Research in Gerontological Nursing

The articles prior to January 2010 are part of the back file collection and are not available with a current paid subscription. To access the article, you may purchase it or purchase the complete back file collection here

Original Research: Descriptive Correlational 

New Parameters for Daytime Wandering

Donna L. Algase, PhD, RN; Cathy L. Antonakos, PhD; Elizabeth Beattie, PhD, RN; Cynthia A. Beel-Bates, PhD, RN; Lan Yao, PhD, RN

Abstract

This study aimed to describe wandering using new parameters and to evaluate parameters as a function of cognitive impairment and mobility. Forty-four wanderers in long-term care settings were videotaped 12 times. Rate and duration of wandering episodes were plotted and used to derive parameters from values above and below case medians, proportion of hours wandering, and time of day. Participants wandered during 47% of observations; on average, the hourly rate was 4.3 episodes, the peak hourly rate was 18 episodes, and the peak hourly duration was 19.9 minutes. Mini-Mental State Examination (MMSE) scores was negatively correlated with overall duration and number of observations during which duration exceeded 15 minutes per hour, was positively correlated with number of observations without wandering, and was not significantly correlated with rate-related parameters. Mobility correlated positively with rate and duration parameters. Interaction of MMSE score and mobility was the strongest predictor of wandering duration. Parameters derived from repeated measures provide a new view of daytime wandering and insight into relationships between MMSE score and mobility status with specific parameters of wandering.

Abstract

This study aimed to describe wandering using new parameters and to evaluate parameters as a function of cognitive impairment and mobility. Forty-four wanderers in long-term care settings were videotaped 12 times. Rate and duration of wandering episodes were plotted and used to derive parameters from values above and below case medians, proportion of hours wandering, and time of day. Participants wandered during 47% of observations; on average, the hourly rate was 4.3 episodes, the peak hourly rate was 18 episodes, and the peak hourly duration was 19.9 minutes. Mini-Mental State Examination (MMSE) scores was negatively correlated with overall duration and number of observations during which duration exceeded 15 minutes per hour, was positively correlated with number of observations without wandering, and was not significantly correlated with rate-related parameters. Mobility correlated positively with rate and duration parameters. Interaction of MMSE score and mobility was the strongest predictor of wandering duration. Parameters derived from repeated measures provide a new view of daytime wandering and insight into relationships between MMSE score and mobility status with specific parameters of wandering.

Wandering is a risky and challenging behavior, occurring at some point for most people with dementia. Owing to the variety of definitions and methods to study wandering, researchers have produced neither a full description nor a complete understanding of the behavior as yet. Our purpose was to extend descriptive knowledge of wandering within an established paradigm. Building from our prior work (Algase, Beattie, Bogue, & Yao, 2001; Algase, Beattie, & Therrien, 2001; Algase, Kupferschmid, Beel-Bates, & Beattie, 1997), in this study we derived a comprehensive empirically based set of parameters for wandering based on serial daytime observations and examined them as functions of cognitive impairment and mobility.

Despite an increasing number of wandering studies, surprisingly few focus on measuring and describing it. Recently, the International Consortium for Research on Wandering proposed a provisional consensus definition of wandering on the basis of analysis and synthesis of terms and definitions from 183 published papers. Accordingly, wandering is “a syndrome of dementia-related locomotion behavior having a frequent, repetitive, temporally-disordered, and/or spatially-disoriented nature that is manifested in lapping, random, and/or pacing patterns, some of which are associated with eloping, eloping attempts, or getting lost unless accompanied” (Algase, Moore, Vandeweerd, & Gavin-Dreschnack, 2007, p. 696). These authors urged that future work be situated within this definition to evolve evidence that supports, refutes, or refines meaning of the term and validity and reliability of its measures. Herein conception and measurement of wandering are so aligned.

Observational Studies of Wandering

Science has long relied on observation as the tool for delineating and describing phenomena, particularly those, such as wandering, about which relatively little is known. Observational approaches to describe wandering as it naturally occurs have been used in eight studies, all in long-term care settings. Definitions, approaches, parameters, and major findings of each are summarized in Table 1. Eight other observational studies were found but excluded from Table 1 because natural conditions were manipulated or augmented as part of an intervention to affect wandering.

Summary of Methodological Aspects and Findings from Eight Observational Studies of WanderingSummary of Methodological Aspects and Findings from Eight Observational Studies of Wandering

Table 1: Summary of Methodological Aspects and Findings from Eight Observational Studies of Wandering

Conception and Definition

All but two studies explicitly defined wandering; all definitions encompassed the notion of locomotion but varied on presence/absence of goal directedness. Rhythm theory, casting each wandering episode as an ultradian rhythm with locomoting and nonlocomoting phases, guided definition in three studies (Algase, Beattie, Bogue, et al., 2001; Algase, Beattie, & Therrien, 2001; Algase et al., 1997).

