D-efficient Design Research Articles

Patient and Physician Preferences for First-Line Treatment of Classical Hodgkin Lymphoma in the United States

▪Introduction: The standard of care for previously untreated classical Hodgkin lymphoma (HL) in the United States has been combination chemotherapy with doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD) alone or with radiotherapy depending on clinical characteristics. Other treatment programs exist, including intensified chemotherapy regimens, and substitution of brentuximab vedotin for bleomycin recently received FDA approval as first-line therapy for advance stage HL. The objective of this study was to understand how specific treatment attributes impact preferences among patients and physicians for choice of first-line treatment of HL.Methods: An online survey including a discrete choice experiment (DCE) was administered both to oncologists who manage patients with HL and to patients diagnosed within HL within the last two years in the United States. Participants were identified via online research panels. The attributes and levels of the hypothetical treatments presented in the DCE were informed by targeted literature review and physician and patient interviews. For physicians, six attributes were included: two-year overall survival (OS); two-year progression free survival (PFS); risk of side effects requiring hospitalization; risk of peripheral neuropathy (PN); risk of pulmonary toxicity; and patient out-of-pocket cost. The patient DCE included four attributes: OS, PFS, risk of PN and risk of pulmonary toxicity. DCE scenarios were developed using a d-efficient design. Participants reviewed 10 scenarios, and selected their preference between two hypothetical treatments. Patients considered themselves when selecting their preference; physicians considered four different advanced HL patient profiles that differed in age (30 or 65 years), smoking status, and the presence of baseline PN. The DCE data were analyzed using a mixed logit model (MXL). The relative importance of each attribute was calculated by determining the differences between the maximum and minimum coefficients of each attribute. These were then normalized and presented as percentages.Results: A total of 200 physicians and 141 patients were included in the analysis. Physicians had a mean of 15 years' experience and 71% practiced in a community setting. Patients had a median age of 35 years (range 19 to 69), 60% were male, and 34% were diagnosed with advanced stage HL. In the DCE, the most important attributes to both patients and physicians were OS and PFS. Based on the coefficients from the MXL model, a 1% increase in OS was more important to both groups than a 1% increase in PFS. The coefficients and level ranges for each attribute were used to calculate preference weights (see methods). Based on preference weights, PFS was the most important attribute for patients, followed by OS, risk of pulmonary toxicity, and risk of PN (Table 1). Compared to male patients, there was a trend for female patients to have a lower preference for a 1% decrease in risk of progression (p=0.077). Patients above the median age of 35 years had a significant preference (p=0.048) for a lower risk of pulmonary toxicity, and a trend for a higher preference for a 1% increase in OS (p=0.059) was observed. OS was also marginally more important to patients diagnosed with advanced stage HL versus those diagnosed in earlier stages. For physicians, preferences for treatment attributes differed based on the patient profile presented. PFS outweighed OS for a healthy 35 year-old patient (Table 2), whereas OS had a higher relative preference weight for a 35 year old smoker and older patients. For smokers, physicians ranked pulmonary toxicity as the most important attribute. Among older patients, side effects requiring hospitalization were more important to physicians' preferences than both OS and PFS. There were no major differences in preferences between academic and community oncologists.Conclusion: Patients are willing to accept treatments with worse short and long-term side effects in exchange for improved OS or PFS. Physicians' treatment preferences are patient-specific, with age and comorbidities impacting both the relative weight of OS and PFS attributes and the importance of pulmonary toxicity and short-term side effects. These results underscore the importance of assessing and sharing patient and physician preferences in creating a treatment plan for the management of newly diagnosed Hodgkin lymphoma. DisclosuresFeliciano:Seattle Genetics: Employment, Equity Ownership. Mueller:ICON plc: Employment. Korol:ICON plc: Employment. He:ICON plc: Employment. Matasar:Seattle Genetics: Honoraria.

