Psychoanalytic/psychodynamic psychotherapy for children and adolescents who have been sexually abused
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
To assess the effectiveness of psychoanalytic/psychodynamic psychotherapeutic approaches in treating the effects of sexual abuse in children and adolescents.
Background
Description of the condition
Although the so‐called universality of the incest taboo had for a long time led people to believe that sexual abuse of children was a minor problem, it is in fact a significant worldwide problem (De Mause 1991), the scale of which was largely unrecognised until the late 1970s (Finkelhor 1994a).
Sexual abuse is traditionally divided into non‐contact and contact abuse (Peters 1986), with contact abuse subdivided into penetrative and non‐penetrative sexual abuse. Non‐contact sexual abuse includes deliberate exposure of breasts or genitalia or witnessing a sexual act, either live or in photographs or films. Non‐penetrative sexual abuse includes touching, kissing and masturbation. Penetrative sexual abuse includes oral, vaginal and anal penetration (Glaser 2002).
Finkelhor's review of prevalence studies of child sexual abuse in 21 countries (mainly English‐speaking and North European) reported rates ranging from 7% to 36% in women and 3% to 29% in men. Such variation reflects differences in the methodologies of studies, such as the definition of sexual abuse used, rather than substantive differences in the rate of childhood sexual abuse (Finkelhor 1994a). Adjusting for sample‐related variation, response rates and differences in definitions, Gorey and Leslie estimated a rate of 16.8% for women and 7.9% for men from 16 cross‐sectional North American community sample surveys (Gorey 1997). Men make up the vast majority of abusers (85 to 90%) (Glaser 2002) and the majority of sexual abuse is perpetrated by either family members or acquaintances (Finkelhor 1994b).
Importantly sexual abuse is an event or a series of events. It is not a disease or medical condition (Finkelhor 1995). It often comes to the attention of services due to psychological consequences which can be remarkably heterogeneous in their nature. This heterogeneity is partly because sexual abuse rarely occurs in isolation: it often occurs alongside other forms of abuse and neglect (Mullen 1994; Ney 1994) and a wide variety of other disadvantages (eg parental mental illness or substance abuse, child disability or illness, economic deprivation etc).
The psychological effects of sexual abuse can range from significant symptoms and emotional distress to little discernable impact (one review of 45 studies looking at the impact of sexual abuse on children found four studies which estimated from 21% to 49% of victims to be free of symptoms (Kendall‐Tackett 1993)). The most common childhood effects of childhood sexual abuse include symptoms of post‐traumatic stress disorder (PTSD) and sexualised behaviour (Kendall‐Tackett 1993) but other common effects are aggressive or disturbed behaviour, depression, anxiety, low self‐esteem, guilt, fear, eating disorders, self‐harm, suicidal ideation and suicide, substance misuse and relationship problems (Browne 1986; Beitchman 1991; Cotgrove 1996; Tyler 2002). Effects are often dependant on the developmental age of the victim. Overall the severity of abuse, use of force, and the victim’s relationship to the perpetrator appear to have the greatest influence on the degree of impact of the sexual abuse to the victim (Tyler 2002).
Many children who are sexually abused go on to experience a wide range of serious adverse effects in adulthood and some only begin to develop symptoms in later childhood or beyond (‘sleeper effects’). Intervention studies sometimes exclude asymptomatic victims as they are likely to underestimate the effect of the intervention for symptomatic victims. It is also unlikely for an asymptomatic victim to receive psychotherapy in everyday clinical practice, however the possibility of an asymptomatic victim having ‘sleeper effects’ needs to be kept in mind. A higher than average rate of childhood sexual abuse is reported by adults suffering personality disorders and other psychiatric illnesses (Zanarini 1989; Brown 1991; Bulik 2001; MacMillan 2001) and besides intrapsychic difficulties, they experience a wide range of physical symptoms (Arnold 1990), occupational, social and sexual difficulties (Mullen 1994).
Description of the intervention
Psychoanalytic and psychodynamic psychotherapy have their origins in the work of Freud (1856 to 1939) but the practice of psychoanalytic and psychodynamic psychotherapy for children and adolescents gained impetus through the theoretical interest and practice of Anna Freud and Melanie Klein (Daws 1987). Nowadays there are a variety of different approaches (eg Kleinian, Freudian) under the umbrella of psychoanalytic and psychodynamic psychotherapy. These terms are often used synonymously (Gabbard 2005), but psychodynamic psychotherapy sometimes refers to a briefer form of therapy where the therapist is more active and focuses on a particular problem, rather than difficulties affecting the whole personality.
