Polysomnography in Patients With Obstructive Sleep Apnea

7/27/2019 Polysomnography in Patients With Obstructive Sleep Apnea

1/22

31.05.2013 Polysomnography in Patients With Obstructive Sleep Apnea

www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/ 1/22

Polysomnography in Patients With Obstructive Sleep ApneaAn Evidence-Based Analysis

Health Quality Ontario

Executive Summary

Objective

The objectiv e of this health tech nology policy assessment w as to ev alu ate the clinical u ti lity an d

cost-effectiv eness of sleep stu dies in On ta rio.

Clinical Need: Target Population and Condition

Sleep disorders are comm on an d obstru ctiv e sleep apnea (OSA) is the predom inan t ty pe.

Obstru ctiv e sleep apnea is the r epetitiv e com plete obstru ction (apn ea) or partial obstru ction

(hy popnea) of the collapsible par t of the u pper airw ay dur ing sleep. The sy ndrom e is associated w ith

excessiv e day tim e sleepiness or ch ronic fatigue. Sev eral studies hav e shown tha t OSA is associated

w it h h y per ten sion , st r oke, a n d oth er ca r di ov a scu la r di sor de r s; m a n y r ese a r ch er s b el ie v e th a t th ese

car diov ascular disorders are consequen ces of OSA. T his has genera ted increasing in terest in recent

y ea r s i n slee p stu di es.

The Technology Being Reviewed

Ther e is no gold stan dard for t he diagn osis of OSA, w hich m akes it diff icult to calibrate a ny test for

diagnosis. Tra ditiona lly , poly som nogra phy (PSG) in an a ttended setting (sleep laboratory ) has been

used as a reference standar d for th e diagnosis of OSA. Poly som nogra phy m easures sev eral sleep

v a r ia bl es, on e of w h ic h is th e a pn ea -h y pop n ea in de x (A HI) or r esp ir a tor y di st u r ba n ce in dex (RDI).

The A HI is defined as the sum of apneas an d hy popneas per h our of sleep; apnea is defined as th e

absence of airflow for 1 0 seconds; and h y popnea is defined as reduc tion in r espiratory effort w ith

4% oxy gen desatur ation. The RDI is defined as the sum of apneas, hy popneas, and a bnorm al

respirat ory ev ents per hour of sleep. Often th e two term s ar e used intercha ng eably . The AHI ha s

be en w id el y u sed to dia g n ose OS A , a lth ou g h w it h di ffe r en t cu t -off le v el s, th e ba sis for w h ic h a r e

often u nclear or arbi trar i ly determ ined. General ly , an A HI of more than f iv e events per hour of

sleep is considered abnorm al an d the pat ient is considered to hav e a sleep disorder. An abnorm al

A HI a ccom pa n ie d by ex cessiv e da y tim e slee pi n ess i s t h e h a llm a r k for OS A di a g n osis. For pa tie n ts

diagnosed with OSA, continu ous positiv e airw ay pressur e (CPAP) thera py is the tr eatm ent of

choice. Poly som nogra phy m ay also used for titratin g CPAP to indiv idual n eeds.

In Jan ua ry 20 05 , th e College of Phy sicians an d Sur geons of Ontar io published the second edition of

Indep en de nt Health Facilities: Clinical Prac tice Param eters and Facility S tan da rds : Slee p Me dic ine,

comm only known as Th e Sleep Book. T he Sleep Book states that OSA is the m ost com m on prim ar y

respirat ory sleep disorder a nd a fu ll ov ernig ht sleep stu dy is considered the cur ren t standar d test for

indiv iduals in wh om OSA is suspected (based on clinical signs and sy m ptoms), particula rly i f CPAP

or sur gical th erapy is being considered.

Poly som nogra phy in a sleep laboratory is tim e-consum ing an d expensiv e. With the ev olution of

techn ology , porta ble dev ices ha v e emerg ed that m easure m ore or less the sam e sleep v ar iables in

sleep labora tories as in th e home. New er CPAP devices also ha v e au to-titra tion featu res and ca n

record s leep v ariables including AHI. These devices, i f equal ly accur ate, m ay reduce the

dependency on sleep laboratories for the diag nosis of OSA an d the titra tion of CPAP, an d thu s ma y

be m or e cost -effect iv e.

Difficulties arise, howev er, w hen try ing t o assess an d com pare th e diagnostic efficacy of in-hom e


2/22



PSG ver sus in-lab. The AH I m easured from portable dev ices in-hom e is the sum of apneas and

hy popneas per hour of tim e in bed, rath er th an of sleep, and th e absolute diagn ostic efficacy of in-

lab PSG is un known . To compar e in-hom e PSG with in-lab PSG, sev eral r esearch ers hav e used

correlation coefficients or sensitivity an d specificity , w hile others hav e used Bland-Altm an plots or

receiv er operatin g char acter istics (ROC) curv es. All these approaches, howev er, hav e potentia l

pitfalls. Corr elation coefficients do not m easure a gr eemen t; sensitiv ity an d specificity ar e not

helpful wh en the tru e disease status is un known ; an d Bland-Altm an plots m easure agr eemen t (but

are h elpful wh en th e ran ge of c l inical equiv alence is known). Last ly , receiver operating

cha ra cteristics cur v es ar e genera ted using logistic regression w ith th e tru e disease statu s as the

dependent v ar iable and test v alu es as the independent v aria ble. Thus, each v alu e of the test is used

as a cut-point to measur e sensitivity an d specificity , wh ich ar e then plotted on an x-y plane. The

cut -point t hat m axim izes both sensitiv ity and specificity is chosen a s the cut -off lev el to

discrim inate between disease and n o-disease states. In the absence of a gold stan dard t o determ ine

the tru e disease status, ROC cur v es are of m inimal v alue.

A t th e r equ est of t h e On ta r io Hea lth Tec h n olog y A dv isor y Com m it tee (OH TA C) , MA S h a s t h u s

rev iewed the l i teratu re on PSG published ov er th e last tw o year s to exam ine new developments.

Methods

Review Strategy Ther e is a lar ge body of l iterat ur e on sleep stu dies an d sev eral r ev iews hav e been

conducted. Tw o lar ge cohort stu dies, th e Sleep Heart H ealth St udy an d the Wisconsin Sleep Cohort

Study , are th e ma in sour ces of ev idence on sleep litera tu re.

To exam ine n ew dev elopm ent s on PSG published in th e past tw o y ear s, MEDLINE, EMBASE, MEDLINE

In-Process & Oth er Non-Index ed Citations, the Cochr an e Dat aba se of Sy stem at ic Rev iews an d

Cochr an e CENTRAL, INAHTA , an d w ebsites of other health techn ology assessment agen cies wer e

searc hed. An y study tha t reported results of in-hom e or in-lab PSG wa s inclu ded. All ar ticles tha t

reported findings from t he Sleep Heart H ealth Stu dy an d the Wisconsin Sleep Cohort Stu dy wer e

also rev iewed.

Diffusion of Sleep Laboratories To estim at e th e diffusion of sleep laboratories, a list of sleep labora tories

licensed under th e Independent Health Facility Act w as obtained. The an nu al nu m ber of sleep

studies per 1 00,00 0 indiv iduals in Onta rio from 20 00 to 2004 w as also estim ated using

administrat iv e databases.

Summary of Findings

Literature Review A tota l of 3 1 5 a r t ic le s w er e id en tifie d th a t w er e pu bli sh ed in th e pa st tw o y ea r s; 2 2 7

w er e ex clu ded a ft er r ev ie w in g t it les a n d a bst r a cts. A tota l of 5 9 a r t ic le s w er e id en t ifie d th a t

reported findings of the Sleep Heart Health Study an d the Wisconsin Sleep Cohort Stu dy .

Prevalence Based on cross-sectiona l data from the W isconsin Sleep Cohort Stu dy of 602 m en a nd

w om en a g ed 3 0 to 6 0 y ea r s, it is est im a ted th a t th e pr ev a le n ce of sle ep -disor de r ed br ea th in g is 9 %

in w om en an d 24 % in m en, on the basis of more than f iv e AHI ev ents per h our of sleep. Am ong the

w om en w it h slee p di sor de r br ea th in g , 2 2 .6 % h a d da y tim e slee pi n ess a n d a m on g th e m en , 1 5 .5 %

ha d day tim e sleepiness. Based on t his, the prev alence of OSA in the m iddle-aged adu lt popula tion is

est imat ed to be 2% in women a nd 4% in m en.

Snoring is present in 94 % of OSA patients, but n ot all snorers hav e OSA. Wom en report day tim e

sleepiness less often compar ed with their m ale count erpar ts (of sim ilar age, body m ass index [BMI],an d AHI). Prev alence of OSA tends to be higher in older ag e groups compar ed with y oun ger ag e

groups.

Diagnostic Value of Polysomnography It is believ ed that PSG in the sleep labora tory is m ore accu ra te tha n in-


3/22



home PSG. In th e absence of a gold stan dard, h ow ev er, claim s of accu ra cy can not be substan tiated.

In gen eral, th ere is poor corr elation betw een PSG v aria bles an d clinical v ar iables. A v ariety of cut-

off points of AHI (> 5 , > 1 0, an d > 1 5 ) ar e arbitr ar ily used to diagnose and ca tegorize sev erity of

OSA, th oug h t he clinical im porta nce of these cut -off points has not been determ ined.

Recently , a study of the u se of a thera peutic tr ial of CPAP to diagn ose OSA w as reported. Th e au thors

studied habitu al snorers with day tim e sleepiness in th e absence of other m edical or psy chiat ric

disorders. Using PSG as the reference standar d, the au thors calcula ted the sensitiv ity of this test to

be 8 0 % a n d it s sp ec ific it y to be 9 7 %. Fu r th er , th ey con clu de d th a t PSG cou ld be a v oid ed in 4 6 % of

this population.

Obstructive Sleep Apnea and Obesity Obstru ctiv e sleep apnea is strongly associated w ith obesity . Obese

indiv iduals (BMI >3 0 kg/m ) are at h igher r isk for OSA compar ed with non-obese indiv iduals and up

to 7 5 % of OSA patients ar e obese. It is hy pothesized tha t obese indiv iduals ha v e lar ge deposits of fat

in th e neck tha t cau se the u pper airw ay to collapse in the supine position dur ing sleep. The

observ ations reported from sev eral studies support th e hy pothesis that AHIs (or RDIs) are

sign ifican tly reduced w ith w eight loss in obese indiv iduals.

