MEASUREMENT AND EXPECTATIONS
* Finalists for the Lee B. Lusted Student Prize
Allocation of inevitably limited financial resources for health care requires judgments about the effectiveness of interventions. Health-related Quality of Life (QoL) scales are commonly used in economic evaluations, but interventions are likely to affect QoL more broadly than is measurable with existent scales. In traditionally used measures, physical dimensions like ‘mobility’ are prominently present. In line with the WHO definition of health, a recent Delphi-procedure shows that value assessment needs to put more emphasis on mental and social dimensions. Our main objective was to identify the core dimensions of subjective well-being (SWB) for a new, more comprehensive outcome measure.
Building on a previous review of the literature and existing QoL and SWB measures, followed by a three-stage online Delphi consensus-procedure among five stakeholder groups (i.e., patients, family of patients, clinicians, scientists and general public) to identify the key domains of QoL, we formulated about 3 items per domain for an initial, Delphi-based set of 21 domains of well-being. We tested these questions in a large sample (N=1143) and used dimensionality analyses to narrow down the initial set of domains and find a smaller number of latent factors.
Exploratory factor analysis suggested retaining 5 dimensions. The five-factor model explained 65% of common variance and revealed dimensions of physical health/daily functioning, positive affect/happiness, personal growth, autonomy, and mental health. We propose this 5-factor model for measuring SWB in economic evaluations.
We identified a set of key dimensions to be included a new, comprehensive measure of SWB which reliably captures these dimensions and fills in the gaps of existent measures used in economic evaluations. These efforts are intended to further develop standardized methodology in economic evaluations by providing a more comprehensive and more accurate estimate of outcomes, resulting in more comparable judgments of outcomes in economic evaluations.
Method: We conducted a vignette based survey over the internet. The vignettes depicted clinical study designs (phase I, II, III trial and observational study) in which uncertainties and six other factors potentially influencing approval of research studies were used. We used a factorial design whereby seven aspects were randomly varied in each vignette to produce unique vignettes for each respondent. Each participant reviewed 9 vignettes (4 for each study design and 1 randomly selected vignette). At the end of each vignette participants responded to the question regarding their likelihood to approve the proposed study on a 7-point Likert-type scale. Participants also responded to the open-ended question, “Please briefly describe what factors influenced your decision to approve or not approve the proposed study.” Two team members independently conducted thematic content analysis of qualitative responses to the open ended question and categorized the responses into the nine a priori themes or an emergent theme. Partially mixed concurrent dominant status design was employed. Quantitative component was assigned more weight and mixing occurred at the data interpretation. We investigated the association between themes and approval decision by adjusting for multiple observations per person by applying the random-effect logit model.
Result: Our study included 230 IRB members from 42 institutions. Majority of the respondents had experience conducting clinical research (74%) and training in research ethics (65%). Ninety percent (N=208) of these participants provided responses to the open ended question which were classified into 16 themes (9 a priori and 7 emerging themes). Uncertainty, adherence, study design, and harms were frequently and intensely cited to influence study approval. Analysis of two extreme subgroups (approvers vs. non-approvers) showed that uncertainty influenced approval decisions, odds ratios (OR) =3.5 (95%CI: 1.3-9.8) and OR=3.2 (95%CI: 1.1-8.9), respectively, based on theme frequency and theme intensity, ignoring multiple observations per person. Taking into consideration multiple observations per person, similar results were obtained for uncertainty: OR= 8.9(95%CI: 0.93-85.4).
Conclusion: Uncertainty regarding therapeutic efficacy is critical in IRB members' approval of clinical trials. This indicates that before approving the trials, IRBs should introduce the policy to better assess the existing uncertainties in a given trial.
Method: 606 participants recruited via Amazon’s Mechanical Turk completed an online survey. Mean age was 33 (SD=11; range=18-80), and the majority (81%) were white. Participants responded to 52 questions designed to assess the 2 proposed preference dimensions (maximizer/minimizer; naturalist/technology enthusiast). An exploratory factor analysis identified items that loaded highly onto factors corresponding to these dimensions. To assess predictive validity, participants reported their healthcare utilization and responded to 4 medical decision scenarios.
Result: Analyses showed that 9 questions captured the maximizer/minimizer distinction (Cronbach’s α=.83), and 8 items captured naturalist orientation (α=.86). No factor corresponded to a general preference for technology.
These preference measures significantly predicted self-reported medical utilization and scenario choices, controlling for demographic factors. Relative to minimizers, maximizers took more medicines (r=0.21), visited the doctor more often in the last year (r=0.36), had more medical scans (r=.010), more overnight hospital stays (r=0.13) and had more lifetime surgeries (r=0.12; all p<0.01). Maximizers were also more likely to choose the more active medical treatments in all of the decision scenarios (all p<0.01). Naturalists were more likely to take herbal remedies and supplements (r=0.52), more likely to have visited an acupuncturist (r=0.15), chiropractor (r=0.12) or healer (r=0.25), and were more likely to refuse medicine prescribed by a doctor (r=0.25; all p<0.01). Naturalist orientation was negatively associated with choosing active treatments for 3 hypothetical scenarios.
Conclusion: We developed and validated a scale that assesses preferences for maximizing vs. minimizing healthcare, and preferences for natural treatments. The maximizer/minimizer distinction, in particular, predicted many aspects of healthcare utilization. We hope that this scale will be used to predict patient outcomes and to better understand real-life medical decisions.
