Effects of Behavioral Interventions on Disruptive Behavior and Affect in Demented Nursing Home Residents Essay

Behavioral interventions might ameliorate them and have a positive effect on residents’ mood (affect). Objectives: This study tested two interventions—an activities of daily living and a psychosocial activity intervention—and a combination of the two to determine their efficacy in reducing disruptive behaviors and improving affect in nursing home residents with dementia. Methods: The study had three treatment groups (activities of daily living, psychosocial activity, and a combination) and two control groups (placebo and no intervention). Nursing assistants hired specifically for this study enacted the interventions under the direction of a master’s prepared gerontological clinical nurse specialist. Nursing assistants employed at the nursing homes recorded the occurrence of disruptive behaviors. Raters analyzed videotapes filmed during the study to determine the interventions’ influence on affect. Results: Findings indicated significantly more positive affect but not reduced disruptive behaviors in treatment groups compared to control groups. Conclusions: The treatments did not specifically address the factors that may have been triggering disruptive behaviors. Interventions much more precisely designed than those employed in this study require development to quell disruptive behaviors. Nontargeted interventions might increase positive affect. Treatments that produce even a brief improvement in affect indicate improved quality of mental health as mandated by federal law. Key Words: affect †¢ Alzheimer’s disease †¢ behavior therapy †¢ dementia †¢ nursing homes Nursing Research July/August 2002 Vol 51, No 4 proximately 1. 3 million older Americans live in nursing homes today (Magaziner et al. , 2000). By 2030, with the aging of the population, the estimated demand for long-term care is expected to more than double (Feder, Komisar, & Niefeld, 2000). Thus, nursing home expenditures could grow from $69 billion in 2000 to $330 billion in 2030 (Shactman & Altman, 2000). About half of new nursing home r esidents have dementia (Magaziner et al. , 2000). The disease has an impact on four major categories of functioning in persons with dementia. These are disruptive behavior (DB), affect, functional status, and cognition (Cohen-Mansfield, 2000). This article will focus on the first two categories. Disruptive behavior has received much more attention than affect has (Lawton, 1997), perhaps for three reasons. First, more than half (53. 7%) of nursing home residents display DB with aggression (34. 3%) occurring the most often (Jackson, Spector, & Rabins, 1997). Second, DB threatens the wellbeing of the resident and others in the environment. Consequences include: (a) stress experienced by other resiCornelia K. Beck, PhD, RN, is Professor, Colleges of Medicine and Nursing, University of Arkansas for Medical Sciences. Theresa S. Vogelpohl, MNSc, RN, is President, ElderCare Decisions. Joyce H. Rasin, PhD, RN, is Associate Professor, School of Nursing, University of North Carolina. Johannah Topps Uriri, PhD(c), RN, is Clinical Assistant Professor, College of Nursing, University of Arkansas for Medical Sciences. Patricia O’Sullivan, EdD, is Associate Professor, Office of Educational Development, University of Arkansas for Medical Sciences. Robert Walls, PhD, is Professor Emeritus, University of Arkansas for Medical Sciences. Regina Phillips, PhD(c), RN, is Assistant Professor, Nursing Villa Julie College. Beverly Baldwin, PhD, RN, deceased, was Sonya Ziporkin Gershowitz Professor of Gerontological Nursing, University of Maryland. A Note to Readers: This article employs a number of acronyms. Refer to Table 1 to facilitate reading. 219 220 Effects of Behavioral Interventions Nursing Research July/August 2002 Vol 51, No 4 TABLE 1. Acronyms Term Activities of daily living Analysis of variance Apparent affect rating scale Arkansas Combined Disruptive behavior(s) Disruptive behavior scale Licensed practical nurse(s) Maryland Mini mental status exam Negative visual analogue scale Nursing home nursing assistant(s) Observable displays of affect scale Positive visual analogue scale Project nursing assistant(s) Psychosocial activity Research assistant(s) Acronym ADL ANOVA AARS AR CB DB DBS LPN MD MMSE NVAS NHNA ODAS PVAS PNA PSA RA decreases in targeted behaviors (Gerdner, 2000; Matteson, Linton, Cleary, Barnes, & Lichtenstein, 1997). However, others reported nonsignificant reductions (Teri et al. , 2000), no change (Churchill, Safaoui, McCabe, & Baun, 1999), or increased behavioral symptoms (Mather, Nemecek, & Oliver, 1997). These studies used