Obstructive Sleep Apnea and Cardiovascular Diseases A ssoc ia t ion s h a v e been sh ow n be tw ee n OS A a n d

comorbidities such a s diabetes m ellitus and hy pertension, w hich are kn ow n r isk factors for

m y ocar dial infarction an d stroke. Patients w ith m ore sev ere form s of OSA (based on A HI) report

poorer quality of l ife an d increased health car e uti lization com pared with patients with m ilder

form s of OSA. From a nim al m odels, i t is hy pothesized that sleep fra gm enta tion r esults in g lucose

intolerance an d hy pertension. There is , however, no ev idence from prospectiv e studies in hu m ans

to establish a cau sal l ink betw een OSA an d hy pertension or diabetes m ellitus. It is also not clear t ha t

th e associations between OSA a nd other diseases ar e independen t of obesity ; in m ost of th ese studies,

patients with high er v alu es of AHI ha d higher v alu es of BMI compar ed with patien ts with lower A HI

v a lu es.

A r ec en t m et a -an a ly sis of ba r ia tr ic su r g er y h a s sh ow n th a t w ei g h t loss i n obe se in di v id u a ls (m ea n

BMI = 46.8 kg/m ; ran ge = 32 .3068.80) s igni ficantly improv ed their health prof ile . Diabetes was

resolv ed in 7 6.8 % of patients, hy pertension w as resolv ed in 6 1 .7 % of patients, hy perlipidemia

im proved in 7 0% of patients, an d OSA resolv ed in 85 .7 % of patients. This sug gests tha t obesity leads

to OSA, diabetes, an d hy pertension, r ather than OSA independently causing diabetes and

hy pertension.

Health Technology Assessments, Guidelines, and Recommendations In A pril 200 5 , the Center s for Medicare an d

Medicaid Serv ices (CMS) in th e United States published its decision a nd r ev iew r egar ding in -home

an d in-lab sleep stu dies for t he diagn osis and t reatm ent of OSA w ith CPAP. In order to cov er CPAP,

CMS requires that a diagn osis of OSA be established using PSG in a sleep labora tory . After r ev iewingthe l i teratur e, CMS concluded that the ev idence w as not adequate to determine th at u nattended

porta ble sleep stu dy w as reasona ble and n ecessary in th e diagnosis of OSA.

In May 2 005 , the Can adian Coordina ting Office of Health Techn ology Assessment (CCOHTA )

published a rev iew of guidelines for r eferra l of patients to sleep labora tories. Th e rev iew inc luded 37

gu idelines and associated rev iews tha t cov ered 1 8 a pplications of sleep labora tory studies. Th e

CCOHTA reported that th e lev el of ev idence for m an y applications was of l im ited qua lity , that som e

cited studies were n ot relev ant to the recom m endations m ade, that m any recom m endations reflect

consensus positions only , an d tha t th ere w as a n eed for m ore good qua lity studies of m an y sleep

laboratory applications.

Diffusion A s of th e t im e of w r it in g , th er e a r e 9 7 li cen sed slee p la bor a tor ies i n On ta r io. In 2 0 00 , th e

nu m ber of sleep studies perform ed in Ont ar io wa s 3 7 6/10 0,000 people. There w as a steady rise in

sleep stu dies in th e follow ing y ears such tha t in 2 004 , 7 69 sleep studies per 1 00,00 0 people were

perform ed, for a total of 96 ,1 34 sleep stu dies. Based on prev alence estima tes of the Wisconsin Sleep

2

2


4/22



Cohort Study , i t wa s est ima ted that 9 27 ,1 05 people aged 30 to 60 y ears hav e sleep-disordered

br ea th in g . Th u s, th er e m a y be a 1 0-fold r ise in th e r a te of sleep tests i n th e n ex t few y ea r s.

Economic Analysis In 2004 , approxim ately 96 ,000 s leep studies were conducted in Ontar io at a tota l

cost of ~$47 m illion (Cdn). Since obesity is associated w ith sleep disordered breath ing, MAS

compar ed the costs of sleep studies to the cost of bar iatric sur gery . Th e cost of bariatr ic surg ery is

$1 7 ,35 0 per patient . In 2 004, Ontar io spent $4.7 m i l lion per y ear for 2 7 0 patients to un dergo

ba r ia tr ic su r g er y in th e pr ov in ce, a n d $8 .2 m il lion for 2 2 5 pa tien ts t o seek ou t-of-cou n tr y

treat m ent. Using a Markov m odel, i t wa s conclu ded that shifting costs from sleep stu dies to bariatr ic

sur gery would benefi t m ore patients with OSA an d m ay a lso prev ent health consequences related to

diabetes, hy pertension, an d hy perlipidemia. It is estim ated tha t the an nu al cost of treating

comorbid conditions in m orbidly obese patients often exceeds $10,0 00 per patient. T hu s, the

down stream cost sav ings could be substan tial.

Considerations for Policy Development Wei g h t loss is a ssoc ia ted w it h a de cr ea se in OS A sev er it y . Tr ea tin g a n d

prev enting obesity would also substantially reduce th e econom ic burden a ssociated w ith diabetes,

hy pertension, hy perl ipidemia, and OSA. Prom otion of healthy weights ma y be achiev ed by a

m ultisectorial a pproach as recom m ended by the Ch ief Medical Officer of Health for On tar io.

Bariatric surg ery has the potentia l to help morbidly obese individuals (BMI > 35 kg/m with anaccom pany ing comorbid condit ion, or BMI > 40 kg/m ) lose weight. In Janu ary 2005 , MAS

completed an assessment of bariat ric surger y , based on w hich OHTA C recom m ended an

im provem ent in a ccess to these sur geries for m orbidly obese patients in Onta rio.

Habitu al snorers with excessiv e day tim e sleepiness ha v e a high pretest probability of ha v ing OSA.

These patients could be offered a t hera peutic tr ial of CPAP to diagnose OSA, r ath er th an a PSG. A

m ajority of these patients are also obese and m ay benefit from w eight loss. Indiv idualized weigh t

loss progr am s shoul d, ther efore, be offered an d patient s w ho ar e m orbidly obese should be offered

ba r ia tr ic su r g er y .

Tha t said, and in v iew of the sti l l ev olv ing u nderstan ding of the cau ses, consequen ces an d optima l

treat m ent of OSA, furt her r esearch is w arr an ted to identify wh ich patients should be screened for

OSA.

Objective

The objectiv e of this health tech nology policy assessment w as to ev alu ate the clinical u ti lity an d

cost-effectiv eness of sleep stu dies in On ta rio.

Background

Clinical Need: Target Population and Condition

Sleep disorders are comm on an d obstru ctiv e sleep apnea (OSA) is the predom inan t ty pe. (1 ) Other

ty pes inclu de insom nia, n ar colepsy , restless leg sy ndr om e, and sleepwalking . Obstru ctiv e sleep

apnea is a repetitiv e com plete obstruction (apnea ) or par tial obstru ction (hy popnea) of the

collapsible part of the u pper airw ay dur ing sleep; the sy ndr om e is associated with excessiv e day tim e

sleepiness or chronic fatigue. (2 ) Sev eral studies ha v e shown tha t OSA is associated w ith accident

risk, cognit iv e impairm ent, an d cardiov ascular disorders. ( 3 ) Intu i t ively , i t could be argu ed that

excessiv e day tim e sleepiness in OSA patients would low er atten tion span and m ight incr ease the

risk of accidents com pared with people w ho do not hav e OSA. Howev er, m any researcher s believ e

tha t t he a ssociated car diov ascular disorders ar e m ore serious consequen ces of OSA. T his ha s raised

aw ar eness on t he im porta nce of OSA diagn osis.

In Can ada, 3 7 0 s leep studies per 1 00,000 population a re perform ed annu ally on a v erage an d

7 7 6/100,000 in Onta rio. Corresponding rates interna tional ly are 42 7 /1 00,000 in the United

2

2
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref03http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref02http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref01


5/22



States, 42 .5 /1 00,000 in the United Kingdom , 17 7 /1 00,000 in Belgium, an d 282 /1 00,000 in

A u str a lia . (4 ) The r ate of sleep stu dies perform ed in Ontar io is thu s ver y high in relat ion to other

provin ces in Can ada, as w ell as other coun tries. This prompted a request t o assess sleep laborat ories.

Technology Being Reviewed

Obstru ctiv e sleep apnea, un like other diseases such a s cancer, ca nn ot be diagnosed by a tissue

bi opsy . Th u s, th e a bsen ce of a g old st a n da r d by w h ic h to det er m in e th e tr u e di sea se st a tu s m a kes it

difficult t o calibrat e any test for OSA diagnosis. Tra ditiona lly , poly som nogra phy (PSG) in an

atten ded setting (sleep labora tory ) ha s been u sed as a reference standar d for the diag nosis of OSA.

This requir es observ ing patient s w hile they ar e asleep. (5 ) A patient stay s ov ernight in the s leep

laboratory and is constantly m onitored by a technician.

Poly som nography includes electroencephalography , electroocculography , subm ental

electromy ography , electrocardiogra phy , respiratory m ov ement or r espiratory effort , n asal or oral

a irf low, pulse oxim etry , and l im b movem ent electrom y ography . (6 ) Thu s, PSG m onitors sleep

stages, respira tory effort, oxy gen satur ation, heart r ate, body position, an d lim b mov em ents. These

data a re u sed to calculate th e apnea-hy popnea index (A HI) or r espiratory distur bance in dex (RDI).

The A HI is the sum of apneas and h y popneas per hour of sleep with a pnea defined as the absence ofairflow for 10 seconds an d hy popnea a s reduction in respira tory effort w ith 4% oxy gen

desatu ra tion. Th e RDI is the sum of apneas, hy popneas, and abn orm al respira tory ev ents per hour

of sleep. Often, th e term s are u sed intercha ng eably .

