Patient decision aids for cancer treatment

Are there any alternatives?

Authors

  • Gillian Spiegle BSc,

    1. University of Toronto, Toronto, Ontario, Canada
    2. Division of General Surgery, Mount Sinai Hospital and Samuel Lunenfeld Research Institute, Toronto, Ontario, Canada
    Search for more papers by this author
  • Eisar Al-Sukhni MD,

    1. University of Toronto, Toronto, Ontario, Canada
    2. Division of General Surgery, Mount Sinai Hospital and Samuel Lunenfeld Research Institute, Toronto, Ontario, Canada
    3. Zane Cohen Centre for Digestive Diseases, Toronto, Ontario, Canada
    Search for more papers by this author
  • Selina Schmocker BSc,

    1. Division of General Surgery, Mount Sinai Hospital and Samuel Lunenfeld Research Institute, Toronto, Ontario, Canada
    2. Zane Cohen Centre for Digestive Diseases, Toronto, Ontario, Canada
    Search for more papers by this author
  • Anna R. Gagliardi PhD,

    1. University of Toronto, Toronto, Ontario, Canada
    2. Toronto General Hospital and Toronto General Research Institute, Toronto, Ontario, Canada
    3. Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
  • J. Charles Victor MSc,

    1. University of Toronto, Toronto, Ontario, Canada
    2. Institute of Clinical Evaluative Sciences (ICES), Toronto, Ontario, Canada
    3. Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
  • Nancy N. Baxter MD, PhD,

    1. University of Toronto, Toronto, Ontario, Canada
    2. Institute of Clinical Evaluative Sciences (ICES), Toronto, Ontario, Canada
    3. Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada
    4. Department of Surgery and Keenan Research Centre, Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada
    Search for more papers by this author
  • Erin D. Kennedy MD, PhD

    Corresponding author
    1. University of Toronto, Toronto, Ontario, Canada
    2. Division of General Surgery, Mount Sinai Hospital and Samuel Lunenfeld Research Institute, Toronto, Ontario, Canada
    3. Zane Cohen Centre for Digestive Diseases, Toronto, Ontario, Canada
    4. Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada
    5. Department of Surgical Oncology, Princess Margaret Hospital, Toronto, Ontario, Canada
    • Division of General Surgery, Mount Sinai Hospital, 600 University Avenue, Suite 455, M5G 1X5, Toronto, Canada

    Search for more papers by this author
    • Fax: (416) 586-5932


Abstract

BACKGROUND:

Although patient decision aids (pDAs) are effective, widespread use of pDAs for cancer treatment has not been achieved. The objectives of this study were to perform a systematic review to identify alternate types of decision support interventions (DSIs) for cancer treatment and a meta-analysis to compare the effectiveness of these DSIs to pDAs.

METHODS:

The inclusion criteria for the study were: 1) all published studies using a randomized, controlled trial design, and 2) DSIs involving treatment decision-making for breast, prostate, colorectal, and/or lung cancer. For this analysis, DSIs were classified as pDAs if: 1) one reported outcome measure mapped onto the International Patient Decision Aids Standards Collaboration effectiveness criterion, and 2) the DSI was evaluated relative to standard consultation. Random effects models were used to compare the effectiveness of pDAs relative to other identified DSIs for reported outcomes.

RESULTS:

A total of 71 studies were reviewed, and 24 met the inclusion criteria. Overall, there were no significant differences in knowledge, satisfaction, anxiety, or decisional conflict scores between pDAs and other DSIs.

CONCLUSIONS:

This study showed that the effectiveness of other DSIs, including question prompt lists and audiorecording of the consultation, is similar to pDAs. This is important because it may be that these less complex DSIs may be all that is necessary to achieve similar outcomes as pDAs for cancer treatment. Cancer 2013. © 2012 American Cancer Society.

