ADVERTISEMENT

Issue Navigator

Volume 14 No. 02
Earn CME
Accepted Papers
Classifieds





Scientific Investigations

Clinical Trial Enrollment Enrichment in Resource-Constrained Research Environments: Multivariable Apnea Prediction (MAP) Index in SCIP-PA Trial

Hyunju Yang, PhD, RN1; Alexa Watach, PhD, RN1,2; Miranda Varrasse, PhD, RN2,3; Tonya S. King, PhD4; Amy M. Sawyer, PhD, RN1,3
1Penn State University College of Nursing, University Park, Pennsylvania; 2University of Pennsylvania Perelman School of Medicine, Center for Sleep & Circadian Neurobiology, Philadelphia, Pennsylvania; 3University of Pennsylvania School of Nursing, Philadelphia, Pennsylvania; 4Penn State University College of Medicine, Department of Public Health Sciences, Hershey, Pennsylvania

ABSTRACT

Study Objectives:

Determine the Multivariable Apnea Prediction (MAP) index predictive utility for enrollment enrichment in a clinical trial wherein enrollment was prior to obstructive sleep apnea diagnosis.

Methods:

Secondary analysis of screening data (n = 264) from randomized, double-blind, pilot trial. Clinical sleep center patients with complete screening and polysomnography data were included. To determine diagnostic test accuracy of the MAP index using apnea-hypopnea index criterion ≥ 10 events/h (primary) and ≥ 5, ≥ 15, and ≥ 30 events/h (secondary), sensitivity, specificity, negative and positive predictive values, likelihood positive and negative ratios, and receiver operating characteristic curves were calculated. Predictive utility was examined by characteristic variables.

Results:

Middle-aged, overweight or obese, men and women were included. Employing a MAP index threshold ≥ 0.5, sensitivity for obstructive sleep apnea (apnea-hypopnea index ≥ 10 events/h) was 83.6%; specificity was 46.4%; area under the curve = 0.74. Sensitivity was higher in males than females (95.3%, 68.7%, respectively); specificity was lower in males than females (30.4%, 57.6%, respectively) with similar area under the curve (0.74 versus 0.72, respectively). MAP accuracy was higher in younger versus older adults (younger than 50 years, or 50 years or older; area under the curve 0.82 versus 0.63, respectively). Varied apnea-hypopnea index criteria produced stable accuracy estimates.

Conclusions:

Recruitment/enrollment is a high-cost endeavor. Screening procedures may confer resource savings but careful evaluation prior to study implementation assures effectiveness and efficiency.

Clinical Trial Registration:

The secondary analysis reports data from the SCIP-PA Trial (NCT 01454830); study information available at: https://clinicaltrials.gov.

Citation:

Yang H, Watach A, Varrasse M, King TS, Sawyer AM. Clinical trial enrollment enrichment in resource-constrained research environments: Multivariable Apnea Prediction (MAP) index in SCIP-PA Trial. J Clin Sleep Med. 2018;14(2):173–181.


BRIEF SUMMARY

Current Knowledge/Study Rationale: In the current resource-constrained research environment, pre-enrollment screening in studies of obstructive sleep apnea is likely to increase enrollment efficiency and may also reduce protocol resource expenditure. An obstructive sleep apnea screening tool with high predictive utility, Multivariable Apnea Prediction (MAP) index, has been shown to have high clinical predictive utility; however, the use of MAP index for research enrollment enrichment is not well understood.

Study Impact: Results suggest the MAP index can be used for clinical trial enrollment enrichment but consideration of the target population characteristics, including prevalence of obstructive sleep apnea at the study-defined apnea-hypopnea index criterion, sex and age, are important prestudy factors to consider so that the overall utility of employing MAP index as an enrollment enrichment strategy in study settings is optimized.

INTRODUCTION

Obstructive sleep apnea (OSA) is associated with increased risk of cardiovascular morbidity and mortality and all-cause mortality.13 Twenty-five percent of adults in the United States are estimated to be at high risk for OSA,4 but up to 90% of this population is undiagnosed.5 Polysomnography (PSG) is considered the gold standard diagnostic test for OSA6; yet, the complexity, burdensome nature for patients, limited access to sleep diagnostics, and expense of the procedure are prohibitive of using PSG for OSA screening.7 Although several simple OSA screening tools exist and have been tested for clinical predictive utility,810 little is known about the use of OSA screening instruments for research enrollment enrichment.