PMD181 - PATIENT PREFERENCE FOR CHRONIC OBSTRUCTIVE PULMONARY DISEASE (COPD) TREATMENT INHALERS: A DISCRETE CHOICE EXPERIMENT (DCE) IN FRANCE

Patients’ preference is increasingly explored for COPD treatments, notably because poor inhaler adherence is a major challenge. Understanding inhaler preferences may contribute to adherence and improve long-term outcomes. This study aimed to identify and quantify preferences for convenience-related inhaler attributes in French moderate-to-severe COPD patients, with DCE methodology. No previous studies report on these characteristics in this population. Attributes were defined from a literature search, expert- and patient interviews. An online DCE with COPD patients aged 40-75, using an inhaler for COPD was conducted, including a 2-phase pilot study and cognitive-debriefing interviews. The choice scenarios presented fictitious inhalers with 7 attributes; shape, dose insertion, dose preparation, dose release, dose confirmation, dose counter, and reusability. The COPD Assessment Test and 18 questions on diagnosis, treatment and socio-demographics were included. A D-efficient design with priors was used. Analyses used a mixed logit regression model with random effects. 3 inhaler profiles were of particular interest: Inhaler A - soft mist inhaler, Inhaler B - single dose dry powder inhaler Inhaler C – multi-dose dry powder inhaler. 153 patients (50 females) with a self-reported COPD stage of 2-4 completed the study. Respondents were 50.4 years old on average, using 13 different inhaler devices. Statistically significant results show that L-shaped inhalers, dose preparation with capsule insertion, inhalers with a dose counter and re-usable inhalers were preferred. Pre-filled inhalers were also preferred versus inhalers requiring cartridge or capsule insertion, with borderline significance. Inhaler A and Inhaler A as a reusable device inhaler profiles had the highest utility (mean of 1.2533 and 0.9578 respectively) compared to inhaler B (0.2200) and inhaler C (0.6315). The results show that a number of convenience-related inhaler characteristics are strong drivers of COPD patient preference in France and should be taken into account by clinicians prescribing these devices.

Public, medical professionals’ and patients’ preferences for the allocation of donor organs for transplantation: study protocol for discrete choice experiments

IntroductionOrgan transplantation is the treatment of choice for patients with severe organ failure. Nevertheless, donor organs are a scarce resource resulting in a large mismatch between supply and demand. Therefore,...

PMU107 - DISCRETE CHOICE EXPERIMENTS IN HEALTH ECONOMICS: PAST, PRESENT AND FUTURE

Discrete choice experiments (DCEs) are increasingly advocated as a way to quantify preferences for healthcare. However, increasing support does not necessitate increasing quality. Although specific reviews have been conducted in certain contexts, there exists no recent description of the current-state-of-play in healthcare DCEs more generally. The aim of this paper was to update previous reviews, which covered the period 1990-2012, to identify all health-related DCEs and provide a description of trends, current practice and future challenges. A systematic literature review was conducted to identify health-related empirical DCE studies published between 2013-2017. The search strategy (terms and electronic database) replicated previous reviews to allow the reporting of trends. Data were also extracted in line with previous reviews and included information on attributes and levels, experimental design, analytical methods, validity and outcome measures. From the generated 3381 possible references, 308 DCEs met the inclusion criteria and were selected for data extraction. The number of DCEs per year has continued to increase and in an expanding number of countries. A trend towards the inclusion of attributes involving money, risk or time was also observed. Studies reported using more sophisticated designs with associated software, for example, Bayesian D-efficient designs generated using Ngene. The trend towards the use of more complex econometric models also continued. However, many reports presented sophisticated methods with insufficient detail. Qualitative methods have become more popular as an approach to selecting attributes and levels and to increase understanding of results, which might improve validity. The use of empirical DCEs in health economics continues to grow, although less rapidly than before. However, inadequate reporting of methodological details inhibits quality assessment, reducing confidence in the results and the ability for decision-makers to act on the results. How and when to integrate health-related DCE outcomes into decision-making remains an important area for future research.

