Medication Adherence and Persistence in Severe Obstructive Sleep Apnea

Abstract

Study Objective:

The aim of this study was to compare 2 groups of patients with severe obstructive sleep apnea (OSA) who were taking medication for cardiovascular disease: those who were compliant with nasal continuous positive airway pressure (CPAP) treatment and those who refused treatment or were noncompliant with CPAP treament.

Methods:

In a cohort of 2158 patients with severe OSA (apnea-hypopnea index > 30) a 2-year prospective longitudinal assessment of adherence and persistence with 3 medication categories (antihypertensives, statins, and antiplatelets) was carried out using the administrative database of the National Health Service. MEdication adherence was evaluated by calculating the medication possession ratio (%MPR = days supply/actual days to refill x 100) for each drug. Medication persistency was defined as the proportion of subjects having filled a prescription in the last 30 days of the 2-year period. CPAP use was assessed at every follow-up visit after the treatment was prescribed. Medication adherence was compared between patients who had adequate CPAP adherence (> 4 h/day) and those who declined CPAP therapy or had discontinued CPAP due to an average use of less than 4 hours per day.

Results:

The average 2-year MPR for antihypertensives, statins, and antiplatelets was not different among patients who used CPAP (88%, 81%, 95%) or did not use CPAP (86%, 77%, 93%). Female sex and increased number of comorbidities were predictors of good medication adherence (MPR > 80%). The rates of persistence for the 3 studied medications after the 2-year observation period were not different between the 2 groups (patients with or without CPAP).

Conclusions:

Medication adherence and persistence during a 2-year period for 3 well-known protective cardiovascular medications were not different in patients with severe OSA, whether or not they were treated with CPAP.

Citation:

Villar I; Izuel M; Carrizo S; Vicente E; Marin JM. Medication adherence and persistence in severe obstructive sleep apnea. SLEEP 2009;32(5):623-628.

OBSTRUCTIVE SLEEP APNEA (OSA) IS CHARACTERIZED BY THE PRESENCE OF AT LEAST 5 UPPER AIRWAY OBSTRUCTIVE EVENTS PER HOUR OF SLEEP. OSA is found in 9% to 26% of middle-aged adults.¹ Excessive daytime sleepiness and inability to sustain attention to different tasks are the main complaints. This condition is also associated with an increased risk of having motor vehicle accidents² and cerebrovascular³ and cardiovascular disease.⁴ The treatment of choice for patients with OSA is the application of continuous positive airway pressure (CPAP) through a nasal mask during sleep. In cohort studies, untreated patients with severe sleep apnea (apnea-hypopnea index [AHI] of more than 30 episodes per hour of sleep) have an increased risk of having a cardiovascular event over a 10-year period when compared with control subjects with similar degrees of obesity.⁵ In addition, treatment with CPAP in severe OSA reduces the risk to the control level among patients who use the devise for at least 5 hours per night.⁵ Although it has been suggested that untreated severe OSA leads to elevated risks of having a cardiovascular event, one cannot rule out the possibility that patients with severe OSA who do not use CPAP are different in other ways and are, for example, noncompliant with medications used to reduce cardiovascular risk.

Medication nonadherence may explain the suboptimal achievement of therapeutic targets in most chronic diseases.⁶ In clinical practice no study has evaluated the association between medication adherence and CPAP adherence in OSA. Given the importance of clarifying the role of OSA as a risk factor for cardiovascular diseases and the impact of CPAP therapy to decrease such risk, it is paramount to assess the adherence to protective medications for cardiovascular events of patients with OSA who do or do not adhere to treatment with CPAP. In this study, we characterized adherence to the use of protective cardiovascular medications in a large, clinical-based cohort with OSA and determined the association with patient compliance with nasal CPAP therapy.

METHODS

Patient Population

We conducted a prospective study of patients in the Zaragoza Sleep Cohort Study. The characteristics of this cohort have been published elsewhere.⁵ For the purpose of this study, we included all patients with severe OSA who had been studied in our Sleep Unit from January 1996 through December 31, 2005. All patients alive on January 1, 2006, were then evaluated for medication adherence and persistence for at least 2 years. According to National Guidelines, severe OSA that merits CPAP as the primary treatment is considered when there are more than 30 episodes of apnea or hypopnea per hour of sleep (ie, AHI > 30), or if the AHI is between 5 and 30 and the patient complains about severe daytime sleepiness, or when the patient has coexisting polycythemia or cardiac failure.⁷ Patients who were not treated with nasal CPAP received conservative advice: weight loss, avoid alcohol and sedatives, stop smoking, avoid sleep deprivation, and, if appropriate, restrict sleep position. Patients with OSA annually attended our clinic. In addition, patients treated with CPAP came to the clinic for follow-up care during the third and sixth month after the beginning of treatment and annually thereafter. During these visits, compliance with CPAP therapy was assessed by the timer built into each CPAP device. A daily mean use of more than 4 hours per day was considered necessary to maintain the CPAP prescription. Otherwise, if, after a reinforcement period of 3 additional months, the patient still used the CPAP during less than 4 hours per night, treatment was stopped and an alternative therapeutic option was offered.

