Microsoft Word - 58 2016.docx   Vol  4,  No  1  (2016)   ISSN  2167-­‐8677  (online)   DOI  10.5195/d3000.2016.58           http://dentistry3000.pitt.edu     New  articles  in  this  journal  are  licensed  under  a  Creative  Commons  Attribution  4.0  United  States  License.     This  journal  is  published  by  the  University  Library  System,  University  of  Pittsburgh  as  part  of  its  D-­‐Scribe  Digital  Publishing  Program  and  is  cosponored   by  the  University  of  Pittsburgh  Press.     Multilevel  modeling  for  dental  caries  among  adolescents  in  a  Bra-­‐ zilian  large  city     Giovana  Daniela  Pecharki  1,  João  Armando  Brancher1,  Márcia  Olandoski1,  Andrea  Duarte  Doetzer1,  Samuel  Jorge  Moyses1,  Paula  Cris%na  Trevi-­‐ la!o1     1  School  of  Health  and  Biosciences,  Pon&'cia  Universidade  Católica  do  Paraná  (PUCPR),  Rua  Imaculada  Conceição,  Curi'ba,  PR,  Brazil Abstract   Dental  caries  is  a  complex  disease,  which  needs  an  approach  that  considers  caries  influenc-­‐ ing  factors  at  different  levels  and  their  integra2on.  Mul$level  Modeling  is  a  clustered  analy-­‐ sis  of  variables  from  the  individual  to  the  community  level.  The  aim  of  this  study  was  to  in-­‐ ves$gate  the  associa$on  of  social  and  biological  factors  grouped  into  hierarchical  levels,  in   students   with   caries.   A   sample   of   687   students   was   evaluated   from   public   and   private   schools  of  Curi,ba.  The  parameters  evaluated  were:   individual   level,   school   level  and  dis-­‐ trict  level.  Individual  variables  had  a  highly  significant  associa6on  with  caries  experience,  al-­‐ so  in  the  presence  of  school  and  district   levels.  Male  sex  nega4vely  associated  with  caries   experience.  However,  the  interac2on  between  male  sex  and  no  fluoride  use  was  posi2vely   associated  with  caries.  Lower  socioeconomic  status,  dental  biofilm,  and  fluorosis  were  as-­‐ sociated  with  caries.  Nevertheless,  the  interac3on  between  dental  biofilm  and  fluorosis  was   nega%vely  associated  with  caries  experience.  The   interac%on  between  no  flossing  and  use   of  public  dental  services  were  also  associated  with  caries  outcome.  Individual  factors  were   associated   with   caries   experience   even   with   the   inclusion   of   contextual   variables   in   the   study  popula+on.     Cita%on:  Pecharki,  et  al.  (2016)     Mul$level  Modeling  for  Dental  Caries  among   Adolescents  in  a  Brazilian  Large  City.     Den$stry  3000.  1:a001   doi:10.5195/d3000.2016.58   Received:  July  22,  2016   Accepted:    July  29  ,  2016   Published:    October  3,  2016   Copyright:  ©2016  Pecharki,  et  al.  This  is  an  open   access   ar!cle   licensed   under   a   Crea!ve   Com-­‐ mons   A!ribu%on   Work   4.0   United   States   Li-­‐ cense.   Email:  paula.trevila+o@pucpr.br   Introduction   Dental  caries  is  an  infec-­‐ tious  and  multifactorial  disease.   The  prevalence  of  caries  has  re-­‐ duced  significantly,  including  in   Latin  America  and  Brazil   (DMFT=2.07  for  12  years-­‐old)  [1,2]   possibly  due  to  the  increased   availability  of  fluoride  and  oral   health  programs  [3,2].   Nevertheless,  groups  of   children  have  still  been  showing   high  caries  activity.  It  is  estimated   that  20  to  25%  of  children  and  ad-­‐ olescents  in  Brazil  concentrate  60   to  80%  of  caries  prevalence  [2].   The  phenomenon  of  disease  con-­‐ centration  in  small  groups  is   termed  polarization  and  repre-­‐ sents  one  of  the  epidemiological   disease  aspects,  where  a  popula-­‐ tion  portion  focuses  most  needs   for  treatment  [2,4].  Thus,  the  early   identification  of  caries  risk  indica-­‐ tors  is  fundamental  for  measures   of  prevention,  control  and  reduc-­‐ tion  of  damage,  with  obvious  epi-­‐ demiological,  human  and  econom-­‐ ical  consequences  [5].   Socioeconomic  status  [6],   oral  health  behavior,  including  diet   [7],  gender,  ethnicity  [3]  and  bio-­‐ logical  factors,  such  as  biofilm   formation  and  saliva  properties   [8,9],  have  been  considered  the   main  risk  factors  influencing  indi-­‐ vidual  susceptibility  to  caries.   However,  there  has  been  a  shift   from  individual  to  population– level  approach  when  researching   the  risk  indicators  for  chronic  dis-­‐ eases  [10].  As  previously  pub-­‐ lished,  individual  factors  are  usual-­‐ ly  influenced  by  the  population   context  [7,10].  Also,  a  higher  car-­‐ ies  experience  and  a  lower  dental   care  index  (the  ratio  between  the   number  of  filled  teeth  and  DMFT)   were  observed  in  children  attend-­‐ ing  public  schools  than  for  those   enrolled  in  private  schools  [3]  and   in  areas  with  lower  levels  of  em-­‐ powerment  and  income  [11,  12].   There  has  been  seen  significant   Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   2   variation  in  the  severity  of  caries   between  low-­‐income  neighbor-­‐ hood  clusters.  Indeed,  neighbor-­‐ hood  features  were  seen  to  be  as-­‐ sociated  with  self-­‐reported  oral   health  [13].     Caries  is  a  complex  disease   and  an  approach  that  can  evaluate   the  factors  influencing  the  disease   at  different  levels  and  their  inte-­‐ gration  is  suitable.  To  date  there   are  only  few  studies  evaluating   caries  in  a  multilevel  approach,   and  they  do  not  analyze  all  factors   that  may  be  contributing  to  its   complex  development  [7].    Multi-­‐ level  Analysis  is  a  model  that  con-­‐ siders  factors  related  to  the  indi-­‐ vidual  together  with  the  commu-­‐ nity  level,  organized  in  hierarchical   levels.  Thus,  the  aim  of  this  study   was  to  investigate  the  association   of  individual  and  community  fac-­‐ tors,  grouped  into  hierarchical   levels  by  Multilevel  Modeling.     