Universidad Central “Marta Abreu” de Las Villas
CURSO 2006-2007 PRIMER SEMESTRE
Intervention 1: “concrete sensory information”:procedural and sensory information.
Tape-recorded message re. what could experience during hospitalization
Intervention 2: “cognitive-coping technique” -cognitive intervention. Recording in- structing patient to distract from negative aspects and focus on positive
Intervention 3: “Behavioural-coping technique” -behavioural instruction. Recording instructing ways to move to minimise pull on incision and reduce pain
All interventions delivered evening before surgery by research nurse who also answered questions and helped to practise coping techniques
Outcomes Pain- scale from 1 to 10, day 3 post-surgery
Negative affect:Profile of Mood States (POMS: anxiety, confusion, anger, depression, fatigue, vigour). Day 3 post-surgery and 1stand 4t hweek post-discharge
Length of stay
Notes Complex factorial design. Also randomized again to postoperative intervention: discharge
information/no discharge information. This occurred on day 4 postoperation (after day 3 results day) and method allowed to look at main effects of other interventions. Length of stay seems to have been measured as a co-variate rather than as an outcome Author provided information regarding ‘Risk of bias’ assessment
Risk of bias
Bias Authors’ judgement Support for judgement
Random sequence generation (selection bias)
Unclear risk p134: “Patients were randomly assigned to study con-
ditions as their names appeared on the operating room schedule. An exception to random assignment occurred when two subjects occupied the same room. In such cases the second subject was assigned to the same con- dition as the first subject to prevent contamination. Pa- tients lost from the study during the hospital phase were replaced, thus, an equal number of subjects per condi- tion was achieved”. Seems unlikely that patients replac- ing lost patients were randomized - otherwise unclear. Author: “We used a randomly generated number table. The study groups were assigned a number, i.e. 1, 2, 3 or 4 as appropriate for the study. With eyes closed we placed a pencil tip on the table, and starting from that point we moved down the column until we came to one
Johnson 1985 (Continued)
of the numbers”
Allocation concealment (selection bias) Unclear risk No information provided. Author: “A slip of paper with the group assignment was sealed into an envelope. This continued until a significant number of envelopes had been prepared. The envelopes were keep in the order they were prepared. The researcher opened the envelope at the time an intervention was to be delivered. This procedure was used for both studies.” Unclear as may have been successful but would not appear envelopes numbered/ opaque
Blinding of participants and personnel (performance bias)
All outcomes
High risk Not stated, but seems unlikely that participants would be blind if fully informed; research nurses delivering tapes and answering questions would not be blind as some- times helped participants with interventions
Blinding of outcome assessment (detection bias)
All outcomes
Low risk p136: “on the third postoperative day, a nurse who was
not informed of patients’ assignment to study conditions collected data…Finally, the open-ended questions were asked…thus, the data collector elicited information that could have revealed patients’ study condition only after all other data were obtained”. Later time points obtained by postal survey so effectively blind outcome assessor (although some phone reminders)
Incomplete outcome data (attrition bias) All outcomes
Unclear risk 31 lost to follow-up (p133): 19: complication during
surgery/major postoperative complication/additional di- agnostic/treatment procedure after surgery; 8: too ill/de- clined for other reasons to continue participating during hospitalization; 4: failure to contact patient during hos- pitalization. Attrition data are not broken down by study condition
Selective reporting (reporting bias) Unclear risk Authors did not use all outcomes in analyses on basis of judgements of redundancy. Not known what techniques were specified a priori. p137: “The precision of the exper- iment was increased by eliminating outcome indicators with minimal variance and those that were redundant. Depression and anger scores from the POMS were elim- inated because 30 to 45% of the scores were zero at each measurement…fatigue and pain distress scores were also eliminated as indicators”. Emailed authors about 2 pa- pers together: “To the best of my memory, all outcomes that we measured were reported in the articles”
‘Intention-to-treat’ High risk Appears to be by intention-to-treat - p142: report that
only 17% of patients in cognitive coping group reported using the technique but the sample sizes suggest that
Johnson 1985 (Continued)
clearly not only 17% of that group used in analysis (e. g. table 2). From procedure, seems unlikely that par- ticipant would be able to withdraw between consenting and completing allocated treatment as all done in same session. However, email from author: “The idea of ana- lyzing data using the notion of intention-to-treat came into use after the studies had been conducted, so no we did not do those analyses”
Other bias Unclear risk Coping conditions: 94% behavioural-coping group re-