Adherence to medications was assessed during 2006 and 2007. Baseline patient demographics, vital signs, and laboratory data were derived from automated cohort databases. The degree of comorbidity was determined for each patient with the validated Charlson index.⁸ The Ethics Committee of our institution approved the study, and all participants gave their informed consent after being fully informed of the nature, design, and potential risks and benefits of the study.

Source of Data

The list of drugs prescribed to each patient was obtained from the patient’s medical records. The information available for each medication included medication name and daily dose. We used the Pharmacy Department of Health Services database from the province of Aragon, Spain, to evaluate the prescription claims submitted from pharmacies for reimbursement. This database registers all prescriptions dispensed from Aragon pharmacies. In Spain, the National Health System covers all inhabitants and completely reimburses drug expenses for retired and disabled people and partly reimburses all other people (with a maximum copayment for each individual of 10%-40% of the market price). Treatment with CPAP is free of charge for any patient. Pharmacies are required to register patient insurance codes for all prescriptions dispensed, which ensures complete registration, and the low copayment for patients minimizes their incentive to obtain medication through other sources. The prescription claims submitted from the pharmacies to the Department of Health Services for reimbursement contain the brand name and dosage of the drug, units dispensed, and the name of the doctor who issued the prescription. Matching the data from the attending clinic and from the Department of Health Services allowed us to assess the concordance between the prescribed medication and what was dispensed. For ethical considerations, all personal identifiers were scrambled by the Department of Health Services before we received the data.

Medication Adherence

Medication adherence was calculated as the proportion of days covered, based on the total number of days supplied for filled prescriptions between the first and the last dispensing date of the observation interval. This term has also been named as the medication possession ratio (MPR) and was calculated, according to the following formula⁹:

MPR was calculated from blind patient files for 3 categories of medications that potentially impact on cardiovascular risk factors: (1) antihypertensives, including angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, β-blockers, diuretics, or calcium channel blockers; (2) statins; and (3) antiplatelet drugs, including aspirin and clopidogrel. Because antihypertensive medication regimens are sometimes changed due to adverse effects or inadequate control and because patients often require more than 1 agent to control blood pressure, adherence was averaged across the different classes of these medications. Therefore, the antihypertensive medication MPR reflects general adherence to the overall antihypertensive medication regimen. The adherence measure was dichotomized so that patients were considered adherent if a value of 80% or higher was attained, which is consistent with the threshold found in most studies.^9,10 To measure persistence with treatment we used the so called “anniversary model,” in which a patients are deemed persistent if they refill a prescription within 30 before or after the 2-year anniversary from the initiation of the observation period (January 2006).^11,12

Statistical Analysis

The aim of this study was the comparison of adherence cardiovascular active medication in patients with OSA who accepted treatment with nasal CPAP and used the device a minimum of 4 hours per night versus patients with OSA who refused the treatment of nasal CPAP or used the devise less than 4 hours per night on average. All individuals included in the analysis had the same 2-year follow-up period. Data from individuals who died before the end of the observation period were censored at the time of the patient’s death. Patients with an MPR greater than 1 because of oversupply had their MPR value truncated at 1¹⁰ Results were expressed as median (interquartile range), and differences between groups were performed with the Mann-Whitney U test. Logistic regression analysis was used to evaluate predictors of “good medication adherence,” as defined by an MPR greater tan 80% for all 3 medication categories. In the univariate analysis, the following variables were regarded as potential predictors: age, sex, body mass index (BMI), smoker status, AHI, Epworth Sleepiness Scale (ESS) score, and the Charlson Index. Variables with P values of less tan 0.1 were included in the multivariate model. Kaplan-Meier analysis was performed to compare the probability of medication persistence over the observed period (2 years) between the patients who accepted or refused CPAP. Data were censored due to death of the patient or end of the study. Data analysis was performed with SPSS, Version 15.1 (SPSS, Inc., Chicago, IL).

