This is an elec onic ep in o he o iginal a icle.
This ep in may di e om he o iginal in pagina ion and ypog aphic de ail.
Au ho (s):
Ti le:
Yea :
Ve sion:
Please ci e he o iginal e sion:
All ma e ial supplied ia JYX is p o ec ed by copy igh and o he in ellec ual p ope y igh s, and
duplica ion o sale o all o pa o any o he eposi o y collec ions is no pe mi ed, excep ha
ma e ial may be duplica ed by you o you esea ch use o educa ional pu poses in elec onic o
p in o m. You mus ob ain pe mission o any o he use. Elec onic o p in copies may no be
o e ed, whe he o sale o o he wise o anyone who is no an au ho ised use .
Hea ing loss and use o heal h se ices : a popula ion-based c oss-sec ional s udy
among Finnish olde adul s
Mikkola, Tuija; Polku, Hannele; Sainio, Päi i; Koponen, Päi ikki; Koskinen, Seppo;
Viljanen, Anne
Mikkola, T., Polku, H., Sainio, P., Koponen, P., Koskinen, S., & Viljanen, A. (2016).
Hea ing loss and use o heal h se ices : a popula ion-based c oss-sec ional s udy
among Finnish olde adul s. BMC Ge ia ics, 16(1), A icle 182.
h ps://doi.o g/10.1186/s12877-016-0356-5
2016
RESEARCH ARTICLE Open Access
Hea ing loss and use o heal h se ices: a
popula ion-based c oss-sec ional s udy
among Finnish olde adul s
Tuija M. Mikkola
1,2*
, Hannele Polku
1
, Päi i Sainio
3
, Päi ikki Koponen
3
, Seppo Koskinen
3
and Anne Viljanen
1
Abs ac
Backg ound: Olde adul s wi h hea ing di icul ies ace p oblems o communica ion which may lead o unde use
o heal h se ices. This s udy in es iga ed he associa ion o hea ing loss and sel - epo ed hea ing di icul y wi h
he use o heal h se ices and unme heal h ca e needs in olde adul s.
Me hods: Da a on pe sons aged 65 and olde (n= 2144) d awn om a popula ion-based s udy, Heal h 2000, we e
analyzed. Hea ing loss was de e mined wi h sc eening audiome y (n= 1680). S uc u ed ace- o- ace in e iews
we e used o assess sel - epo ed hea ing di icul y (n= 1962), use o heal h se ices (physician and nu se isi s,
heal h examina ions, men al heal h se ices, physical he apy, heal h p omo ion g oups, ision es , hea ing es ,
mammog aphy, PSA es ) and pe cei ed unme heal h ca e needs. Mul i a iable logis ic eg ession analyses we e
used.
Resul s: A e adjus ing o socio-economic and heal h- ela ed con ounde s, pe sons wi h hea ing loss (hea ing
le el o be e ea 0.5–2 kHz > 40 dB) we e mo e likely o ha e used men al heal h se ices han hose wi h non-
impai ed hea ing (OR = 3.2, 95 % CI 1.3–7.9). Sel - epo ed hea ing di icul y was also associa ed wi h highe odds
o men al heal h se ice use (OR = 2.1 95 % CI 1.2–3.5). Hea ing was no associa ed wi h use o he o he heal h
se ices s udied, excep p esen ing o a hea ing es . Pe sons wi h sel - epo ed hea ing di icul y we e mo e likely
o pe cei e unme heal h ca e needs han hose wi hou hea ing di icul y (OR = 1.7, 95 % CI 1.4–2.1).
Conclusions: Olde adul s wi h hea ing loss o sel - epo ed hea ing di icul y a e as likely o use mos heal h
se ices as hose wi hou hea ing loss. Howe e , sel - epo ed hea ing di icul y is associa ed wi h expe iencing
unme heal h ca e needs. Adequa e heal h se ices should be ensu ed o olde adul s wi h hea ing di icul ies.
Keywo ds: Hea ing loss, Heal h se ices, Heal h se ices needs and demand, Pu e- one audiome y, Aged, Aging
Backg ound
Hea ing loss is ela i ely a e in young adul s bu i s
p e alence ises exponen ially in olde popula ion [1]
due o age- ela ed senso ineu al hea ing loss [2]. I
has been epo ed ha in he Uni ed S a es mo e
han 50 % o adul s aged 70 and olde and o e 80 %
o pe sons aged 80 and olde ha e hea ing loss [1, 3].
Due o he aging o he popula ion, he numbe o
pe sons wi h hea ing loss is expec ed o ise subs an-
ially in he coming yea s.
The diagnosis and ea men o hea ing loss may equi e
se e al isi s o di e en heal h ca e p o essionals, such as
a gene al p ac i ione , o o hinola yngologis and audiolo-
gis . On he o he hand, o a ma ked p opo ion o hose
wi h hea ing loss, he condi ion emains undiagnosed, and
e en a e diagnosis does no ecei e ehabili a ion [4].
Ce ain diseases, such as ca dio ascula diseases [5] and
diabe es [6], may inc ease he isk o de eloping hea ing
loss. The ea men o hese diseases, al hough no a esul
o hea ing loss pe se, may lead o highe use o heal h
se ices in his g oup. Fu he , hea ing loss may also indi -
ec ly inc ease he need o o he heal h se ices, as hea ing
loss ele a es he isk o de eloping cogni i e impai men
[7], dep ession [8] and disabili y in ac i i ies o daily li ing
* Co espondence: [email p o ec ed]
1
Ge on ology Resea ch Cen e and Depa men o Heal h Sciences, Uni e si y
o Jy askyla, P.O. Box 35Vi eca, Jy askyla 40014, Finland
2
Folkhälsan Resea ch Cen e , Helsinki, Finland
Full lis o au ho in o ma ion is a ailable a he end o he a icle
© The Au ho (s). 2016 Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and
ep oduc ion in any medium, p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o
he C ea i e Commons license, and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e
(h p://c ea i ecommons.o g/publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed.