The A HI ha s been w idely used to diagnose OSA, a lthough with different cut -off levels, the basis for

w h ic h a r e oft en u n cle a r or a r bit r a r il y de ter m in ed. Gen er a lly , a n A HI of g r ea ter th a n fiv e ev en ts

per h our of sleep is considered abnorm al a nd t he pat ient is considered to ha v e a sleep disorder. An

abnorm al AH I accompa nied by excessiv e day tim e sleepiness is the h allm ar k for th e diagnosis of

OSA. For patient s diagnosed with OSA, continu ous positiv e airw ay pressur e thera py (CPAP) is the

treat m ent of choice. Poly som nogra phy m ay also used for titr ating CPAP to indiv idual needs.

In Jan ua ry 20 05 , th e College of Phy sicians an d Sur geons of Ontar io published the second edition of

Indep en de nt Health Facilities: Clinical Prac tice Param eters and Facility S tan da rds : Slee p Me dic ine,

comm only known as Th e Sleep Book. T he docum ent w as designed to assist phy sicians in th eir

clinical decision-ma king by providing a fram ework for assessing a nd tr eating clin ical conditions

comm only car ed for by a v ar iety of specialties. The prim ary pur pose w as to assist phy sicians in

developing their own qual i ty m anag ement program and to act as a g uide for a ssessing the qual i ty of

patient car e prov ided in t hese facilities. Th e Sleep Book r eports th at OSA is the m ost com m on

prim ary respira tory sleep disorder an d full over nigh t sleep study is the cu rr ent standar d for those

indiv iduals in wh om OSA is suspected (based on clinical signs and sy m ptoms), particula rly i f CPAP

or surg ical thera py a re being considered.

Poly som nogra phy in a sleep laboratory is tim e-consum ing an d expensiv e. With the ev olution of

technology , portable devices hav e emerg ed that m easure m ore or less the same s leep v ariables

w h et h er in slee p la bor a tor ie s or in -h om e. Th e A m er ic a n Sl ee p Disor de r s A ssoc ia t ion cla ssi fies th ese

dev ices into four ty pes: Ty pe I dev ices ar e considered th e standar d laboratory -based PSG. Ty pe II

devices are comprehensiv e portable PSG dev ices with a m inimu m of sev en chan nels that m easur e

the sam e param eters as those by Ty pe I dev ices, inclu ding sleep stag ing. Ty pe III dev ices ha v e a

m inimu m of four chan nels and m easur e only the cardiorespiratory param eters of sleep. The AHI

calcu lated from these dev ices is calcu lated per hour of time in bed, ra ther tha n per h our of sleep.

Ty pe IV devices m easur e only oxy gen satur ation or a irf low. Newer CPAP devices a lso hav e aut o-titra tion featur es. These dev ices, i f equa lly a ccu rat e, ma y reduce th e dependency on sleep

laboratories for t he diagn osis of OSA an d th e titra tion of CPAP an d th us m ay be m ore cost-effectiv e.

Difficulties arise, howev er, w hen try ing t o assess an d com pare th e diagnostic efficacy of in-hom e


6/22



PSG ver sus in-lab. The AH I m easured from portable dev ices in-hom e is the sum of apneas and

hy popneas per hour of tim e in bed, rath er th an of sleep, and th e absolute diagn ostic efficacy of in-

lab PSG is un known . To compar e in-hom e PSG with in-lab PSG, sev eral r esearch ers hav e used

correlation coefficients or sensitivity an d specificity , w hile others hav e used Bland-Altm an plots or

receiv er operatin g char acter istics (ROC) curv es. All these approaches, howev er, hav e potentia l

pitfalls. Corr elation coefficients do not m easure a gr eemen t; sensitiv ity an d specificity ar e not

helpful wh en the tru e disease status is un known ; an d Bland-Altm an plots m easure agr eemen t (but

are h elpful wh en th e ran ge of c l inical equiv alence is known). Last ly , receiver operating

cha ra cteristics cur v es ar e genera ted using logistic regression w ith th e tru e disease statu s as the

dependent v ar iable and test v alu es as the independent v aria ble. Thus, each v alu e of the test is used

as a cut-point to measur e sensitivity an d specificity , wh ich ar e then plotted on an x-y plane. The

cut -point t hat m axim izes both sensitiv ity and specificity is chosen a s the cut -off lev el to

discrim inate between disease and n o-disease states. In the absence of a gold stan dard t o determ ine

the tru e disease status, ROC cur v es are of m inimal v alue.

Literature Review on Effectiveness

Research Questions

1 . Wha t is the clinical uti lity of sleep laboratory studies?

2 . Wha t is the diffusion of sleep labora tory techn ology in Onta rio?

3 . Ar e sleep laboratory studies cost-effectiv e?

Review Strategy

The objective of the litera tu re rev iew w as to address the question: Wh at is the clinical u ti lity of sleep

laboratory studies?

Ther e is a lar ge body of l i teratu re on sleep studies, and sev eral r ev iews hav e been conducted. Tw o

larg e cohort studies, the Sleep Heart Health Stu dy and th e Wisconsin Sleep Cohort Stu dy , are th e

m ain sour ce of ev idence on sleep litera tu re. Th e MAS r ev iewed all l i teratu re published on PSG ov er

th e past tw o y ear s to exa m ine n ew dev elopmen ts in th e diagn osis of OSA. MEDLINE, EMBASE,

MEDLINE In-Process & Other Non-Index ed Cita tions, th e Cochr an e Dat aba se of Sy stem at ic Rev iews

an d Cochr an e CENTRAL, INAHTA , an d websites of other health techn ology assessment agen cies

w er e sea r ch ed . A ll st u di es r ep or tin g r esu lt s of a PSG in a slee p la bor a tor y or in -h om e w er e in clu de d.

Studies that did not use PSG wer e excluded. In a ddition, t o un derstand th e clinical im porta nce of

diagnosing and tr eating OSA, a rt icles that reported f indings from the Sleep Heart Health Study and

the Wisconsin Sleep Cohort Stu dy wer e also rev iewed.

Results of Literature Review

A tota l of 3 1 5 a r t ic les w er e id en tif ied th a t w er e pu bli sh ed in th e pa st 2 y ea r s; 2 2 7 w er e ex clu de d

after rev iewing titles an d abstracts. A t otal of 5 9 a rticles were identified that reported findings of

the Sleep Heart Hea lth Stu dy an d the Wisconsin Sleep Cohort Stu dy . Table 1 shows th e qua lity of

ev idence of inclu ded stu dies. Please note th at the table does not apply to diagnostic studies.

Table 1:

Quality of Evidence of Included Studies

Summary of Existing Health Technology Assessments

*
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab01/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab01/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab01/?report=objectonly


7/22



In A pril 20 05 , the CMS in th e United States published its decision a nd r ev iew reg ar ding sleep

studies in-hom e com pared w ith in-lab for th e diagnosis an d treatm ent of OSA w ith CPAP. (7 ) In

order to cov er CPAP, CMS requir es that a diag nosis of OSA be established u sing PSG in a sleep

laboratory . After rev iewing th e l i teratu re, CMS concluded that th e evidence was not adequate to

determ ine tha t un atten ded porta ble sleep study wa s reasona ble and necessar y in th e diagn osis of

OSA.

In May 2 005 , th e CCOHTA pu blished a r ev iew of guidelines for r eferr al of patients to sleep

laboratories. (6 ) The r eview included 37 guidel ines and associated reviews that cov ered 18

applications of sleep labora tory studies. Th e CCOHTA r eported th at t he lev el of evidence for m any

applications wa s of l im ited qua lity , tha t som e cited stu dies w ere not relev an t to the

recomm endations m ade, that m an y recomm endations reflect consensus positions only , and that

ther e wa s a need for m ore good qua lity studies of ma ny sleep labora tory applications.

Medical Advisory Secretariat Review

The findings of the MAS r ev iew ar e presented un der the followin g th em es:

Prevalence A w el l-c it ed a r t ic le ba sed on cr oss-sect ion a l da ta fr om th e Wisc on sin Sl ee p Coh or t Stu dy

reported findings on 6 02 m en and w om en aged 30 to 60 y ears . (2 ) The au thors est ima ted that the

prev alence of sleep disordered breath ing w as 9% in w omen a nd 2 4% in m en on the basis of more

than f iv e AHI ev ents per hour of sleep. Am ong the w om en w ith s leep disorder breathing, 22 .6% h ad

day time s leepiness and am ong the m en, 1 5 .5 % had day time s leepiness. Based on this, th e

prev alence of OSA in th e middle-aged adu lt populat ion is estimat ed to be 2% in w om en an d 4% in

m e n .

Snoring is present in 94 % of OSA patients, but n ot all snorers hav e OSA. Wom en report day tim e

sleepiness less often compar ed with their m ale count erpar ts (of sim ilar age, BMI, an d AHI). The

prev alence of OSA also tends to be higher in older age gr oups compar ed with y oun ger a ge gr oups.

Man y patients w ith suspected OSA a lso hav e positiona l sleep apnea. ( 8 ) Positiona l sleep apnea isdefined as a 5 0% redu ction in A HI dur ing n on-supine sleep in r elation t o supine sleep. In one study ,

it w as estim ated th at 2 6% of patients with a positiv e sleep test h ad positiona l sleep apnea. ( 8 )

Patients with positiona l sleep apnea m ay benefit from positiona l ther apy designed to prev ent th e

supine position dur ing sleep.

Diagnostic Value of Polysomnography It is believed th at PSG in-lab is more accu ra te tha n PSG in-hom e. In th e

absence of a gold stan dard, howev er, claim s of accu rac y can not be substantiated. In gener al, ther e

is poor corr elation betw een PSG v ar iables an d clinical v ar iables. A v ariety of cut-off points of AHI (>

5 , > 10, a nd > 15 ) are ar bi trari ly used to diagnose and categorize sev eri ty of OSA. Th us, these cut-

off points hav e undeterm ined clinical im porta nce. (2 )

Recently , one stu dy used a ther apeutic tr ial of CPAP to diagn ose OSA. (9 ) The au thors stu died

habitual snorers with day time s leepiness that did not ha v e any other m edical or psy chiatric

disorders. Using PSG as the reference standar d, the au thors calcula ted the sensitiv ity of the test to

be 8 0 % a n d spec ific it y to be 9 7 %. Th ey con clu de d th a t PSG cou ld be a v oid ed in 4 6 % of t h is

population.