Methods   Six  districts  in  Curitiba-­‐  PR,   Brazil  with  similar  socioeconomic   aspects  were  assessed  in  this   study  (Table  1).  In  Brazil,  the  soci-­‐ oeconomic  status  generally  im-­‐ plies  in  the  type  of  school,  public   (lower  income)  or  private  (higher   income).  Then,  two  larger  schools   from  each  district,  one  public  and   one  private,  with  a  minimum  of   1500  students,  were  chosen  using   a  table  of  random  numbers  totaliz-­‐ ing  twelve  schools.  All  12-­‐year-­‐old   students  or  those  which  would   complete  this  age  in  the  year  of   the  study  were  invited  to  partici-­‐ pate  on  each  school.  An  informed   consent  form  was  given  to  every   student  caregiver  in  each  school.   Adolescents  whose  caregiv-­‐ ers/parents  did  not  return  the   consent  form  were  not  included   for  study  along  with  smokers,  or-­‐ thodontic  appliances  users,  and   Table 1. Districts demographic and socioeconomic aspects. ! DISTRICTS' ! 1! 2! 3! 4! 5! 6! ! DEMOGRAPHIC! ASPECTS! 248,698! inhabitants!or! 14.2%!of!the! total! Municipality! 155,794! inhabitants! or!8.9%!of! total! Municipality! ! 215!503! inhabitants!or! 12.3%!of!the! total! Municipality! 197!346! inhabitants! or!11.3%!of! the!total! Municipality! 168,425! inhabitants!or! 9.6%!of!total! Municipality! ! 243!506! inhabitants!or! 13.9%!of!the! total! Municipality! 47427! children!!!!0M 14!years!old! (19.1%!of!the! regional! population)! ! 42125! children! 0M14!years! old!(40.4%!of! the!regional! population)! ! 45711! children!!!!!!!! 0M14!years! old!(21.2%!of! the!regional! population)! 41297!M! children!!!!!!0M 14!years! (20.9%!of!the! regional! population)! ! 42087! children!!!!!!!!!!!!! 0M14!years!old! (25.0%!of!the! regional! population)! 42125! children!!!!!!!!!! 0M14!years! (17.3%!of!the! regional! population)! ! SOCIOECONOMIC! ASPECTS! !!!!!!!!!!!!!!INCOME! PER!CAPITA!–! 2010*!!!!!!!!!!!(*1! minimum!wage,!! MW=!$130)! 45.9%! received!over! 1!to!3!MW! 42.5%! received!over! 1!to!3!MW! 44.3%! received!over! 1!to!3!MW! 50.4%! received!over! 1!to!3!MW! 44.9%! received!over! 1!to!3!MW! 40.0%! received!over! 1!to!3!MW! WATER!SUPPLY! 99.2%!of! households! connected!to! the!water! network! 98.8%!of! households! connected!to! the!water! network! 99.4%!of! households! connected!to! the!water! network! 99.7%!of! households! connected!to! the!water! network! 99.0%!of! households! connected!to! the!water! network! 98.9%!of! households! connected!to! the!water! network! SEWAGE! NETWORK! 82.7%!of! households! connected!to! sewage! network! 86.6%!of! households! connected!to! sewage! network! 94.8%!!of! households! connected!to! sewage! network! 93.9%!of! households! connected!to! sewage! network! 89.4%!of! households! connected!to! sewage! network! 95.8%!of! households! connected!to! sewage! network! Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   3   individuals  taking  antibiotics  in  the   last  three  months.  The  study  was   approved  by  the  Ethical  Commit-­‐ tee  on  Research  of  Pontifical   Catholic  University  of  Paraná   (PUCPR)  under  register  n°  487.   They  were  distributed  about  1100   consent  forms,  and  687  12-­‐year-­‐ old  unrelated  adolescents  agreed   to  participate,  approximately  55   students  from  each  school,  both   sexes.   The  687  students   were  diagnosed  according  to   the  decayed,  missing  and   filled  teeth  index  (DMFT).   White  spot  lesions  in  dental   surfaces  were  considered   decayed  according  to  clinical   criteria  described  by  Assaf  et   al.  [14].     Examinations  were   conducted  in  schoolrooms  in   accordance  with  interna-­‐ tional  standards  established   by  World  Health  Organiza-­‐ tion  (WHO)  [15].  All  evalua-­‐ tions  were  performed  by   two  calibrated  examiners.   Inter-­‐  and  intra-­‐examiner   reproducibility  was  taken  on   10%  of  the  sample  and  the   Kappa  test  was  used  to   measure  reliability.  The  ob-­‐ tained  values  for  Kappa  test   were  0.93  for  inter-­‐  and  0.99   for  intra-­‐examiner.   A  multilevel  study   was  designed  to  assess  the   individual  and  contextual   effects  on  caries  experience   (DMFT=0  or  DMFT≥1).  Data   were  hierarchically  struc-­‐ tured  in  three  levels:  indi-­‐ viduals  (level  1),  schools   (level  2),  and  districts  (level   3).  The  choice  of  variables   was  based  in  caries  multifac-­‐ torial  model  proposed  by   Fejerskov  &  Manji  [16].   Individual  level  (level  1)   Individual  demographic   variables  included  sex   (male/female)  and  ethnicity:   white/  light-­‐  and  dark-­‐ Table&2.&Individual)and)contextual)characteristics)of)students)(n=687))and)caries)status.) Variables& n& Frequency& (%)& Proportion&of&DMFT≥1&(%)& ! 687) 100.0) ) Level!1:!Student! ) ) ) Sociodemografic) ) ) ) Ethnic'group' ) ) ) ) White)) 606) 88.2) 48.5a) ) LightD)and)darkDskinned)black) 62) 9.0) 69.3b) ) Yellow) 19) 2.8) 63.2a,b) Sex' ) ) ) ) Female)) 377) 54.9) 52.2ª) ) Male) 310) 45.1) 49.0a) Behavioural) ) ) ) Toothbrushing'frequency) ) ) ) ) 2)or)more/day) 644) 93.7) 49.8ª) ) Until)once/day) 43) 6.3) 65.1ª) Flossing' ) ) ) ) Yes) 443) 64.5) 47.2a) ) No) 244) 35.5) 57.4b)) Fluoride'use'(solution,varnish,gel)' ) ) ) ) Yes) 405) 59.0) 47.4a)) ) No) 282) 41.0) 55.7b) Dental'visits'frequency' ) ) ) ) 2)times)or)more/year) 304) 44.2) 48.7a) ) Once/year) 315) 45.9) 50.8a) ) No) 68) 9.9) 60.3a) Sugar'consumption'between'meals' ) ) ) ) No) 69) 10.0) 50.7ª) ) Yes) 618) 90.0) 50.8ª) Socioeconomic) ) ) ) Dental'access' ) ) ) ) Private) 427) 62.2) 41.4ª) ) Public) 260) 37.8) 66.1b) Individual'socieconomic'status' ) ) ) ) A1/A2)(highest)) 175) 25.5) 32.6ª) ) B1/B2) 310) 45.1) 51.9b) ) C) 166) 24.2) 65.1c) ) D/E)(lowest)) Clinical) 36) 5.2) 63.9b,c) Plaque'Index'(modified)' ) ) ) ) 0) 95) 13.8) 38.9ª) ) >0)and)<1) 475) 69.2) 50.1ª) ) ≥1) 117) 17.0) 63.2b) Gengivitis' ) ) ) ) No) 404) 58.8) 47.0a) ) Yes) 283) 41.2) 56.2b) Fluorosis' ) ) ) ) No) 497) 72.3) 52.3ª) ) Yes) 190) 27.7) 46.8a) Stimulated'salivary'flow'rate' ) ) ) ) >0.5)mL/min) 522) 76.0) 51.0a) ) ≤0.5)mL/min) 165) 24.0) 50.3a) Salivary'buffering'capacity' ) ) ) ) pH>3.9) 670) 97.5) 50.9ª) ) pH≤3.9) 17) 2.5) 47.1a) Level!2:!School! ) ) ) Type'of'school' ) ) ) )))Private) 334) 48.6) 41.3ª) )))Public) 353) 51.4) 59.8b) Oral'health'education'' ) ) ) )))Yes) 240) 34.9) 52.5a)) )))No) 447) 65.1)) 49.9a) Permission'for'sweet'consumption'' ) ) ) )))Non)permitted) 228) 33.2)) 37.3a) )))Permitted) 459) 66.8) 57.5b) Level!3:!District& ) ) ) Fluoride'concentration'in'water'supply' ) ) ) )))≥0.7)mL/L) 333) 48.5) 49.2a) )))<0.7)mL/L) 354) 51.5) 52.3a) Socioeconomic'position' ) ) ) )))better)living)conditions) 447) 65.1) 47.2a) )))worse)living)conditions) 240) 34.9) 57.5b) Fisher’s)exact)or)chiDsquare)test.)Distinct)lowerDcase)superscript)letters)indicate)statistical)significance)(p<0.05).) Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   4   skinned/black  and  yellow  (Asian   descent).The  parent/caregiver  an-­‐ swered  self-­‐completed  question-­‐ naires  about  oral  health  behavior:   tooth  brushing  frequency,  dental   flossing,  topical  fluoride  (solution,   varnish,  gel),  dental  attendance   pattern  (frequency  of  dental   check-­‐ups  and  public/private  ac-­‐ cess)  and  sugar  consumption  be-­‐ tween  meals  [12,6].     A  standard  Brazilian  socio-­‐ economic  classification  based  on   household  items  and  on  the  level   of  education  of  the  head  of   household  was  adopted  [12].  A  set   of  points  is  assigned  to  these  indi-­‐ cators  and  a  final  score  defines  the   socioeconomic  groups;  A  (highest),   B,  C,  D,  and  E  (lowest).  Because  of   the  small  number  of  observations   in  class  E,  data  were  categorized   into  four  groups:  high  social  class   (class  A),  high-­‐middle  social  class   (class  B),  middle  social  class  (class   C)  and  low  social  class  (classes  D   and  E).     The  biofilm  accumulation   was  verified  by  the  Plaque  Index   (PI)  [17]  modified,  which  adopted   the  same  criteria,  but  evaluating  6   teeth  surfaces:  16  [buccal  (B)],  12   B,  26  [lingual  (L)],  36  B,  32  L,  and   46  L.  The  PI  was  categorized  as:  no   plaque  accumulation  (PI=0),  regu-­‐ lar  plaque  accumulation  (PI>0  and   <1,  up  to  a  third  of  the  surface   with  accumulated  biofilm)  and   high  plaque  accumulation  (PI≥1,   more  than  a  third  of  the  surface   with  accumulated  biofilm).  Indi-­‐ viduals  were  considered  positive   for  gingivitis  when  all  teeth  from   at  least  one  sextant  were  affected,   with  bleeding  and  inflammation.   The  presence  or  not  of  any  degree   of  fluorosis  was  established  ac-­‐ cording  to  Dean’s  index  (WHO)   [15]  .   The  stimulated  salivary   flow  rate  (SSFR)  was  measured  as   previously  described  [18].  To  clas-­‐ sify  the  SSFR  the  following  numer-­‐ ical  scores  were  attributed:  0  for   low  caries  risk  >0.5  mL/min  and  1   for  high  caries  risk  ≤0.5  mL/min,   because  the  cut  line  for  dichoto-­‐ mization  was  based  on  the  data   distribution.  The  buffering  capaci-­‐ ty  (BC)  was  performed  as  de-­‐ scribed  previously  [19].  Final  pH  of   the  mixture  was  determined  using   a  pH-­‐meter  and  BC  was  consid-­‐ ered  good  if  the  final  pH  was  >3.9   and  deficient  for  pH≤3.9.     School  level  (level  2)   Twelve  large  schools  were   randomly  chosen,  being  1  public   and  1  private  from  each  of  the  6   health  districts  studied.  Besides   the  type  of  school,  it  was  also  veri-­‐ fied  permission  or  not  for  sweet-­‐ ies’  consumption  in  the  classroom   and  other  areas  in  the  institution   and  the  presence  or  not  of  oral   health  education  programs.   District  level  (level  3)   Means  of  fluoride  concen-­‐ tration  in  water  supply  in  2006   were  obtained  for  6  Health  Dis-­‐ tricts  from  the  Municipal  Health   Secretary  of  Curitiba  and  dichot-­‐ omized  into  <0.7  and  ≥0.7  mL/L,   because  it  is  the  minimum  concen-­‐ tration  allowed  in  the  city.   The  classification  of  the  so-­‐ cioeconomic  conditions  in  the   health  districts  was  described  by   Moraes  &  Ribeiro  [20],  and  con-­‐ sidered  income  and  education  of   the  household  head.  The  score   was  then  organized  in  descending   order  and  grouped  into  5  socioec-­‐ onomic  categories.  The  first  cate-­‐ gory  was  A  (best  living  conditions)   and  the  last  category  was  E  (worst   living  conditions).  Because  of  data   distribution  in  this  study,  it  was   chosen  to  join  B/C  (better  living   conditions),  and  D/E  (worse  living   conditions).  None  of  the  evaluated   districts  obtained  score  A.         Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   5   Firstly,  an  exploratory   study  was  performed  using  Fisher   and  chi-­‐square  test.  The  multilevel   software  MLwiN  version  2.01   (Centre  for  Multilevel  Modeling,   Bristol,  UK)  was  used  to  estimate   the  effects  of  individual,  school   and  district  variables  on  DMFT.   The  multilevel  analyzes  were  used   to  determine  the  relative  size  of   the  variance  at  each  level  [21].  As   the  outcome  was  binary,  a  multi-­‐ level  logistic  model  was  used.  Both   the  Marginal  Quasi  Likelihood   (MQL)  first-­‐order  approximation   procedures  and  Predictive  Quasi   Likelihood  (PQL)  second-­‐order  ap-­‐ proximation  procedures  were   used.  However,  because  they  pro-­‐ duced  similar  results,  only  the  re-­‐ sults  based  on  PQL  second-­‐order   procedures  are  reported  and  dis-­‐ cussed  [12].  Five  models  were   specified  for  the  outcome.  First,  a   null  model  (Model  1),  which  took   into  account  the  structure  from   levels  1,  2  and  3.  The  null  model   was  followed  by  Model  2  in  which   variables  from  level  1  were  includ-­‐ ed.  In  Model  3,  it  was  included   variables  from  levels  1,  2  and  3.   