The Use of Pro Re Nata or Statim Medications for Behavioral Control: A Summary of Experience at a Tertiary Care Children's Mental Health Center

Abstract

The present study aims to identify patterns for use of medication given pro re nata (PRN or “on an as needed [preordered] basis”) or statim (STAT [a new order] or “at once, immediately”) and their efficacy in controlling aggressive behavior in the mental health (MH) services environment. PRN and STAT medication data were combined and referred to as PRN throughout this article, as the data were not collected in a manner required to differentiate between PRN and STAT medication administration. Analyzed data were extracted from the clinical records of a sample of children and youth admitted for the first time to a tertiary MH center. MH Program patients (characterized by at least one Axis I psychiatric diagnosis [Axis I group]) were compared to Dual Diagnosis Program patients (characterized by an Axis I diagnosis in addition to an Axis II diagnosis of mental retardation [Axis II group]). Age, gender, Program (Axis I or II group), and the length of stay for treatment produced significant differences in the use of PRNs between the two groups. Further, the study investigated the precipitating factors leading to use of PRNs, in conjunction with the level of supervision and the de-escalation techniques used to avoid the use of PRNs. Axis I patients were more likely to endanger others, whereas Axis II patients were more likely to endanger themselves. Both groups of patients demonstrated a need for an increased level of supervision prior to the crisis. Olanzapine, chlorpromazine, and lorazepam were effective in calming patients and preventing further aggressive outbursts.

Introduction

I n the mental health (MH) environment, restraints are used to stop harmful behavior to oneself or to others and to prevent serious property damage. When child self-control and other less-restrictive alternatives are ineffective in avoiding individual crises, restraints are applied to the disruptive person. Sedative medication given on a pro re nata (PRN) or “as-needed” basis is one of the common intrusive measures available to staff working at children's MH residential facilities. An empirical study on three psychiatric inpatient units for children revealed that up to 80% of patients receive PRN medication at least once during admission (Kaplan and Busner 1997). Dean et al. (2006) reported a study in which 71% of patients admitted to a child and youth service had a PRN prescribed and that 50.8% actually had received one or more doses of the PRN. Despite the frequent use of psychotropic medications for PRNs, there are few studies examining their effectiveness (Baker et al. 2007).

Controlling aggression is the most common purpose of using restraints. A significant proportion of children's MH center patients are referred because of aggression problems (Hage et al. 2009). The developmental theory linking age and aggression is not easy to grasp. This literature suggests that most humans experience an increase in physical aggression in the very early years of life (1–3 years old), and afterward, it gradually fades away as they grow older and their cognitive capacity takes over the instinctual impulses (Tremblay et al. 1996; Tremblay 2000; Brame et al. 2001). This relationship becomes more complex because among the general population of children and youth there is a small group (between 5% and 20%) that shows persistent high levels of aggression (Nagin and Trembley 1999; Tremblay et al. 1999; Brame et al. 2001). In characterizing this persistently aggressive group, the current research suggests two pathways: early age onset aggression, and adolescence age onset aggression (Brame et al. 2001; Moffit and Caspi 2001). The early childhood onset of aggression is explained by the inherent inability of some children to learn to control physical aggression during childhood, whereas the adolescence onset of aggression is explained by the learning of offensive behaviors during periods of time when adult supervision is inadequate or absent. Therefore, in general, one should expect a negative relationship between age and physical aggression with the addition that there is a small group of highly aggressive children and youth who will maintain or escalate their aggressive behavior over time. The early age aggression is more reactive in nature and thus easier to control, whereas the later age aggression is more planned and more difficult to control (Kempes et al. 2005). This complex picture often makes it difficult to identify the relationship between age and aggressiveness and ultimately to extrapolate it to the use of PRNs. It might simply not be a linear relationship.

The developmental relationship between aggression and psychopathology is also complicated by biological insults and neurological issues. Moffit and Caspi (2001) suggest that a child/youth's risk of following the highly aggressive path in life is determined by inherited or acquired neuropsychological issues. These initial problems are eventually identified as slight cognitive deficits, problematic temperament, or hyperactivity, but with time progression they may be aggravated by environmental factors such as neglect and abuse, weak family ties, adverse social environment, and poverty. Therefore, during their life-course development these highly aggressive children and youth have been affected by neurological issues, which not properly detected, or treated, led to the imminent development of a disordered personality.

More specifically, children and youth having a developmental disability (mental retardation) were more likely to exhibit aggression due to their executive functioning impairments (Valois et al. 2002; Hage et al. 2009). Certain psychopathologies were associated with youth aggression, such as attention-deficit/hyperactivity disorder, oppositional defiant disorder, conduct disorder, mood disorders, trauma and abuse, schizophrenia, and personality disorders (Hage et al. 2009). Dean et al. (2006, 2009) reported fairly similar findings regarding the associations between patients' MH characteristics and utilization of PRN medication.

In addition to these predictors of child and youth aggression, gender has been also the focus of the research seeking to explain aggression. Boys have been identified as more physically aggressive than girls, especially in the preschool period (Keenan and Shaw 1997; Loeber and Hay 1997; Loeber and Stouthammer-Loeber 1998). Moffitt and Caspi (2001, p.357) suggest that “girls have lower rates than boys of symptoms of nervous dysfunction, difficult temperament, late milestones in verbal and motor development, hyperactivity, learning disabilities, reading failure, and childhood conduct problems.” Most likely, girls develop the disruptive aggressive behaviors during adolescence if they have been exposed to antisocial models and perceive rewards for the consequences of their delinquent behavior. Also, girls are more likely to develop aggressive behaviors during adolescence if they have been victims of physical abuse. Females are prone to display less physical aggression, but more covert forms of aggressive behavior such as bullying (Loeber and Stouthamer-Loeber 1998). Boys are suspected of displaying mostly overt (reactive) aggression, whereas girls are for more relational (proactive) aggression (Kemps 2005). Recent reports have suggested that girls engage in more relational aggression such as withholding friendships, gossiping about peers, excluding others from social activities, whereas boys tend to engage in more physical aggression such as pushing, shoving, and hitting (Bowie 2007; Crapanzano et al. 2010). These latter forms of physical aggression are more likely to result in the use of intrusive measures (Stewart et al., 2010).

