INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sleep apnea is a disorder characterized by repetitive partial or complete occlusions of the upper airway during sleep. Common nocturnal effects include hypoxemia, arousals from sleep, and brief increases in blood pressure (1, 2). Common daytime effects include excessive sleepiness, irritability, decreased concentration and memory, decreased energy, and depressive symptoms (3, 4). The most widely used therapy is nasally applied continuous positive airway pressure (CPAP), which decreases sleepiness and improves neurocognitive function and vigilance (5). Questionnaires to assess severity of daytime sleepiness have been developed, but they have not taken into account the impact of this symptom on patients' lives (6). No questionnaires have been developed to assess the specific health-related quality of life issues that are important to patients with sleep apnea. Generic health measures such as the Nottingham Health Profile (NHP) (7) and the Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36) (8) have been used in patients with sleep apnea, but they may not detect more subtle effects of the disease on quality of life and were not constructed to evaluate within-subject change after treatment (5, 9, 10).

We therefore developed and tested a disease-specific instrument to evaluate health-related quality of life in patients with sleep apnea for use in clinical trials. We followed published guidelines that have been used to construct health-related quality of life scales in other disease states (11).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Item Generation

Five approaches were used to identify the many ways that sleep apnea might contribute to impairment of quality of life that were important to patients suffering from this disorder. First, we conducted a comprehensive review of the sleep apnea literature. Second, we drew on our clinical experience and prior research with hundreds of patients with sleep apnea during the previous 8 yr. Third, we sought the opinions of eight sleep apnea content experts from across North America. Fourth, we used five focus groups involving approximately 50 patients with sleep apnea and their partners to discuss the impact of sleep apnea and its treatment on their quality of life. Fifth, we conducted semistructured interviews, lasting approximately 2 h each, with 40 patients with sleep apnea, most of whom had been treated with nasal continuous positive airway pressure (CPAP). These interviews included open-ended questions about all possible aspects of a patient's life that might be adversely affected from having (or having had) sleep apnea. From content analysis based on all of these approaches a list of 133 items likely to be important to patients with sleep apnea was developed. This list included 24 symptoms associated with treatment, primarily with CPAP, that might have some adverse impact on quality of life.

Item Selection

The purpose of this phase was to determine which of the 133 items were most commonly experienced and were most important to patients with sleep apnea. We identified the last 300 patients to undergo polysomnography in our sleep laboratory, and then excluded patients in whom a disorder other than sleep apnea had been diagnosed or who lived outside of Calgary. A random sample of the remaining patients was asked to participate. The 133-item questionnaire was administered individually to these 100 patients by a trained interviewer. Because women were not adequately represented, we interviewed 13 additional, randomly selected, women from the patient list. We also administered the questionnaire to 50 patients referred by their physician to an otolaryngologist for a primary complaint of snoring. Of the 133 questions in the questionnaire, all patients were asked three questions about 109 items that pertained to quality of life issues that may have been as a result of them having sleep apnea and/or snoring: (1) was it a problem for them? (yes/no); (2) if it was a problem how important was it to their quality of life? (five-point scale ranging from "not very important" [zero] to "extremely important" [five]); (3) if they had undergone treatment, how much did it improve? Patients with sleep apnea who had undergone treatment were asked an additional 24 questions about treatment-related symptoms. For each item the frequency that respondents answered "yes, it was a problem" was recorded (frequency). Next the mean importance score for each item was determined (importance), and finally the product of these two numbers (frequency×importance) was calculated. Patients who had undergone treatment for their sleep apnea were also asked how much improvement they had experienced in a particular problem after treatment. Patterns of response were analyzed to determine if there were differences according to sex (unpaired t tests) or clinical sleep apnea severity (analysis of variance).

A rating of each patient's severity of sleep apnea was assigned by physicians experienced in assessing patients with sleep apnea after reviewing the patient's chart and based on a combination of the apnea-hypopnea index, degree of oxygen desaturation and extent of self-reported daytime symptoms.