Although patient decision aids (pDAs) are considered the gold standard for assisting patients in treatment decision-making, widespread use of pDAs for cancer treatment has not been achieved.1-4 The main reasons for this include lack of physician awareness, inapplicability of pDAs to the individual patient, and concerns regarding physician autonomy.4, 5 Other more practical issues affecting the implementation of pDAs include the relatively high financial cost to develop, evaluate, and maintain and the fact that many pDAs become obsolete as new treatment options become available.6 Because pDAs are not being routinely used for cancer treatment, our group felt it would be useful to identify and evaluate alternative strategies to pDAs to assist patients with treatment decision-making. Therefore, the objectives of this article were to: 1) perform systematic review to identify alternate types of decision support interventions (DSIs) used for breast, prostate, colorectal, and lung cancer, and 2) perform a meta-analysis to compare the effectiveness of these DSIs to pDAs.

MATERIALS AND METHODS

For this study, the term “decision support interventions” was used, because it recognizes the existence of different types of support methods including pDAs and was defined as any intervention that assists the individual to understand their treatment options and assist them to deliberate, either independently or in collaboration with others, in order to arrive at a meaningful choice consistent with their own personal values.7

Literature Search and Data Abstraction

A literature search was performed using MEDLINE, EMBASE, CINAHL, HEALTHSTAR, PSYCHINFO, and Cochrane Database of Systematic Reviews. The search was limited to articles that included human subjects, written in English language, and published between January 1990 and June 2011. The search included the following Medical Subject Headings: decision making, decision support techniques, decision support systems, shared decision making, decision tool, education aid, education tool, decision counseling, patient participation, physician-patient relations, patient choice, and patient preference. Abstracts identified from the literature search were retrieved and independently evaluated by 2 reviewers (G.S., E.D.K.). The inclusion criteria set a priori for the study were: 1) all published studies using a randomized controlled trial (RCT) design comparing DSIs to standard care or alternative interventions, and 2) DSIs involving treatment decision-making for breast, prostate, colorectal, and/or lung cancer. The full articles of abstracts for which the inclusion criteria were judged to be either fully met or unclear were retrieved and reviewed in full by 3 reviewers (G.S., S.S., E.D.K.). Data from these articles were abstracted onto a standardized data abstraction form, and the reviewers met at regular intervals to discuss the articles and resolve any discrepancies by consensus.

Assessment of Methodological Quality

The quality of the RCTs was assessed using the criteria set out by the Cochrane Collaboration.8 The minimum criteria for fulfilling each design element was independently reviewed by the 3 reviewers, and discrepancies were resolved by consensus (S.S., E.A., E.D.K.). Each of the 7 items were scored as “yes,” “unclear,” or “no.”

Statistical Analysis and Data Synthesis

Because it was expected that the reported outcomes would be measured on a number of different scales and time points, a priori we planned to summarize these outcomes using standard mean differences (SMD) for the most immediate postintervention follow-up interval. For studies reporting only medians and interquartile ranges, means and standard deviations were estimated.9, 10 Standard inverse-variance random effects meta-analysis models were used to combine the trials and report the overall SMD and 95% confidence intervals (CIs) for each outcome.11 Based on this model, SMD values of 0.20, 0.50, and 0.80 standard deviation units are considered small, moderate, and large differences, respectively.12, 13

Heterogeneity between the studies included in the model were evaluated using chi-square tests for the tau-squared statistic and quantified using the I2 statistic.14, 15 Where significant heterogeneity was found, sensitivity analyses removing studies was performed to determine factors related to the heterogeneity and the effect on the pooled outcome.

The primary outcome for the meta-analysis was to compare the effectiveness of pDAs relative to other identified DSIs for reported outcomes. For this analysis, DSIs were classified as pDAs if: 1) at least 1 reported outcome measure mapped onto the International Patient Decision Aids Standards Collaboration (IPDAS) effectiveness criterion for specified decision and decision process attributes, as described by the Cochrane Collaboration,2 and 2) the DSI was evaluated relative to the standard consultation (ie, not to an alternative intervention).