As emphasized earlier, the risk of aggression varies among groups of patients and PRNs are needed to prevent harm and damage. Serious dangers are associated with the use of restraints in MH treatment facilities. The very unlikely outcome of using restraints in an MH facility is the death of the patient, but such tragic events have changed the present approach of using restraints (Masters et al. 2002). Briefly, current research focuses mainly on reducing the need for PRN medication among inpatients. Many studies highlight particular psychiatric diagnoses that are associated with PRN use, namely developmental disabilities, disruptive behavior disorders, learning disorders (Dean et al. 2007), multiple diagnoses, and pervasive development disorders (Dean et al. 2006). Studies have found that, demographically, PRN use is associated with older (i.e., preadolescent to adolescent age), male children (Baker et al. 2007). PRN use is found to be higher during “transition and unstructured times,” when aggressive behavior is common (Measham 1995). Despite these various analyses, no studies have reported precipitating factors that led to PRN medication use. One study suggested that children's limited capacity to learn new behaviors maybe a more central feature in individuals with the highest PRN use (Dean et al. 2007).

The objectives of this study were to identify patterns of use of PRN medication in a large children's MH center. By examining features associated with the use of PRNs, it might lead to the potential prediction and reduction of their use.

Methods

This study was a small part of a much larger study on the use of intrusive measures, specifically PRN medication, physical restraint (PR), and secure isolation for behavioral control, at a regional children's MH center (Stewart et al. 2008). This facility, directly operated by the Province of Ontario, provides highly specialized, tertiary care, MH services for a 17-county catchment area in southwestern Ontario. Youth admitted to residential treatment are referred to the children's MH center by their local “hard-to-serve committee” only after ensuring that all the local MH services have been exhausted and that the youth needed more specialized services than could be provided in their home community.

This study includes all events associated with medication given to children and youth inpatients on a PRN basis for behavioral control whether it was preordered as part of a treatment plan or given following a new order obtained after the behavioral issue arose.

Approval for the quality assurance study was given by the Administrator at the children's MH center pursuant to Section 182(2) of the Child and Family Services Act RSO 1990 C. 11 (last revised 2008). The Office of Research Ethics on the Use of Human Subjects from the University of Western Ontario approved this study.

All residential patients, ages 6 through 18 years, who were residents between January 1, 2007 and June 30, 2009, were included in this study whether they received PRN medication for behavioral control or not. If a child had more than one admission during the study period, only data from the first admission were included in the study. The children's MH center has a residential section for technology-dependent children. These youth have a severe to profound level of mental retardation, usually requiring feeding tubes (75%), commonly have intractable epilepsy (50%, have a tracheostomy (20%), and are non-ambulatory (100%). The severe physical disabilities of these children preclude them from having disruptive behaviors that would lead to the possible use of a PRN for behavioral control. Therefore, they were excluded from the study.

Starting in March 2006, the children's MH center started requiring resident care staff to complete an electronic report every time an intrusive measure was used for behavioral control. This form required demographic information about the patient, the circumstances leading to the use of the intrusive measure, the actions undertaken to deescalate the situation to prevent the need for the intrusive measure, and any other intrusive measures that might have been used in the same occurrence such as PR or secure isolation. Clinical information regarding the specific medication used, the dose given, the route of administration, the time to settling, and side effects noticed were recorded. Other details of the occurrence were also recorded such as if there was any injury to the patient or staff and the type of PR used if any. The electronic report was reviewed and signed off by the residential manager the next business day for accuracy and completeness as well as to look at possible alternate ways to handle the situation should it arise again. These data were electronically stored in the client database. For this study, the data regarding PRN use were exported from the client database into an Excel spread sheet for coding and were merged with additional information from the client database, including demographic and diagnostic information. The data were electronically imported into and analyzed using SPSS 17.

We conducted parametric group analyses for continuous variables (t-test and one-way analysis of variance) and nonparametric group analyses for discrete variables (χ ²). In addition, a logistic regression and a Cox proportional hazards regression were used to test simultaneous effects of various variables. Most of the reported findings are exploratory in nature and thus the benefits of using a “marginal” significance (0.10 < p < 0.05) has been considered (Ziliak and McCloskey 2008). Odds ratios were described as an exponentiation of the B coefficient [Exp(B)]. This value is given by default because odds ratios can be easier to interpret than the coefficient, which is in log-odds unties. The Wald test is a parametric test used to test the true value of the parameter based on the sample estimate. Cohen's d is used to indicate the standardized difference between two means. The data were structured in multiple events per patient, and consequently, in certain instances, an aggregation was performed. For continuous variables (e.g., time to settle), the aggregation across events was done using the average statistical function, whereas for categorical variables (e.g., reason for administrating the PRN), the aggregation across events was done using the mode statistical function (the most frequent event). For the cases when multiple modes result from the aggregation, we created a variable category named “combination of two or more ….” The variables resulting from this aggregation will be denominated as “the most common … ” (e.g., the most common technique, the most common reason).

Results

The results presented here are based on two relevant samples: the “entire sample” including all children and youth patients that met the earlier announced selection criteria, indifferent if they received or not a PRN during their admission, and the “PRN sample” containing only the youth who received at least one PRN during their admission. For the analysis purpose, the identification of the two samples was done by a dichotomous variable named “PRN status” (received/not received PRNs).

Entire sample

During the study period, 338 youth were qualified for the study according to the criteria described in the prior section. Approximately three quarters of the sample were male youth. The mean age across the sample was 12.3 years (SD = 2.68 years), and half of the sample was children between 10 and 13 years old. The regional children's MH center runs two programs: an Axis I program for patients diagnosed with at least one Axis I disorder, and a Axis II program for patients diagnosed with at least one Axis II disorder (mental retardation) in addition to one or more Axis I diagnoses. From the 338 children and youth included in the study, 254 (75.1%) were assigned to the Axis I program and 84 (24.9%) were assigned to the Axis II program (Table 1).

Table 1.

