Published Ahead of Print on February 17, 2016 as 10.1212/WNL.0000000000002434

How best to obtain consent to thrombolysis Individualized decision-making

Jingjing Gong, PhD Yan Zhang, PhD Jun Feng, BM Weiwei Zhang, MD Weimin Yin, MM Xinhuai Wu, BM Yanhong Hou, PhD Yonghua Huang, MD Hongyun Liu, PhD Danmin Miao, MP

Correspondence to Dr. Gong: [email protected] or Dr. Huang: [email protected] or Dr. Liu: [email protected] or D. Miao: [email protected]

ABSTRACT

Objective: To investigate the factors that influence the preferences of patients and their proxies concerning thrombolytic therapy and to determine how best to convey information.

Methods: A total of 613 participants were randomly assigned to a positively or negatively framed group. Each participant completed a series of surveys. We applied latent class analysis (LCA) to explore participants’ patterns of choices of thrombolysis and to classify the participants into different subgroups. Then we performed regression analyses to investigate predictors of classification of the participants into each subgroup and to establish a thrombolytic decision-making model. Results: LCA indicated an optimal 3-subgroup model comprising intermediate, favorable to thrombolysis, and aversion to thrombolysis subgroups. Multiple regression analysis revealed that 10 factors predicted assignment to the intermediate subgroup and 4 factors predicted assignment to the aversion to thrombolysis subgroup compared with the favorable to thrombolysis subgroup. The x2 tests indicated that the information presentation format and the context of thrombolysis influenced participants’ choices of thrombolysis and revealed a framing effect in different subgroups.

Conclusions: The preference for thrombolysis was influenced by the positive vs negative framing scenarios, the format of item presentation, the context of thrombolysis, and individual characteristics. Inconsistent results may be due to participant heterogeneity and the evaluation of limited factors in previous studies. Based on a decision model of thrombolysis, physicians should consider the effects of positive vs negative framing and should seek a neutral tone when presenting the facts, providing an important reference point for health persuasion in other clinical domains. Neurology® 2016;86:1–8 GLOSSARY AIS 5 acute ischemic stroke; IDM 5 individualized decision-making; LCA 5 latent class analysis; LRT 5 likelihood ratio test; MMSE 5 Mini-Mental State Examination; OR 5 odds ratio; SDM 5 shared decision-making.

Supplemental data at Neurology.org

IV thrombolytic therapy for acute ischemic stroke (AIS) can increase the overall odds of a good stroke outcome with acceptable safety.1–5 There has been reluctance to use thrombolysis because of a treatment delay beyond 4.5 hours and an increased risk of fatal intracranial hemorrhage.3 In North America, the rates of thrombolysis are extremely low (2.4%–5.2%).6 The most common reason for in-emergency-department delay is the time needed to obtain consent (43.24%)7 due to delay or refusal by patients or their proxies. It is crucial to help patients/proxies understand the advantages and disadvantages of thrombolysis and make an informed decision based on appropriate information. The mode of presentation of clinical trial results, or frame, may affect the perception of a treatment and the final decisions by physicians and patients.8 In combination with many other factors,9–11 the framing effect can influence choices for treatment, medical prevention, and screening.12 Studies of the influence of such effects on patients’ preferences for thrombolytic therapy and their medical decision-making during the consent process have been limited. In the present study, we attempted to explore (1) the preferences of patients and their proxies regarding thrombolytic therapy; (2) whether the framing effect, along with other factors, might influence From the Departments of Neurology (J.G., J.F., W.Z., W.Y., Y. Huang) and Radiology (X.W.), General Hospital of Beijing Command PLA; Centre of Psychology (Y.Z.), Air Force Aviation Medicine Research Institute; Department of Psychological Medicine (Y. Hou), 309 Hospital of PLA, Beijing; School of Psychology (H.L.), Beijing Normal University; and Department of Medical Psychology (D.M.), Fourth Military Medical University, Xi’an, China. Go to Neurology.org for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article. © 2016 American Academy of Neurology

ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

1

Table 1

Formats of presentation of information related to thrombolytic therapy

Contexts of thrombolysis Outside the context of thrombolysisa

Within the context of thrombolysisb

Item no.

Hospital no.