RESULTS

Search Results and Study Characteristics

The literature search (not limited to RCTs), identified 1176 citations, of which 71 were considered potentially eligible for the study and were fully reviewed. Of these 71 articles, 24 met the inclusion criteria and are shown in Table 1.16-39 The main reason for exclusion from the study was that the study design was not a RCT. The reference lists of the 71 articles were reviewed, and no further articles were identified.

Table 1. Overview of Randomized, Controlled Trials Included in Review
Ref.Study, Year (Location)NContextInterventionControlDesign CategoryTimingAdministration
Adjuvant Treatment for Primary Breast Cancer
 
16Hack, 2003 (Canada)628Adjuvant chemotherapy and radiotherapy for breast cancerAudiorecordingStandard careReinforcementAfterPatient
17Whelan, 2003 (Canada)176Adjuvant chemotherapy for node negative breast cancerDecision boardStandard careInformation givingDuringNurse
18Siminoff, 2006 (USA)405Adjuvant chemotherapy for breast cancerComputer program (Adjuvant!)PamphletInformation givingDuringHealth educator
19Peele, 2005 (USA)386Adjuvant chemotherapy for breast cancerComputer program (Adjuvant!)PamphletInformation givingDuringHealth educator
20Brown, 2004 (Australia)65Newly referred female patients to medical oncology clinic (predominantly early stage breast cancer)Booklet + videoPamphletInformation givingBeforePatient
Surgery for Primary Breast Cancer
 
21Vodermaier, 2009 (Germany)111BCS for T1 breast cancer OR preoperative chemotherapy for T2 or T3 breast cancerDecision board + coaching + pamphletStandard careEnhanced participationBeforePsychologist
22Street, 1995 (USA)60Breast-conserving surgery for Stage I and Stage II breast cancerComputer program (Options for treating breast cancer)PamphletInformation givingBeforePatient
23Whelan, 2004 (Canada)201Breast conserving surgeryDecision boardStandard careInformation givingDuringPhysician
24Goel, 2001 (Canada)136Breast conserving surgeryDecision aid (audiotape + workbook + values clarification exercise)PamphletInformation givingAfterPatient
25Jibaja-Weiss, 2011 (USA)100Breast conserving surgeryDecision aid (interactive computer program + values clarification exercise)Standard careInformation givingBeforePatient
Primary Prostate Cancer
 
26Hack, 2007 (Canada)425Initial consultation with radiation oncologist for prostate cancerAudiorecordingStandard careReinforcementAfterPatient
27Davison, 2007 (Canada)324Initial consultation with urologist for prostate cancerIndividualized computer program + output + information packageGeneric video + information packageInformation givingBeforePatient
28Davison, 1997 (Canada)60Initial consultation with urologist for prostate cancerInterview + individualized question list + audiorecording of consultInterview + information packageEnhanced participationBeforeNurse
29Feldman-Stewart, 2006 (Canada)308Initial consultation for prostate cancer (includes patients and family members)Prostate cancer information bookletPamphletInformation givingNot recordedPatient
30Auvinen, 2004 (Finland)203Initial consultation with urologist for prostate cancer (all comers)Decision boardStandard careInformation givingDuringPhysician
31Mishel, 2009 (USA)256Initial consultation for prostate cancer (includes patients and family members)Workbook + prompt list + coachingStandard careEnhanced participationBeforeNurse
Locally Advanced Cancer (Locoregional, Recurrent and Metastatic)
 