Observational Approach

Direct real-time observation was the usual method but may have been supported with technology (e.g., bar coders, portable computers) to capture data and/or augmented with other forms of data (e.g., field notes, audiotapes, activity monitors). Overall time frames for observing a given participant were as short as one 24-hour day to as long as 3 months. Within the time frame, time sampling strategies varied markedly. Individual observation periods were as short as one ambulation episode to as long as 24 continuous hours. Total time for observing a given participant was between 180 and 2,037 minutes. In general, time frames and time sampling strategies lacked explicit rationale, although circadian rhythm was considered in some studies.

Observed Parameters of Wandering. In all studies but one (Lucero, Hutchinson, Leger-Krall, & Wilson, 1993), locomotion episodes were coded or classified. Four used the same typology of pattern (i.e., random, lapping, pacing, direct studies) (Algase, Beattie, Bogue, et al., 2001; Algase, Beattie, & Therrien, 2001; Algase et al., 1997; Martino-Saltzman, Blasch, Morris, & McNeal, 1991); otherwise, codes were unique to each study (Cohen-Mansfield, Werner, Marx, & Friedman, 1991; Matteson & Linton, 1996; Snyder, Rupprecht, Pyrek, Brekhus, & Moss, 1978). Aspects of gait and posture, as well as associated behaviors (e.g., exit seeking, doorknob fiddling), were reported by Lucero et al. (1993); two others reported location (Cohen-Mansfield et al., 1991; Snyder et al., 1978); and three reported temporal elements associated with locomotion (e.g., duration of episodes, lag between episodes, hour of the day) (Algase, Beattie, Bogue, et al., 2001; Algase, Beattie, & Therrien, 2001; Algase et al., 1997).

Descriptive Findings. In all studies, data about individual wandering episodes were collapsed within and/or across observations. In quantitative studies, data were considered as proportions of observations containing wandering or with metrics (e.g., rate) reflecting individual patterns (e.g., random). Due to differences in time sampling and quantification, findings cannot be readily compared across studies; rather, each contributes a unique description of wandering. Further, in several studies, wandering was compared with other variables, such as social interaction, cognitive impairment, and environmental features. Capacity for independent ambulation was an inclusion criterion in most studies, yet none evaluated mobility relative to wandering parameters.

Although all studies used repeated observations, resulting descriptions reduced multiple observations to a few overall variables for wandering. Data reduction is appealing because it enables straightforward analysis of large complex datasets and makes sense clinically, as care providers seldom concern themselves with individual wandering episodes, focusing more on overall volume, timing, or problematic kinds of wandering. However, data reduction also misses potentially valuable information and overlooks the clinical need to examine patterns of change within the phenomenon over time. Approaches to analysis depicting characteristics of wandering over time, such as during the course of a day, can provide a more detailed view of the behavior and may also contribute to development of clinically useful assessment parameters. No observational study has used multiple observation periods, viewed as a set across the course of a day, to derive parameters for describing wandering based on its frequency and duration. Our study addresses this gap.

Theoretical Framework

This study was constructed within the context of the Need-driven Dementia-compromised Behavior (NDB) model (Algase et al., 1996) in which stable (in the short term) characteristics of people with dementia (e.g., age, gender, mobility status, cognitive impairment, circadian rhythm) are regarded as putative risk factors for wandering and other need-driven dementia-related behaviors. While circadian rhythm was not evaluated as a predictor of wandering in this study, the rhythmic nature of wandering (i.e., alternating phases of locomoting and nonlocomoting behavior) was used to guide measurement, consistent with the provisional definition of wandering.

Thus, our aims were to describe frequency/repetition and temporal characteristics of wandering in long-term care settings as it occurs across daytime hours, and to evaluate the extent to which cognitive impairment and mobility affect new parameters. Research questions were:

  • How do new parameters of daytime wandering and direct (nonwandering) ambulation of people with dementia in long-term care settings compare?
  • Do values vary as a function of cognitive impairment and mobility level?

Method

Design

A repeated measures approach was taken within the context of a larger study to test the NDB model (NR04569, principal investigator: D.L.A.). In the parent study, 183 ambulatory people with dementia from 22 nursing homes and 6 assisted living facilities were videotaped in their natural surroundings. Twelve observations occurred according to a schedule that distributed observations randomly across 2 days such that each participant was observed once during each hour between 8:00 a.m. and 8:00 p.m. Participants were randomly assigned to one of four such schedules. Planned length of observations was 20 minutes; the average length was 19.85 minutes (SD = 0.44, range = 18.85 to 20.27). Study procedures were approved by institutional review boards of two participating universities; each study site received a single project assurance federally.

Sample

Participants with a full complement of 12 observations in the parent project (n = 54) were identified; 44 who demonstrated wandering during one or more observations were chosen for this analysis. These 44 participants met the following inclusion criteria: spoke English, met Diagnostic and Statistical Manual of Mental Disorders, fourth edition (American Psychiatric Association, 1994), criteria for medical diagnosis of dementia, scored less than 24 on the Mini-Mental State Examination (MMSE) (Folstein, Folstein, & McHugh, 1975), were ambulatory (with or without an assistive device), and maintained a stable regimen of psychotropic medications, if any, during the prior 30 days.