nursing home staffs to collect data, had sample sizes below 100, and measured an array of DB with different assessments. Only in the last decade have researchers investigated affect. Compared to studies to reduce DB, far fewer studies have measured interventions using affect as an outcome measure. Studies reported positive outcomes on affect from such interventions as simulated presence therapy (Camberg et al. , 1999), Montessori-based activities (Orsulic-Jeras, Judge, & Camp, 2000), advanced practice nursing (Ryden et al. , 2000), music (Ragneskog, Brane, Karlsson, & Kihlgren, 1996), rocking chair therapy (Watson, Wells, & Cox, 1998), and pet therapy (Churchill et al. , 1999). The studies on affect used global measures that relied on observer interpretation, which could have compromised objectivity. Theoretical Bases A number of conceptual frameworks have guided intervention research on persons with cognitive impairment (Garand et al. , 2000). The theoretical basis for this study was that individuals have basic psychosocial needs, which, when met, reduce DB (Algase et al. , 1996) (Table 2). The interventions, one focusing on activities of daily living (ADL) and the other focusing on psychosocial activity (PSA), and a combination (CB) of the two, were developed to meet most of the basic psychosocial needs that Boettcher (1983) identified. These included territoriality, privacy and freedom from unwanted physical intrusion; communication, opportunity to talk openly with others; self-esteem, respect from others and freedom from insult or shaming; safety and security, protection from harm; autonomy, control over one’s life; personal identity, access to personal items and identifying material, and cognitive understanding, awareness of surroundings and mental clarity. The section on study groups specifies which interventions were designed to meet which needs. Positive affect usually accompanies interventions that meet basic psychosocial needs (Lawton, Van Haitsma, & Klapper, 1996). Several researchers and clinicians have suggested that displays of affect may offer a window for revealing demented residents’ needs, preferences, aversions (Lawton, 1994), and responses to daily events (Hurley, Volicer, Mahoney, & Volicer, 1993). The study reported here dents and staff; (b) increased falls and injury; (c) economic costs, such as property damage and staff burn-out, absenteeism, and turnover; (d) emotional deprivation such as social isolation of the resident; and (e) use of physical or pharmacologic restraints (Beck, Heithoff, et al. 1997). Third, before the Nursing Home Reform Act (Omnibus Budget Reconciliation Act, 1987), nursing homes routinely applied physical and chemical restraints to control DB with only moderate results (Garand, Buckwalter, & Hall, 2000). However, the Act mandated that residents have the right to be free from restraints imposed for discipline or convenience and not required to treat the residents’ medical symptoms. Thus, research ers have tested a wide range of behavioral interventions to reduce DB and replace restraints. The Act (1987) also stipulated that all residents are entitled to an environment that improves or maintains the quality of mental health. Interventions that promote positive mood or affect fulfill this entitlement. Therefore, this article will report the effects of an intervention to increase functional status in activities of daily living (Beck, Heacock, et al. , 1997), a psychosocial intervention, and a combination of both on reducing DB and improving affect of nursing home residents with dementia. TABLE 2. Basic Psychosocial Needs Relevant Literature Literature suggests that behavioral interventions offer promise in managing DB. A wide range of modalities and approaches have been tested: (a) sensory stimulation (e. g. , music); (b) physical environment changes (e. g. , walled garden); (c) psychosocial measures (e. g. , pet therapy); and (d) multimodal strategies. Many studies found significant Territoriality Communication Self-esteem Safety and security Autonomy Personal identity Cognitive understanding Nursing Research July/August 2002 Vol 51, No 4 Effects of Behavioral Interventions 221 adopted the inference by Lawton et al. (1996) that frequent displays of positive affect when basic psychosocial needs are met might indicate improved emotional wellbeing. is leg continually and without apparent reason needs redirection. This intervention lasted 45–60 minutes a day during various ADL. PSA Intervention. A PNA also conducted the PSA intervention, which involved 25 standardized modules designed to meet the psychosocial needs for communication, selfesteem, safety and security, personal identity, and