Information presentation format of thrombolysis

First presented therapy among the items

Description of parenchymal hemorrhage in pros and cons

1

1,002

Percentage

Nonthrombolytic therapy

No

2

1,003

Specific figures (20,000 persons as a base)

Nonthrombolytic therapy

No

3

1,006

Percentage

Nonthrombolytic therapy

Yes

4

1,007

Literal odds ratios

Thrombolytic therapy

Yes

5

1,008

Literal odds ratios

Nonthrombolytic therapy

Yes

6

1,009

Specific figures (20,000 as a base)

Nonthrombolytic therapy

Yes

7

4,003

Numeric odds ratios

Thrombolytic therapy

Yes

8

4,004

Numeric odds ratios

Nonthrombolytic therapy

Yes

9

4,005

Specific figures (1,000 persons as a base)

Nonthrombolytic therapy

Yes

10

9,001

Percentage

Nonthrombolytic therapy

Yes

11

9,002

Literal odds ratios

Thrombolytic therapy

Yes

12

9,003

Specific figures (20,000 as a base)

Nonthrombolytic therapy

Yes

13

9,004

Numeric odds ratios

Thrombolytic therapy

Yes

14

9005

Percentage

Nonthrombolytic therapy

Yes

a The participants were presented with no detailed thrombolysis introduction or description for the first 9 items, in which the option of thrombolytic therapy was not specified. b The participants were presented with a detailed thrombolysis introduction and description before making choices for items 10–14, in which the 2 available options were specified as “nonthrombolytic therapy” and “thrombolytic therapy.”

participants’ consent to thrombolytic therapy; and (3) how these influencing factors in combination affect the perceptual judgement of thrombolytic therapy and how to best convey this risk/benefit information. METHODS Participants. A total of 613 Chinese inpatients in the Department of Neurology or their next of kin were consecutively recruited between August 9, 2013, and September 16, 2014. The general eligibility criteria for the participants included (1) age $18 years; (2) normal cognition, as indicated by a Mini-Mental State Examination (MMSE; Chinese revised version) score of either .20 (for those with #6 years of education) or .24 (for those with .6 years of education); (3) the ability to communicate verbally and complete the questionnaires; (4) no history of severe mental disorders; (5) no disability (modified Rankin Scale score of 0–1); (6) no history of thrombolytic therapy; and (7) diagnosis via a brain MRI scan (old, lacunar, or acute cerebral infarction) if categorized into the group of stroke inpatients. The exclusion criteria included (1) cerebral haemorrhage, (2) medical work experience, and (3) failure to complete the assessment. Among the 613 participants, 46 were excluded (2 due to loss of the questionnaires, 3 due to age ,18 years, 7 due to a history of cerebral hemorrhage, 9 due to medical work experience, and 25 due to failure to complete the assessments). Detailed sociodemographic characteristics, health status, attitudes, and emotional state of the participants are presented in table e-1 on the Neurology® Web site at Neurology.org.

Materials. The assessments included instructions, a numeracy scale,13 14 items consisting of different presentation formats of information about thrombolysis in different contexts5 (table 1), sociodemographic, health, and attitude questionnaires, the 2

Neurology 86

SCL-90-R,14 and assessments of decision-making in other domains (appendix e-1).15,16

Study design and procedures. First, the participants were randomly assigned to the positively or negatively framed group the day after their hospitalization; the participants were not subjected to time pressure regarding the completion of the 14 items about thrombolysis and other factors. Second, regardless of the framing scenario, all participants’ responses to the 14 items were analyzed by latent class analysis (LCA), which is similar to cluster analytic methods17 and can be used to identify different subgroups of participants (latent classes) according to their item response patterns.18 Specifically, those participants with similar response patterns for the 14 thrombolytic items were categorized into the homogeneous subgroup, indicating that they reported similar attitudes toward thrombolysis. As many different classifications can be developed via LCA (e.g., subgroup model 1, subgroup model 2, subgroup model 3) (table 2), another goal of LCA was to identify the optimal model, which contained the smallest number of subgroups necessary to adequately describe the association of the choice of thrombolysis with the format of the items and the context of thrombolysis (tables 1 and 2). Third, we employed univariate and multivariate logistic regression analyses of sociodemographic data to identify factors (independent variables) that might predict the classification of the participants into different subgroups (dependent variables), establish a thrombolytic decision-making model, and reveal the combined influence and mutual relationships of the positive vs negative framing scenarios, the formats of items related to information about thrombolysis, the contexts of risk decisionmaking, and sociodemographic factors. After completing all of the surveys, the participants were informed that thrombolytic therapy was strongly recommended by neurologists to treat AIS when possible.

March 15, 2016

ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

Table 2

Model fit statistics for different subgroup models in the negative, the positive, and both framing scenarios

