Breast cancer patients’ visual attention to information in hospital report cards: An eye-tracking study on differences between younger and older female patients

Abstract

To (1) explore how women visually attend to a hospital report card (HRC), (2) explore whether visual attention of younger and older women (patients and non-patients) differs. Eye-tracking study with a short survey. Participants (N = 37) were provided with a hypothetical realistic HRC. Total dwell times and fixation counts were measured while participants viewed the information. Overall, no differences existed between younger and older women. Visual attention to the hospital of choice (vs not of choice) and to indicators perceived as most important (vs least important) did not differ. However, women with higher health literacy looked longer at the HRC than women with lower health literacy. Also, per fixation, older patients (vs younger patients) looked longer at the hospital of choice and at indicators perceived most important. Pre-existing conceptions of what information is relevant might result in more in-depth information processing among older patients than younger patients. In general, differences in level of health literacy, rather than (chronological) age, seem to be relevant to take into account when designing and/or updating HRCs.

Keywords

Breast cancer comparative performance information eye-tracking hospital report cards probabilistic information

Introduction and background

Over the past few decades, online hospital report cards (HRCs) have been made available for patients,¹ and have also gained ground in oncology.² HRCs are assistive frameworks based on comparative performance information (CPI) that contribute to patient empowerment,³ and could help patients to compare and choose hospitals on provided services and quality.^4,5 HRCs typically contain information about general hospital characteristics, the quality of provided care (e.g. number of patients treated, and patient recommendations), and the availability of particular technology/resources (e.g. mammography, MRI).¹

To apply HRCs in hospital choices, it is important that patients perceive the HRC as useable,⁶ that they believe they are able to use the HRC,⁶ and that they appropriately use the information provided.^7–9 However, information in HRCs is often difficult to process because typically multiple choice options with conflicting attributes need to be compared (i.e. multi-attribute choices).^10,11 Cognitive overload is known to occur quickly in multi-attribute choices,¹² because people can only hold a fixed number of elements in their working memory.¹³ Consequently, patients tend to rely on heuristics.^14–17 In the simplified decision-making process most people use, they usually eliminate options that do not seem attractive or relevant based on a few attributes, and subsequently weigh remaining information, which is a limited part of total information available.^17–20 This might lead to hospital choices that do not correspond with patients’ preferences and values, which may consequently lead to undesired outcomes such as decisional conflict.²¹

With regard to HRCs in oncology, it is especially important to carefully test HRCs in the group of older adults (≥ 65 years), who make up a substantial part of the cancer population.²² Older adults are at risk of suboptimal information processing due to an age-related reduction in working memory capacity.^23–26 As a result, older patients experience cognitive overload more easily than younger patients, likely leading to more reliance on shortcuts in information processing.^27,28 In previous studies outside the field of HRCs, but within the health domain, such cognitive overload has been associated with suboptimal comprehension and suboptimal decision-related outcomes.^29,30

In attempts to (re)design HRCs for use by older patients, an important first step is to gain insight into how older patients visually attend to this information. This study aimed to explore how women focus their visual attention to core (as indicated by experts and patients in previous studies) decision-relevant information in a HRC about breast cancer care, and whether this differs between younger and older women by using eye-tracking. Eye-tracking has been applied in health communication research, for example showing that older people, compared to younger people, need more time to process online information,³¹ and switch less often between website elements.³² However, to date, it remains unclear how older patients visually attend to multi-attribute information in HRCs, and how this compares to younger patients. Breast cancer care was chosen as a case example within oncology, as HRCs have become common in this specific setting. The research questions were:

1. How do women focus their visual attention to core decision-relevant information in a HRC?

a. What attributes (i.e. quality indicators) do women visually attend to?

b. What choice options (i.e. hospitals) do women visually attend to, and how does this relate to their hypothetical hospital choice?

c. How does the visual attention to attributes and choice options relate to the perceived relevance of those attributes and options?

2. Are there differences between younger women (<65 years) and older women (≥65 years) in visual attention devoted to:

a. the attributes (i.e. quality indicators)?,

b. the choice options (i.e. hospitals)?, and

c. attributes and choice options of perceived relevance?

3. Are there differences between patients and non-patients in visual attention devoted to the measures mentioned under RQ2a to RQ2c?

Methods

Design

Eye-tracking measures were combined with a short survey. Eye-tracking is a technology that monitors and records eye movements while subjects pay visual attention to stimuli.^33,34 Attention to information precedes information encoding and storage in working memory,^35–37 and is therefore of importance regarding patients’ information processing.

Stimulus materials concerned a fictitious HRC version on breast cancer care, based on realistic CPI from the Dutch ‘Monitor Borstkankerzorg’ (see Materials). Prior to data collection, this work was examined by [details omitted for double-anonymised peer review] as a non-WMO (Medical Research Involving Human Subjects Act) research, and approved by [details omitted for double-anonymised peer review]. Written informed consent was obtained from participants.

