* Candidate for the Lee B. Lusted Student Prize Competition
Purpose: This study investigated the role of facility attributes as well as respondent demographics in influencing preference for treatment at local health care clinics.
Method: A discrete choice experiment (DCE) was conducted as part of a population-based, random cluster survey of 1,434 adults aged 18 years or older in Liberia's Nimba County. Respondents were asked to select their preferred health facility to use for their next episodic illness based on six attributes: waiting time, cost, respectfulness of provider, availability of drugs, thoroughness of medical examination, and facility manager (government v. non-governmental organization). Two mixed logit models were used to estimate utility functions: without and with respondent characteristics (e.g., gender, age, education, wealth status, and post-traumatic stress disorder (PTSD) status). Based on significance of demographic characteristics, stratified mixed logit models were estimated among subgroups to characterize how preferences vary in this population.
Result: In the full model, a thorough examination (β 3.31, P < 0.01), available drugs (β 1.55, P < 0.01), respectful providers (β 0.40, P < 0.01) and lower cost (β -0.25, P < 0.01) were significant drivers of preference. Gender, relative wealth (richest 20% versus rest), and having presumptive PTSD influenced preference. Stratified models showed that females valued non-governmental organization management (β 0.51, P < 0.01) while males showed no significant preference for management. The wealthiest group's preferences were influenced by high quality examinations (β 3.52, P < 0.01) and availability of medicines (β 1.88, P < 0.01) only, while the remainder of the population also valued the remaining attributes (except for waiting time). Preferences of people with PTSD were not influenced by cost, unlike those of people without PTSD.
Conclusion: For the residents of Nimba County, Liberia, markers of technical quality of health care, i.e. a thorough examination and available drugs, were most important in choosing a health facility. Health system reforms aimed at increasing service utilization should focus on investing in health workers and ensuring that dispensaries and health centers are supplied with drugs. Gender, wealth status, and PTSD status were associated with differences in health facility attribute preferences. The factors should be considered in initiatives aimed at increasing utilization of essential health services.
Purpose: Examine the relationship between factors that influence parent decision-making about treatment of their child with Attention-Deficit/Hyperactivity Disorder (ADHD) and their child’s subsequent medication use.
Method: Longitudinal cohort of children newly diagnosed with ADHD who completed an N-of-1 trial of stimulant medication (i.e. 4-week randomized double-blind placebo-controlled trial of methylphenidate at low, medium, and higher dosages). Parent and teacher ratings of the child’s ADHD symptoms were obtained at baseline (e.g. before starting medicine) and after each dosage and placebo week. After completing the medication trial, parents completed validated surveys about their perceptions about ADHD, their beliefs about medicine, and their decisional conflict about continuing medicine (predictor variables). Six months later, pharmacy records were obtained for each child to calculate the number of days covered with medicine (outcome measure). Pearson correlation coefficients (r) between predictors and the outcome measure were calculated. 24 children (mean age = 7.9+/-1.2 years; 68% male, 78% Caucasian, 19.5% African-American) completed the N-of-1 trial and had 6 months of pharmacy dispensing records. Parents who completed scales were mainly mothers (89%) mean age 35.9 +/- 5.6 years.
Results: Greater parent belief that ADHD severely affected their child’s life (r = 0.45, p=0.05) or their own life (r = 0.55, p=0.02) predicted more days covered with medicine. Interestingly, concerns specific to ADHD medicines (e.g. side effects, long-term effects, etc.) was not a significant predictor. However, the more strongly parents endorsed beliefs about the harmfulness of medication (r = -0.60, p=0.004) and overuse of medicines in general (r = -0.43, p=0.05) the fewer days covered with medicine. The higher the parent’s decisional conflict about continuing medicine (r = -0.5, p=0.02) and the higher the decisional conflict resulting from not feeling supported in their decision to continue medicine (r = -0.66, p=0.001), the fewer days covered with medicine. Of note, the child’s initial response to medication (e.g. difference in symptom ratings from baseline to the best week of medication during the N-of-1 trail) did not predict continued medication use.
Conclusions: Parent decisions about continued medication use for their child with ADHD may be driven more by parent beliefs about ADHD/medication and decisional conflict than the child’s actual response to medication. Interventions to support family/self management of ADHD need to address these factors.