Demographics

	PRN status
	Entire sample		Received		Not received
Demographic	Frequency	Percent	Frequency	Percent	Frequency	Percent
Gender
Male	258	76.3	132	77.6	126	75
Female	80	23.7	38	22.4	42	25
Total	338	100	170	100	168	100
Age
6–9 years	75	22.2	33	19.4	42	25
10–13 years	165	48.8	85	50	80	47.6
14–18 years	98	29	52	30.6	46	27.4
Total	338	100	170	100	168	100
Program
Axis I	254	75.1	108	63.5	146	86.9
Axis II	84	24.9	62	36.5	22	13.1
Total	338	100	170	100	168	100

PRN = pro re nata.

In the entire sample, a marginally significant association was observed between age and gender (χ ²(2) = 5.42, p = 0.07; Table 2). Male patients were more likely to be younger than female patients, 6–9 years old (24% vs. 16.3%) and 10–13 years old (50% vs. 45%), respectively, whereas female patients were more likely to be 14–18 years old than male patients (38.8% vs. 26%). This was confirmed when using age as a continuous variable: on average, females are older than male patients (M _F = 12.78 years, SD_F = 2.78 years vs. M _M =12.16 years, SD_M = 2.64 years, t(336) = 1.830, p = 0.07; Table 3).

Table 2.

χ ² Test for Demographic × Demographic for the Entire Sample (n = 338)

		Test
Demographic	× Demographic	df	χ²	p
Age	Gender	2	5.42	0.07^a
Gender	Program	1	0.85	0.36
Age	Program	2	5.75	0.06^a
Gender	PRN status	1	0.33	0.57
Age	PRN status	2	1.59	0.45
Program	PRN status	1	24.73	<0.01^b

Marginally statistically significant at p < 0.10.

Statistically significant at p < 0.01.

Table 3.

t-Test Results for Demographic × Demographic for the Entire Sample (n = 338)

					t-Test
Demographic	Mean	SD	Mean	SD	df	t	p	Mean difference	SD difference	d
Age	× Gender
	Female (n = 80)		Male (n = 258)
	12.78	2.78	12.16	2.64	336	1.83	0.07^a	0.63	0.34	0.23
Age	× PRN status
	Received (n = 170)		Not received (n = 168)
	12.58	2.64	12.03	2.70	336	1.88	0.06^a	0.55	0.29	0.21
Length of admission	× PRN status
	Received (n = 170)		Not received (n = 168)
	93.89	66.62	66.2	40.55	336	4.61	0.01^b	27.69	6.01	0.50
Length of admission	× Gender
	Female (n = 80)		Male (n = 258)
	89.05	71.36	77.36	51.39	336	1.61	0.11	11.69	7.26	0.19
Length of admission	×Program
	DD (n = 84)		MH (n = 254)
	105.0	67.25	71.87	50.47	336	4.79	<0.01^c	33.20	6.93	0.56

Marginally statistically significant at p < 0.10.

Statistically significant at p < 0.05.

Statistically significant at p < 0.01.

DD = dual diagnosis; d = Cohen's standardized effect measure.

Among the 338 total patients, no significant difference was observed between male and female patients in the length of admission (M _M = 77.36 days, SD_M = 51.39 days vs. M _F = 89.05 days, SD_F = 71.36 days, t(336) = 1.61, p = 0.11; Table 3). However, the Axis II patients spent on average significantly more days in treatment than the Axis I patients (M _II = 105.07 days, SD_II = 67.25 days vs. M _I = 71.87 days, SD_I = 50.47 days, t(336) = 4.79, p < 0.01; Table 3).

There were no significant differences between the youth who received a PRN and those who did not receive a PRN based on gender (χ ²(1) = 0.33, p = 0.57; Table 2), or based on age (χ ²(2) = 1.59, p = 0.45; Table 2). On average, youth who received a PRN were older than those who did not receive a PRN (M _PRN = 12.58 years, SD_PRN = 2.64 years vs. M _noPRN = 12.03 years, SD_noPRN = 2.70 years, t(336) = 1.88, p = 0.06; Table 3). Youth who received PRNs had a significantly longer period of residential treatment than youth who did not receive PRNs (M _PRN = 93.89 days, SD_PRN = 66.62 days vs. M _noPRN = 66.2 days, SD_noPRN = 40.55 days, t(336) = 4.61, p < 0.01; Table 3). Further, youth in the Axis II program were more likely than youth in the Axis I program to receive PRN medication (73.8% vs. 42.5%) (χ ²(1) = 24.73, p < 0.01; Table 2).

These descriptive results were used for modeling a logistic regression with PRN status as a dependent variable and gender, age, program, and length of admission as independent variables (Table 4). The regression results suggest that female patients have 95% lower odds than male patients in receiving a PRN (Exp.(B) = 0.05, marginally significant at p = 0.07). Also, the Axis II program patients have significantly higher odds than the Axis I program patients to receive a PRN (Exp.(B) = 34.69, p = 0.02). At the same age level, female patients are less likely to receive a PRN (Exp.(B) = 0.81, p = 0.07). These results suggest that gender, age, program, and length of admission are significant predictor candidates for the use of PRNs.

Table 4.

The Logistic Regression Results with Dependent Variable Pro Re Nata Status for the Entire Sample (n = 338)

							95% CI for Exp(B)
	B	SE	Wald	df	f	Exp(B)	Lower	Upper
Constant	−1.16	0.68	2.94	1	0.09	0.31	0.08	1.18
Gender (female)	−2.91	1.59	3.36	1	0.07^a	0.05	0.002	1.23
Age	0.02	0.06	0.09	1	0.76	1.02	0.91	1.14
Program (II)	3.55	1.57	5.10	1	0.02^b	34.70	1.60	753.3
Program (II) × Gender (female)	0.98	0.69	1.99	1	0.16	2.66	0.68	10.36
Gender (female) × Age	0.18	0.12	2.23	1	0.14	1.19	0.95	1.51
Program (II) × Age	−0.21	0.12	3.31	1	0.07^a	0.81	0.65	1.02
Length of admission	0.01	0.003	10.05	1	<0.01^c	1.01	1.00	1.02

Goodness of fit (χ ²(330) = 3.57/df = 330): 1.09; Omnibus test for the model: χ ²(7) = 50.59, p < 0.001.