Despite there being different schools of psychoanalytic/psychodynamic psychotherapy used for a wide range of psychological problems, there are some common theoretical concepts: psychoanalytic/psychodynamic therapy involves a therapist listening to an individual and observing their behaviour. People are referred or seek help when they are in conflict over aspects of themselves or their relationships which are problematic. Such conflicts are generally held to originate from difficulties in past relationships and experiences (eg sexual abuse). These conflicts are thought to cause anxiety or psychic pain and are pushed out of consciousness and into the unconscious through the use of defence mechanisms (Bateman 2000). Some defence mechanisms may be helpful (eg humour, altruism), but others may be developmentally immature and harmful (eg denial, splitting, projection). Psychoanalytic/psychodynamic psychotherapy attempts to explore, through talking, play (with younger children) and the formation of a therapeutic relationship, how earlier experiences influence and perhaps seriously distort current feelings, actions and relationships (McQueen 2008). Eventually therapy aims to help patients have a better understanding of difficulties about which they may previously have been unaware and this is thought to allow resolution of their problems.
Therapy is usually one‐to‐one, but it can be done with families or with groups. One‐to‐one therapy tends to happen at a fixed time once or twice a week, but occasionally it can be more intensive and up to five times a week. Parents/carers are often provided with parallel supportive work which is vital to help parents/carers to understand and manage their children’s behaviour safely (McQueen 2008).
Therapy can be brief, lasting only a few sessions, particularly when working with young children or focused on a particular problem. It can also last for months or even years, particularly when the hope is for deep resolution of difficulties affecting the whole personality. When funded by state healthcare systems, insurance companies or for research purposes, therapy may be offered for a fixed amount of time (usually up to one or two years in the UK) and may target a particular problem, but it can sometimes be more open‐ended, especially when paid for privately. Therapists have usually completed a lengthy training which includes undergoing their own psychoanalytic/psychodynamic psychotherapy. When providing treatment therapists tend to get outside supervision from a more experienced therapist.
Following recognition or disclosure of sexual abuse, the first priority is ensuring the child’s safety. If conservative measures are not sufficient in order to protect the child, the potential benefit of removing a child from the home in order to protect their safety has to be balanced against the risk the secondary traumatisation of the child through such separation from their family, particularly if the suspicion of abuse is subsequently proven to be unfounded. When the child's safety is ensured, good practice then involves providing the child with an adult with whom they can talk to about the abuse and their feelings (Glaser 1991). For children who develop symptoms, treatment should be tailored to the developmental age and needs of the individual child (Finkelhor 1995;Ramchandani 2003).
How the intervention might work
There are a variety of theories about how the intervention actually works but, as with other forms of psychotherapy, the relationship between therapist and patient is felt to be important (Gabbard 2005; Midgely 2005). Insight (conscious understanding) of unconscious conflicts is thought to be particularly significant (Traux 1973). Insight is gained through interpretations offered by the therapist based on what the patient says and their behaviour. The patient is encouraged to talk about whatever comes to their mind ('free association'), but some therapists are more active and may ask questions. Therapists usually reveal very few details about themselves so that they are a 'blank screen'. The patient is thought likely to unconsciously behave towards the therapist in the same way that they experienced earlier relationships (eg a parent or an abuser), which also involves them re‐experiencing some of the original anxieties and distress. The transfer of feelings from previous relationships onto the therapist ('transference') allows the therapist to hypothesise about the unconscious conflicts and defence mechanisms of the patient. The therapist can then begin to give interpretations of the unconscious conflicts and defence mechanisms, helping the patient to gain conscious understanding of these.
Play is seen as the primary way for younger children to communicate their unconscious conflicts. The child expresses himself or herself through imaginative and symbolic play, drawing and games, providing the therapist with a window to understanding the child's anxieties, conflicts and defences. Children also talk while they play which provides a safe background for talking about painful subjects. While interpretation is important, the primary agent of change is "enhancing the child's symbolic, imaginative and mentalising capacities by increasing the range, depth and emotional richness of his play" (Target 2005). Thus through play they become able to express their internal conflicts at a more developmentally mature level and this is thought to be curative (Passey 1994; Target 2005; Emanuel 2006).