Participants

Participants (N = 37) were females recruited through several channels (see Table 1). Participants were analogue patients³⁸ with (n = 17) and without (n = 20) a history of cancer. 20 participants (10 patients, 10 non-patients) were younger (i.e. <65 years old) and 17 participants (seven patients, 10 non-patients) were older (i.e. ≥65 years old). An age cut-off of 65 years was used, as common in health-related studies into aging.^39,40 Participants were included if they: (1) were female, (2) had a sufficient mastery of the Dutch language (both reading and speaking), and – in case the participant was a patient – (3) had already completed primary treatment. The third inclusion criterion aimed to avoid burden for those who might still have to choose a hospital.

Table 1.

Overview of inclusion of participants per sampling channel.

Channel	Younger participants		Older participants		Total
Channel	Patient	Non-patient	Patient	Non-patient	Total
Dutch breast cancer Association (BVN)	2	0	5	0	7
PanelCom (online access panel)	8	2	0	2	12
Snowballing (i.e. personal networks, colleagues, and social media of the researchers)	0	8	2	8	18

Procedure

Potential participants with interest emailed the first author, who sent an information letter to the potential participant and scheduled an appointment. The information letter described the study aim/content, potential advantages and disadvantages of participation, privacy, and the procedure concerning participation cessation or complaints.

The study was performed in a research lab of [details omitted for double-anonymised peer review]. Participants came to the lab, where they were welcomed by two researchers (NGY and AL). Before the actual start of the study, participants were informed about what to expect from the session, and about the eye-tracking procedure. They were then asked to take place in front of the computer and to position their chair in such a way that the SMI eye-tracker would recognize their eyes. Textbox 1 shows the specifics of the eye-tracker. Participants viewed the stimulus materials on the computer screen. They were instructed that they could do this at their own pace. When participants indicated that they were finished, the researcher shut down the eye-tracker. Socio-demographic variables (age, educational level, health literacy) were assessed via a short survey to describe the sample.

Textbox 1. Specifics of the eye-tracker.

In iView X2 the researcher could see a green bar when the participant was positioned correctly. Hence, participants were installed at a distance of 60–80 cm from a 22-inch display (100 Hz) with a resolution of 1680 × 1050 pixels. The computer had an Intel Core i7 – 4610m CPU (3 Ghz) processor. Both eyes were monitored using non-invasive, video-based dark-pupil eye-tracking, with 0.03° RMS (spatial). All participants were calibrated on the eye-tracker using a 5-point calibration routine. Simultaneously, visual attention was recorded using the Experiment Center software.

Stimulus materials

The stimulus material consisted of a simplified, fictitious HRC providing CPI about three hospitals (Hospital A, Hospital B, and Hospital C). Per hospital, 10 quality indicators were shown. Hospital A performed best on five indicators when compared to Hospital B and C; Hospital C performed best on four indicators when compared to Hospital A and B. Hospital B did not perform best on any of the indicators, and scored equal to Hospital C on one indicator. Hence, there was no dominant best option, making the hypothetical hospital choice preference sensitive. To ensure that the most relevant indicators for patients were included, they were derived from the HRCs from the Dutch ‘Monitor Borstkankerzorg’ and were selected together with the BVN based on their previous evaluations with patients. Hence, although the HRC was fictitious, it was a realistic representation of existing and core CPI for breast cancer care according to experts and patients (see www.monitorborstkankerzorg.nl).

Two versions of the fictitious HRC providing CPI were developed (Figure 1 shows one version). Both versions contained the same information (i.e. the same attributes), and had the same design, but the order of attributes differed. As previous studies suggested that the order of indicators can influence attention paid to them, having two different versions enabled us to avoid a potential order effect.^41–43 Half of the participants viewed the first version of the stimulus material (n = 19), and the other half the second version (n = 18). Both younger and older women, and patients and non-patients were equally divided over the versions.

Figure 1.

Stimulus material (version 1).

Measures

Background characteristics

Age, educational level, and health literacy^44,45 were assessed.

Survey measures

The survey measures were related to how participants comprehended and used the information. Our central premise was that the “best option” did not exist, and that participants’ choices are mainly (if not entirely) preference-sensitive. As such, choosing hospital A, B or C was not considered an optimal or suboptimal decision.

Hypothetical hospital choice

“If I had to choose between the hospitals right now, I would choose:”. The response options were: “Hospital A”, “Hospital B”, and “Hospital C”.

Importance attached to indicators

“You saw different indicators of quality of care on which hospitals are being compared. Which of these indicators are the most important in making your choice?” (4-point Likert scale, 0 = Not important to 3 = Of utmost importance).

Perceived cognitive load

Perceived cognitive load was measured by 4 items (5-point Likert scale, 0 = ‘yes’ to 4 = ‘no’).⁴⁶ The mean was taken to calculate a mean perceived cognitive load score per age group (α = 0.67). The median split was used to dichotomize perceived cognitive load to low and high perceived cognitive load.

