Although observation largely takes into account the needs and abilities of individuals with profound intellectual and multiple disabilities, several difficulties are related to this assessment method as well. Our aim in this study was to investigate what possibilities the use of physiological measurements make available to validate alertness observations. Measurements of five physiological parameters were compared with video observations of three individuals with profound intellectual and multiple disabilities. Because our first findings are broadly in line with those of studies involving individuals in the general population, we hypothesize that physiological measurements can be used to validate alertness observations in individuals with profound intellectual and multiple disabilities. Future studies are needed to compensate for the limitations of this study and to answer ensuing questions.
Assessing the functioning of individuals with profound intellectual and multiple disabilities (PIMD) presents special challenges for researchers and assessors. Individuals in the target group experience a combination of severe intellectual and motor disabilities (Nakken & Vlaskamp, 2007). These disabilities are complicated by severe sensory disabilities and a broad range of health problems such as epilepsy, chronic pulmonary infections, and gastroesophageal reflux (Arvio & Sillanpää, 2003; Van Schrojenstein, Van den Akker, Maaskant, & Haveman, 1997; Van Splunder, Stilma, Bernsen, & Evenhuis, 2006). As a consequence of their disabilities, individuals with PIMD often do not use spoken language. They communicate by means of body language, using subtle gestures, sounds, or movements to express themselves (Petry, Maes, & Vlaskamp, 2005). Because of these disabilities, individuals with PIMD do not meet the requirements of standardized assessment instruments in terms of motor and speech abilities (Vlaskamp, 2005). At the same time, differences in combination and severity of disabilities result in a broad heterogeneity for this target population in terms of abilities and constraints (Nakken & Vlaskamp, 2007). This diversity also makes the development of specialized and standardized instruments that might be usable with this target population difficult (Vlaskamp & Cuppen-Fonteine, 2007).
One method that largely takes into account the needs and abilities of individuals with PIMD is observation. Although the subtle expressions of individuals in the target population are difficult to register, even more important is the ability to combine any registration with an interpretation of the signal. The meaning of a signal may be different for different individuals; even the same expression on the part of one individual may be interpreted differently in another situation. Consequently, assessors need to take the individual's communication into account as well as any possible contextual influences (Hogg, Reeves, Roberts, & Mudford, 2001; Petry & Maes, 2006). Only observation makes this possible.
However, several disadvantages are also related to observation as an assessment method. First, making observations is very time consuming. Although assessors may often confirm the value of the results of observation, they may not always feel able to carry out observations as part of their daily work (Petry & Maes, 2006). Second, all observations entail subjectivity on the observer's part (Hogg et al., 2001). Although it is important for assessors to take the meaning of their clients' behavior into account, several factors can bias their interpretation. Assessors interpret the behavior of the individual with PIMD on the basis of their knowledge of the individual, taking into account their own previous experiences with the individual in similar situations. Because this knowledge differs for each assessor, different assessors observing the same situation may give different scores. At the same time, general expectations concerning reactions and contextual factors in a specific situation can also influence the assessor's judgment. This subjectivity is a threat to the reliability of this kind of assessment method. As a result, researchers find themselves continually involved in discussions about influencing factors and explanations (Vlaskamp, 2005). Moreover, studies comparing proxy reports to the self-reports of individuals whose job it is to formulate them have shown that those reports are not always in agreement with each other (Lewis & Morrissey, 2010; Lunsky & Bramston, 2006; Perkins, 2007). Proxies' judgments should, therefore, be interpreted with caution.
In support and education of individuals with PIMD, observation is also used to determine alertness. Alertness has been described as an individual's level of interaction and engagement with the environment, which becomes manifest and observable in the individual's behavior (Munde, Vlaskamp, Ruijssenaars, & Nakken, 2009). Direct support professionals (DSPs) need to know the individual's alertness level before they begin to offer stimulation. Only if individuals are alert and focused on the environment can they be expected to process the stimuli presented to them on a conscious level (Nelson, Van Dijk, McDonnell, & Thompson, 2002). Consequently, alertness is one of the most important preconditions for learning and development (Munde et al., 2009; Munde, Vlaskamp, Ruijssenaars, & Nakken, 2011).