Obstructive Sleep Apnea and Obesity Obstru ctiv e sleep apnea is strong ly associated with obesity . (1 0 ;1 1 )

Obese indiv iduals (BMI > 30 kg/m ) are a t a h igher risk for OSA. For exa m ple, Up to 75 % of OSA

patients seen at th e Univ ersity Hea lth Netw ork in Toronto are obese (Persona l com m un ication,

Mar ch 2 005 ). It is hy pothesized that obese indiv iduals hav e larg e deposits of fat in the n eck tha tcau ses the upper a irw ay to collapse in th e supine position du ring sleep. The observ ations reported

from sev eral stu dies supports th e hy pothesis tha t A HIs (or RDIs) are significantly reduced w ith

w ei g h t loss i n obese in di v idu a ls . (1 2 -1 4 ) For exa m ple, Dixon et al. ( 1 4 ) prospectiv ely followed 2 5

sev erely obese patients for 1 7 10 m onths fol lowing bariatric surgery . The m ean BMI wa s 52 .7

2
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref14http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref14http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref12http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref11http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref10http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref09http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref02http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref08http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref08http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref02http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref06http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref07


8/22



9 . 5 k g /m a t ba se li ne com pa red wi th 3 7 . 2 7 . 2 k g /m a t th e end of study (P< .001 ); mean AHI was

61 .6 31 .9/hr com pared with 13 .4 1 3/hr at the end of study (P< .001 ); 23 of the 25

patients(92 %) needed CPAP at baseline com pared with 2 4% at th e end of study (P< .001 ).

Wei g h t loss i s on e of t h e few in ter v en tion s t h a t m a y cu r e OS A . (1 5 ) This m ay be achiev ed by

m odification of li festy le, diet, m edication, an d bariatr ic sur gery . The cu rr ent epidem ic of obesity is

likely to driv e an increa se in obesity -related sleep disorders, inclu ding OSA, as well as other

comorbid conditions. Thus, the Chief Medical Officer of Health for Ontario has recognized the

ov erw eight an d obesity epidem ic as one of the biggest cha llenges, and has recomm ended a

com prehensiv e and m ult isectoria l s trategy to help the people of Ontario achiev e and m aintain a

h ea l th y w eig h t . (1 6 ) In Janu ar y 2 005 , MAS completed an assessment of baria tric surg ery , based on

w h ic h th e On ta r io Hea lt h Tec h n olog y A dv isor y Com m it tee (O HTA C) r ec om m en de d im pr ov in g

access to these surg eries for m orbidly obese patient s in Ont ar io.

Obstructive Sleep Apnea and Cardiovascular Diseases A ssoc ia t ion s b et w ee n OS A a n d h y per ten sion h a v e bee n

demonstrat ed: patients with a m ore sev ere form of OSA (based on AH I) hav e a higher pr ev alence of

hy pertension compared w ith patients who hav e m i lder form s. (1 7 -2 1 ) It is, as y et, un clear,

w h et h er th ese a ssoc ia t ion s a r e in de pe n de n t of ob esi ty . In m ost of t h ese st u di es, pa tie n ts w it h h ig h er

A HI v a lu es a lso h a d h ig h er BMI v a lu es com pa r ed to t h ose pa tie n ts w it h low er A HI v a lu es. Fr oman im al m odels, i t was initially hy pothesized that OSA can lead to sustained hy pertension. (2 2 ) In a

rev iew published in 2000, h owev er, Youn g an d Peppard (2 3 ) concluded that th ere was no evidence

from prospectiv e stu dies in hu m an s to establish a ca usal l ink betw een OSA an d hy pertension.

Since then , few studies ha v e reported findings from prospectiv ely collected data . In 2 000, Peppard

et al. (2 4 ) published their f indings from the Wisconsin Sleep Cohort Stu dy on 8 93 part icipants on

w h om th ey h a d fol low -u p da ta for a t le a st 4 y ea r s. Th e a u th or s def in ed h y per ten sion a s a bl ood

pressure of at least 1 40/90 m m Hg, or the u se of antihy pertensiv e m edications. They div ided the

cohort by baseline v alu es of AHI into four gr oups: 1) A HI = 0; 2) A HI = 0.1 4.9 ; 3) A HI = 5 1 4.9 ;

an d 4) AHI 1 5 . Using th e first group as the reference group, th ey compar ed the other gr oups for

ra tes of hy pertension a t th e end-point v ia logistic regr ession. A fter adjusting for ba seline

hy pertension, BMI, alcohol, and cig ar ette u se, th ey com puted odds ra tios (ORs) and 9 5 % confidence

interv als (CIs). They foun d that th e odds of hy pertension w ere high er in gr oups 2 to 4 compa red

w it h g r ou p 1 [O R = 1 .4 2 (2 v s. 1 ), 2 .0 3 (3 v s. 1 ), 2 .8 9 (4 v s. 1 ); P= .002 for t rend]. A s seen in

Table 2 , howev er, i t is ev ident tha t BMI also tended to be higher in the gr oups w ith hig her v alu es of

A HI com pa r ed to g r ou ps w it h low er A HI v a lu es. Th e a u th or s a lso a ckn ow ledg ed th a t th e m ea su r es

of body habitu s (BMI an d wa ist an d neck circu m ferenc e) were strong confoun ding v ar iables.

Table 2:

Characteristics of the Participants who Completed One or Both Follow-Up Slee p Studies, According to the

Apnea-Hypopnea Index at Baseline

In 2 003 , Kaneko et al. (2 5 ) published findings from a r an dom ized contr olled trial compar ing CPAP

to medical treatm ent in only 24 patients with heart fa ilure an d OSA. The m ean sy stol ic blood

pressure was 1 28 [standard err or (SE) = 7 ] at basel ine, and 13 4 (SE = 8) at 1 m onth in the m edical

treatm ent group, com pared with 1 26 (SE = 6) at baseline and 11 6 (SE = 5 ) at 1 m onth in th e CPAP

group (P= .008). Th ere w ere n o significant differences in diastolic blood pressur e. Howev er, th e

au thors did not r eport im pact on h y pertension u sing the conv entional definition of blood pressur e

greater th an 1 40/90 m m Hg; thu s, the cl inical importan ce of these f indings is un clear.

In 2 004 , Gotsopoulos et al. ( 2 6 ) published findings from a ra ndomized crossov er tr ial com parin g

m andibular a dv ancem ent splint for 4 weeks to oral a ppliance (control) for 4 weeks, in 6 1 patients

w it h OS A . A t th e en d of stu dy , m ea n A HI w a s 1 2 (S E = 2 ) in th e spli n t g r ou p com pa r ed w it h 2 4 (S E

2 2
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref26http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref25http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref24http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref23http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref22http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref21http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref17http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref16http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref15http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab02/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab02/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab02/?report=objectonly


9/22



= 2 ) in th e contr ol group (P< .0001 ). Mean sy stolic blood pressure w hile aw ake w as 13 1 .6 (SE = 1 .5 )

at basel ine, which r educed to 12 6.7 (SE = 1 .7 ) in the spl int group com pared with 1 30.1 (SE = 1 .5) in

the control group at th e end of stu dy (P= .003 ). Similar ly , m ean diastolic blood pressure redu ced

from 80.9 (SE = 1 .0) at basel ine to 77 .2 (SE = 1 .2) in the spl int group com pared with 8 0.7 (SE = 1 .0)

in th e contr ol group (P< .0001 ). Aga in, th e clinical im porta nce of these findings is not clear.

In 2 005 , Dursun oglu et a l . ( 2 7 ) inv estigat ed acu te effects of au toma tic CPAP on blood pressur e in 1 2

patients with OSA an d hy pertension. Th ey compar ed sy stolic and diastolic blood pressure

m easurem ents after ov ernig ht CPAP with baseline v alu es. There wer e no sign ifican t differences.

A lso i n 2 00 5 , Her m ida et a l. (2 8 ) published the results of CPAP ther apy on a m bula tory blood

pressur e at 2 an d 4 m onths post-CPAP. In th is study , 64 of 83 patients (77 %) treated with CPAP had

hy pertension at basel ine. After 4 m onths, 61 (74 %) were st il l hy pertensiv e (P> .05 ). The auth ors

sugg ested that OSA patients m ust be ev aluated for h y pertension an d treated with an tihy pertensive

drugs rath er tha n CPAP alone.

Sev eral studies hav e docum ented an association between sleep disordered breath ing a nd diabetes.

(2 9 -3 1 ) Th oug h, as is the ca se with hy pertension, these associations are based on cross-sectiona l

data an d hence pr ovide no ev idence for a cau se-effect relat ionship. Only one of these thr ee studies

reported BMI v alu es stra tified by dia betes stat us. Resnick et a l. (2 9 ) studied 4,87 2 part icipants in

the Sleep Heart Health Stu dy . They reported that the m ean BMI wa s 31 .3 [standar d dev iat ion (SD)

= 6.0 ] in 47 0 participan ts with diabetes compar ed with a BMI of 2 8.1 (SD = 5 .1 ) in 4,4 02 n on-

diabetic participants (P< .00 1 ). Th e au th ors also reported a positiv e association betw een BMI an d

RDI.

In 2 005 , Reichm ut h et al. published findings from a longitudina l ana ly sis of the Wisconsin Sleep

Cohort study. (3 2 ) Of the 1 ,3 82 part icipants studied at baseline, BMI tended to increase acr oss each

category of AHI. That is, m ean BMI wa s 27 .9 in th e group with an AHI


10/22



A HI 3 0 a n d 2 3 w it h a n A HI 5 3 0 ), a n d 2 6 4 h ea lt h y m en . Un tr ea ted pa t ien ts w er e th ose

w h o r efu sed CPA P t h er a py . Hea lt h y m en w er e m a tch ed for a g e a n d BMI w it h u n tr ea ted pa t ien ts

w h o h a d sev er e OS A . Baseli n e ch a r a cter ist ic s a r e sh ow n in Table 4 .

