Background
Law enforcement is an occupation with specific characteristics that can act as work stressors. There are higher sickness absence rates among police officers compared to other occupations and the two major causes of long term absence are musculoskeletal conditions, and depression and other stress-related disorders. Among police officers, 26% of the medical retirements are due to psychological or mental health disorders. Police medical retirements are a focus of particular concern due to substantially heightened costs for the State (HMIC 1997). In a study of burnout in police managers, 34% were labelled "distressed police managers" because they had the highest level of emotional exhaustion and depersonalization scores in the sample (Loo 2004). In a survey of the Royal Canadian Mounted Police only female officers indicated high levels of burnout on these subscales while, both men and women showed high burnout scores on the "personal accomplishment" subscale (Stearns 1990).
A variety of symptoms and reactions can occur, such as deteriorating work performance (absenteeism, low morale), negative psychological states (emotional burnout, frustration, depression, anxiety, anger), and psychosomatic and physical conditions such as headaches and ulcers (Burke 2000). These can also include a wide range of emotional and behavioral problems and psychosomatic symptoms such substance abuse, adjustment disorders, personality disorder (Saathoff 1990), depression, burnout, frustration, anger, acute stress, anxiety and stress (Burke 2000; Violanti 1985). Although one study has shown that female officers have statistically higher levels of somatization and depression, work-family conflicts (spillover) and destructive coping mechanisms are among the strongest and most consistent stressors in both men and women (He 2002). The estimated rate of probable major depression among 822 Taiwanese police was 21.6% (Chen 2006) and among 873 UK police officers, 41% were deemed to be 'cases' on the 12 item General Health Questionnaire (Collins 2003).
Past and recent studies have shown a high prevalence of excessive alcohol consumption among police officers. A study with a large-scale sample found that a significant number of police officers displayed at-risk alcohol consumption behaviour. By comparison to earlier studies, it has been shown that this proportion is not decreasing (Davey 2000). Another study found that almost half (48%) of male and 40% of females police officers consumed alcohol excessively, including continuous hazardous or harmful consumption and binge drinking (Richmond 1999). Stress has been shown to have a strong effect on alcohol use among the police and alcohol use is used, along with cynicism, as a coping mechanism during different phases of the police stress process (Violanti 1985).
Stressors among police offers can be readily organized into two categories. Those linked to operational factors (job content) and those linked to organizational factors (job context).
Operational stressors
The police service is one of the few occupations where people are expected to face physical dangers and, if necessary, to risk their lives (HMIC 1997). The FBI report published on law enforcement officers feloniously killed, shows that there was no statistical decrease in the five years through the turn of the century. In 1998, 61 officers were killed; in 1999, 42; in 2000, 51; in 2001, 70 and 56 in 2002 (FBI 2002).The Buffalo Cardio-Metabolic Occupational Police Stress (BCOPS) pilot study of 100 police officers identified sub-clinical levels of depression using the Centre for Epidemiologic Studies Depression scale but clinical levels of stress using the Impact of Events Scale (Violanti 1985). Among 157 members of an elite Brazilian police unit employed in critical incidents, the rate of full or partial post-traumatic stress disorder was 8.9% and 15.9%, respectively (Maia 2007).
Other job content stressors include exposure to physical risk such as being a victim of violence, witness to the murder of a companion or having to kill when necessary (Burke 2000). An examination of stressors in Norwegian police officers identified a difference between 'content' stressors such as 'fellow police hurt on duty' as being the more stressful but also being infrequent, while more frequent stressors were 'context' stressors such as 'working overtime' or 'lack of support' (Berg 2005). Similar responses were also given in a survey of 371 Australian police officers: 'witnessing the death of partner', 'participating in an act of corruption' and 'shooting somebody in the line of duty' were the most stressful job content events (Evans 1993).
Organizational stressors
In the Australian study, the most stressful job context events were 'failing police training course', 'failure on promotional examination' and 'unsatisfactory personnel evaluation' (Evans 1993). Other stressors found to be just as important include public contact that may sometimes result in a lack of respect, hostility, many hours of inactivity suddenly shifting to overwhelming responsibility (Burke 2000; Kroes 1974b), pressure for quick crucial decisions, lack of perceived value of their role in society, low income (in most countries), equipment in disrepair, internal organisation structure (Burke 2000) and dealing with the criminal justice system (Burke 2000; Kroes 1974; Kroes 1974b).
Helplessness and feelings of a lack of control in the work environment might also be a major source of stress for police officers. The organisational culture and workload are the key issues in officer stress where the degree of symptomatology worsens each year (Collins 2003). In a study examining the sources of stress-related symptoms among 1206 police officers, the highest ranking occupational stressors were not specific to policing, but to organisational issues such as demands of work impinging upon human life, lack of consultation and communication, lack of control over workload, inadequate support and excess workload in general (Kirkaldy 1995).
Shift work is also a factor, police officers reported significantly worse sleep quality and less average sleep time compared with a control group not involved in the police. Although the life-threatening aspects of police work were related to nightmares, the routine stressors of police service seemed to the most common cause affecting global sleep quality (Neylan 2002).
Another very important aspect of being a police officer is the perhaps unique subculture where, on the one hand, there exists a fraternity-subculture with high social support from peers but, on the other hand, there are also the expectations about behaviours and attitudes of a 'macho culture' (e.g., control, dominance and authority) which include a lack of willingness to seek help in dealing with stress, difficulty in admitting psychological weakness and a constant pressure for police officers to control their emotions and to appear efficient (Burke 2000). However despite this, when support is offered, it is generally accepted (Karlsson 2003).