At the same time, assessing alertness presents challenges to both researchers and assessors. Although authors of previous studies have confirmed that it is indeed possible to determine alertness on the basis of the behavior of individuals with PIMD (Munde et al., 2009), alertness observations of individuals with PIMD run up against problems similar to those related to observation in general. No consensus has been arrived at for the scoring categories that determine alertness levels. The position of self-injurious behavior, for example, has been discussed in several studies of alertness observations. Although some researchers expect this behavior to occur at all alertness levels (Woodyatt, Marinac, Darnell, Sigafoos, & Halle, 2004), others have suggested that self-injurious behavior should not be described as an alertness level at all but as a form of communication (Siegel-Causey & Bashinski, 1997). Moreover, researchers and assessors still wonder which expressions of an individual with PIMD should be interpreted as an indication of which particular level of alertness. When an individual looks at a stimulus, for example, it can be unclear to the observer whether the individual is alert and concentrating on the stimulus or just staring in the direction of the stimulus without any actual focus. Differentiating between being alert and being withdrawn is especially difficult in such circumstances (Munde et al., 2009). Another point of discussion revolves around scoring frequency. Because of quick and irregular changes in alertness levels, some authors have pled the case for continuous scoring (Guess, Roberts, & Guy, 1999; Mudford, Hogg, & Roberts, 1997). However, the difference in content information when using continuous as compared with interval scoring is not so obvious. Moreover, continuous scoring is difficult to do in actual clinical practice. To measure and compare the impact of these differences, researchers have calculated the reliability of their observations. In many studies, the result was that reliability did not exceed the formulated criterion (Mudford et al., 1997; Woodyatt et al., 2004). Although different explanations for these results have been discussed (Arthur, 2000; Guess, Roberts, Behrens, & Rues, 1998; Mudford, Hogg, & Roberts, 1999), no solution for the problem of low reliability in alertness observations has been offered.
Because of the ongoing discussion about the difficulties and disadvantages of alertness observations, several studies have stated the need for complementary assessment methods to validate the observations (Mudford et al., 1997; Munde et al., 2011). However, finding methods that are in line with the needs and abilities of individuals with PIMD presents another challenge. In the literature, two alternative methods of determining alertness in the general population are regularly described: brain measurements and physiological measurements (Beauchaine, 2001; Dykman, Ackerman, Holcomb, & Boudreau, 1983). Because severe brain damage is the cause of their disabilities, however, brain measurements in individuals with PIMD are rarely carried out, and interpretations of the results remain a point of discussion (Kemner, Van der Gaag, Verbaten, & Van Engeland, 1999). Similarly, to date no studies using physiological measurements to determine alertness levels in individuals with PIMD have been available.
At the same time, many studies can in fact be found of individuals in the general population (e.g., Beauchaine, 2001; Boiten, 1998; Dykman et al., 1983; Thayer & Lane, 2000). Looking at the literature on physiological measurement in detail, several relationships between alertness and physiological expressions have been described. Although higher alertness accompanies a higher heart rate (Dykman et al., 1983; Thayer & Lane, 2000), changes in alertness are also often accompanied by changing heart rate levels (Bradley, 2009). Studies focusing on breath as a measurement of alertness have found that individuals in the general population show a quick and shallow respiration when they are alert (Beauchaine, 2001; Boiten, 1998). Consequently, high respiration rate, low tidal volume, and minute ventilation have been linked to high alertness levels. Those findings also accompany differences in respiratory sinus arrhythmia (RSA; Beauchaine, 2001), combining the measurement of both heart beat and respiration into one parameter. RSA measurements describe heart rate variability by using the ratio of the heart rate during an individual's expiration (Berntson, Cacioppo, & Quigley, 1993).
Moreover, the only study focusing on the target group of individuals with PIMD found that participants' physiological measurements differed for positive and negative emotions (Vos, De Cock, Petry, Van, & Maes, 2010). Vos et al. (2010) also suggested a relationship between emotions and alertness. Although participants focused their attention when presented with positive stimuli, they withdrew their alertness when presented with negative stimuli.
On the basis of these previous findings, our aim in this study was to explore the possibility of validating alertness observations of individuals with PIMD by using physiological measurements. Similar patterns in both methods could indicate convergent validity for the alertness observations. Moreover, the results could shed more light on the feasibility of physiological measurements when assessing individuals with PIMD.