Eventually, in psychoanalytic/psychodynamic psychotherapy, the relationship with the therapist and the new understanding of the self is thought to be internalised and neural networks based on earlier childhood relationships and experiences are modified (Amini 1996; Schore 1994; Schore 1997; Gabbard 2000; Westen 2002a; Westen 2002b).
Why it is important to do this review
In commenting on a Cochrane systematic review of ‘Cognitive behavioural interventions for children who have been sexually abused’ (Macdonald 2006), which confirmed “CBT’s potential as a means of addressing the adverse consequences of child sexual abuse”, Eamon McCrory observed there is a “need for evaluation of other therapeutic approaches, notably psychodynamic psychotherapy” (McCrory 2007).
To date, there has not been a systematic review of high quality evidence (ie randomised controlled trails) of psychoanalytic psychotherapy for young people who have been sexually abused. This is in contrast to greater availability of evidence for other psychotherapies (especially CBT).
If psychoanalytic/psychodynamic psychotherapy is to continue to receive funding from state healthcare systems and insurance companies, there is a need to continue to review the available literature as new research is published. This review sets out to serve that purpose.
Objectives
To assess the effectiveness of psychoanalytic/psychodynamic psychotherapeutic approaches in treating the effects of sexual abuse in children and adolescents.
Methods
Criteria for considering studies for this review
Types of studies
Randomised controlled trials or quasi‐randomised trials (ie in which participants are allocated by means such as alternate allocation, person's birth date, the day of the week or month, case number or alphabetical order). Studies comparing a psychoanalytic or psychodynamic psychotherapy versus treatment as usual (eg treatment by a psychiatrist) or versus no treatment control/waiting list control will be included.
Types of participants
Children and adolescents up to age 18 years who have experienced sexual abuse at any time prior to the intervention. The participants must be symptomatic at the time of entry into the study.
Types of interventions
Interventions of any duration, delivered to individuals and/or groups, described by the authors as psychoanalytic/psychodynamic or ‐ in the judgement of the review authors ‐ that describe the use of predominantly psychoanalytic/psychodynamic interventions. The following are forms of psychotherapy using psychoanalytic/psychodynamic principles which will be considered for inclusion: child psychotherapy, child and adolescent psychotherapy, child analysis, child psychoanalysis, Freudian therapy, Jungian therapy, Kleinian therapy, Winnicottian therapy, brief psychodynamic psychotherapy, and object relations based therapy.
Studies where psychodynamic/psychotherapy is delivered as an adjunctive treatment (ie, trials in which, for example, drug treatments are used alongside psychoanalytic/psychodynamic interventions (eg psychoanalytic/psychodynamic interventions plus medication versus medication alone) will be included). Studies may or may not include separate parallel supportive work with parents/carers of participants, but must not include therapy with both the child and parent seen together.
Types of outcome measures
Primary outcomes
-
Post traumatic stress disorder (PTSD)
-
Depression
-
Sexualised behaviour
-
Aggression/Conduct problems
-
Self Harm
Adverse outcomes
-
Suicide
Secondary outcomes
Measures of symptoms/psychiatric diagnosis:
-
eg generalised anxiety, panic disorder, social phobia, somatisation symptoms, conversion symptoms, eating disorder symptoms, dissociation.
Measures of underlying processes (relevant to psychoanalytic/psychodynamic psychotherapy):
-
Defence mechanisms (eg Comprehensive Assessment of Defence Style (CADS), Defence Mechanism Inventory (DMI) ‐ Children's Version)
-
Relationship with therapist (eg Child Therapeutic Alliance Scale (CTAS))
-
Transference (eg Core Conflictual Relationship Theme (CCRT) ‐ Child Version)
-
Level/maturity of functioning
Measures of psychosocial functioning:
-
Quality of life
-
Global functioning
-
Social functioning (including peer relationships)
-
Educational functioning
-
Victim‐perpetrator cycle (eg conviction for sexual offences)
-
Disturbed/externalising behaviour (eg 'acting out’)
-
Drug and Alcohol use
Measures of service use:
-
(Psychiatric) number of hospital admissions/days spent in hospital
-
Emergency psychiatric contacts
Other measures:
-
Views of treatment
-
Satisfaction/acceptability with treatment of both participants and parents/carers
-
Withdrawal from treatment (dropout)
-
-
Parental (carer) relationship with the child
Any scales will be accepted for the purpose of the review. Outcomes will be assessed immediately post intervention, for the short‐term (up to one year post‐intervention) and long term (over one year post‐intervention).