Mikkola e al. BMC Ge ia ics (2016) 16:182
DOI 10.1186/s12877-016-0356-5
[9]. The associa ion be ween hea ing loss and use o
heal h se ices may also be con ounded by sociodemo-
g aphic ac o s. Highe age [10, 11] and low socioeco-
nomic s a us [5, 11–13] a e isk ac o s o bo h hea ing
loss and poo heal h, which in u n inc ease he need o
heal h se ices. Male gende is ano he isk ac o o
hea ing loss [11] and gende has been shown o ha e com-
plex e ec s on heal h and heal h ca e use o e he li e-
cou se [14, 15].
On he o he hand, pe sons wi h hea ing loss may
a oid con ac s wi h heal h ca e pe sonnel owing o com-
munica ion p oblems. I has been shown ha olde
pe sons su e ing om hea ing p oblems a e socially less
ac i e, i.e. pa icipa e in a ious social ac i i ies less o en
and mee o he people less o en han pe sons wi hou
hea ing p oblems [16, 17], mos p obably due o he
dis ess caused by di icul ies in communica ion [18].
Simila dis ess and a oidance beha io may also apply
o isi s o heal h ca e p ac i ione s, leading o lowe
heal h se ice use by pe sons wi h hea ing loss.
Li le is known abou whe he olde adul s wi h hea -
ing loss use heal h se ices less o mo e han o he s.
Only one hea ing loss s udy has ocused on olde adul s
(who u ned 65 du ing he yea o he s udy) [19]. The
au ho s ound ha hea ing loss was associa ed wi h
inc eased odds o making a isi o a heal h ca e p ac i-
ione . The esul emained signi ican e en a e exclud-
ing hea ing- ela ed isi s, sugges ing ha pe sons wi h
hea ing loss also use mo e heal h se ices o he han
hose needed speci ically o ea ing hei hea ing loss.
Al hough he au ho s adjus ed o ch onic condi ions,
hey did no adjus he analyses o socioeconomic
ac o s which may con ound he associa ion be ween
hea ing and use o heal h se ices. P e ious s udies wi h
a wide age ange, om adolescence o young adul hood
o old age, ha e ound a highe han a e age u iliza ion
o physician se ices among pe sons wi h hea ing loss;
howe e , hese s udies did no con ol o socioeco-
nomic s a us [20] o diseases ha may con ound he
associa ion be ween hea ing and use o heal h se ices
[21, 22]. One p e ious s udy on adul s below age 65
adjus ed o socioeconomic and heal h- ela ed con-
ounde s and ound signi ican ly mo e con ac s wi h
heal h ca e among pe sons wi h hea ing loss han
among pe sons wi h no mal hea ing [23]. Howe e ,
when hea ing- ela ed isi s we e excluded he di e -
ence disappea ed. This sugges s ha , apa om hose
ela ed di ec ly o hei hea ing loss, pe sons wi h
hea ing loss do no use mo e heal h se ices han
o he people. In summa y, i appea s ha he p e i-
ous s udies on he associa ions be ween hea ing loss
and use o heal h se ices ha e con olled o con-
ounding ac o s o a a ying ex en and ha he e-
sul s a e con adic o y.
One explana ion o he di e ences in he esul s o
he p e ious s udies may lie in he use o ei he objec i e
o subjec i e me hods o measu ing hea ing. I has been
shown ha objec i e and subjec i e measu es o hea ing
do no co ela e e y highly and hey co e pa ly di e -
en aspec s o hea ing pe o mance [24]. Pu e- one audi-
ome y, mos widely used objec i e assessmen o
hea ing, measu es mos ly pe iphe al senso y unc ioning
in op imal condi ions whe eas subjec i e measu es a e
in luenced by pe sonal ac o s and e e yday acous ic
en i onmen o a pe son. The e o e, i is impo an o
s udy bo h objec i e and subjec i e measu es o hea ing
simul aneously.
Despi e adjus ing use o heal h se ices o he needs
a iables (i.e. diseases) pe sons may a y in pe cei ed
access o, and adequacy o , heal h se ices. Pe sons wi h
hea ing p oblems may be dissa is ied wi h he quali y o
heal h se ices due o communica ion p oblems wi h
heal h ca e p o ide s and hence, hey may expe ience
unme heal h ca e needs. Consequen ly, in addi ion o
absolu e use o heal h se ices, pe cei ed unme need
should be s udied. To he bes o ou knowledge, only
one s udy has in es iga ed he associa ion be ween hea -
ing p oblems and pe cei ed unme heal h ca e needs.
The esul s showed ha , among Canadians om 12 yea s
o old age, pe sons who epo ed hea ing p oblems also
epo ed mo e unme heal h ca e needs han hose no
epo ing hea ing p oblems [20].
In p e ious s udies in es iga ing he associa ions be-
ween hea ing and heal h se ices, he pa icipan s ha e
mainly been ei he younge adul s [23] o d awn om a
wide age ange [20–22], making i ha d o d aw conclu-
sions conce ning olde adul s. Fu he , p e ious s udies
ha e no explo ed objec i ely measu ed and sel - epo ed
hea ing simul aneously. The e o e, mo e knowledge is
needed on whe he olde adul s wi h hea ing loss use
heal h se ices mo e o less han hose wi hou hea ing
loss and whe he hey pe cei e ha heal h se ices mee
hei needs. The pu pose o he p esen s udy was o
explo e he associa ions o hea ing loss, de e mined by
bo h audiome y and sel - epo , wi h he use o a ious
heal h se ices and pe cei ed unme heal h ca e needs
among Finnish adul s aged 65 and olde .
Me hods
The Heal h 2000 S udy, conduc ed in 2000–2001, is a
c oss-sec ional su ey comp ising a comp ehensi e
heal h in e iew and a de ailed heal h examina ion in a
popula ion-based la ge sample o Finnish adul s [25].