Because police officers have unique organizational and cultural issues, specific interventions may be necessary for them. Two types of initiative are often cited as the most effective in interventions to alleviate the stressors of police officers. According to Brown and Campbell (Brown 1994), these are the organizational initiatives such as recruitment and selection interventions, counselling interventions, training interventions and work/family initiatives or physical fitness initiatives, and individual initiatives such as the use of self-help, peer-support and active coping responses because of their association with greater job satisfaction (Burke 2000).
Psychosocial interventions include all psychological interventions, social interventions (such as social skills training and befriending) and packages of interventions that have a psychosocial focus even if they also include some more biological interventions (DH 2001). Psychotherapies involving cognitive-behavioural approaches, exposure, exposure plus ritual prevention (ERP), cognitive therapy and stress inoculation have a promising record of experimental evidence of effectiveness in dealing with anxiety and stress disorders (Hunot 2007). Empirically supported psychological interventions for depression are behaviour therapy, brief psychodynamic therapy and cognitive behaviour therapy (CBT) (Chambless 2001). Cognitive behaviour therapy or cognitive therapy focuses on recognizing distorted thinking and learning to replace it with more realistic substitute ideas. Exposure therapy is a form of behaviour therapy in which the person confronts feelings, phobias or anxieties about a traumatic event and relives it in the therapy situation. Stress inoculation is a type of behaviour therapy that tries to cope with stressful situations and consists of three phases: the educational, rehearsal and application phases. Psychodynamic therapy is a term given to approaches used in surfacing true feelings, enabling understanding. The therapy assumes the existence of an unconscious mind which stores over-painful feelings. It works to unravel the natural defences to these feelings, helping to diminish them. Supportive therapy focuses on supporting reality testing, providing ego support, and maintains or re-establishes a usual level of functioning.
Physical activity and fitness have been identified as possible moderators of the stress-illness relationship. Physical activity may play an important role in the management of mild-to-moderate mental health diseases, especially depression and anxiety. In a systematic review on physical activity and mental health, increased aerobic exercise or strength training was found to reduce depressive symptoms significantly. Anxiety symptoms and panic disorder also improve with regular exercise, and beneficial effects appear to be similar to those from meditation or relaxation (Paluska 2000). Sufficient evidence now exists for the effectiveness of exercise in the treatment of clinical depression. Furthermore, exercise has a moderate reducing effect on state and trait anxiety and can improve physical self-perceptions and in some cases global self-esteem. There is also now good evidence that aerobic and resistance exercise enhance mood states, and weaker evidence that exercise can improve cognitive function (primarily assessed by reaction time) in older adults. Together, this body of research suggests that moderate regular exercise should be considered as a viable means of treating depression and anxiety and improving mental well-being in the general public (Fox 1999; Larun 2006). In a meta-analysis with 34 studies, aerobically fit subjects had a reduced psychosocial stress response compared to either their baseline values or the control group (Crews 1987).
In view of the importance of the functions performed by law enforcement officers, and the fact that there is no definitive approach to deal with this problem, a systematic review of the documented evidence regarding the effectiveness of psychosocial interventions for the prevention of psychological problems in this select population is vital.
Objectives
We performed a systematic review assessing the effectiveness of psychosocial interventions for the prevention of psychological disorders in law enforcement officers. This included primary prevention (before disorders occurred) to reduce its incidence, secondary prevention (in a population with known risk factors) and tertiary prevention (after the disorder had occurred).
Methods
Criteria for considering studies for this review
Types of studies
Randomised or quasi randomised controlled trials. Randomised crossover trials were also eligible, but only the first phase of treatment would be included.
Types of participants
Individuals employed directly in law enforcement, including all levels of police officers and military police, regardless of gender, age, and country of origin, and whether or not they have experienced some psychological trauma. Civilians working in police organisations were excluded. Army personnel in combat situations were excluded.
Types of interventions
Psychosocial interventions, as long as they were manualised or described by study's authors, allowing replication. These could include:
- CBT, cognitive and behavioural interventions (e.g. mental imaging training, stress reduction programs, behaviour reduction programs, psycho-education interventions, relaxation).
- Supportive therapies (e.g. counselling sessions, social support).
- Psychodynamic therapies.
- Exercise therapies (e.g. anaerobic and aerobic training, short exercise program, circuit weight training, physical activity, fitness and training).
- Non-pharmacological alternative strategies (e.g. acupuncture, Reiki, diet, meditation).
Psychosocial intervention could be compared to another psychosocial intervention, no treatment (including treatment as usual, attention placebo, psychological placebo or waiting list), pharmacotherapy, or combined pharmacotherapy and psychosocial interventions.
Types of outcome measures
A wide range of instruments were available to measure possible behavioural and psychosocial problems associated with police officers. These instruments varied in quality and validity. For our analyses, the minimum standards for the inclusion of data were that the psychometric properties of the instrument had been described in a peer-reviewed journal.