The sample was 79.6% women and had a mean age of 83.7. Mean MMSE score (n = 26) was 10.5 (SD = 5.7), at the cusp of moderate to severe impairment. Participants did not differ from others in the parent study on gender, age, MMSE scores, mobility, or residence (proportion in nursing homes versus assisted living facilities). A comparison of wandering in nursing homes and assisted living facilities did not reveal differences in the behavior (Beattie, Song, & LaGore, 2005).

Measures

Cognitive Impairment. This was assessed using the MMSE (Folstein et al., 1975), a valid and reliable tool (Folstein & McHugh, 1979; Tsai & Tsuang, 1979) used in standard dementia assessments. It taps orientation, registration, attention and calculation, recall, language, and construction tasks to yield a global performance score between 0 (severely impaired) and 30 (no impairment). To minimize missing data and floor effects of the MMSE, participants too impaired to complete testing were assigned a score of −1.

Mobility. On the basis of their most recent Minimum Data Set (MDS) 2.0 data, participants were designated as independent if records indicated independent performance on all items in Section G (i.e., bed mobility, transfer, walk in room, walk in corridor, locomotion on unit and off unit) or as assisted if not independent in performing on any item. The MDS is reasonably reliable and valid for assessing a wide range of resident conditions (Hawes et al., 1997; Hawes et al., 1995; Phillips & Morris, 1997). In assisted living facilities, where the MDS is not required, research nurse practitioners completed mobility items for participants. After obtaining consent, MMSE and mobility data were collected by a doctorally prepared research nurse, trained by a consulting neuropsychologist.

Wandering. Wandering was quantified from videotapes of participants’ behavior. All observations were videotaped using a rolling tripod-mounted camera focused on the participant’s entire body from approximately 10 to 30 feet. Two research assistants conducted the videotaping, avoiding unwarranted invasion of privacy.

Videotapes were later coded using Noldus Information Technology’s The Observer® 5.0 software. The start of each ambulation episode (three steps forward) was identified and timed to its cessation (change of position to sitting or standing with no forward movement after 15 seconds). All episodes were coded for pattern using Martino-Saltzman et al.’s (1991) typology:

  • Lapping (a repeating path involving at least three points).
  • Pacing (a repeating path back and forth between two points).
  • Random (a continuous path with multiple points in no particular order).
  • Direct (a single straightforward path from one location to another).

Coders were trained to meet a gold standard of 95% accuracy in recognizing and coding behaviors against a training videotape. Quarterly reliability checks sustained coding standards. Consistent with prior studies using this typology, lapping, pacing, and random patterns were considered collectively as wandering; the direct pattern signified nonwandering ambulation (Algase, Beattie, Bogue, et al., 2001; Algase, Beattie, & Therrien, 2001; Algase et al., 1997).

Data Analysis

Description of New Wandering Parameters. Rate (frequency of episodes within an observation period) and duration (total length of episodes within an observation period) were calculated for each observation and extrapolated to an hourly basis (rate = episodes per hour and duration = minutes per hour) to standardize for variation in observation lengths. Resulting rates and durations of wandering were plotted separately for each observation and compared with a participant’s median values across all observations to derive new parameters for this study. The Figure provides an example for one participant.

Hourly Plots of Wandering Rate (frequency of Episodes Within an Observation Period Extrapolated to 1 Hour) and Wandering Duration (sum of Episode Lengths Within an Observation Period Extrapolated to 1 Hour) for One Participant Compared with the Participant’s Median Values Across All Observations.

Figure. Hourly Plots of Wandering Rate (frequency of Episodes Within an Observation Period Extrapolated to 1 Hour) and Wandering Duration (sum of Episode Lengths Within an Observation Period Extrapolated to 1 Hour) for One Participant Compared with the Participant’s Median Values Across All Observations.

As the Figure shows, hours with high rates (0900, 1100, 1300, 1600, 1700, 1800) and high durations (0800, 1100, 1200, 1400, 1500, 1600) may not coincide. Cut-off points for low/moderate and high rates and durations were based on clinical judgment and used to derive additional parameters. Percentage of observations with any wandering also was calculated. All parameters are defined in Table 2.

Wandering Parameters Based on 12 Observations per ParticipantWandering Parameters Based on 12 Observations per Participant

Table 2: Wandering Parameters Based on 12 Observations per Participant

Direct Walking. As applicable, we constructed parallel parameters to summarize direct walking for comparison purposes. These included percentage of hours with any direct walking by observation and averaged for the participant, rate (frequency per hour), duration (minutes per hour), and average duration of each direct episode (in minutes).

Wandering rate and duration parameters were compared with MMSE scores and mobility categories using analysis of variance to compare categorical with continuous variables. Chi-square tests of association were used to compare categorical variables. Pearson’s correlations were used to estimate association between continuous variables. The p value was set at < 0.05.