32Ford, 1995 (United Kingdom)117Newly referred patients to medical oncology (mixed cancers)Audiorecording of initial consultation and second follow-up visitStandard careReinforcementAfterPatient
33Butow, 2004 (Australia)164Newly referred patients to medical and radiation oncology clinics (mixed cancers)Information package + question prompt sheet + audiorecordingPamphletEnhanced participationBeforePatient
34Brown, 1999 (Australia)60Newly referred medical oncology patients (mixed cancers)Question prompt sheet ± coachingStandard careEnhanced participationBeforePatient
35De Lorenzo, 2004 (Italy)300Newly referred medical oncology patients (mixed cancers)Booklet + videorecordingStandard careInformation givingBeforePatient
36Bruera, 1999 (USA)60Newly referred patients with locally advanced and metastatic cancer to Pain and Symptom Clinic (mixed cancers)Audiorecording + written summary of consultWritten summary of consultReinforcementAfterPatient
37Clayton, 2007 (Australia)174Newly referred palliative care patientsQuestion prompt listStandard careEnhanced participationBeforePatient
38Brown, 2001 (Australia)318Newly referred medical and radiology clinics (mixed cancers)Question prompt list ± physician endorsementStandard careEnhanced participationDuringPatient
39Leighl 2011 (Canada)207Newly referred patients to medical oncology (colorectal cancer and liver metastases)Booklet + DVDStandard careInformation givingAfterPatient

The 24 RCTs were published between 1995 and 2011 and included a total of 5244 patients (range, 60-628). Overall, 10 RCTs assessed DSIs for breast cancer (5 adjuvant therapy, 5 surgery), 6 initial treatment for prostate cancer, and 8 locally advanced and/or metastatic disease in mixed cancers including breast, prostate, colorectal, and/or lung cancer patients. Thirteen RCTs assessed DSIs relative to standard care, and the other 11 RCTs assessed DSIs relative to alternative interventions such as pamphlets or other similar materials.

The quality of the RCTs was assessed using the criteria set out by the Cochrane Collaboration and showed that <50% of the trials clearly described critical study design elements in particular sequence generation and allocation concealment (Table 2).8

Table 2. Quality of Selected Randomized, Controlled Trials
Quality IndicatorsNumber of Studies
YesUnclearNo
Sequence generation9150
Allocation concealment6180
Blinding of patients and personnel0024
Blinding of outcome assessment0024
Incomplete outcome data accounted for6153
Free of selective outcome reporting2202
Free of other potential threats to validity2400

pDAs

Seven of the selected RCTs evaluated DSIs that met the specified criteria for pDAs,17, 21, 23, 25, 30, 31, 39 and the effectiveness of these 7 pDAs for each of the reported outcomes was compared with the effectiveness of the remaining other 17 DSIs identified in the literature review.

Other Types of DSIs

During the review process, 3 specific design strategies emerged, and a dominant DSI was identified within each of these design strategies. The 3 design strategies were: 1) information giving, 2) enhanced participation, and 3) reinforcement; the dominant DSI within each of these respective design strategies were: 1) pDAs and patient decision boards, 2) question prompt lists, and 3) audiorecording of the consultation. When this classification was applied to the 24 RCTs included in the study, 13 DSIs were information giving, 7 DSIs were participating enhancing, and 4 DSIs used a reinforcement strategy (Table 1).

Method of Administration

Sixteen of the DSIs evaluated were self-administered by the patient, 6 DSIs were administered by an allied health professional or nurse, and 2 DSIs were administered by the treating physician. Overall, 11 DSIs were administered before the consultation, 6 during the consultation, and 6 after the consultation (Table 1).

Reported Outcomes

The most commonly reported outcomes were knowledge and recall, satisfaction, anxiety, decisional conflict, and patient question asking (Table 3).

Table 3. Summary of Reported Randomized, Controlled Trial Outcomes
Ref.NInterventionKnowledgeSatisfactionAnxiety, Depression, MoodDecisional ConflictQuestion Asking
  • Abbreviations: Beck, Beck Depression Inventory; CES-D, Centre for Epidemiologic Studies Depression Scale; DCS, Decisional Conflict Scale; POMS, Profile of Mood States; SSAI, Spielberger State Anxiety Inventory; STAI, State Trait Anxiety Inventory; VAS, visual analog scale.