Modeling departure time choice of metro passengers with a smart corrected mixed logit model - A case study in Beijing

It is critical to improve the effectiveness of demand management in metro systems with passenger departure time choice exactly learned during peak hours. In this study, a practical framework is developed to model departure time choice of metro passengers during peak hours. First, various attributes that influence departure time choice of metro passengers are investigated and the technique for order preference by similarity to ideal solutions (TOPSIS) is used to identify these main attributes. Then, a mixed logit (ML) model of departure time choice that accounts for price endogeneity is developed. To calibrate the model, a stated preference (SP) survey based on D-efficient design is conducted in the Beijing metro system. It is proved that the corrected ML model outperforms the uncorrected ML model according to the collected 1152 sample data. An elasticity analysis of these main attributes is further conducted, which indicates that metro fare and departure time change influence passenger departure time choice more than crowdedness in Beijing metro. Knowledge of these preferences assists traffic managers in balancing passenger departure time to mitigate overcrowding during peak hours. Heterogeneity of passenger socioeconomic and trip characteristics is also concerned taking advantage of ML model. Finally, a ML-based fare discount strategy to ease the crowdedness in Batong Line of Beijing metro is presented and evaluated via an existing simulation tool.

Can a Repeated Opt-Out Reminder mitigate hypothetical bias in discrete choice experiments? An application to consumer valuation of novel food products

Can a Repeated Opt-Out Reminder mitigate hypothetical bias in discrete choice experiments? An application to consumer valuation of novel food products

Threshold incorporating freight choice modeling for hinterland leg transportation chain of export containers

This study investigated the influence of the time upper threshold on the choice behavior of shippers. The time upper threshold is the longest hinterland leg transportation chain time that can be accepted by a shipper. Based on D-efficient design, data were collected from a freight corridor between a hinterland container terminal and a seaport in China. Model estimation results implied that exogenous threshold models are more realistic as they distinguish preference differences of shippers. Share simulation indicates that railroad share will sharply increase when time reduces and approaches a threshold; furthermore, only when the reduction of railroad transportation chain time exceeds 35%, reducing cost or time fluctuation will significantly increase its share.

Assessment of the societal and individual preferences for fertility treatment in Australia: study protocol for stated preference discrete choice experiments

IntroductionIn Australia, societal and individual preferences for funding fertility treatment remain largely unknown. This has resulted in a lack of evidence about willingness to pay (WTP) for fertility treatment by...

Rural households’ preferences for nutri-cereals supplied through public distribution system: a discrete choice experiment

The Public Distribution System (PDS) is one of the largest safety net programmes in India, set up to provide subsidised grains to the poor population. This paper examines the rural households’ preferences of specific attributes of millets supplied through the PDS. To assess their stated preferences, a Discrete Choice Experiment was conducted in the Sonipat district of Haryana and Baghpat district of Uttar Pradesh in 2018. The sample included 100 respondents who hold BPL (below poverty line) ration card. In all, six attributes of millets, that is product type, form, price, quantity, frequency of supply, and substitution with existing products were identified and a D-efficient fractional factorial design was used to construct pairs of preferences over the existing foodgrains supply through PDS. Respondents’ preference for nutri-millets’ attributes was analysed using a mixed logit model. The results reveal sorghum as the most preferred choice, followed by pearl millet. The respondents expressed their strong preference for supply of nutri-millet flour at least once a month in addition to the existing provision of rice or wheat. Respondents are willing to pay Rs 6.57 more for sorghum and Rs 10.77 more for pearl- millet compared to other minor millets. The acceptance is higher if it is supplied as flour and the willingness to pay was Rs 5.83 more over the grain.

Singular prior distributions in Bayesian D-optimal design for nonlinear models

For Bayesian D-optimal design, we define a singular prior distribution for the model parameters as a prior distribution such that the determinant of the Fisher information matrix has a prior geometric mean of zero for all designs. For such a prior distribution, the Bayesian D-optimality criterion fails to select a design. For the exponential decay model, we characterize singularity of the prior distribution in terms of the expectations of a few elementary transformations of the parameter. For a compartmental model and several multi-parameter generalized linear models, we establish sufficient conditions for singularity of a prior distribution. For the generalized linear models we also obtain sufficient conditions for non-singularity. In the existing literature, weakly informative prior distributions are commonly recommended as a default choice for inference in logistic regression. Here it is shown that some of the recommended prior distributions are singular, and hence should not be used for Bayesian D-optimal design. Additionally, methods are developed to derive and assess Bayesian D-efficient designs when numerical evaluation of the objective function fails due to ill-conditioning, as often occurs for heavy-tailed prior distributions. These numerical methods are illustrated for logistic regression.