Results

Parameters of Daytime Wandering

Wandering Rate. Estimated values for rate parameters are presented in Table 2. People with dementia wandered during almost half (47%) of the observations at an average rate of 4.37 episodes per hour. Maximum rate averaged 18.39 episodes per hour (range = 2.92 to 56.83) and occurred in the afternoon for half of the participants, between 2:00 and 3:00 p.m. on average. A majority (70%) wandered at a rate of more than 6 episodes per hour during at least one observation. On average, they wandered less than 6 episodes per hour during 26.7% of observations, and more than 6 episodes per hour during 20.64% of observations.

Wandering Duration. For estimated values of duration parameters see Table 2. Total wandering duration (summed across observations) averaged 63.29 minutes per participant (range = 2.5 to 358.21 minutes). Individual wandering episodes averaged 0.57 minutes in duration. Maximum duration was 19.87 minutes per hour (range = 2.28 to 54.9 minutes) on average.

As for peak rate, peak duration occurred in the afternoon for almost half (47.7%) of the participants, again between 2:00 and 3:00 p.m. People with dementia wandered fewer than 15 minutes per hour during 35.8% of observations and more than 15 minutes per hour during 11.6% of observations. Nearly half (47.73%) of them wandered for fewer than 15 minutes per hour during one or more observations.

Direct Walking. On average, participants displayed direct walking during fewer observations (37.69%) than they wandered (47%). The average rate of direct walking, 3.48 episodes per hour (range = 0.25 to 16.58), was also lower when compared with wandering (4.37 episodes per hour). Total direct walking across observation periods averaged 12.58 minutes per participant (range = 0.42 to 37.83), approximately one fifth that of wandering. Average duration of a direct walking episode, 0.37 minutes (or 22 seconds), was less than half that of a wandering episode. Descriptive statistics for direct walking appear in Table 2.

Daytime Wandering as a Function of Cognitive Impairment and Mobility Status

Parameters for wandering and direct walking were compared with cognitive impairment and mobility. MMSE score was treated both as categorical and as an ordered-categorical (continuous) variable. Mobility was treated categorically. Two-sample t tests and one-way analysis of variance were used to estimate associations between continuous and categorical variables. Pearson’s correlations were used to estimate associations between continuous variables. Two participants were missing cognitive impairment scores, so they were not included in analysis involving cognitive impairment.

Cognitive Impairment. No rate parameter was significantly associated with cognitive impairment; however, several duration parameters were. The percentage of observations with low/moderate wandering duration (more than 0 minutes per hour to fewer than 15 minutes per hour) was significantly lower among participants with lower MMSE scores (greater impairment) (r = 0.43, p = 0.005, n = 42). The percentage of observation periods with high wandering duration (more than 15 minutes per hour) was significantly higher among participants with lower MMSE scores (r = −0.33, p = 0.03, n = 42). Participants who ever wandered 15 or more minutes per hour had lower MMSE scores (greater impairment) (F[1,40] = 4.06, p = 0.051), approaching significance. No measure of direct walking was significantly associated with cognitive impairment.

Mobility. Mobility was not significantly associated with overall wandering rate. More hours with peak wandering rates (above the participant’s median rate) occurred among participants with independent mobility (t = 1.87, p = 0.068, N = 44). Participants with independent mobility also had more duration peaks (above their median durations) than participants requiring assistance (t = 2.82, p = 0.007, N = 44). The association between mobility and total wandering duration approached significance, with longer duration among independently mobile participants (t = 1.72, p = 0.093, N = 44). Although not significant, independently mobile participants had a higher percentage of observations with wandering duration at or more than 15 minutes (t = 1.96, p = 0.057, N = 44). Direct walking was not significantly associated with mobility.

Interaction Effects of Cognition and Mobility. Because people with dementia who have the most severe cognitive impairment (untestable) and those who are independently mobile had the highest percentage of observations in the high duration range, interaction effects were evaluated. To properly contrast participants with these characteristics, cognitive impairment was recoded into two categories for MMSE (greater than 0 and untestable). At the first order (without interaction term), the model was significant (F = 5.07, df = 2, p = 0.0110), with an adjusted R2 of 0.1656; also significant were both the recoded MMSE score (F = 6.88, df = 1, p = 0.0123) and mobility (F = 4.35, df = 1, p = 0.0437). At the higher order, adding the interaction term, again the model was significant (F = 4.97, df = 3, p = 0.0052), with an adjusted R2 of 0.2252; the interaction term approached significance (F = 4.00, df = 1, p = 0.0527); neither the recoded MMSE score nor mobility alone reached significance. In regression analysis, coefficients obtained for the higher order model were −0.0060 for MMSE score, 0.0026 for mobility status, and 0.2024 for the interaction term.

Discussion

All new wandering parameters for people with dementia demonstrated adequate variability, supporting an assertion that, as a set, these parameters are sensitive enough to portray a detailed picture of wandering. As such, they may be suitable for use in subsequent studies to detect subgroups or kinds of wandering or as a basis for a clinical assessment protocol.

