time in the U.S., oral health promotion and prevention for Somalis should highlight the importance of a diet low in refined sugar, frequent brushing, retention of beneficial traditional practices such as the miswak stick brush, linkages with cleanliness rituals before prayer, and the importance of preventive visits to the dentist.
This study was funded by the National Institute of Dental and Craniofacial Research (Paul L. Geltman, Principle Investigator; Grant number 1R01DE017716-01A2). Study findings were presented in part at the Health Literacy Annual Research Conference on October 17, 2011. The authors wish to acknowledge and thank our Somali study staff, Fadumo Egal and Ahmed Hassan, for the efforts in making this project successful and also the Somali community of Eastern Massachusetts for its enthusiastic embrace of the study.
Human Participant Protection: The study was approved and monitored by the Institutional Review Board of the Massachusetts Department of Public Health.
Contributor Statement: Dr. Geltman, Ms Hunter Adams, Ms Cochran, Dr. Barnes, Dr. Henshaw, and Dr. Paasche-Orlow were all involved with the study design and implementation; data collection, analyses, and interpretation; and manuscript preparation and editing. Dr. Doros and Mr. Rybin were involved with study implementation; data collection, analyses, and interpretation; and manuscript preparation and editing.
Paul L. Geltman, Department of Pediatrics, Boston University School of Medicine Refugee and Immigrant Health Program, Massachusetts Department of Public Health.
Jo Hunter Adams, Refugee and Immigrant Health Program, Massachusetts Department of Public Health.
Jennifer Cochran, Refugee and Immigrant Health Program, Massachusetts Department of Public Health.
Gheorghe Doros, Department of Biostatistics, Boston University School of Public Health.
Denis Rybin, Department of Biostatistics, Boston University School of Public Health.
Michelle Henshaw, Department of Health Policy and Health Services Research, Boston University Henry M. Goldman School of Dental Medicine.
Linda L. Barnes, Department of Family Medicine, Boston University School of Medicine.
Michael Paasche-Orlow, Department of Medicine, Boston University School of Medicine.