cogni tive understanding through engagement in meaningful activity while respecting the individual’s unique cognitive and physical abilities (Baldwin, Magsamen, Griggs, & Kent, 1992). The intervention was chosen because it: (a) provided a systematic plan for the PNA to address some of the participant’s basic psychosocial needs; and (b) represented clinical interventions that many long-term care facilities routinely used, but had not been formalized into a research protocol or systematically tested. Each module contained five psychosocial areas of content (expression of feelings, expression of thoughts, memory/recall, recreation, and education) and stimulated five sensory modalities (verbal, visual, auditory, tactile, and gustatory/olfactory). For instance, Activity Module I involved life review, communicating ideas visually (identifying and making drawings), clapping to different rhythms, massaging one’s face, and eating a snack. Initially, many participants tolerated less than 15 minutes of the activity but eventually habituated and participated 30 minutes. CB Intervention. This treatment consisted of both the ADL and PSA interventions and lasted 90 minutes daily. Placebo Control. This involved a one-to-one interaction between the participant and PNA. It controlled for the effect of the personal attention that the PNA provided to the three treatment groups. The PNA asked the participant to choose the activity, such as holding a conversation or manicuring nails. It lasted 30 minutes a day. No Intervention Control. This condition consisted of routine care from a NHNA with no scheduled contact between participants and the PNA. Instruments: Disruptive Behavior Scale. The 45-item disruptive behavior scale (DBS), designed to construct scores based on the occurrence and severity of behaviors, assessed the effect of the interventions on DB (Beck, Heithoff et al. 1997). Gerontological experts (n 29) established content validity, and interrater reliability tests yielded an interclass correlation coefficient of . 80 (p . 001). Geropsychiatricnursing experts weighted the behaviors using a Q-sort to improve the scale’s capacity to predict perceived patient disruptiveness. Factor analysis identified four factors (Beck et al. , 1998). Two corresponded to two—physically aggressive and physically nonaggressive—of the three categories from the factor analysis of the Cohen-Mansfield Agitation Inventory (Cohen-Mansfield, Marx, & Rosenthal, 1989). The third category of the Inventory was verbally agitated; in contrast, the factor analysis of the DBS produced a third and fourth category—vocally agitated and vocally aggressive. To obtain a score for the DBS, a trained individual completed a DBS form for every hour of a shift by check- Methods The primary aim was to conduct a randomized trial of the ADL and PSA interventions individually and in combination (CB) for their effect on DB and affect on a large sample of nursing home residents. The experimental design consisted of three treatment groups (ADL, PSA, and Combined) and two control groups (placebo and no intervention). Individual residents were assigned to one of the five groups at each of seven sites in Arkansas and Maryland, which controlled for site differences. To demonstrate the practicability of the interventions and assure adherence to the treatment protocols, certified nursing assistants were hired and trained as project nursing assistants (PNA). They implemented the interventions Monday–Friday for 12 weeks. Afterward, one-month and two-month follow-up periods occurred. Nursing assistants employed by the nursing homes (NHNA) recorded DB. To measure affect, raters were hired for the study to analyze videotapes filmed during intervention. Research Subjects: The sample initially consisted of 179 participants. The study design allowed for the detection of an improvement in DB scores on the Disruptive Behavior Scale (DBS) (Beck, Heithoff et al. , 1997) across time of at least 1. 6 units with a power of 80%. This power calculation assumed that the repeated measures would be correlated with one another at 0. 