Purpose: Physicians are reluctant to use decision aids despite the aids' ability to improve patient care. One potential reason is that physicians lend more credence to human than decision aid advice. Do internal medicine residents lend more credence to a human specialist or a validated decision aid when the advice contradicts their own decision?
Method: A randomized controlled trial. Internal medicine residents from two programs read a scenario of a patient with community acquired pneumonia and were asked whether they would admit the patient to the intensive care unit or the floor. The residents were randomized to receive contrary advice from either an unspecified pulmonologist or a validated decision aid. They were then asked, in light of this new information, to which location they would admit the patient. The primary endpoint was the number of subjects who switched their admission decision depending on whether they received human or decision aid advice, adjusting for their initial admission location. The secondary endpoint was the change in confidence in the decision.
Result: 108 internal medicine residents responded. 28 of 52 residents (53.8%, 95% CI: 39.5-67.8%) who received contradictory advice from the decision aid and 20 of 56 residents (35.7%, 95% CI: 23.3-49.6%) who received contradictory advice from the pulmonologist changed their decision. The odds ratio for a location change, adjusting for the initial admission location, was 2.27 (95% CI: 1.04 – 5.08, P = 0.04) favoring the decision aid. Respondents changed their confidence more after hearing the decision aid’s recommendation (mean -36.0%) than the pulmonologist’s recommendation (mean -23.0%) adjusting for their initial admission decision (adjusted difference -12.9%, 95% CI -3.0% to -22.8%, P = 0.011). Residents also lowered their confidence more if their initial admission location was to the floor (mean -32.8%) than to the ICU (mean -22.2%) adjusting for the source of advice (adjusted difference -12.0%, 95% CI -1.5% to -22.4%, P = 0.025).
Conclusion: Physicians in training treating a scenario depicting community acquired pneumonia were more influenced by the recommendation of a validated decision aid than the recommendation of an unnamed specialist, each of which provided advice that conflicted with the initial admission decision. This suggests that greater willingness to adhere to human over decision aid advice is not a cause of decision aid non-use.
Purpose: Combining individual health state utilities (single states, SS) into joint states (JS) is necessary for accurate decision modeling. We previously showed our linear index model surpasses other theoretical models for estimating JS from SS related to prostate cancer. Some subjects give logically inconsistent ratings in utility surveys, however, which may affect data-based predictions. This research measures the impact of excluding inconsistent ratings on aggregate linear index prediction values, and identifies socio-demographic features linked with inconsistent utility ratings.
Method: Men completed a utility elicitation survey after prostate biopsy (n=279), during which they rated SS and JS related to prostate cancer treatment using time tradeoff. Ratings were tested for logical inconsistency: a JS rating should not be higher than either composing SS. An alternate definition to accommodate measurement error considered ratings inconsistent only if the JS exceeded the population average for the SS by 1+ SD. Aggregate mean SS utilities with and without inconsistent responses were entered into the linear index model to predict JS values. Univariate and multivariate regression showed associations between socio-demographic features and rating inconsistently.
Result: Linear index prediction values are lower by 4 to 7 points on a 0-100 utility scale for JS of Impotence with each of Asymptomatic Localized Disease, living Post-Prostatectomy, and Incontinence when inconsistent responses for these health states are removed. Excluding ratings by people who give 1 SD rating inconsistencies drops prediction values slightly less, by 3 to 4 points for each of the three CS. Univariate regression analyses indicate associations between rating inconsistency and being married, anxious, living at home, and taking more time with each rating. Large errors are made by African-Americans, people with lower education, worse current health, and lower income. In multivariate analysis, marriage and anxiety significantly predict inconsistency by both definitions. Lower education and rating time were also significant for giving any inconsistent rating, while worse current health was significant for rating inconsistently by the 1 SD criterion.
Conclusion: Excluding inconsistent responses lowers JS utilities predicted using our linear index model. Clinically important differences have not been established for utilities. Whether excluding inconsistent responses will change a decision depends on preference sensitivity of the decision. Associations with inconsistent ratings indicate that elicitation methods should account for patient emotions, marital status, and education.