There  were  adjustments  in  Models   4  considering:  i)  all  the  school  and   district  variables;  ii)  all  the  varia-­‐ bles  at  the  individual  level,  which   presented  statistically  significant   Table&3.&Multilevel(modeling(of(121yr1old(students,(considering(individual,(school(and(district(hierarchical(levels(in(relation(to(caries(experience( & Model&1&(null&model)& Model&2& Model&3& Model&4& Model&5& Variables& p" value& OR& CI& &&95%& p&value& OR& CI& &&95%& p&value& OR& CI& &&95%& p"" value& OR& CI& &&95%& p&value& OR& CI& &&95%& " LEVEL"1:"INDIVIDUAL" ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( " ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Sociodemografic" ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Ethnic&group& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( White((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Light1(and(dark1skinned( black( ( ( ( 0.096( 1.66( (0.911 3.02)( 0.080( 1.71( (0.94(–( 3.11)( 0.060( 1.82( (0.981 3.38)( ( ( ( ( Yellow( ( ( ( 0.129( 2.21( (0.791 6.17)( 0.109( 2.32( (0.83(–( 6.49)( 0.137( 2.24( (0.771 6.52)( ( ( ( Sex& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Female((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Male( ( ( ( 0.133( 0.77( (0.54(1( 1.08)( 0.107( 0.75( (0.53(–( 1.06)( 0.002*( 0.49( (0.311 0.77)( 0.003*( 0.51( (0.33(–( 0.79)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Behavioural" ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Toothbrushing&frequency& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( 2(or(more/day((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Until(once/day( ( ( ( 0.859( 1.07( (0.521 2.21)( 0.929( 1.03( (0.50(–( 2.14)( 0.685( 0.85( (0.401 1.83)( ( ( ( Flossing& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Yes((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( No(( ( ( ( 0.161( 1.30( (0.901 1.87)( 0.215( 1.26( (0.87(–( 1.82)( 0.189( 0.59( (0.261 1.30)( 0.635( 0.89( (0.56(–( 1.42)( Fluoride&use&(solution,& varnish,&gel)& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Yes((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( No( ( ( ( 0.652( 1.09( (0.761 1.55)( 0.698( 1.07( (0.75(–( 1.54)( 0.087( 0.66( (0.401 1.06)( 0.082( 0.66( (0.41(–( 1.05)( Dental&visits&frequency& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( 2(times(or(more/year((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Once/year( ( ( ( 0.441( 0.87( (0.611 1.24)( 0.510( 0.87( (0.63(–( 1.26)( 0.922( 0.98( (0.681 1.41)( ( ( ( ( No( ( ( ( 0.614( 0.85( (0.461 1.59)( 0.736( 0.90( (0.48(–( 1.68)( 0.285( 0.70( (0.361 1.35)( ( ( ( Sugar&consumption&between& meals& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( No((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Yes( ( ( ( 0.481( 1.22( (0.701 2.11)( 0.474( 1.22( (0.70(–( 2.13)( 0.425( 1.27( (0.711 2.26)( ( ( ( Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   6   association  (SSA)  with  the  out-­‐ come,  and  iii)  interactions  at  level   1,  which  have  made  sense  as  hy-­‐ potheses  and  also  demonstrated   SSA.  Interactions  at  second  and   third  levels  did  not  reach  statistical   significance  and  thus  were  not  in-­‐ cluded  in  this  model.  In  Model  5,  it   was  included  i)  all  variables  at  lev-­‐ el  1  which  showed  significance  in   Model  4  ii)  all  variables  at  level  2   and  3,  and  iii)  all  individual  inter-­‐ actions  that  have  made  sense  as   hypotheses  and  showed  signifi-­‐ cance  in  Model  4.  The  results  were   presented  as  odds  ratios  (OR)  and   their  95%  confidence  intervals  (CI).   For  each  logistic  model  the  intra-­‐ class  correlation  (ICC)  was  calcu-­‐ lated  using  an  approach  described   by  Hox  [21].   Results   Individuals  from  level  1   (n=687),  schools  from  level  2   (n=12)  and  districts  from  level  3   (n=6)  were  evaluated  in  this  study.   Caries  status  according  to  individ-­‐ ual  and  contextual  aspects  is   shown  in  Table  2.     Out  of  the  students,  338   subjects  (49.2%)  were  caries-­‐free   (DMFT=0)  and  349  (50.8%)  had   caries  experience  (DMFT≥1).  The   mean  DMFT  for  students  with  car-­‐ ies  experience  was  2.88±1.79  and   the  general  mean  DMFT  was   1.46±1.92,  being  1.82±2.05  for   public  and  1.08±1.70  for  private   schools.   In  the  exploratory  study,  it   was  observed  an  association  of   caries  experience  with  the  varia-­‐ bles  i)  ethnic  group  (light-­‐  and   dark-­‐skinned  black),  no  flossing,   no  fluoride  use  (solution,  varnish,   gel),  public  dental  access,  lower   individual  socioeconomic  status,   higher  plaque  index,  presence  of   gingivitis  (level  1),  ii)  public  school,   permission  for  candy  or  gum  con-­‐ sumption  (level  2),  and  iii)  worse   living  conditions  in  the  district  so-­‐ cioeconomic  condition  (level  3).  3   presents  the  findings  of  a  multi-­‐ level  logistic  regression  with  the   variable  outcome   DMFT=0⁄DMFT≥1.     Model  1  (Null  Model)   showed  that  the  variation  be-­‐ Table&3&(Continued).&Multilevel(modeling(of(121yr1old(students,(considering(individual,(school(and(district(hierarchical(levels(in(relation(to(caries(experience( Socioeconomic( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Dental&access& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Private((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Public( ( ( ( 0.003*( 1.91( (1.251 2.92)( 0.014*( 1.75( (1.12(–( 2.75)( 0.401( 1.26( (0.731 2.19)( 0.139( 1.47( (0.88(–( 2.46)( Individual&socieconomic& status& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( A1/A2((ref)(highest( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( B1/B2( ( ( ( 0.002*( 1.92( (1.271 2.92)( 0.009*( 1.80( (1.16(–( 2.78)( 0.099( 1.54( (0.921 2.57)( 0.009*( 1.79( (1.15(–( 2.77)( ( C( ( ( ( 0.013*( 2.05( (1.171 3.61)( 0.038*( 1.88( (1.03(–( 3.41)( 0.045*( 2.13( (1.021 4.44)( 0.047*( 1.83( (1.01(–( 3.32)( ( D/E(lowest( ( ( ( 0.368( 1.50( (0.621 3.60)( 0.458( 0.40( (0.57(–( 3.45)( 0.155( 2.47( (0.711 8.59)( 0.402( 1.47( (0.60(–( 