Optimal and efficient production of rose coco beans through the twenty-four points second order rotatable design

The yield results of the twenty four points response surface methodology (RSM) design permitted a response surface to be fitted easily and provided spherical information contours besides the realizations of an optimum combination of the fertilizers in rose coco beans, which resulted in economic use of scarce resources for optimal production of rose coco beans. In this study an existing A-optimum and D-efficient second order rotatable design in three dimensions was used to produce rose coco beans optimally and efficiently. The general objective of the study was to produce rose coco beans (Phaseolus vulgaris) optimally and efficiently using an existing A-optimum and D-efficient twenty four points second order rotatable design in three dimensions in a greenhouse setting using three inorganic fertilizers, namely, nitrogen, phosphorus and potassium. Thus the study was accomplished using the calculus optimum value of the free/letter parameter f=1.1072569. The specific objectives were to estimate the linear parameters, thereby making available for the yield response of rose coco beans at calculus optimum value design for the first time, fitted and tested the model adequacy via lack of fit test, and then found the setting of the experimental factors that produces optimal response using contour plots to assist visualizes the response surfaces. This study demonstrated the importance of statistical methods in the optimal and efficient production of rose coco beans. The results showed that the three factors: nitrogen, phosphorus, and potassium contributed significantly on the yield of rose coco beans (p<0.05). The regression coefficients were determined by employing least square's techniques to predict quadratic polynomial model for group 3 greenhouse (GP3G) for the three fertilizer combinations. In GP3G, the second-order model was adequate at 1% level of significance with a p-value of 0.0034. The analysis of variance (ANOVA) of response surface for rose coco yield showed that this design was adequate due to satisfactory level of a coefficient of determination, R2, 0.8066 (GP3G) and coefficient variation, CV was 10.30. The canonical analysis showed that there was the saddle point for GP3G, meaning there was no unique optimum; therefore, ridge analysis was used to overcome the saddle problem. The result from ridge analysis provided the maximum yield of 70.25grams for the three fertilizer combinations at radii of one. We, therefore, recommend the use of GP3G design since it gave the required coefficient of determination (R2=80.66) and the maximum yield (70. 25grams) was achieved.

D-efficient or deficient? A robustness analysis of stated choice experimental designs

This paper is motivated by the increasing popularity of efficient designs for stated choice experiments. The objective in efficient designs is to create a stated choice experiment that minimizes the standard errors of the estimated parameters. In order to do so, such designs require specifying prior values for the parameters to be estimated. While there is significant literature demonstrating the efficiency improvements (and cost savings) of employing efficient designs, the bulk of the literature tests conditions where the priors used to generate the efficient design are assumed to be accurate. However, there is substantially less literature that compares how different design types perform under varying degree of error of the prior. The literature that does exist assumes small fractions are used (e.g., under 20 unique choice tasks generated), which is in contrast to computer-aided surveys that readily allow for large fractions. Further, the results in the literature are abstract in that there is no reference point (i.e., meaningful units) to provide clear insight on the magnitude of any issue. Our objective is to analyze the robustness of different designs within a typical stated choice experiment context of a trade-off between price and quality. We use as an example transportation mode choice, where the key parameter to estimate is the value of time (VOT). Within this context, we test many designs to examine how robust efficient designs are against a misspecification of the prior parameters. The simple mode choice setting allows for insightful visualizations of the designs themselves and also an interpretable reference point (VOT) for the range in which each design is robust. Not surprisingly, the D-efficient design is most efficient in the region where the true population VOT is near the prior used to generate the design: the prior is $20/h and the efficient range is $10–$30/h. However, the D-efficient design quickly becomes the most inefficient outside of this range (under $5/h and above $40/h), and the estimation significantly degrades above $50/h. The orthogonal and random designs are robust for a much larger range of VOT. The robustness of Bayesian efficient designs varies depending on the variance that the prior assumes. Implementing two-stage designs that first use a small sample to estimate priors are also not robust relative to uninformative designs. Arguably, the random design (which is the easiest to generate) performs as well as any design, and it (as well as any design) will perform even better if data cleaning is done to remove choice tasks where one alternative dominates the other.