60. Inclusion criteria were age 65; a dementia diagnosis; a Mini Mental Status Exam (MMSE) (Folstein, Folstein, & McHugh, 1975) score of 20; and a report of DB in the previous two weeks. To form a more homogeneous group for generalizing findings, exclusion criteria were a physical disability that severely limited ADL; a psychiatric diagnosis; and a progressive or recurring medical, metabolic, or neurological condition that might interfere with cognition or behavior. Study Groups: ADL Intervention. A PNA used the ADL intervention during bathing, grooming, dressing, and the noon meal based on successful protocols that improved functional status in dressing (Beck, Heacock et al. , 1997). It attempted to meet residents’ psychosocial needs for territoriality, communication, autonomy, and self-esteem to promote their sense of safety and security. The intervention also tried to respect participants’ cognitive and physical abilities by prescribing three types of strategies specific to the individual participant. First, strategies to complete an ADL address specific cognitive deficits. For example, the person with ideomotor apraxia needs touch or physical guidance to start movements. Second, standard strategies are behaviors and communication techniques that work for almost everyone with dementia. For example, the caregiver gives a series of one-step commands to guide the resident to put on her shoe. Third, problem-oriented strategies address particular disabilities such as fine motor impairment, physical limitations, or perseveration. For example, a subject who rubs his hand back and forth on 222 Effects of Behavioral Interventions ing the behaviors that occurred. The score for a behavior was the frequency (0–8) times the weight. The item scores were summed to obtain each of the four subscale scores. Mini Mental Status Exam. The Mini Mental Status Exam (MMSE) (Folstein et al. 1975) provided a global evaluation of participants’ cognitive statuses for screening subjects for the study. Test-retest reliability of the MMSE is . 82 or better (Folstein et al. ). Cognition is assessed in seven areas, and scores lower than 24 out of 30 indicate dementia. Nursing Research July/August 2002 Vol 51, No 4 Observable Displays of Affect Scale. The Observable Displays of Affect Scale (ODAS) (Vogelpohl & Beck, 1997), designed to rate videotaped data, contains 41 behaviors categorized into six subscales of positive and negative facial displays, vocalizations, and body movement/posture. Raters indicate presence/absence of each behavior during five 2-minute intervals from a 10minute videotape. Scores range from 0–5 for each item. The range of scores for each scale is: facial positive (0–20), Aggression during bathing facial negative (0–20), vocal positive (0–45), vocal negative (0–50), body could stem from physical positive (0–30), and body negative discomfort or rough (0–40). Interrater reliabilities (Kappa handling coefficients) for the ODAS range from . 68–1. 00, and intrarater reliability is . 97–1. 00. Ten gerontological nursing experts established content validity (Vogelpohl & Beck). Apparent Affect Rating Scale. The Apparent Affect Rating Scale (AARS) (Lawton et al. , 1996) is designed for direct observation of persons with dementia and contains six affective states: pleasure, anger, anxiety/fear, sadness, interest, and contentment. (In later work, Lawton, Van Haitsma, Perkinson, & Ruckdeschel [1999] deleted contentment). Each item has a noninclusive list of behaviors that might signal the presence of the affect from which observers infer the affect. The observer assigns a score of 1 to 5 to measure the duration of the behavior. Visual Analogue Scales. The Positive Visual Analogue Scale (PVAS) and Negative Visual Analogue Scale (NVAS) (Lee & Kieckhefer, 1989; Wewers & Lowe, 1990) are two 10centimeter lines on separate pages for rating positive and negative affect. The PVAS has end anchors of â€Å"no positive affect† and â€Å"a great deal of positive affect. † The NVAS has end anchors of â€Å"no negative affect† and â€Å"a great deal of negative affect. † Scores range from 0 to 100. Procedure: The study consisted of six phases: (a) preliminary activities, (b) a three-week normalization/desensitization period, (c) a 12-week intervention period, (d) a onemonth follow-up period, (e) a two-month follow-up period, and f) a videotape analysis. Preliminary Activities. The institutional review boards at the University of Arkansas for Medical Sciences and the Univer- sity of Maryland approved the research. Each nursing home identified residents with dementia and sent letters informing persons responsible for the residents that researchers would be contacting them. Responsible persons could return a signed form if they did not want to participate. Willing responsible persons received a telephone call explaining the study followed by a mailed written description along with two consent forms. Those willing kept one consent form for their records and signed and mailed back the other. Screening involved a review of the residents’ charts, recording their diagnoses, and interviews with the staff to find evidence of DB during the previous two weeks. Each resident took the MMSE to meet inclusion criteria. Within each home, female residents who passed these