Frame type c

Negative

d

Positive

Bothe

No. subgroupsa

No. parametersb

Log likelihood

AIC

BIC

Adjusted BIC

Entropy

LRT

Ddf

p

1

14

22,370.578

4,769.156

4,819.325

4,774.937

—

—

—

—

2

29

22,191.242

4,440.484

4,544.405

4,452.459

0.831

354.440

15

0.0081

3

44

22,086.249

4,260.498

4,418.172

4,278.667

0.890

207.508

15

0.0000

4

59

22,028.340

4,174.681

4,386.107

4,199.044

0.903

114.451

15

0.0009

5

74

21,993.337

4,134.674

4,399.853

4,165.231

0.899

69.180

15

0.1809

1

14

22,583.325

5,194.649

5,246.549

5,202.149

—

—

—

—

2

29

22,431.230

4,920.460

5,027.967

4,935.995

0.828

300.676

15

0.0001

3

44

22,301.728

4,691.456

4,854.569

4,715.026

0.861

256.014

15

0.0041

4

59

22,239.530

4,597.061

4,815.780

4,628.666

0.834

122.959

15

0.3749

5

74

22,188.367

4,524.734

4,799.060

4,564.375

0.863

94.012

15

0.0913

1

14

24,984.404

9,996.808

10,057.573

10,013.130

—

—

—

—

2

29

24,677.351

9,412.703

9,538.573

9,446.512

0.768

607.716

15

0.0000

3

44

24,454.043

8,996.086

9,187.062

9,047.383

0.826

441.970

15

0.0185

4

59

24,333.981

8,785.963

9,042.044

8,854.746

0.851

237.625

15

0.2285

5

74

24,262.795

8,673.590

8,994.777

8,759.862

0.851

140.891

15

0.0860

Abbreviations: AIC 5 Akaike Information Criterion; BIC 5 Bayesian Information Criterion; LRT 5 Lo-Mendell-Rubin likelihood ratio test. a Many different classifications can be developed via LCA (e.g., subgroup model 1, subgroup model 2, subgroup model 3). b Concerning the fit measures (parsimony and goodness-of-fit), the model with fewer parameters (or subgroups), relatively lower BIC value and AIC value, and significant p value for LRT (,0.05), which should also be interpretable in medical practice, would be preferable to other models. c After LCA in the negative framing scenario, subgroup model 3 was preferred according to the fit measures. d After LCA in the positive framing scenario, subgroup 3 model was preferred according to the fit measures. e After LCA in the negative and positive framing scenarios, the participants in both negative and positive scenarios were combined and all their responses to items were again analyzed by LCA as a whole, and subgroup model 3 was optimal according to the fit measures, which also corresponded to clinical practice.

Statistical analysis. LCA was performed using Mplus 7.0,19 followed by univariate and multiple logistic regression analyses (stepwise regression) using SPSS 19.0 (SPSS, Chicago, IL). x2 Tests were performed to evaluate the differences in the rates of consent to thrombolysis between different framing scenarios and item formats. A significance level, from 0.05 to 0.00054, was set according to the partitioning of the x2 method (a’ 5 a ÷ [k (k21)/2 1 1], K 5 14).

Standard protocol approval, registration, and patient consent. This study was approved by the Academic Committee of the General Hospital of Beijing Command PLA. The ethics committee of this institution approved the experimental procedures and the written consent form. Each participant provided informed consent to participate in the experiment. RESULTS After randomization of positive or negative framing and completion of 14 items about thrombolysis, the classification of participants into subgroups was determined based on LCA of their response patterns for the 14 items assessed. In the LCA, latent subgroup models were fit to the data starting with the most parsimonious one-subgroup model (all participants classified into the same subgroup) followed by progression to less parsimonious models. The optimal number of subgroups of participants was determined by the Akaike Information Criterion and the Bayesian Information Criterion, as well as the Lo-MendellRubin likelihood ratio test (LRT) of the fitness of the model. The LRT, which compared the estimated

model with a model containing one fewer subgroup than the estimated model,20 demonstrated that a 3-subgroup model was optimal and easily interpretable in medical practice (table 2). The conditional probability of consent to thrombolysis in each item among the 3 subgroups was also calculated by LCA (figure). The 3 subgroups of participants were designated as favorable to thrombolysis, aversion to thrombolysis, and intermediate subgroups based on their patterns of choices regarding consent to thrombolysis (conditional probability of consent to thrombolysis), especially in the context of thrombolysis (from item 10 to item 14) (figure). Compared to the aversion to thrombolysis subgroup (36.7% of the sample), the participants in the favorable to thrombolysis subgroup (39.5%) exhibited overwhelming preferences for thrombolysis in the context of thrombolysis, but the intermediate subgroup (23.8%) showed a moderate probability of consent to thrombolytic therapy. For the class probability of subgroups in different framing scenarios, the constituent ratios differed between the negative and positive framing scenarios (intermediate subgroup/favorable to thrombolysis subgroup/aversion to thrombolysis subgroup: 37.1%/ 25.2%/37.7% vs 17.1%/48.8%/34.1%, x2 5 43.32, p , 0.001). This result indicated that substantially more participants preferred thrombolysis in the Neurology 86