Questionnaire Construction and Testing

Items with a high frequency×importance rating were selected for inclusion in the final questionnaire, the Calgary Sleep Apnea Quality of Life Index (SAQLI). When it appeared that two questions were probing for similar types of information we determined the concordance rate for the response that both items either were or were not a problem. If the concordance rate was greater than 85% then the questions were combined, or the one that appeared to us to be the inferior question was dropped. Items were eliminated from the final questionnaire if most patients reported that it had not improved after a therapeutic intervention.

The items that were chosen through this process to be included in the Calgary SAQLI were organized into four domains: A. Daily Functioning (11 items): B. Social Interactions (13 items): C. Emotional Functioning (11 items): D. Symptoms (five items). An additional domain, E, entitled Treatment-related Symptoms (five items) was added for use when some active therapy such as surgery, CPAP, or a dental appliance has been tried. Provision was made in Domains D and E for patients to select and rate the five symptoms most relevant to their experience. Patients were also given an opportunity to add other symptoms to this list. The questionnaire was designed for administration by a trained interviewer using Likert scales with seven response options (see APPENDIX). Color coded cards, as used in similar types of quality of life questionnaires (12, 15), were used so that subjects did not have to remember the response options.

The questionnaire was sent to six clinical experts from across North America, distinct from the first group of experts, who were asked to comment on its sensibility, clarity, and comprehensiveness.

The SAQLI was then tested in 24 patients with newly diagnosed sleep apnea who were about to start a home trial of CPAP to determine: (1) that the questions and instructions were understood and made sense to the patient, (2) that a complete range of responses on the Likert scales was used, (3) the degree of correlation with the SF-36, (4) the degree of internal consistency (reliability), and (5) that the items were responsive to change after a 4-wk trial of CPAP. The SF-36 was chosen for assessment of convergent validity because it is widely accepted as a generic measure of quality of life, and some of the scales such as Vitality and Physical Functioning may be particularly sensitive to the symptoms of sleep apnea and improvement with CPAP (10).

CPAP compliance was monitored covertly in most patients with hour timers that were activated only when the machine was generating a pressure within 2 cm H₂O of the set pressure. All patients had a minimum 4-wk home trial of CPAP (Healthdyne Tranquility Quest 7300; Healthdyne, Marietta, GA); occasional patients had slightly longer trials because of scheduling difficulties. Patients who were found not to be compliant with therapy were given an additional 2-wk trial. All patients initially had complete orientation to CPAP from trained respiratory therapists and had easy access to them for any difficulties they had with CPAP. Compliance was defined as greater than 3.5 h average use per night, which has been previously shown to be the average use by patients with sleep apnea (16, 17), for at least 2 wk prior to their final assessment.

Prior to starting CPAP the patients completed the SF-36 and the SAQLI. At the completion of the CPAP trial subjects completed the SF-36 and a global rating of change for each SAQLI item. Subjects were asked if they had experienced a change in the item. If it had not changed the item was given a score of zero. If it had changed they were asked if it had improved or worsened. The degree of improvement or worsening was assessed with a seven-item Likert scale that ranged from "hardly any improvement (worsening)" to a "a very large improvement (worsening)." A rating of 2, "a little improvement (worsening)," or more was taken to indicate a clinically significant change (17, 18). Negative numbers indicated worsening and positive numbers indicated improvement, so in total there were 15 response options (7 to +7). An average rating of change was calculated for each of the domains A through D of the SAQLI. Domain E from the SAQLI, Treatment-related Symptoms, was included with this testing. After the administration of the questionnaire respondents were asked to identify any questions they had difficulty understanding or made them feel uncomfortable. Suggestions for revision were sought for any questions identified as problematic. Patient feedback led to only minor revisions. Internal consistency of the SAQLI was evaluated by calculating Cronbach's alpha.

The research was approved by the University of Calgary Conjoint Medical Research Ethics Board, and all subjects signed informed consent forms. Subjects who had their CPAP compliance monitored covertly signed a poststudy consent form that informed them of this monitoring and requested they allow their data to be used in the final analysis. No patients refused this request.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Baseline characteristics of the 113 patients with sleep apnea and 50 snorers who completed the 133-item questionnaire and the 24 patients selected to test the SAQLI are listed in Table 1. As expected for a referral population to a sleep laboratory, patients tended to be middle-aged men. The vast majority of the patients had regular sleeping partners, and there was a good spectrum of disease severity. Of the 113 patients with sleep apnea, 80% had attempted some type of therapy; 42% reported regular CPAP use for a median of 6 mo (interquartile range, 1 to 14 mo).