screens were randomized to one of the five groups by a drawing, but males were assigned to the five groups to ensure even distribution of their small number. Simultaneously, research staff members were hired and trained. Normalization/Desensitization. For the next three weeks, each PNA accompanied a NHNA to learn the routines of the facility but did not help care for potential study participants. A videotape technician placed a camera that was not running in the dining and shower rooms to desensitize residents and staff to its presence. In addition, nursing home staffs participated in two-hour training sessions on the DBS. Throughout the study, a gerontological clinical nurse specialist trained any new NHNA and retrained if behaviors reported on the DBS differed from those she observed during randomized checks. Intervention. During the 12-week intervention period, the first three weeks were considered baseline and the last two weeks postintervention. The PNA administered the treatment/s or placebo five days a week. Every day, they asked participants to give their assent and espected any dissents. During weeks 11–12 (postintervention), the PNA prepared the participants for their departure by telling them that they were leaving soon. To facilitate data collection, a separate form of the DBS for each of the three eight-hour daily shifts was developed. Eight one-hour blocks accompanied each item of the scale. The NHNA placed a check mark in the block that corresponded to the hour when the NHNA observed the behavi or. The NHNA completed the DBS on all participants during or at the end of a shift. In addition, a technician videotaped participants in the treatment and placebo groups every other week during an interaction with the PNA and no intervention group monthly during an ADL. The technician monitored positioning and operation of the camera from outside the room or behind a curtain to respect the participants’ privacy. One-Month and Two-Month Follow-up. One month and two months after the research team left the nursing home, Nursing Research July/August 2002 Vol 51, No 4 Effects of Behavioral Interventions 223 esearch assistants (RA) retrained nursing home staffs on the DBS. The NHNA then collected DB data on their shifts Monday–Friday for one week. tervention, week 16 as one-month follow-up, and week 20 as two-month follow-up. Participants with fewer than six observations at any time period were omitted. For each period, a total DBS score represented an averVideotape Analysis. The videotapes ranged in length from age of the participant’s data for the t hree shifts of each day less than five minutes to 40 minutes, depending on the across the five days of the observation week. Therefore, activity and the participant’s willingness to cooperate with total DBS scores were obtained for baseline (M of weeks the treatment (baseline and control participants’ tapes 1–3), intervention (M of weeks 4–10), postintervention (M tended to be shorter). To standardize the opportunity for of weeks 11–12), first follow-up (M of week 16), and secbehaviors to occur, an editor took 10-minute segments ond follow-up (M of week 20). The same procedure from the middle of baseline and final treatment eek tapes yielded subscale scores for physically aggressive, physically and randomized them onto videotapes for rating. Because nonaggressive, vocally aggressive, and vocally agitated videotaping occurred to ensure appropriate implementabehaviors for each of the five time periods. tion of interventions, the treatment groups had more A repeated measures analysis of variance (ANOVA) usable videotapes than the control groups did. consisted of two between- subjects and one within-subjects A master’s prepared gerontological factors. The between-subjects factors nurse specialist intensively trained six were intervention group and state (AR raters on the Observer III Software or MD) to account for regional differSystem (Noldus Information Technolences in scoring DB, and the withinogy, 1993) for direct data entry and subjects factor represented DBS scores the affect rating scales. The raters for the five different time periods. Each reached . 80 agreement with the speanalysis allowed for testing by intervencialist on practice tapes before they tion group, time period, and state. The Screaming may started rating the study videotapes. nalysis of the interaction effect of She monitored reliability for each tape intervention group by time period express pain or monthly, retrained as needed, and rantested the hypothesis that the intervenself-stimulation domized the sequence of rating the tions would decrease DB across time in scales. The raters entered the ODAS treatment conditions as compared to and AARS data directly into a comcontrol conditions. The analysis was puter using the Observer. The system repeated five times, once for each suballowed raters to watch videos repeatscale of the DBS and once for the total edly in actual time and slow action to score. Level of significance was set at document behaviors objectively and 0. 