Outcome measures of interest were:
Primary outcomes
- Ocurrence of psychological disorders
- Psychological symptoms (change in symptoms between baseline and endpoint)
Secondary outcomes
- Reduction in other psychological or behavioural symptoms, such as hostility, flashbacks, intrusive thoughts, mood swings, rage attacks, social isolation, suicide, accidents, alcohol abuse, sleep problems, drug abuse, smoking habits
- Specific outcomes related to the job: absenteeism, job satisfaction and mood, job stress, and burnout
- Quality of life
- Adverse events experienced by the police officer
- Acceptability of intervention as assessed directly by questioning trial participants and indirectly by the drop-out rates
- Physiological or biochemical analysis related to psychological health
Whenever possible, studies were grouped into time periods according to the evaluation of outcomes: short term (less than 6 weeks), medium term (between 6 weeks and 6 months), and long term (over 6 months); and analysed separately.
Search methods for identification of studies
We attempted to identify all relevant published and unpublished trials, irrespective of language or country.
Electronic searches
The following databases were searched to identify randomised or quasi-randomised controlled trials Cochrane Depression, Anxiety and Neurosis Controlled Trials References Register (CCDANCTR-References), Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE, Lilacs and PsycINFO.
CCDANCTR-References was searched on 12/5/2008 using the following terms
Free-text = Police* or "Law Enforce*"
The MEDLINE search strategy was as follows:
1 law enforcement.mp. [mp=title, original title, abstract, name of substance word, subject heading word] (1889)
2 police$.mp. [mp=title, original title, abstract, name of substance word, subject heading word] (5073)
3 law officer$.mp. [mp=title, original title, abstract, name of substance word, subject heading word] (12)
4 1 or 2 or 3 (6661)
5 exp Mental Processes/ (380854)
6 exp Mental Disorders/ (554765)
7 exp PSYCHOLOGICAL TECHNIQUES/ (67607)
8 5 or 6 or 7 (909754)
9 randomized-controlled-trial.pt. (198152)
10 controlled clinical trial.pt. (67773)
11 randomized controlled trials.sh. (36154)
12 random allocation.sh. (52597)
13 double blind method.sh. (80589)
14 single blind method.sh. (8737)
15 clinical trial.pt. (400002)
16 exp Clinical trials/ (162754)
17 (clin$ adj25 trial$).ti,ab. (107224)
18 ((singl$ or doubl$ or tripl$ or trebl$) adj25 (blind$ or mask$ or dummy$)).mp. [mp=title, original title, abstract, name of substance word, subject heading word] (109275)
19 placebos.sh. (23474)
20 placebo$.ti,ab. (87368)
21 random$.ti,ab. (302881)
22 research design.sh. (39968)
23 comparative study.sh. (1174165)
24 exp evaluation studies/ (512022)
25 follow up studies.sh. (296255)
26 prospective studies.sh. (184088)
27 (control$ or prospectiv$ or volunteer$).ti,ab. (1506746)
28 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25 or 26 or 27 (3249416)
29 (animals not humans).sh. (2847046)
30 28 not 29 (2516742)
31 4 and 8 and 30 (572)
Descriptors and synonyms for police, as well as descriptors and synonyms for psychosocial interventions were used and modified according to specific requirements of each electronic database.
The EMBASE, LILACS, PsycINFO and CENTRAL searches are shown in Table 1.
Searching other resources
1. Handsearching: We handsearched the Journal of Police Science and Administration (volumes 1 to 17: 1973 to 1990).
2. Reference lists of review articles and included studies, and books related to police officers were checked.
3. Personal Communication: The first author of reports of controlled studies and reviews on the subject was approached, as were experts in the field, for information on trials in progress or currently unpublished studies.
Data collection and analysis
Selection of studies
Two authors (VP and JRL) checked the titles and abstracts of the articles identified by the search. The full text for each potentially relevant study was obtained and independently selected for critical appraisal.
Data extraction and management
Data on patients, methods, interventions used, randomisation process, outcomes, and results were extracted by the first author (VP) using a specially designed form, and checked by a co-author (HMG). Possible disagreements were resolved through discussion. Missing data were obtained from trialists, whenever possible.
Assessment of risk of bias in included studies
In order to ensure that variation is not due to systematic errors in the design of a study, two authors (VP and JRL) assessed the methodological quality of the selected trials independent of each other. Quality was assessed using the criteria described in the Cochrane Reviewers Handbook at the time that much of this work was done (Alderson 2004). These are based on evidence for a strong relationship between the potential for bias in the results and allocation concealment in the trial, as defined below:
Selection bias
A. Adequate allocation concealment
B. Mentioned as randomised, but details of randomisation unreported and impossible to be acquired from the author of primary study
C. Randomised by a process where the allocation could be predictable in some way (birth date, arriving order, etc)
Attrition bias/Detection bias
Three other methodological criteria were considered according to three conditions, MET, UNCLEAR and NOT MET:
- Were the data assessors blinded?
- Was there no systematic difference between the comparison groups regarding the withdrawals and was it less than 15%?
The following answers were used for the above methodological questions:
Met: criteria appropriately applied and either described or obtained from the author of the study
Unclear: criteria not described and impossible to be obtained from the author of the study
Not met: criteria inappropriately applied
Included studies were classified as
(a) low risk of bias: all criteria met
(b) moderate risk of bias: one or more unclear issues
(c) high risk of bias: one or more unmet issues
Performance bias was considered according to the criterion mentioned above, for those study designs in which the blinding of care providers could be applied.