3.63)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Clinical( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Plaque&Index&(modified)& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( 0((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( >0(and(<1( ( ( ( 0.151( 1.43( (0.881 2.34)( 0.130( 1.46( (0.89(–( 2.38)( 0.016*( 2.11( (1.151 3.88)( 0.008*( 2.21( (1.23(–( 3.96)( ( ≥1( ( ( ( 0.014*( 2.28( (1.181 4.40)( 0.011*( 2.35( (1.22(–( 4.56)( 0.000*( 4.42( (1.971 9.91)( 0.000*( 4.94( (2.36(–( 10.37)( Gengivitis& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( No((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Yes( ( ( ( 0.482( 1.14( (0.791 1.65)( 0.442( 1.16( (0.80(–( 1.68)( 0.397( 1.18( (0.801 1.73)( ( ( ( Fluorosis& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( No((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Yes( ( ( ( 0.554( 0.90( (0.621 1.29)( 0.565( 0.90( (0.62(–( 1.29)( 0.026*( 3.03( (1.141 8.01)( 0.019*( 3.12( (1.21(–( 8.08)( Stimulated&salivary&flow& rate& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( >0.5(mL/min((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ≤0.5(mL/min( ( ( ( 0.732( 0.94( (0.641 1.37)( 0.729( 0.93( (0.64(–( 1.37)( 0.753( 0.94( (0.631 1.40)( ( ( ( Salivary&buffering& capacity& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( pH>3.9((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( pH≤3.9( ( ( ( 0.484( 0.69( (0.251 1.94)( 0.543( 0.72( (0.25(–( 2.05)( 0.459( 0.65( (0.211 2.03)( ( ( ( Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   7   tween  districts  (7.8%)  was  much   smaller  than  the  variation  at  the   school  level  (92.3%).  The  intraclass   correlation  coefficient  for  schools   was  0.073  and  for  districts  was   0.006.     In  Model  2,  it  was  ob-­‐ served  that  socioeconomic  status   (class  B,  p=0.002  and  class  C,   p=0.013),  public  dental  access   (p=0.003),  and  high  PI  (p=0.014)   were  associated  with  caries  expe-­‐ rience.     In  Model  3,  it  was  verified   that  the  characteristics  of  schools   and  districts  have  not  affected  car-­‐ ies  experience.  Significant  findings   remained  similar  to  those  from   Model  2.   In  Model  4,  the  lack  of  as-­‐ sociation  between  community-­‐ level  variables  (schools  and  dis-­‐ tricts)  and  caries  experience  was   still  maintained.  Socioeconomic   class  C  (p=0.045),  dental  plaque   accumulation  (regular  PI,  p=0.016   and  high  PI,  p=0.000),  and  fluoro-­‐ sis  (p=0.026)  were  significantly  as-­‐ sociated  with  caries  experience.   The  interaction  between  fluorosis   and  dental  biofilm  accumulation   was  negatively  associated  with   caries  experience  [regular  PI  (β=-­‐ 1.265,  p=0.020)  and  high  PI  (β=-­‐ 2.032,  p=0.003)].  In  addition,  the   interactions  between  no  flossing   and  use  of  public  dental  services   were  associated  with  caries  expe-­‐ rience  [β=+0.809;  p=0.007;   OR=3.79;  95%  CI:1.45-­‐9.92]  (Table   3).   All  these  findings  remained   significant  even  after  statistical   adjustment  in  the  final  model   (Model  5).       Discussion   Multilevel  Modeling  is  ap-­‐ propriate  for  analyzing  hierarchical   data  and  provides  an  efficient  way   to  link  the  conventionally  distinct   community  and  individual-­‐level   approaches  [21].    Moreover,  it   could  avoid  a  loss  of  statistical   power  in  the  use  of  variables  from   different  levels,  and  the  risk  of   ecological  fallacy,  which  is  a  kind   of  error  when  only  the  population   level  is  considered  [22].     Table&3&(Continued).&Multilevel(modeling(of(121yr1old(students,(considering(individual,(school(and(district(hierarchical(levels(in(relation(to(caries(experience( ! LEVEL!2:!SCHOOL! ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Type&of&school& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Private((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Public( ( ( ( ( ( ( 0.894( 0.96( (0.54(–( 1.70)( 0.796( 0.92( (0.501 1.69)( 0.948( 1.02( (0.56(–( 1.85)( Oral&health&education&& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Yes((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( No( ( ( ( ( ( ( 0.517( 0.85( (0.53(–( 1.38)( 0.850( 0.95( (0.561 1.61)( 0.676( 0.90( (0.54(–( 1.50)( Permission&for&candy&or&gum& consumption&& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Non(permitted((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Permitted( ( ( ( ( ( ( 0.136( 1.51( (0.88(–( 2.61)( 0.193( 1.48( (0.821 2.68)( 0.228( 1.43( (0.80(–( 2.56)( ! LEVEL!3:!DISTRICT& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Fluoride&concentration&in&water& supply& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ≥0.7(mL/L((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( <0.7(mL/L( ( ( ( ( ( ( 0.789( 1.06( (0.69(–( 1.64)( 0.609( 1.14( (0.691 1.89)( 0.624( 1.13( (0.69(–( 1.88)( Socioeconomic&position& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Better(living(conditions((ref)( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Worse(living(conditions( ( ( ( ( ( ( 0.623( 1.14( (0.68(–( 1.89)( 0.575( 1.16( (0.691 1.97)( 0.562( 1.17( (0.69(–( 1.98)( & Interactions& ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Male(sex(and(no(fluoride(use( ( ( ( ( ( ( ( ( ( 0.002*( ( ( 0.003*( ( ( PI(score(1(and(fluorosis( ( ( ( ( ( ( ( ( ( 0.020*( ( ( 0.018*( ( ( PI(score(2(and(fluorosis( ( ( ( ( ( ( ( ( ( 0.003*( ( ( 0.002*( ( ( Class(B(and(no(flossing(use( ( ( ( ( ( ( ( ( ( 0.112( ( ( ( ( ( Class(C(and(no(flossing(use( ( ( ( ( ( ( ( ( ( 0.826( ( ( ( ( ( Classes(D/E(and(no(flossing(( ( ( ( ( ( ( ( ( ( 0.269( ( ( ( ( ( No(flossing(and(public(access( ( ( ( ( ( ( ( ( ( 0.007*( ( ( 0.028*( ( ( ( σ2(District((( ( 0.020( ( ( ( 0.075(( ( ( ( 0.020( ( ( ( 0.043( ( ( ( 0.045( ( ( σ2(School(( 0.238( ( ( 0( ( ( 0( ( ( 0( ( ( 0( ( ( ρ(District(( 0.006( ( ( 0.022( ( ( 0.006( ( ( 0.013( ( ( 0.013( ( ( ρ(School(( 0.073( ( ( 0.022( ( ( 0.006( ( ( 0.013( ( ( 0.013( ( ( * Association with dental caries outcome. Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   8   The  mean  DMFT  in  the   study  population  was  1.46,  con-­‐ sidered  low  by  WHO  standards   and  similar  to  the  levels  observed   in  Europe  [23],  where  the  DMFT   indexes  among  12-­‐year-­‐old  chil-­‐ dren  range  from  1.2  to  2.6.  How-­‐ ever,  the  mean  DMFT  for  students   with  caries  experience  was  2.88,   which  demonstrates  higher  needs   for  treatment  in  this  population,   evidencing  the  dental  caries  polar-­‐ ization  in  the  present  study.     Regarding  variables  from   the  individual  level,  ethnicity   seems  to  be  less  relevant  than  so-­‐ cioeconomic  status  to  determine   caries  outcome  [24].  In  Brazil,  eth-­‐ nic  differences  in  dental  health   may  be  further  due  to  socioeco-­‐ nomic  status  and  access  to  ser-­‐ vices  than  to  biological  back-­‐ ground  [25].   In  this  study,  male  sex  was   negatively  associated  with  caries   experience,  differently  from  the   findings  by  Aida  et  al.  [7],  who  did   not  observe  any  association  be-­‐ tween  gender  and  dental  caries.   Nevertheless,  our  results  corrobo-­‐ rate  the  study  by  Antunes  et  al.   [3],  who  argued  that  an  early   chronology  of  permanent  tooth   eruption  in  women  might  deter-­‐ mine  the  higher  risk  to  tooth  de-­‐ cay  for  girls  among  Brazilians.   Higher  access  to  dental  care  by   girls  may  lead  to  an  increase  in   DMFT  index,  especially  regarding   the  number  of  filled  teeth,  as  a   result  of  overtreatment.  The  in-­‐ teraction  between  male  sex  and   no  use  of  fluoride  was  positively   associated  with  caries  experience.   This  is  due  to  the  fact  that  boys,  in   general,  have  a  lower  care  index,   which  reflects  their  decreased   concern  about  health  and  aesthet-­‐ ics  [6].  In  relation  to  caries-­‐ inhibiting  effect  of  fluoride   sources,  Cochrane  systematic  re-­‐ views  found  that  fluoride  rinses   are  responsible  for  26%,  fluoride   gel  for  28%,  and  fluoride  varnish   for  46%  of  the  reduction  in  the   DMFS  index  [26],  demonstrating   the  relevance  of  these  topical   methods.  Lower  individual  socio-­‐ economic  status  was  associated   with  caries  experience.  Indeed,   low  social  class  has  been  associat-­‐ ed  with  high  DMFT  [27,12].  The   higher  socioeconomic  positions   have  shown  the  better  health  sta-­‐ tus  measurements.     The  present  study  also   showed  that  higher  degrees  of   plaque  index  were  positively  asso-­‐ ciated  with  caries  outcome.  Caries   lesions  develop  where  oral  bio-­‐ films  are  allowed  to  mature  and   remain  on  teeth  for  long  periods   [9].     Fluorosis  presence  was  also  asso-­‐ ciated  with  caries  experience,  in   accordance  to  other  studies  re-­‐ sults  [28,  29].  Severe  fluorosis  has   been  reported  to  increase  suscep-­‐ tibility  to  dental  caries  by  hypo-­‐ mineralization  [30].  However,   there  has  been  some  controversy   on  whether  dental  fluorosis  in-­‐ creases,  decreases,  or  has  no  ef-­‐ fect  on  caries  experience  [31].   When  plaque  index  and  fluorosis   were  analyzed  in  interaction,  fluo-­‐ rosis  seemed  to  protect  against   caries  only  in  the  presence  of  den-­‐ tal  biofilm.  The  interaction  be-­‐ tween  fluorosis  and  dental  biofilm   with  caries  experience  had  never   been  carried  out  in  vivo.  One  hy-­‐ pothesis  is  that  fluorotic  teeth   could  release  some  fluoride  ions   to  the  dental  biofilm  in  a  low  pH   condition  which,  consequently,   reduce  demineralization  [32].  On   the  other  hand,  enamel  changes  in   experimental  teeth  with  signs  of   fluorosis  that  had  been  exposed  to   plaque  accumulation  were  exam-­‐ ined  in  vitro  and  specimens   showed  signs  of  surface  deminer-­‐ alization  [33].   Other  significant  interac-­‐ tion  in  the  present  study  was   shown  between  no  flossing  and   use  of  public  dental  services,   which  means  an  increase  in  odds   of  caries  experience  if  these  two   socioeconomic-­‐dependent  varia-­‐ bles  act  together.  This  finding   points  at  a  role  of  health  services   on  improvement's  individual  be-­‐ havior,  which  has  also  been  re-­‐ ported  for  other  chronic  diseases   [33].  In  relation  to  school  level,  it   was  not  found  a  SSA  between  the   type  of  school  and  oral  health  ed-­‐ ucation  with  caries  experience.   The  type  of  school  has  been  con-­‐ sidered  a  risk  indicator  of  the  dis-­‐ ease  in  a  Brazilian  study  [27]  and   not  in  another  one  [13].  However,   public  school  enrollment  was  de-­‐ terminant  in  having  one  or  more   untreated  decayed  permanent   teeth  [3].  The  lack  of  association   between  oral  health  education  in   schools  and  dental  caries  may  be   due  to  the  fact  that  programs  con-­‐ sidering  specific  individual  needs   are  similar  in  both  types  of  school.     With  regard  to  district  lev-­‐ el,  an  association  between  fluo-­‐ Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   9   ride  concentration  in  water  supply   and  caries  experience  was  not   found.  Curitiba  has  a  water  fluori-­‐ dation  scheme  which  covers  the   whole  population  for  decades.   This  homogeneity  in  the  fluoride   distribution  could  be  the  reason   for  the  lack  of  association  with   caries  experience.  Besides,  the  6   evaluated  districts  present  a  ho-­‐ mogeneous  income  distribution,   which  might  contribute  for  the   lack  of  influence  in  caries  experi-­‐ ence.     The  results  from  a  few   multilevel  studies  on  dental  caries,   taking  into  account  two  levels,   have  been  reported.  Neighbor-­‐ hood  aspects  influenced  oral   health  [34]  and  dental  caries  expe-­‐ rience  in  Brazilians  [12].  Moreo-­‐ ver,  contextual  and  individual  so-­‐ ciodemographic  characteristics   influenced  dental  caries  experi-­‐ ence  in  12-­‐year-­‐old  schoolchildren   from  Brazil  when  using  a  conven-­‐ tional  multivariate  logistic  regres-­‐ sion  (1).  