The sampling was ca ied ou using wo-s age s a i ied
clus e sampling. The sample consis ed o 8028 adul s
aged 30 and olde , and was ep esen a i e o he same-
age Finnish gene al popula ion. In he oldes age g oup,
80 and olde , he sampling p obabili y was doubled o
Mikkola e al. BMC Ge ia ics (2016) 16:182 Page 2 o 11
ensu e a su icien numbe o he adul s in his g oup
we e included in he sample. The p esen analysis was
con ined o adul s aged 65 and olde ( ange 65 o
99 yea s, N= 2 144). The sample pa icipa ion a e o
hose aged 65 and o e was 89 % o he in e iews, and
83 % o he heal h examina ion.
Compu e -assis ed pe sonal heal h in e iews (ques-
ions a ailable a h p://www. e eys2000. i/ o ms.h ml)
we e ca ied ou a he pa icipan s’homes o in esi-
den ial ins i u ions. In he in e iews, he in e iewe
ead he ques ions om he lap op sc een and en e ed
he answe s di ec ly in o he lap op. Heal h examina-
ions we e pe o med in public heal h cen e s o in em-
po a y examina ion acili ies. I he pa icipan was
unable o a el o he heal h cen e o empo a y acil-
i y, an abb e ia ed examina ion was ca ied ou in he
pa icipan ’s home [25].
Assessmen o hea ing
Pu e- one ai -conduc ion hea ing h esholds we e
assessed o bo h ea s wi hou a hea ing aid using a
sc eening audiome e (Mic oma e 304, Madsen Elec-
onics) a he equencies o 0.5, 1 and 2 kHz in a
silen oom [25]. Headphones wi h padded ea pieces
we e used o minimize any en i onmen al noise. The
lowes signal in ensi y was 5 dB. The es s a ed
om he be e -hea ing ea o he igh ea (i he
pa icipan epo ed no di e ence be ween he ea s)
a a equency o 1 kHz (a 25 dB, o mo e o olde
people and hose who seemed ha d o hea ing). In en-
si y was hen educed in dec emen s o 10 dB un il
he pa icipan could no longe hea he signal. The
in ensi y was hen inc eased by in inc emen s o 5 dB
un il he pa icipan was able o hea he signal. The
lowes in ensi y ha he pa icipan could hea was
de e mined as he hea ing h eshold a 1 kHz. Nex ,
hea ing h esholds a equencies o 2 and 0.5 kHz
we e simila ly assessed, a e which he o he ea was
assessed. I he pa icipan could no hea a he in-
ensi y o 90 dB, 99 dB was ma ked as he hea ing
h eshold. Be e ea hea ing le el (BEHL
0.5–2kHz
)was
calcula ed as he mean alue o e he measu ed
equencies. Tes - e es epea abili y was excellen
(in aclass co ela ion coe icien = 0.97) [25]. Hea ing
le el was dicho omized by ca ego izing hose wi h
BEHL
0.5–2kHz
> 40 dB as ha ing hea ing loss and hose
wi h BEHL
0.5–2kHz
≤40 dB as ha ing no hea ing loss.
A hea ing h eshold o 40 dB is he lowe limi o
mode a e hea ing loss [26] and was chosen as he
cu -o since mild hea ing loss (26 o 40 dB) is
unlikely o ma kedly a ec communica ion si ua ions
in heal h ca e, as hese mos ly ake place be ween
wo pe sons in ai ly quie su oundings. Audiome ic
da a we e a ailable o 1680 (78 %) pe sons.
Sel - epo ed hea ing di icul y was assessed wi h he
ques ion “Can you hea wi hou di icul ies wha is said
in a con e sa ion be ween se e al people (wi h o wi h-
ou a hea ing aid)?”. The esponse ca ego ies we e 1) I
can hea wi hou di icul ies 2) I can hea , bu i causes
di icul ies and 3) I canno hea a all. The la e wo
ca ego ies we e combined as he numbe o men in he
hi d ca ego y was oo low o pe mi analysis o some o
he ou come measu es. Da a on sel - epo ed hea ing
di icul y was a ailable o 1962 (92 %) pe sons.
Use o heal h se ices
In Finland, public heal h se ices, unded by local and
cen al go e nmen , a e a ailable o all ci izens. Some
se ices (e.g. p e en ion, such as cance sc eening) a e
ee o cha ge and o some se ices (e.g. physician’s
se ices, physio he apy), small ees a e cha ged. I a
ci izen uses p i a e heal h se ices a mino p opo ion
o he cos s is usually bo ne by he Social Insu ance
Ins i u ion o Finland. Audiologic ehabili a ion, in-
cluding i ing a hea ing aid, is p o ided ee o cha ge by
he public specialized heal h ca e se ice a e e e al
om p ima y heal h ca e.
As a pa o he home in e iew, he pa icipan s we e
asked how many imes du ing he las 12 mon hs hey
had isi ed a physician (no including hospi aliza ions) a
a heal h cen e (p ima y ca e), hospi al ou pa ien clinic
(seconda y ca e), occupa ional heal h ca e clinic, p i a e
clinic, in connec ion wi h a home isi , o elsewhe e.
The numbe s o isi s epo ed we e summa ized.
Pa icipan s who epo ed hea ing loss we e also asked
o s a e he numbe o physician isi s wi hin he las
12 mon hs due o hei hea ing loss. This numbe was
sub ac ed om he o al numbe o physician isi s o
yield he numbe o isi s no ela ed o hea ing loss.
