Intravenous infusion lasting 24 h
Intravenous infusion lasting 12 h
A Bayesian D-efficient design consisting of 2-unlabeled alternative-forced-choice tasks was constructed for the DCE using Ngene (ChoiceMetrics, Sydney, Australia). 16 , 51 A total of 36 unique choice tasks were generated, which were divided over three 12-choice task blocks. Respondents were randomly assigned to complete 1 of those blocks. Attribute prior information for DCE design optimization was generated using previously published literature and best guesses. The survey was pilot tested among respondents in Italy ( N = 50), with the outcomes of a conditional logit model used to inform the final experimental design. Interactions between the attributes “5-y survival” and, respectively, “Risk of long-lasting skin problems,”“Risk of extreme tiredness,” and “Mode of administration” were accounted for in this design. An example of a DCE choice task can be found in Figure 1 .
Illustration of survey elements. (A) Swing weighting (SW) ranking task in which respondents sort the swings in attributes from worst level to best level by priority for improvement in a treatment. (B) SW direct rating task in which patients rate the swings relative on a scale from 0 (not at all important) to 100 (as important as the most important improvement). (C) Discrete choice experiment choice task in which respondents choose their most preferred treatment (pop-up shown to explain risk attribute).
An SW-DR task was developed using the same attributes and levels used in the DCE. In the SW section, respondents were first asked to choose the attribute they preferred to swing from the lowest (worst) to the highest (best) level. Respondents were asked to rank all other swings subsequently from most to least preferred (see Figure 1A ). The order in which the swings were presented was randomized in this section. In the DR section, respondents were asked to rate each of the swings relative to the others by giving it between 0 and 100 points, except for the highest ranked swing, which automatically received 100 points 25 (see Figure 1B ). This reflects the relative valuation of the importance of the different swings. Respondents were instructed on what this relative rating means as follows: “If you give 50 points to improve a feature, it means that you think improving it is half as important as improving the top ranked attribute because you gave it half as many points.” This unrestricted valuation is assumed to be simpler for respondents than PA from a fixed pool and has been found to be more reliable than restricted PA methods, 52 – 55 making it more suitable for this study population, who may have more fragile physical states or diminished cognitive status. 41 – 44
Both the DCE and SW-DR tasks were included as parts of a one-time online survey with respondents able to pause and return to the survey. The survey was programmed in Sawtooth software (lighthouse studio 9.13) and consisted of 6 parts. First, respondents were informed about the study and provided consent for data collection prior to answering sociodemographic and medical history–related questions. Second, respondents watched 2 different educational videos consisting of text and animations with voiceovers giving 1) an introduction with information on lung cancer and detailed descriptions of the attributes and levels included and (2) instructions on how to complete the first-choice task. Third, respondents were randomly assigned to receive either the DCE or SW task first to avoid any ordering effects. Fourth, respondents completed quality of life–related questions (EQ-5D). 56 , 57 Fifth, respondents watched a video with instructions on how to complete the second choice task. Finally, respondents were asked to complete psychosocial measures including measures of health literacy. 58 , 59
After each choice task, respondents were given 2 feedback questions about ease of understanding and answering the choice tasks on a 5-point Likert-type scale ranging from very easy to very difficult. The survey was pretested with 5 lung cancers patients in think-aloud interviews.
Contrary to common practice in applied preference research, in this study, only surveys of respondents who completed both DCE and SW choice tasks were included in the analysis to facilitate within-person comparisons. One respondent was excluded from the data set due to flatlining behavior (defined as always choosing A or always choosing B). Statistical analysis was performed with Nlogit version 6 and R version 4.0.4. A significance level of P < 0.05 was used for all analyses. All analyses were performed separately for data from Italy and Belgium to ensure most accurate methods comparison measures and avoid conflating potential scale heterogeneity between countries. 60
Respondent background characteristics (including general demographic and medical history information) were categorized and are presented as counts with percentages.
Random parameter logit models (RPLs) were used to analyze the DCE data. Such models adjust for the fact that panel data were collected and adjusted for the multilevel structure of the data. 9 , 13 In addition, these models allow to include attribute (levels) as random parameters to adjust for the effect of preference heterogeneity. 9 , 13 All risk and benefit attributes were assumed to be linear, and the categorical attributes were dummy coded. The significance level of the standard deviation of the attributes was used to test which attributes should be included in the final model as random parameters (assuming normal distributions) to account for preference heterogeneity. The utility equation below formulates the outcomes of these procedures and displays the final utility model tested in the analysis. The systematic utility component (V) describes the measurable utility of a specific treatment based on the attributes included in the DCE. The β 1 –β 7 coefficients represent the attribute-level estimates indicating the relative importance of each attribute level for individual i . A constant term was included in model exploration (i.e., to test for reading order bias), but it was found to be insignificant and removed from the final model.