Recruitment is critical to the success of any research study. Clinical research is conducted in practice settings, includes patients representing the target population, and research procedures are situated within the context of clinical services.11 Though this would seem to be directly supportive of research recruitment, target population characteristics and particular nuances of how, when, and where patients interface with the clinical enterprise are essential to understand at the outset of study planning. In the current cost-constrained health care environment, largely dependent on third-party payer reimbursement policy, face-to-face clinical engagement with patients in sleep centers is increasingly curtailed,12 which further complicates traditional recruitment approaches.

Among adults clinically suspected of having OSA, the diagnostic pathway is one example of curtailed clinical engagement. In-laboratory PSG for OSA is less often utilized and more commonly, OSA diagnostic testing has transitioned to the use of home sleep apnea tests.13 This health care delivery model minimizes direct clinical engagement within the sleep clinic.12,14 The advent of telemedicine also moves clinical care from a traditional face-to-face model to care intervals outside the sleep clinic walls.15 Though these trends are progressive and may be cost-conserving, such system-level advances necessitate that clinical researchers develop and align effective recruitment strategies.

The conduct of clinical trials and large prospective, observational studies are essential to generating knowledge and supporting evidence-based practice recommendations for OSA. Though both study designs produce high levels of evidence,16 study resources are often constrained. Recruitment and enrollment significantly escalate study-related costs. For example, the Best Apnea Interventions for Research (BestAIR study), a National Institutes of Health-supported planning study to address challenges of large-scale trials of sleep apnea treatment, estimated expenses for recruiting and screening potential OSA participants through face-to-face contact to be $2,139 per randomized subject and $647 per randomized subject with a mail-based recruitment strategy.17 The BestAIR planning study recruited both patients who were clinically suspicious for OSA and patients with confirmed OSA; similarly, many OSA studies necessitate enrolling participants prior to confirming the presence of OSA. Prediagnosis enrollment can be risky, as a potentially large number of participants may be enrolled and then excluded when OSA is ruled out.

The Multivariable Apnea Prediction (MAP) index is a screening tool developed for predicting sleep apnea risk in clinical populations.18 The predictive probability of the MAP index has been shown to have high sensitivity and specificity. Studies examining the MAP index predictive probability have focused on clinical screening of OSA in commercial truck drivers,19,20 general clinical populations in sleep centers,18,21 cardiovascular patients,22 and a population with mild cognitive impairment.23 Employment of the MAP index may enrich study enrollment by screening out participants not likely to have OSA while screening in patients likely to have OSA; such a pre-enrollment screening strategy is likely to identify those who will progress per-protocol in the setting of a study sampling for OSA.

The objective of this secondary analysis of data from a single-site, randomized, double-blind pilot clinical trial24 was to examine the overall utility of the MAP index as an enrollment enrichment strategy. We examined the predictive utility of the MAP index, including sensitivity and specificity, in the setting of clinical trial recruiting/enrolling adult clinical sleep center patients prior to OSA diagnosis. Next, we examined the utility of the MAP index stratified by characteristics (ie, sex, body mass index, age) and by varied apnea-hypopnea index (AHI) outcome criteria.

METHODS

Secondary analysis data were derived from a registered (NCT 01454830) randomized, double-blind, parallel pilot controlled trial that examined the feasibility and established an effect size of a treatment self-efficacy tailored intervention to improve adherence to positive airway pressure (PAP) therapy in adults with newly diagnosed OSA.24 The study was approved by the Institutional Review Board; participants gave their informed consent and permission for use of private health information.

Sample

Study participants were recruited at a clinical sleep center at a suburban tertiary academic medical center. Clinical patients referred to OSA diagnostics were approached to participate in the trial (n = 431) by (1) a study-affiliated research assistant at the time of diagnostic PSG; or (2) by mailed letter of invitation sent to patients scheduled for in-laboratory diagnostic PSG for OSA. Complete data for the evaluable outcomes of the secondary analysis were available for 264 study volunteers. In the parent trial, convenience sampling was used with the following inclusion criteria: (1) newly diagnosed, with AHI ≥ 10 events/h; (2) PAP-naïve; (3) age 18 years or older; (4) able to read/speak English. Exclusion criteria were: (1) previous OSA diagnosis and/or treatment; (2) medical record documented new psychiatric diagnosis within 6 months of enrollment; (3) requirement of supplemental oxygen or bilevel PAP identified on PAP titration PSG; (4) diagnosis of another sleep disorder in addition to OSA based on PSG (ie, periodic limb movement disorder, central sleep apnea, insomnia, sleep hypoventilation syndrome, or narcolepsy). All patients screened for study enrollment were eligible for inclusion in the secondary analysis. Secondary analysis inclusion criteria were (1) complete MAP index; and (2) PSG data (Figure 1).