March 15, 2016

ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

3

Figure

Conditional probability of consent to thrombolysisd for 14 items among 3 subgroups of participants

a

The participants were presented with no detailed thrombolysis introduction or description for the first 9 items, in which the option of thrombolytic therapy was not specified. bThe participants were presented with a detailed thrombolysis introduction and description before making choices for items 10–14, in which the 2 available options were specified as “nonthrombolytic therapy” and “thrombolytic therapy.” cThe subgroup classification was determined based on latent class analysis (LCA) after randomization of positive or negative framing and the completion of the 14 items. dThe conditional probability of consent to thrombolysis for each item was calculated via LCA, and high conditional probability indicated that the participants had more favorable attitudes toward thrombolysis, e.g., the participants in the favorable to thrombolysis subgroup in the context of thrombolysis.

positive framing scenario (48.8%) than in the negative framing scenario (25.2%). We performed univariate analyses of the association of variables with subgroups of participants; the factors nationality, smoking status, and stroke in relatives (p . 0.05) were excluded from the multivariate model (table e-2). Multiple logistic regression analysis demonstrated that compared to the favorable to thrombolysis subgroup (serving as the reference subgroup), the intermediate subgroup could be predicted by the following factors: being presented with a negative framing scenario, identities of stroke patients’ relatives, self-rating their health as poor (or intermediate), having an intermediate or less focus on health, having no history of hypertension or drinking, exhibiting an introverted personality, and not disagreeing with the statement that “One’s quality of life is more important than his or her lifespan” (table 3). Compared to the favorable to thrombolysis subgroup, the aversion to thrombolysis subgroup could be predicted by being presented with a negative framing scenario, being single (unmarried/divorced/widowed), having no hypertension, and reporting an intermediate opinion about whether “One’s quality of life is more important than his or her lifespan” (table 3). The pseudo-R2 value (0.460) indicated that the factors in the model mentioned above accounted for 46% of 4

Neurology 86

the variation in the prediction of the subgroup classification of all participants, suggesting fine goodness-of-fit of the model. The estimated probabilities of classification into different subgroups of participants could be calculated according to our thrombolytic decision-making model and the corresponding probability equations. In addition, x2 tests indicated that the format of the presentation of items and the context of thrombolysis influenced the participants’ choices of thrombolysis (table e-3). There were effects of the positive vs negative framing scenarios among different subgroups; in most cases, the rates of consent to thrombolysis were higher in the positive framing scenario than in the negative framing scenario (table 4). DISCUSSION Informed consent to thrombolytic therapy, while not always required if time is pressing, can have profound consequences on patient and proxy decisionmaking. In particular, the presentation of the risks and benefits can be framed positively or negatively, but there is little evidence to guide physicians in effectively sharing these data with patients and proxies. Although studies of the effect of framing on decision-making in other medical domains are accumulating, reliable evidence to guide the communication between clinicians and patients remains insufficient due to the generation of mixed or contradictory results.21 First, differences among individuals or groups of

March 15, 2016

ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

Table 3

Multivariate analysis of sociodemographic, health status, and attitude predictors for the classification of participants into subgroups Aversion to thrombolysis subgroup: Favorable to thrombolysis subgroupa

Intermediate subgroup: Favorable to thrombolysis subgroupa Variables Frame type

Category

B

Negative framing

SE 1.541

0.393 4.668 (2.163–10.076)

Positive framing Marital status

Unmarried/divorced/ widowed

0.864

0.734 2.372 (0.563–9.995)

Stroke patients

1.022

Stroke patients’ relatives

1.493

Nonstroke patients

,0.001

SE

0.581 0.256 1.787 (1.083–2.949)

p 0.023

1

0.578 2.778 (0.894–8.632)

0.077

0.184 0.382 1.202 (0.569–2.541)

0.630

0.559 4.450 (1.488–13.313)

0.008

0.043 0.400 1.044 (0.477–2.288)

0.914

0.605 0.631 (0.193–2.065)

0.446

0.343 0.363 1.410 (0.692–2.873)

0.345

1

20.461

OR (95% CI)b

0.955 0.378 2.598 (1.240–5.444)

0.011

1

1

1

Poor (very poor or poor)

2.963

1.182 19.357 (1.908–196.407)

0.012

0.229 0.392 1.257 (0.583–2.709)

0.559

Intermediate

3.106

1.095 22.329 (2.610–190.994)

0.005

0.050 0.325 1.051 (0.556–1.986)

0.878

Good (good or best) Focus on health

B

0.239

Nonstroke patients’ relatives Health self-rating

p

1

Married/living with partner Participant types

OR (95% CI)b

1

1

Less (not at all or less)

1.986

0.842 7.287 (1.400–37.923)

0.018 20.244 0.329 0.783 (0.411–1.492)

0.458

Intermediate

2.361

0.805 10.603 (2.187–51.409)

0.003 20.552 0.295 0.576 (0.323–1.026)

0.061

More (more or extremely)

1

1

Anamnesis Hypertension

No

2.203

0.462 9.052 (3.658–22.400)