Although nearly half of the observations captured some wandering, overall rate and duration was in the low/moderate range, less than 6 episodes per hour, and lasted fewer than 15 minutes per hour. However, both overall peak rate and peak duration were well within the high range (6 or more episodes per hour and 15 or more minutes per hour) and occurred in the afternoon, mostly between 2:00 and 3:00 p.m., for approximately half of the participants. Yet, a large number of participants peaked at other times of the day. Time of day for maximum peak may be an important parameter in differentiating wandering types in future studies and also may serve to isolate those whose wandering reflects circadian rhythm disturbances, which have been documented in a subset of people with dementia (Satlin et al., 1991; Satlin, Volicer, Stopa, & Harper, 1995).

While revealing some similarities, rate and duration parameters were not parallel in their distributions. A large majority (70%) demonstrated a high wandering rate at least once, where slightly less than half wandered for a long duration at least once. Furthermore, in just more than 20% of observations on average, people with dementia demonstrated high rates of wandering, where in only slightly more than 11% on average did they wander for a long time. Thus, rate and duration, while related, reflect a somewhat different picture of wandering and thereby, each group of parameters is essential to provide a full view of it. In comparison, direct walking (nonwandering) occurs less frequently and for shorter durations on every comparable parameter, clearly reflecting a different kind of walking.

Differences in the pictures reflected by wandering rate and duration parameters are further demonstrated in their relationship to cognitive impairment and mobility. Rate parameters were not a function of either cognitive impairment or mobility. Conversely, participants with mild cognitive impairment had a greater proportion of observations classified to the low/moderate range of duration and a lower proportion to the high range, whereas participants who wandered in the high range for duration at least once had significantly lower MMSE scores. Mobility also had significant (or nearly so) effects on several duration parameters. Participants who were independently mobile had longer overall wandering durations, more duration peaks, and a higher proportion of observations in the high range for duration. Mobility itself enhances capacity to wander. Consistent with our previous studies, direct walking was not a function of cognitive impairment (Algase, Beattie, Bogue, et al., 2001; Algase, Beattie, & Therrien, 2001; Algase et al., 1997) or mobility.

An interaction effect of cognitive function and mobility status was demonstrated on the proportion of observations in the high range for duration, where the higher order model explained a greater proportion of the variance than did the first order model, and only the interaction term was (nearly) significant. This finding indicates that better mobility further enhances the capacity of the most severely impaired people with dementia to wander. Overall, findings on cognition and mobility support the NDB model, which posits that cognitive impairment and mobility have direct effects on wandering (Algase et al., 1996), and demonstrate that effects are greater in combination than when either independent variable is evaluated alone. Nonetheless, cognitive impairment and mobility accounted for less than one quarter of wandering. Other characteristics of wanderers (e.g., health status or stamina, personality, need-state) or of the environment (e.g., noise, crowding), which were not evaluated in this study, likely contribute to the behavior.

Limitations

While rate and duration parameters presented somewhat differing pictures of wandering in this study, four points should be recognized. First, lacking norms for setting cut-off points to establish low/moderate and high groups for rate and duration parameters, we based them on clinical judgment. Resulting distributions across these groups were examined and appeared consistent with our experience. Post hoc comparisons of means for overall wandering rate and duration revealed that means for both parameters were well within the low/moderate group and cut-off points were 0.37 SD above the mean and just below the 75th percentile for rate and 2.02 SD above the mean and at the 90th percentile for duration. If standardized at the same SD or percentile for both parameters, differences in the pictures presented by rate and duration parameters would have been reduced, as fewer observations would have been in the high rate group and more in the high duration group.

Second, all parameters rely on coders classifying ambulation by pattern and are subject to error. Training coders to a standard and monitoring performance as we did reduces this problem. The potential for error was mitigated further by combining patterns indicative of wandering together, eliminating misclassification among such patterns, but not between wandering and direct nonwandering ambulation. Notable differences between parameters for direct ambulation and wandering suggest these classifications represent two distinct kinds of walking.

Third, in univariate analyses for MMSE and mobility, some rate-related parameters approached significance; a larger sample could have provided greater power to detect such effects among more parameters for both rate and duration. Post hoc power analyses revealed that samples between 52 and 111 would have been needed for 80% power to detect differences at p < 0.05 for parameters that were nearly significant in this study. However, as a secondary analysis, the sample was limited by availability of qualifying participants from the parent study.

Finally, standardization of hourly rate and duration may not reflect actual hourly rate and duration. It is possible that observation periods happened to capture that segment of an hour where wandering was highest or lowest, thus inflating or deflating rate or duration for that hour. While rate and duration for any particular hour may be so affected, such effects are likely evened out over the 12 observations.