Material and Methods
Participants were selected from among individuals attending two day care facilities in Flanders, the Dutch-speaking part of Belgium, on the basis of the following criteria. All participants had to be diagnosed as having PIMD according to the description of key characteristics of the target group provided by Nakken and Vlaskamp (2007). We excluded individuals with uncontrolled seizure disorder, hypersensitivity of the skin, and behavioral problems because any interference with the measurements might lead to invalid results. Moreover, for each client, one DSP who had known the client for at least 1 year had to be willing to participate in the study as well.
At the two facilities, DSPs were asked to select three clients who met these criteria. Participant 1 was a 23-year-old man with a developmental age of less than 6 months. His motor abilities were restricted by severe spasticity; he was not known to have any sensory problems. Because he experienced difficulties with swallowing, he received only mashed food. The second participant was a 23-year-old woman. Her mental age had been determined to be 3.5 months. After research, it was supposed that she most likely also had a visual impairment. However, this had not as yet been confirmed. In addition to severe motor disabilities, she had also been diagnosed with feeding and airway problems. She received her food via a stomach tube. Participant 3 was a 52-year-old man. His developmental age was between 6 and 12 months. Because of serious hypertension in his whole body, he was restricted in his movements. No sensory problems were known. He was given thickened food and drinks to avoid swallowing problems. All three participants had epilepsy that was kept under control with medical treatment. Informed consent for participation in this study and video registration was obtained from the parents or legal representatives.
Alertness Observation List (
Vlaskamp, Fonteine, & Tadema, 2005)
DSPs scored alertness levels using the Alertness Observation List. The four levels of alertness were each assigned a color: (a) active, focused on the environment (green); (b) inactive, withdrawn (orange); (c) sleeping, drowsy (red); and (d) agitated, discontented (blue). When four different observation forms had been completed, an individual alertness profile was formulated. The individual descriptions of each alertness level were complemented by concrete examples of behavior. Previous research has shown that the Alertness Observation List is a reliable instrument for determining alertness levels in individuals with PIMD (Munde et al., 2011; Petitiaux, Elsinga, Cuppen-Fonteine, & Vlaskamp, 2006).
The noninvasive ambulatory LifeShirt technology (VivoMetrics, Inc., Ventura, CA) was used to collect the physiological data. Various sensors woven into the LifeShirt continuously register a range of cardiopulmonary parameters. This study included measurements of pulmonary functioning and the electrical activity of the heart. These measurements were based on respiratory bands using inductance plethysmography and electrocardiography, respectively. Additionally, participants' activity levels were recorded using an accelerometer. An overview of the parameters is given in Table 1.
The study was conducted over 4 weeks. The first week was the preparation phase. The participants wore the LifeShirt for 30 min every day to become accustomed to the shirt and to minimize its effects on their functioning. During these sessions, DSPs filled in a questionnaire about the clients' reactions to the LifeShirt. The clients initially showed two types of reaction. Although they disliked having the skin conductance sensor attached to their finger, they enjoyed the additional individual attention from the DSPs. After the first week, the participants no longer showed any effects from wearing the shirt.
In the subsequent 3 weeks, six different situations of 15 min each were videotaped for each of the participants while they were wearing the LifeShirt. The aim was to gather an at-random selection of situations that were illustrative of the participants' everyday life. Although all the situations included stimulation of the participant, the content was chosen separately according to each participant's preferences and possibilities. To make the situations comparable for all participants, one situation involving individual stimulation, one involving a group activity, one involving a meal, and one in which the participant was alone with a stimulus were included. Each type of situation was videotaped and recorded three times. In the end, one situation of each type (six situations per person) was chosen at random to be used for further analysis. The two observers involved in this study were trained in the use of the Alertness Observation List and familiar with the aim of the study. DSPs had previously formulated individual alertness profiles that were now used as a framework for the scoring. The scoring differentiated among the four alertness levels described for the Alertness Observation List (see Instruments section). The Media Coder (Bos & Steenbeek, 2008) was the computer tool that facilitated the registration of the observations. To calculate the interobserver agreement, 20% of the videotapes were scored by two observers. The agreement, based on the general agreement formula (Mudford et al., 1997), was 82.6%.
To prepare the physiological data for analysis, all irregularities were removed. The data were calibrated using the qualitative calibration method available in the Vivologics software (VivoMetrics, Ventura, CA). Because some of the sensors were susceptible to movement, all values higher than 2.5 on the accelerometer (indicating the same amount of movement as walking) were excluded. All periods that included vocalization were also removed to avoid artifacts in the data. Finally, values for all parameters were extracted per breath.