Marginally statistically significant at p < 0.10.

Statistically significant at p < 0.05.

Statistically significant at p < 0.01.

B = estimated regression coefficient.

PRN sample

Of the 338 youth, 170 (50.3%) received one or more PRNs for behavioral control. Of the youth who received a PRN, 132 (77.6%) were male, 33 (19.4%) were between the ages of 6–9 years, 85 (50%) were between 10 and 13 years, and 52 (30.6%) were 14–18 years (Table 1).

Gender and age

For those who received PRNs, younger male patients (6–9 or 10–13 years old) were more likely to be among those who received PRNs than female patients (21.2% vs. 13.2% and 53% vs. 39.5%), whereas older (14–18 years old) female patients were more likely than male patients to be among the patients who received PRNs (47.4% vs. 25.8%) (χ ²(2) = 6.58, p = 0.04; Table 5). These results were supported when using age as a continuous variable: on average, female patients were older than male patients (M _F = 13.50 years, SD_F = 2.72 years vs. M _M = 12.31 years, SD_M = 2.57 years, t(168) = 2.47, p = 0.01; Table 6).

Table 5.

χ ² Test for Demographic × Demographic for the Received Pro Re Nata Sample (n = 170)

		PRN status received
Demographic	× Demographic	df	χ²	p
Age	Gender	2	6.58	0.04^a
Age	Program	2	5.10	0.08^b
Gender	Program	1	2.51	0.11
Number of doses	Gender	2	3.38	0.19
Number of doses	Age	4	8.71	0.07^b
Number of doses	Program	2	7.50	0.02^a
PRN only	Program	1	3.25	0.07^b
Level of supervision	Program	3	7.37	0.06^b
Prevention method	Program	2	.80	0.67
Reason for administering a PRN	Program	2	10.39	0.01^a
Medication type	Time to settle	3	9.34	0.03^a

Statistically significant at p < 0.05.

Marginally statistically significant at p < 0.10.

Table 6.

t-Test Results for Demographic × Demographic for the Received Pro Re Nata Sample (n = 170)

					t-Test
Demographic	Mean	SD	Mean	SD	df	t	p	Mean difference	SD difference	d
Age	× Gender
	Female (n = 38)		Male (n = 132)
	13.50	2.72	12.31	2.57	168	2.47	0.01^a	1.19	0.48	0.45
Age	× Program
	Axis II (n = 62)		Axis I (n = 108)
	12.78	2.80	12.46	2.55	168	0.765	0.45	0.32	0.42	0.12
Length of	× Gender
admission	female (n = 38)		male (n = 132)
	115.7	86.95	87.60	58.41	168	2.32	0.02^a	28.14	12.11	0.38
Length of	× Program
admission	Axis II (n = 62)		Axis I (n = 108)
	108.6	67.54	85.44	65.89	168	2.21	0.03^a	23.15	10.50	0.35
PRN count	× Gender
	Female (n = 38)		Male (n = 132)
	9.21	9.63	10.02	19.25	168	0.25	0.80	0.81	3.24	0.00
PRN count	× Program
	Axis II (n = 62)		Axis I (n = 108)
	14.01	25.83	7.09	9.21	70	2.21	0.03^a	7.52	2.74	0.36

Statistically significant at p < 0.05.

d = Cohen's standardized effect measure.

Number of doses

Among those receiving PRNs, 76 patients (44.7%) received <4 doses, 57 patients (33.5%) received 4–10 PRN doses, and 37 patients (21.8%) received >10 doses. On average, the patients receiving <4 doses had a shorter admission than those receiving >10 doses (81.35 vs. 112.65 days, p = 0.05; F(2, 167) = 3.01, p = 0.05; Table 7). No significant differences were observed by gender (χ ²(2) = 3.38, p = 0.19; Table 5) across the “number of doses” categories. Marginal differences were observed when cross-tabulation was conducted for the number of doses grouping by the age categories (χ ²(4) = 8.71, p = 0.07; Table 5). Overall, on average there were no differences among age groups in the number of PRNs received (Table 8). Older patients (10–13 and 14–18 years old) were more likely than younger patients (6–9 years old) to get <4 PRNs (49.4% and 44.2% vs. 33.3%) or >10 PRNs (18.8% and 30.8% vs. 15.2%). Table 9. In contrast, younger patients (6–9 years old) were more likely than older patients (10–13 and 14–18 years old) to get 4–10 PRN doses.

Table 7.

Post Hoc Analysis of Variance Results for Length of Admission × Number of Doses for the Received Pro Re Nata Sample (n = 170)

Demographic	Mean	SD	Mean	SD	Mean difference	SD difference	GT2 p₁	G-H p₂
Length of admission	×Number of doses
	<4 (n = 76)		4–10 (n = 57)
	81.36	42.50	98.42	97.45	−17.07	11.54	0.37	0.44
Length of admission	×Number of doses
	4–10 (n = 57)		>10 (n = 37)
	98.42	97.45	112.65	39.65	−14.23	13.90	0.67	0.59
Length of admission	×Number of doses
	<4 (n = 76)		>10 (n = 37)
	81.36	42.50	112.65	39.65	−31.29	13.20	0.06^a	<0.01^b

F test: F(2,169) = 3.01, p = 0.05 (marginally statistically significant at p < 0.10). GT2 refers to Hochberg's GT2 post hoc procedure applied when sample sizes are different, and GH refers to Games-Howell post hoc analysis applied when population variances are unequal.

Statistically significant at p < 0.01.

Marginally significant differences on the length of admission were observed among the age groups (Table 9).

Table 8.

Post Hoc Analysis of Variance Results for Length of Admission × Age for the Received Pro Re Nata Sample (n = 170)

Demographic	Mean	SD	Mean	SD	Mean difference	SD difference	GT2 p₁	G-H p₂
Length of admission	×Age
	6–9 (n = 33)		10–13 (n = 85)
	82.42	41.53	86.92	64.40	−4.49	13.50	0.982	0.896
Length of admission	×Age
	10–13 (n = 85)		14–18 (n = 52)		−25.64	11.59	0.08^a	0.125
	86.92	64.40	112.6	79.19
Length of admission	×Age
	6–9 (n = 33)		14–18 (n = 52)
	82.42	41.53	112.6	79.19	−30.13	14.65	0.118	0.06^a

F test: F(2, 169) = 3.01, p = 0.05 marginally statistically significant at p < 0.10). GT2 refers to Hochberg's GT2 post hoc procedure applied when sample sizes are different, and G-H refers to Games-Howell post hoc analysis applied when population variances are unequal.