Comprehension

Gist comprehension⁴⁷ was measured by seven questions related to the indicators, and consisted of five response options. For each question, only one response option was correct, and one point could be achieved per correct response (range 0–7). For example: “For Nina it is of utmost importance that many patients are treated in a hospital. Which hospital would be the best option for her?” Answer options for this question were: (a) Hospital A, (b) Hospital B, (c) Hospital C, (d) It doesn’t matter, (e) I don’t know. Verbatim comprehension was measured by six questions related to the indicators, and consisted of five response options. For some of these questions, multiple response options were correct. Participants were asked to select all correct options, and one point could be achieved for a correct response (range 0–6). For example: “In which hospital, less than 25 direct breast reconstruction surgeries were performed?”, with response options: (a) Hospital A, (b) Hospital B, (c) Hospital C, (d) None of the hospitals, (e) I don’t know.

Eye-tracking measures

The eye-tracker registered dwell times, i.e. the total amount of time a participant spends looking at a predetermined area of interest (AOI). A variable representing the dwell time per fixation was calculated by dividing the dwell times by fixation counts. Fixation counts refer to the number of revisits at an AOI and a higher number of revisits can be evaluated as a proxy for greater interest.⁴⁸ As fixation counts (in general, but also in our sample) strongly correlate with dwell times, we did not report the findings on fixation counts separately. The calculated variable (hereafter called dwell time/fixation), compared to registered dwell times, was thought to indicate more appropriately how efficiently participants used the information. It was expected that older women would show longer dwell times and less fixation counts than younger women. Per fixation, older women were expected to look longer at information than younger women, which was assumed to be an indicator for less efficient information use.

We also constructed variables indicating dwell times and dwell time/fixation for: (1) the hospital of choice; (2) the hospitals not chosen; (3) the indicators reported to be most relevant for hospital choice (i.e. ‘of utmost importance’; score 3, see Survey measures); (4) the indicators reported to be least relevant for hospital choice (i.e. ‘not very important’; score 0 or 1, see Survey measures). These variables were thought to indicate more appropriately how participants used the information in the context of what matters most or least to them. For the indicators reported to be most relevant, per participant it was checked how many quality indicators they reported to be ‘of utmost importance’. The dwell times and dwell time/fixation for these indicators were summed and divided by the number of indicators reported to be ‘of utmost importance’. For example, one participant reported all indicators to be ‘of utmost importance’. For this participant, dwell times and dwell time/fixation were summed and divided by 10. For the indicators reported to be least relevant, the same procedure was repeated. For several women, who did not indicate any indicator as ‘not important’ (score 0), this was calculated with indicators perceived to be ‘somewhat important’ (score 1).

Statistical analyses

For data registered with the eye-tracker, 45 areas of interest (AOIs) were created beforehand. These AOIs enabled to derive the dwell times and fixation counts. These data were imported to the statistical software program SPSS, version 26. All survey data were inserted into the SPSS data file. Descriptive statistical analyses, paired samples t-tests, and independent samples t-tests were performed. Normal distribution was tested with the Shapiro-Wilk Test. For not normally distributed variables, the non-parametric alternatives (i.e., Wilcoxon test or Mann-Whitney U test, respectively) were used. Correlation analyses were performed to explore the relationship between total dwell times and total dwell time/fixation, and survey measures. One significant correlation was found between health literacy and total dwell times (see Results). All p-values ≤0.05 were considered statistically significant. Effect sizes have been provided for statistically significant findings.

Results

Survey findings

Table 2 displays the study population’s characteristics. There were no significant differences between younger and older participants in background characteristics, except for age.

Table 2.

Characteristics study population.

Characteristics	Younger (<65 years) (N = 20)	Older (≥65 years) (N = 17)	Total (N = 37)
Age (M ± SD)	44.54 ± 15.81	69.68 ± 5.14 ***	56.09 ± 17.46
Education (n (%))
Low	—	1 (5.9%)	1 (2.7%)
Moderate	6 (30.0%)	3 (17.6%)	9 (24.3%)
High	14 (70.0%)	13 (76.5%)	27 (73.0%)
Hypothetical hospital choice (n (%))
Hospital A	6 (30.0%)	7 (46.7%)	13 (37.1%)
Hospital B	—	—	—
Hospital C	14 (70.0%)	8 (53.3%)	22 (62.9%)
Health literacy (M ± SD)^a	17.70 ± 3.98	18.24 ± 2.25	17.95 ± 3.27
Importance attached to indicator (Utmost importance (%)/Least importance (%))
Number of patients who underwent surgery	5.0%/10.0%	17.6%/0.0%	10.8%/5.4%
Number of direct-to-implant breast reconstructions	5.3%/21.1%	17.6%/29.4%	11.1%/25.0%
Structure of weekly multidisciplinary meeting	30.0%/10.0%	17.6%/11.8%	24.3%/10.8%
Waiting time examination-diagnosis	47.4%/0.0%	52.9%/0.0%	50.0%/0.0%
Waiting time diagnosis-treatment	52.5%/0.0%	41.2%/0.0%	47.2%/0.0%
Remaining cancer tissue	47.4%/0.0%	70.6%/0.0%	58.3%/0.0%
Regular contact person	36.8%/0.0%	47.1%/0.0%	41.7%/0.0%
Discussing consequences of treatment	31.6%/0.0%	41.2%/0.0%	36.1%/0.0%
Counseling recovery after treatment	10.5%/5.3%	18.8%/0.0%	14.3%/2.9%
Recommendation by patients	55.0%/10.0%	29.4%/0.0%	43.2%/5.4%
Perceived cognitive load (M ± SD)^c	7.20 ± 4.32	9.41 ± 6.44	8.22 ± 5.43
Gist comprehension (M ± SD)^d	6.10 ± 1.07	6.18 ± 1.24	6.14 ± 1.13
Verbatim comprehension (M ± SD)^e	5.75 ± 0.55	5.47 ± 0.72	5.62 ± 0.64