A choice task-order variable was included in the model as an interaction term with the attribute levels to test whether the task order (i.e., DCE first or SW first) influenced the outcomes, which turned out insignificant. Prespecified interaction terms that significantly contributed to model fit (as assessed using a log-likelihood [LL] ratio test) were included in the model. Individual specific conditional parameter estimates were estimated for each respondent using the final model. Individual attribute weights and rankings were calculated with these parameter estimates (by calculating the total impact of each attribute on utility and standardizing to a total of 100, where the highest weight represents highest rank) and averaged to estimate the mean population weights and rankings.
The SW analysis was performed by analyzing the patients’ rankings of the attributes and the points allocated to the different attributes. The individual attribute relative importance weights were calculated using both the rank-ordered centroid (ROC) weight method and the DR weight method per patient. The ROC weight method calculates a relative weight representing the distance between adjacent ranks on an ordinal or normalized scale. 61
The ROC weight for an attribute with rank i equals (in case of 5 attributes):
The DR method is used to generate individual proportional weights for an attribute with rank i and allocated points p i and equals (in case of 5 attributes):
These individual weights were averaged over all patients per country to obtain the average weights, which are the equivalent of the attribute relative importance weights resulting from DCEs.
Respondent feedback.
Frequencies and chi-square tests were conducted to compare the feedback of respondents regarding their perceived difficulty in understanding and answering the DCE and SW questions.
Based on the outcomes of the RPL of the DCE and the SW-ROC, attribute ranking was compared. Ranking agreement (based on individual-level estimates from the DCE and SW-ROC) was evaluated with Cohen’s weighted kappa, which measures interrater reliability while accounting for chance similarities in rating. 62 , 63 Differences in the ranking based on DCE and SW-ROC were analyzed and tested with an ordered logit model. 64
Based on the outcomes of the RPL of the DCE and the SW-DR, attribute weighting was compared. Differences in the weighting based on DCE and SW-DR were analyzed and tested using Dirichlet regression models. 65 Dirichlet regression models can be regarded as a generalization of beta regression models for proportions and percentages and are particularly suited for the analysis of compositional data (i.e., for weights that add up to 1). 66 In a Dirichlet regression model, the aggregate attribute weights are assumed to be distributed with a Dirichlet distribution with parameters µ i , i = 1,…, 5, mean attribute weights that add up to 1, and a precision parameter ϕ (according to the so-called alternative parametrization). 67 The mean attribute weights are modeled with a logit link function similar to logistic regression:
Here, the logit of μ for individual i is equal to the linear predictor η and is modeled with an intercept β 0 , i , representing the DCE, and with a dummy variable D SW for the method as covariate. We defined the attribute 5-y survival as the base category, with β 0 , survival = β 1 , survival = 0 . In this way, the corresponding values of µ i equal
The precision parameter is modelled with a log link function with method as covariate:
The parameter estimates β 1,i can be interpreted as odds ratios after exponentiation relative to survival as base category. 66 Maximum likelihood estimation is used for obtaining the parameter estimates. 68 Finally, covariates were added to the models to correct for possible effects of method, for educational level, health literacy, gender, age, cancer stage, and treatment history.
A sample of n = 307 NSCLC patients was obtained from N = 560 requests to patients ( n = 159 declined invite; n = 94 withdrew consent). No significant differences were found between the countries in respondents’ gender, age, cancer stage, or family history of cancer. Respondents in both countries differed significantly in family and relationship status, χ 2 (3) = 8.1, P = 0.045; education level, χ 2 (2) = 7.248, P = 0.027; and health literacy, t (305) = −6.591, P < 0.001. Patient demographic information can be found in Table 2 .