Yes Drinking

No

1.639

0.582 5.152 (1.646–16.128)

1.823

0.419 6.188 (2.721–14.071)

Yes Personality

0.576 0.269 1.780 (1.050–3.018)

0.032

1 0.005 20.072 0.271 0.930 (0.547–1.581)

1

Introverted Extroverted

Attitudes toward quality of lifec

,0.001

1

0.790

1 ,0.001 0.185

0.251 1.204 (0.736–1.970)

1

0.460

1

Strongly disagree or disagree

22.432

1.096 0.088 (0.010–0.753)

0.027 20.619

0.380 0.538 (0.256–1.133)

0.103

Intermediate

21.514

0.846 0.220 (0.042–1.156)

0.074 1.203

0.418 3.331 (1.469–7.555)

0.004

Strongly agree or agree

1 210.829 1.605

Constant Pseudo R2d

0.460

Model pe

,0.001

1 ,0.001 20.830

0.537

0.122

Abbreviations: B 5 nonstandardized coefficient B; CI 5 confidence interval; OR 5 odds ratio. a The favorable to thrombolysis subgroup acted as the reference subgroup. b Multiple logistic regression models were generated using backward stepwise selection. c Your attitudes toward the statement “One’s quality of life is more important than his or her lifespan.” d Pseudo R2 was an index of goodness-of-fit of a model, and the pseudo R2 value (0.460) indicated that the factors in the model accounted for 46% of the variation in the prediction of assignment of subgroups, suggesting fine goodness-of-fit of the model. e Model p value indicated the significance of the model (p , 0.05).

participants play important roles in the susceptibility to framing,22 but much greater attention has been paid to differences in the demographic characteristics of participants23,24 than to differences in the patterns of their responses to clinical decisions. Second, there is a widely held belief that medical decision-making is influenced by many internal and external factors, such as differences in samples, information formats,25 and message framing,12 but investigation of the combined effects and mutual relationships of

these influencing factors is rare, possibly due to methodologic limitations. The present study applies LCA to analyze consent to thrombolysis, which is rarely reported. LCA assumes that the population from which a sample is drawn comprises underlying subgroups of individuals who can be grouped according to their response patterns.17 Our 3-subgroup model, which included intermediate, favorable to thrombolysis, and aversion to thrombolysis subgroups, revealed the evident Neurology 86

March 15, 2016

ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

5

Table 4

Rates of consent to thrombolysis in the negative and positive framing scenarios among the different subgroups of participants Intermediate subgroup (n 5 135)

Favorable to thrombolysis subgroup (n 5 224)

Aversion to thrombolysis subgroup (n 5 208) Neg. (107)

Total (n 5 567)

Item no.

Hospital no.

Neg. (76)

Pos. (59)

p

Neg. (83)

Pos. (141)

p

1

1,002

100.00

100.00

—

62.20

68.15

0.370

50.49

63.27

0.068

68.46

72.95

0.2473

2

1,003

100.00

100.00

—

51.22

70.68

0.004a

46.60

49.49

0.681

63.60

69.42

0.1480

3

1,006

85.33

89.83

0.437

44.44

47.76

0.637

21.43

30.30

0.155

47.64

50.34

0.5283

49.42

65.98

,0.001a

49.03

47.90

0.7934

Pos. (101)

p

Neg. (266)

a

Pos. (301)

p

4

1,007

18.67

15.79

0.666

70.73

79.10

0.162

55.00

77.00

0.001

5

1,008

86.49

83.33

0.620

40.96

42.11

0.869

28.00

36.36

0.207

6

1,009

92.00

93.10

0.811

35.37

47.48

0.079

13.73

30.69

0.004a

43.24

50.67

0.0799

a

62.45

74.58

0.0022a

48.58

34.74

0.0012a

27.34

25.76

0.6749

a

7

4,003

90.41

96.55

0.168

59.76

75.91

0.012

43.88

60.00

0.023

8

4,004

69.44

64.91

0.585

40.00

20.45

0.002a

40.00

36.46

0.615 a

9

4,005

10.81

5.26

0.256

33.33

37.96

0.493

34.65

20.79

0.028

10

9,001

74.65

77.97

0.658

91.57

93.43

0.606

38.24

27.72

0.111

65.63

68.01

0.5517

11

9,002

40.85

48.28

0.398

97.59

97.81

0.916

64.71

55.45

0.178

68.75

73.65

0.2042

12

9,003

69.01

72.41

0.673

97.59

95.62

0.449

13.73

13.86

0.978

56.25

63.18

0.0977

a

56.86

71.96

,0.001a

64.84

63.27

0.7000

13

9,004

34.29

37.93

0.669

97.59

97.81

0.916

39.22

56.44

0.014

14

9,005

80.28

77.19

0.670

98.80

94.12

0.091

26.47

13.86

0.025a

Abbreviations: Neg. 5 negative framing scenario; Pos. 5 positive framing scenario. a Significant.