*p ≤ 0.05 **p ≤ 0.01 ***p ≤ 0.001.

^aRange of possible scores: 0–22.

^bRange of possible scores: 0–3

^cRange of possible scores: 0–20.

^dRange of possible scores: 0–7

^eRange of possible scores: 0–6

Overall, both younger and older women reported the majority of indicators to be of relevance for their hospital choice, and ‘Remaining cancer tissue’ and ‘Waiting time examination-diagnosis’ were reported by (over) half of the women to be of utmost importance (Table 2). The majority of both younger and older women chose Hospital C (i.e. 70.0% and 53.3%, respectively). Younger women perceived a low level of cognitive load (M = 7.20), while older women perceived a high level of cognitive load (M = 9.41). Both younger and older women showed relatively high gist comprehension (i.e. M = 6.10 and M = 6.18, respectively, on a scale of 0–7) and verbatim comprehension (i.e. M = 5.75 and M = 5.47, respectively, on a scale of 0–6). There were no significant differences between younger and older participants concerning the perceived relevance of indicators, hypothetical hospital choice, or comprehension. We only found a significant positive correlation between health literacy and total dwell times, r = 0.35, p = 0.038, meaning that women with higher health literacy looked longer at the HRC than women with lower health literacy.

Eye-tracking findings

Table 3 shows all dwell times and Table 4 all dwell times/fixation for the total sample, for younger and older women separately, and for patients and non-patients separately within the groups of younger and older women.

Table 3.

Average dwell times (milliseconds) for the total sample, per age group, and per patient group.

	Total sample	Younger women			Older women
	(N = 36) *^,1,2	Total (n = 19)	Non-patients (n = 10)	Patients (n = 9)	Total (n = 17)	Non-patients (n = 10)	Patients (n = 7)
Indicators
Number of patients who underwent surgery	6294.16 ± 3867.25^a,b	6156.21 ± 4557.46	5926.29 ± 5244.14	6411.68 ± 3956.69	6448.33 ± 3049.72	5751.25 ± 2671.49	7444.16 ± 3482.82
Number of direct-to-implant breast reconstructions	7313.20 ± 5328.70^c	6905.54 ± 5585.72	7336.00 ± 7255.16	6427.26 ± 3239.53	7768.81 ± 5157.14	8062.79 ± 6553.27	7348.84 ± 2480.94
Structure weekly multidisciplinary meeting	8680.21 ± 6044.97^a,c,d,e	7705.41 ± 6177.33	7815.17 ± 7544.87	7583.46 ± 4667.54	9769.69 ± 5883.53	8494.29 ± 4745.26	11,591.69 ± 7207.56
Waiting time examination-diagnosis	8388.29 ± 7015.53^f	8183.02 ± 7788.00	8501.17 ± 9051.17	7829.51 ± 6636.10	8617.72 ± 6270.81	5962.77 ± 3204.13	12,410.51 ± 7805.84
Waiting time diagnosis-treatment	7130.04 ± 4488.42^d	6477.66 ± 4565.55	6785.11 ± 5584.42	6136.06 ± 3400.89	7859.16 ± 4421.46	6964.44 ± 3710.81	9137.33 ± 5313.76
Remaining cancer tissue	7497.64 ± 5733.82^g	7098.16 ± 5837.10	7460.32 ± 7352.31	6695.77 ± 3937.23	7944.11 ± 5760.70	5804.14 ± 3715.76	11,001.21 ± 7017.75
Regular contact person	8237.94 ± 5689.40^b,h	8087.04 ± 6210.72	7761.49 ± 6442.40	8448.76 ± 6310.09	8406.59 ± 5230.29	8361.75 ± 4424.27	8470.64 ± 6601.50
Discussing consequences of treatment	8104.41±5498.01ⁱ	7530.48 ± 4775.64	5825.64 ± 3832.55	9424.74 ± 5207.76	8745.85 ± 6295.89	7897.64 ± 4421.18	9957.59 ± 8571.43
Counseling recovery after treatment	7744.22 ± 4872.67^j	7135.33 ± 4707.58	7314.95 ± 4895.26	6935.76 ± 4776.90	8424.74 ± 5106.25	7268.17 ± 4428.72	10,076.99 ± 5890.28
Recommendation by patients	5901.71 ± 4414.55^e,f,g,h,i,j	5378.55 ± 3756.24	6443.03 ± 4515.40	4195.79 ± 2412.14	6486.41 ± 5105.81	5111.50 ± 3074.36	8450.57 ± 6905.73
Hospitals
Hospital A	10,159.91 ± 6445.14	8613.56 ± 6280.15	7578.35 ± 6234.43	9763.80 ± 6495.09	11,888.19 ± 6362.77	10,645.19 ± 5580.21	13,663.90 ± 7416.05
Hospital B	9302.91 ± 6534.44	9460.78 ± 7999.60	9771.71 ± 9442.16	9115.31 ± 6590.28	9126.46 ± 4620.43	8348.37 ± 4281.09	10,238.03 ± 5194.90
Hospital C	10,311.59 ± 6896.15	10,021.57 ± 7853.34	9118.95 ± 7384.41	11,024.49 ± 8676.00	10,635.72 ± 5868.12	9902.27 ± 5037.89	11,683.50 ± 7181.81
Hospital of choice	10,801.25 ± 6813.65	9631.87 ± 6649.19	9202.59 ± 7271.72	10,108.84 ± 6285.09	12,282.47 ± 6955.72	10,818.40 ± 4743.37	14,478.58 ± 9477.58
Hospitals not chosen	9470.59 ± 6456.23	9232.03 ± 7474.21	8633.21 ± 7806.91	9897.38 ± 7495.39	9772.76 ± 5123.72	8445.97 ± 5033.30	11,762.95 ± 5005.07
Indicators of most perceived relevance	7973.29 ± 5534.77	8832.37 ± 5581.78	9599.87 ± 6282.24	7936.95 ± 5062.44	8901.15 ± 6298.05	5408.41 ± 1459.27	12,393.89 ± 7439.58
Indicators of least perceived relevance	7243.24 ± 3065.39	7432.31 ± 3845.46	8075.00 ± 4157.31	6682.50 ± 3673.89	7993.29 ± 3348.84	7832.59 ± 2800.55	8153.99 ± 4049.10
Overall	77,097.49 ± 42,674.86	72,561.29 ± 47,987.54	72,454.33 ± 56,038.93	72,680.14 ± 40,600.15	82,167.35 ± 36,624.47	71,716.73 ± 22,040.77	97,096.80 ± 49,052.86