Enrollment and evaluable data flowchart.

AHI = apnea-hypopnea index, MAP index = Multivariable Apnea Prediction index.

jcsm.14.2.173a.jpg

jcsm.14.2.173a.jpg
Figure 1

Enrollment and evaluable data flowchart.

(more ...)

Measures

Multivariable Apnea Prediction Index

The MAP index is a screening tool for OSA.18 The tool takes less than 1 minute to complete and combines self-reported apnea symptoms and age, sex, and body mass index (BMI) to derive an index of probability for OSA. The MAP items for self-reported apnea symptoms ask respondents to identify the frequency of OSA symptoms during the last month; symptoms include “loud snoring,” “snorting or gasping,” and “cessation of breathing or choking or struggling for breath.” Item responses are rated 0 to 4: 4 = always (5–7 times per week); 3 = frequently (3–4 times per week); 2 = sometimes (1–2 times per week); 1 = rarely (less than once a week); 0 = never; and do not know (missing data point). Responses for the 3 symptom questions are calculated as the mean of the nonmissing values.

The formula for the MAP index is:

jcsm.14.2.173a-e1.jpg
where x = −8.160 + 1.299 × Index 1 + 0.163 × BMI − 0.028 × Index 1 × BMI + 0.032 × Age + 1.278 × Sex, and where Sex = 1 if male and 0 if female, and Index 1 is the mean score of subjective symptom responses.

The MAP index values range from 0–1; with 0 representing the lowest probability of OSA and 1 representing the highest risk of OSA. For the purposes of enrollment enrichment in our pilot clinical trial that was designed, in part, to subsequently provide evidence-based and experience-based data for an optimal MAP cutoff point for a larger clinical trial, a MAP cutoff point of 0.5 was selected for enrollment in the parent pilot clinical trial. This cutoff point was selected based on prior published evidence suggesting high sensitivity and moderate specificity in a clinical sleep population21 and site-specific clinical characteristics of the study target population.

Polysomnography

A single-night, in-laboratory PSG was conducted to diagnose OSA according to established standards for PSG conduct and scoring.25 PSG included 2 electroencephalography leads (C3-A2, C3-O1), bilateral electrooculography, single lead II electrocardiography, surface electromyography of submental and bilateral anterior tibialis muscles, thoracic and abdominal plethysmography, oronasal thermistor and pressure transducer, oxyhemoglobin saturation, and snore microphone. The alternate hypopnea definition (ie, 3% oxygen desaturation from pre-event baseline and/or arousal) was used in scoring hypopneic events.25 The AHI was defined as the total number of apneas and hypopneas per hour of sleep.

Protocol

The parent trial protocol has been previously reported.24 Specific to this secondary analysis, pre-enrollment screening was conducted by two trained study research assistants by telephone interview for patients responding to mailed letters of invitation or face-to-face interview for patients recruited at diagnostic PSG or during a pre-PSG clinical visit. A study-developed automated electronic MAP index scoring program was used to support real-time screening results and minimize risk of calculation errors. Patients were advised to proceed to PSG as scheduled regardless of study eligibility. PSG data was extracted from scored records, including AHI.

Analysis

All research variables and characteristic variables were descriptively analyzed (mean ± standard deviation [SD]), frequencies [n (%)]) for description purposes and assessment of distribution. To determine the diagnostic test accuracy of the MAP, sensitivity, specificity, negative and positive predictive values (NPV, PPV, respectively), likelihood positive ratio, and likelihood negative ratio were calculated. A receiver operating characteristic (ROC) curve was calculated and area under the curve (AUC) with 95% confidence interval (CI) estimated using MAP index as a predictor for AHI ≥ 10 events/h; this AHI threshold was a priori selected as the parent trial sought to enroll adults with OSA who were likely to proceed to PAP treatment. For secondary analysis purposes, we also examined the diagnostic test accuracy of the MAP index for OSA defined by varying AHI cutoffs including ≥ 5 events/h (any OSA), ≥ 15 events/h (moderate or severe OSA), and ≥ 30 events/h (severe OSA). Separate predictive utility analyses (sensitivity, specificity, NPV, PPV) stratified by characteristic variables, including sex, age (younger than 50 or 50 years or older), and BMI (18.5 to < 25; ≥ 25 to < 30; ≥ 30 kg/m2) were also examined using MAP index ≥ 0.5 and AHI ≥ 10 events/h. Statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, North Carolina, United States).