Demographic Characteristics of the Sample
Italy ( = 158) | Belgium ( = 149) | ||||
---|---|---|---|---|---|
% | % | ||||
Sex | Male | 88 | 55.7 | 89 | 59.7 |
Age at survey completion, y | ≥71 | 56 | 35.4 | 45 | 30.2 |
Education | No degree | 0 | 0 | 6 | 4 |
Primary school | 12 | 7.6 | 6 | 4 | |
Middle school | 37 | 23.4 | 30 | 20.1 | |
Secondary school | 51 | 32.3 | 51 | 34.2 | |
Professional degree | 19 | 12 | 26 | 17.4 | |
Bachelor’s degree | 4 | 2.5 | 0 | 0 | |
Master’s degree | 26 | 16.5 | 14 | 9.4 | |
Postgraduate degree | 5 | 3.2 | 2 | 1.3 | |
Other | 4 | 2.5 | 14 | 9.4 | |
Family and relationship status | Single no children | 15 | 9.5 | 18 | 12.1 |
Single with children | 12 | 7.6 | 17 | 11.4 | |
Partner with children | 64 | 40.5 | 38 | 25.5 | |
Partner no children | 67 | 42.4 | 76 | 51 | |
Family history of lung cancer | Yes | 45 | 28.5 | 37 | 24.8 |
Cancer stage | I, II | 78 | 49.4 | 65 | 43.6 |
III, IV | 80 | 50.6 | 84 | 56.4 | |
Type of treatment | No treatments | 21 | 13.3 | 0 | 0 |
Surgery | 94 | 59.5 | 78 | 52.3 | |
Chemotherapy | 55 | 34.8 | 88 | 59.1 | |
Immunotherapy | 35 | 22.2 | 78 | 52.3 | |
Radiotherapy | 35 | 22.2 | 46 | 30.9 | |
Other | 18 | 11.4 | 12 | 8.1 | |
Don’t know | 3 | 1.9 | 0 | 0 | |
Lines of treatment | No treatment | 72 | 45.6 | 72 | 48.3 |
1 treatment | 34 | 21.5 | 14 | 9.4 | |
2 treatments | 14 | 8.9 | 15 | 10.1 | |
3 treatments | 17 | 10.8 | 48 | 32.2 | |
>3 treatments | 21 | 13.3 | 0 | 0 | |
Age when diagnosed, y | <55 | 28 | 17.7 | 21 | 14.1 |
55–64 | 48 | 30.4 | 57 | 38.2 | |
65–74 | 57 | 36.1 | 59 | 39.6 | |
≥75 | 25 | 15.8 | 12 | 8.1 | |
Health literacy (newest vital sign) | Very limited literacy | 7 | 4.4 | 12 | 8.1 |
Limited literacy | 40 | 25.3 | 32 | 21.5 | |
Adequate literacy | 111 | 70.3 | 105 | 70.5 |
Most respondents found the DCE and SW tasks very easy or easy to understand and answer (74.6% and 64.5% for DCE and 73.0% and 69.7% for SW, respectively, in Italy and Belgium). The ease of understanding and answering the DCE and understanding the SW task was associated with educational level, with those who had higher levels of education reporting greater ease ( P < 0.001).
Table 3 shows attribute ranks for the 2 methods separately per country ( Appendix Tables A1 and A2 show the original attribute-level estimates of the DCE and the ROC and DR estimates of the SW that were used for these calculations). Five-year survival was the most important attribute for most of the respondents, irrespective of the method. Agreement between the ranking of the DCE and SW-ROC was moderate with weighted Kappa correlation coefficients varying between 0.53 and 0.55. Despite the similar ranking of the 5-y survival and tiredness attributes, the overall ranking of the attributes differed significantly between DCE and SW-ROC tasks for both countries (χ 2 = 2042.9, 4 df, P < .0001 for Italy; χ 2 = 1932.5, 4 df, P < .0001 for Belgium; Table 4 ). For the Italian respondents, the attributes of mode and hair swapped their rank (third or fifth) depending on the method. For the Belgian respondents, the attributes of mode, skin problems, and hair changed ranking between being third, fourth, or fifth most important.