In the analysis, we compared the alertness observations with the same-time physiological measurements. On the basis of previous studies, we explored two patterns in the data: Differences in physiological measurements were observed for the various alertness levels among samples of the general population and changes in alertness levels were associated with changes in heart rate and RSA.
We used SPSS Version 16 (SPSS, Inc., Chicago, IL) to explore the descriptive statistics of the data. Means of the physiological measurements of heart rate, respiration rate, RSA, tidal volume, and minute ventilation were compared with each other for the different alertness levels. In addition, mean heart rate and RSA during changes to the “alert” alertness level were compared with moments without changes. For this purpose, 5 s before and 5 s after the actual change were scored as the moment of change. Only those moments without any changes between the other alertness levels were coded as “no change.” Explorative analyses were conducted per person per situation.
Because none of the participants was asleep during the situations, this particular alertness level was excluded from further analysis. In addition, only those situations that revealed information relevant to this study were described in the analysis. Concerning the first part of the analysis, Participant 1 did not show withdrawn behavior in Situations 2, 4, and 6. In Situations 2 and 4, no changes in alertness level on the part of Participant 1 were observed for the second part of the analysis. Similarly, Participant 2 did not show any changes in terms of the alert alertness level in Situation 2 either. Consequently, six situations had to be excluded in total.
Comparing the mean heart rates for the different alertness levels with each other, Participant 1 had higher scores when he was alert than when he was withdrawn in two of the three situations. When he was agitated, heart rate scores were even higher in two situations and lower in one situation. In two of the three situations, the respiration parameters respiration rate and RSA were lower for the alert moments than for those moments when Participant 1 was withdrawn, whereas tidal volume and minute ventilation were higher during the alert moments. When Participant 1 was agitated, the means of the respiration parameters differed in all situations.
Moreover, heart rate during changes to the alert alertness level was higher than during moments without changes in two of the four situations. RSA scores, in contrast, were lower for three of the four situations.
Participant 2 had a higher heart rate during alert moments than in the withdrawn moments in three of the six situations. As with heart rate, Participant 2 had a higher respiration rate in three of the six situations. Scores for tidal volume and minute ventilation were higher during alert moments than during withdrawn moments in five of the six situations. RSA scores, in contrast, were lower in five situations when the participant was alert and higher when she was withdrawn. Participant 2 showed agitated behavior in only one situation.
Similar to Participant 1, Participant 2 showed higher heart rate during changes to the alert alertness level than during moments without changes in three of the five situations. In four of the six situations, RSA scores were lower when she was changing alertness levels.
Heart rate levels were higher in alert than in withdrawn moments for Participant 3 in four of the six situations. Mean scores for tidal volume, minute ventilation, and RSA were lower for the alert than the withdrawn alertness level in four of the six situations. For respiration rate, scores were higher. When Participant 3 was agitated (in three of the six situations), he showed even higher scores for all four parameters in two of these situations.
When he changed to the alert alertness level, heart rate was lower than during moments without changes in four of the six situations. RSA scores, in contrast, were higher in five of the six situations.
Overviews of the results for each participant and for each situation are given in Tables 2–7. Two tables per participant are provided, including the different alertness levels and changes to the alert alertness level, respectively.
The aim of this study was to explore the possibility of validating alertness observations by using physiological measurements. On the basis of the results, the following links were found. Heart rate was higher when the participants were alert than when they were withdrawn from the environment in 10 of the 15 situations. Furthermore, no clear differences in heart rate were found during changes to the alert alertness level compared with moments without changes. In seven situations, mean heart rate was higher during changes, whereas in eight situations, scores were lower. For the respiration parameters, a tendency became apparent for three parameters. Participants showed higher tidal volume and minute ventilation (nine of the 15 situations) and lower RSA (11 situations) when they were alert. Because respiration rate was only higher in eight of the situations, no conclusions can be drawn for this parameter. For the agitated moments, the participants' mean respiration parameters did not show any pattern.