RESULTS

During the inclusion period (January 1 1996 to December 31, 2005), among 2240 patients with severe OSA considered as candidates for CPAP therapy, 1360 (60.7%) accepted having a second sleep study for CPAP titration, took home the CPAP device, and demonstrated good CPAP compliance. Because 4 patients were lost to follow-up, in the final analysis, 1356 patients with severe OSA effectively treated with CPAP were included in the analysis. Among the rest of the patients with severe OSA who were eligible for CPAP therapy, 480 did not accept to have a second night’s sleep study for CPAP titration, 213 rejected the use of CPAP after the titration study, and 187 were not compliant with CPAP. Because 70 patients chose an alternative therapy and 8 were lost to follow-up, in the final analysis, 802 were included. Throughout the inclusion time period, the rate of CPAP recommendation, acceptance, and hours of compliance did not differ, ranging from 32%, 75%, 5.74 hours per night in 1996 to 26%, 80%, and 5.41 hours per night in 2005, respectively. The range of CPAP use over the observation period was 4.2 to 10.6 mean hours per night, with no differences among patients depending on the date of inclusion. Table 1 shows the baseline characteristics of the participants from each group who either were treated with CPAP or had no active treatment for their OSA condition. No differences were found between the groups at baseline in sex, age, BMI, AHI, daytime hypersomnolence, prevalence of current smokers, and education level. Prevalence of concomitant comorbid condition, as quantified by the Charlson Index, and the percentages of patients treated with antihypertensives, statins, and antiplatelet medications were also similar. At the end of the 2-year observation period, BMI was 0.6 kg/m² higher in the CPAP group compared with the non-CPAP group (30.9 ± 5.4 vs 30.3 ± 5.6 kg/m², P = 0.02) with no difference in current smokers (23.8% vs 24.7% respectively, P < 0.05).

Table 1.

Patient Characteristics

Characteristic	OSA with CPAP (n = 1356)	OSA without CPAP (n = 802)	P Value
Men, %	86.1	88.5	0.11
Age, y	49.8 ± 8.4	49.4 ± 9.4	0.31
BMI, kg/m2	30.7 ± 5.3	30.3 ± 6.1	0.11
AHI, no/h	44.1 ± 5.7	43.6 ± 7.2	0.07
Sao2 nadir, %	77.5 ± 8.5	78.1 ± 7.8	0.10
ESS, score	12.5 ± 6.5	11.9 ± 7.5	0.05
Charlson Comorbidity Index	1.81 ± 1.61	1.67 ± 1.95	0.07
Current smoker, %	24.2	26.4	0.26
Completed high school, %	62.2	60.8	0.57
Prescribed medications, No. (%)
Antihypertensives	420 (31)	239 (30)	0.61
Statins	112 (8)	63 (8)	0.80
Antiplatelets	68 (5)	41 (5)	0.92

Data are presented as mean ± SD unless otherwise indicated. BMI refers to body mass index; AHI, apnea-hypopnea index; ESS, Epworth Sleepiness Scale.

Table 2 shows the medication adherence of participants from each group for the 3 categories of studied medications. The average 2-year MPR for antihypertensives, statins, and antiplatelets in the CPAP group was 88%, 81%, and 95%, respectively. These values were not different from those in the non-CPAP group: 86%, 77%, and 93%, respectively. Again, among patients taking medications from all 3 medications categories, adherence was not different between CPAP and non-CPAP groups (82 and 79%, respectively). The percentages of patients showing good medication adherence, as considered when the MPR was 80% or greater, did not differ between groups. In the CPAP group, there was no correlation between medication adherence for the 3 studied medications and daily CPAP use. To explore predictors for good medication adherence in severe OSA, we used logistic regression modeling. In the univariate analysis, female sex and comorbidity index were positive predictors, and active smoking was a negative predictor (Table 3). Once the model was fully adjusted, women were 2 times more likely than men to adhere to medication (adjusted odds ratio 2.05, confidence interval [CI]: 1.22–5.19). In this model, the comorbidity index was also a predictor of good medication adherence, with an odds ratio of 1.85 per score unit (adjusted odds ratio 1.85, CI 1.07-3.33). There was no significant effect of the treatment for OSA (CPAP versus non-CPAP); neither was significant group-by-sex interaction (P = 0.15).

Table 2.