Attribute Rank and Weight (95% Confidence Interval) Based on DCE and SW-DR and Rank and Weight ( s ) for SW-ROC Separately for Italy and Belgium
Italy | Belgium | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
DCE | SW-DR | SW-ROC | DCE | SW-DR | SW-ROC | |||||||
Rank | Weight (95% CI) | Rank | Weight (95% CI) | Rank | Weight ( ) | Rank | Weight (95% CI) | Rank | Weight (95% CI) | Rank | Weight ( ) | |
Mode of administration | 5 | 0.05 (0.04–0.06) | 3 | 0.18 (0.15 – 0.19) | 3 | 0.16 (0.12) | 5 | 0.02 (0.02–0.02) | 4 | 0.16 (0.14–0.17) | 3 | 0.14 (0.12) |
5-y survival | 1 | 0.63 (0.61–0.66) | 1 | 0.33 (0.31–0.34) | 1 | 0.43 (0.08) | 1 | 0.59 (0.57–0.62) | 1 | 0.31 (0.30–0.33) | 1 | 0.42 (0.10) |
Risk of long-lasting skin problems | 4 | 0.08 (0.07–0.08) | 4 | 0.16 (0.15–0.17) | 4 | 0.14 (0.07) | 4 | 0.08 (0.08–0.09) | 3 | 0.18 (0.17–0.19) | 4 | 0.14 (0.06) |
Risk of extreme tiredness | 2 | 0.16 (0.14–0.17) | 2 | 0.19 (0.18–0.20) | 2 | 0.18 (0.09) | 2 | 0.20 (0.18–0.22) | 2 | 0.22 (0.21–0.23) | 2 | 0.20 (0.08) |
Hair loss | 3 | 0.08 (0.08–0.09) | 5 | 0.14 (0.13–0.15) | 5 | 0.10 (0.08) | 3 | 0.10 (0.09–0.11) | 5 | 0.13 (0.12–0.14) | 5 | 0.11 (0.10) |
CI, confidence interval; DCE, discrete choice experiement; s , standard deviation; SW-DR, swing weighting with direct rating; SW-ROC, swing weighting with rank order centroid method.
Rank-Ordered Logit Model a Beta Parameters (Mean and SE) Comparing SW-ROC and DCE Including Likelihood Ratio Test and Dirichlet Regression Odds Ratios (SW-DR Compared with DCE and 95% CI) Including the Dispersion Parameter (Ln Phi)
5-y Survival | Mode of Administration | Risk of Long-Lasting Skin Problems | Risk of Extreme Tiredness | Hair Loss | Ln Phi | ||
---|---|---|---|---|---|---|---|
Rank-ordered logit | — | ||||||
Italy | Mean | 0.09 | 2.17 | 1.12 | −0.14 | Ref | — |
(SE) | (0.37) | (0.23) | (0.21) | (0.22) | — | ||
Likelihood ratio test, χ = 2,042.9(df = 4), < 0.001 | |||||||
Belgium | Mean | 0.62 | 3.58 | 1.48 | 0.70 | Ref | — |
(SE) | (0.31) | (0.35) | (0.22) | (0.23) | — | ||
Likelihood ratio test, χ = 1,932.5(df = 4), < 0.001 | |||||||
Dirichlet regression | |||||||
Italy | OR | Ref | 4.87 | 3.80 | 2.46 | 2.58 | 0.75 |
(95% CI) | (4.13, 5.74) | (3.24, 4.45) | (2.14, 2.83) | (2.22, 3.03) | (0.64, 0.87) | ||
Belgium | OR | Ref | 6.41 | 3.41 | 2.15 | 1.67 | 1.02 |
(95% CI) | (5.27, 7.80) | (2.90, 4.01) | (1.87, 2.48) | (1.42, 1.98) | (0.87, 1.20) |
The weights of all the attributes differed substantially between DCE and SW-DR ( Table 3 and Figure 2 ). The largest difference was found for the weight of “5-y survival,” which was much greater for the DCE (59%–63% of total weight) than for the SW-DR methods (31%–33%). The differences in the weights are evidenced in their 95% confidence intervals, which minimally overlap between methods ( Table 3 ). The less important attributes had different weights but were more comparable across methods.
Relative attribute ranking and weights for Belgium (A) and Italy (B) calculated based on the discrete choice experiment (DCE) data and the swing weighting (SW) data, using both direct ranking (DR) and rank-ordered centroid (ROC).
The outcomes of the Dirichlet regression models are shown in Table 4 . The odds ratio refers to the attribute weights of all attributes relative to 5-y survival of the SW-DR (with the DCE being considered the base case). The aggregate attribute weights of the DCE and SW-DR were significantly different (LL ratio = 466.4 for Italy, P < 0.0001; LL ratio = 435.0 for Belgium, P < 0.0001). Weights of the SW-DR were more equally divided over the included attributes as compared with the DCE (in the DCE, most of the weight was allocated to the 5-y survival attribute; Figure 3 ). Relative to survival, the attribute importance weights calculated from the SW-DR for skin problems, mode of administration, tiredness, and hair problems were significantly larger compared wit the DCE weights ( P < 0.001). Moreover, for Italy, the weights based on the SW-DR were significantly less dispersed (i.e., weighted more equally) compared with the DCE (ϕ = 0.75, CI: 0.64–0.87; P < 0.001). These differences remained highly significant even after correcting for educational level, health literacy, gender, age, cancer stage, and treatment experience.