Modeling of rose coco beans using twenty four points optimum second order rotatable design

The response surface methodology (RSM) is a collection of mathematical and statistical techniques useful for the modeling and analysis of problems in which a response of interest is influenced by several variables, and the objective is to optimize the response. The objective of the study was to model the rose coco beans (Phaseolus vulgaris) through an existing A-optimum and D-efficient second order rotatable design of twenty four points in three dimensions in a greenhouse setting using three inorganic fertilizers, namely, nitrogen, phosphorus and potassium. Thus, the objective of the study was accomplished using the calculus optimum value of the free/letter parameter f=1.1072569. This was done by estimating the parameters via least square's techniques, by making available for the yield response of rose coco beans at calculus optimum value design for the first time. The results showed that, the three factors: nitrogen, phosphorus, and potassium contributed significantly on the yield of rose coco beans (p<0.05). In GP3G, the second-order model was adequate for 1% level of significance with p value of 0.0034. The analysis of variance (ANOVA) of response surface for rose coco yield showed that this design was adequate due to satisfactory level of a coefficient of determination, R2, 0.8066 and coefficient variation, CV was 10.30. This study demonstrated the importance of statistical methods in the optimal and efficient production of rose coco beans. We do recommend a randomize screening of all the fertilizer components with which it has influence on rose coco beans be done to ascertain the right initial amount of each fertilizer that could achieve maximum yield than this study realized.

Patients’ Preferences Related to Treatments for Endometriosis-Related Pain: Results of a Discrete Choice Experiment in the UK

To generate evidence on patient preferences concerning medication attributes among UK patients with endometriosis-related pain. An online discrete choice experiment among endometriosis patients diagnosed 0.5-10 years ago and without confounding pain conditions. Attributes and levels were defined from literature search, expert and patient interviews. The design included 12 choice sets, each consisting of two alternative fictitious treatments and a No treatment option. A pilot study was conducted and a subset did a cognitive-debriefing interview. A D-efficient design was used for the full study. Analyses included a conditional logit model with interactions. The attributes following qualitative research were: mode of administration, ability to conceive, worst level of (i) pelvic pain, (ii) period pain, (iii) pain during intercourse, need for pain medication (days/month), fracture risk and out-of-pocket costs. Following pilot analysis (n=30) and patient interviews (n=10), minor wording and format changes were made. The full study included 200 participants. Participants’ average age was 32.5 (19-42), 58.5% used hormonal contraceptives, 71% wanted (more) children, and 45% were dissatisfied with current pain treatment; 78% of patients always took pain treatment during their periods, 79% reported a period pain-level of 7-10. Of pain types, period pain had the most influence; patients’ WTP was £36.39 for a period pain-level of 4-6. “Ability to conceive” was valued at £20.51/month and participants preferred options without fracture risk (£13.51/month). Both of these are limitations with current treatments. Treatment at a lower cost was preferred, as were oral tablets over vaginal rings and weekly injections. The most important pain type was period pain, perhaps influenced by the fact that 75% of participants experienced severe period pain. Overall lack of preference for vaginal rings should be interpreted with caution as interviews highlighted patients’ lack of awareness of this device. A treatment allowing patients to conceive was appreciated.