RESULTS

The sample (n = 264) included males and females (n = 133; 50.4%) of middle age (48.8 years ± 12.6) with obesity (mean BMI 35.2 kg/m2 ± 8.9) and mild to moderate OSA (median AHI, 14.3 events/h [interquartile range 32.4]). Seventy percent of screened patients (n = 187) met MAP index threshold criteria (≥ 0.5) (Table 1).

Baseline characteristics of patients.

jcsm.14.2.173.t01.jpg

table icon
Table 1

Baseline characteristics of patients.

(more ...)

Predictive Utility: MAP Index ≥ 0.5 for OSA (AHI Criterion ≥ 10 events/h)

Employing a MAP index score ≥ 0.5 for OSA defined by AHI ≥ 10 events/h, the MAP index sensitivity was 83.6% (95% CI 76.7%, 89.1%) and specificity was 46.4% (95% CI 37.0%, 56.1%). The associated PPV and NPV indicated that more than 65% of those screened with MAP index ≥ 0.5 were accurately identified as having OSA (PPV, 67.9%), or not having OSA (NPV, 67.5%), defined by AHI ≥ 10 events/h (Figure 2, Table 2). The AUC for the ROC curve was 0.74 (95% CI 0.67, 0.80). The predictive function of the continuum of MAP index threshold scores, 0.1 through 0.9, was examined for the diagnostic level of OSA ≥ 10 events/h with optimal diagnostic performance inclusive of 0.4, 0.5 and 0.6 MAP threshold scores (Table 2). As the MAP index score criterion increases (ie, 0.1 to 0.9), the qualifying number of eligible patients, defined by MAP index score only, decreases relative to the PPV; similarly, the proportion that meet the PSG-confirmed AHI and MAP index score criterion for enrollment also decreases (Figure 2).

The relationship between positive predictive value and the number of population retained.

AHI = apnea-hypopnea index, n AHI_10 = the number of population retained who met the AHI ≥ 10 events/h criteria, PPV = positive predictive value.

jcsm.14.2.173b.jpg

jcsm.14.2.173b.jpg
Figure 2

The relationship between positive predictive value and the number of population retained.

(more ...)

MAP index for prediction of OSA (AHI ≥ 10 events/h).

jcsm.14.2.173.t02.jpg

table icon
Table 2

MAP index for prediction of OSA (AHI ≥ 10 events/h).

(more ...)

Among male clinical sleep center patients screened by MAP index ≥ 0.5, employing a threshold for OSA ≥ 10 events/h, sensitivity was higher (95.3%; 95% CI 88.4%, 98.7%) and specificity was lower (30.4%; 95% CI 17.7%, 45.8%) than for all screened patients and females (Table 3). Both PPV and NPV were higher for male than female clinical sleep center patients screened by MAP index. For both younger and older patients (younger than 50 years or 50 years or older), sensitivity was greater than 80%; specificity was lower among screened older sleep center patients (37.5%) compared to younger patients (51.4%). When BMI characteristics of clinical sleep center patients were considered, predictive utility of the MAP index for AHI ≥ 10 events/h was most accurate among overweight clinical sleep center patients (AUC = 0.74) compared to obese clinical sleep center patients (AUC = 0.69); this is particularly true for positive case prediction (positive likelihood ratio, 2.29). Due to rare cases of normal BMI (18.5 to < 25 kg/m2) in the sample (n = 16), MAP predictive utility was estimated with less precision.

MAP index ≥ 0.5 for OSA (AHI ≥ 10 events/h) by characteristics.

jcsm.14.2.173.t03.jpg

table icon
Table 3

MAP index ≥ 0.5 for OSA (AHI ≥ 10 events/h) by characteristics.

(more ...)