Table 4:

Participant Characteristics (Marin et al. 2005)

By m atch ing for age an d BMI, the a ut hors were able to balance th e groups (as seen in Table 4 ). A

sign ifican tly high er proportion of OSA patient s w ere foun d to ha v e hy pertension, diabetes, and

cardiov ascular disease, com pared with h ealthy m en (control group). New car diov ascular ev ents

occurr ed more frequently in un treated patients with severe OSA com pared with healthy m en (

Figur e 1 ). Th e au thors used a m ultiple logistic reg ression m odel to adjust for age, pr esence of

car diov ascular disease, hy pertension, diabetes, l ipid disorders, sm oking status, a lcohol u se, sy stolic

an d diastolic blood pressur e, blood glu cose, total cholesterol, trig ly cerides, a nd cur ren t u se of

2
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref35http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref34http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig01/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab04/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab04/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab04/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab04/?report=objectonly


11/22



an tihy pertensiv e, l ipid lower ing, an d antidiabetic dru gs. After adju sting for these va riables, OR

w a s 2 .8 7 (CI = 1 .1 7 7 .5 1 ) for th e u n tr ea ted sev er e OS A g r ou p com pa r ed w it h th e con tr ol g r ou p.

A g e (OR = 1 .0 9 ; CI = 1 .04 1 .1 2 ) a n d pr e-ex ist in g ca r di ov a scu la r di sea se (O R = 2 .5 4 ; C I = 1 .3 4 .9 9 )

w er e a lso si g n ific a n t pr ed ic tor s of n ew ca r di ov a scu la r ev en ts in th is m ode l. Th e a u th or s c on clu de d

that sev ere OSA patients are at higher r isk of cardiov ascular ev ents com pared with h ealthy m en

and th at CPAP treatm ent reduces this r isk.

Figure 1:

Cumulative Percentage of Individuals With New Fatal (A) and Nonfatal (B) Cardiovascular Events in Each

of the Five Groups

The r esults of this study should be seen in the context of its l im itations as it w as not a r an dom ized

trial. Obstru ctiv e sleep apnea pat ients had poorer h ealth pr ofiles at baseline com pared w ith hea lthy

m en. Howev er, th e reported basel ine char acterist ics of patients w ith severe u ntreated OSA were

sim ilar to CPAP-treat ed patients. Thu s, it could be argu ed that th e lower ev ent ra te in CPAP

treatm ent gr oup w as due to CPAP therapy . It could a lso be argu ed that th e patients wh o were

com pliant with CPAP therapy were a lso com pliant with t he m edical ma nagem ent of com orbid

conditions, w hile patients wh o refused CPAP ther apy w ere also noncompliant w ith other form s of

therapy . This could hav e biased the results in favour of CPAP therapy . In a ddit ion, th ere m ay be

correlat ions am ong m any of the v ariables included in the m ult iv ariate ana ly ses. The au thors did

not report w hether th ey checked for m ult icol l ineari ty or w hether th ey perform ed any other m odel

diagnostics. These are standar d procedur es for com plex ana ly ses to ensure t ha t th e robustness of

results.

Ya g g i et a l. (3 6 ) enrol led 1 ,02 2 patients wh o underwent PSG and recorded subsequent ev ents(stroke and death). Of their tota l sam ple, 69 7 (68%) had OSA (AHI > 5 ) and 3 25 did not hav e OSA

(contr ols). Th eir baseline ch ara cteristics ar e shown in Table 5 .

Table 5:

Baseline Characteristics of Patients With Obstructive Sleep Apnea Syndrome and Controls

Out of 69 7 patients in the OSA gr oup, 1 24 (1 8%) w ere lost to follow-up an d out of 32 5 patients in

the control group, 5 6 (1 7 %) were lost to follow -up. Th e ev ent r ate of stroke or death w as 3.4 8/100

person-y ears in the OSA group an d 1.6 0/100 person-y ears in the contr ol gr oup. After adju sting for

age, sex, ra ce, smoking statu s, alcohol consum ption, BMI, diabetes, hy perlipidemia , atr ial

f ibri l lat ion, a nd hy pertension, th e hazard rat io was 1 .97 (CI = 1 .1 23.48). T he au thors concluded

tha t OSA significantly increa ses the risk of stroke and death an d the inc rease is independent of other

risk factors.

These results should also be used in th e sam e context of l imitat ions as of Mar in et al. T he OSA gr oup

ha d a poorer h ealth pr ofi le com pared w ith th e contr ol group. Man y of the v ar iables inclu ded in th e

m ult iv ariate an aly ses m ight be correlated but the au thors m ake no mention of m odel checking;

ther efore, th e differen ce in stroke or death can not be solely att ribut ed to OSA.

The effect of obesity ha s been furth er substantiat ed by a m eta-analy sis (1 3 ) of bariatric sur gery ,

w h ic h de m on str a ted th a t w ei g h t loss i n obese in di v idu a ls (m ea n BMI = 4 6 .8 5 kg /m ; r a n g e =

32 .3068.80) s igni ficantly improves health prof iles; hy pertension wa s resolv ed in 61 .7 % of

2
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref13http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref36http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab05/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab05/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab05/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig01/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig01/?report=objectonly


12/22



patients, diabetes w as resolv ed in 7 6.8 % of patients, hy perlipidemia im proved in 7 0% of patients,

an d OSA resolv ed in 85 .7 % of patients. This suggests that obesity leads to OSA, diabetes, and

hy pertension, rath er th an OSA independently causing h y pertension, diabetes, or stroke.

Economic Analysis

Economic Literature Review: Summary

The Medical Adv isory Secretar iat l i teratu re search identified 3 a rticles that contain ed some form of

econom ic ana ly sis in OSA pa tients. In th e first ar ticle, Pelletier-Fleur y et al. (3 7 ) compa red costs

an d sleep outcomes between 8 2 pat ients randomized to im m ediate PSG, with 8 9 patient s

ra ndomized to PSG with in 6 m onth s. Costs (in Euros) w ere r elated to com orbid conditions (an d

m edications) including h y pertension, stroke, angin a, diabetes, hy perlipidemia, a nd depression.

Outcom es were sleepiness as m easured by the Epworth Sleepiness Scale, percent age of positiv e

responses to Notting ha m Healt h Profile item s, and scores of the fiv e dimen sions of th e Nottin gh am

Health Profile. The a ut hors stra tif ied OSA patients into tw o subg roups: 1) A HI < 30 ev ents/hour ;

an d 2) A HI 30 ev ents/hour , an d calcu lated costs per pa tient a ssociated w ith a differenc e of 1-point

decrease in th e Epworth score, 1 % decrease in positiv e responses to the Nottingh am Health Profile,

or 1 point decrease in the fiv e dim ensions of the Notting ha m Health Profi le. They foun d that t he

increm enta l cost-effectiv eness rat ios were low er in th e subg roup with an A HI 30 ev ents/hour . The

auth ors argu ed for early m ana gement of patients with a m ore sev ere form of OSA.

A lba r r a k et a l. (3 8 ) com pared 1 0-year uti lization ra tes of health resour ces in 34 2 pat ients with

OSA (cases), to patients w ithout OSA (a ge m atch ed contr ols), using t he Man itoba Health Data base.

They ha d data from 5 y ears prior to the diagn osis of OSA to 5 y ears post-CPAP in OSA pat ients.

Ther e was a significant difference in phy sician v isits (m ean = 1 .85 , SE = 0.5 2 ; P< .05 ) and

phy sician fees between ca ses an d contr ols (m ean = $61 .44 (Cdn), SE = 2 9.5 1 ; P< .05 ). Mean v isits

an d fees wer e higher in cases compar ed with contr ols, howev er, th ere wa s a significant drop in

phy sician v isits and phy sician fees in th e cases from 1 y ear prior to diagn osis to 2 and 5 y ears post-diagnosis. This w as mostly du e to a reduction in u ti lization of psy chiat ric an d respirat ory serv ices.

A y a s e t a l. (3 9 ) assessed cost-effectiv eness of CPAP ther apy in r elation to no ther apy in OSA

patients . They assum ed that CPAP therapy would reduce accident rates in OSA patients and u sed a

Markov m odel to relate costs with qual i ty of l ife ov er 5 y ears . From a third-party pay ers

perspectiv e, the incr em enta l cost of CPAP wa s $3,3 5 4 (US) per qua lity -adjusted life y ear (QALY)

gain ed; from a societal perspectiv e this v alu e was $31 4 (US). The au thors conclu ded tha t CPAP

therapy wa s economical ly attra ct ive in OSA patients.

The r esults published by Pelletier-Fleur y et al. (3 7 ) ar e not u seful for clinical or policy decision-

m aking as the clinical relev an ce of the reported outcomes is ambigu ous. The findings of Albarr ak et

a l . (3 8 ) suggest tha t u ntreated OSA patients may un necessari ly ut i l ize psy chiatric an d respiratory

serv ices. Thu s, CPAP m ay be cost-sav ing becau se w hen OSA patient s ar e treated, other r esources

are freed up. The A y as et a l . (3 9 ) m odel did not captu re th is aspect of cost-sav ing. T hey m odelled

the effect of CPAP on accident ra tes only . To estim ate t his effect th ey used before-after da ta on

accident ra tes in patients on CPAP an d condu cted a meta -an aly sis. Intu itiv ely , this approach

ov erest ima tes the effect as m ost indiv iduals would apply greater caution wh i le driv ing a f ter

be com in g in v olv ed in a n a ccid en t ; t h u s, r ed u ction in a cciden t r a tes in a befor e-a n d-a ft er de sig n

can not be solely a ttr ibuted to CPAP ther apy .

Ontario-Based Economic Analysis

Notes and Disclaimer

Th e MAS u ses a stan dar dized costing m eth odology for a ll of its econom ic an aly ses of tech nologies.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref39http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref38http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref37http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref39http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref38http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref37


13/22



The m ain cost cat egories an d the associated m ethods from the prov inces perspectiv e are as

follows:

Hospital: Onta rio Case Costing Initiativ e (OCCI) cost da ta is used for all pr ogr am costs

w h en th er e a r e 1 0 or m or e h ospit a l sepa r a tion s, or on e-t h ir d or m or e of h ospit a l

separa tions in the m inistry s data w areh ouse are for th e design ated Inter na tiona l

Classificat ion of Diseases-1 0 dia gn osis codes and Ca na dian Classificat ion of Healt h

Int erv ent ions procedur e codes. Wh ere appr opriate, costs ar e adju sted for h ospital-specific

or peer-specific effects. In cases wh ere th e techn ology un der r ev iew falls out side the

hospitals that r eport t o the OCCI, Progr am Assignm ent Code (PAC)-1 0 w eights conv erted

into monetary un its are u sed. Adjustments m ay need to be made to ensure the relev ant

case m ix g roup is reflectiv e of the diagnosis and procedures un der consideration. Due to

the difficulties of estim ating indirect costs in hospitals associated w ith a particu lar

diagnosis or pr ocedure, MAS norm ally defau lts to considering direct t reatm ent costs

only . Historical costs hav e been adjusted upwar d by 3% per an nu m , representing a 5 %

inflation r ate assum ption less a 2 % im plicit expectation of efficiency gain s by hospitals.