Marginally statistically significant at p < 0.10.

Table 9.

Post Hoc Analysis of Variance Results for Pro Re Nata Count × Age for the Received Pro Re Nata Sample (n = 170)

Demographic	Mean	SD	Mean	SD	Mean difference	SD difference	GT2 p₁	G-H p₂
PRN count	×Age
	6–9 (n = 33)		10–13 (n = 85)
	6.67	7.37	8.64	14.32	−1.96	3.57	0.927	0.593
PRN count	×Age
	10–13 (n = 85)		14–18 (n = 52)
	8.64	14.32	13.81	24.95	−7.14	3.88	0.188	0.137
PRN count	×Age
	6–9 (n = 33)		14–18 (n = 52)
	6.67	7.37	13.81	24.95	−5.17	3.79	0.255	0.365

F test: F(2, 169) = 2.10, p = 0.13. GT2 refers to Hochberg's GT2 post hoc procedure applied when sample sizes are different, and G-H refers to Games-Howell post hoc analysis applied when population variances are unequal.

Length of admission

On average, male patients who received PRNs had shorter admissions than female patients who received PRNs (M _M = 87.60 days, SD_M = 58.41 days vs. M _F = 115.74 days, SD_F = 86.95 days, t(168) = 2.32, p = 0.02; Table 6). Marginally significant differences on the length of admission were observed among the age groups (Table 9).

Single versus multiple type of intrusive measures

Of the 170 youth who received a PRN in the study, 44 (25.9%) received PRN only and 126 (74.1%) received PRN plus at least another type of restraint (χ ²(1) = 39.55, p < 0.01; Table 5). A large majority (∼3/4) of patients who received PRNs received multitype intrusive measures during their admission. Further, the patients receiving <4 PRNs were more likely to be receiving PRN only (68.2% vs. 36.5%), whereas those receiving 4–10 PRNs or >10 PRNs were more likely to get multitype of restraints (35.7% vs. 27.3% and 27.8% vs. 4.5%) (χ ²(2) = 16.09, p < 0.01; Table 5).

MH program

Further, among the 170 patients receiving PRNs, 108 (63.5%) youth were Axis I and 62 (36.5%) youth were Axis II. Youth with a developmental disability had a greater average number of PRNs than the youth without a developmental disability (M _{Axis II} = 14.01 PRNs, SD_{Axis II} = 25.83 PRNs vs. M _{Axis I} = 7.09 PRNs, SD_{Axis I} = 9.21 PRNs, t(168) = 2.21, p = 0.03; Table 6).

Among those who received at least one PRN, there is almost no difference between the Axis II and Axis I patients with respect to receiving <4 PRNs during their admission (45.4% vs. 43.5%), but Axis I patients were more likely than Axis II patients to receive 4–10 PRNs (39.9% vs. 24.2%), whereas Axis II patients were more likely than Axis I patients to get more than 10 PRNs during their admission (32.3% vs. 15.7%) (χ ²(2) = 7.50, p = 0.02; Table 5). Further, Axis II patients were more likely than Axis I patients to receive “PRNs only” (33.9% vs. 21.3%), whereas Axis I patients were more likely than Axis II patients to receive “multitype restraints” (78.7% vs. 66.1%) (χ ²(1) = 3.25, p = 0.07; Table 5).

Level of supervision

The majority of children and youth were most commonly on the general level of supervision (66.5%) followed by unobtrusive observation or within staff eyesight at all times (18.8%), constant supervision requires the staff to be within arms length (9.4%), or a combination of various levels of supervision (5.3%) (χ ²(4) =243.94, p < 0.01; Table 5). Youth in the Axis II program were significantly more likely to be under constant supervision with staff at arms length than youth in the Axis I program, 14.5% versus 6.5%, whereas the Axis I youth were more likely to be under unobtrusive supervision than the Axis II youth (24.1% versus 9.7%, χ² (3) = 7.37, p = 0.06; Table 5).

Methods used to prevent using PRNs

The most common technique used to prevent using PRN medication was allowing the child or youth space (73.7%) followed by counseling (9.9%), prompt to calm (6.5%), redirection (4.3%), planned ignoring (2.5%), offering alternative choices (2.3%), and reminder of consequences (0.8%) (χ ²(6) = 651.88, p < 0.01). After collapsing these categories into only three, “allow space,” “counseling,” and “combinations of two or more,” no significant differences were observed between Axis I and Axis II patients regarding the most common technique used to prevent using PRN (χ² (2) = 0.80, p = 0.67; Table 6).

Reasons for administering a PRN medication

The most common reason for administering PRN was a gestural threat of physical aggression (34.7%) followed by a combination of two or more reasons (24.1%). Next on the preset list was “other” (14.1%) followed by leaving the area with risk to self or others (10.6%) and physical aggression (8.8%) (χ ²(10) = 239.33, p < 0.01). When the reason listed in “other” was reviewed, agitation, anxiety, or anger accounted for 11.8% of all the uses of PRN. Collapsing the reasons into fewer categories lead to the following distribution: 108 (63.5%) risk of aggression, 37 (21.8%) combination of two or more reasons, and 25 (14.7%) risk of injury to self and space issues (χ ²(2) = 71.02, p < 0.01). There was no statistically significant gender difference among the reasons for the administration of PRNs (χ ²(2) = 0.13, p = 0.94). However, the Axis I patients were more likely to receive PRNs because of the risk of aggression than the Axis II patients, 70.4% versus 51.6%, whereas the Axis II patients were more likely to receive PRNs because of the risk of self-injury and space issues, 25.8% versus 8.3% (χ ²(2) = 10.39, p < 0.01). A statistically significant difference (F(170, 2) = 3.23, p = 0.04) was observed in the average age of the patients by reason of PRN administration. The average age is decreasing in the following order by the reason for administration: combinations of two reasons, risk of aggression, and risk of self-injury and space issues (M ₁ = 12.96, SD₁ = 2.21; M ₂ = 12.72, SD₂ = 2.79; M ₃ = 11.38, SD₃ = 2.33). This indicates that older children, on average, were more likely to present both risks, or risk of aggression, whereas younger children, on average, were more likely to be associated with the risk of self-injury or space issues.