heterogeneity of the participants in terms of their medical decision-making. Regression analysis revealed the comprehensive influence of individual sociodemographic factors, participants’ values and preferences, and the framing and format of information on thrombolytic decision-making. These findings permitted us to draw relatively systematic and unbiased conclusions. The identification of different influencing factors in decision-making in the intermediate, favorable, and averse subgroups may partially explain the previous inconsistent results regarding medical decision-making. Our data confirm that participants’ choices of thrombolysis are regulated by the influence of positive vs negative framing scenarios. Consent to thrombolysis is more prevalent in the positive framing scenario than in the negative framing scenario under certain circumstances. This finding is consistent with the conclusion that the framing effect varies by the type of health question (such as screening, prevention, or therapy)26,27 and that participants are more likely to accept a treatment (e.g., surgery) when described positively than when described negatively.16,28 The framing effect has also been hypothesized to be affected by the type of framing structure,29 and some researchers have proposed a taxonomy classifying the framing effect into 3 categories: risky choice framing, goal framing, and attribute framing.27,30 Attribute framing, which affects the encoding and evaluation of the characteristics of an object or an event,30 refers to the positive vs negative description of a specific 6

Neurology 86

attribute of a single item or state27; for example, “The survival rate of thrombolysis is 89.9%” vs “The mortality rate of thrombolysis is 10.1%.” Attribute framing was applied in our framing messages. In attribute framing, attributes are judged more favorably when described in the positive scenario than the negative scenario29; this observation is consistent with our results. Our data indicate that framing effects on consent to thrombolysis (following distinct descriptions according to attribute framing) are not consistent. This result suggests that this phenomenon may not be simply explained by the above 2 hypotheses. In brief, we suggest that the framing effect is a pooled consequence modulated by various factors. Our findings also demonstrate a substantial influence of the format of presentation,31 in combination with the framing effect and the introduction and description of thrombolysis, on thrombolytic treatment decision-making. Such an influence is inconsistent with the previous claim that problems in communicating risks occur not because of cognitive biases (framing effect) but rather because of inappropriate information formats.32 In other words, we suggest that all of these factors mutually affect the decision-making process. The odds ratios (OR) formats of presentation, including literal and numeric ORs, indicate the optimal choices for obtaining consent to thrombolysis. Additionally, the decision to consent to thrombolysis is influenced by the order of the presentation of therapies among the items being presented, particularly in the positive framing scenario. Moreover, consent to

March 15, 2016

ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

thrombolysis is lowest when the item is presented as a small-base-specific figure (1,000 as a base), which might be avoided in the process of describing treatment to patients. Contrary to prior reports, many common sociodemographic factors, such as numeracy,9 sex,33 and age, were eliminated from the decision model. We speculate that factors with a limited influence on decision-making are assumed to be involved in designs of experiments, resulting in the inclusion of these factors. When more influential factors are considered and input into the model, these nonessential factors are eliminated from the model. Our data imply that age, sex, and numeracy did not contribute significantly to the decision-making process in our participants. This novel finding requires replication in other datasets. Other reasons for this inconsistency may include differences in race, ethnicity, and culture among the patient samples; hence, it will be necessary to replicate these results in other samples. Our results, if confirmed, could significantly contribute to clinical care in several ways. (1) Our findings can guide clinicians concerning how best to obtain consent to thrombolysis. A standard process for obtaining informed consent to thrombolysis and a standard content of such consent requests are lacking in many countries. These observations imply that the process of obtaining informed consent to thrombolysis is arbitrary and subjective in clinical practice. If the clinician wishes to obtain consent prior to thrombolysis, even though this is not required in all cases, the physician should consider the effects of positive vs negative framing and should seek a neutral tone when presenting the facts. Both the benefits and the risks of thrombolytic therapy are of great importance to doctors and patients. Our decision model can reduce such uncertainties by providing theoretical and practical bases for a standardized clinical pathway for thrombolytic decisionmaking. (2) Our framing messages and information formats were derived from large clinical studies.5 Thus, these findings build a bridge to the outcomes of clinical studies that directly examined patient care. (3) The establishment of a thrombolytic decision model facilitates individualized decision-making (IDM). (4) The combination of LCA and regression analysis provides important reference points for the establishment of decision models in other clinical domains, such as the prevention and detection of diseases. Advocacy of shared decision-making (SDM), a patient-centered process of collaboration between clinicians and patients in which common decisions are based on the best available evidence regarding the risks and benefits of all available options as well as on the patient’s values and preference, is increasing.34 Some investigators argue against SDM because not all patients wish to be told about their illness or to make their own decisions, and the stress, panic, and