Non-Hospital: These include phy sician serv ices costs obtained from th e Prov ider Serv ices

Bra nch of the Onta rio Ministry of Health an d Long-Term Care, dev ice costs from the

perspectiv e of local h ealth car e institutions, and dru g costs from the On tar io Dru g Benefit

form ulary l i st price.

Discounting: For a ll cost-effective a na ly ses, discoun t r ates of 5% and 3 % ar e used as per t he

CCOHTA an d th e Washin gton Pan el of Cost-Effectiv eness, r espectiv ely .

Downstream Cost Sav ings: All cost av oidance a nd cost sav ings ar e based on a ssum ptions of

ut i lization, care patter ns, fun ding, an d other factors. These m ay or m ay not be realized

by th e sy st em or in di v idu a l in st it u tion s.

In cases w here a dev iation from th is standar d is used, an explan ation has been giv en as to the

reasons, the assum ptions, an d the rev ised approach .

The econom ic ana ly sis represents an estima te only , based on a ssum ptions and costing m ethods

tha t hav e been explicitly stated above. Th ese estim ates will chan ge if differen t assum ptions and

costing m ethods ar e applied for the pu rpose of dev eloping im plement ation plan s for the

technology .

Diffusion of Sleep Laboratories The objectiv e of this analy sis wa s to address the second r esearch question:

Wh a t is th e di ffu sion of sleep la bor a tor y tec h n olog y in On ta r io?

A li st of sle ep la bor a tor ies l ic en sed u n de r th e Indep endent He alth Facilities Ac twas obtained. In

addit ion, th e ann ual nu m ber of sleep studies per 1 00,000 indiv iduals in Ont ario from 2000 to

2004 wa s est ima ted using adm inistrat iv e databases.

Cur rent ly , there ar e 97 licensed sleep laboratories in Ontar io in independent health facilities an d

sev eral in On tar io hospitals. In 2 000, t he n um ber of sleep stu dies perform ed in Ontar io w as

37 6/100,000 people. Since then, there h as been a steady rise in the an nu al v olum e of studies, such

tha t in 2 004 , 7 69 per 1 00,00 0 people were perform ed, for a total of 96 ,1 3 4 sleep studies. Based on

prev alence estim ates of the Wisconsin Sleep Cohort Stu dy , it is estim ated tha t in Ontar io, 92 7 ,1 05

people aged 30 to 60 y ears hav e sleep disordered breath ing. Th us, ther e ma y be a 1 0- fold rise in the

ra te of sleep tests ov er th e next few y ears.

Of the 7 2,9 41 patients (mean ag e = 48 y ears) wh o un derw ent s leep studies between 2000 a nd

20 04 , the nu m ber of studies/patients ran ged from tw o (qua rti le 1 ) to four (qua rti le 3). In 6 0,82 2(83 %) patients, PSG w as perform ed. Man y patients had m ultiple diagn oses. Of 83 ,2 5 4 patient

diagnoses, 38 ,383 (46 %) were unkn own , 31 ,27 3 (3 7 %) were related to psy chiatric condit ions (e.g . ,

anxiety , depression), 1 1 ,827 (14 %) were r elated to congenita l condit ions, and th e rest w ere related

to other system s.


14/22



In 2 004, a t least one PSG ( lev el 1) w as done in 6 2,4 98 patients . Of these, 1 0,7 02 (17 %) patients

un derwent CPAP t i trat ion study (wh ich indicates that th ey were diagnosed with OSA), 12 (0.02%)

patients had lev el 2 PSG, 2,6 7 7 (4%) patients had m ult iple sleep latency tests ( indicated when

na rcolepsy is suspected), an d 7 62 (1 .2 %) patients had m ainten an ce of sleep w akefulness tests

(indicated to determ ine the ability to stay aw ake in select cases, for exam ple, factory workers/tru ck

driv ers). Thus, the uti lity of PSG in 48, 34 5 (7 7 %) patients is un clear. Th is ra ises the question

w h et h er PSG is bei n g a pp r opr ia tel y u t il ize d in On ta r io.

Budget Impact Analysis In 2 004 , approxim ately 96 ,000 sleep stu dies w ere condu cted in Onta rio at a total

cost of ~$47 m illion (Cdn). The cost of baria tric surg ery is $17 ,35 0 (Cdn) per patient. In 200 4,

Ontario spent $4.7 m i l lion (Cdn) per y ear for 2 7 0 patients to undergo bariatric surgery in the

provin ce, and $8.2 m illion (Cdn) for 2 2 5 patients to seek out -of-countr y trea tm ent. Shifting costs

from sleep studies to bariatric surgery would benefi t m ore patients with OSA and m ay a lso prev ent

health consequen ces related to diabetes, hy pertension, an d hy perlipidem ia. It is estim ated tha t the

an nu al cost of treating comorbid conditions in m orbidly obese patients often ex ceeds $10,0 00 (Cdn )

per patient. Thu s, the dow nstream cost sav ings could be substan tial.

Cost-Effectiveness Analysis The objectiv e of this ana ly sis was to address the th ird research question: Ar e

sleep labora tory studies cost-effectiv e?

The a na ly sis focused on OSA, t he predomina nt t y pe of sleep disorder, w hich, in contra st to

literatu re-based estimat es, is diagn osed in approxim ately 2 3% of all patients tested w ith PSG in

Ontario. The mean age of OSA patients is 50 10 y ears and m ean BMI is 29 4.5 kg/m . Using

cum ulativ e density function a nd assum ing th at BMI are n orm ally distributed, i t was est ima ted

that out of these 23 %, 11 % hav e a BMI greater th an 3 5 kg/m (morbid obesi ty ) . The treatm ent of

choice for OSA pat ients is CPAP and th e trea tm ent of choice for m orbidly obese patients is bariatr ic

sur gery . The treatm ent of com orbid condit ions is usual ly v ia phar m acological m easur es.

Thr ee strategies wer e com pared:

1 . The cu rr ent pra ctice of referr ing all sleepy patients to a sleep labora tory for PSG. Patients in

w h om OS A is di a g n osed a r e fol low ed u p w it h a CPA P t it r a t ion test a n d li fe-lon g CPA P t h er a py

w it h y ea r ly slee p test s a n d CPA P dev ic e r ep la cem en t ev er y 5 y ea r s.

2. An a lternate strategy that l inks the curr ent practice with obesi ty control strategy : 8% of

OSA patients wh o are also m orbidly obese are offered baria tric surg ery each y ear a s per

current ca pa city .

3 . A n ew stra tegy in w hich sleep tests are n ot offered but a CPAP trial is offered. Patients

diagnosed with OSA are treated with CPAP therapy and 9 0% of OSA patients who are a lso

m orbidly obese are offered bariatric surgery each y ear.

Using a Mar kov m odel (Figur e 2 ), a cohort (m ean a ge = 5 0 y ears) wa s followed for its entire li fe

span, i .e. , from 5 0 to 85 y ears of age (a total follow -up of 35 y ears). It w as assum ed that:

Figure 2:

Markov Model

CPAP tria l is as accur ate as PSG in diagn osing OSA ;

Patients w ho are treated w ith CPAP would ha v e im prov ed qua lity of l i fe but would require

lifelong CPAP thera py ;

1 % to 5% of patients on CPAP ma y be cur ed of OSA th rough li festy le m odification (diet an d

2

2
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig02/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig02/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig02/?report=objectonly


15/22



exercise);

87 % of patients who would undergo bariatric surgery would be cured of OSA a nd w ould no

long er require CPAP (they would also no long er require m orbid obesity -related car e); and

A ll pa t ien ts w ou ld be a li v e du r in g 3 5 y ea r s of fol low -u p.

The costs of sleep tests ($506 /test), CPAP dev ices ($81 7 /dev ice), and of bariat ric sur gery

($17 ,000/patient) ar e all expressed in Cana dian dollar s. Ann ua l costs related to m orbid obesity

($1 0,000/patient) w ere a lso included; howev er, the m odel was run both w ith an d without m orbid

obesity -related costs. The outcome w as QALY, wh ich w as com puted u sing the Tu fts-New Eng land

Medical Center, Institute for Clinica l Research an d Health Policy Stu dies Catalog of Preference

Sc ores . (40) Thu s, the uti l i ty v alue of untreated OSA w as 0.63, CPAP-treated OSA w as 0.87, a nd

cured OSA or n o OSA w as 1 .00.

The m odel ev alu ated thr ee stra tegies in a sleepy patient (ag e = 5 0 y ears) in a Mark ov process of

t ime cy cle. Each cy cle was of 1-year duration. In th e fi rst s trategy , the patient went t hrough

standar d sleep laboratory testing, follow ing w hich the patient could tr an sit into one of tw o Mar kov

states: OSA or No OSA. T he probability of enter ing an OSA state wa s 2 3%, a nd the probability of

entering No OSA w as 77 % (a com plem entary probabi li ty denoted by the # sign). No OSA w as

an a bsorbing state; the patient could not retur n from th at state. If the patient h ad OSA, then th epatient could get CPAP therapy with 9 9% probabi l ity . There wa s a 1% chan ce that the patient

w ou ld n ot r ec ei v e th er a py (e. g ., in ca se th e pa tie n t r efu sed) . Th er e w a s a 1 % ch a n ce th a t th e

patient m ight be cu red of OSA (th rough li festy le modification). If the patient w as cured, th e patient

be g a n th e n ex t t im e cy cle in th e No OS A sta te. If t h e pa tie n t w a s n ot cu r ed, th e pa tie n t be g a n th e

next t im e cy cle in th e OSA state and w ent thr ough the same process . At the end of each t im e cy cle,

the patient ac cum ula ted som e v alu e for QALY depending u pon th e cour se the patient took durin g

th a t cy c le.