  • a

    Actual treatment decision was the only outcome evaluated for this study

Adjuvant Therapy for Primary Breast Cancer
16628AudiorecordingInformed Communication Scale: (5 items, maximum score 25): at 3 mo Side effects 4.68 DSI vs 4.46 control, P = .01Patient Satisfaction with Communication: NSPOMS: NS  
17176Decision board25-item scale, (maximum score 100): 80.2 DSI vs 71.7 control, P < .001)Satisfaction with Decision Making (DCS subscale, maximum score 5 ): Increased satisfaction DSI vs control at weeks 1, 12, 26, and 52, P = .032SSAI: NS  
18a405Computer program (Adjuvant!)     
19a386Computer program (Adjuvant!)     
2065Booklet + video Satisfaction with consultation: NS Satisfaction with TDM :NSSSAI: 32.3 DSI vs 40.3 control at 2 weeks, P = .01 Beck: NSDCS: NSMean number of questions asked: NS
Surgery for Primary Breast Cancer
21111Decision board Satisfaction with TDM: NS Satisfaction with consultation (ZUF8): NS DCS: NS 
2260Interactive computer program “Options for Treating Breast Cancer”11 multiple choice questions: NS Optimism: NS Mean number of questions asked: NS (raw data not provided)
23201Decision board44-item scale, (maximum score 100): 66.9 DSI vs 58.7 control, P < .001Satisfaction with Decision Making (DCS subscale, maximum score 5): After consult: 4.50 DSI vs 4.32 control, P = .05 At 6 mo: NS At 12 mo: NSSSAI:NS CES-D: NSDCS: After consult: 1.40 DSI vs 1.62 control, P = .02 At 6 mo: NS At 12 mo: NS 
24136Decision aid (audiorecording and workbook)BCIT-R (18 items): NS SSAI: NSDCS: NS 
25100Decision aid (interactive computer program + values clarification exercise)16-item scale (maximum score 16): After intervention 9.9 DSI vs 6.4 control, P < 0.001 At 12 mo: NSSatisfaction with Decision Making Process: After intervention: NS At 12 mo: NS Satisfaction with Decision: After intervention: NS At 12 mo: NS DCS (Low Literacy) After intervention: NS At 12 mo: NS 
Primary Prostate Cancer
26425AudiorecordingInformed Communication Scale (5-items, maximum score 25): at 3 mo Overall: 21.83 DSI vs 21.01 control, P = .04 Treatment alternatives: 4.73 DSI vs 4.56, P = .04 Side effects: 4.63 DSI vs 4.37, P = .01Patient satisfaction with communication: NSPOMS: NS  
27324Computer program Preparation for decision making: NS Satisfaction with information and treatment decision, (5 items, maximum score 5 per item): Amount of info (P = .004), type of info (P = .001), administration (P = .004) better with DSI (no mean scores provided) DCS: NS 
2860Information + Question prompt list + Audiorecording  SSAI: NS CES-D: NS  
29308Prostate-specific educational booklet (patients and family members) Preparation for Decision Making Scale (10 items, maximum score 5): 2.8 DSI vs 2.5 control, P = .047SSAI: NS Stanford Inventory of Cancer Patient Adjustment: NSDCS: NS 
30a203Decision board     
31256Workbook + prompt list + coachingProstate Cancer Knowledge Scale; 20 items, maximum score 20 After DSI: 15.02 DSI vs 13.88 control; P < .05 At 3 mo: 15.21 DSI vs 14.51 control; P < .05Patient-Provider Communication Scale: 5 items on 5-point scale; high scores represent better communication After DSI: 19.43 DSI vs 17.81 control; P < .05 At 3 mo: NS Medical Communication Competence; 5-point Likert scale After DSI: NS At 3 mo: 70.55 DSI vs 69.89 control; P < .05POMS: NS  
Locally Advanced Cancer (Locoregional, Recurrent, Metastatic)
32117Audiorecording of initial consultation and first follow-up visitInformation retention questionnaire (9 items, maximum 100/item): at 6 mo Recall improved in DSI for tests and results, treatment name, treatment alternatives, side effects, and self-care instructions (P < .05) Hospital Anxiety and Depression Scale: NS Mean number of questions asked: Initial consultation: NS Second follow-up visit: NS
Locally Advanced Cancer
33164Information package + question prompt list Satisfaction with the consultation: NS Satisfaction with decision making: NSSSAI: NS Beck: NS Mean number of questions asked: 13 DSI vs 9 control, P = .009
3460Question prompt list ± coaching Satisfaction with the consultation: NSSSAI: NS Mental Adjustment to Cancer: NS Mean number of questions asked: 14 DSI vs 8.5 control, P = .043
35300Booklet + video  Psychologic Distress Inventory: NS Anxiety (VAS): NS  
3660Audiorecording + written summary of consult10 items (maximum score 100): 88 DSI vs 80 control, P = .02Usefulness of clinic (VAS): 8.7 DSI vs 7.7 control, P = .04   
37174Question prompt list Satisfaction with the consultation: NSSSAI: NS Mean number of questions asked: 5.4 DSI vs 2.3 control, P < .0001
38318Question prompt list ± Physician endorsement Satisfaction with consultation: NSSSAI: Immediately after consult: 37 DSI vs 32 control; 31 DSI + physician At 1 week: NS Median number of questions: NS
39207Booklet + DVD10-item scale (maximum score 16): median 11.6 DSI vs 9.6 control (P < .001)Satisfaction with the consultation (immediately after consultation): NS Satisfaction with decision-making (4 weeks postdecision): NSSTAI: NSDCS: NS 