anxiety caused by severe diseases can render some individuals unfit to make rational decisions.35,36 Furthermore, some researchers argue for authoritarian decision-making (ADM) by clinicians. For instance, when Franz Ingelfinger, a past Editor of the New England Journal of Medicine, was diagnosed with cancer and was in pain, he became frustrated with being presented with options and desired ADM rather than SDM.36 Clinicians should take greater responsibility as medical experts; patients should be fully respected as decision-makers; and consent must be obtained before performing certain risky treatments that may become the focus of conflicts between clinicians and patients. IDM might be the key to solving such problems; this hypothesis will require further investigation to validate our findings in future studies. To more objectively assess the participants’ cognition, we should use a more professional instrument than the MMSE, whose scores are mainly reflective of severity of dementia in clinical practice. The exclusion of participants with medical work experience may cause bias to some extent. Certain questions in the surveys should be presented with clear definitions to avoid ambiguity, especially for the questions related to attitudes toward life. For instance, from the clinician’s or patient’s perspective, quality of life is the fundamental motivation to undertake any form of intervention, but some people may have a different understanding of the meaning of quality of life, as this term may substantially reflect the difference between expectation and reality rather than an objective impact as observed by an external evaluator. Future research should recruit more participants with various backgrounds, utilize optimized items and questions, and be performed in a clinical context. AUTHOR CONTRIBUTIONS Jingjing Gong and Yonghua Huang designed the study and performed the experiments. Hongyun Liu, Jingjing Gong, and Danmin Miao designed the study and analyzed the data. Yan Zhang searched the literature, collected the data, and wrote the manuscript. Jun Feng, Weiwei Zhang, Weimin Yin, and Xinhuai Wu performed the experiments and collected and interpreted the data. Yanhong Hou prepared the psychological assessment/materials/analysis tools and revised the manuscript. Jingjing Gong supervised the study.

STUDY FUNDING Supported by the Beijing Natural Science Foundation (grants 7153176 and 7123230) and the National Natural Science Foundation of China (grants 31000461 and 81171100).

DISCLOSURE The authors report no disclosures relevant to the manuscript. Go to Neurology.org for full disclosures.

Received May 20, 2015. Accepted in final form October 7, 2015. REFERENCES 1. Hacke W, Kaste M, Fieschi C, et al. Intravenous thrombolysis with recombinant tissue plasminogen activator for acute Neurology 86

March 15, 2016

ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

7

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15. 16. 17.

8

hemispheric stroke: The European Cooperative Acute Stroke Study (ECASS). JAMA 1995;274:1017–1025. Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke: The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. N Engl J Med 1995;333:1581–1587. Emberson J, Lees KR, Lyden P, et al. Effect of treatment delay, age, and stroke severity on the effects of intravenous thrombolysis with alteplase for acute ischaemic stroke: a meta-analysis of individual patient data from randomised trials. Lancet 2014;384:1929–1935. Hacke W, Kaste M, Bluhmki E, et al. Thrombolysis with alteplase 3 to 4.5 hours after acute ischemic stroke. N Engl J Med 2008;359:1317–1329. Lees KR, Bluhmki E, von Kummer R, et al. Time to treatment with intravenous alteplase and outcome in stroke: an updated pooled analysis of ECASS, ATLANTIS, NINDS, and EPITHET trials. Lancet 2010;375:1695–1703. El Khoury R, Jung R, Nanda A, et al. Overview of key factors in improving access to acute stroke care. Neurology 2012;79:S26–S34. Wang Y, Liao X, Zhao X, et al. Using recombinant tissue plasminogen activator to treat acute ischemic stroke in China: analysis of the results from the Chinese National Stroke Registry (CNSR). Stroke 2011;42:1658–1664. Jasper JD, Goel R, Einarson A, Gallo M, Koren G. Effects of framing on teratogenic risk perception in pregnant women. Lancet 2001;358:1237–1238. Reyna VF, Nelson WL, Han PK, Dieckmann NF. How numeracy influences risk comprehension and medical decision making. Psychol Bull 2009;135:943–973. Frith CD, Singer T. The role of social cognition in decision making. Philos Trans R Soc Lond B Biol Sci 2008; 363:3875–3886. Moxey A, O’Connell D, McGettigan P, Henry D. Describing treatment effects to patients. J Gen Intern Med 2003;18:948–959. Gong J, Zhang Y, Yang Z, Huang Y, Feng J, Zhang W. The framing effect in medical decision-making: a review of the literature. Psychol Health Med 2013;18:645–653. Lipkus IM, Samsa G, Rimer BK. General performance on a numeracy scale among highly educated samples. Med Decis Making 2001;21:37–44. Jalenques I, Galland F, Malet L, et al. Quality of life in adults with Gilles de la Tourette syndrome. BMC Psychiatry 2012;12:109. Tversky A, Kahneman D. The framing of decisions and the psychology of choice. Science 1981;211:453–458. Haward MF, Murphy RO, Lorenz JM. Message framing and perinatal decisions. Pediatrics 2008;122:109–118. Wong YN, Egleston BL, Sachdeva K, et al. Cancer patients’ trade-offs among efficacy, toxicity, and out-of-pocket cost in the curative and noncurative setting. Med Care 2013;51: 838–845.