Close HER2 Home: Breast Cancer Patients Preferences For Treatment Delivery

This study is currently being undertaken to identify the preferences of people diagnosed with HER2 positive breast for delivery of subcutaneous trastuzumab (Herceptin) therapy. The introduction of a product with a less invasive administration method holds out the prospect of delivering care in non hospital based environments. The design of care pathways that incorporate patient preferences is a key stage in achieving the best clinical outcomes from available resources. Important attributes for the delivery of trastuzumab therapy have been identified through two focus groups undertaken with breast cancer survivors attending a cancer support centre. The group will act as a steering group for the implementation of a discrete choice experiment: confirming the delivery of attribute levels; piloting of the questionnaire and commenting on choice sets; providing feedback on the interpretation of data achieved from respondent participation. Given the number of attributes and levels, the number of combinations is 128. A series of 16 choices will be employed using a D-efficient main effects design. The model will be assessed using conditional logistic regression and mixed logit regression in Stata. If possible, latent class modelling will be performed. A community pharmacy pathway will be implemented using data generated from the study and the feasibility of treatment delivery assessed using a realist evaluation approach. Predicted uptake will be estimated and compared to the current situation. Attributes identifed by the patient group for delivery of treatment are: travel time; travel cost; identity of provider; treatment site; "They know your name". The potential to deliver cancer care in primary care settings is large, given the rise in cancer prevalence due to an ageing population and increasing survivorship amongst cancer sufferers. Design of care pathways that explicitly meet patients' preferences for care delivery is therefore a key step to successful service redesign.

PSS44 - Preferences of The General Population To Avoid Oral Health Outcomes: Results of A Bayesian Discrete Choice Experiment

Paucity of data on quality of life associated with oral health conditions and concerns over using generic measures has led to alternative approaches being used to value prevention of oral health problems. The purpose of this study was to obtain willingness-to-pay (WTP) values for preventing oral ill-health to inform economic modelling, using a discrete choice experiment (DCE). The first stage of the DCE was to identify attributes and levels associated with specific oral health problems (tooth decay and gum disease). This was informed by health states expected to be included in an oral health economic model and a focussed literature review. Pretesting was conducted, followed by two surveys administered online to UK general population panel. The DCE study included a cost attribute to estimate respondents’ WTP to avoid specific oral health problems. A Bayesian D-efficient design was employed using estimates from first survey as informative priors in the final statistical design. Attributes were defined according to the type of tooth affected (molar, pre-molar, anterior), gum disease and cost. The levels within tooth attributes were: no problem, decay without pain, decay with pain and tooth requiring removal. Coefficients and standard errors from the first survey (N=944), were used to inform the second survey (N=1047). Conditional logit model reflecting repeated observations from the same individuals was fitted to the data. The model was statistically significant. Avoiding problems in anterior teeth was most highly valued followed by premolar and molar teeth. Avoiding decay with pain in an anterior tooth generated the highest WTP (mean £245; 95% CI £216 to 272) followed by removal of anterior tooth (mean £203; £169 to £235). Results demonstrate that people have stronger preferences to avoid problems with anterior teeth compared to pre-molars and molars. Prevention of gum problems are also highly valued by respondents.

Why Do People Participate In Health-Related Preference Studies? A Discrete-Choice Experiment

Patient and citizens are increasingly asked to participate in surveys about health to promote community centeredness. Ethical standards govern how one might interact with respondents, but there is a paucity of research asking what potential respondents want out of this type of research. We sought to document the preferences and motivations of potential respondents to survey research. Respondents from a national panel completed a discrete-choice experiment comparing pairs of potential studies respondents could participate in at a local hospital. Studies were defined across six attributes (validity, relevance, bias, burden, time, and reimbursement) with three possible levels each. A D-efficient design resulted in three blocks of 12 tasks. A choice model was estimated using a continuous coded mixed logit and latent class analysis (LCA). After completing the survey respondents were asked which motivating factors they used to justify their choices from pre-defined list of factors that were identified through community engagement. : 629 people participated in the survey. Participants valued validity (OR=2.4), relevance (OR=1.8), and minimizing bias (OR=1.7) the most. A 2-class LCA confirmed that the majority of participants (76%) valued quality indicators, but 24% of respondents strongly valued incentives such as increasing reimbursement (OR=3.4) or decreasing time (OR=1.3). While both groups had similar motivations, the quality-focused class was more likely to be motivated by “measuring real preference” (p<0.001) and “benefits to society” (p=0.009). Given the increase in studies focused on patient and community centeredness, the paucity of preference-based research focused on what patients and citizens want out of research is surprising. Understanding the motivations of respondents is not only important in designing future studies, but also in interpreting the results of existing studies.