The o al numbe o isi s o a nu se wi hin he las
12 mon hs was ob ained om h ee dis inc ques ions
(occupa ional nu se; o he nu se; home isi s). As phys-
ician isi s (median 2, minimum 0, maximum 50), phys-
ician isi s due o hea ing loss (median 2, minimum 0,
maximum 50) and nu se isi s (median 0, minimum 0,
maximum 1095) we e non-no mally dis ibu ed hey we e
ca ego ized. Physician isi s and physician isi s due o
hea ing loss we e classi ied in o ca ego ies no isi s/1–4
isi s/5 o mo e isi s) and nu se isi s we e classi ied in o
ca ego ies no isi s/1–5 isi s/6 o mo e isi s. The p o-
po ion o he pa icipan s epo ing a high numbe o
nu se isi s was la ge han he p opo ion epo ing a
high numbe o physician isi s. Applying cu poin s o
i e o physician isi s and six o nu se isi s yielded
compa able dis ibu ions be ween hese a iables. Pa ici-
pan s we e also asked whe he hey had ecei ed physical
he apy ( ia a e e al om a physician) and whe he hey
had used men al heal h se ices wi hin he las 12 mon hs.
Mikkola e al. BMC Ge ia ics (2016) 16:182 Page 3 o 11
They we e u he asked whe he hey had unde gone any
heal h examina ions (o ganized wi hin occupa ional heal h
ca e; o wa e e ans; ela ed o he d i e ’slicense;in
connec ion wi h unemploymen ; o o o he any easons)
wi hin he las 5 yea s. Pa icipan s we e also asked o
epo any ision es , hea ing es , mammog aphy, o
p os a e-speci ic an igen (PSA) es aken du ing he las 5
yea s. The mammog aphy analysis included women youn-
ge han 70, as hey o m he a ge g oup o he mam-
mog aphy sc eening a anged by municipal public heal h
se ices. Pa icipa ion in heal h p omo ion g oups was
de ined as ha ing a ended a g oup a ge ing weigh man-
agemen , smoking cessa ion, neck/back ehabili a ion,
o he physical exe cise, men al wellbeing, suppo o
pa ien s’ ela i es, o p oblems wi h alcohol o o he ad-
dic ions, pa en hood, sel -ca e/managemen o an illness,
o o he heal h p omo ing ac i i ies wi hin he las 5 yea s.
The pa icipan s we e u he asked: “Do you ha e a
ch onic illness o which you would like o ge con inuous
ea men by a doc o bu do no ecei e i ?”and “Do you
ha e a ch onic condi ion o which you would like o ge
o he ype o ca e bu do no ecei e i ?”. Pa icipan s
answe ing yes o ei he ques ion we e conside ed o ex-
pe ience unme heal h ca e needs.
Po en ial con ounde s
Po en ial con ounde s we e selec ed acco ding o he
c i e ia sugges ed by McNamee [27]. Acco dingly, a
con ounde mus be a cause o he ou come (use o
heal h se ices), be co ela ed wi h he exposu e (hea -
ing loss/di icul y) and no be a ec ed by he exposu e.
Sociodemog aphic ac o s included age, sex, mo he
ongue (Finnish/Swedish/o he ), income, yea s o educa-
ion, and li ing alone (yes/no). Age and sex we e
ob ained om he popula ion egis e and income om
axa ion eco ds. Household ne income was di ided by
he numbe o consump ion uni s in he household ( i s
adul wi h weigh 1, o he adul s 0.7 and child en unde
18 yea s 0.5) o yield he pa icipan ’s income. Sel - e-
po ed diseases and heal h beha io ha ha e been ound
o be associa ed wi h hea ing loss, namely ca dio ascula
disease (myoca dial in a c ion, angina pec o is, hype en-
sion, lowe limb a e ial embolism) [5], s oke [28], a h-
i is ( heuma oid o os eoa h i is) [29], diabe es [30],
alcohol use (8+ uni s/week s. less) [31] and smoking
( o me o cu en s. ne e ) [32] we e ob ained om he
home in e iew and sel -adminis e ed ques ionnai e.
Hea ing aid use was de ined as daily o almos daily use,
and was based on wo ques ions: “Do you ha e a hea ing
aid?”(yes/no) and “Do you use i daily o almos daily?”
(yes/no). Fo calcula ion o body mass index (BMI), body
weigh and heigh we e measu ed using s anda d p oce-
du es. I measu ed da a we e no a ailable o a pa ici-
pan , sel - epo s we e used o calcula e BMI. Binocula
a ision acui y was assessed on a decimal scale wi h eye-
glasses on (i he pa icipan usually wo e hem) using an
illumina ed (>350 lx) le e cha (P ecision Vision Le e
Cha Acui y Tes s) [33]. Fa ision acui y was dicho o-
mized in o he ca ego ies <0.5 (low ision), co esponding
o <20/40 in 20/20 scale, and ≥0.5 (no mal ision) [33].
Da a analysis
The sampling design, i.e. s a i ica ion and clus e ing,
was aken in o accoun in all analyses. Obse a ions
we e weigh ed o educe bias due o non- esponse and
o co ec o o e sampling o hose aged >80 yea s
using in e se p obabili y weigh s cons uc ed using
egis e da a on geog aphical a ea (uni e si y hospi al
and heal h cen e dis ic ), age, sex and mo he ongue
[25]. As age has a s ong e ec on hea ing loss, back-
g ound cha ac e is ics o hose wi h and wi hou a hea -
ing loss we e calcula ed con olling o he e ec o age.
P- alues o compa isons we e ob ained om age-
adjus ed logis ic and linea eg ession analyses using
S a a e sion 14. Con inuous a iables we e s anda dized
o he analyses. Hea ing loss was ound o ha e signi i-
can in e ac ions wi h sex on physician isi s, physical
he apy, and hea ing es . Sel - epo ed hea ing di icul y
showed a signi ican in e ac ion wi h sex on physical
he apy. The e o e, he esul s on hese ou comes a e
epo ed sepa a ely o men and women.
