Critique of the Richards-Campbell Sleep Questionnaire reliability and validity assessment

Sleep Questionnaire (RCSQ) Instrument Analysis

Research to evaluate interventions to promote sleep in critically ill patients has been restricted by the lack of brief, inexpensive outcome measures (Richards, O'Sullivan, & Phillips, 2000). Usually the critically ill are given different medications that help them sleep. However, in many cases this is insufficient to achieve enough rest and sleep periods are generally fragmented and with decreased restrictiveness. Most researchers believe that a few days of partial sleep or even complete sleep deprivation in a healthy adult for brief periods is completely benign. However, for patients in critical care settings the effects of sleep deprivation can contribute to major complications.

Sleep is divided into two distinctive states, rapid eye movement (REM) and nonREM states. The former is defined by periods of episodic burst of rapid eye movements and the later (NREM) has a set of sub-stages that include for separate phases that can be identified through EEG activation. As an individual progress through the NREM cycles, each cycle last approximately eighty minutes each. Currently available techniques for measuring sleep include polysomnography (PSG), actigraphy, observation, and patient perception. PSG is the best tool for measurement and data collection on sleep, yet this requires expensive equipment and highly trained operators. Furthermore, it is especially challenging in a critical care setting because it takes over an hour to set up. Because this tool is so difficult to use in this setting, the Richards-Campbell Sleep Questionnaire (RCSQ) has been proposed as an effective proxy tool to collect data in critical care patients.

Description of the Instrument

The researchers who designed the RCSQ questionnaire were trying to find a valid alternative to the more comprehensive PSG tools since these instruments are simply not feasible in a critical care setting. There are five categories of information that the RCSQ tries to capture. These items are:

1. Sleep Depth

2. Falling Asleep

3. Number of Awakenings

4. Percent of Time Awake

5. Overall Quality of Sleep

The RCSQ items are constructed as visual analogs so that the patients could easily provide answers. To answer the questions the patients need only mark an "X" to the answer they are seeking. The patients' place the "X" on a visual scale in which must the data point must be recorded by measuring the distance of the "X" in millimeters from the low end of the scale. The content included in these categories were constructed by a panel of experts as well as items that were identified in a literature review of Medline articles that were published within the range of 1976 to May of 1999 using the key word "sleep." They also identified the most common sleep problems that patients in these settings are likely to encounter.

Psychometric Assessment

A pilot test was performed on the RCSQ from nine patients in critical care who also received a PSG test. The item on the questionnaire regarding light/deep sleep correlated significantly with the PSG sleep characteristic of percent stage 4 NREM (r=.59) and percent stage 3 NREM (r=.56). The item "A good night's sleep/A bad night's sleep correlated 0.64 for percent stage 2 and 0.55 for percent stage REM. The item "Fell asleep immediately/Never could fall asleep" was strongly associated with latency of sleep onset, although not statistically significant (r=.51, p=.07). With a sample size of only nine patients, the correlation must be ? .53 to be statistically significant at the 0.05 level (Richards, O'Sullivan, & Phillips, 2000). Thus with a larger sample size it is reasonable to suspect that latency question might be statistically significant as well.

Validity and Reliability Assessment

Although the preliminary data collected with the pilot study provided some insights that the scale could effectively achieve the objectives that it was intended for, the sample size was relatively small and the did not represent a good comparison with PSG data. Therefore a secondary and more comprehensive study was designed to include a bigger sample. In a private room of an eight-bed medical critical care unit, seventy men with the mean age of 65.81 agreed to participate in a new study. A Grass 16 Channel Electroencephalogram Machine (Grass Instrument Company, Quincy, Mass.) was used to collect a range of PSG data. Each subject was studied from approximately 10PM to 5AM and data from the RCSQ was collected as well during this time period.