One outlier excluded from analyses.

¹For a, and c to j, the Wilcoxon test was run.

²For b, the paired samples t-test was run.

^aZ = −2.49, p = 0.013, r = −0.42.

^bt(35) = -2.05, p = 0.048, r = −0.34.

^cZ = −2.00, p = 0.046, r = −0.33.

^dZ = −2.20, p = 0.028, r = −0.37.

^eZ = −2.64, p = 0.008, r = −0.44.

^fZ = −2.16, p = 0.031, r = −0.36.

^gZ = −2.03, p = 0.042, r = −0.34.

^hZ = −2.69, p = 0.007, r = −0.45.

ⁱZ = −0.2.06, p = 0.040, r = −0.34.

^jZ = −2.99, p = 0.003, r = −0.50.

Table 4.

Average dwell times/fixation (milliseconds) for the total sample, per age group, and per patient group.

	Total sample	Younger women			Older women
	(N = 34) *^,1^,2^,3	Total (n = 17)	Non-patients(n = 8)	Patients (n = 9)	Total (n = 17)	Non-patients(n = 10)	Patients (n = 7)
Indicators
Number of patients who underwent surgery	240.28 ± 49.41^a	244.39 ± 56.57	266.33 ± 55.15	232.61 ± 57.09	233.61 ± 41.64	227.55 ± 46.82	242.27 ± 34.47
Number of direct-to-implant breast reconstructions	239.04 ± 45.96^b,c	240.97 ± 58.61	273.28 ± 75.22	222.59 ± 29.55	233.65 ± 31.15	231.56 ± 38.66	236.64 ± 18.13
Structure weekly multidisciplinary meeting	249.96 ± 43.71^b	246.62 ± 53.79	274.47 ± 55.61	231.12 ± 46.20	249.84 ± 33.65	237.79 ± 37.42	267.05 ± 18.23
Waiting time examination-diagnosis	248.56 ± 52.52	239.30 ± 42.38	257.65 ± 32.37	226.36 ± 48.00^f	256.56 ± 59.91	240.43 ± 73.08	279.60 ± 22.50^f
Waiting time diagnosis-treatment	256.69 ± 50.89	241.52 ± 49.51	263.68 ± 41.65	230.09 ± 51.79	267.88 ± 51.33	257.24 ± 53.38	283.09 ± 47.89
Remaining cancer tissue	251.34 ± 60.06	241.40 ± 58.25	260.10 ± 56.88	230.27 ± 61.62^g	258.89 ± 61.31	239.57 ± 65.15	286.49 ± 46.32^g
Regular contact person	260.29 ± 68.82^c	250.10 ± 59.37	281.48 ± 45.61	235.29 ± 58.74	264.80 ± 79.98	268.68 ± 87.45	259.27 ± 74.34
Discussing consequences of treatment	250.47 ± 52.61	248.50 ± 53.52	259.22 ± 32.06	248.33 ± 64.66	248.00 ± 54.93	234.18 ± 60.33	267.75 ± 42.58
Counseling recovery after treatment	262.12 ± 60.14^a	260.50 ± 60.39	263.25 ± 52.02	266.15 ± 67.88	259.53 ± 62.43	251.42 ± 67.37	271.11 ± 57.62
Recommendation by patients	249.48 ± 77.32	235.36 ± 52.10	253.29 ± 43.03	227.96 ± 57.13	259.31 ± 96.23	256.73 ± 114.52	263.00 ± 70.64
Hospitals
Hospital A	229.61 ± 46.76^d	218.84 ± 59.52	235.87 ± 71.90	211.49 ± 33.19	236.26 ± 38.03	230.21 ± 36.58	244.89 ± 41.25
Hospital B	234.75 ± 52.31^e	228.83 ± 57.36	241.72 ± 66.15	220.02 ± 46.18	239.28 ± 49.94	231.03 ± 57.26	251.06 ± 38.19
Hospital C	257.40 ± 60.57^d,e	248.37 ± 66.66	275.91 ± 73.04	232.43 ± 54.59	261.91 ± 56.56	248.45 ± 63.92	281.14 ± 40.85
Hospital of choice	248.67 ± 63.00	242.83 ± 63.60	275.91 ± 73.04	213.43 ± 36.81^h	255.29 ± 63.86	242.01 ± 66.15	275.21 ± 60.20^h