In th e second strateg y , there w ere thr ee Mark ov states, the th ird of which arose from a subdiv ision

of OSA state into OSA (w ithout m orbid obesity ) state a nd OSA_MO state to distingu ish

m orbidly obese patients from non-morbidly obese OSA patients. Th e m orbidly obese patient could

get bariatric surgery , w hich w ould cur e or n ot cur e the condit ion. I f cured, the patient began the

next cy cle in the No OSA state, otherw ise in the OSA state or OSA_MO state, depending u pon

the patient s cur rent state. In the third strategy , the patient w ent through the same branch ing

cascade as in the second strateg y .

To accoun t for un certainty in para m eter est ima tes, probabi l ist ic sensit iv i ty analy ses were

perform ed by ca rr y ing out 1 0,000 Monte Car lo sim ula tions upon the Mar kov m odel. In th is process,

v a lu es w er e sim u lt a n eou sly sa m pl ed for a ll u n cer ta in pa r a m et er s fr om a pp r opr ia te di st r ib u tion s.

A ll fu tu r e cost s a n d QA LY s w er e di scou n ted a t a 3 % a n n u a l r a te. Th e soft w a r e pa cka g e u sed for

these analy ses wa s Tree Age Pro 2005. The r esults are sum m arized in Table 6 and graphical ly

presented in Figur e 3 .

Table 6:

Results of the Cost-Effectiveness Analyses

Figur e 3:

Probabilistic Analyses

*
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/#A01ref40http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig03/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig03/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab06/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab06/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig03/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab06/?report=objectonly


16/22



The r esults show th at w hen m orbidity costs were included, the m ean incr emental cost (Cdn) of the

second strategy com pared with th e fi rst w as -$3,1 68 (95 % probabi li ty interv al = -$2,57 0 to -

$3,7 61 ) , and the m ean increm ental cost of the third strategy com pared with the f irst strategy wa s -

$6,9 08 (-$6,038 to -$7 ,7 65 ). When m orbidity costs were excluded, the corresponding m ean

increment al costs were -$142 (-$1 1 6 to -$1 68) a nd -$2,5 1 3 (-$2,465 to -$2,56 3). Th us, both th e

second and th ird stra tegies ar e cost-sav ing compar ed w ith the first stra tegy (cur rent pra ctice) and

this conclusion does not chan ge by inclu sion or exclu sion of m orbid obesity costs, alth ough cost

sav ings are gr eater w hen these costs are included.

The r esults a lso show th at th e m ean incr ementa l QALYs for the second strategy com pared with the

f irst was 0.26 (0.24 0.27 ), and the m ean increm ental QALYs for th e third strategy com pared with

the first stra tegy w as 0.28 (0 .2 7 0.3 0). Hence, l inking sleep clinics to obesity clinics would not

only result in g ains in QALYs but also cost-sav ing.

Comparison of Ontario-Based Economic Analysis With Other Economic Studies

The Ont ario results ar e not directly compar able to prev ious economic a na ly ses because of

differen ces in an aly tic approaches. That said, both Pelletier-Fleur y et al. (3 7 ) and Albarra k et a l .

(3 8 ) foun d that costs w ere high er in OSA patients partly due to high er uti lization of health car eresour ces secondar y to com orbid conditions. It wa s demonstrated t ha t t hese costs could be

m inim ized by linking sleep clinics to obesity clinics. Ay as et al. (3 9 ) exam ined the effect of CPAP

therapy on a ccident ra tes com pared with n o therapy. Accident r ates were not modeled in th e MAS

m odel because in th e three exam ined stra tegies, patients receiv ed CPAP ther apy . Thu s, in this

m odel all patients had a sim ilar att ention span.

Conclusions

Obesity , rath er tha n OSA, leads to car diov ascular consequen ces. Treatin g an d prev enting obesity

w ou ld su bsta n tia ll y r ed u ce th e ec on om ic bu r den a ssoc ia ted w it h di a be tes, h y pe r ten sion ,

hy perl ipidemia, an d OSA. Prom otion of healthy weight m ay be achieved by a m ult isectoria l

approach as recomm ended by t he Chief Medical Officer of Health for Ontar io. Bar iatric sur gery ha s

a m ajor role in m orbidly obese indiv iduals (BMI > 3 5 kg/m an d a comorbid condition, or BMI > 40

k g/m ).

Habitu al snorers with excessiv e day tim e sleepiness ha v e a high pretest probability of ha v ing OSA.

These patients ma y be offered a th erapeut ic trial of CPAP to diagnose OSA, ra ther tha n a PSG. A

m ajority of these patients are also obese and m ay benefit from w eight loss. Thus, indiv idualized

w ei g h t loss p r og r a m s sh ou ld be offe r ed , a n d pa tie n ts w h o a r e m or bi dl y obe se sh ou ld be offe r ed

ba r ia tr ic su r g er y . Th a t sa id , a n d in v ie w of t h e st il l ev olv in g u n de r st a n di n g of t h e ca u ses,

consequen ces an d optima l treatm ent of OSA, furth er research is w ar ran ted to identify w hich

patien ts shou ld be screen ed for OSA.

Appendix

Sea rch da te : Februa ry 28 , 2 0 0 6

Dat ab ases sea rc h ed: OV ID MEDLINE, MEDLINE In -Pr ocess & Oth er Non-In dexed Cita tion s, EMBA SE,

Cochran e Library , INAHTA

Data base: Ov id MEDLINE(R)

Sea rch Stra teg y :

--------------------------------------------------------------------------------

1 . *Sleep Apnea, Obstru ctive/ (25 66 )

2

2


17/22



2 . (sleep adj (apnea or apnoea) adj3 (r esistan ce or obstru cti$)).m p. [mp=title, origin al title,

abstract , nam e of substance word, subject h eading word] (482 8)

3 . 1 or 2 (4 8 2 8 )

4. exp Poly som nography / (494 3)

5 . 3 a n d 4 (1 7 3 1 )

6 . l im i t 5 to (h um a ns a nd eng l ish la ng ua g e a nd y r=20 0 4 - 20 0 6 ) (4 0 9 )

7 . (sy stem atic rev iew$ or m etaanaly sis or m eta-ana ly sis) .mp. [m p=ti t le, original t i t le ,

abstract , n am e of substance word, subject heading w ord] (22 1 1 3)

8 . 6 a n d 7 ( 2 )

9 . 6 ( 4 0 9 )

1 0. limit 9 to (case reports or comm ent or editorial or letter or rev iew ) (80)

1 1 . 9 n ot 1 0 ( 3 2 9 )

1 2 . 8 or 1 1 ( 3 3 0 )

1 3. l imit 12 to diagnosis (sensi t iv i ty ) (21 5 )

Data base: EMBASE

Sea rch Stra teg y :

--------------------------------------------------------------------------------

1 . *Sleep A pnea Sy ndrom e/ (7 8 5 7 )

2 . (obstructi$ or resistan ce).m p. [mp=title, abstra ct, subject heading s, heading w ord, dru g

tra de na m e, o r ig i na l t it l e, dev i ce m a nuf a cturer , drug m a nuf a cturer na m e] (4 0 8 4 0 4 )

3 . 1 a n d 2 (5 3 6 1 )

4. exp POLYSOMNOGRAPHY/ (61 87 )

5 . 3 a n d 4 (1 9 2 9 )

6 . l im i t 5 to (h um a n a nd eng l ish la ng ua g e a nd y r=20 0 4 - 20 0 6 ) (4 0 5)

7 . (sy stem atic rev iew$ or m eta-analy sis or m etaanaly sis) .mp. [m p=ti t le, abstract , subject

headings, heading word, dru g tra de nam e, original t i t le, dev ice m anu facturer, drug

m a n u fa c t u r er n a m e ] (3 6 8 5 0 )

8 . 6 a n d 7 ( 4 )

9 . 6 ( 4 0 5 )

1 0. l imit 9 to (edi toria l or letter or note or r ev iew) (85 )

1 1 . C a se Report/ (8 7 5 3 5 1 )

1 2 . 9 not (1 0 or 1 1 ) (28 8 )

1 3. l imit 12 to diagnosis (sensi t iv i ty ) (15 3)

Glossary

A pn ea -h y pop n ea in dex Th e su m of a pn ea s a n d h y pop n ea s per h ou r of sle ep .

Bod y h a b it u s T h e ph y si qu e or b od y b u i ld .

Body m ass index An index that relates body w eight to height. The body m ass index (BMI) is

obtained by dividing a persons weight in ki logram s (kg) by their height in m eters (m ) squar ed.

C ont i nuous posi t iv e a irwa y pressure A tech ni que of respira tory th era py i n wh i ch a irwa y

pressure is ma intained abov e atm ospheric pressure thr ough out the r espiratory cy cle by

pressur ization of the v entilatory circu it.

Obstru ctiv e s leep apnea The repeti t ive com plete obstru ction (apnea) or part ia l obstru ction

(hy popnea) of the collapsible part of the u pper airw ay dur ing sleep.

Poly s om nog ra ph y Sim ul ta neous a nd cont i nuous m o ni tor i ng of norm a l a nd a bnorm a l

phy siological activ ity dur ing sleep, inclu ding the apnea-hy popnea index (AHI) and respiratory

disturbance index (RDI).

Respira tory d isturba nce i ndex Th e sum of a pnea s, h y popnea s, a nd a bnorm a l respira tory events

per h our of sleep.


18/22



Notes

Suggested Citation

Th is report should be cited a s follow s:

Medical Adv isory Secretar iat. Poly som nogra phy in patients with obstru ctiv e sleep apnea: an

ev idence-based analy sis. Ontario Health Technology As ses sm ent Series 2 0 0 6 ; 6 ( 1 3 ) .

Permission Requests

A ll in qu ir ie s r eg a r di n g pe r m ission to r ep r odu ce a n y con ten t in th e Ontario Health Technology

As ses sm en t Series should be directed to MASinfo.moh/at/ontario.ca.

How to Obtain Issues in the Ontario Health Technology Assessment Series

A ll r ep or ts i n th e Ontario Health Technology As ses sm ent Series are freely av ai lable in PDF form at a t

th e followin g URL:w w w .h ea lt h .g ov .on .ca /oh ta s.

Print copies can be obtained by contacting MASinfo.moh/at/onta rio.ca.