Patient Knowledge and Recall

Of the 11 RCTs evaluating the effect of the DSI on patient knowledge and recall, all but 1 used nonvalidated scales.24 Overall, 9 of the 11 trials showed a significant improvement in patient knowledge or recall with the DSI, compared with control. All 4 trials evaluating audiorecording resulted in a significant improvement in patient recall compared with standard care, and this was due to improved recall for treatment alternatives and side effects.16, 26, 32, 36

Patient Satisfaction With the Consultation

Patient satisfaction was assessed in 16 studies on a variety of validated and nonvalidated scales. Overall, 10 trials reported no significant difference in patient satisfaction with the consultation or decision-making process between the DSI and control groups. Two trials showed a significant improvement in patient satisfaction with treatment decision-making on the subscale of the Decisional Conflict Scale (DCS),17, 23 and although this effect continued for up to 52 weeks in 1 trial, it was only seen immediately after the consult in the second trial.

Anxiety, Mood, and Depression

Seventeen RCTs assessed anxiety, depression, and/or mood, and 10 of these trials used the Spielberger State Anxiety Inventory. Overall, 15 of these trials found no significant difference in anxiety between the DSI and control groups. In the remaining RCTs, the first showed a significant decrease in anxiety with the DSI compared with the control group 2 weeks after administration, and the second showed a significant increase in anxiety with the DSI compared with control immediately after the consultation, but this effect was not found 1 week after administration of the DSI.20, 38

Decisional Conflict

Eight RCTs evaluated decision conflict using the DCS developed by O'Connor. Overall, 7 of these trials found no significant difference on the DCS between the DSI and control groups. In the remaining study, a significant decrease on the DCS was found immediately after administration of the DSI, but this effect was not maintained at 6 or 12 months.23

Patient Question Asking

Seven RCTs evaluated the number of questions asked by the patient during the consultation. Of these, 3 trials evaluating the question prompt list showed a significant increase in the mean number of questions patients asked during the consultation. These trials also found that patients asked more questions with respect to prognosis, tests, and side effects than patients in the control group.33-34, 37