Assessment of heterogeneity
Heterogeneity in the results of the trials was assessed both by inspection of graphical presentations and by observing the I-square test of heterogeneity (Higgins 2003). For quantifying inconsistency, the statistic I²=[(Q - df)/Q] x 100%, where Q is the chi-squared statistic and df is its degrees of freedom would be considered. A value greater than 50% in this test may be considered indicative of substantial heterogeneity. Possible reasons for heterogeneity were expected and pre-specified as follows:
- That treatment effects differ according to the different kinds of intervention strategies
- That response differs when psychosocial therapies are provided in conjunction with prescribed drugs
- According to the characteristics of participants in the trials (administrative police, mounted police, vice squad, drug squad, riot police, beat policeman, military police, female police, etc
- According to length of follow-up.
These sources of heterogeneity would be assessed by looking at separate subgroups of trials.
Since considerable clinical and methodological diversity was anticipated across studies, we planned to use the random effect model for any meta-analyses.
Assessment of reporting biases
The possibility of publication bias could not be examined using a funnel plot because there were only five included studies with data.
Data synthesis
For dichotomous outcomes, the measure of treatment effect was the relative risk (RR) with a 95% confidence interval. The relative risk from individual trials was to be combined when possible through meta-analysis. If the overall estimate was significant, the number needed to treat (NNT) to produce one outcome would be calculated.
For continuous data, we applied the following standards: (a) standard deviations and means had to be reported or obtained from authors; and (b) for data with finite limits, such as endpoint scale data, the standard deviation, when multiplied by two, had to be less than the mean; otherwise the mean is unlikely to be an appropriate measure of the centre of the distribution (Altman 1996). Only non-skewed data was analysed. For any meta-analyses, weighted mean difference would be calculated if the same instrument had been used for the outcome measure. Standardised mean difference would be used if different instruments were used for the same outcome.
It was not possible to obtain all missing data, so we did not carry out an ITT analysis and the data was based on completers only (Unnebrink 2001).
Subgroup analysis and investigation of heterogeneity
Sub-group analysis were planned on different kinds of interventional strategies:
1. different categories of psychosocial intervention(psychotherapy, counselling, brief interventions, coping strategies,psycho-education interventions,fitness, training, physical activity, etc)
2. individual versus group therapies (familial or with workmates)
3. psychosocial versus non-pharmacological intervention (acupuncture, Reiki, religion, diet,meditation,etc)
Sensitivity analysis
Sensitivity analyses were planned to estimate the robustness of our results according to the judgements about the methodological quality of the included studies.
Results
Description of studies
See: Characteristics of included studies; Characteristics of excluded studies.
The electronic searches identified a total of 25 studies and handsearches identified three more. Of these, 10 were included, 16 excluded and two are awaiting classification as the hard copies have not yet been obtained.
Studies were excluded for the following reasons: 14 were not randomised controlled trials or quasi randomised controlled trials (Arokoski 2002; Bar 2004; Carlier 1998; Carlier 2000; Danish 1970; Godschalx 1984; Harrel 1993; Meadows 1985; Norris 1990; Pinfold 2003; Richmond 1999b; Schmit 1996; Tolin1999; Wallace 1977) and the aims of the intervention in the final study were outside the scope of current review (Gersons 2000; McMurray 1998).
Of the included studies, most were described as 'randomised', although there is uncertainty about the randomisation method used in one (Norvell 1993) so it was classified as 'quasi-randomised' (CCT). McNulty 1984 was described its randomisation method as 'equally spread' so it was classified as a quasi-randomised controlled trial (CCT). O'Neill 1982 was described as a controlled clinical trial (CCT).
Patient selection
The majority of these trials involved populations that were comparable. All patients were serving law enforcement officers, although there were three trials with police trainees (Backman 1997; Sarason 1979; Shipley 2002). Five trials included both men and women (Backman 1997; Doctor 1994; Sarason 1979; Shipley 2002; Wilson 2001). In the majority of the trials, clinical diagnosis was not used to include to exclude participants (Aremu 2006; Backman 1997; Doctor 1994; McNulty 1984; Norvell 1993; O'Neill 1982; Sarason 1979; Shipley 2002; Wilson 2001). In one study, the diagnosis for inclusion was to be overweight by 20% (Short 1984). No characteristics of patient selection were mentioned in one trial (O'Neill 1982).
Cultural Setting
Five studies were undertaken in United States (O'Neill 1982; Norvell 1993; Sarason 1979; Short 1984; Wilson 2001). The others were carried out in Sweden (Backman 1997), Australia (McNulty 1984), Canada (Shipley 2002), Nigeria (Aremu 2006) and the United Kingdom (Doctor 1994).
Sample size
The number of patients randomised in the trials ranged from 18 (Sarason 1979) to 90 (Aremu 2006) with a total of 583 individuals in all ten trials combined.
Outcomes measures
In nine trials, psychological measures were assessed in the form of stress, anxiety, anger, physical symptoms, self concept, job stress, hostility, burnout, cynicism and marital relationship (Aremu 2006; Backman 1997; Doctor 1994; Norvell 1993; O'Neill 1982; Sarason 1979; Shipley 2002; Short 1984; Wilson 2001). In two trials, physiological measures were assessed (Backman 1997; McNulty 1984).
Interventions
The psychosocial interventions that were evaluated are outlined here and discussed in more detail below.
Stress management program versus psycho-educational interventions (McNulty 1984; O'Neill 1982; Sarason 1979; Wilson 2001).