Medication Adherence

	Antihypertensives		Statins		Antiplatelets		All 3 medications
	CPAP n = 420	Non CPAP n = 239	CPAP n = 112	Non CPAP n = 63	CPAP n = 68	Non CPAP n = 41	CPAP n = 54	Non CPAP n = 35
% MPR, median	89	88	83	80	97	96	84	82
(interquartile range)	(45 – 99)	(44 – 99)	(44 – 99)	41 – 98)	(67 – 99)	65 – 99)	(50 – 99)	(41 – 99)
Good Adherence (MPR < 80%), no. (%)	348 (83)	193 (81)	76 (68)	40 (63)	63 (93)	37 (90)	49 (91)	31 (89)

CPAP refers to continuous positive airway pressure; MPR, medication possession ratio.

Table 3.

Logistic Regression Model Predicting Good Medication Adherence in Patients with OSA

Variable	Univariate OR (95% CI)	P Value	Multivariate OR (95% CI)	P Value
Age, y	0.96 (0.94 – 1.08)	0.41
Sex, female	2.98 (1.50 – 4.08)	< 0.001	2.05 (1.22 – 5.19)	0.02
BMI, kg/m2	0.94 (0.88 – 1.05)	0.07
ESS, units	1.05 (1.01 – 1.19)	0.04	1.02 (0.96 – 1.56)	0.09
Charlson Index, units	2.65 (1.85 – 4.82)	< 0.001	1.85 (1.07 – 3.33)	0.03
Current smoker	0.85 (0.71 – 0.98)	0.01	0.91 (0.65 – 1.58)	0.15
Treatment with CPAP	1.05 (0.85 – 2.11)	0.11

CPAP refers to continuous positive airway pressure; BMI refers to body mass index; AHI, apnea-hypopnea index; ESS, Epworth Sleepiness Scale; OR, odds ratio; CI, confidence intervals.

The rates of persistence in medication adherence between patients using CPAP and patients without CPAP were compared using Kaplan-Meier analysis (Figure 1). Over the 2-year observation period, the persistence rates of the 3 medication categories decreased slightly, with no statistical differences between groups. The average best persistence rates were for antiplatelets (96% and 94%, respectively). Statins showed the worst rates (92% and 89%, respectively) and antihypertensives had intermediate rates (94% and 92%, respectively).

Figure 1.

Kaplan-Meier curves of the probability of persistence to antihypertensives, statins, and antiplatelets during the 2-years follow-up. CPAP refers to those patients who were compliant with continuous positive airway pressure (CPAP) use; non-CPAP, those patients who refused treatment with CPAP or did not use CPAP for at least 4 hours per day.

DISCUSSION

In this large cohort of patients with severe OSA, medication adherence and persistence in using antihypertensive, lipid-lowering, and antiplatelet drugs did not differ whether the patients received treatment with CPAP. This is the first study to compare general adherent behavior in patients with OSA and to demonstrate that not using CPAP is not a marker for nonadherence to other medical therapies with preventive cardiovascular value.

Recently published prospective population studies from Australia and the United States have shown that untreated moderate-to-severe OSA is associated with increased mortality.^13,14 The results of these studies are in agreement with large-cohort, clinical-based series, which also showed that severe OSA is an independent risk factor for cardiovascular morbidity and mortality.^3,5,15,16 When cohort studies are used to evaluate the effect of a specific therapy, the main criticized point is the self-selection of motivated patients who choose the active treatment, as compared with the “control” or untreated group. In OSA, such motivated patients, who, in parallel with adhering to CPAP treatment, could demonstrate a healthy lifestyle, higher smoking cessation rate, or be more adherent to medications prescribed for comorbid conditions, as compared with patients who refused to use CPAP. Consequently the reduced cardiovascular morbidity and mortality observed in patients with CPAP over a period of years could be partly explained by a more adherent behavior toward medications such as statins or antihypertensives, which also reduce cardiovascular outcomes. In our long-term cohort study, at baseline, important health variables such as comorbid conditions, active smoking, or education level did not differ between the group treated with CPAP and the group who refused or was not compliant with this therapy. Physical activity level was not collected, but the fact that both groups at matching examination had similar age, sex distribution, and BMI suggests that engagement in active exercise did not differ. Furthermore, throughout the 2-year observation period, the CPAP group showed an increase in weight with no change in current smoking. If anything, these observations suggest a lower healthy lifestyle and several higher risk factors in the CPAP group throughout the observation period.