Predictive Utility: MAP Index ≥ 0.5 for OSA Defined by Common AHI Criteria

The use of the MAP index in the parent study employed an AHI criterion commensurate with third-party treatment recommendations for progressing to PAP therapy. Other commonly employed AHI criteria were also examined in relationship to MAP index predictive utility (Table 4). Across other AHI criteria examined (AHI ≥ 5 events/h [any OSA], ≥ 15 events/h [moderate and severe OSA], ≥ 30 events/h [severe OSA]), sensitivity of MAP index ≥ 0.5 was at least 79%, with a specificity range of 35.5% to 51.4%. Positive likelihood ratios for all AHI criteria were greater than 1.0 (range 1.33–1.63) and negative likelihood ratios for all AHI criteria were lower, as expected (range 0.35–0.41). AUC for predicting OSA at any criteria was relatively stable (0.73–0.74). When sex characteristics were examined, the MAP index ≥ 0.5 overall predictive utility was similar for males and females, regardless of AHI criteria (Table 5). Negative case identification of OSA was better (ie, lower) in males than females (negative likelihood ratio range, 0.12–0.37 versus range, 0.51–0.59, respectively). Among screened males, specificity was low (range 19.5% to 28.0%) across AHI criteria. When stratified by age (younger than 50 years or 50 years or older), MAP index was more accurate across AHI criteria among younger sleep center patients who were screened, as is indicated by AUC range (0.77–0.82), compared to AUC range (0.61–0.67) for older screened sleep center patients. Sensitivity for MAP index ≥ 0.5 was > 78% for both younger and older screened patients, regardless of AHI criteria (range 80.0% to 88.2%; 78.6% to 85.7%, respectively); specificity had more variability between age strata. Positive likelihood ratio was > 1.0 for all age groups across all AHI criteria (Table 6). When stratified by BMI (overweight; obese), sensitivity was consistently > 89% for obese only regardless of AHI criteria; specificity had greater variability, ranging from 17.0% to 67.6% (overweight and obese). Because of the rarity of cases for normal BMI in the clinical sample screened, imprecise estimates limit any summary conclusions about MAP predictive utility based on this characteristic variable (see Table S1 in the supplemental material).

MAP index for OSA across AHI criteria.

jcsm.14.2.173.t04.jpg

table icon
Table 4

MAP index for OSA across AHI criteria.

(more ...)

MAP index for OSA across AHI criteria by sex.

jcsm.14.2.173.t05.jpg

table icon
Table 5

MAP index for OSA across AHI criteria by sex.

(more ...)

MAP index for OSA across AHI criteria by age.

jcsm.14.2.173.t06.jpg

table icon
Table 6

MAP index for OSA across AHI criteria by age.

(more ...)

DISCUSSION

Using MAP index for AHI ≥ 10 events/h demonstrated adequate discrimination based on AUC of 0.74 (95% CI 0.67, 0.80).26 Our diagnostic accuracy findings are consistent with the MAP validation study that reported MAP index as a clinical OSA screening tool wherein AHI criterion was ≥ 10 events/h (reported AUC = 0.786).18 Discriminatory power of MAP for AHI ≥ 10 events/h was similarly adequate for males (AUC = 0.74) and females (AUC = 0.72), younger adults (younger than 50 years; AUC = 0.82), and overweight adults (BMI ≥ 25 to < 30 kg/m2; AUC = 0.74). For older adults (50 years or older) and obese adults, discriminatory accuracy of the MAP with the study-applied criteria was not acceptable (ie, AUC < 0.70). Varying the OSA outcome criteria (AHI ≥ 5, ≥ 15, and ≥ 30 events/h) did not substantially alter the discriminatory power of the MAP index (0.73, 0.74, and 0.73, respectively). For any OSA (AHI ≥ 5 events/h), discriminatory power was lower in males (AUC = 0.70) than females (AUC = 0.74) but this was not found for any other OSA outcome criterion examined. A recently published review of diagnostic and treatment approaches in OSA summarized the accuracy of screening with MAP based on 2 observational studies; AUCs 0.68 (95% CI 0.67, 0.70) and 0.78 (95% CI 0.71, 0.85) were reported for severe OSA (AHI ≥ 30 events/h and Epworth Sleepiness Scale score > 10); AUC 0.61 (no confidence interval reported) for any OSA (AHI ≥ 5 events/h and Epworth Sleepiness Scale score > 10) was reported.9 Though our outcome criterion was OSA by AHI only, the currently reported diagnostic accuracy is consistent with prior reports.