05. The researchers believed that the precisely. The raters indicated their small group sizes justified the liberal perception of the participants’ positive and negative level of significance. For the videotape analysis, analyses of affect by placing a vertical mark at some point between covariance occurred for the 14 variables observed from the the two end anchors of the PVAS and NVAS. They videotapes during intervention. The baseline score served marked neutral affect as negative. s a covariate for the final score. While a multivariate analysis would have been desirable, it would have had Intervention Integrity: The PNA and video camera techniinsufficient power with this number of variables and subcian underwent two weeks of intensive training on general jects. The 14 univariate analyses do inflate the Type I error aging topics, stress management, information on dementia, rate. and confidentiality/privacy issues. Training also involved instruction on the study interventions, DBS, and research Results protocols. Of the 179 initial participants, 36 did not finish; the greatA gerontological clinical nurse specialist viewed treatest attrition occurred in the no intervention control group. ment and placebo videotapes biweekly in a private office to Attrition resulted from death (39%), withdrawal of fammonitor PNA compliance with research protocols, provide ily’s consent or at nursing home staff’s request (26%), discorrective feedback to PNA, and help PNA recognize and charge (18%), and change in health status/medications meet participants’ needs as they changed during treatment. hat did not meet inclusion criteria (17%). This left 143 The possibility for contamination appeared to be low participants: 29 in the ADL, 30 in PSA, 30 in CB, 30 in the because NHNA were unlikely to change their care practices placebo, and 24 in the no intervention, but 16 with incomand had little opportunity to observe PNA. Further, NHNA plete data were dropped. Table 3 gives the demographic were b linded to the hypothesis of the study, the nature of the statistics for the 127 participants with complete data. No interventions, and the participants’ group assignments, statistically significant demographic differences emerged although they probably could identify the no intervention among the five groups. In short, this sample primarily conparticipants. sisted of elderly, white females with severe cognitive impairment. Analysis: Reviewers checked for completeness of all data. For the videotape analysis, the final number was 84 The researchers designated intervention weeks 1–3 as baseparticipants with 168 videotape segments. Most were line, weeks 4–10 as intervention, weeks 11–12 as postin- 224 Effects of Behavioral Interventions Nursing Research July/August 2002 Vol 51, No 4 TABLE 3. Description of the Sample by Intervention Group No Intervention 19 89. 5 78. 9 84. 2 86. 47 (6. 37) 11. 47 (6. 43) ADL Number in group Percent female Percent white Percent widowed Mean age (SD) M MMSE (SD) 28 78. 6 82. 1 64. 3 82. 29 (8. 40) 11. 44 (7. 69) PSA 29 82. 1 85. 7 66. 7 82. 18 (7. 64) 10. 65 (6. 76) CB 22 81. 8 77. 3 77. 3 82. 82 (9. 81) 7. 91 (5. 41) Placebo 29 75. 9 86. 2 75. 9 86. 45 (6. 92) 11. 11 (6. 39) Total 127 81. 0 82. 5 72. 8 83. 64 (7. 97) 10. 55 (6. 64) Note. ADL = activities of daily living; PSA = psychosocial activity; CB = combination. emale (79%) and widowed (69%) with a mean age of 83 (SD 7. 44). Participants had a mean score of 10 (SD 6. 34) on the MMSE, indicating moderate to severe cognitive impairment. Table 4 displays the means and standard deviations for the DBS overall and the four subscales across the five time periods for the five groups. No significant differences emerged for the interventi on-by-time interaction for any of the dependent variables. Thus, the results failed to support the hypothesis that the interventions would decrease DB across time in treatment groups as compared to control groups (statistical analysis tables on Website at: http://sonweb. nc. edu/nursing-research-editor). However, the main effect of state was significant in three analyses. Arkansas recorded significantly more behaviors than Maryland did for the dependent variables of physically nonaggressive (p . 001), vocally agitated (p . 001), and overall DBS (p . 002). Further, the main effect of time was significant for overall DBS (p . 