Comparison of the rankings derived from the ROC and DCE in Italy (A) and Belgium (B), and the attribute weighting from the point task and the DCE in Italy (C) and Belgium (D).
This study used empirical evidence to compare the relative importance of NSCLC treatment profile attribute ranking and weighting when assessed using a DCE or SW-DR task. Significant differences were found in the relative ranking and weights of the attributes between the SW-DR and the DCE. Similar results were found in the 2 countries included in this study, supporting the theoretical validity of these study outcomes. In addition, respondents generally indicated that both DCE and SW-DR tasks were easy or very easy to understand and answer.
The difference in relative attribute weights and ranking is likely in part due to the differences in how the 2 methods assess patient preferences and how respondents engage with the tasks. In an optimally designed DCE, respondents are forced to weigh all attributes when choosing and cannot directly state their individual attribute valuations. In contrast to this multiattribute nature of a DCE, in which the total utility of all attributes guides choices, the SW-DR method is unrestricted, allowing respondents to assign any number of points to attributes (excluding the most important attribute, which automatically receives 100 points). 25 This may have induced an equivalence bias, leading to a relative undervaluing of the more important attributes and overvaluing of the less important attributes. 53 The potential presence and impact of equivalence bias in SW experiments should be tested in future research, as the current study was not powered to test conclusions in this regard. Nevertheless, a small explorative post hoc add-on study was conducted (see Appendix B ) to explore whether a restricted PA task (forcing respondents to consider all attributes in assigning points) results in more equivalent relative importance weights to the DCE than the unrestricted DR task. In this study, 14 (randomly selected) Italian patients who previously completed the full survey were asked to complete the SW-DR task from the original survey as well as an additional restricted PA task. Respondents were asked to divide a total of 100 points over 5 attributes rather than simply rate each swing on a 100-point scale, thus forcing respondents to account for all attributes when allocating points. 26 While small and underpowered due to the explorative nature of this study, the results indicate that weights based on this restricted PA task more closely resemble the DCE study outcomes than those from the unrestricted DR task, which replicate previous findings 53 (see Appendix B ). Further studies are needed to confirm if findings from this exploratory analysis hold with larger samples, different sample composition, and different choice contexts to see whether differences remain and compare the outcomes with DCE outcomes.
Surprisingly, respondents did not report the SW-DR method being easier to understand and answer compared with the DCE. While, on one hand, this supports the use of SW-DR in future research on treatment preferences in similar patient populations, it does not favor this method over the DCE. Contrary, one could question whether DCE choice tasks really are as difficult as previously has been assumed. Respondents might be perfectly capable to accurately complete such choice tasks, which would “call for a partial change in perspective toward this method as being (too) complex and time consuming to complete.” 37 In part, this might be affected by the steep increase in the use of DCEs to elicit preferences, 6 which has undoubtfully led to increased familiarity among researchers with accurate design and conduct of DCE studies. Given that the SW method is relatively unexplored, this calls for further investigation into how best to design such studies, with specific attention for the validity and reliability of this method in studies aiming to measure attribute relative importance ranking and weights. While awaiting this evidence, the current study outcomes support the use of DCE over SW-DR in preference assessment.
A primary strength of this study is that the empirical evidence used to compare the 2 methods was generated in a 1-time survey of NSCLC patients who completed both methods, allowing for direct comparison of results. The within-subjects design reduced the chance of confounding factors playing a role in different preference outcomes. This survey was developed after an extensive qualitative study in close collaboration with a multidisciplinary team of clinicians, patients, and researchers. The tasks were explained using informational videos designed for the study, and the online setting allowing respondents to pause the educational material or the survey and return to it at a later time in. The online setting also allowed for multicountry, location-independent data collection and access for those with more serious disease complications or fatigue to participate, increasing the generalizability of the findings to other NCSLC populations and reducing the chance of bias.