Summary

This study proposed and estimated new parameters for wandering and evaluated effects of cognitive impairment and mobility on them. Findings reveal that rate-related and duration-related parameters present somewhat different pictures, both of which are important to a full view of wandering and may be relevant to delineating wandering types in subsequent studies or as a basis for developing clinical assessment protocols. Further, evaluating effects of cognitive impairment and mobility revealed greater sensitivity among duration-related than rate-related parameters and greater impact in combination than individually. Results support and extend major propositions of the NDB model.

References

  • Algase, DL, Beattie, ER, Bogue, EL & Yao, L2001. The Algase Wandering Scale: Initial psychometrics of a new caregiver reporting tool. American Journal of Alzheimer’s Disease and Other Dementias, 16(3),141–152. doi:10.1177/153331750101600301 [CrossRef]
  • Algase, DL, Beattie, ER & Therrien, B2001. Impact of cognitive impairment on wandering behavior. Western Journal of Nursing Research, 23, 283–295. doi:10.1177/01939450122045159 [CrossRef]
  • Algase, DL, Beck, C, Kolanowski, A, Whall, A, Berent, S & Richards, K et al. . 1996. Need-driven dementia-compromised behavior: An alternative view of disruptive behavior. American Journal of Alzheimer’s Disease and Other Dementias, 11(6),10–19. doi:10.1177/153331759601100603 [CrossRef]
  • Algase, DL, Kupferschmid, B, Beel-Bates, CA & Beattie, ER1997. Estimates of stability of daily wandering behavior among cognitively impaired long-term care residents. Nursing Research, 46, 172–178. doi:10.1097/00006199-199705000-00008 [CrossRef]
  • Algase, DL, Moore, DH, Vandeweerd, C & Gavin-Dreschnack, D2007. Mapping the maze of terms and definitions in dementia-related wandering. Aging & Mental Health, 11, 686–698. doi:10.1080/13607860701366434 [CrossRef]
  • American Psychiatric Association. 1994. Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.
  • Beattie, ER, Song, J & LaGore, S2005. A comparison of wandering behavior in nursing homes and assisted living facilities. Research and Theory for Nursing Practice, 19, 181–196.
  • Cohen-Mansfield, J, Werner, P, Marx, MS & Friedman, L1991. Two studies of pacing in the nursing home. Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 46, M77–M83.
  • Folstein, MF, Folstein, SE & McHugh, PR1975. “Mini-mental state.” A practical method for grading the cognitive state of patients for the clinician”. Journal of Psychiatric Research, 12, 189–198. doi:10.1016/0022-3956(75)90026-6 [CrossRef]
  • McHugh, PR & Folstein, MF1979. Psychopathology of dementia: Implications for neuropathology. In Katzman, R (Ed.), Congenital and acquired cognitive disorders (pp. 17–30). New York: Raven Press.
  • Hawes, C, Morris, JN, Phillips, CD, Fries, BE, Murphy, K & Mor, V1997. Development of the nursing home Resident Assessment Instrument in the USA. Age and Ageing, 26(Suppl. 2),19–26.
  • Hawes, C, Morris, JN, Phillips, CD, Mor, V, Fries, BE & Nonemaker, S1995. Reliability estimates for the minimum data set for nursing home resident assessment and care screening (MDS). The Gerontologist, 35, 172–178.
  • Lucero, M, Hutchinson, S, Leger-Krall, S & Wilson, HS1993. Wandering in Alzheimer’s dementia patients. Clinical Nursing Research, 2, 160–175. doi:10.1177/105477389300200206 [CrossRef]
  • Martino-Saltzman, D, Blasch, BB, Morris, RD & McNeal, LW1991. Travel behavior of nursing home residents perceived as wanderers and nonwanderers. The Gerontologist, 31, 666–672.
  • Matteson, MA & Linton, A1996. Wandering behaviors in institutionalized persons with dementia. Journal of Gerontological Nursing, 22(9),39–46.
  • Phillips, CD & Morris, JN1997. The potential use of administrative and clinical data to analyze outcomes for the cognitively impaired: Evaluating the minimum data set for nursing homes. Alzheimer Disease and Associated Disorders, 11(Suppl. 6),162–167.
  • Satlin, A, Teicher, MH, Lieberman, HR, Baldessarini, RJ, Volicer, L & Rheaume, Y1991. Circadian locomotor activity rhythms in Alzheimer’s disease. Neuropsychopharmacology, 5(2),115–126.
  • Satlin, A, Volicer, L, Stopa, EG & Harper, D1995. Circadian locomotor activity and core-body temperature rhythms in Alzheimer’s disease. Neurobiology of Aging, 16, 765–771. doi:10.1016/0197-4580(95)00059-N [CrossRef]
  • Snyder, LH, Rupprecht, P, Pyrek, J, Brekhus, G & Moss, T1978. Wandering. The Gerontologist, 18, 272–280.
  • Tsai, L & Tsuang, M1979. The mini-mental state test and computerized tomography. American Journal of Psychiatry, 136(4A),436–438.