Search methods for identification of studies
Electronic searches
The following databases will be searched:
-
Cochrane Central Register of Controlled Trials (CENTRAL)
-
MEDLINE
-
PsycINFO
-
EMBASE
-
CINAHL
-
LILACS
-
ASSIA
-
Sociological Abstracts
-
Dissertation Abstracts International
-
Social Science Citation Index
-
metaRegister of Controlled Trials (mRCT)
-
Child Welfare Information Gateway
The search strategy that will be used for MEDLINE is reproduced below. It will be modified, where necessary, for the other databases listed. Appropriate trials filters (where necessary) will be used in the search of the other databases and full details of these will be reported in the completed review. No language or date restrictions will be applied to the searches.
1 (child$ adj5 abuse$).tw.
2 (sex$ adj5 abuse$).tw.
3 incest$.tw.
4 (sex$ adj5 offenc$).tw.
5 (sex$ adj5 child$).tw.
6 (sex$ adj5 offens$).tw.
7 incest/
8 Child Abuse, Sexual/
9 Incest/
10 or/1‐9
11 exp Psychoanalytic Therapy/
12 psychoanaly$.tw.
13 psychodynamic$.tw.
14 (child$ adj3 analys$).tw.
15 Psychoanalysis/
16 exp Psychotherapy/
17 psychotherap$.tw.
18 (freud$ adj (therap$ or psychotherap$)).tw.
19 (jung$ adj (therap$ or psychotherap$)).tw.
20 (klein$ adj (therap$ or psychotherap$)).tw.
21 (winnicott$ adj (therap$ or psychotherap$)).tw.
22 (object relations adj (therap$ or psychotherap$)).tw.
23 Psychotherapy, Group/
24 (group adj (therap$ or technique$ or psychotherap$)).tw.
25 adolescent/ or child/
26 Child, Preschool/
27 (child$ or girl$ or boy$ or schoolchild$ or adolescen$ or teen$ or pre‐school$ or preschool$).tw.
28 26 or 27
29 11 or 21 or 17 or 12 or 20 or 15 or 14 or 22 or 18 or 24 or 23 or 13 or 16 or 19
30 28 and 10 and 29
31 randomized controlled trial.pt.
32 controlled clinical trial.pt.
33 randomized.ab.
34 placebo.ab.
35 drug therapy.fs.
36 randomly.ab.
37 trial.ab.
38 groups.ab.
39 31 or 32 or 33 or 34 or 35 or 36 or 37 or 38
40 humans.sh.
41 39 and 40
42 30 and 41
Searching other resources
References of previous reviews and studies will be checked to identify any missing studies. Authors and known experts will be contacted to identify any additional unpublished and/or ongoing relevant trials. Efforts will be made to establish contacts in countries in which English is not the dominant language.
Data collection and analysis
Selection of studies
The two authors (BP and WT) will independently review titles and abstracts retrieved from the searches and will identify studies for potential inclusion. Copies of these articles will be obtained, and reviewed independently by the same authors against the inclusion criteria of the study. In the case of disagreement which cannot be resolved by discussion, a third researcher (MZ (Morris Zwi)) will arbitrate and where necessary, the authors of the studies will be contacted to gather further information about the relevance and/or the methodology of the study. All decisions will be recorded.
Data extraction and management
BP and WT will extract data independently using pre‐designed data extraction sheets covering study design (eg randomised or quasi‐randomised), characteristics of the intervention, participants (total number, number per group, age distribution, gender, symptomatology), randomisation method, method of allocation concealment, outcome measures, duration of follow‐up, attrition, handling of missing data, and method of analysis (including intention‐to‐treat analyses). Any disagreement will be resolved by discussion. All decisions will be recorded and if necessary, the authors of the included studies will be contacted to gather further information.
Assessment of risk of bias in included studies
The two authors will independently assess the risk of bias for each included study, and report an agreed view in a 'Risk of Bias' table. The authors will describe and assign a judgement of quality for each of the six domains as recommended in the Cochrane Handbook, (Higgins 2008a) namely sequence generation; allocation concealment; outcome assessors; incomplete outcome data; selective outcome reporting; and ‘other sources of bias’ in a ‘Risk of bias’ table. In all cases, an answer ‘Yes’ indicates a low risk of bias, and an answer ‘No’ indicates a high risk of bias. ‘Unclear’ will indicate an unclear or unknown risk of bias.