Hospitals not chosen	235.02 ± 48.59	231.63 ± 56.80	238.79 ± 67.94	225.25 ± 48.14	238.87 ± 38.88	231.16 ± 43.97	250.44 ± 29.51
Indicators of most perceived relevance	262.26 ± 79.38	244.94 ± 55.67	262.97 ± 43.63	226.91 ± 64.30	296.11 ± 99.73	299.67 ± 140.61	292.56 ± 41.82
Indicators of least perceived relevance	250.31 ± 44.73	255.09 ± 57.11	289.25 ± 49.18	220.93 ± 44.23	248.05 ± 34.12	247.41 ± 46.88	248.69 ± 18.01
Overall	183.40 ± 32.75	176.95 ± 38.30	196.30 ± 31.08	168.61 ± 36.71	185.91 ± 30.10	179.28 ± 36.12	195.39 ± 16.74

Three outliers excluded from analyses.

¹For a to e, the paired samples t-test was run.

²For f and h, the independent t-test was run.

³For g, the Mann-Whitney U test was run.

^at(34) = −2.11, p = .042, r = −.36.

^bt(32) = −2.20, p = .035, r = −.38.

^ct(34) = −2.08, p = .045, r = −.35.

^dt(33) = −3.45, p = .002, r = −.59.

^et(33) = −3.09, p = .004, r = −.53.

^ft(14) = -2.70, p = .017, d = 1.42.

^gZ = -2.06, p = .042, η² = .28.

^ht(13) = -2.48, p = .027, d = 1.24.

Visual attention to quality indicators

Dwell times for quality indicators

Women looked longest at the indicator ‘Structure weekly multidisciplinary meeting’ (M = 8,680, SD = 6045) and shortest at ‘Recommendation by patients’ (M = 5,902, SD = 4415). This difference was significant, Z = −2.64, p = 0.008, r = −0.44. Other significant differences in dwell times to quality indicators can be found in Table 3.

Dwell time/fixation for quality indicators

Per fixation, women looked longest at ‘Counseling recovery after treatment’ (M = 262, SD = 60) and shortest at ‘Number of direct-to-implant breast reconstructions’ (M = 239, SD = 46). This difference was not significant, t(34) = -1.72, p = 0.095. Significant differences in dwell time/fixation to quality indicators can be found in Table 4.

Visual attention to hospitals

Dwell times for hospitals

Women looked longest at Hospital C (M = 10,312, SD = 6896) and shortest at Hospital B (M = 9,303, SD = 6534). This difference was not significant, Z = -1.60, p = 0.109.

On average, women looked 10,801 ms (SD = 6814) at the hospital of choice, and 9471 ms (SD = 6456) at the hospitals not chosen. This difference was not significant, Z = -1.72, p = 0.086.

Dwell time/fixation for hospitals

Per fixation, women looked longest at Hospital C (M = 257, SD = 61) and shortest at Hospital A (M = 230, SD = 47). This difference was significant, t(33) = -3.45, p = 0.002, r = −0.59. Women also looked significantly longer per fixation at Hospital C (M = 257, SD = 61) than at Hospital B (M = 235, SD = 52), t(33) = -3.09, p = 0.004, r = −0.53.