The highest scoring items from the initial questionnaire are shown in Table 2. The highest scores were typically related to problems with alertness, concentration, and the feeling that most activities required an extraordinary effort to accomplish. A majority of patients reported that their snoring or sleep apnea was contributing to significant problems in their relationships. Patients also reported many effects on their mood that were probably secondary to chronic sleep deprivation.

Some questions had a very high concordance rate with other questions. For example, 90 patients indicated that being told by their partner that their snoring was bothersome/irritating to them was a problem. Of those 90 patients, 87 also indicated that being told by their partner that their snoring disturbs their partner's sleep was a problem. The overall concordance rate for the percentage of respondents who indicated that the item either was or was not a problem for them between these two questions was very high (96%). Therefore, only one of the questions was included in the final version of the SAQLI.

Other items, for example "concerns patients had about their physical appearance," had a reasonably high frequency × importance product but were excluded from the SAQLI because almost all patients reported that it did not improve after therapy.

With few exceptions there was very little difference in the order of the frequency × importance product between men and women and between the three levels of sleep apnea clinical severity (Table 3). The only exception to this was in the domain Social Interactions where there was general agreement among the first four items, but the last four items in that domain were more variable. There was very good agreement between snorers and patients with sleep apnea in their ranking of items selected for most of the domains. Although the rank order of items was similar between different groups, the mean frequency × importance scores were higher in women than in men (Table 4). There were also differences between snorers and patients with sleep apnea and between different levels of sleep apnea severity. As expected, snorers had the lowest scores, and patients with severe sleep apnea had the highest scores.

SAQLI Testing

Results from the SAQLI and SF-36 in 24 patients with sleep apnea prior to commencement of CPAP are shown in Table 5. The scores in each of the SAQLI domains and the total score range from 1 (poor quality of life) to 7 (excellent quality of life). The scores in each of the SF-36 domains and total score range from 0 (poor quality of life) to 100 (excellent quality of life). Symptom scores (Domain D) were clearly lower than the other domain scores (A, B, and C). There was a wide range of scores in these patients with no obvious floor or ceiling effects. Results from the SF-36 were below values reported for a normal population, especially in the domains of Role-Physical, and General Health, and Vitality. The correlation between the SAQLI and SF-36 total scores (r = 0.21) was not statistically significant (p = 0.30); there was a higher, statistically significant, correlation between the SAQLI total score and the Vitality domain of the SF-36 (r = 0.45, p = 0.03) (Figure 1).

View larger version (11K): [in this window] [in a new window]   Figure 1.   Comparison of SAQLI and SF-36 Total (r = 0.21; p = 0.3) and Vitality Domain (r = 0.45; p = 0.03) scores in 24 patients with sleep apnea prior to treatment with continuous positive airway pressure.

Twenty of these patients completed at least a 4-wk trial of CPAP using a machine capable of monitoring compliance. Of these, 15 patients were compliant with therapy. Their rating of change of each of the items in the SAQLI, summarized by mean domain scores and the change (before-after) in SF-36 scores is shown in Table 6. The rating of change scores were greatest for symptoms and least for social functioning and emotional functioning. For each of the domains and the total score of the SAQLI rating of change, 33% or more of the patients reported being at least somewhat better (score of 3 or greater). In the symptoms domain, 75% of patients reported being at least somewhat better. When treatment-related scores (Domain E) were included in the calculation of total SAQLI rating of change score the total fell by 0.4. Treatment related symptoms were consistently rated less important to patients than the other items in domains A to D. 