Fi s , we analyzed he p opo ions o heal h se ice
use s, adjus ed o sex (in case o no in e ac ion) and
age, using logis ic eg ession analysis and he p edic i e
ma gins unc ion in S a a. In he case o o dinal eg es-
sion analysis, he odds a io desc ibes how likely pe sons
wi h hea ing loss a e o ha e a ce ain alue ( e sus all
he lowe alues) o he o dinal ou come a iable com-
pa ed o hose wi hou hea ing loss. Then, mul i a iable-
adjus ed logis ic eg ession models we e un using
MPlus e sion 7 [34]. In he mul i a iable-adjus ed
models, sociodemog aphic a iables and hea ing aid use
we e used as co a ia es o all he ou come a iables. In
addi ion, diseases, smoking, alcohol use, and BMI we e
used as addi ional co a ia es o physician and nu se
isi s and pa icipa ion in a heal h p omo ion g oup. Fo
use o physical he apy, ca diac diseases, s oke, and
heuma oid a h i is/os eoa h i is we e he addi ional
co a ia es. Fo ision examina ion, a ision, diabe es
and s oke we e he addi ional co a ia es. B eas cance
and p os a e cance we e used as addi ional co a ia es
o mammog aphy and he PSA es , espec i ely. The
same analyses we e epea ed wi h sel - epo ed hea ing
di icul y as he main p edic o . Hea ing aid use was no
en e ed in o he model since pe sons who had a hea ing
aid we e asked o e alua e hei hea ing when wea ing
he hea ing aid. The analyses employed he maximum
likelihood es ima o which au oma ically akes in o
Mikkola e al. BMC Ge ia ics (2016) 16:182 Page 4 o 11

accoun missing da a in he dependen a iables bu no
in he independen a iables. Auxilia y a iables we e
no used in he maximum likelihood es ima ion.
Nex , he maximum likelihood me hod was applied in
ano he way o u he es ha he esul s we e no
biased by missing da a in he independen a iables. This
was done by epea ing he abo e men ioned eg ession
analyses and simul aneously eques ing means and a i-
ances o he independen a iables and using Mon e
Ca lo in eg a ion wi hou analysis weigh s in MPlus. In
his p ocedu e, he analysis akes in o accoun missing
da a also in he independen a iables. Maximum likeli-
hood me hod does no ill in missing alues bu uses
obse ed da a o es ima e he pa ame e s o he a iables
wi h missing da a. Based on he a ailable da a o he
a iables in he eg ession model (comple e and incom-
ple e), i iden i ies pa ame e es ima es ha ha e he
highes likelihood o unde lying he obse ed da a. The
maximum likelihood me hod, along wi h mul iple im-
pu a ion, is among he wo missing da a analysis ech-
niques ha a e conside ed as e icien o accoun ing o
missing da a when he da a a e missing a andom o
missing comple ely a andom [35]. E en i he da a is
no missing a andom maximum likelihood yields less
biased es ima es han dele ion echniques [35].
Resul s
Cha ac e is ics
In o al, 43 % (n= 837) o he pa icipan s (n= 1962) e-
po ed di icul y hea ing a con e sa ion be ween se e al
people.Among hose es edbyaudiome y(n= 1680), he
weigh ed p e alence o hea ing loss (BEHL
0.5–2kHz
>40dB)
was 16 % (non-weigh ed N= 328) (Table 1). Pe sons wi h
hea ing loss we e olde (mean 79 SE .13 s. 72 SE .39 yea s,
Table 1 Backg ound cha ac e is ics o he s udy pa icipan s o he whole sample and acco ding o hea ing s a us
All, N= 2144 No hea ing loss (BEHL
0.5–2kHz
≤40 dB), N= 1352 Hea ing loss (BEHL
0.5–2kHz
> 40 dB), N= 328
Mean
a
SE
a
Mean
a
SE
a
Mean
a
SE
a
p
b
BEHL
0.5–2kHz
(dB), n= 1680 26 0.37 21 .25 52 1.0 <.001
Body mass index (kg/m
2
), n= 1867 27 0.12 28 .13 27 .30 .047
N%
a
N%
a
N%
a
p
c
Male, n= 2144 766 39 515 39 108 45 .063
Language, n= 2144 .338
Finnish 1739 87 1237 92 283 90
Swedish 150 6.7 96 7.1 40 8.7
O he 299 5.8 19 1.3 5 1.6
Li es alone, n= 1864 892 44 570 43 172 41 .648
Educa ion, n= 1980 .029
Highes 180 9.3 136 10 24 7.0
Middle 329 17 247 18 37 14
Lowes 1471 73 966 72 264 79
Income, n= 2144 .004
Highes e ile 729 36 533 39 65 30
Middle e ile 730 35 470 35 109 36
Lowes e ile 729 29 349 26 154 34
Smoke , n= 1891 637 36 491 37 87 37 .999
Alcohol use ≥8uni s/week,n= 1742 114 7.5 105 8.3 9 6.6 .489
Ca dio ascula disease, n= 1985 1168 59 812 60 192 60 .990
S oke, n= 1993 171 7.7 92 6.7 35 8.0 .427
Diabe es, n= 1996 256 12 159 12 52 15 .236
A h i is, n= 1989 895 45 594 44 167 50 .067
Low ision, n= 1689 298 14 173 12 111 16 .034
Hea ing aid use , n= 1975 118 5.1 12 0.8 94 33 <.001
BEHL be e ea hea ing le el h eshold
a
Weigh ed and age-adjus ed
b
p- alue o compa ison be ween Hea ing loss and No hea ing loss om linea eg ession analysis
c
p- alue o compa ison be ween Hea ing loss and No hea ing loss om logis ic eg ession analysis
Mikkola e al. BMC Ge ia ics (2016) 16:182 Page 5 o 11
p< .001) han hose wi hou hea ing loss. When adjus ed
o age, pe sons wi h a hea ing loss (n= 328) had lowe
educa ion (79 % s. 72 % in he lowes e ile, p= .029),
income (34 % s. 26 % in he lowes e ile, p= .004) and
BMI (mean 27 s. 28 kg/m
2
,p= .047), and we e mo e likely
o ha e low ision (16 % s. 12 %, p= .034) han pe sons
wi hou hea ing loss (n= 1352). O he pe sons wi h
hea ing loss, 33 % epo ed using a hea ing aid daily while
o hose wi hou hea ing loss 0.8 % used a hea ing aid daily
(p< .001). Hea ing aid use s we e p esen among hose
ca ego ized as wi hou hea ing loss (<40 dB) as he ca ego y
also included pe sons wi h mild hea ing loss (25–40 dB).