Effectiveness of pDAs Versus Other DSIs

A random effects model was used to determine the effectiveness of pDAs compared with other DSIs for the reported outcomes. The results of this analysis showed a significant improvement in patient knowledge with both pDAs and other DSIs (pDAs: SMD = 0.38 [95% CI = 0.17, 0.59], P = .0005 versus Other DSIs: SMD = 0.22 [95% CI = 0.12, 0.33], P < .0001) (Fig. 1). Although there was no significant improvement with either pDAs or other DSIs in terms of patient satisfaction (pDAs: SMD = 0.17 [95% CI = −0.03, 0.36], P = .11 versus Other DSIs: SMD = 0.11 [95% CI = −0.02, 0.23], P = .11), there was a trend toward improved satisfaction in both subgroups (Fig. 2). Despite increased overall knowledge, there was no significant change in anxiety with either pDAs or Other DSIs (pDAs: SMD = −0.09 [95% CI = −0.26, 0.08], P = .31 versus Other DSIs: SMD = 0.03 [95% CI = −0.10, 0.17]. P = .62) (Fig. 3). Similarly, pDAs did not significantly affect DCS scores compared to Other DSIs (pDAs: SMD −0.19 [95% CI = −0.55, 0.17], P = .30) versus Other DSIs: SMD = 0.00 [95% CI = −0.22, 0.21], P = .97) (Fig. 4). Although none of the 7 pDAs evaluated patient question asking, this outcome was significantly improved with the other DSIs (SMD = 0.16 [95% CI = 0.02, 0.29], P = .02) (Fig. 5).

Figure 1.

Results are shown for patient knowledge. CI indicates confidence interval; DSI, decision support intervention; pDA, patient decision aid; SD, standard deviation; SMD, standard mean difference.

Figure 2.

Results are shown for patient satisfaction. CI indicates confidence interval; DSI, decision support intervention; pDA, patient decision aid; SD, standard deviation; SMD, standard mean difference.

Figure 3.

Results are shown for patient anxiety. CI indicates confidence interval; DSI, decision support intervention; pDA, patient decision aid; SD, standard deviation; SMD, standard mean difference.

Figure 4.

Results are shown for Decisional Conflict Scale scores. CI indicates confidence interval; DSI, decision support intervention; pDA, patient decision aid; SD, standard deviation; SMD, standard mean difference.

Figure 5.

Results are shown for patient question asking. CI indicates confidence interval; DSI, decision support intervention; pDA, patient decision aid; SD, standard deviation; SMD, standard mean difference.

The I2 scores ranged from 0% to 59% representing low to moderate heterogeneity and were not statistically significant,15 except for the analysis of DCS among pDAs (I2 = 75%, P = .02; Fig. 4). For this comparison, sensitivity analysis showed that omission of the study by Leighl et al39 (for which means and standard deviations had to be estimated from medians and ranges) led to a significant, positive effect of pDAs on DCS scores (SMD = −0.37 [95% CI = −0.60, −0.15], P = .001, I2 = 0%) based on the 2 remaining studies in the analysis.

DISCUSSION

This study identified other DSIs, including question prompt lists and audiorecording of the consultation, for treatment of breast, prostate, lung, and colorectal cancer and showed that the effectiveness of these other DSIs for reported outcomes is similar to pDAs. This finding is particularly important, because it suggests that less complex DSIs may be all that is necessary to achieve similar outcomes as pDAs for cancer treatment. Some of the advantages of question prompt lists and audiorecording are that: 1) by design, they are specifically tailored to the individual patients and are therefore applicable to all patients and settings, 2) they have a negligible effect on the flow of the clinic, 3) they can easily be administered by the treating physician, 4) they are relatively inexpensive, and 5) they do not require regular updating. These advantages are important, because they avoid some of the implementation challenges faced by pDAs such as cost, disruption of the flow of clinic, and inapplicability of the pDA to the clinical setting and/or individual patient.5, 6 Furthermore, Sepucha et al40, 41 reported a high level of information overload in patients with breast cancer; therefore, it may be that DSIs that combine information giving with either enhanced participation and/or reinforcement design strategies may be more effective than information giving alone.