We used the MAP index as a clinical trial enrollment enrichment tool. The following sequence of a priori considerations and decisions guided MAP index implementation in the pilot trial: (1) MAP index threshold of ≥ 0.5 was selected for protocol enrollment based on prior reports in similar target populations18,21 with reported AHI criterion ≥ 10 events/h, (2) estimated prevalence of OSA in the study population at a suburban tertiary academic medical center sleep center wherein recruitment focused on patients referred for diagnostic PSG for OSA with high clinical likelihood for OSA, (3) study resources for recruitment/enrollment (research assistant time, access to clinical space, and use of clinical PSG data, not research PSG, for study purposes) and (4) phase one of the exposure condition delivered prior to PSG-confirmed OSA criteria being met to continue in the parent trial protocol. Given the a priori considerations, we prioritized sensitivity over specificity. More simply, we prioritized enrolling prediagnosis patients likely to have PSG-confirmed OSA (AHI ≥ 10 events/h) while accepting the risks and associated study costs of enrolling a smaller proportion of prediagnosis patients who met MAP criterion but who would subsequently be determined by PSG to not meet study OSA inclusion criteria (ie, AHI < 10 events/h).

For clinical trial enrollment enrichment purposes, we used the MAP index to screen for OSA prior to PSG-confirmed diagnosis. Because sleep center patient volunteers were scheduled for PSG regardless of study-associated MAP screening, the “risks” associated with the screening procedure were wholly assumed by the trial. In other words, unlike clinical applications of screening for OSA wherein there is a high risk associated with screening when “missed OSA cases” result in undiagnosed OSA, in the case of this trial, all screened patients were already scheduled to progress to diagnostic procedure by PSG. The “risks” for the purposes of this reported trial were such that study resources, including recruitment and screening and first-phase exposure condition resources, would be expended with enrolled subjects who were later excluded for failing to meet AHI inclusion criterion.

Approximately 33% of positive MAP screened patients were enrolled and failed to meet AHI inclusion criterion. As our objective was to enrich enrollment with pre-enrollment OSA screening, a reasonable retention failure rate benchmark for our enrollment enrichment was ≤ 50%. This benchmark is based on the prevalence of newly diagnosed OSA in the site-specific target population (ie, adult patients undergoing diagnostic PSG). In the absence of a known, site-specific diagnostic rate or prevalence rate for any given target population, other investigators may consider nondiagnostic PSG rates for OSA reported in population-based studies, such as the Wisconsin Sleep Cohort Study (WSCS).27 The WSCS oversampled for likely OSA and the prevalence of all OSA AHI ≥ 10 events/h ranged from 5% to 15% for men and women, respectively, aged 30–60 years. A retention failure benchmark based on the WSCS findings would be set at ≤ 70%. Our retention failure rate of 33% demonstrates effectiveness of enriched enrollment with MAP index.

Our experience is derived from a pilot clinical trial, preparatory for a larger trial, and as such, we employed “best available data” to guide decisions for the pilot trial. Our pilot trial was designed to purposefully guide, in part, subsequent MAP index criterion decisions. In preparation for a trial that intends to use the MAP index to enrich enrollment, important considerations for selecting an optimal MAP index criterion will include site-specific, experience-based, or evidence-based MAP utility metrics. However, there are also feasibility considerations that must be taken into account for recruitment of the target population to the study. For example, ideally investigators will select a MAP index criterion that maximizes the PPV; yet, by selecting a MAP index criterion with high PPV, the number of retained patients for enrollment is necessarily lower than other MAP index criteria with slightly lower PPV (see Figure 2). The investigators will then anticipate longer enrollment duration to meet sample size requirements while also assuming responsibility for expenditure of study resources (ie, budget implications) to span a lengthy enrollment time line. In addition, feasibility considerations for enrolling the target population may be dampened by the degree of difficulty that the investigators anticipate to recruit and subsequently enroll eligible patients at the study site. With a MAP index criterion that maximizes PPV, the number of patients retained is smaller and will necessitate access to a larger target population to reasonably argue feasibility at the study stage of enrollment. By developing a comprehensive and well thought out enrichment plan, enrollment progression can be accurately estimated and reduce concerns, from the outset of the study, for feasibility.