However, this study also had some limitations. First, SW tasks were originally designed to be conducted in person via a trained facilitator. 26 , 31 The current study was administered online, with respondents completing the survey on their own. While online surveys are less costly and time-consuming than interviewer-led studies and SW surveys have previously been done online, the presence of an interviewer allows for assistance and clarification of questions or issues that could arise while the participant is completing the choice task. 69 This can be especially helpful when attributes are complex or the target population experiences cognitive impairments. 31 The patient feedback questions indicated that the online setting was not a problem for this study. Second, the sample was composed of relatively old and “fragile” NSCLC patients, reducing generalizability to younger or less fragile patient populations. Generalizability is also limited by the fact that the digital format of the survey may have discouraged those patients with lower digital literacy from participating as well as those who lack access to computer equipment or to the internet. 70 Third, the current study focused on medical decision making along the medical product life cycle, which did not include clinical or shared decision making. Because other outcome measures and potential methodological considerations might be important when selecting a preference method, the current findings might have limited generalizability toward those situations. Finally, it is unclear whether patients received support from relatives while completing the survey. If this occurred, those supporting the patient in completing the survey could have influenced the outcomes of the survey such that the values measured did not solely reflect the true values of the patient.
In conclusion, this study found significant differences in attribute importance between DCE and SW-DR as well as a greater spread in the DCE-derived relative importance of the attributes. Respondents reported both methods being relatively easy to understand and answer. Further studies confirming these findings as well as SW studies with restricted PA tasks are warranted to enable the provision of accurate guidance for methods selection when studying relative attribute importance across a wide array of preference-relevant decisions. Such studies will contribute to the knowledge base around the validity and reliability of SW in health preference assessment, support guidance for good research practices when using this method, and help researchers decide which method to use when assessing attribute relative importance ranking and weights. While awaiting this evidence, the current study outcomes support the use of DCE over SW-DR in preference assessment.
Acknowledgments.
The authors of this article would like to acknowledge the following people for their contributions to the study: the patients for participating and sharing their valuable insights in the interviews and survey; Luca Bailo for his work in developing the study and writing the protocol; Gabriella Pravettoni for her aid in recruiting patients; Reinhard Arnou for his assistance in conducting and analyzing the patient discussions; Elise Schoefs for her help with the development of the survey and the recruitment of patients; our clinical partners in the Department of Respiratory Oncology, (KU Leuven), Department of Thoracic Surgery (KU Leuven), Thoracic Surgery Division (IEO, Milan), Department of Medical Oncology (Fondazione IRCCS Istituto Nazionale dei Tumori, Milan), and the Department of Medicine Section Hematology/Oncology (University of Chicago) for their contributions throughout the survey development and patient recruitment; and the PREFER consortium for their advice and guidance throughout the study.
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: J. Veldwijk, I. P. Smith, S. Oliveri, S. Petrocchi, L. Lanzoni, Isabelle Huys, Rosanne Janssens, Ardine de Wit, and C. G. M. Groothuis-Oudshoorn declare no conflict of interest. M. Y. Smith is a full-time employee of Alexion AstraZeneca Rare Disease and is a shareholder in the company. The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study formed part of the PREFER project. The Patient Preferences in Benefit-Risk Assessments during the Drug Life Cycle (PREFER) project has received funding from the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement No. 115966. This Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation program and EFPIA. The PREFER project aims to strengthen patient-centric decision-making through evidence-based recommendations guiding stakeholders on how and when patient preference studies should inform medical product development and evaluation. Financial support for this study was provided entirely by a grant from Innovative Medicines Initiative 2 Joint Undertaking under grant No. 11966. The funding agreement ensured the authors’ independence in designing the study, interpreting the data, writing, and publishing the report.
Disclaimer: This article and its contents reflect the view of the presenter and not the view of PREFER, IMI, the European Union or EFPIA.
J. Veldwijk, Erasmus School of Health Policy & Management, Erasmus University, Rotterdam, the Netherlands. Erasmus Choice Modelling Centre, Erasmus University, Rotterdam, the Netherlands. Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Julius Centrum, Utrecht, the Netherlands.
I. P. Smith, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Julius Centrum, Utrecht, the Netherlands.
S. Oliveri, Applied Research Division for Cognitive and Psychological Science, IEO, European Institute of Oncology IRCCS, Milan, Italy.
S. Petrocchi, Applied Research Division for Cognitive and Psychological Science, IEO, European Institute of Oncology IRCCS, Milan, Italy.
M. Y. Smith, Alexion AstraZeneca Rare Disease, Boston, MA, USA. Department of Regulatory and Quality Sciences, School of Pharmacy, University of Southern California, Los Angeles, CA, USA.