002) and the four subscales of physically aggressive (p . 001), physically nonaggressive (p . 027), vocally aggressive (p . 021), and vocally agitated behaviors (p . 008). The significance resulted from increased DB after the PNA had left the home (generally from intervention or postintervention to first follow-up). For the videotape analysis, the hypothesis stated that treatment groups, compared with control groups, would display more indicators of positive affect and fewer indicators of negative affect following behavioral interventions. In general, neither the positive nor the negative affect scores were particularly high, indicating that this sample had relatively flat affect. Results from the analysis of covariance tests supported increased positive affect but not decreased negative affect. Compared to the control groups, the treatment groups had significantly more positive facial expressions (p . 001) and positive body posture/movements (p . 001), but not more positive verbal displays on the ODAS. The treatment groups displayed significantly more contentment (p . 037) and interest (p . 028) than the control groups did on the AARS. For the negative affects on the AARS, the treatment groups had a shorter duration of sad behaviors (p . 007) than the control groups did. Comparison of VAS scales likewise showed that the treatment groups displayed more positive affect (p . 012). Discussion In contrast to other studies (e. . , Hoeffer et al. , 1997; Kim & Buschmann. , 1999; Whall et al. , 1997), this study found no treatment effect on DB. The interventions were a synthesis of approaches believed to globally address â€Å"triggers† of DB and meet psychosocial needs (Boettcher, 1983). They did not address the specific factors that might have been triggering the particula r behavior (Algase et al. , 1996). Such triggers include under/over stimulation, unfamiliar or impersonal caregivers, and particular individual unmet psychosocial needs. For example, aggression during bathing could stem from physical discomfort or rough handling (Whall et al. 1997). Interventions much more individually designed require development. Increasing DB across all groups was reflected in the DBS scores at 1-month follow-up. Two factors may explain this increase. First, the PNA had warned participants that they would be leaving. Second, the ADL and CB participants no longer received care from the familiar PNA, and PSA, CB, and placebo participants no longer had a daily activity or visit. The increased stress and time constraints for NHNA as they resumed caregiving of the ADL and CB participants may explain the heightened DB in the control groups. Such changes may trigger increased behavioral symptoms in persons with dementia (Hall, Gerdner, Zwygart-Stauffacher, & Buckwalter, 1995). Two measurement issues may have affected outcomes. First, observers view behaviors differently (Whall et al. , 1997) and come to expect particular behaviors from certain residents (Hillman, Skoloda, Zander, & Stricker, 1999). If the NHNA were accustomed to a participant’s DB pattern, such as persistent screaming, they may have overlooked decreases in that behavior. Initial training and retraining of raters occurred as needed; however, some Nursing Research July/August 2002 Vol 51, No 4 Effects of Behavioral Interventions 225 TABLE 4. Weighted Scores for Disruptive Behavior by Intervention Group and Time Period No Intervention (n = 19) Mean (SD) 408. 71 (427. 24) 303. 69 (408. 44) 281. 97 (410. 85) 418. 31 (630. 58) 292. 85 (405. 15) 114. 66 (202. 89) 90. 85 (182. 70) 77. 98 (173. 15) 130. 92 (257. 12) 128. 20 (195. 67) 191. 97 (157. 75) 117. 11 (112. 30) 118. 23 (137. 08) 154. 46 (225. 05) 100. 45 (153. 30) 55. 16 (74. 70) 42. 89 (54. 54) 33. 26 (47. 06) 64. 72 (77. 89) 28. 09 (37. 02) (continues) DB Category Time Period DBS total Baseline ADL (n = 28) Mean (SD) 172. 51 (191. 47) 182. 45 (181. 3) 164. 56 (154. 95) 207. 22 (205. 58) 190. 70 (291. 06) 20. 67 (30. 52) 32. 59 (51. 29) 15. 02 (26. 10) 44. 18 (100. 62) 21. 45 (36. 47) 95. 50 (105. 28) 87. 58 (87. 58) 85. 04 (89. 60) 88. 81 (85. 69) 148. 75 (187. 28) 22. 85 (32. 10) 28. 37 (32. 50) 21. 15 (26. 54) 30. 72 (48. 95) 18. 28 (24. 55) PSA (n = 29) Mean (SD) 348. 02 (467. 50) 306. 81 (393. 03) 303. 24 (367. 54) 373. 17 (533. 05) 300. 20 (366. 42) 85. 87 (199. 01) 83. 94 (167. 53) 82. 82 (166. 93) 113. 49 (235. 71) 81. 30 (151. 85) 162. 41 (206. 65) 130. 82 (142. 72) 133. 92 (145. 97) 141. 47 (188. 99) 164. 92 (223. 63) 49. 64 (93. 15) 43. 80 (64. 6) 37. 90 (53. 43) 54. 47 (90. 33) 40. 26 (45. 26) CB (n = 22) Mean (SD) 287. 66 (373. 73) 300. 84 (379. 33) 286. 21 (365. 78) 374. 10 (510. 10) 312. 