We report here a secondary study conducted within a randomized experiment studying the efficacy of an assessment and treatment protocol for people with moderate to severe dementia that have a behavior change (Kovach, Noonan, Schlidt, & Wells, 2006
). All nurses participating in this secondary study received enhanced education on behavior change and the assessment and treatment protocol being tested in the larger efficacy study (http://projectreporter.nih.gov/project_info_details.cfm?aid=7743788&icde=4752118
). Nurses were taught that dementia behaviors may be symptoms of unmet needs and were taught about various types of dementia behaviors such as nonspecific vocalizations, motor agitation, passive behavior and body part cues. In addition, nurses were taught that descriptions of verbal symptoms and dementia behaviors may be important clues for physical problems such as pain, infection or exacerbations of chronic conditions. The need for behavior change or verbal symptoms to be followed by physical assessment was emphasized.
Within one week of attending the 8-hour training program, an advanced practice nurse with extensive experience working with residents with dementia, met with the consenting staff nurse. The staff nurse provided a written description of the baseline behaviors of each resident participant on their unit. For example, a resident’s baseline behavior may have been described as “frequent calling out in the morning,” “frequently restless,” or “resistive during baths and transfers.” The expert nurse then shadowed the staff nurse, following him or her as the person performed their nursing roles. This shadowing of the staff nurse occurred on one day for approximately two hours during the midmorning time period after breakfast and before lunch is served. Both the staff and expert nurse independently recorded a description of any behavior demonstrated by a resident enrolled in the study that was perceived to be a change from the baseline information provided by the staff nurse. The staff and expert nurse independently determined whether or not the behavior change specifically indicated a need for further physical assessment of the resident.
Resident participants were from eleven nursing homes in the Midwest and a total of 155 people with dementia participated. Written consent was obtained from the Durable Power of Attorney and verbal assent was obtained from the resident. Residents had an average age of 87.21 (SD = 7.08) and length of stay of 33.06 months (SD = 30.11). The sample had severe cognitive impairment, with an average mini mental status exam score of 6.64 (SD = 5.90) (Folstein, Folstein, & McHugh, 1975
). Most subjects were Caucasian (n = 151) and there were 125 females and 30 male participants. Number of years of formal education ranged from 0 to 22 years (M = 12.0, SD =3.11). Excluded from the study were people with chronic psychiatric illnesses other than the condition identified as causing dementia and those who had been admitted to the nursing home within four weeks.
Thirty eight nurses (26 LPNs, 12 RNs) were asked to participate based on working day shift fulltime on one of the units that housed resident participants. All 38 provided written informed consent. The majority were female (n = 37), Caucasion (n = 28), and over age 45 (n = 24). Only 8% (n = 3) had been nurses for less than five years but 34% (n =13) had been employed in a setting providing care to people with dementia for less than five years.
Coding and Analysis of Shadowing Data
The staff and expert nurse’s descriptions of behaviors were coded by a research assistant into SPSS as present or absent for each participant using the following a priori categories: nonspecific vocalization, verbal symptom, motor agitation, passive behavior and body part cues. Nonspecific vocalizations were defined as sounds that do not clearly communicate a specific message or need and do not appear to be a by-product of events experienced by the individual at that time (e.g. crying, calling out, repetitive verbalizations). Verbal symptoms were verbal communications of a logical and decipherable complaint, request or need (e.g. my back hurts) (Scherder and Bouma 2000
; Closs et al. 2004
). Motor agitation was movement that did not appear to be purposeful or a by-product of events experienced by the individual at that time. (e.g. pacing, resistive, restlessness) (Feldt 2000
; Mahoney et. al, 1999
). Passive behaviors were defined as a decrease in movement, response to human emotions, interaction with people or the environment (e.g. withdrawal) (Colling, 2000
). Body part cues were defined as movements or tension in specific muscles that did not appear connected to current events (e.g. facial grimacing, clenched jaw, rubbing or holding body parts) (Kovach, Cashin & Sauer, 2006
). Interrater reliability of coding was .80 for nonspecific vocalizations and motor agitation, .90 for body part cues and 1.0 for passive behavior, verbal symptoms, and decisions regarding implementation of the STI.
Assessments done by the expert were used to describe prevalence of behaviors and if behaviors co-occurred. Prevalence of each category of behavior was described using frequencies and percents. The association between different types of behavior was tested for each pair of types of behavior by first constructing a two by two contingency table, and then applying Fisher’s Exact test to determine pairwise associations.
The relationship between the staff nurse assessment and the expert advanced practice nurse identifications of behaviors change were described using sensitivity and specificity, with the expert as the gold standard. Sensitivity refers to the proportion of residents identified as having a particular behavior by the staff nurse, among the residents identified as having that behavior by the expert. Specificity refers to the proportion of residents identified as not having a particular behavior by the staff nurse, among the residents identified as not having that behavior by the expert.