Neurology 86

18.

19.

20. 21.

22.

23.

24.

25.

26.

27.

28.

29.

30. 31.

32.

33. 34.

35. 36.

Sutfin EL, Reboussin BA, McCoy TP, Wolfson M. Are college student smokers really a homogeneous group? A latent class analysis of college student smokers. Nicotine Tob Res 2009;11:444–454. McCrae R, Costa T Jr, Muthén LK, Muthén BO. Mplus User’s Guide, Seventh Edition (1998–2012). Los Angeles: Muthén & Muthén; 2012. Lo Y, Mendell NR, Rubin DB. Testing the number of components in a normal mixture. Biometrika 2001;88:767–778. Garcia-Retamero R, Galesic M. How to reduce the effect of framing on messages about health. J Gen Intern Med 2010;25:1323–1329. Latimer AE, Rivers SE, Rench TA, et al. A field experiment testing the utility of regulatory fit messages for promoting physical activity. J Exp Soc Psychol 2008;44:826–832. Apanovitch AM, McCarthy D, Salovey P. Using message framing to motivate HIV testing among low-income, ethnic minority women. Health Psychol 2003;22:60–67. Schneider S, Levin I, Gaeth G. The three faces of framing: a typology of framing effects. Presented at the 36th annual meeting of the Psychonomic Society. Los Angeles; 1995. Garcia-Retamero R, Galesic M. Using plausible group sizes to communicate information about medical risks. Patient Educ Couns 2011;84:245–250. Rothman AJ, Salovey P. Shaping perceptions to motivate healthy behavior: the role of message framing. Psychol Bull 1997;121:3–19. Akl EA, Oxman AD, Herrin J, et al. Framing of health information messages. Cochrane Database Syst 2011: CD006777. McNeil BJ, Pauker SG, Sox HC Jr, Tversky A. On the elicitation of preferences for alternative therapies. N Engl J Med 1982;306:1259–1262. Levin IP, Schneider SL, Gaeth GJ. All frames are not created equal: a typology and critical analysis of framing effects. Organ Behav Hum Decis Process 1998;76:149–188. Wang XT. Framing effects: dynamics and task domains. Organ Behav Hum Decis Process 1996;68:145–157. Almashat S, Ayotte B, Edelstein B, Margrett J. Framing effect debiasing in medical decision making. Patient Educ Couns 2008;71:102–107. Gigerenzer G, Gaissmaier W, Kurz-Milcke E, Schwartz LM, Woloshin S. Helping doctors and patients make sense of health statistics. Psychol Sci Public Interest 2007;8:53–96. Huang Y, Wang L. Sex differences in framing effects across task domain. Pers Individ Dif 2010;48:649–653. Charles C, Gafni A, Whelan T. Decision-making in the physician-patient encounter: revisiting the shared treatment decision-making model. Soc Sci Med 1999;49: 651–661. Schattner A, Tal M. Truth telling and patient autonomy: the patient’s point of view. Am J Med 2002;113:66–69. Schattner A. Shared decision making: an alternative view. Mayo Clinic Proc 2014;89:276.

March 15, 2016

ª 2016 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.

How best to obtain consent to thrombolysis: Individualized decision-making Jingjing Gong, Yan Zhang, Jun Feng, et al. Neurology published online February 17, 2016 DOI 10.1212/WNL.0000000000002434 This information is current as of February 17, 2016 Updated Information & Services

including high resolution figures, can be found at: http://www.neurology.org/content/early/2016/02/17/WNL.0000000000 002434.full.html

Supplementary Material

Supplementary material can be found at: http://www.neurology.org/content/suppl/2016/02/18/WNL.000000000 0002434.DC1.html

Subspecialty Collections

This article, along with others on similar topics, appears in the following collection(s): All epidemiology http://www.neurology.org//cgi/collection/all_epidemiology Decision analysis http://www.neurology.org//cgi/collection/decision_analysis Infarction http://www.neurology.org//cgi/collection/infarction

Permissions & Licensing

Information about reproducing this article in parts (figures,tables) or in its entirety can be found online at: http://www.neurology.org/misc/about.xhtml#permissions

Reprints

Information about ordering reprints can be found online: http://www.neurology.org/misc/addir.xhtml#reprintsus

Neurology ® is the official journal of the American Academy of Neurology. Published continuously since 1951, it is now a weekly with 48 issues per year. Copyright © 2016 American Academy of Neurology. All rights reserved. Print ISSN: 0028-3878. Online ISSN: 1526-632X.