Pilot studies are imperative for good decision-making preparatory to the execution of large prospective studies wherein resources (ie, budget, personnel, study space and equipment) are constrained. Prior studies reporting experience with OSA screening tools are insightful to planning trials. More insightful is preexisting data, or data developed for planning purposes, that is site-specific. As we have demonstrated, these data include study population characteristics, such as prevalence of OSA by study AHI criterion, and other characteristics if the study seeks to examine any particular subpopulation. As the field addresses OSA outcomes and treatment efficacy in specific subgroups (ie, women, nonobese, or mild OSA), investigators will be better prepared to successfully plan and execute respective studies by considering the selected enrollment enrichment screening instrument utility that minimizes unnecessary study resource consumption. These considerations must be evaluated in terms of available study resources but also, in the case of applied or clinical studies, the clinical services pathway, if clinical procedures will provide source data for the study. Screening for disease also poses inherent risks; if screening for disease is assumed by the study then ethical considerations for study-associated screening results, including referral to diagnosis, must be addressed by study investigators.28 All of these considerations are also balanced with volunteer burden for pre-enrollment screening and successful study recruitment.

Though the reported results are consistent with published MAP index accuracy estimates, the current results are based on a smaller sample size than prior reports, which contributes to less precise estimates. As the MAP index was used in a clinical trial recruiting from an adult sleep center population, the generalizability is limited and should not be construed as applicable to general adult clinical populations without high likelihood for OSA. With current research trends moving to community-based settings, wherein the OSA prevalence and population characteristics are likely to be different than in clinical sleep centers, it is increasingly important to pilot test the MAP index as an enrollment enrichment strategy prior to study implementation. This affords researchers insight for addressing key questions preparatory to expending constrained study resources.

DISCLOSURE STATEMENT

The parent trial from which the data for the secondary analysis were derived was supported by Award Number R00NR011173 (Sawyer AM, PI) from the National Institutes of Health/National Institute of Nursing Research. The content is solely the responsibility of the authors and does not represent the official views of National Institute of Nursing Research or National Institutes of Health. The parent trial was also supported by American Nurses Foundation and Sigma Theta Tau International (Sawyer AM, PI). The authors report no conflicts of interest.

ABBREVIATIONS

AHI

apnea-hypopnea index

AUC

area under the curve

BestAIR

Best Apnea Interventions for Research

BMI

body mass index

CI

confidence interval

MAP

Multivariable Apnea Prediction

NPV

negative predictive value

OSA

obstructive sleep apnea

PAP

positive airway pressure

PPV

positive predictive value

PSG

polysomnography

ROC

receiver operating characteristic

ACKNOWLEDGMENTS

Author contributions: Yang: Study design, statistical analysis and manuscript writing; Watach: Study design, data preparation and manuscript writing; Varrasse: Study design, data collection, and manuscript writing; King: Principal biostatistician and manuscript writing; Sawyer: Principal investigator (parent study and secondary study) and manuscript writing.

REFERENCES

1 

Marshall NS, Wong KK, Cullen SR, Knuiman MW, Grunstein RR. Sleep apnea and 20-year follow-up for all-cause mortality, stroke, and cancer incidence and mortality in the Busselton Health Study cohort. J Clin Sleep Med. 2014;10(4):355–362. [PubMed Central][PubMed]

2 

Dewan NA, Nieto FJ, Somers VK. Intermittent hypoxemia and OSA: implications for comorbidities. Chest. 2015;147(1):266–274. [PubMed Central][PubMed]

3 

Dong JY, Zhang YH, Qin LQ. Obstructive sleep apnea and cardiovascular risk: meta-analysis of prospective cohort studies. Atherosclerosis. 2013;229(2):489–495. [PubMed]

4 

Hiestand DM, Britz P, Goldman M, Phillips B. Prevalence of symptoms and risk of sleep apnea in the US population: results from the national sleep foundation sleep in America 2005 poll. Chest. 2006;130(3):780–786. [PubMed]

5 

Young T, Evans L, Finn L, Palta M. Estimation of the clinically diagnosed proportion of sleep apnea syndrome in middle-aged men and women. Sleep. 1997;20(9):705–706. [PubMed]

6 

Kushida CA, Littner MR, Morgenthaler T, et al. Practice parameters for the indications for polysomnography and related procedures: an update for 2005. Sleep. 2005;28(4):499–521. [PubMed]

7 

Chai-Coetzer CL, Antic NA, McEvoy RD. Ambulatory models of care for obstructive sleep apnoea: Diagnosis and management. Respirology. 2013;18(4):605–615. [PubMed]

8 

Balk EM, Moorthy D, Obadan NO, et al. Diagnosis and Treatment of Obstructive Sleep Apnea in Adults. AHRQ Comparative Effectiveness Reviews. Rockville, MD: Agency for Healthcare Research and Quality (US); 2011.