Use o heal h se ices
Table 2 p esen s he p opo ions o heal h se ice use s
o pe sons wi h and wi hou hea ing loss adjus ed o
age, and also o sex in cases whe e he e was no in e -
ac ion be ween sex and hea ing loss. Men wi h hea ing
loss we e mo e likely o ha e isi ed a physician i e
imes o mo e (31 % s. 21 %) and less likely o no ha e
isi ed a physician (17 % s. 25 %) du ing he las
12 mon hs (p= .020). Pe sons wi h hea ing loss we e less
likely o ha e pa icipa ed in a heal h p omo ion g oup
du ing he las 5 yea s han pe sons wi hou hea ing
loss(16% s.23%,p= .009). Bo h men (46 % s. 29 %,
p= .003) and women (42 % s. 13 %, <.001) wi h hea -
ing loss we e mo e likely o ha e been o a hea ing es
du ing he las 5 yea s han hose wi hou hea ing loss.
In he mul i a iable-adjus ed logis ic eg ession ana-
lyses, hea ing loss was associa ed wi h inc eased odds
o ha ing used men al heal h se ices wi hin he las
12 mon hs (OR 3.2, 95 % CI 1.3–7.9, p= .034). Women
wi hhea inglosswe emo elikely oha ebeen o a
hea ing es wi hin he las 5 yea s han women wi hou
hea ing loss (OR 3.4, 95 % CI 2.3–5.2, p< .001). When
maximum likelihood me hod was used o accoun o
missing da a in bo h independen and dependen
a iables, he esul s we e pa allel o he o iginal esul s
Table 2 P opo ions o heal h se ice use s and odds a ios o hea ing loss explaining use o heal h se ices
Age-sex adjus ed p opo ion % Mul i a iable
adjus ed odds a io
All no HL, N= 1352 HL, N= 328 p o no HL s. HL OR 95 % CI p
Physician isi s, all (las 12 mon hs), n= 1925
Men 25/52/22
a
25/54/21
a
17/52/31
a
.020 1.3
c
0.8;2.2 .328
Women 19/57/24
a
18/58/24
a
17/58/26
a
.695 1.0
c
0.7;1.3 .882
Physician isi s, no ela ed o hea ing loss
(las 12 mon hs), n= 1914
22/55/22
a
21/57/22
a
19/56/24
a
.257 1.0
c
0.7;1.3 .981
Nu se isi s (las 12 mon hs), n= 1980 55/26/19
b
55/27/18
b
51/29/20
b
.291 1.2
c
0.9;1.7 .261
Heal h examina ion (las 5 yea s), n= 2006 41 44 40 .382 0.7 0.5;1.1 .201
Men al heal h se ice (las 12 mon hs), n= 1907 2.2 2.0 4.4 .059 3.2 1.3;7.9 .034
Physical he apy (las 12 mon hs), n= 1909
Men 10 9.9 15 .172 2.0
d
1.0;3.9 .088
Women 14 15 12 .298 0.9
d
0.5;1.5 .744
Heal h p omo ion g oup (las 5 yea s), n= 1677 22 23 16 .009 0.8
c
0.6;1.2 .385
Vision es (las 5 yea s), n= 1882 64 67 62 .153 0.7
e
0.5;1.0 .085
Hea ing es (las 5 yea s), n= 1870
Men 32 29 46 .003 1.1 0.6;2.0 .759
Women 17 13 42 <.001 3.4 2.3;5.2 <.001
Mammog aphy (women <70 yea , las 5 yea s), n= 259 63 64 65 .973 1.7
0.3;10.5 .624
PSA es (men, las 5 yea s), n= 628 28 29 25 .485 1.0
g
0.5;1.8 .896
Unme heal h ca e needs, n= 1929 26 26 32 .100 1.3 0.9;1.7 .232
Resul s a e gi en sepa a ely o men and women whe e in e ac ion o sex is signi ican
Only models ha include bo h sexes a e adjus ed o sex. All mul i a iable adjus ed models a e con olled o age, mo he ongue, li ing alone, income,
educa ion, and hea ing aid use
HL hea ing loss, be e ea hea ing h eshold le el 0.5–2 kHz > 40 dB; OR odds a io
a
P opo ions o 0/1–4/5+ isi s
b
P opo ions o 0/1–5/6+ isi s
c
Model addi ionally adjus ed o diseases, smoking, alcohol use, and BMI
d
Model addi ionally adjus ed o ca dio ascula diseases, s oke, and a h i is
e
Model addi ionally adjus ed o a ision, diabe es and s oke
Model addi ionally adjus ed o b eas cance
g
Model addi ionally adjus ed o p os a e cance
Mikkola e al. BMC Ge ia ics (2016) 16:182 Page 6 o 11
(men al heal h se ices OR 2.7, 95 % CI 0.9–7.2, p=.052;
hea ing es among women OR 2.5, 95 % CI 1.6–4.1, p
< .001) (Addi ional ile 1). Hea ing loss was no associa ed
wi h use o he o he heal h se ices s udied o unme
need o heal h ca e. We also an supplemen a y analyses
o see whe he he esul s emained unchanged when a
pu e- one hea ing le el o 25 dB was used as he cu
poin . Women wi h a leas mild hea ing loss
(>25 dB) we e mo e likely (OR 1.7, 95 % CI 1.2–2.4,
p= .010) o ha e been o a hea ing es han hose
wi h no mal hea ing (≤25 dB) whe eas he associa-
ions be ween mild hea ing loss and use o he o he
heal h se ices s udied we e s a is ically nonsigni ican
(Addi ional ile 2).