On average, per fixation women looked 249 ms (SD = 63) at the hospital of choice, and 235 ms (SD = 49) at the hospitals not chosen. This difference was not significant, Z = -1.20, p = 0.231.

Visual attention and perceived relevance of indicators

Dwell times

On average, women looked 7973 ms (SD = 5535) at indicators perceived most relevant, and 7243 ms (SD = 3065) to indicators perceived least relevant. This difference was not significant, Z = -1.07, p = 0.285.

Dwell time/fixation

On average, per fixation women looked 262 ms (SD = 79) at indicators perceived most relevant, and 250 ms (SD = 45) to indicators perceived least relevant. This difference was not significant, Z = -0.95, p = 0.341.

Differences between younger and older women

No statistically significant differences existed between younger and older women in dwell times, nor in the dwell time/fixation for indicators. Likewise, no significant differences appeared for visual attention to hospitals, nor for the attributes of perceived relevance or the choice options.

Differences between patients and non-patients

Patients versus non-patients

No statistically significant differences existed between the total group of patients and the total group of non-patients for any of the eye-tracking variables.

Younger patients versus older patients

No statistically significant differences existed between younger and older patients on dwell times. However, per fixation, older patients looked longer at ‘Waiting time examination-diagnosis’, t(14) = -2.70, p = 0.017, d = 1.42, and ‘Remaining cancer tissue’, Z = -2.06, p = 0.042, r = −0.52 compared to younger patients. Moreover, per fixation, older patients looked longer at the hospital of choice than younger patients, t(13) = -2.48, p = 0.027, d = 1.24.

Younger non-patients versus older non-patients

No statistically significant differences existed between younger and older non-patients for any of the eye-tracking variables.

Discussion

This study aimed to explore how women focus their visual attention to core decision-relevant information in a hospital report card (HRC) about breast cancer care, and whether this differs between younger and older women by using eye-tracking. In the general eye-tracking literature, dwell times less than 100 ms are thought to indicate a limited depth of information processing, and dwell times longer than 500 ms an appropriate depth of information processing.⁴⁸ Longer dwell times dedicated to a choice option are known to increase the odds of choosing that option.⁴⁹ Our results showed that, overall, both younger and older women showed dwell times corresponding with adequate information processing (i.e. dwell times longer than 500 ms). Most visual attention in terms of dwell times was paid to the quality indicator ‘Structure of weekly multidisciplinary meeting’, and least visual attention to ‘Recommendation by patients’. Most visual attention in terms of dwell times per fixation was paid to the hospital that was chosen most often by participants. As for differences between younger and older women, only in the subgroup of patients several differences existed in visual attention. Per fixation, older patients looked longer at the hospital of choice, and the indicators ‘Waiting time examination-diagnosis’ and ‘Remaining cancer tissue’ than younger patients.

Per fixation, older patients looked longer at the hospital of their choice than younger patients. This implies that women spent visual attention foremost to information that they were interested in. This finding was expected based on previous eye-tracking literature that suggests that longer visual attention for a choice option increases the odds of choosing that option.⁴⁹ The finding that more visual attention was paid to information that is most interesting to women, however, should be interpreted with caution. While it was found that, per fixation, older patients looked longer at the hospital of choice than younger patients, this finding did neither hold for the quality indicators nor for total dwell times. For example, women looked longest at the indicator ‘Structure of weekly interdisciplinary meeting’. This corresponds with previous research, as this information has previously been shown to be important for patients⁵⁰ and is generally seen as an indicator of high quality healthcare.⁵¹ Women looked shortest at ‘Recommendation by patients’. However, ‘Structure of weekly interdisciplinary meeting’ was reported by 24.3% of the women to be of utmost relevance for hospital choice, while ‘Recommendation by patients’ was reported by 43.2% of women to be of utmost importance. A similar pattern was observed for, among others, ‘Structure of weekly interdisciplinary meeting’ (24.3%) versus ‘Waiting time diagnosis-treatment’ (47.2%). These patterns imply that longer dwell times are not necessarily a result of perceived relevance of indicators.

An interesting finding was that dwell times were associated with women’s health literacy: the higher one’s health literacy, the longer she viewed the information. It has been shown previously that health literacy, together with numeracy, is positively correlated with comprehension, and even is the strongest predictor of the comprehension and use of CPI.^52,53 Hence, a possible mechanism behind the correlation we found might be that women with a higher level of health literacy were better able to understand the information than women with a lower level of health literacy, leading to a higher ability and motivation to visually attend (longer) to the information. Combined with the finding that, overall, no statistically significant differences existed between younger and older women, including the non-patients, the correlation between health literacy and dwell times might suggest that the level of health literacy, rather than (chronological) age, might be an important characteristic to take into account when developing HRCs.