Otherwise,  Aida  [7]   showed  that  90.8%  of  variance  in   dmft  index  among  Japanese  3-­‐ year-­‐old  children  occurred  at  the   individual  level  and  9.2%  of  the   variance  occurred  at  the  commu-­‐ nity  level.     The  main  findings  of  this   study  were  that  individual  varia-­‐ bles  had  a  highly  significant  asso-­‐ ciation  with  caries  experience   even  in  the  presence  of  school  and   district  levels  (contextual  varia-­‐ bles).     The  variables  which  de-­‐ termine  the  risk  indicators  for  car-­‐ ies  development  varies  significant-­‐ ly  for  different  groups.  The  moni-­‐ toring  of  contrasts  in  dental  health   outcomes  is  relevant  for  pro-­‐ gramming  socially  appropriate  in-­‐ terventions.  The  analysis  of  differ-­‐ ent  risk  indicators,  which  influence   dental  caries,  and  its  early  identifi-­‐ cation,  is  relevant  for  planning   suitable  interventions  for  target   population  groups  whom  could   present  higher  levels  of  disease,   re-­‐orienting  attention  on  dental   prevention  and  reducing  costs  in   dental  treatments  offered  by  the   health  systems.     Overall  effective  health  in-­‐ terventions  might  be  more  effec-­‐ tive  in  that  part  of  a  population   which  concentrates  disease  [7].   On  the  other  hand,  groups  of  low-­‐ er  caries  risk  should  not  be  under-­‐ estimated,  considering  that  risk   indicators  may  not  be  easily  rec-­‐ ognized  [35].  Maybe,  a  good  tool   to  preserve  equity  and  universality   principles  for  dental  caries  should   be  the  combination  of  both  polari-­‐ zation  and  global  attention  ap-­‐ proaches.           The  Multilevel  Analysis  al-­‐ lows  the  evaluation  of  different   risk  indicators  from  distinct  hierar-­‐ chical  levels,  which  makes  this  ap-­‐ proach  valuable  for  the  integration   of  variables  influencing  the  dis-­‐ ease  outcome,  especially  for  com-­‐ plex  diseases,  such  as  dental  car-­‐ ies.  It  was  observed  that  individual   factors,  such  as  female  sex,  the   interaction  of  male  sex  with  no   fluoride  use,  lower  socioeconomic   status,  dental  plaque  accumula-­‐ tion,  fluorosis,  and  the  integration   of  no  flossing  with  public  health   service  use,  even  in  the  presence   of  contextual  variables  (schools   and  districts),  were  associated   with  caries  experience  in  the  study   population.     Acknowledgements   We  would  like  to  thank  the   Municipal  Health  Secretary  of  Cu-­‐ ritiba  for  their  cooperation  to  this   study.  This  work  was  based  on  a   thesis  submitted  by  the  first  au-­‐ thor  to  the  Catholic  University  of   Paraná  (PUCPR),  Brazil,  in  partial   fulfillment  of  the  requirements  for   the  Doctor's  Degree  in  Health  Sci-­‐ ences.  The  first  and  fourth  authors   were  supported  by  scholarship   from  the  Brazilian  Government  –   CAPES  (Coordination  for  the  Im-­‐ provement  of  Higher  Level  –  Edu-­‐ cation  –  Personnel).  We  would  al-­‐ so  like  to  thank  Kamilla  Gabriella   dos  Santos  Medeiros  and  Carlos   Alberto  Cordeiro-­‐Jr,  for  the  contri-­‐ bution  with  the  sample  collection.   References   1. Brasil.  Ministery  of  Health.   Oral  Health  Brasil  2010  (SB   Brasil  2010).  National  Oral   Health  Research:  main  re-­‐ sults.  2012.  available  at:   http://dab.saude.gov.br/C NSB/sbbrasil/  arquivos/   proje-­‐ je-­‐ to_sb2010_relatorio_final. pdf  (last  acess  12  February   2015).   2. Decline,  polarization,  ine-­‐ quality  and  social  exclu-­‐ sion.  Narvai  PC,  Frazao  P,   Roncalli  AG,  Antunes  JL.   Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   10   Dental  caries  in  brazil:  Rev   Panam  Salud  Publica.  2006;   19:385-­‐93.  PMID:   16968593.   3. Ethnic  disparities  in  the   prevalence  of  dental  caries   and  restorative  dental   treatment  in  brazilian  chil-­‐ dren.  Antunes  JL,  Pegoretti   T,  de  Andrade  FP,  Junquei-­‐ ra  SR,  Frazao  P,  Narvai  PC.   Int  Dent  J.  2003;  53:7-­‐12.   PMID:  12653333.   4. Prevalence  and  polariza-­‐ tion  of  dental  caries  among   young,  healthy  adults:   Cross-­‐sectional  epidemio-­‐ logical  study.  Tanner  T1,   Kämppi  A,  Päkkilä  J,  Pat-­‐ inen  P,  Rosberg  J,  Kar-­‐ jalainen  K.  Acta  Odontol   Scand.  2013;  71(6):1436-­‐ 42.  doi:   10.3109/00016357.2013.7 67932.  Epub  2013  Apr  30.   PMID:  23627898.   5. Factors  associated  with   dental  caries  experience   and  oral  health  status   among  New  South  Wales   adolescents.  Skinner  J1,   Johnson  G,  Blinkhorn  A,   Byun  R.  Aust  N  Z  J  Public   Health.  2014:  Oct;   38(5):485-­‐9.  doi:   10.1111/1753-­‐6405.12245.   Epub  2014  Aug  28.  PMID:   25169434.   6. The  relation  between  fami-­‐ ly  socioeconomic  trajecto-­‐ ries  from  childhood  to  ado-­‐ lescence  and  dental  caries   and  associated  oral  behav-­‐ iours.  Peres  MA,  Peres  KG,   de  Barros  AJ,  Victora  CG.  J   Epidemiol  Community   Health.  2007;  61:141-­‐5.   PMID:  17234873.   7. Contributions  of  social  con-­‐ text  to  inequality  in  dental   caries:  A  multilevel  analysis   of  japanese  3-­‐year-­‐old   children.  Aida  J,  Ando  Y,   Oosaka  M,  Niimi  K,  Morita   M.  Community  Dent  Oral   Epidemiol.  2008;    36:149-­‐ 56.  PMID:  18333879.   8. Risk  indicators  of  dental   caries  in  5-­‐year-­‐old  brazili-­‐ an  children.  Cortellazzi  KL,   Pereira  SM,  Tagliaferro  EP,   Tengan  C,  Ambrosano  GM,   Meneghim  Mde  C,  et  al.   Community  Dent  Health.   2008;  25:253-­‐6.  PMID:   21367823.   9. Dental  Caries.  Selwitz  RH,   Ismail  AI,  Pitts  NB.  Lancet.   2007;  369:51-­‐9.  PMID:   17208642.   10. Area  characteristics,  indi-­‐ vidual-­‐level  socioeconomic   indicators,  and  smoking  in   young  adults:  The  coronary   artery  disease  risk  devel-­‐ opment  in  young  adults   study.  Diez  Roux  AV,  Mer-­‐ kin  SS,  Hannan  P,  Jacobs   DR,  Kiefe  CI.  