Conflict of Interest Statement

A ll a n a ly ses i n th e On ta r io Hea lt h Tec h n olog y A ssessm en t Ser ies a r e im pa r tia l a n d su bj ec t to a

system atic ev idence-based assessment process. There ar e no competing interests or conflicts of

interest to declare.

Peer Review

A ll Medic a l A dv isor y Sec r et a r ia t a n a ly ses a r e su bj ec t to ex ter n a l ex pe r t pe er r ev ie w . A dd it ion a ll y ,

the public consultation process is also av ailable to indiv iduals wishing to com m ent on an a na ly sis

prior to f ina lization. For m ore in form ation, please v isit

htt p://w ww .health .gov .on.ca/eng lish/prov iders/program /ohtac/public_enga ge_ov erv iew.h tm l.

Contact Information

The Medical Adv isory Secreta riat

Ministry of Health an d Long-Term Care

20 Dun das Street West, 1 0th floor

Toronto, Onta rio

CANADA

M5 G 2N6

Em ail: MASin fo.m oh/at /onta rio.ca

Telephone: 416-314-1092

ISSN 1 9 1 5- 7 3 9 8 (Onl i ne)

ISBN 97 8-1-42 49 -43 1 6-6 (PDF)

About the Medical Advisory Secretariat

The Medical Adv isory Secretar iat is part of the Ontar io Ministry of Health a nd Long -Term Car e. The

m anda te of the Medical Adv isory Secretar iat is to prov ide ev idence-based policy adv ice on th e

coordina ted uptake of health serv ices an d new health techn ologies in Onta rio to the Ministry of
http://www.health.gov.on.ca/english/providers/program/ohtac/public_engage_overview.htmlhttp://www.health.gov.on.ca/ohtas


19/22



Health a nd Long -Term Care an d to the hea lthca re system . The aim is to ensure th at r esidents of

Ontario hav e access to the best av ai lable new health technologies that wi l l im prov e patient

outcomes.

The Medical Adv isory Secretar iat also prov ides a secretariat funct ion a nd ev idence-based health

technology pol icy a naly sis for r ev iew by the Ontar io Health Technology Adv isory Com m ittee

(OHTAC).

The Medical Adv isory Secretar iat condu cts sy stemat ic rev iews of scientific ev idence and

consultat ions with experts in th e health care serv ices com m un ity to produce th e Ontario Health

Technology Asses sment S eries.

About the Ontario Health Technology Assessment Series

To conduct i ts com prehensiv e ana ly ses, th e Medical Adv isory Secretariat systematical ly rev iews

av ai lable scienti f ic l i teratur e, col laborates with partners across relev ant govern m ent bra nches,

and consults with cl inical an d other external experts and m anu facturers , and sol icits any necessary

adv ice to gath er inform ation. T he Medical Adv isory Secretariat m akes ev ery effort to ensure th at

al l relev ant r esearch, n ational ly and interna tional ly , i s included in the systematic l i teratur e

rev iews conducted.

The inform ation gat hered is the foun dation of the ev idence to determ ine if a technology is effectiv e

an d safe for u se in a particu lar clinical popula tion or setting. Inform ation is collected to un derstand

how a new technology f its within cu rrent practice and treatm ent a lternat ives . Detai ls of the

techn ology s diffusion into curr ent prac tice and input from pra cticing m edical experts and industry

add importan t inform ation to the rev iew of the prov ision an d deliv ery of the hea lth tech nology in

Onta rio. Inform ation concer ning the healt h benefits; econom ic and hu m an r esources; an d ethical,

regu latory , social an d legal issues relatin g to the techn ology assist policy m akers to m ake tim ely

an d relev an t decisions to optim ize patient out com es.

If you ar e awa re of any cur ren t additiona l ev idence to inform an existing ev idence-based an aly sis,

please conta ct th e Medical Adv isory Secretar iat: MASinfo.moh/at/onta rio.ca. Th e public

consulta tion process is also av ailable to indiv iduals wishing to com m ent on an an aly sis prior to

publication. For m ore inform ation, please visit

htt p://w ww .health .gov .on.ca/eng lish/prov iders/program /ohtac/public_enga ge_ov erv iew.h tm l.

Disclaimer

This evidence-based an aly sis w as prepar ed by the Medical Adv isory Secretar iat, Onta rio Ministry of

Health an d Long-Term Care, for t he Ontar io Health Technology Adv isory Com m ittee and developed

from an aly sis, interpreta tion, a nd compa rison of scientific research and/or tech nology assessment s

conducted by other org an izations. It also incorpora tes, w hen a v ailable, Ontar io data, a nd

inform ation prov ided by experts and applicants to the Medical Adv isory Secretar iat to inform the

an aly sis. While ev ery effort h as been m ade to reflect all scientific research av ailable, this docum ent

m ay not ful ly do so. Addit ional ly , other relev ant scienti f ic f indings m ay hav e been reported s ince

completion of the rev iew. Th is ev idence-based an aly sis is curr ent to the date of publication. Th is

an aly sis may be superseded by an u pdated publication on the sam e topic. Please check th e Medical

A dv isor y Sec r et a r ia t Web sit e for a li st of a ll ev id en ce-b a sed a n a ly ses:

http://ww w.h ealth.gov .on.ca/ohtas.

List of Tables

Table 1: Qualit y of Evi dence of Incl uded Studies

Table 2: Characteri stics of the Part ic ipants wh o Completed One or Both Foll ow-Up Sleep Studies, According to the A pnea-

Hypopnea Index at Baseli ne

Table 3: Parti cip ant Characteri stics (Arzt et al . 2005)

*
http://www.health.gov.on.ca/ohtashttp://www.health.gov.on.ca/english/providers/program/ohtac/public_engage_overview.htmlhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab03/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab02/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab01/?report=objectonly


20/22



Table 4: Parti cip ant Characteri stics (Marin et al. 2005)

Table 5: Baseli ne Characteri stics of Patients Wi th Obstructi ve Sleep A pnea Syndrome and Controls

Table 6: Results of th e Cost-Effecti veness Analy ses

List of Figures

Figure 1 : Cumulati ve Percentage of Indiv iduals Wi th New Fatal (A) and Nonfatal (B) Cardiovascul ar Events i n Each of the Fi ve

Groups

Figure 2: Mar kov Model

Figure 3: Probabilistic Analyses

Abbreviations

A HI A pn ea h y pop n ea in de x

BMI Body m a ss in dex

C C OHTA C a na dia n C oordina t i ng O ffi ce of Hea l th Tech nolog y A s sessm ent

CI Con fiden ce in t er v a l

C MS C enters for Medi ca re a nd Medi ca i d Serv i ces

C PA P C on t in u ou s posi ti v e a ir w a y p r essu r e

O CC I O n ta r i o C a se C ost in g In i ti a ti v e

OR Odds r a t io

OS A O bst r u ct iv e sl eep a pn ea

PSG Poly som n og ra ph y

QA LY Qu a l it y -a dj u st ed life y e ar

RDI Re sp ir a t or y d ist u r b a n ce in d ex

RO C Re ce iv e r op er a t in g c h a r a ct er i st ic s

SD St an da rd dev i at ion

SE Sta nda rd er ror

Article information

Ont Health Technol Assess Ser. 2006 ; 6(13): 1 38.

Published online 2006 June 1 .

PMCID: PMC3379160

Health Quality Ontario

Copyright 2006, The Me dical Advisory Secretariat

Articles from Ontario Hea lth Techno logy Assessment Series are provided here courtesy of Health Quality Ontario

References

1. Y oun g T, Pal ta M, Dempsey J, Skatru d J, Weber S, Badr S. The occurrence of sleep-disordered breathin g a mong

mi ddle-aged adul ts. N Engl J Med. 1993;328(17):12301235. [PubMed]

2. Coll op NA . Obstru ctive sl eep apnea: wha t does the cardiovascu la r phy sici an need to know? Am J Cardi ovasc Drug s.

2005;5(2):7181. [PubMed]

3. Tran D, Wal la ce J. Obstru ctive sleep apnea syn drome in a publ icl y fu nded health care system . J Na tl Med Assoc.

2005;97(3):370374. [PMC free arti cle] [PubMed]

4. Flemons WW, Doug la s N J, Kuna ST, Rodenstein DO, Wheatley J. Access to diagn osis a nd treatm ent of patients wi th

su spected sleep apnea . Am J Respir Cri t Care Med. 2004;169(6):668672. [PubMed]

5. Am erican Aca demy of Sleep Medicine Task Force. Sleep-related breathin g disorders in adul ts: recomm endations for

syndrome definition an d measurement techniques in cli nical research. The Report of an A merican Academy of Sleep

Medici ne Tas k Force. Sleep. 1999;22(5):667689. [PubMed]

*
http://www.ncbi.nlm.nih.gov/pubmed/10450601http://www.ncbi.nlm.nih.gov/pubmed/15003950http://www.ncbi.nlm.nih.gov/pubmed/15779501http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2568633/?report=readerhttp://www.ncbi.nlm.nih.gov/pubmed/15725038http://www.ncbi.nlm.nih.gov/pubmed/8464434http://www.ncbi.nlm.nih.gov/pmc/about/copyright.htmlhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig03/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig02/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/figure/A01fig01/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab06/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab05/?report=objectonlyhttp://www.ncbi.nlm.nih.gov/pmc/articles/PMC3379160/table/A01tab04/?report=objectonly


21/22



6. Hail ey D, Jacobs P, Mayers I, Mensinkai S. Auto-titrati ng nasal contin uous positive a irway pressure systems i n the

ma na gement of obstru ctive sl eep apnea [report on th e Internet] [Technology Report Issue 39]. Ottawa , Onta ri o:

Can adi an Coordi na tin g Office for Health Technology A ssessm ent (CCOHTA); Sep, 2003. [[cited 2006 Mar . 15]].

Avai l abl e at: http://www.cadth.ca/media/pdf/202_autocpap_tr_e.pdf .

7. Centers f or Medicar e and Medicai d Services (CMS) Conti nu ous positi ve ai rway pressure (CPAP) thera py for

obstru ctive sl eep apnea (OSA) [report on th e Internet] 2005. [[cited 2006 Jul y 1]]. Ava il able a t:

http://www.cms.hhs.gov/transmittals/downloads/R35NCD.pdf.

8. Mador MJ, Kuf el TJ, Maga la ng

Polysomnography in Patients With Obstructive Sleep Apnea

Documents