Pediatric influenza can place significant burden on caregivers. The specific aim of our research was to estimate the contribution of a parent’s perceived burden as a caregiver on their level of expectations to receive antibiotic prescriptions for their influenza infected child.
A prospective study was conducted using a self-administered survey of parents at West Virginia University Pediatric Clinic, Morgantown, WV. Parents of children (age ≤ 5 years) who are suffering from influenza/influenza like illnesses (ILI) were considered for this study. Parents of those children without chronic medical conditions perceived by parents but with signs and/or symptoms of influenza perceived by the parents were recruited for this study at the patient waiting room. A data collection instrument was prepared using pre-validated questionnaires. The Modified Caregiver Strain Index (MCSI), a 13 items tool that measures strain (major domains: financial, physical, psychological, social, and personal) related to care provision was used to measure caregiver burden. Parents’ expectation to receive antibiotic prescription was measured on a scale of 0 (‘No Expectation’) to 100 (‘High Expectation’); a 100mm visual-analog-scale was used for this purpose. Socio-demographic information of the parents and their knowledge related to antibiotic use was also measured. Descriptive and multivariable regression analyses were performed using SAS®9.3 (significance level = 0.05).
A total of 53 completed surveys were analyzed. Mean age of the sample was 30.3±7 years and 75% of them were female. The median expectation score to receive antibiotic prescription for children was 50 on a scale of 0 to 100. Unadjusted correlation analysis revealed that parents’ expectation for antibiotic prescription was highly correlated (correlation coefficient = 0.75, p <0.0001) with the burden a parent perceived as a caregiver when the child was suffering from influenza/ILI. Covariate adjusted multivariable regression analysis indicated that caregiver burden was the most influencing predictor (p<0.0001) of parents’ expectation for antibiotic expectation. The only covariate which has significant effect (p=0.03) on expectation was parents’ believe that “in the past they have received antibiotic prescription for their children because they requested it from their pediatricians”.
There was significant effect of perceived caregiver burden on parents’ expectation to receive antibiotic prescription for their children. Policy makers and intervention programs should consider this factor to boost effective reduction of antibiotic expectations.
Surgical site infections (SSI) after orthopedic surgery have large personal and societal costs. The success of interventions to decrease SSI risk, such as perioperative protocols, may be undermined by discordant perceptions of SSI risk among different surgical team members. We sought to quantify these perceptions on the spine surgery service of a major orthopedic hospital.
We developed, piloted, and administered a survey of potential SSI risk factors composed of selected binary comparisons of 23 potential pre-, post-, and intraoperative sources of infection risk. Respondents held one of 5 team roles: surgeon; operating room (OR) nurses or technician; infection control (IC) nurse; physician assistant (PA); or infectious disease (ID) physician. We calculated the first eigenvector of pairwise comparisons to quantify group-specific risk rankings and also assessed the strength of belief in these rankings across groups on a 0-1 scale (1=strongest belief).
The survey was conducted at one of the largest orthopedic hospitals in the United States; calculations are based on responses from 147 spine surgery service workers (122 OR staff, 10 surgeons, 9 PAs, and 6 IC staff). OR and IC staff consider proper pre-surgical sterile preparation of patients to be the most important risk factor for SSI. In contrast, the most important risk factors chosen by surgeons and PAs were patient comorbidities and post-operative soiling of the surgical site, respectively. Surgeons held the strongest views regarding individual SSI risk factors, followed by IC staff, PAs, and OR staff (relative strength of belief 0.62, 0.60, 0.58, and 0.33, respectively)
Risk perceptions for post-spinal surgery SSI vary widely in content and strength of belief among surgical team members at our high-volume orthopedic hospital. Addressing these role-specific health beliefs may aide in implementing infection control interventions.
Method: We developed 14 short scenarios in which people imagined facing a well-defined risk presented in 1-in-X format (e.g., a 1 in 10,000 risk of experiencing a stroke from a medication or a 1 in 10,000,000 chance of winning a lottery). Both probability and valence (positive / negative) varied across scenarios. Participants rated how worried or excited they would be about the possibility of being “the 1 person” who had the event happen. We recruited 2464 Mturk participants to complete both this and other scales to examine scale reliability, factor structure, and discriminant validity versus other measures. In addition, we assessed these questions’ ability to predict (as part of another study) attitudes regarding the potential risks associated with various degrees of maternal alcohol use in pregnancy.
Result: The overall scale (i.e., average ratings of worry and excitement) was highly reliable (Cronbach’s alpha=0.84) and its factor structure included a main scale factor, a valence factor, and a probability level (rare vs. common) factor. It was only modestly correlated with subjective numeracy (r=-0.19) and tolerance for medical ambiguity (r=0.17) and uncorrelated with optimism (r=0.03) and rational-experiential thinking (r=0.02). Furthermore, the scale showed predictive validity: participants scoring higher rated a pregnant mother taking a single sip of alcohol as more likely to have caused fetal harm (p<0.001). A 3 question reduced scale focused on rare negative events had good reliability (alpha=0.75) and equivalent predictive validity.
Conclusion: Certain people appear more emotionally reactive to the possibility that they might be the 1 person out of many who would experience rare risk events. This sensitivity has unique relevance for predicting medical decisions about risk, especially rare complications risks. The “Be the 1” items tested here appear to be a reliable way of identifying such people. Additional research is needed to further establish its validity and assess its usefulness in research about risky decision making (both medical and non-medical).