Table 3 p esen s he p opo ions o heal h se ice
use s o pe sons wi h and wi hou sel - epo ed hea ing
di icul y adjus ed o age, and also o sex in cases wi h
no in e ac ion be ween sex and sel - epo ed hea ing
di icul y. Pe sons wi h sel - epo ed hea ing di icul y
we e mo e likely o ha e isi ed a physician i e imes o
mo e (26 % s. 22 %) and less likely o no ha e isi ed a
physician (20 % s. 24 %) du ing he las 12 mon hs han
pe sons epo ing no hea ing di icul y (p= .020).
Fu he , pe sons wi h hea ing di icul y we e mo e likely
o ha e been o a hea ing es (29 % s. 19 %, p< .001)
and mo e likely o pe cei e unme need o heal h ca e
(32 % s. 21 %, p< .001) compa ed o hose epo ing no
hea ing di icul y. In he mul i a iable-adjus ed logis ic
eg ession analyses, sel - epo ed hea ing di icul y was
associa ed wi h inc eased odds o ha ing used men al
heal h se ices du ing he las 12 mon hs (OR 2.1, 95 %
CI 1.2–3.5, p= .025) and he odds o ha ing been o a
hea ing es wi hin he las 5 yea s (OR 1.8, 95 % CI
1.4–2.3, p< .001). Hea ing di icul y also inc eased he
likelihood o epo ing pe cei ed unme needs o heal h
ca e (OR 1.7, 95 % CI 1.4–2.1, p< .001). When maximum
likelihood me hod was used o accoun o missing da a in
bo h independen and dependen a iables, he esul s we e
pa allel o he o iginal esul s (men al heal h se ices OR
2.0, 95 % CI 1.1–3.6, p= .029; hea ing es OR 2.0, 95 % CI
1.5–2.5, p< .001; unme heal h ca e needs OR 1.7, 95 % CI
1.3–2.1, p< .001) (Addi ional ile 3). Sel - epo ed hea ing
di icul y was no associa ed wi h physician isi s no
ela ed o hea ing loss, nu se isi s, heal h examina ions,
physical he apy, heal h p omo ion g oup, ision es , mam-
mog aphy o PSA es .
As hea ing loss was no associa ed wi h mos o he
s udied heal h se ices we explo ed whe he his was
due o hea ing aid use accoun ing o a majo pa o
he co a iance be ween hea ing loss and use o heal h
se ices. This was no he case, as hea ing aid use was
signi ican ly associa ed only wi h he likelihood o ha ing
Table 3 P opo ions o heal h se ice use s and odds a ios o sel - epo ed hea ing di icul y explaining use o heal h se ices
Age-sex adjus ed p opo ions % Mul i a iable adjus ed odds a io
no HD, N= 1125 HD, N= 837 pOR 95 % CI p
Physician isi s, all (las 12 mon hs) 24/55/22
a
20/55/26
a
.020 1.2
c
1.0;1.5 .072
Physician isi s, no ela ed o hea ing loss (las 12 mon hs) 24/55/21
a
21/55/24
a
.112 1.1
c
0.9;1.4 .291
Nu se isi s (las 12 mon hs) 56/26/18
b
53/27/20
b
.109 1.1
c
0.9;1.4 .280
Heal h examina ion (las 5 yea s) 40 40 .736 1.1 0.9;1.3 .655
Men al heal h se ice (las 12 mon hs) 1.7 3.1 .058 2.1 1.2;3.5 .025
Physical he apy (las 12 mon hs)
Men 8.8 12 .145 1.3
d
0.8;2.1 .298
Women 14 14 .826 1.1
d
0.8;1.5 .647
Heal h p omo ion g oup (las 5 yea s) 22 22 .728 0.9
c
0.7;1.1 .518
Vision es (las 5 yea s) 65 65 .979 1.0
e
0.8;1.2 .709
Hea ing es (las 5 yea s) 19 29 <.001 1.8 1.4;2.3 <.001
Mammog aphy (women <70 yea , las 5 yea s) 64 64 .946 1.1
0.6;2.0 .870
PSA es (men, las 5 yea s) 28 28 .950 0.9
g
0.7;1.2 .608
Unme heal h ca e needs 21 32 <.001 1.7 1.4;2.1 <.001
Resul s a e gi en sepa a ely o men and women whe e in e ac ion o sex is signi ican . Only models ha include bo h sexes a e adjus ed o sex. All mul i a iable
adjus ed models a e con olled o age, mo he ongue, li ing alone, income and educa ion
HD sel - epo ed hea ing di icul y, OR odds a io
a
P opo ions o 0/1–4/5+ isi s
b
P opo ions o 0/1–5/6+ isi s
c
Model addi ionally adjus ed o diseases, smoking, alcohol use, and BMI
d
Model addi ionally adjus ed o ca dio ascula diseases, s oke, and a h i is
e
Model addi ionally adjus ed o a ision, diabe es and s oke
Model addi ionally adjus ed o b eas cance
g
Model addi ionally adjus ed o p os a e cance
Mikkola e al. BMC Ge ia ics (2016) 16:182 Page 7 o 11
been o a hea ing es in bo h men (OR 6.6, 95 % CI
3.5–12.6, p< .001) and women (OR 4.7, 95 % CI 2.6–8.4)
and no associa ed wi h he o he heal h se ices s udied
o pe cei ed unme needs o heal h ca e.
Discussion
The esul s o he p esen analysis showed ha olde
adul s wi h hea ing loss o sel - epo ed hea ing di icul y
we e as likely o use mos heal h se ices as o he olde
adul s. Howe e , bo h g oups we e mo e likely o ha e
used men al heal h se ices and o ha e been o a hea ing
es . Ne e heless, pe sons wi h sel - epo ed hea ing di i-
cul y we e mo e likely o expe ience unme needs o
heal h ca e han pe sons epo ing no hea ing di icul y.