The SF-36 domain scores that improved in patients who successfully used CPAP were role-physical, vitality, and role-emotional. Overall there was a reasonably high correlation between the SAQLI rating of change (total) and the change in total SF-36 score (r = 0.60; p = 0.018) as well as five of eight SF-36 domains: Vitality (r = 0.81; p = 0.0002), Mental Health (r = 0.76; p = 0.0009), General Health (r = 0.69; p = 0.005), Physical Functioning (r = 0.61; p = 0.02), and Social Functioning (r = 0.51; p = 0.05) (Figure 2). Internal consistency as measured by Cronbach's alpha for the SAQLI total score was 0.92 and for each domain was: A, 0.88; B, 0.92; C, 0.92; D, 0.92. 

View larger version (11K): [in this window] [in a new window]   Figure 2.   SAQLI Rating of Change Score compared with the change in the SF-36 Total (r = 0.60; p = 0.018) and Vitality Domain (r = 0.81; p = 0.0002) scores in 15 patients with sleep apnea after successful completion of a trial of continuous positive airway pressure.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The development of the Calgary SAQLI was comprehensive and systematic so that all items that were potentially relevant to the lives of patients suffering from sleep apnea would be identified. We took advantage of published guidelines and recommendations for developing a disease-specific quality of life questionnaire that have been used successfully in developing indices in many other related disease states (12). The items included in the final version of the SAQLI were those that a random sample of patients with sleep apnea identified as important. We have demonstrated that although the mean frequency × importance product was different between men and women (Table 4), their ranking of items was extremely similar (Table 3), as in the Asthma Quality of Life Questionnaire (12). Thus, we believe the SAQLI is appropriate for all patients regardless of sex. A sex difference in reporting health-related quality of life has been previously reported, but we are not aware of investigations that have addressed the possible causes (12, 20). We did not perform a separate analysis for age since the vast majority of patients who present to our sleep laboratory are between 35 to 65 yr of age. Almost all of the items in the SAQLI are applicable to people across this age range. Not unexpectedly, there were significant differences in the frequency × importance product between snorers and between patients with differing levels of sleep apnea severity. However, as with sex differences, the ranking of items was very similar across the different levels of sleep apnea severity and snorers. Thus, we believe the SAQLI is an appropriate outcome measure for patients with sleep apnea and snorers. We did not test the snorers to exclude the possibility that some of them may also have had sleep apnea. However, they were selected at random from the office of an otolaryngologist for the treatment of snoring. We believe these patients are representative of the types of patients whose major problem is only snoring and this is reflected by their substantially lower frequency×importance product scores than even the patients with mild sleep apnea. If we had studied these patients and excluded those with evidence of sleep apnea it is likely their frequency×importance product scores would have been even lower. Although the SAQLI was developed to be an evaluative instrument capable of measuring subtle, but important, changes in quality of life after a therapeutic intervention, the distinct differences in the frequency×importance product scores between snorers, and patients with mild, moderate, or severe sleep apnea suggests that these items may well have discriminative capabilities as well.

Content validity, as reflected by the comprehensiveness and range of items included, was established through the five approaches used to generate items, including the extensive involvement of patients with sleep apnea. Face validity was checked by clinicians and clinical researchers from across North America as well as by patients. Construct validity of the SAQLI as a measure of disease-specific quality of life was supported by the differences in the mean frequency×importance product that increased predictably from lowest in snorers to highest in patients with severe sleep apnea (Table 4). Construct validity was investigated further by asking 24 patients who had undergone a 4-wk trial of CPAP whether each item in the SAQLI had changed (yes or no) and if yes how much improvement or worsening there had been. There was a moderate correlation between this rating of change of the SAQLI items and the change in five of the eight SF-36 domains. The rating of change in SAQLI items showed that they did improve in the majority of patients after a trial of CPAP (Table 6). The change in each of the SAQLI domains and in total SAQLI scores post-CPAP was not evaluated in the initial testing of the questionnaire. The determination of its responsiveness in different study settings will be an important addition to the ongoing evaluation of the SAQLI's validity as an outcome measure in sleep apnea clinical trials.