Adherence-to-medication rate is defined as the percentage of the prescribed medication actually taken by the patient over a specific period, whereas persistence indicates whether or not a patient stays on therapy.⁶ For chronic conditions, the average rates of adherences in clinical trials range from 43% to 78%.¹⁷ Most authors consider rates higher than 80% to be adequate.^9–13 Poor adherence to medication regimens is common, contributing to substantial worsening of disease, death, and increased health care costs.^18–20 In clinical trials, participants who do not follow the active medication or the placebo regimens have poorer prognosis than do adherent subjects.²¹ Such data are not available in cohort studies. Furthermore, in OSA, no one has previously studied whether patients who refuse or are not compliant with CPAP treatment are also not adherent to other therapies. If this were the case, this nonadherent behavior would be, per se, a risk factor for cardiovascular outcomes that precludes the protective effect of CPAP. Our data indicate that, in patients with severe OSA, adherence and persistence to antihypertensive, lipid-lowering, and antiplatelet medications were not different among patients who did or did not use CPAP. We also found higher persistence rates than those reported previously in nonintervention studies. In these other studies assessing persistence in patients taking antihypertensive and lipid-lowering drugs, the 1-year persistence ranged from 35% to 83%^22,23 and from 33% to 69%,^24,25 respectively. In our series, the 2-year persistence for antihypertensives and statins was 94% and 92% in the CPAP group and 92% and 89% in the non-CPAP group. Antiplatelet-medication adherence and persistence are even higher in our patients, but no data are available from the literature that allow us to compare our figures. Some reasons could explain the differences between medication adherence of our patients and those from the mentioned reports, the most important being the close follow-up visit with the family practitioner in charge of the prescription in our Health System and our own additional follow-up protocol, which included annual visits to the sleep clinic. That is, reinforcing continuity of care may promote long-term adherence by shortening or eliminating gaps in medication use.

Major predictors associated with poor adherence are related to the presence of psychological and cognitive impairment problems and barriers to care about medication.^26–28 Patients with fewer comorbidities or asymptomatic diseases are also the least adherent to medications.²⁷ This is in accordance with our results, since we found that the Charlson Index, a well-validated system to grade comorbidities, was an independent predictor of good medication adherence using logistic regression analysis. Again, being treated or not with CPAP was not a predictor for adherence in the multivariate model.

Our study does have some limitations. First, the patients studied are included in a large cohort study from a sleep unit with a specific protocol therapy for OSA and a close follow-up system for all patients attending the clinic. This will not be the case for patients with OSA whose care is provided by solo-practice physicians or small clinics. Second, all of our patients belong to the National Health System, most of them with full drug coverage and a minority having coinsurance with a high deducible. These results could differ in countries with different health systems that involve social and economic barriers to medication adherence. Third, ideal direct methods to measure adherence, such as biochemical measures of drug or metabolites in plasma or urine, and directly observed therapy, are precise but expensive and burdensome to the patient and to the health care provider. Calculation of refill adherence from administrative data can be an alternative and useful indirect method to assist in evaluating patient medication adherence, particularly in a closed pharmacy system.⁹ This is the case for countries or provinces with universal drug coverage.^29–32 Use of the databases has a number of limitations, including the inability to determine if the patient actually consumed the dispensed medication. However, the efficiency of using the automated pharmacy-dispensing data for studies of adherence and persistence in large populations in a real-world setting is highly advantageous if data are deemed complete.^9,10 Our health care system uses electronic medical and pharmacy records that provide us ready and objective information on rates of refilling and persistence over time. If not perfect, this system allows us to accurately compare medication adherence with potential protective effects on cardiovascular outcomes among patients with OSA grouped according to CPAP use. Fourth, we cannot exclude a “survival effect” resulting from the non-CPAP survivors being more adherent to medications during the period 1996-2005. Finally, it could be argued that an observation period of 2 years is not enough to reflect general adherence behavior of an individual. We made an intermediary analysis after the first year of observation period, and the figures we obtained were almost identical to those after 2 years. There are no standards in the literature showing how long adherence must be evaluated. Most of the available studies on this subject have evaluated 1-year periods when comparing medication adherence between patient groups or medication categories. In the only study evaluating the persistent use of an antihypertensive medication for 4 years, the percentages of persistent patients after the first 2 years were almost identical to those at the end of the period.¹²

In conclusion, in patients with severe OSA enrolled in a large cohort study, we have evaluated medication adherence and persistence to antihypertensive, lipid-lowering, and antiplatelet drugs through a 2-year period. Using an administrative pharmacy database, we calculated the average MPR and persistence. Female sex and coexistent comorbid conditions were predictors of good adherence. Patients treated with CPAP were not more adherent to the studied medication categories than were patients who did not use CPAP. Our findings do not support the hypothesis that nonuse of CPAP could be a marker for nonadherent behavior.³³ Rather, they reinforce the current evidence of the effect of untreated severe OSA as an independent cardiovascular risk factor.

DISCLOSURE STATEMENT

This was not an industry supported study. The authors have indicated no financial conflicts of interest.