The highe use o men al heal h se ices wi hin he
las 12 mon hs in pe sons wi h a hea ing loss o sel -
epo ed hea ing di icul y is in line wi h p e ious
esea ch showing mo e men al heal h se ice use among
wa e e ans wi h hea ing loss aged 18 and olde [36]. A
me a-analysis o longi udinal s udies showed ha pe -
sons wi h a hea ing loss ha e inc eased isk o dep es-
sion [37], al hough his has no been con i med by
subsequen longi udinal s udies [38, 39]. Howe e , in he
p esen s udy, adding men al illness as a co a ia e did
no educe he s eng h o he associa ion, implying ha
he e ec was no media ed h ough mani es men al
illness (da a no shown). Ins ead, pe sons wi h hea ing
loss may ha e had subclinical dep essi e symp oms o
which a leas some had ecei ed men al heal h se ices.
I should be no ed, howe e , ha a e y low p opo ion
(2.3 %) o he p esen pa icipan s had used men al
heal h se ices wi hin he las 12 mon hs. This is consid-
e ably less han he p e alence o any dep essi e dis-
o de (6.7 %) in Finnish adul s olde han 65 [40].
Unexpec edly, a e adjus ing o socioeconomic ac o s
and hea ing aid use, hea ing loss was associa ed wi h ha -
ing been o a hea ing es only among women. Howe e ,
in bo h sexes hea ing aid use was associa ed wi h in-
c eased likelihood o ha ing been o a hea ing es . Men
we e mo e likely han women o ha e been o a hea ing
es among bo h hose wi h hea ing loss and hose wi hou
hea ing loss. The pa e n was simila wi h ega d o sel -
epo ed hea ing di icul y. O he s ha e also epo ed ha
men a e mo e likely o ha e hei hea ing es ed han
women [4]. These esul s sugges ha men a e mo e
equen ly e e ed o a hea ing es han women i e-
spec i e o ac ual hea ing loss o sel - epo ed hea ing di -
icul y. The au ho s specula e ha he sex di e ence may
pa ly de i e om job his o y. Men a e mo e likely o be
exposed o occupa ional noise [41], which may lead o
mo e ollow-up o hea ing in olde men.
The p esen s udy showed ha pe sons wi h sel -
epo ed hea ing di icul y we e mo e likely o pe cei e
unme heal h ca e needs compa ed o hose epo ing
no hea ing di icul y, al hough hey we e equally o mo e
likely o ha e used heal h se ices. In he p esen s udy,
owing o he gene al na u e o he ques ion posed, un-
me needs may no be ela ed solely o hea ing heal h
se ices bu also include o he heal h se ices. Ou ind-
ing in olde adul s is in line wi h hose o a p e ious
s udy among Canadians ac oss a e y wide age ange in
which he likelihood o unme heal h ca e needs among
hose wi h sel - epo ed hea ing di icul y was 1.3 imes
highe han among hose who did no epo hea ing
p oblems [20].
In e es ingly, only sel - epo ed hea ing di icul y, bu
no measu ed hea ing loss, was associa ed wi h unme
heal h ca e needs in he p esen s udy. We specula e ha
he explana ion may lie in di e ences in communica ion
s a egies, which a e in insically aken in o accoun in he
sel - epo ed hea ing ques ion bu no in he measu ed
pu e- one hea ing le els. E ec i e communica ion s a -
egies may alle ia e he nega i e in luence o hea ing loss
on speech unde s anding [42] and he eby educe sel -
epo ed hea ing di icul y. Fu he , he ques ion on sel -
epo ed hea ing ea ed all hose wi h hea ing p oblems
equally, i espec i e o hea ing aid use. This means ha i
a pe son pe cei ed communica ion di icul ies, wi h o
wi hou a hea ing aid, he was classi ied as ha ing hea ing
di icul y whe eas he analysis o measu ed hea ing, ad-
jus ed o hea ing aid, ea ed all hea ing aid use s as a uni-
o m g oup. Pe cei ed unme needs o heal h ca e may be
explained by g ea e dissa is ac ion wi h he a ailabili y o
specialis heal h ca e and ollow-up ea men o olde
adul s wi h hea ing loss han hose wi hou disabili ies [43]
al hough e idence has also been shown o simila le els
o sa is ac ion wi h access be ween pe sons wi h and wi h-
ou hea ing loss [44]. Unme needs o and dissa is ac ion
wi h heal h ca e may esul om a pa ien wi h hea ing
p oblems ecei ing inadequa e in o ma ion owing o com-
munica ion p oblems wi h he heal h ca e p o ide [45].
This s udy has se e al s eng hs. Fi s , he sample was
popula ion-based wi h a ela i ely high pa icipa ion a e,
which inc eases he gene alizabili y o he esul s. Second,
a wide a ie y o heal h se ices and pe cei ed access o
heal h se ices we e analyzed. Thi d, he use o bo h
objec i e and subjec i e hea ing assessmen s p o ides a
deepe unde s anding on whe he use o heal h se ices is
mo e dependen on physiological hea ing loss o pe -
cei ed hea ing. On he one hand, objec i e hea ing assess-
men may e eal hea ing p oblems e en i he pe son has
no no iced hea ing loss himsel . On he o he hand, o
wo pe sons wi h equal physiological hea ing loss one may
ind hea ing p oblems mo e disabling due o en i onmen-
al and pe sonal ac o s. Fou h, we adjus ed he analyses
o se e al impo an ac o s, i.e. ce ain como bidi ies
and socioeconomic s a us, ha a e likely o con ound he
associa ion be ween hea ing and use o heal h se ices.
Mikkola e al. BMC Ge ia ics (2016) 16:182 Page 8 o 11