Choice of medication

The attending child and adolescent psychiatrist or developmental pediatrician (six physicians) chose the medication that was used as a PRN and reflected the individual physician's practice. Three medications were used frequently enough to be able to analyze a low-potency neuroleptic, chlorpromazine (n = 665, 39.8% of all doses administered); a short-acting anxiolytic, lorazepam (n = 348, 20.8% of all doses administered); and an atypical neuroleptic, olanzapine (n = 220, 19.1% of all doses administered). Several other drugs were used but much less frequently and, hence, merged together in category called “other” in the further analyses. These included trazodone 71 doses (4.2%), quetiapine 37 doses (2.2%), dimethydramine 21 doses (1.3%), methotrimeprazine 19 doses (1.1%), loxapine 14 doses (0.8%), clonazepam 13 doses (0.8%), and risperidone 11 doses (0.7%) (Table 10).

Table 10.

Choice of Medication

PRN medication	Frequency	Percentage
Chlorpromazine	665	39.77
Clonazepam	13	0.78
Diphenhydramine	21	1.26
Haloperidol	81	4.84
Lorazepam	348	20.82
Lorazepam + Haloperidol	17	1.02
Loxapine	14	0.84
Methotrimeprazine	19	1.14
Olanzapine	220	19.10
Quetiapine	37	2.21
Risperidone	66	3.95
Trazodone	71	4.25
Total	1,672	100

The side effect profile of chlorpromazine and olanzapine were similar and only included drowsiness in a few youth. This is not totally adverse when the intent of using the medication is to help the child calm him or herself. Lorazepam caused drowsiness in a few youth, but in some youth it led to bizarre behaviors and agitation. Lorazepam is known to disinhibit some people in small doses and this is likely what happened. Because disinhibition is a known adverse effect of lorazepam, it might be less useful as a PRN. The major problems with daily, long-term use of neuroleptics such as metabolic syndrome, hematological problems, and movement disorders, were not seen with the infrequent PRN usage. Common anticholinergic side effects seen with the regular use of chlorpromazine and olanzapine, including dry mouth, constipation, urinary retention, blurred vision, memory impairment, and confusional states, were not seen. Orthostatic hypotension, known to happen with chlorpromazine, was not reported. Other side effects such as Parkinsonian-like effects, dystonias, akesthesias, and tardive dyskinesia were also not seen. Neuroleptic malignant syndrome was not reported.

Antipsychotics can prolong the QTc interval (the length of time between the start of Q wave and the end of T wave of the electrocardiogram corrected for heart rate) and a prolonged QTc interval can lead to torsade de pointes and sudden death. Glassman and Bigger (2001) reviewed the literature and indicated that olanzapine and quetiapine can prolong the QTc interval but neither drug has been reported to produce torsade de pointe or sudden death and that risperidone lengthens the QTc interval but there has been only one reported sudden death, which was not due to torsade de pointe. Reilly et al. (2000) reported no association between prolongation of the QTc interval and chlorpromazine, haloperidol, risperidone, or benzodiazepines. Most of the medications used as a PRN for behavioral control in this study have not been reported to cause a significant prolongation of the QTc interval. If the child had a history of cardiac problems, syncopy, near syncopy, or a family history of sudden cardiac death or medication-induced prolongation of the QTc, an electrocardiogram was obtained before the PRN was prescribed.

A statistically significant association has been observed between medication type and the time to settle (χ ²(9) = 19.97, p = 0.02). Chlorpromazine and lorazepam were more likely to produce settling effects in “more than 30 minutes” than in “30 minutes or less” (41.6% vs. 37.4%, and 22.6% vs. 19.1%), whereas olanzapine was more likely to produce settling effects in “30 minutes or less” than in “more than 30 minutes” (20.8% vs. 18.1%, and 22.7% and 17.6%, Table 11) (χ ²(3) = 9.34, p = 0.03; Table 5). Chlorpromazine, lorazepam, and olanzapine are available as oral tablets, a liquid for injection, and either an oral liquid or a quick-dissolve sublingual tablet, offering a variety of routes for administration.

Table 11.

Cross Tabulation Results of Type of Medication × Time to Settle

	Time to settle
Medication type	30 minutes or less	more than 30 minutes	Total
Chlorpromazine
Count	309	250	559
% within medication type	55.3	44.7	100
% within time to settle	37.4	41.6	39.1
% of total	21.6	17.5	39.1
Lorazepam
Count	158	136	294
% within medication type	53.7	46.3	100
% within time to settle	19.1	22.6	20.6
% of total	11.1	9.5	20.6
Olanzapine
Count	172	109	281
% within medication type	61.2	38.8	100
% within time to settle	20.8	18.1	19.7
% of total	12.0	7.6	19.7
Other
Count	188	106	294
% within medication type	63.9	36.1	100
% within time to settle	22.7	17.6	20.6
% of total	13.2	7.4	20.6
Count	827	601	1,428
Total
% within medication type	57.9	42.1	100
% within time to settle	100	100	100
% of total	57.9	42.1	100

χ ²(3) = 9.34, p = 0.03 (statistically significant at p < 0.05).

Time of the first PRN

Finally, a Cox proportional hazards regression was used to explore the likelihood of receiving the first PRN earlier versus later during the admission. In this model, the dependent variable is “time to the first PRN,” and the independent variables were gender, age, and program.

The results presented in Table 12 suggest that overall female patients were more likely to receive their first PRN later than male patients (lower hazard rate, Exp.(B) = 0.11, p = 0.02). Axis II patients were more likely to receive their first PRN sooner than Axis I patients (Exp.(B) = 4.79, p = 0.05). However, the interaction effects show that on the same age level, female patients had 15% higher odds to receive their first PRN sooner than male patients (Exp.(B) = 1.15, p = 0.04).