83 (433. 18) 68. 84 (126. 18) 67. 14 (137. 79) 61. 04 (127. 78) 92. 68 (205. 52) 60. 40 (131. 54) 136. 67 (189. 03) 124. 64 (164. 49) 125. 99 (157. 78) 159. 97 (202. 75) 146. 53 (201. 83) 34. 49 (55. 91) 40. 73 (52. 60) 31. 18 (33. 85) 36. 95 (42. 70) 32. 82 (51. 32) Placebo (n = 29) Mean (SD) 325. 96 (337. 14) 337. 60 (328. 94) 336. 80 (366. 55) 389. 92 (434. 43) 319. 15 (384. 59) 49. 26 (90. 24) 62. 10 (112. 71) 59. 67 (106. 37) 76. 79 (165. 45) 48. 25 (101. 4) 167. 01 (177. 80) 164. 62 (161. 48) 175. 36 (189. 80) 201. 68 (212. 06) 87. 67 (127. 38) 47. 20 (79. 70) 39. 55 (57. 74) 32. 69 (55. 77) 29 . 30 (47. 60) 30. 18 (52. 85) Intervention Postintervention 1 month follow-up 2 month follow-up Physically aggressive Baseline Intervention Postintervention 1 month follow-up 2 month follow-up Physically nonaggressive Baseline Intervention Postintervention 1 month follow-up 2 month follow-up Vocally aggressive Baseline Intervention Postintervention 1 month follow-up 2 month follow-up 226 Effects of Behavioral Interventions Nursing Research July/August 2002 Vol 51, No 4 TABLE 4. Weighted Scores for Disruptive Behavior by Intervention Group and Time Period (Continued) NoIntervention (n = 19) Mean (SD) 47. 65 (97. 22) 68. 32 (103. 13) 68. 01 (116. 62) 84. 50 (112. 48) 73. 07 (117. 12) DB Category Time Period Vocally agitated Baseline ADL (n = 28) Mean (SD) 33. 49 (84. 39) 33. 91 (62. 52) 43. 17 (72. 10) 43. 48 (64. 39) 50. 53 (117. 95) PSA (n = 29) Mean (SD) 46. 92 (98. 70) 52. 84 (96. 03) 52. 50 (90. 78) 68. 22 (98. 89) 48. 89 (92. 33) CB (n = 22) Mean (SD) 62. 49 (98. 97) 70. 43 (110. 85) 69. 08 (107. 29) 82. 14 (118. 97) 75. 80 (129. 67) Placebo (n = 29) Mean (SD) 50. 0 (92. 05) 48. 25 (81. 63) 48. 59 (72. 20) 63. 74 (95. 30) 54. 11 (80. 61) Intervention Postintervention 1 month follow-up 2 month follow-up Note. Scores were created by assigning each behavior with a severity weight prior to summing and then averaging across day and then week(s). DBS = disruptive behaviors; ADL = activities of daily living intervention; PSA = psychocial activity inte rvention; CB = combination of the two interventions. NHNA appeared to continue to consider participants’ behaviors, such as repetitive questioning, to be personality characteristics or attention-seeking efforts rather than DB. Thus, they may have under-reported behaviors. Further, staff may prefer withdrawn behaviors, such as isolating self and muteness (Camberg et al. , 1999), and view them as nonproblematic. Second, categorizing a behavior as disruptive without understanding its meaning to the person with dementia may be conceptually flawed. For example, screaming may express pain or self-stimulation. Two design features may explain differences between the findings of this study and others. First, this study had both placebo and no intervention control conditions. Just a few other studies randomized subjects to treatment or control groups or included two control groups (e. g. , Camberg et al. , 1999). In most studies, control conditions preceded or followed treatment conditions (e. g. , Clark, Lipe, & Bilbrey, 1998). In both designs, subjects served as their own controls, which limits examination of simultaneous intra- and extra-personal events that might affect DB frequency. Second, many control groups came from separate units or different nursing homes (e. g. , Matteson et al. , 1997), which makes it difficult to control for differences in environment, staff relationships, and personalities. This study occurred at seven sites in two different geographical areas, but at each site, the randomization of female participants distributed the groups across all nursing units to control for environmental and staff characteristics. Acknowledged limitations include the following. First, in spite of the large overall sample, the group sizes were small (range 19–30) with the greatest loss in the no inter- vention group. Larger groups might have provided more definitive findings on the relationship between behavioral interventions and DB frequency as Rovner et al. (1996) did (treatment group 42; control group 39). Second, NHNA served as data collectors because using independent observers would have been cost-prohibitive. These results suggest that future intervention research should consider the individual characteristics of the person with dementia (Maslow, 1996) and the triggers of the behavior (Algase et al. , 1996). Studies that have individualized interventions have demonstrated decreased DB (Gerdner, 2000; Hoeffer et al. , 1997). Researchers need to continue to refine methods for identifying what works for whom (Forbes, 1998) to minimize the prevalent trial-anderror approach to DB management.