Am  J  Epidemi-­‐ ol.  2003;  157:315-­‐26.   PMID:  12578802.   11. Measuring  inequalities  in   the  distribution  of  dental   caries.  Antunes  JL,  Narvai   PC,  Nugent  ZJ.    Community   Dent  Oral  Epidemiol.  2004;   32:41-­‐8.  PMID:  14961839.   12. The  potential  impact  of   neighborhood  empower-­‐ ment  on  dental  caries   among  adolescents.  Pat-­‐ tussi  MP,  Hardy  R,  Sheiham   A.  Community  Dent  Oral   Epidemiol.  2006;  34:344-­‐ 50.  PMID:  16948673.   13. Women's  health:  beyond   reproductive  years.  Vibha,   Laskar  AR.  Indian  J  Public   Health.  2011;  55(4):247-­‐51.   PMID:  22298132.   14. Influence  of  initial  carious   lesions  on  oral  health   planning.  Assaf  AV,  Tengan   C,  Tagliaferro  EPS,   Meneghim  ZP,  Meneghim   MC,  Pereira  AC,  Ambro-­‐ sano  GMB.  Braz  J  Oral  Sci.   2006;  5(19):1217-­‐1222.     15. WHO.  Oral  health  survey.   Basic  methods.  Fourth  edi-­‐ tion.  Geneva:  World  Health   Organization.  1997;66.   16. Risk  assessment  in  dental   caries.  In:  Risk  assessment   in  dentistry.  Fejerskov  O,   Manji  F.  Bader  JD.  Chapel   Hill:  University  of  North  Ca-­‐ rolina  Dental  Ecology.   1990;  215–217.     17. The  gingival  index,  the   plaque  index  and  the  re-­‐ tention  index  systems.  Loe   H.  J  Periodontol.  1967;  38   suppl:610-­‐6.  PMID:   5237684.   Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   11   18. Effects  of  antidepressants   and  benzodiazepines  on   stimulated  salivary  flow   rate  and  biochemistry   composition  of  the  saliva.   de  Almeida  Pdel  V,  Gregio   AM,  Brancher  JA,  Ignacio   SA,  Machado  MA,  de  Lima   AA,  et  al.  Oral  Surg  Oral   Med  Oral  Pathol  Oral  Radi-­‐ ol  Endod.  2008;  106:58-­‐65.   PMID:  18329910.   19. Clinical  investigations  of   the  salivary  buffering  ac-­‐ tion.  Ericsson  Y.  Acta   Odontol  Scand.  1959;   17:133-­‐65.     20. Social  inequalities  and  vac-­‐ cination  coverage:  Utiliza-­‐ tion  of  household  surveys.   Moraes  JCD,  Ribeiro   MCSdA.  Braz  J  Epidemiol.   2008;  11:113-­‐24.     21. Introduction  to  multilevel   analysis.  Hox  J:  Multilevel   analysis,  techniques  and   applications.  Hox  J.  Mah-­‐ wah:  Lawrence  Erlbaum   Associates.  2002;  1-­‐36.     22. The  social  determinants  of   oral  health:  New  ap-­‐ proaches  to  conceptualiz-­‐ ing  and  researching  com-­‐ plex  causal  networks.  New-­‐ ton  JT,  Bower  EJ.  Commu-­‐ nity  Dent  Oral  Epidemiol.   2005;  33:25-­‐34.  PMID:   15642044.   23. The  world  oral  health  re-­‐ port  2003:  Continuous  im-­‐ provement  of  oral  health  in   the  21st  century-­‐-­‐the  ap-­‐ proach  of  the  who  global   oral  health  programme.   Petersen  PE.  Community   Dent  Oral  Epidemiol.  2003  ;   31  suppl  1:3-­‐23.  PMID:   15015736.   24. Inequalities  in  oral  health:   A  review  of  the  evidence   and  recommendations  for   action.  Watt  R,  Sheiham  A.   Br  Dent  J.  1999;  187:6-­‐12.   PMID:  10452185.   25. Multilevel  assessment  of   determinants  of  dental  car-­‐ ies  experience  in  Brazil.  An-­‐ tunes  JL,  Peres  MA,  de   Campos  Mello  TR,  Wald-­‐ man  EA.  Community  Dent   Oral  Epidemiol.  2006;   34:146-­‐52.  PMID:   16515679.   26. Evidence-­‐based  effective-­‐ ness  of  topical  fluorides.   Marinho  VC.  Adv  Dent  Res.   2008;  20:3-­‐7.  PMID:   18694869.   27. Factors  associated  with   dental  caries  in  Brazilian   children:  a  multilevel  ap-­‐ proach.  Martins  MT,  Sar-­‐ denberg  F,  Abreu  MH,  Vale   MP,  Paiva  SM,  Pordeus  IA.   Community  Dent  Oral  Epi-­‐ demiol.  2014;  42:  289–299.   PMID:  24354434.   28. The  relationship  between   dental  caries  and  dental   fluorosis  in  areas  with   moderate-­‐  and  high-­‐ fluoride  drinking  water  in   ethiopia.  Wondwossen  F,   Astrom  AN,  Bjorvatn  K,   Bardsen  A.  Community   Dent  Oral  Epidemiol.  2004;   32:337-­‐44.  PMID:   15341618.   29. Prevalence  and  severity  of   dental  caries  in  adoles-­‐ cents  aged  12  and  15  living   in  communities  with  vari-­‐ ous  fluoride  concentra-­‐ tions.  Pontigo-­‐Loyola  AP,   Medina-­‐Solis  CE,  Borges-­‐ Yanez  SA,  Patino-­‐Marin  N,   Islas-­‐Marquez  A,  Maupome   G.  J  Public  Health  Dent.   2007;  67:8-­‐13.  PMID:   17436973.   30. Caries  and  dental  fluorosis   in  a  western  saharan  popu-­‐ lation  of  refugee  children.   Almerich-­‐Silla  JM,  Montiel-­‐ Company  JM,  Ruiz-­‐Miravet   A.  Eur  J  Oral  Sci.  2008;   116:512-­‐7.  PMID:   19049520.   31. Dental  fluorosis  increases   caries  risk.  Cunha-­‐Cruz  J,   Nadanovsky  P.  J  Evid  Based   Dent  Pract.  2005;  5:170-­‐1.   PMID:  26092033.   32. Biochemical  composition   and  cariogenicity  of  dental   plaque  formed  in  the  pres-­‐ ence  of  sucrose  or  glucose   and  fructose.  Cury  JA,   Rebelo  MA,  Del  Bel  Cury   AA,  Derbyshire  MT,  Tab-­‐ choury  CP.  Caries  Res.   2000;  34:491-­‐7.  PMID:   11093024.   33. A  light  and  scanning  elec-­‐ tron  microscopic  study  of   enamel  decalcification  in   children  living  in  a  water-­‐ fluoridated  area.  Thylstrup   A,  Boyar  RM,  Holmen  L,   Mul$level  modeling  for  dental  caries  among  adolescents  in  a  Brazilian  large  city     Vol  4,  No  1  (2016)        DOI  10.5195/d3000.2016.58    http://dentistry3000.pitt.edu   12   Bowden  GH.  J  Dent  Res.   1990;  69:1626-­‐33.  PMID:   2212207.   34. The  independent  contribu-­‐ tion  of  neighborhood  dis-­‐ advantage  and  individual-­‐ level  socioeconomic  posi-­‐ tion  to  self-­‐reported  oral   health:  A  multilevel  analy-­‐ sis.  Turrell  G,  Sanders  AE,   Slade  GD,  Spencer  AJ,  Mar-­‐ cenes  W.  Community  Dent   Oral  Epidemiol.  2007;   35:195-­‐206.  PMID:   17518966.   35. Life  course  dental  caries   determinants  and  predic-­‐ tors  in  children  aged  12   years:  A  population-­‐based   birth  cohort.  Peres  MA,   Barros  AJ,  Peres  KG,  Araujo   CL,  Menezes  AM.  Commu-­‐ nity  Dent  Oral  Epidemiol.   2009;  37:123-­‐33.  PMID:   19250294.