ported using techniques cf only 17% in cognitive-coping group. As result, not clear how to interpret finding where cognitive coping associated with longer stay in hospital Authors discussed allocating patients who were seen to- gether to the same allocation to reduce risk of contami- nation
Klos 1980
Methods Randomized controlled trial (2 x 2 design)
Participants 50 patients undergoing cholecystectomy at a 500-bed hospital, St Joseph Mercy Hospital,
Ann Arbor, Michigan, USA (dates not provided). Mean age: 43.6 years (range: 20 to 72) . 38 (76%) female; 12 male
Interventions Control: no treatment control
Intervention 1: pamphlet containing procedural information: information about operative events; alsobehavioural instructione.g. instructions in deep breathing, leg exercises
Intervention 2: nurse visit providing same content as 1; also assisted patient in practising exercises until mastery achieved
Intervention 3:both pamphlet and nurse visit
Outcomes Length of stay: number of days between surgery and discharge
Negative affect:day 2 post-surgery. Mood Adjective Checklist: 15 adjectives describing the 5 mood dimensions: well-being, happiness, fear, helplessness, anger
Notes Data only presented grouped by preoperative fear level, adjusted for age
Behavioural recovery and negative affect were measured, but using scales without pub- lished psychometric information
Author provided some information regarding risk of bias (stage 1 response only) Risk of bias
Bias Authors’ judgement Support for judgement
Random sequence generation (selection bias)
Unclear risk p7: “Patients were randomly assigned to
Klos 1980 (Continued)
Allocation concealment (selection bias) Unclear risk No information provided in paper
Blinding of participants and personnel (performance bias)
All outcomes
High risk p7: “surgeons and the nursing staff were un-
informed as to the specific dependent vari- ables being studied and the assignment of patients to the experimental groups”. But the participants would have known and all interventions were provided by the same nurses
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk p8: “Patients…were visited both pre-and
postoperatively by the same nurse who vis- ited the patients in the other conditions. The purpose of the visits was to explain the study, secure patient consent and collect data. The data collection and preoperative interventions were carried out by the same two nurses over a period of 2 ½ months.” However, as the outcome of interest to us is length of stay it is not clear how lack of blinding could influence outcome assess- ment. Therefore rating of ‘unclear’ rather than ‘high’ risk
Incomplete outcome data (attrition bias) All outcomes
Unclear risk Attrition and exclusions were reported (p7)
: 1 had an atypical cholecystectomy; for 5 patients the nurse-experimenter was unable to collect all observations; 1 patient in the pamphlet-only group did not read the pam- phlet. The authors do not report how many participants were lost from each group, so the impact in terms of bias is unknown. However, the participant who was excluded for not reading the pamphlet is unlikely to have been excluded if they were in a differ- ent group
Selective reporting (reporting bias) High risk The analyses did not allow the authors’
primary aims to be addressed. The stated purposes were (p7): “1) to test the effects of providing instruction about the usual events of surgery and instructions in leg ex- ercises, turning in bed, getting out of bed, and coughing and deep breathing; and 2) to compare the relative impact of two in- formation-delivery methods on various in- dicators of postoperative recovery, using as delivery methods a preoperative nurse visit
Klos 1980 (Continued)
and/or pamphlet”. However, data address- ing these aims are not provided; all data are split according to low or high preoper- ative fear. This splitting was not only not pre-specified, but seems likely to have re- sulted in unreliably small sample sizes. Data appropriate for meta-analysis are not pro- vided for either length of stay or negative affect outcome
‘Intention-to-treat’ High risk Not intention-to-treat: p7: the patient who
reported not having read the pamphlet in the pamphlet-only condition was excluded from analysis
Other bias Unclear risk Potential for contamination across groups
(e.g. pamphlets could be passed around)
Kulkarni 2010
Methods Randomized controlled trial
Participants 80 patients undergoing major abdominal surgery at a hospital in the UK (dates not
provided). Age: Control: 65, Intervention 1: 64, Intervention 2: 65, Intervention 3: 60 (not clear whether median or mean). Gender: Control: 12 male, 8 female; Intervention 1: 10 male, 10 female; Intervention 2: 5 male, 15 female; Intervention 3: 11 male, 9 female. Overall: 38 male, 42 female, 52.5% female
Interventions Control:no training
Intervention 1: deep breathing training
Intervention 2:incentive spirometry
Intervention 3: specific inspiratory muscle training
All interventions: asked to train x 2 per day for at least 2 weeks prior to surgery (be- havioural instruction, each session 15 minutes)
Outcomes Length of stay
Postoperative pain(no information on how/when measured)
Notes Only Intervention 1 is purely behavioural instruction; Intervention 2 and 3 also involved
a device
Attempted to contact authors; no reply received Risk of bias
Bias Authors’ judgement Support for judgement