Sequence generation
The authors will assess whether the allocation concealment sequence was adequately generated (ie assignment to groups was truly random) and 'randomisation' will receive the following judgements:
'Yes' when participants were allocated to treatment conditions using randomisation based on computer‐generated numbers, table of random numbers or coin‐tossing.
'No' when randomisation did not use any of the above methods.
'Unclear' when randomisation method is not known or not clearly stated.
Allocation concealment
Allocation concealment will receive the following judgements from reviewers:
'Yes' when participants and researchers were unaware of participants' future allocation to treatment conditions until decisions about eligibility/suitability were made and informed consent was obtained.
'No' when allocation was not used, or allocation was not concealed either from participants before informed consent or from researchers before decisions about inclusion were made .
'Unclear' when information regarding allocation concealment is not known or not clearly stated.
Blinding
With regards to blinding, it is not expected that either the therapist or patient could be kept blind to the intervention. Given that many of the outcome measures are likely to be be self‐report, it is therefore probable that blinding of outcome assessments will be low in the included studies. Quality of blinding will be determined primarily by whether those who assessed and coded outcome measures were blind to treatment conditions. Blinding will receive the following judgements:
'Yes' (low risk of bias) when assessors were blind to therapy conditions.
'No' (high risk of bias) when assessors were not blind to therapy conditions.
'Unclear' (uncertain risk of bias) when blinding of assessors is not reported and this information is not available from the study authors.
Incomplete outcome data
Risk of bias due to incomplete data will be assessed in relation to all reported sources of attrition and exclusions, and whether or not these were adequately addressed by the study authors.
Data on attrition and exclusions will be extracted as well as numbers involved (compared to total randomised) and reasons for attrition/exclusion where reported or obtained from the study authors. When attrition between groups differs within studies, sensitivity analyses will be used to determine the extent to which these studies may bias the results of the meta‐analyses. The adequacy of the way the authors of studies dealt with missing data will receive the following judgements:
'Yes' when all participants were included in outcome measure analyses (including those who withdrew from the study) or intention‐to‐treat analysis can be performed using the available data.
'No' when intention‐to‐treat analyses were not performed and cannot be performed using the available data.
'Unclear' when information about whether intention‐to‐treat analyses were performed is not available and cannot be acquired by contacting the authors of the study.
Selective outcome reporting
The possibility of selective outcome reporting will be determined and will receive the following judgements:
'Yes' when all collected data appear to be reported.
'No' when the data from some measures used in the study are not reported.
'Unclear' when it is not clear whether other data were collected and not reported.
Other sources of bias
Assessment will determine whether the study is free from other problems that could put it at a high risk of bias, such as stopping the study early, changing methods during the study or other anomalies.
Measures of treatment effect
Dichotomous data
It is not anticipated that measures will be reported in categorical level measurements. In cases of binary outcomes, Relative Risk (RR) estimations with 95% confidence intervals (CI) (Higgins 2006) will be calculated.
Continuous data
When outcomes in the included studies are measured using the same scale, mean difference (MD) will be calculated. To combine outcomes across studies that have used different scales, standardised mean differences (SMD) will be calculated using Hedges g.
If means and standard deviations are not made available and cannot be calculated from the available information, study authors will be contacted to provide the required information.
Unit of analysis issues
Considering the nature of the intervention examined, it is not anticipated that included studies will deviate from the simple parallel group design.
Multiple interventions per participant
If the participants in some trials receive psychoanalytic or psychodynamic psychotherapy plus treatment as usual (eg treatment by a psychiatrist), those studies will be meta‐analysed separately, with the psychoanalytic or psychodynamic psychotherapy plus treatment as usual arm compared to treatment as usual alone. The treatment effects of these studies will be reported separately and discussion of these results will consider the extent to which additional treatments may have influenced outcomes.
Multiple time points
Studies of the effectiveness of psychotherapeutic interventions sometimes measure outcomes at multiple time points post‐intervention. Ideally time points for assessing the impacts of treatments would be taken from randomisation, but given the likely variability in the duration of interventions between studies we have chosen to group measurements into those taken immediately post‐intervention, those at short‐term follow‐up (up to one year post‐intervention) and those at long‐term follow‐up (the final measure, greater than one year post‐intervention).