Table 12.

The Cox Regression Results for the Time Until the First-Received Pro Re Nata

							95% CI for Exp(B)
	B	SE	Wald	df	P	Exp(B)	Lower	Upper
Gender (female)	−2.20	0.98	5.08	1	0.02^a	0.11	0.02	0.75
Age	0.02	0.04	0.19	1	0.66	1.02	0.94	1.10
Program (Axis II)	1.57	0.79	3.90	1	0.05^a	4.79	1.01	22.68
Program (Axis II) × gender (female)	0.66	0.38	2.95	1	0.09^b	1.93	0.91	4.09
Gender (female) × age	0.14	0.07	4.15	1	0.04^a	1.15	1.00	1.32
Program (Axis II) × age	−0.08	0.06	1.78	1	0.18	0.92	0.82	1.04

Model χ ²(6) = 25.01, p < 0.001 (statistically significant at p < 0.01).

Statistically significant at p < 0.05.

Marginally statistically significant at p < 0.10.

B = estimated regression coefficient.

Discussion

This is a large comprehensive study of PRN medication used for behavioral control at a tertiary care children's MH center. This children's MH center is unique in that although all children and youth have Axis I diagnoses, a significant number also have an Axis II diagnosis of mental retardation. Data were collected in a prospective manner and were contemporaneously checked for accuracy and completeness. These are major strengths of the study.

The fact that our sample is homogeneous with respect to the high level of MH needs in these children and youth could explain some of our marginally significant results regarding the relationships among age or gender, and the use of PRNs (Hage et al. 2009). However, most of our findings confirm the earlier hypothesized theory about aggression and demographics. Younger male patients and older female patients are more likely to receive PRNs. Compared with patients with shorter admission, children and youth staying longer in residence were more likely to receive PRNs.

Several studies support our differential results in the use of PRNs based on the MH treatment program, Axis I versus Axis II. Dean et al. (2006, 2009), Hage et al. (2009), and Valois et al. (2002) reported fairly similar findings regarding the associations between patients' diagnostic characteristics. Axis I youth were significantly more likely to receive PRN medication for aggressive behaviors and Axis II youth were significantly more likely to receive a PRN for risk of self-injury or space issues.

No prior studies describe the level of supervision as related to the use of PRNs. In our study, youth in the Axis I program who received a PRN had been already identified as requiring more supervision than usual by virtue of a significant number who were receiving enhanced supervision requiring the youth to be within the eyesight of staff. Youth in the Axis II program who received a PRN also had been identified as requiring significantly more supervision at the level of being within arms length, compared with children in their program who did not receive a PRN. Looking at the combination of reason for administering the PRN in conjunction with the level of supervision at the time of the PRN and the program, youth in both cases (Axis I and II) were identified as raising the level of staff concern, but the Axis I youth were at more risk for being a danger to others and the Axis II youth were more likely at risk for being in danger to themself. If a youth had a mental retardation (Axis II) and received one dose of a PRN, they were significantly more likely to receive a future dose of PRN medication than a child without a developmental disability. These differences are likely a function of the impairment in learning potential and executive function in children and youth having developmental disabilities. Youth with an Axis II diagnosis do not foresee the possibility of risk to self and are slower to learn new ways to handle situations (Valois et al. 2002; Hage et al. 2009).

No prior studies describe deescalation techniques tried to prevent the use of PRN medication. In a significantly higher frequency of situations, the staff was able to attempt to deescalate the situation through the use of one or more approaches before resorting to a PRN.

This study included a large number of youth who received PRN medication and a wide range of medications were used. There was a significant tendency for youth receiving olanzapine to settle a bit more quickly than with chlorpromazine or lorazepam. Chlorpromazine and olanzapine were used frequently and both were effective. Drowsiness was the most common side effect seen in a few patients. This might not be an “adverse” side effect when the reason for giving the medication is to settle seriously disruptive and aggressive behaviors. A few youth who received lorazepam became disinhibited, resulting in worsening of the disruptive behaviors.

Conclusions

It is possible to collect prospective data on the use of intrusive measures and these data are useful in describing who is at risk for receiving an intrusive intervention. In a tertiary-level children's MH center, the usual pattern of aggression based upon age and gender is no longer present and is likely the result of the level of severity of MH problems in these children and youth.

The presence or absence of the Axis II diagnosis of mental retardation does affect the reasons for using PRN medication, the level of supervision the youth is receiving at the time that the PRN is used, and the total number of times the youth receives a PRN during their admission.

Chlorpromazine and olanzapine were used frequently and appeared to be effective and produced almost no adverse reactions when used as a PRN for behavioral control. When an individual child or youth in treatment received one or more intrusive measures, it comes to note quickly and the written individual plan of care can be reviewed to determine if changes are needed. Making the changes at the level of the individual has been reflected in a reduction in the frequency and pattern of use of intrusive measures (unpublished internal data). Presenting the data regarding the use of intrusive measures at all the monthly plan of care meetings with the child or youth, their guardian, and community clinical treatment partners present makes the residential treatment team more accountable in their use of intrusive measures.

Limitations

There were a number of limitations to this study. First, this is a post hoc data analysis study. Therefore, the results reported do not follow the standards of a randomized controlled trial. Confounding effects might be captured by our analyses and reported findings.

Second, most of the patients are unique in terms of their MH needs. They were admitted to a regional tertiary care MH treatment center because they have exhausted any other available treatment provided by other local or regional agencies. This unique characteristic of our sample limits the degree of generalization of our results to larger and more diverse populations with less-severe MH problems.

Third, aggregating data across multiple events per patient and then across the sample of patients might hide important relationships among the discussed variables. The nature of the data makes statistical analysis difficult. As described earlier, 37 of 170 cases (21.76%) contribute the most (received more than 10 PRNs) to explaining the hypothesized relationships. These issues indicate the need of further analysis based on methods specific to count data and Poisson distribution.

Fourth, it should be also noted that three medications were used frequently enough for comment. A number of other medications were used infrequently and were omitted from further analysis regarding time between administration and effectiveness and adverse side effects. If a larger sample were available, perhaps one or more of these medications may prove to be useful.