Mental imaging training (at home) (Backman 1997).
Counseling group sessions (Doctor 1994).
Circuit weight training (Norvell 1993).
Visuo-Motor Behavior Rehearsal (Shipley 2002).
Social Skills Training versus Problem-solving skills training versus control (Aremu 2006).
Aerobic program (including theory) versus theory on exercise program (Short 1984).
Risk of bias in included studies
The full text of all potentially relevant studies was obtained and critically appraised by two authors (VP and HMG).
Allocation
Two authors (VP and HMG) assessed the methodological quality of the included studies using the criteria described in the Cochrane Reviewers' Handbook (Alderson 2004). We wrote to the authors of the ten studies requesting more details on the randomisation process that they used. We received responses from three (Backman 1997; Doctor 1994; Wilson 2001). Only one study had adequate allocation concealment and this was scored A (Doctor 1994). The method of randomisation was unclear or not reported in eight trials which were scored B (Aremu 2006; McNulty 1984; Norvell 1993; O'Neill 1982; Sarason 1979; Shipley 2002; Short 1984; Wilson 2001).The method of randomisation was inadequate, scoring C, in one study (Backman 1997).
Blinding
In three trials (Backman 1997; Doctor 1994; Wilson 2001) blinding was appropriately applied and the criteria of 'assessors concealed' was met. In six trials (McNulty 1984; Norvell 1993; O'Neill 1982; Sarason 1979; Shipley 2002; Short 1984) the necessary information was not described and so these were considered to be 'unclear'.
Incomplete outcome data
In three studies (Backman 1997; Doctor 1994; Wilson 2001) the criteria were met. It was considered 'not met' only for the body symptom scale assessment in Backman 1997. In six studies (Aremu 2006; McNulty 1984; O'Neill 1982; Sarason 1979; Shipley 2002; Short 1984) the criteria was not described so was considered 'unclear'. One study (Norvell 1993) was classified as not met.
Overall methodological quality
In eight trials, data were not available for all methodological variables, none of the authors could provide the data and the necessary information was not included in the trial report (Aremu 2006; McNulty 1984; Norvell 1993; O'Neill 1982; Sarason 1979; Shipley 2002; Short 1984; Wilson 2001). The necessary information regarding assessor concealment and attrition bias was provided for one study (Wilson 2001). Information on all the methodological data was provided for Backman 1997. The necessary information was obtained in part from the trial report and also from communication from the author for the Doctor 1994 trial.
Overall, the criteria for good methodological quality of the included studies were partly met. The Doctor 1994 trial was the only one to meet the criteria of low risk of bias. Wilson 2001 came closest of the other trials to fulfilling the criteria of partly met. Aremu 2006; Backman 1997; McNulty 1984; Norvell 1993; O'Neill 1982; Sarason 1979; Shipley 2002; and Short 1984 had unclear issues, which was partly due to incomplete reporting. We are unsure of the randomisation procedure in Norvell 1993 and O'Neill 1982. Follow-up was of 3 months (Wilson 2001) and 18 months (Backman 1997).
Included studies were classified as
(a) low risk of bias: all criteria met: Doctor 1994.
(b) moderate risk of bias: one or more criteria partly met and unclear issues: Aremu 2006; McNulty 1984; O'Neill 1982; Sarason 1979; Shipley 2002; Short 1984; and Wilson 2001.
(c) high risk of bias: one or more unmet issues; Backman 1997 and Norvell 1993.
Effects of interventions
Quantitative meta-analyses were not possible.
Primary Prevention - Intervention versus Control
Only one trial reported data for primary intervention (intervention versus control) (Backman 1997). Ten outcomes were measured. There was significant difference in favour of the intervention group on the depression, sleep, cynicism and symptoms scale measures. There were significant differences in favour of the control group in general health and coping; and no differences between intervention and control groups in bodily symptoms scale, Jenskins activity, burnout and vital exhaustion.
Primary outcomes
Depression: One study (Backman 1997) comparing Mental Imaging Training versus control, with a sample size of 60 participants at endpoint (N = 60) and at follow-up (N = 60) contributed to this outcome. There was a significant difference in favour of the intervention at the endpoint (MD (fixed) -2.14, 95%CI -4.00 to -0.28) but at 18 month follow-up there was no statistical difference between the two groups (MD (fixed) -0.97, 95%CI -2.43 to 0.49).
Secondary Oucomes
General Health Questionnaire: One study (Backman 1997) comparing Mental Imaging Training versus control, with a sample size of 63 participants at endpoint (N = 63) and at follow-up (N = 63) contributed to this outcome. There was a significant difference in favour of the control group at the endpoint (MD (fixed) 2.74, 95% CI 0.78 to 4.70) but at the 18 month follow-up the difference was not statistically significant (MD (fixed) 1.30, 95% CI -0.61 to 3.21).
IBM Coping: One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 65 participants at endpoint (N = 65) and at follow-up (N = 65). There was a statistical difference in favour of the control group at the endpoint (MD (fixed) 2.83, 95% CI 1.51 to 4.15) and again at the 18 month follow-up (MD (fixed) 2.24, 95%CI 1.06 to 3.42).
Jenkin's Activity Survey: One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 63 participants at the endpoint (N = 63) and at follow-up (N = 63). There was no statistical difference between the groups at the endpoint (MD (fixed) 1.1, 95%CI -2.83 to 0.61) but by the 18 month follow-up there was a statistical difference in favour of the intervention group (MD (fixed) 2.77, 95% CI -4.66 to -0.88).