CPAP is a highly successful physiologic treatment for sleep apnea, but it has many potential side effects and drawbacks, which may in part explain the modest level of compliance documented in patients. We evaluated the characteristics of the SAQLI in patients after a 4-wk trial of CPAP, their use of which was documented by covert compliance meters. We have attributed improvements in the SAQLI and the SF-36 to the physiologic effects of CPAP; however, because we did not have a control group, we cannot entirely exclude a partial placebo response. We consider placebo responses to CPAP to be unlikely because of the obtrusive and at times uncomfortable nature of the treatment. In addition a satisfactory CPAP placebo has not been devised in which to compare the clinical response to treatment. If a successful CPAP placebo is devised it will be another important method of testing the SAQLI for construct validity.

The initial testing of the questionnaire indicates that this population of patients with sleep apnea had a relatively normal distribution of responses without floor or ceiling effects (Table 5). The SAQLI showed a high degree of internal consistency as reflected by intraclass correlation coefficients (Cronbach's alpha) of 0.88 to 0.92.

During the item identification phase we assumed that patients would identify problems they had performing activities like their usual job, or with leisure activities. We had not anticipated that for many patients the problems they actually have is not whether they could do an activity or perform their job, but rather it was how much they had to push themselves to accomplish them and the lack of energy they had left over for other activities. The wording in the SAQLI reflects this in that most questions ask either how much difficulty patients have with something or how often they can do it rather than whether they can do it. It is possible that this characteristic of the SAQLI is population-specific, that is, it may be more or less important for people in different countries and in different cultures. It is also possible that responses to treatment will vary across cultures. Thus, it will be important for the SAQLI to be validated in different ethnic groups and trials of different modes of treatment.

We constructed the SAQLI to be administered by a trained interviewer using color-coded response cards with a seven-level Likert scale, similar in design to other quality of life questionnaires (12, 15). We found that patients reported a large number of symptoms caused by sleep apnea or snoring and by treatment of these disorders. We therefore elected to have patients select the five most important symptoms that affect them from a list of commonly reported symptoms. Provision was also made for patients to add other symptoms that are not on the list. This procedure has been used successfully in other disease-specific quality of life questionnaires (12, 13).

Initially, we elected to include a patient-based weighting factor for each domain that reflected how important the patients felt the domain was to their overall quality of life. However, we found it made a negligible difference in scores and added a level of complexity that some patients found confusing. Therefore, we elected not to weight the importance of individual questions or individual domains in the SAQLI, again in keeping with the construction of other disease-specific quality of life scales (12). The exception to this is Domain E (Treatment-related Symptoms). It was clear from the item selection phase that these symptoms had a much lower frequency×importance product than did the other domains (Table 2). Because we incorporated this domain to reflect the potential for adverse consequences on quality of life from treatment, it is subtracted from the total of Domains A to D prior to the total SAQLI score being calculated. This strategy has the effect of reducing the apparent benefit of treatment to reflect what we believe to be the "net" benefit. This is a new concept in the design of a quality of life instrument that we felt was important because all of the available treatments for sleep apnea and snoring are known to have substantial adverse effects in many patients. Because patients reported that on average this negative impact of treatment was less than the positive impact, we felt it was important for each patient to rate the relative positive and negative aspects of treatment and use this as a "weighting factor" for treatment-related symptoms. We believe that by including Domain E (Treatment-related Symptoms) and weighting it according to its relative impact on quality of life, the SAQLI more adequately reflects the net effect of treatment. As shown in Table 6 the average weighted reduction in the SAQLI score by Domain E is 0.4.

Although further testing of validity, reliability, and responsiveness are required we believe the methodology we used, plus the data on 24 patients with sleep apnea, indicates that this questionnaire, the only disease-specific sleep apnea quality of life instrument that has been developed, can be used in clinical trials. Our methodology was almost identical to other instruments that have been published and validated for disorders such as asthma and chronic obstructive lung disease, heart failure, and rhinoconjunctivitis (12). In each of these other disease states these indices have proved to be valid, responsive, and reliable and thus important outcome measures in clinical trials.

The true utility of the Calgary SAQLI will only be established by its use in different types of clinical trials and by several different investigators. We hope that other investigators will include it as an outcome measure in future clinical trials, especially those assessing a patient's response to treatment for sleep apnea.