Summary of Methodological Aspects and Findings from Eight Observational Studies of Wandering

Study N/na Observational Approach Concepts/Codes/Variables Description of Wandering Behavior
Snyder et al. (1978) 16/8 Direct observation; one 10-minute period/hour for 2 days; 18 total observations (180 minutes); movement traced on floor plan. Initial location, subsequent movement, zone, social behaviors. Wanderers moved about an average of 32.5% (high = 51.7%) of observed time (39%, excluding sleep)b; traversed 5.6 zones per observation periodb, higher proportion of observation periods alone or sleeping, 41% versus 26.1% for controls.b
Martino-Saltzman et al. (1991) 40/24 Videorecorder triggered by electronic tag in select detection areas; excludes private rooms and dining rooms; data were collected 24 hours/day, 7 days/week for 4 weeks. Ambulation was coded using four travel patterns, developed by consensus among observers after initial viewing of videotapes. Inefficient patterns were random, lapping, and pacing; efficient travel was direct. Among wanderers, inefficient travel (wandering) constituted 19.7% of all ambulation, lapping 17.3%, random 1%, and pacing 0.1%. Perceived wanderers made 56.6% of all travel events, perceived nonwanderers the remainder. Nonwanderers also showed inefficient travel but at lower rates: lapping 4.3%, random 0.4%, and pacing 1.3%. Proportion of inefficient travel grew with worse cognitive impairment from 5.3% for mildly impaired to 23% for severely impaired.
Cohen-Mansfield et al. (1991) n = 6 Direct observation; 3 minutes/hour; stratified random sampling of hours over 24 hours. Observations spanned 3 months; 679 observations (2,037 minutes) per participant. Frequency of pacing (defined as walking aimlessly, wandering, constant searching); social and physical environment; interpersonal distance; physical location. Pacing occurred in 55% of observations; in 77% of these, pacing was continuous. Compared with other times between 7:30 a.m. and 11 p.m., pacing occurred less during meal times, when near (within 3 feet) of others, and when it was dark, and it occurred more when in corridors and when noise levels were low.b
Lucero et al. (1993) 10/10 Videorecorded and supplemented with field notes and/or narration by observer for 4 hours/participant over 2 months, covering various times of the day between mid-morning and midnight; time sampling methods not described. Time of day, body position, activity, and social interaction. Qualitative description of activity and abilities observed among participants in middle- and late-stage dementia during unstructured time. Wandering behaviors noted in middle-stage dementia included stooped posture, mild shuffling gait, good balance, steps over small obstacles in path, may explore, walks around communal areas (sometimes socializing), may enter rooms at ends of hallways, rummages, may attempt to leave during outbursts of other residents or with the coming and going of visitors or staff. In late-stage dementia, posture more stooped, gait more shuffling, balance problems exist, hesitation to step over obstacles, prefers activities based on a single repeated motion, walks around perimeters of large areas (no initiation of interaction or socialization), may exit unintentionally by fiddling with doorknobs, often carry stuffed animals or pillows while walking, follows staff.
Matteson & Linton (1996) N = 49 Direct observation; behavior logged every 5 minutes within a randomly selected 1-hour period on each of three shifts; 36 discrete observation points per participant. Activity, time of day, O2 saturation, and blood pressure. Overall, participants stood (5%) or paced (3%) little; 50% were observed pacing at some point in time. Pacing was highest during the day shift and lowest at night. O2 saturation correlated positively with standing and pacing; systolic blood pressure was lower in those who paced.b
Algase et al. (1997) 25/25 Direct observation recorded on portable computers using time-study methods; each participant observed continuously for one 24-hour period and two 2-hour periods (1200–1400 and an individually determined period of highest activity) on the third day following. Each ambulation episode coded for start and stop time, pattern (random, lapping, pacing, direct), and lag to next episode; and impetus (self- or other-initiated). Percentages of episodes, episodes per hour, and ratio of walking to nonwalking time by pattern was calculated for each observation period. Over 24 hours, 25 wanderers averaged 19.7 episodes of wandering (17% random, 13.2% lapping, 9.6% pacing); wandered for 42.9 minutes; and averaged 36.1% of wandering cycles in motion. Their direct ambulation averaged 22 episodes over 29 minutes and 28.8% of an ambulation cycle in motion. The 24-hour distribution varied by pattern and hour. The standard period (1400–1600) and individualized periods were moderately correlated to one another for frequency and total duration (standard period only). Either 2-hour period correlated well (approximately 0.70) with the 24-hour total.
Algase, Beattie, & Therrien (2001) 25/23 Direct observation recorded on portable computers using time-study methods; each participant observed continuously for one 24-hour period. As above for Algase et al. (1997) Participants who wandered within a 24-hour period displayed between 1 and 120 episodes (mean = 22.4, SD = 28.3); wandered between 0.7 and 199.1 minutes (mean = 48.7, SD = 54.1). Single episodes lasted between 0.7 and 30.4 minutes (mean = 3.8, SD = 6.1). 88% of wanderers displayed wandering; 52% showed lapping or pacing. Increased frequency of episodes and higher total duration were predicted by higher levels of cognitive impairment.b
Algase, Beattie, Bogue, et al. (2001) N = 151 Direct observation documented by scanning a bar-coded code sheet for two 4-hour periods (0800–1200, 1200–1600, 1600–2000) on each of 2 days; participants were observed for 16 hours each. Each ambulation episode coded for start and stop time, pattern (random, lapping, pacing, direct), and lag to next episode. Data were pooled across observations to yield percentage of episodes, episodes per hour, and ration of walking to nonwalking time by pattern. All observed parameters of random and overall wandering correlated with the overall Algase Wandering Scale (AWS) and four of five subscales with r values between 0.19 and 0.46b; some lapping and pacing parameters also correlated with the overall AWS and several of its subscales.b