Random sequence generation (selection bias)
Low risk p701: “Patients were allocated to four groups by com-
Kulkarni 2010 (Continued)
numbered sealed envelopes” Allocation concealment (selection bias) Unclear risk No further information to above Blinding of participants and personnel
(performance bias) All outcomes
High risk Unlikely that participant was blind; all patients were “as- sessed and trained by the researcher” (p701), indicating that the person administering interventions was also not blind across conditions
Blinding of outcome assessment (detection bias)
All outcomes
High risk p701: patients were “assessed and trained by the re-
searcher” - implies not blind at assessment
Incomplete outcome data (attrition bias) All outcomes
Unclear risk No./% participants lost to follow-up: 14 (17.5%). By
group: Control n = 3, Intervention 1 n = 3, Intervention 2 n = 5, Intervention 3 n = 3. Consort flow chart (p702) clearly accounts for participants for primary outcomes; loss to follow-up is similar across conditions and seems unlikely to be a cause of bias, although given the size of the sample impact is not impossible. However, it is not clear whether this also applies to the assessments of secondary outcomes (those outcomes of interest in this review)
Selective reporting (reporting bias) Unclear risk No protocol mentioned, but clearly state primary out- comes and these are all accounted for. However, less con- sistent with secondary outcomes
‘Intention-to-treat’ Low risk The Consort diagram (p702) reports that, for Group D
(Intervention 3), 2 participants were lost to follow-up because they discontinued intervention - but also states that these patients did not have surgery (which seems likely to explain why they discontinued intervention)
Other bias Unclear risk The authors do not state whether the groups were com-
parable in terms of baseline demographics. Given the small sample size this could be important. From the de- mographics table (Table 1), group D (Intervention 3) seems young (60 years, compared with 65, 64, 65 for A (Control), B (Intervention 1), C (Intervention 2)); also different patterns with gender (12:8; 10:10; 5:15; 11:9)
Lam 2001
Methods Randomized controlled trial
Participants 60 patients undergoing major gynaecological surgery at a hospital in Shatin, Hong Kong
(dates not provided). Control group mean age: 40; intervention group 43. 100% female
Interventions Control:“standard information”
Intervention: education about patient-controlled analgesia (PCA). 15-minute session (verbal instruction), demonstration of PCA device and pamphlet. Includessensory in- formation and behavioural instruction
Outcomes Painseverity (before discharge from recovery room and 24 and 48 hours post-surgery)
Length of stay
Notes Attempted to contact authors; no reply received
Risk of bias
Bias Authors’ judgement Support for judgement
Random sequence generation (selection bias)
Unclear risk No information provided in paper. p466: “half of the
patients (n=30) were also randomly selected to receive additional structured preoperative education”
Allocation concealment (selection bias) Unclear risk No information provided in paper
Blinding of participants and personnel (performance bias)
All outcomes
High risk No information about blinding of study participants -
unlikely given the design. The intervention was admin- istered by the first author (p466) so also not blinded. Anaesthesiologists who provided general anaesthesia “were unaware of the purpose of the study” (p466) Blinding of outcome assessment (detection
bias) All outcomes
Low risk p466: “all patient interviews were conducted in a stan-
dardized fashion by investigators who were blinded as to study group allocation”
Incomplete outcome data (attrition bias) All outcomes
Low risk p467: “All patients completed the study”
Selective reporting (reporting bias) Unclear risk It is not apparent that any outcomes are not reported, but there is no reference of a protocol to which to refer
‘Intention-to-treat’ Low risk All participants are reported to have completed the study;
the same numbers of participants who were randomized were reported per group in Results so unlikely that group- ings could have changed in analysis
Lamarche 1998
Methods Randomized controlled trial
Participants 54 patients undergoing first-time cardiac surgery (CABG) at “a large metropolitan ter-
tiary care teaching hospital in Canada” (dates not provided). Control group: 63.7 years; intervention: 63.5 (assuming means). Control group: 20 male, 6 female; intervention group: 25 male, 3 female. Overall, 83% male
Interventions Control:preadmission teaching session, “cognitive and affective information about hos-
pitalization, along with information about coronary artery disease and lifestyle adjust- ment”.Procedural information, behavioural instruction
Intervention:also received phone call 1 week after teaching session to give additional, personalized information and to discuss feelings.Emotion-focused and procedural information
Outcomes Negative affect: anxiety prior to discharge (10 cm VAS)
Notes Author confirmed suitable for inclusion (general anaesthesia received)
Risk of bias
Bias Authors’ judgement Support for judgement
Random sequence generation (selection bias)
Unclear risk Not stated - p394: “Fifty-four patients were randomly
assigned to the experimental or control group”