Karolinska Sleep Questionnaire (Impaired Awakening:subscale): One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 66 participants at endpoint (N = 66) and at follow-up (N = 66).There was a statistical difference in favour of the intervention group at the endpoint (MD (fixed) -1.58, 95%CI -2.46 to -0.70) and at the 18 month follow-up (MD (fixed) -1.06, 95%CI -1.84 to -0.28).
Karolinska Sleep Questionnaire (Sleep Quality subscale): One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 64 participants at endpoint (N = 64) and at follow-up (N = 64). There was a statistical difference in favour of the intervention group at the endpoint (MD (fixed) -1.97, 95% CI -3.55 to -0.39) and at the 18 month follow-up (MD (fixed) -1.68, 95% CI -3.28 to -0.08).
Karolinska Sleep Questionnaire (Sleepiness): One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 63 participants at endpoint (N = 63) and at follow-up (N = 63). There was a statistical difference in favour of the intervention group at the endpoint (MD (fixed) -1.82, 95% CI -3.05 to -0.59) and at the 18 month follow-up (MD (fixed) -1.19, 95% CI -2.34 to -0.04).
Bodily Symptom Scale - Worry Depression: One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 60 participants at endpoint (N = 60) and at follow-up (N = 60). There was no statistical difference between the groups at the endpoint (MD (fixed) -2.14, 95% CI -4.00 to 0.28) or at the 18 month follow-up (MD (fixed) -0.97, 95% CI -2.43 to 0.49).
Burnout (Emotional subscale): One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 67 participants at follow-up. There was no statistical difference between the groups (MD (fixed) 0.02, 95% CI -1.78 to 1.82).
Burnout (Despersonalization subscale): One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 67 participants at follow-up. There was no statistical difference between the groups (MD (fixed) 0.16, 95% CI -1.00 to 0.68).
Burnout (Personal Accomplishment subscale): One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 67 participants at follow-up. There was no statistical difference between the groups (MD (fixed) 0.58, 95% CI -0.85 to 2.01).
Vital Exhaustion: One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 67 participants at endpoint (N = 67) and at follow-up (N = 67). There was no statistical difference between the groups at the endpoint (MD (fixed) -2..49, 95% CI -5.57 to 0.59) but there was a statistical difference in favour of the intervention group at the 18 month follow-up (MD (fixed) -3.94, 95% CI --6.95 to 0.93).
Cynicism: One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 64 participants at endpoint (N = 64) and at follow-up (N = 64). There was a statistical difference in favour of the intervention group at the endpoint (MD (fixed) -6.99, 95% CI -12.13 to -1.85) and at the 18 month follow-up (MD (fixed) -8.22, 95% CI -14.11 to -2.33).
Symptom Scale: One study (Backman 1997) comparing Mental Imaging Training versus control contributed to this outcome, with a sample size of 60 participants at endpoint (N = 60) and at follow-up(N = 60). There was a statistical difference in favour of the intervention group at the endpoint (MD (fixed) -2.14, 95% CI -4.00 to -0.28) but not at 18 month follow-up (MD (fixed) -0.97, 95% CI -2.43 to 0.49).
Dropouts: One study (Backman 1997) contributed to this outcome. Drop-out was more common in the intervention group (2/37) than in control group (1/38). This difference is not statistically significant (RR 2.05, 95%Cl 0.19 to 21.70).
Adverse effects: No data were reported for this outcomes.
Primary Prevention - Intervention versus Intervention
No studies were identified for this comparison.
Secondary Prevention - Intervention versus Control
There was only one study of secondary prevention (intervention versus control) (Norvell 1993) that reported data and two studies that contributed to the dropouts analysis (Doctor 1994; Norvell 1993). Eight outcomes were measured. There were significant differences in favour of the intervention group in depression, anxiety, general subscale, physical scale symptoms scale. There were no statistically significant differences between the intervention and control groups in hostility, somatization, job descriptive scale and perceived stress scale.
Primary outcomes
SCL-90
Hostility subscale: One study (Norvell 1993) contributed to this outcome (N = 29). There was no statistical difference between the groups at the endpoint (MD (fixed) -3.84, 95% CI -8.48 to 0.80) .
Depression subscale: One study (Norvell 1993) contributed to this outcome (N = 29). There was a significant difference in favour of the intervention group at the endpoint (MD (fixed) -7.32, 95% CI -11.79 to -2.85).
Anxiety subscale: One study (Norvell 1993) contributed to this outcome (N = 29). There was a significant difference in favour of the intervention group at the endpoint (MD (fixed) -6.80, 95% CI -9.22 to -4.38).
Perceived Stress Scale: One study (Norvell 1993) contributed to this outcome (N = 29). There was no statistical difference between the groups at the endpoint (MD (fixed) -3.1, 95%CI -6.94 to 0.60)
Secondary Outcomes
General subscale: One study (Norvell 1993) contributed to this outcome (N = 29). There was a statistical difference in favour of the intervention at the endpoint (MD (fixed) -6.22, 95% CI -10.78 to -1.66).
Somatization subscale: One study (Norvell 1993) contributed to this outcome (N = 29) There was no significant difference between the groups at the endpoint (MD (fixed) -1.84, 95%CI -5.68 to 2.00).