Wandering Parameters Based on 12 Observations per Participant

Parameter Method of Calculation % Mean (SD) Range
Percentage of hours wandering Percentage of observations with any wandering 47% (21) 17% to 92%
Rate-Related Parameters
Overall mean rate Total frequency across all observations divided by 12 observations and extrapolated to an hourly rate 4.37/hour (4.43) 0.33 to 19.9
Maximum rate Highest hourly rate 18.39/hour (14.22) 2.92 to 56.83
Number of rate peaks Number of observations in which a high rate (above participant median) was flanked by two lower rates 2.5 (1.07) 1 to 4
Maximum rate peak time Military time with highest peak for rate 1455 (3 hours 19 minutes) 0836 to 1936
Maximum rate peak time period Time period of maximum peak rate
Morning: 8:00 a.m. to 11:59 a.m. 29.55
Afternoon: 12:00 noon to before 5:00 p.m. 50
Early evening: 5:00 p.m. to 8:00 p.m. 20.45
Relative rate Percentage of observations where rate fell into each of three categories
None: 0 wandering 52.65% (20.94) 8.33% to 83.33%
Low/moderate: More than 0 and fewer than 6 times/hour 26.7% (15.83) 0% to 66.67%
High: More than 6 times/hour 20.64% (20.76) 0% to 83.33%
Duration-Related Parameters
Total duration Sum of all minutes wandering across 12 observation periods, extrapolated to an hourly basis 63.29 (74.69) 2.5 to 358.21
Overall mean episode duration Total duration divided by sum of all wandering episodes 0.57 (0.86) 0.02 to 5.41
Mean duration per observation Average number of minutes wandering per observation period 5.27 (6.22) 0.21 to 29.85
Maximum duration Longest duration (in minutes) 19.87 (15.69) 2.28 to 54.9
Number of duration peaks Number of observations in which a long duration (above participant median) was flanked by two shorter durations 2.39 (1.17) 0 to 4
Maximum duration peak time Military time of the highest peak for duration 1443 (3 hours 21 minutes) 0818 to 1948
Maximum duration peak time period Time period of maximum peak duration
Morning: 8:00 a.m. to 11:59 a.m. 31.82
Afternoon: 12:00 noon to before 5:00 p.m. 47.73
Early evening: 5:00 p.m. to 8:00 p.m. 20.45
Relative duration Percentage of observations for which a participant’s duration fell into each of three categories
None: 0 wandering 52.65% (20.94) 8.3% to 83.33%
Low/moderate: More than 0 and fewer than 6 times/hour 35.8% (79.55) 14.44% to 58.33%
High: More than 6 times/hour 11.55% (17.63) 0% to 66.67%
Direct Parameters
Percentage of hours with direct Percentage of observations with any direct walking 37.69% (16.91) 8.33% to 66.67%
Overall mean rate Total frequency of direct walking across all observations divided by 12 observations, extrapolated to an hourly rate 3.48 (3.04) 0.25 to 16.58
Total duration Sum of all minutes of direct walking across 12 observation periods, extrapolated to an hourly rate 12.58 (9.14) 0.42 to 37.83
Overall mean episode duration Total duration of all direct divided by sum of all direct episodes 0.37 (0.25) 0.14 to 1.27
Authors

Dr. Algase is Professor and Faculty Associate, Institute of Gerontology, and Director, Center on Frail and Vulnerable Elders, Dr. Antonakos is Assistant Research Scientist, Statistical Consulting Team, and Dr. Yao is Assistant Research Scientist, University of Michigan School of Nursing, Ann Arbor, Michigan. Dr. Beattie is Professor of Nursing, Dementia Collaborative Research Centre: Consumers, Carers and Social Research, School of Nursing, Faculty of Health, Queensland University of Technology, Queensland, Australia, and Dr. Beel-Bates is Associate Professor of Nursing, Grand Valley State University, Grand Rapids, Michigan.

Address correspondence to Donna L. Algase, PhD, RN, Professor and Faculty Associate, Institute of Gerontology, University of Michigan School of Nursing, 400 North Ingalls Building, Room 2303, Ann Arbor, MI 48109-0482; e-mail: dalgase@umich.edu.

10.3928/19404921-20090101-02

Sign up to receive

Journal E-contents