Future research

This study is part of a comprehensive management plan targeting reduction of the use of restraints at this facility. Recently, an intensive Therapeutic Crisis Intervention training was deployed across the entire organization. Therefore, the findings reported here would serve as pretraining comparison data. There is the intent to determine the effects of such training on the use of restraints in the organization.

This study is exploratory and descriptive in nature, and consequently, further examination of the relationships among predictor variables should be done using generalized linear models. Such an approach allows the exploration of the potential interaction effects across the variables advanced in the study.

Footnotes

Disclosures

Drs. Swart, Siman, and Stewart have no conflicts of interest or financial ties to report.

References

Baker

, Lovell

, Harris

. A best-evidence synthesis review of the administration of psychotropic PRN medication in in-patient mental health settings. J Clin Nurs, 17:1122–1131. 2007.

Bowie

. Relational aggression, gender, and the developmental process. J Child Adolesc Psychiatric Nurs, 20:107–115. 2007.

Brame

, Nagin

, Tremblay

. Developmental trajectories of physical aggression from school entry to late adolescence. J Child Psychol Psychiatry, 42:503–512. 2001.

Crapanzano

, Frick

, Terranova

. Patterns of physical and relational aggression in a school-based sample of boys and girls. J Abnorm Child Psychol, 38:433–445. 2010.

Dean

, Duke

, George

, Scott

. Behavioural management leads to reduction in aggression in a child and adolescent psychiatric inpatient unit. J Am Acad Child Adolesc Psychiatry, 46:711–720. 2007.

Dean

, McDermott

, Marshall

. PRN sedation—Patterns of prescribing and administration in a child and adolescent mental health inpatient service. Eur Child Adolesc Psychiatry, 15:277–281. 2006.

Dean

, Scott

, McDermott

. Changing utilization of pro re nata (as needed) sedation in a child and adolescent psychiatric inpatient unit. Aust NZ J Psychiatry, 43:360–365. 2009.

Glassman

, Bigger

Jr . Antipsychotic drugs: Prolonged QTc interval, Torsade de Pointes, and sudden death. Am J Psychiatry, 158:1774–1782. 2001.

Hage

, Van Meijel

, Fluttert

, Berden

. Aggressive behavior in adolescent psychiatric settings: what are risk factors, possible interventions and implications for nursing practice? A literature review. J Psychiatr Ment Health Nurs, 16:661–669. 2009.

10.

Kaplan

, Busner

. The use of prn and stat medication in three child psychiatric inpatient settings. Psychopharmacol Bull, 33:161–164. 1997.

11.

Keenan

, Shaw

. Developmental and social influences on young girls' early problem behaviour. Psychol Bull, 121:95–113. 1997.

12.

Kempes

, Matthys

, De Vries

, Van Engeland

. Reactive and proactive aggression in children. A review of theory, findings and the relevance for child and adolescent psychiatry. Eur Child Adolesc Psychiatry, 14:11–19. 2005.

13.

Loeber

, Hay

. Key issues in the development of aggression and violence from childhood to early adulthood. Annu Rev Psychol, 48:371–410. 1997.

14.

Loeber

, Stouthammer-Loeber

. Development of juvenile aggression and violence. Some common misconceptions and controversies. Am Psychol, 53:242–259. 1998.

15.

Masters

, Bellonci

, Bernet

, Arnold

, Beitchman

, Benson

, Bukstein

, Kinlan

, McClellan

, Rue

, Shaw

, Stock

, Kroeger

. Practice parameter for the prevention and management of aggressive behaviour in child and adolescent psychiatric institutions, with special reference to seclusion and restraint. J Am Acad of Child Adolesc Psychiatry, 41:4S–25S. 2002.

16.

Measham

. The acute management of aggressive behaviour in hospitalized children and adolescents. Can J Psychiatry, 40:330–336. 1995.

17.

Moffitt

, Caspi

. Childhood predictors differentiate life-course persistent and adolescence-limited antisocial pathways among males and females. Dev Psychopathol, 13:355–375. 2001.

18.

Nagin

, Trembley

. Trajectories of boys' physical aggression, opposition, and hyperactivity on the path to physically violent and non-violent juvenile delinquency. Child Dev, 70:1181–1196. 1999.

19.

Reilly

, Ayis

, Ferrier

, Thomas

SHL

. QTc-interval abnormalities and psychotropic drug therapy in psychiatric patients. Lancet, 355:1048–1052. 2000.

20.

Stewart

, Cullion

, Wallace-Piccin

, Swart

, Fry

, Okeefe

, Currie

. Tracking intrusive measures during residential treatment: An innovative, clinically relevant approach to assessment, data collection, clinical programming. Poster presented at the 69th Annual Convention of the Canadian Psychological Association, Halifax, NS, June 2008.

21.

Stewart

, Theall-Honey

, Armieri

, Cullion

. Predicting the utilization of intrusive interventions at a tertiary residential treatment centre. Residential Treatment for Children and Youth, 27:175–190. 2010.

22.

Tremblay

, Boulerice

, Harden

, McDuff

, Perusse

, Phil

, Zoccolillo

. Do children in Canada become more aggressive as they approach adolescence? Growing up in Canada: National Longitudinal Survey of Children and Youth. Human Resources Development Canada & Statistics Canada: Ottawa: Statistics Canada, 127–137. 1996.

23.

Tremblay

, Jaspel

, Perusse

, McDuff

, Boivin

, Zoccolillo

, Montplaisir

. The search for the age of “onset” of physical aggression: Rousseeau and Bandura revisited. Crim Behav Ment Health, 9:8–23. 1999.

24.

Tremblay

. The development of aggressive behaviour during childhood: What have we learned in the past century? Int J Behav Dev, 24:129–141. 2000.

25.

Valois

, MacDonald

, Bretous

, Fisher

, Drane

. Risk factors and behaviours associated with adolescent violence and aggression. Am J Health Behav, 26:454–464. 2002.

26.

Ziliak

, McCloskey

. The Cult of Statistical Significance. Ann Arbour, MI: University of Michigan Press, 2008.