One of the other striking findings of this study was that only 2 DSIs were administered by the treating physician.23, 30 In both of these trials, there was a significant increase in the proportion of patients choosing the less invasive treatment option for breast cancer surgery and prostate cancer treatment.23, 30 More recently, there has also been increasing evidence to support improved outcomes when the treating physician is actively involved in the use of DSIs. Nannenga et al performed an RCT that showed an increase in both patient knowledge and trust in the physician when a pDA for statin choice was administered to diabetic patients by the treating physician during the consultation compared with a non–treating physician prior to the consultation.42 Finally, Schroy et al performed an RCT that found use of a pDA led to 59% concordance between the patient's preferred screening option for colorectal cancer and the screening test ordered by the physician, indicating that effectiveness of the DSI seems to be largely determined by the extent that physicians comply with patient preferences.43

There is also some evidence to suggest that physician satisfaction and length of the consultation is not significantly changed with use of DSIs. Whelan et al found no difference in physician satisfaction, or length of the initial consultation or follow-up visit when a decision board was administered by a nurse during the initial consultation for adjuvant chemotherapy for node-negative breast cancer.17 Similarly Butow et al also found no difference in physician satisfaction with the consult or length of the consultation with use of a question prompt list during the initial palliative care consultation.33 Interestingly, although Clayton et al found no difference in physician satisfaction with use of a question prompt list, the length of the consultation increased by approximately 7 minutes compared with controls (37.8 minutes for intervention versus 30.5 minutes for control, P = .002). Despite this, the participating physicians were not concerned about the increased length of the consultation, because they felt important issues were addressed as a result of the intervention.23

The main limitation of this study is the method that was used to select and classify the DSIs into the 2 subgroups: pDAs and other DSIs. However, we used the same method as the Cochrane Collaboration and only included pDAs that were evaluated relative to standard care, not alternative interventions. Because this led to the selection of higher quality pDAs, it is likely that our results overestimated the effectiveness of pDAs and biased the results toward pDAs. Therefore, it is unlikely that the overall or “big picture” finding that other DSIs such as question prompt lists and audiorecording of the consultation have similar reported outcomes as pDAs would change significantly. The other main limitation of the study is that even though 24 RCTs were included in this meta-analysis, the quality of the RCTs (ie, inadequate data) and degree of heterogeneity (ie, reporting of different outcomes) did not allow for all 24 studies to be included in each of the individual analyses. As a result, only 4 to 6 studies were included in each of the subgroup analyses, and this smaller sample size may lead to less stable estimates. Furthermore, it is possible that other important trials may have been missed in our search strategy; however, our search identified the same RCTs for pDAs for cancer treatment as the Cochrane Collaboration.

Conclusions

In summary, this study showed that the effectiveness of other DSIs, including question prompt lists and audiorecording of the consultation, is similar to pDAs. This is important because it may be that less complex DSIs such as question prompt lists and/or audiorecording may be all that is necessary to achieve similar outcomes as pDAs for cancer treatment. Furthermore, despite evidence suggesting that physician involvement with DSIs leads to improved patient outcomes, this study showed that the majority of DSIs evaluated in RCTs were not administered by the treating physician. Future studies should strongly consider administration of the DSI by the treating physician to facilitate shared decision-making, combining DSI design strategies and direct comparison of different DSI strategies using validated measures.

FUNDING SOURCES

This study was funded by the Canadian Institutes of Health Research, grant number 94129.

CONFLICT OF INTEREST DISCLOSURE

The authors made no disclosure.

Ancillary