9 

Jonas DE, Amick HR, Feltner C, et al. Screening for obstructive sleep apnea in adults: Evidence report and systematic review for the us preventive services task force. JAMA. 2017;317(4):415–433. [PubMed]

10 

Myers KA, Mrkobrada M, Simel DL. Does this patient have obstructive sleep apnea? The rational clinical examination systematic review. JAMA. 2013;310(7):731–741. [PubMed]

11 

Portney LG, Watkins MP. A concept of clinical research. Foundations of Clinical Research: Applications to Practice. 3rd ed. Philadelphia, PA: F.A. Davis Company; 2015:3–26.

12 

Kundel V, Shah N. Impact of portable sleep testing. Sleep Med Clin. 2017;12(1):137–147. [PubMed Central][PubMed]

13 

Collop NA, Anderson WM, Boehlecke B, et al. Clinical guidelines for the use of unattended portable monitors in the diagnosis of obstructive sleep apnea in adult patients. Portable Monitoring Task Force of the American Academy of Sleep Medicine. J Clin Sleep Med. 2007;3(7):737–747. [PubMed Central][PubMed]

14 

Safadi A, Etzioni T, Fliss D, Pillar G, Shapira C. The effect of the transition to home monitoring for the diagnosis of OSAS on test availability, waiting time, patients' satisfaction, and outcome in a large health provider system. Sleep Disord. 2014;2014:418246. [PubMed Central][PubMed]

15 

Singh J, Badr MS, Diebert W, et al. American Academy of Sleep Medicine (AASM) position paper for the use of telemedicine for the diagnosis and treatment of sleep disorders. J Clin Sleep Med. 2015;11(10):1187–1198. [PubMed Central][PubMed]

16 

Ho PM, Peterson PN, Masoudi FA. Evaluating the evidence: is there a rigid hierarchy? Circulation. 2008;118(16):1675–1684. [PubMed]

17 

Gleason K, Shin D, Rueschman M, et al. Challenges in recruitment to a randomized controlled study of cardiovascular disease reduction in sleep apnea: an analysis of alternative strategies. Sleep. 2014;37(12):2035–2038. [PubMed Central][PubMed]

18 

Maislin G, Pack AI, Kribbs NB, et al. A survey screen for prediction of apnea. Sleep. 1995;18(3):158–166. [PubMed]

19 

Gurubhagavatula I, Maislin G, Nkwuo JE, Pack AI. Occupational screening for obstructive sleep apnea in commercial drivers. Am J Respir Crit Care Med. 2004;170(4):371–376. [PubMed]

20 

Gurubhagavatula I, Nkwuo JE, Maislin G, Pack AI. Estimated cost of crashes in commercial drivers supports screening and treatment of obstructive sleep apnea. Accid Anal Prev. 2008;40(1):104–115. [PubMed]

21 

Gurubhagavatula I, Maislin G, Pack AI. An algorithm to stratify sleep apnea risk in a sleep disorders clinic population. Am J Respir Crit Care Med. 2001;164(10 Pt 1):1904–1909. [PubMed]

22 

Gurubhagavatula I, Fields BG, Morales CR, et al. Screening for severe obstructive sleep apnea syndrome in hypertensive outpatients. J Clin Hypertens. 2013;15(4):279–288.

23 

Wilson G, Terpening Z, Wong K, et al. Screening for sleep apnoea in mild cognitive impairment: the utility of the multivariable apnoea prediction index. Sleep Disord. 2014;2014:945287. [PubMed Central][PubMed]

24 

Sawyer AM, King TS, Weaver TE, et al. A tailored intervention for PAP adherence: the SCIP-PA Trial. Behav Sleep Med. 2017;1–25.

25 

Iber C, Ancoli-Israel S, Chesson A, Quan SF; for the American Academy of Sleep Medicine. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications. 1st ed. Westchester, IL: American Academy of Sleep Medicine; 2007.

26 

Hosmer DW, Lemeshow SL, Sturdivant RX. Applied Logistic Regression. 3rd ed. New York, NY: Wiley; 2013.

27 

Young T, Palta M, Dempsey J, et al. The occurrence of sleep-disordered breathing among middle-aged adults. N Engl J Med. 1993;328(17):1230–1235. [PubMed]

28 

Coughlin SS. Ethical issues in epidemiologic research and public health practice. Emerg Themes Epidemiol. 2006;3(1):16.


Supplemental Material

Supplemental Material

(.pdf | 37 KB)