Physical Symptom Scale: One study (Norvell 1993) contributed to this outcome (N = 29). There was a significant difference in favour of the intervention group at the endpoint (MD (fixed) -5.23, 95% CI -8.02 to -2.44).
Job Descriptive Scale: One study (Norvell 1993) contributed to this outcome (N = 29). There was no significant difference between the groups at the endpoint (MD (fixed) -8.82, 95% CI -21.73 to 4.09).
Perceived Stress Scale: One study (Norvell 1993) contributed to this outcome (N = 29). There was no significant difference between the groups at the endpoint (MD (fixed) -3.17, 95%CI -6.94 to 0.60).
Dropouts: Two studies contributed to this outcome (Doctor 1994; Norvell 1993). Drop-out was more common in the intervention groups (23/64) than in the control groups (7/44), with a significant difference in favour of the control group (RR 2.27, 95% CI 1.01 to 5.10).
Adverse effects: No data were reported for this outcome.
Secondary Prevention - Intervention versus Intervention
In secondary prevention (intervention versus intervention) only one study (Wilson 2001) reported data on the primary outcomes and two studies contributed to the analysis of dropouts (Short 1984; Wilson 2001). Seven outcomes were measured. There was a significant difference in favour of the intervention (eye movement desensitization and reprocessing (EMDR)) at endpoint in Anger(Trait), PTSD, marital relationship and Distress Scale. There was no significant difference between the groups in Anger (State), Tennesse Self Concepto Scale and Job Stress Scale.
Primary Outcomes
Anger - Trait: One study (Wilson 2001) contributed to this outcome at the endpoint (N = 62) and at follow-up (N = 62). There was a significant difference in favour of EMDR over Brief Eclectic Psychotherapy (BEP) at the endpoint (MD (fixed) -1.50, 95% CI -2.71 to -0.29) and at the six month follow-up (MD (fixed) -3.46, 95% CI -4.90 to -2.02)
Anger - State: One study (Wilson 2001) contributed to this outcome at the endpoint (N = 62) and at follow-up (N = 62). There was no significant difference at the endpoint (MD (fixed) 0.25, 95% CI -0.78 to 1.28) but there was a significant difference in favour of EMDR over BEP at six month follow-up (MD (fixed) -1.28, 95% CI -2.40 to -0.16)
Diagnosis of PTSD: One study (Wilson 2001) contributed to this outcome at the endpoint (N = 62) and at follow-up (N = 61). There was a significant difference in favour of EMDR over BEP at endpoint (OR (fixed) 0.42, 95% CI 0.04 to 4.91) and again at six month follow up (OR (fixed) 0.28, 95%CI 0.03 to 2.985).
Subjective Units of Distress Scale: One study (Wilson 2001) contributed to this outcome at the endpoint (N = 62). There was a significant difference in favour of EMDR over BEP at the endpoint (MD (fixed) -2.28, 95% CI -2.57 to -1.99)
Secondary Outcomes
Tennessee Self Concept Scale: One study (Short 1984) contributed to this outcome at the endpoint (N = 45). There was no significant difference at the endpoint (MD (fixed) 3.70, 95% CI -14.86 to 22.26)
Job Stress Scale: One study (Wilson 2001) contributed to this outcome at the endpoint (N = 62). There was no significant difference between the groups at endpoint (MD (fixed) -0.66, 95% CI -3.43 to 2.11)
Marital Relationship: One study (Wilson 2001) contributed to this outcome at the endpoint (N = 62). There was a significant difference in favour of EMDR over BEP at the endpoint (MD (fixed) 22.70, 95% CI 9.67 to 35.73) .
Dropouts: There were no reported dropouts in either study (Short 1984; Wilson 2001).
Adverse effects: No data were reported for this outcome.
Tertiary Prevention- Intervention versus Control
No studies were identified for this comparison.
Tertiary Prevention- Intervention versus Intervention
No studies were identified for this comparison.
All prevention versus control - post hoc selected outcomes
Primary Outcomes
Depression: Two studies (Backman 1997; Norvell 1993) contributed to this outcome at endpoint (N = 89). There was a significant difference in favour of psychosocial intervention (MD (standard) -0.80, 95%CI -1.36 to -0.24) at endpoint. The heterogeneity was within accepted limits (I
Missing data
Doctor 1994 used the General Health Questionnaire but did not report the data.
Doctor 1994 measured job stress but did not report the data.
Sarason 1979 measured anxiety but did not report the data.
Sarason 1979 measured hostility but did not report data.
McNulty 1984 measured hormone levels but did not report the data.
Wilson 2001 used the Police Stress Inventory and did not report data.
Wilson 2001 measured Coping Responses but did not report the data.
Wilson 2001 used the Symptom Checklist (SCL-90) but did not report the data.
O'Neill 1982 measured SCL 90, State-Trait Anxiety Index, Michigan Alcoholism Screening Test, Myers-Briggs Type Indicator and Police Job Stress Inventory but did not report these data.
Shipley 2002 measured CSAI-2 State Anxiety Inventory-2, Subjective evaluations and Performance evaluations but did not report these data.
Aremu 2006 measured Police Interpersonal self image but did not report data.
Sub-group analyses
Insufficient data were available for the a priori sub-group analyses.
Sensitivity analysis
Insufficient data were available for the a priori sensitivity analyses.
Publication bias
Insufficient data were available for the a priori analysis of publication bias.