Dealing with missing data
Missing dichotomous data will be managed through intention to treat (ITT) analysis, in which it will be assumed that patients who dropped out after randomisation had a negative outcome. Best/worse case scenarios (Gamble 2005) will also be calculated for the clinical response outcome, in which it will be assumed that dropouts in the active treatment group had positive outcomes and those in the control group had negative outcomes (best case scenario), and that dropouts in the active treatment group had negative outcomes and those in the control group had positive outcomes (worst case scenario), thus providing boundaries for the observed treatment effect. As this approach is not without its problems in drawing inferences about the pooled effect estimate (Higgins 2008b), we will seek additional input from a statistician.
Missing continuous data will either analysed on an endpoint basis, including only participants with a final assessment, or analysed using last observation carried forward to the final assessment (LOCF) if LOCF data were reported by the trial authors. Where SDs are missing, attempts will be made to obtain these data through contacting trial authors. Where SDs are not available from trial authors, they will be calculated from t‐values, confidence intervals or standard errors, where reported in articles (Deeks 1997a; Deeks 1997b). If these additional figures are not available or obtainable, the study data will not be included in the comparison of interest.
Assessment of heterogeneity
Statistical heterogeneity among included studies will be assessed by using the chi2 test, which provides evidence of variation in effect estimates beyond that of chance. Since the chi2 test has low power to assess heterogeneity when a small number of participants or trials are included, the p‐value will be conservatively set at 0.1. Additionally, the I2 statistic (Higgins 2002,Higgins 2003) will be used to determine the percentage of variability that is due to heterogeneity rather than to sampling error or chance (where a value greater than 50% suggests moderate to substantial heterogeneity). The possible reasons for any heterogeneity will be discussed and sensitivity analyses will be conducted accordingly, where data permit. Subgroup analyses and meta‐regression may be used to investigate this further (see below).
Assessment of reporting biases
Funnel plots will be drawn to investigate relationship between effect size and standard error when possible (Egger 1997). When such a relationship is found, clinical diversity will be examined as a possible explanation. Asymmetry could be attributed to publication bias or related biases.
Data synthesis
Data will be entered into RevMan 5.0 by the two review authors (double data entry). A random‐effects model meta‐analysis will be used due to expected heterogeneity among included studies. We will synthesize results in a meta‐analysis providing there is not significant clinical heterogeneity (ie in terms of participants, interventions, methodology, and outcome measurement. If significant heterogeneity is found (for example, when value of the I2 statistic exceeds 75% and/or when studies are dissimilar in terms of important participant factors (e.g. age) or study factors (e.g. definition of sexual abuse used)) and there is inconsistency in the direction of effect, meta‐analyses will not be not deemed appropriate (Higgins 2008) and the results of the each study will be provided.
Subgroup analysis and investigation of heterogeneity
Further investigation of the causes of heterogeneity will be conducted using subgroup analyses. The following subgroup analyses will be conducted with all outcomes that have a sufficient number of studies (normally greater than 10) to see if there are any differences in response between:
-
Child (under 13 years) and adolescent (13‐18 years) participants
-
Male and female participants
-
Studies of group treatment versus studies of individual treatment
-
Studies where psychodynamic/psychoanalytic psychotherapy is delivered as an adjunctive treatment versus studies of psychodynamic/psychoanalytic psychotherapy alone.
-
Studies with the intervention lasting an average of 25 or fewer sessions versus studies with the intervention lasting an average of more than 25 sessions
If sufficient number of studies is identified, meta‐regression will be used to examine study’s effect size variation as a function of potential sources of clinical heterogeneity.
Sensitivity analysis
If the methodologies or analyses of the studies may have affected the robustness of the results of the review, sensitivity analyses will be undertaken to examine the effects of:
-
The removal of studies with quasi‐randomisation
-
The removal of studies with inconsistencies in the definition, measurement, and/or reporting of results (for example, differential attrition, dropouts, lack of intention‐to‐treat analysis, outcome measures not taken at consistent time point for all participants).
-
Method used by review authors to impute values for missing data (for example, participants with a final assessment versus last value carried forward (LOCF)).
-
Reanalysing the data using a different statistical approaches (for example, using a fixed‐effect model instead of a random‐effects model (Higgins 2008a).