Investigating the Heterogeneity in Women’s Preferences for Breast Screening: Does the Communication of Risk Matter?

BackgroundThe relative benefits and risks of screening programs for breast cancer have been extensively debated. ObjectivesTo quantify and investigate heterogeneity in women’s preferences for the benefits and risks of a national breast screening program (NBSP) and to understand the effect of risk communication format on these preferences. MethodsAn online discrete choice experiment survey was designed to elicit preferences from female members of the public for an NBSP described by three attributes (probability of detecting a cancer, risk of unnecessary follow-up, and out-of-pocket screening costs). Survey respondents were randomized to one of two surveys, presenting risk either as percentages only or as icon arrays and percentages. Respondents were required to choose between two hypothetical NBSPs or no screening in 11 choice sets generated using a Bayesian D-efficient design. The trade-offs women made were analyzed using heteroskedastic conditional logit and scale-adjusted latent class models. ResultsA total of 1018 women completed the discrete choice experiment (percentages-only version = 507; icon arrays and percentages version = 511). The results of the heteroskedastic conditional logit model suggested that, on average, women were willing-to-accept 1.72 (confidence interval 1.47–1.97) additional unnecessary follow-ups and willing-to-pay £79.17 (confidence interval £66.98–£91.35) for an additional cancer detected per 100 women screened. Latent class analysis indicated substantial heterogeneity in preferences with six latent classes and three scale classes providing the best fit. The risk communication format received was not a predictor of scale class or preference class membership. ConclusionsMost women were willing to trade-off the benefits and risks of screening, but decision makers seeking to improve uptake should consider the disparate needs of women when configuring services.

Conducting a Discrete-Choice Experiment Study Following Recommendations for Good Research Practices: An Application for Eliciting Patient Preferences for Diabetes Treatments

ObjectivesTo consolidate and illustrate good research practices in health care to the application and reporting of a study measuring patient preferences for type 2 diabetes mellitus medications, given recent methodological advances in stated-preference methods. MethodsThe International Society for Pharmacoeconomics and Outcomes Research good research practices and other recommendations were used to conduct a discrete-choice experiment. Members of a US online panel with type 2 diabetes mellitus completed a Web-enabled, self-administered survey that elicited choices between treatment pairs with six attributes at three possible levels each. A D-efficient experimental design blocked 48 choice tasks into three 16-task surveys. Preference estimates were obtained using mixed logit estimation and were used to calculate choice probabilities. ResultsA total of 552 participants (51% males) completed the survey. Avoiding 90 minutes of nausea was valued the highest (mean −10.00; 95% confidence interval [CI] −10.53 to −9.47). Participants wanted to avoid low blood glucose during the day and/or night (mean −3.87; 95% CI −4.32 to −3.42) or one pill and one injection per day (mean −7.04; 95% CI −7.63 to −6.45). Participants preferred stable blood glucose 6 d/wk (mean 4.63; 95% CI 4.15 to 5.12) and a 1% decrease in glycated hemoglobin (mean 5.74; 95% CI 5.22 to 6.25). If cost increased by $1, the probability that a treatment profile would be chosen decreased by 1%. ConclusionsThese results are consistent with the idea that people have strong preferences for immediate consequences of medication. Despite efforts to produce recommendations, ambiguity surrounding good practices remains and various judgments need to be made when conducting stated-preference studies. To ensure transparency, these judgments should be described and justified.

Determinants of treatment duration in the prevention of recurrent venous thromboembolism: a protocol for a balanced vignette experiment

IntroductionVenous thromboembolism (VTE) is a condition that annually occurs in approximately 1‰ of the world’s population. Patients who have already had a VTE are at elevated risk for a recurrent...