Table 7 Hypotheses for convergent validity and responsiveness
From: Instrument completion and validation of the patient-reported apnea questionnaire (PRAQ)
PRAQ domain | Comparison instrument | Expected correlation strength (direction) | Explanation (hypothesis nr between brackets) |
|---|---|---|---|
Sleepiness | ESS | 0.5–0.8 (+) | The ESS asks about current daytime sleep propensity, while the PRAQ domain asks to look back on the past month and indicate how much of a problem sleepiness or falling asleep was. The domains do not cover the exact same construct, but are relatively similar. We expect a moderately strong correlation (h1). “Sleep-related impairment” is a domain with questions covering both sleepiness and tiredness. We expect a moderate to strong correlation with the PRAQ domain “sleepiness” (h2) because of the overlapping items on sleepiness, but also because the concept of sleepiness itself correlates with the concept of tiredness and how tiredness affects daily activities in our study population (correlation strength 0.57 as found in our principal component analysis). Dissimilar constructs We expect the correlations with the PROMIS fatigue and RAND-36 vitality domains to be lower than the correlations with the similar constructs of the PROMs mentioned above (h3). |
PROMIS: Sleep-related impairment | 0.5–0.8 (+) | ||
Different constructs | |||
PROMIS fatigue (+) RAND-36 vitality (−) | |||
Energy & daily activities | PROMIS: Sleep-related impairment | 0.6–0.9 (+) | “Sleep-related impairment” is a domain with questions covering both sleepiness and tiredness, in the context of daily activities. Therefore we expect a strong correlation with the PRAQ domain (h1). The RAND-36 vitality domain and the PROMIS fatigue both contain questions about tiredness. We expect both to have a strong correlation with the PRAQ domain (h2 + 3). We expect a somewhat stronger (h4) correlation with the PROMIS fatigue domain because this domain is focused on to what extent one feels tired (similar to the PRAQ), while the RAND focuses on how often one feels tired. The PROMIS domain “ability to participate in social roles and activities” asks question about to what extent someone is able to do their usual daily or social activities. We expect a strong correlation with the PRAQ domain because the constructs are similar (h5). Dissimilar constructs We expect the correlations with the ESS and the “PROMIS satisfaction with social roles and activities” domain to be lower than the correlations with the similar constructs of the PROMs mentioned above (h6). |
RAND-36 vitality | 0.6–0.9 (−) | ||
PROMIS Fatigue | 0.6–0.9 (+) | ||
PROMIS ability to participate in social roles and activities | 0.6–0.9 (−) | ||
Different constructs | |||
PROMIS satisfaction with social roles and activities (−) ESS (+) | |||
Emotions | PROMIS anger | 0.6–0.9 (+) | The three PROMIS domains contain items asking about how often certain emotions are felt, which is the same approach as the PRAQ-domain. The PRAQ-domain also contains all three of these types of emotions. Therefore we expect a strong correlation with all three of these domains (h1–3). Dissimilar constructs We expect the correlations of the PRAQ “emotions” domain with the ESS, RAND-36, and the PROMIS domains ability to participate in social roles and activities and “satisfaction with social roles and activities” to be lower than the correlations with the similar constructs of the PROMs mentioned above (h4). |
PROMIS anxiety | 0.6–0.9 (+) | ||
PROMIS depression | 0.6–0.9 (+) | ||
Different constructs | |||
ESS (+) RAND-36 vitality (−) PROMIS ability to participate in social roles and activities (−) PROMIS satisfaction with social roles and activities (−) | |||
Symptoms at night | PROMIS sleep disturbance | 0.2–0.6 (+) | The most similar domain that we included for the PRAQ domain “symptoms at night” is the PROMIS “sleep disturbance” domain. This domain contains items about whether patients are sleeping well. Even though a majority of the items in the PRAQ domain “symptoms at night” will affect the quality of sleep, the content of the two domains is very different. We therefore will not make a very precise hypothesis, and expect at least a low to moderate correlation (h1). Dissimilar constructs are not included for this domain. Due to the domain’s clinimetric nature it covers several topics that are related to severity of OSA and that might therefore have unpredictable correlations with other domains. |
Social interactions | PROMIS: Sleep-related impairment | 0.2–0.6 (+) | The “social interactions” domain of the PRAQ contains a collection of items about different social problems that apnea patients might experience due to their snoring, sleepiness, tiredness, or emotions. Because this domain does not clearly cover one single construct, we expect it to have no more than low to moderate correlations with any of the comparator PROMIS domains (h1–3). Dissimilar constructs are not included for this domain. Due to the domain’s clinimetric nature it covers several topics that are related to severity of OSA and that might therefore have unpredictable correlations with other domains. |
PROMIS ability to participate in social roles and activities | 0.2–0.6 (−) | ||
PROMIS satisfaction with social roles and activities | 0.2–0.6 (−) | ||