L. Lanzoni, Applied Research Division for Cognitive and Psychological Science, IEO, European Institute of Oncology IRCCS, Milan, Italy.
R. Janssens, Department of Pharmaceutical and Pharmacological Sciences, KU Leuven, Leuven, Belgium.
I. Huys, Department of Pharmaceutical and Pharmacological Sciences, KU Leuven, Leuven, Belgium.
G. A. de Wit, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Julius Centrum, Utrecht, the Netherlands. Department of Health Sciences, Faculty of Science, Vrije Universiteit Amsterdam & Amsterdam Public Health Research Institute, Amsterdam, the Netherlands.
C. G. M Groothuis-Oudshoorn, Health Technology and Services Research (HTSR), Faculty of Behavioural Management and Social Sciences, University of Twente, Enschede, the Netherlands.
IMAGES
VIDEO
COMMENTS
An 81-year-old man presented with fever, cough, and shortness of breath. Within a few hours after presentation, chest pain and respiratory distress developed. A chest radiograph showed bilateral pa...
Certain limits on patient choice are obviously objectionable. ... but which are overly risky. In this case, the burdens the patient must bear are health burdens generated by failure to treat a medical condition. ... G., Edwards, A., Gwyn, R. & Grol, R. (1999). Towards a feasible model for shared decision making: Focus group study with general ...
Abstract. Choice is what we all want, as most would say. There is a growing cognisance that patients can and should play an important role in deciding their own care, in defining optimal care, and in improving healthcare delivery. Popular concepts such as patient-centred care, patient empowerment, and patients as partners, shared decision ...
In the past decade, the relationship between patient choice and clinical decision making has been seen as increasingly significant by healthcare systems around the globe. ... Expand 32 Case study: Letting patients decide—A novel distribution strategy in primary care, Massachusetts General Hospital. Leigh Simmons and others. View chapter.
Conflicting Values: A Case Study in Patient Choice and Caregiver Perspectives. Decisions related to births in the "gray zone" of periviability are particularly challenging. Despite published management guidelines, clinicians and families struggle to negotiate care management plans. Stakeholders must reconcile conflicting values in the ...
In light of a case study involving a severely autistic adolescent, ... Relatedly, a clinician's questioning of the rationality and the authenticity of a patient's choice of treatment may also arise when, regardless of her specific religious, cultural, or social characteristics, she chooses to pursue a new practical identity. ...
In this article, I offer a detailed case study on the relationship between specific measures of patient-centered care and the ethical principle of respect for autonomy. Decision Quality Instruments (DQIs) are patient-centered care measures that were developed by Karen Sepucha and colleagues. ... Achieving evidence-based patient choice, ed ...
32 Case study: Letting patients decide—A novel distribution strategy in primary care, Massachusetts General Hospital Notes. Notes. 33 ... Adrian Edwards, and Rachel Thompson (eds), Shared Decision Making in Health Care: Achieving evidence-based patient choice, 3rd edn (Oxford, 2016; ...
This study looks at the factors that can shape patients' choice of healthcare providers. Understanding this process can help with making high quality healthcare more accessible for all. We focus on distance, patient's health status, (perceived) quality of healthcare facility, and referrals to investigate how these factors compete in shaping patients' choice of hospitals.
Request PDF | On Jul 28, 2016, Claudia Zeballos-Palacios and others published Case study: Achieving evidence-based patient choice | Find, read and cite all the research you need on ResearchGate
In several northwest European countries, a demand-driven healthcare system has been implemented that stresses the importance of patient healthcare provider choice. In this study, we are conducting a scoping review aiming to map out what is known about the determinants of patient choice of a wide range of healthcare providers. As far as we know, not many studies are currently available that ...
Conflicting Values: A Case Study in Patient Choice and Caregiver Perspectives Narrat Inq Bioeth. Summer 2015;5(2):167-78. doi: 10.1353/nib.2015.0054. ... Even skilled clinicians may struggle to guide the patient in making value-laden decisions without imposing their own values. Exploring the experiences of one pregnant woman and her caregivers ...
This chapter aims to explore the use of a case-study approach to a specific policy. The use of narrative approaches to analysis provides a way of understanding how policy evolves but also a framework for the analysis of policy. ... The concept of patient choice in England emerged gradually, and over time there was a shift in how patient choice ...