Comparing our results with those of previous studies, our study is in line with the findings for the general population that revealed a relationship between high alertness levels and high heart rate (Dykman et al., 1983; Thayer & Lane, 2000). Freeman, Horner, and Reichle (1999) even stated that heart rate measurements were the most reliable method of determining alertness. On the basis of this study's results, we can hypothesize that this measurement might be reliable for individuals with PIMD as well. When the different alertness levels were linked to the respiration parameters, a similar pattern became apparent for two participants. During alert moments, two participants showed a higher tidal volume and higher minute ventilation. Scores on the two parameters were lower during the withdrawn moments. Although this relationship contradicts the literature, previous studies have also shown that patterns may differ for individuals with respiratory disorders (Beauchaine, 2001; Boiten, 1998). Because such disorders had been diagnosed in Participant 2, this might explain the different results found in the respiration parameters. For Participant 1, the small number of situations (three) that could be included in this comparison might explain the differences between the findings in the literature and the findings for him. Furthermore, Participant 3's scores were in line with the pattern previously revealed in studies of individuals in the general population. During alert moments, he had a decreased depth of breathing that became visible in the form of a lower tidal volume and lower minute ventilation. Looking at the different alertness levels and means of RSA, the patterns found in our study were again similar to the results of studies of the general population. Although sustained high alertness levels in these studies were related to low RSA levels, different RSA scores were found during changes in the alert alertness level compared with moments without changes (Beauchaine, 2001; Thayer & Lane, 2000). We also found those results in this study for 11 and 10 of the 15 situations, respectively. However, the position of the agitated alertness level in relation to the different parameters needs to be discussed. The different results for participants and parameters would seem to confirm the adjunct position of this alertness level (Vlaskamp, Fonteine, Tadema, & Munde, 2010). At the same time, however, RSA has been described as a measurement of the regulation of attention and emotion (Beauchaine, 2001; Thayer & Lane, 2000). Possibly, then, this regulation may be linked to the different functions of the agitated and sometimes self-injurious behavior of individuals with PIMD (Barrera, Violo, & Graver, 2007). Although individuals in the target group can show self-injurious behavior when they withdraw their attention from the environment in some situations, they may also use self-injurious behavior as a means to increase their alertness levels in other situations.
Although the results of this first exploratory study of ours are promising, we should mention several limitations. This study included no baseline measurements. Consequently, comparisons were not possible between participants, only within participants. To strengthen the results and gather general information about physiological measurements in individuals with PIMD, baseline measurements are necessary. Therefore, research into an individual's health problems should precede the measurements. When the individual with PIMD experiences airway problems, for example, the results of the respiration parameters have to be interpreted with caution. Future studies including all alertness levels would be especially useful to determine whether an ongoing line in physiological measurements can be found along the three different alertness levels. Moreover, because of the large heterogeneity of the target group and our small sample, this study's results cannot be generalized to the entire population. Therefore, future studies with larger samples are needed.
Although this study was only an exploratory case study, we feel able to hypothesize that heart and respiration measurements can be used to validate alertness observations in individuals with PIMD. The similarities between the results of the observations and the physiological measurements confirm that alertness and changes in alertness levels are observable. For assessment purposes, this study's results can be a valuable contribution to the ongoing discussion concerning the disadvantages and difficulties of observations. Just as individuals in the target group need pervasive support from the environment (Nakken & Vlaskamp, 2007), they also require assistance during the assessment. This need for assistance may, in turn, lead to a biased picture of the functioning of the individual with PIMD. When the assessors judge their client's functioning, they have to judge their own behavior at the same time. Physiological measurements can thus be used as a more objective complement to observations. In clinical practice, physiological measurements can support the DSP's alertness observations. Physiological measurements can be especially helpful in differentiating between alert, inactive, and withdrawn alertness levels, because these levels are difficult to detect simply through observation (Mudford et al., 1997). Including these measurements in scientific studies can greatly increase the validity of the research. Our study may be seen as a first step in the area of physiological research on individuals with PIMD; however, a large number of resulting questions will need to be answered in the future.
Editor-in-Charge: Glenn Fujiura
Vera Munde (e-mail: V.S.Munde@rug.nl), University of Groningen, Department of Special Needs Education and Youth Care, Grote Rozenstraat 38, 9712 TJ, Groningen, the Netherlands; Carla Vlaskamp, University of Groningen, Groningen, the Netherlands; Pieter Vos and Bea Maes, Katholieke Universiteit, Leuven, Belgium; and Wied Ruijssenaars, University of Groningen, Groningen, the Netherlands.