Discussion
Psychosocial interventions are a common treatment for stress related symptoms and psychological disorders in police officers. The strength of this review lies in its rigorous methods, which include thorough searching for evidence, systematic appraisal of study quality, and systematic and well defined data synthesis. Its main limitation is the lack of evidence with which to inform its results and conclusions. This is similar to another Cochrane review of preventing occupational stress in healthcare workers which found 14 randomised trials, three cluster-randomised trials and two crossover trials, including a total of approximately 2,800 participants and concluded 'limited evidence is available for the effectiveness of interventions to reduce stress levels in healthcare workers. Larger and better quality trials are needed' (Marine 2006). In our review, it is disappointing that despite thorough searching, only 10 randomised studies - with a total of 583 individuals - of psychosocial interventions for the prevention of psychological disorders in police officers were identified. Only four studies contributed outcome data (Backman 1997; Norvell 1993; Short 1984; Wilson 2001) and one further study reported on dropouts only (Doctor 1994).
None of the psychosocial interventions targeted the "job context" stressors, which are far greater the "job content" stressors.
The ten included studies investigated different interventions, participants, and outcomes. For this reason, the studies were analysed in three separate groupings (primary, secondary and tertiary intervention), and the data could not be integrated in a meta-analysis. For an overview of the results of included studies, see the Figures and the Table 'Characteristics of included studies'.
If the results of more trials of high quality become available, our estimates of the effectiveness of psychosocial interventions for the prevention of psychological disorders in law enforcement officers may be made more precise by the quantitative synthesis of trial data, if there is not excessive heterogeneity.
The limited evidence base limits the precision of our results. It also means that a single unidentified trial, or new trials, could have a substantial effect on the findings and conclusions of this review. As so few studies are available, indirect methods of identifying publication bias such as funnel plots are of very limited value, and were therefore not conducted.
The main findings of this review are as follows:
Primary outcomes at endpoint
Psychological symptoms were significantly improved by psychological interventions when compared with a control intervention for depression (Backman 1997). Anger trait, diagnosis of PTSD, marital relationship and distress, depression, anxiety and physical symptoms were all statistically significantly improved (Wilson 2001). Our analyses of drop-outs revealed no statistically differences between any interventions.
Primary outcomes at follow-up of 6 or 18 months
There were statistically significant differences in psychological symptoms (including anger and PTSD) for eye movement desensitization and reprocessing (EMDR) when compared with brief eclectic psychotherapy (BEP) (Wilson 2001). There were no statistically significant differences between the intervention and control group as regards depression (Backman 1997).
Most of the psychological symptoms measured were statistically significantly improved by an intervention when compared with the control group at the endpoint and at the follow up. There were no significant differences for some psychological symptoms including stress (job, stress, burnout, perceived stress) (Backman 1997).
Our findings are based on psychosocial interventions tested in individual trials with small number of participants, which tried to help police officers to cope with stressful events in the workplace rather than interventions aimed at the provision of social and organizational support in their operational experiences which might have changed the conditions or circumstances that may actually causes their occupational stress. There is a huge need for organization-based psychosocial interventions to be examined in police forces.
Authors' conclusions
Implications for practice There is insufficient evidence to show whether police officers benefit from psychosocial interventions including exercise-based interventions. There is lack of evidence that psychosocial interventions can reduce stress-related psychological symptoms. |
Implications for research Further high quality and appropriately large and statistically powered studies of psychosocial interventions in law enforcement officers are required. Specifically, randomised controlled trials with adequate methodological quality (minimising performance, attrition and detection biases). This review would be greatly improved if missing data from the included studies and additional data from new high quality studies were added. Future research is needed to assess organization-based psychosocial interventions. |
Acknowledgements
The authors would like to thank the staff of the Brazilian Cochrane Centre for their help with daily difficulties. We would also like to show our gratitude to Regis Andriolo, Bernardo Soares and Humberto Saconato for their useful comments on the protocol of this review, the CCDAN editorial team for their advice on the protocol and Professor Simon Wessely for guidance and comments on drafts of the protocol and full review.
This review is sponsored by the Associação Fundo de Incentivo à Psicofarmacologia (AFIP) .
Data and analyses
- Top of page
- Background
- Objectives
- Methods
- Results
- Discussion
- Authors' conclusions
- Acknowledgements
- Data and analyses
- What's new
- History
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||
What's new
Last assessed as up-to-date: 11 May 2008.
| |||||||||
History
Protocol first published: Issue 1, 2006
Review first published: Issue 3, 2008
| |||||||||||||||||||||
Contributions of authors
Valentina Penalba: planned, designed and coordinated the review, extracted data, analysed and interpreted the data, and wrote the results and discussion
Jose Roberto Leite: helped develop the protocol, checked studies, and the Abstract and discussion.
Hugh McGuire: checked studies, extracted data, and wrote the results section
Declarations of interest
None known.
Sources of support
Internal sources
- AFIP - Associaçao Fundo de Incentivo a Psicofarmacologia, Brazil.
External sources
- Brazilian Cochrane Centre, Brazil.
Index terms
Medical Subject Headings (MeSH)
*Police; Burnout, Professional [prevention & control]; Depression [prevention & control]; Mental Disorders [*prevention & control]; Occupational Diseases [*prevention & control; psychology]; Randomized Controlled Trials as Topic; Stress, Psychological [psychology]
MeSH check words
Humans
* Indicates the major publication for the study