Stockholm has many alternative caregivers to which to refer patients (both private and public), a positive political climate for patient choice and highly motivated patients. In the case study, policy documents from the past 25 years concerning patient choice issues were collected and analysed to determine how the policy had been implemented in ...
In one study, 69.8% of the patients had a partial or ... 12,37,38 With regard to choice of endocrine therapy, the STAGE study showed that ... Case fatality rate of cancer patients with COVID-19 in ...
This study examines the factors that influence patient choice of medical provider in the three-tier health. care system in rural China: village health posts, township health centres, and county (and higher level) hospitals. The model is estimated using a multinomial logit approach applied to a sample of 1877 cases of outpatient treatment from a ...
Sep 2015. CURR SOCIOL. Jonathan Gabe. Kirsten Harley. Michael Calnan. Request PDF | Patient choice: A contemporary policy story | This chapter aims to explore the use of a case-study approach to a ...
Mr Davey is a 70-year-old man with type 2 diabetes, which he has had for ten years. He has a history of alcohol abuse, but in the last five years has abstained from drinking. He recently registered with your new GP practice as he moved to a new house. He has made an appointment to see you, as he needs a repeat prescription. Start questions.
The objective of the current national cohort study was to analyze the correlation between choice and competition on outcomes after cancer surgery using prostate cancer as a case study. Methods: The analyses included all men who underwent prostate cancer surgery in the United Kingdom between 2008 and 2011 (n = 12,925). Multilevel logistic ...
Abstract. This study examines the factors that influence patient choice of medical provider in the three-tier health care system in rural China: village health posts, township health centres, and county (and higher level) hospitals. The model is estimated using a multinomial logit approach applied to a sample of 1877 cases of outpatient ...
We will identify key bottle necks across phases of the testing process (e.g., between test request - test performance - test result - and test response) and degree of guideline concordance to reveal implementation strategies to optimise the primary or secondary care assessment of patients with possible cancer symptoms.
Messages from adolescent patient portal accounts at Stanford Children's Health between June 1, 2014, and February 28, 2020, were sampled and manually reviewed for authorship as described in the study by Ip et al. 3 Two prompts were iteratively engineered on a stratified random subset of 20 messages until perfect performance (100% sensitivity ...
These studies often seek to collate and correlate Breuer's flattened write-up of the case with historical reality, trying to reconstruct both Anna O.'s illness and her medical treatment.
A 57 year-old male lorry driver, presented to his local emergency department with a 20-minute episode of diaphoresis and chest pain. The chest pain was central, radiating to the left arm and crushing in nature. The pain settled promptly following 300 mg aspirin orally and 800 mcg glyceryl trinitrate (GTN) spray sublingually administered by ...
A Q&A and case study show how this Surescripts medication pricing tool helps patients save money on their prescriptions. A Q&A and case study show how this Surescripts medication pricing tool helps patients save money on their prescriptions. ... So, my advice is two-fold: (1) make the best choice in medication obvious; and (2) strike a balance ...
Most studies of patients' preferences when selecting a hospital have been conducted in public or private hospitals in which the majority of patients have had either social security insurance or Iran Health Insurance and private medical insurance coverage. ... Tengilimoglu D, Parsons A. Hospital choice factors: a case study in Turkey. Health ...
Human Subjects Office / IRB Hardin Library, Suite 105A 600 Newton Rd Iowa City, IA 52242-1098. Voice: 319-335-6564 Fax: 319-335-7310
Myasthenia gravis (MG) is a B cell-driven autoimmune disease (AID) that can coincide with rheumatoid arthritis (RA). 1 Here, we report on a 37-year-old woman who was diagnosed with generalised, acetylcholine receptor (AChR)-antibody positive MG in 2013 and developed anticitrullinated protein antibody (ACPA) positive RA in 2020. Arthritis affected multiple hand and ankle joints that showed ...
Norepinephrine and phenylephrine are commonly used vasoactive drugs to treat hypotension during the perioperative period. The increased release of endogenous norepinephrine elicits prothrombotic changes, while parturients are generally in a hypercoagulable state. Therefore, this trial aims to investigate whether there is a disparity between equivalent doses of prophylactic norepinephrine ...
Details on the study design have been published elsewhere. 39,40 This case study was identified as suitable for the comparison of DCE and SW-DR due to the potentially fragile physical state or diminished cognitive status of the patients. 41 -44 The current study included DCE and SW exercises in the ways these methods typically would be ...