Allocation concealment (selection bias) Unclear risk No information provided in paper
Blinding of participants and personnel (performance bias)
All outcomes
High risk Not stated if participants blind but unlikely given nature of intervention. The person delivering the intervention was not blind (p394)
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk No information provided in paper
Incomplete outcome data (attrition bias) All outcomes
Unclear risk No attrition is reported, but no flow-chart to clearly view this process - nor any statement that there was no attri- tion. However, on p398 it is reported that 1 control par- ticipant died during the postoperative period - it is not clear whether these data were included
Selective reporting (reporting bias) Unclear risk All outcomes mentioned in Methods were reported in
Results, but no mention of a protocol with which to examine this in detail
‘Intention-to-treat’ Unclear risk As no attrition or problems with fidelity are mentioned
Lamarche 1998 (Continued)
Other bias High risk Unintended postoperative impact of intervention (p401)
: the investigator who delivered the telephone interven- tion “observed that during hospitalization, patients in the experimental group asked for her by name, and they reported finding comfort in speaking with a nurse with whom they had already established a link through the telephone contact” - so not clear whether any effects on anxiety at discharge due to phone intervention or this established relationship post-surgery
Langer 1975
Methods Randomized controlled trial - analysed as 2 x 2 factorial design
Participants 60 adults undergoing mixed surgery - including hysterectomy, hernia repair, cholecys-
tectomy, transurethral resection, tubal ligation, D and C - at Yale - New Haven Hospital, USA (dates not provided). No information on age or gender
Interventions Control: attention control; administered as all conditions: 20-minute interview a short time after admission
Intervention 1: “Coping device” - trained in cognitive reframing - focusing on positive aspects of a situation.Cognitive intervention
Intervention 2: “Preparatory information” - discussed practices e.g. skin preparation, anaesthesia and what could expect after surgery, e.g. nausea, pain; reassured re. high quality of staff.Procedural and sensory information
Intervention 3: “Combination” - combined components of Intervention 1 and Inter- vention 2, in briefer format so still approximately 20 minutes long.Cognitive interven- tion, procedural and sensory information
Outcomes Length of stay
Notes Author provided details regarding risk of bias
Risk of bias
Bias Authors’ judgement Support for judgement
Random sequence generation (selection bias)
Unclear risk p158: “Subjects were assigned to condi-
tions on a stratified random basis, so that the experimental groups were equated on five relevant background factors: type of operation, seriousness of operation, sex, age, and religious affiliation”
Langer 1975 (Continued)
Blinding of participants and personnel (performance bias)
All outcomes
High risk It is possible that participants were blind, as
all had interview of same length. However, all delivered by same investigator - investi- gator could not be blind
p160: “All physicians, nurses, and others on the hospital staff were kept blind” Blinding of outcome assessment (detection
bias) All outcomes
Low risk No information as to who collected length
of stay data from patient record. Author confirmed outcome assessor blind to treat- ment
Incomplete outcome data (attrition bias) All outcomes
Low risk Lost 1 participant from Coping Device
group (p161), so it would appear that at- trition is being reported - and that the data set was almost complete
Selective reporting (reporting bias) Low risk All measures reportedly taken were re-
ported in Results, but no mention of a pro- tocol document we can refer to. Author confirmed no measures used that were not reported in Results
‘Intention-to-treat’ Unclear risk It seems unlikely that patients could have
changed groups, but it is not stated that intention-to-treat was followed
Other bias Low risk No other concerns
Lauder 1995
Methods Randomized controlled trial
Participants 226 participants randomized - undergoing total abdominal hysterectomy, Southampton,
UK (dates not provided). Control group: mean age 43.0, range 29 to 70; intervention mean age 43.8, range 28 to 77 100% female
Interventions Control:no discussion re. perioperative nausea and vomiting
Intervention:“positive suggestion” but from description seems likeprocedural and sensory information
During a preoperative interview, participants were informed of use of anti-emetics and told about the expected effect of this
Outcomes Pain:0 to 10-point scale, in recovery room and on ward 4, 8 and 24 hours post-surgery
Notes Note:primary endpoint: symptoms of nausea and vomiting; pain scores taken as one
variable which might impact nausea/vomiting scores
Lauder 1995 (Continued)
access to records Risk of bias
Bias Authors’ judgement Support for judgement
Random sequence generation (selection bias)
Low risk p266: “Patients were allocated to a study (positive sug-
gestion) or control group by means of random numbers generated by a computer program”
Allocation concealment (selection bias) Unclear risk No information provided in paper
Blinding of participants and personnel (performance bias)