Purpose: Clinical decision making concepts are present in “evidence based treatment” (EBT) definitions; decision making includes utilizing empirical information to provide best clinical care. Naturalistic decision making theory postulates that “experts” in a given field make decisions significantly differently than do “novice” individuals and experts are more accurate and efficient. Experts use specific cognitive tools (i.e. forward reasoning, heuristics) in their decision making that lead to successful conclusions. This study examines the decision making strategies of pediatric mental health clinicians in usual care community practice who have and have not been trained in an EBT.
Methods: Forty-nine clinicians (EBT=14, Non-EBT=35) participated. Samples were comparable on clinicians’ years of experience, demographics, and discipline (psychology, LCSW & MFT). A think aloud protocol was used where clinicians read clinical vignettes and verbalized case conceptualization and treatment planning thoughts out-loud without theorizing about their cognitive processes. Responses were coded according to decision-making process models.
Results: Results revealed that there is a lot of variation on what clinicians attend to and how they make clinical decisions. The presence of significant parent and family factors (e.g., clinical or demographic details) in the case impacted clinicians’ responses such that responses to vignettes describing more complex families resulted in more biases and cognitive errors, especially for non-EBT clinicians. Overall, EBT clinicians used cognitive decision making strategies more similar to “expert” decision-making processes and there were statistically significantly differences compared to non-EBT clinicians. EBT clinicians used forward reasoning (asked fewer assessment questions), organized information (sequential movement from case conceptualization to treatment), made less cognitive errors (attended to correct/relevant parent, family and child factors), generated a small number of accurate hypotheses (diagnoses), found solutions quickly (discussed treatment plan earlier), spent more time discussing treatment (percent of total response time), and provided extensive, detailed treatment plans. Non-EBT clinicians asked significantly more questions in total, made significantly more diagnoses per case (including significant number of inaccurate diagnoses) and spent significantly less time discussing treatment.
Conclusions: Results imply that clinical decision making practices are related to clinician training. Targeting mental health clinicians’ cognitions and decision-making regarding case conceptualization and treatment planning, may be an important avenue towards dissemination-implementation efforts within community pediatric mental health services, rather then the current focus on EBT content.
Purpose: A “contextual error” is a medical error that occurs when a physician fails to take into account information that is expressed outside of a patient´s physical boundaries – i.e. their context – that is essential to planning appropriate care. Using incognito standardized patients, we measured the propensity of physicians to elicit contextual (vs. biomedical) information and to make contextual (vs. biomedical) errors in treatment plans, as well as the avoidable direct costs of these errors to patient care.
Method: Over 18 months, 98 internal medicine attending physicians at 8 Midwestern VA and non-VA practices were visited in vivo by incognito actors presenting variants of 4 previously-validated cases that jointly manipulated the presence or absence of contextual and biomedical factors that could lead to errors in treatment. For each visit, we obtained data on whether the physician elicited necessary information and incorporated it in the treatment plan recorded in the visit note. Mixed models were fitted to examine factors associated with failed elicitation or incorporation of information. Costs of missed services or unnecessary services in each visit were computed using Medicare cost-based reimbursement data.
Result: Biomedical and contextual information were elicited equally often when available (70% vs. 69%). In baseline variants, treatment plans were appropriate in 76% of visits; when errors were possible, 29% of plans were appropriate when only a biomedical error was possible, 15% of plans were appropriate when only a contextual error was possible, and 2% of plans were appropriate when both types of errors (all differences significant). Most errors involved failure to order necessary services, but contextual errors were more likely to result in simultaneous underuse of necessary services and provision of unnecessary services. Contextual errors alone resulted in an estimated additional cost of care of $74,697 (median $194/visit), which was significantly greater than the $14,967 associated with biomedical errors alone (median $23/visit; Wilcoxon p<.001).
Conclusion: Inattention to contextual information, such as patients´ transportation, economic situation, or caretaker responsibilities can have dramatic and measurable implications for quality and cost of care, in some cases beyond those resulting from inattention to laboratory values and medication dosages when delivering care. This study suggests a need for greater prioritization of contextual information in planning patients´ care to reduce medical errors and costs.