It should be noted, however, that in light of the more general categories of quality of care (i.e. process, structure, and outcome), and especially in terms of perceived relevance of indicators, our explorative results suggest that some non-significant differences between younger and older women may nevertheless be (clinically) relevant. For example, as for the outcome indicators, older women seemed to attach most importance to hospitals’ performance regarding remaining cancer tissue (i.e., this indicator was reported most often to be of utmost importance), which is information that concerns the patient directly, while younger women seemed to attach most importance to performance in terms of other patients’ experiences (see Table 1). This implies that gaining insight into what general category of quality of care matters most to the individual patient, for instance by including a values clarification exercise in HRCs, and to use this insight in developing the content of HRCs might enhance patients’ motivation to process the information.

Importantly, within the group of breast cancer patients (but not within the group of non-patients), several differences existed in visual attention. These findings show that older patients look longer at the information of interest: older women – compared to younger women – looked longer at their hospital of choice, and at two indicators that were the top 2 most important indictors to them (see Table 1). Hence, older patients might particularly efficiently concentrate on and process information that they are most interested in. It might be that their previous experience played a role in this, as we did not find this pattern overall nor in the group of non-patients. Non-patients might know less well what is important to them, leading them to divide their attention over all information elements. Younger patients may have experience with the indicators, but may rely less on heuristics compared to their older counterparts.^27,28 Partly because of a decline in the efficiency of deliberative reasoning, older adults tend to rely more on intuitive reasoning, with affect playing a more prominent role.⁵⁴ Relying on intuitive reasoning does not necessarily lead to suboptimal decision-making, since older adults generally have accumulated experience in the health domain, and seem to know quite well what they find important.^55–58 Hence, reliance on intuitive reasoning might facilitate older patients’ judgment of relevance of options and attributes.⁵⁷ Our explanations of findings should be interpreted with caution, as the subgroups of patients versus non-patients were overall quite small.

Limitations

First, although the sample size was adequate for an eye-tracking study, our sample was relatively homogeneous on background characteristics (e.g. educational level), limiting the generalizability of results. Second, in analyzing perceived relevance of indicators, we used a somewhat arbitrary cut-off for ‘irrelevant’ indicators. For some women we needed to make the calculations with indicators both perceived to be ‘not important’ and ‘somewhat important’. Third, the survey used a rating scale instead of a ranking scale to measure perceived relevance, which might have led to the ceiling effect mentioned before.⁵⁹ Overall, our exploratory study focused on one specific HRC type under specialized controlled circumstances, which may limit generalizability of findings to HRCs in general.

Practice implications and Future research

To our knowledge, the question of visual attention to information in HRCs has remained unexplored, and our study can therefore inform new research into this field. Given the conditions in which we provided the information, i.e. a well-ordered concise table with a limited number of hospitals, the information in the HRC seemed to be adequately processed by both younger and older women. Besides, women with a higher level of health literacy looked longer at the HRC than women with a lower level of health literacy. Practically, this implies that differences in level of health literacy, rather than (chronological) age, need to be taken into account when designing and/or updating HRCs. This can be done by testing HRCs among people with lower levels of health literacy. It should be noted, however, that typical CPI in HRCs is more elaborate and complex.^8,42,60–65 Moreover, we found that in the group of patients, older participants seemed to rely more on elaborate use of their knowledge about what is important to them. This might mean that CPI can easily become too complex for younger patients, as they often lack accumulated experiences on what matters most to them. Hence, this information should always be carefully tested among patients before making it publicly available. As younger and older women also seem to attach importance to different general categories of quality of care, adding a values clarification exercise to HRCs might enhance patients’ motivation and ability to process the information.

Conclusion

Visual attention was not related to hospital choice and the perceived relevance of quality indicators. Visual attention of younger and older women did not seem to differ. However, in the group of patients, older women seemed to process information more elaborately than younger women, which may be related to a reliance on heuristics in light of their prior experience and knowledge.

Footnotes

Acknowledgements

We thank the Dutch Breast Cancer Association and PanelCom for their help with recruiting participants. We also thank drs. Anne de Lange for her help with the data collection.

Authors’ contributions

NGY contributed to the design of the work, the acquisition, analysis and interpretation of data, and writing the manuscript.

DT contributed to the design of the work, interpretation of data, and giving feedback on the manuscript.

JW contributed to the design of the work, interpretation of data, and giving feedback on the manuscript.

OD contributed to the design of the work, the acquisition, analysis and interpretation of data, and giving feedback to the manuscript. All authors read and approved the final manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was granted by the Dutch Cancer Society (KWF). Financial support for this study was provided entirely by a grant from the Dutch Cancer Society (KWF). The funding agreement ensured the authors’ independence in designing the study, interpreting the data, writing, and publishing the report.

Ethical approval

Ethical approval was obtained from the Institutional Review Board of the VU Medical Center as a non-WMO (Medical Research Involving Human Subjects Act) research (FWA00017598), and the Ethics Committee of the Amsterdam School of Communication Research, University of Amsterdam (2017-PC-7570). Written informed consent was obtained from participants.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

ORCID iD

Nida Gizem Yilmaz

Appendix

References

Emmert

Wiener

. What factors determine the intention to use hospital report cards? The perspectives of users and non-users. Patient Educ Couns 2017; 100: 1394–1401. DOI: 10.1016/j.pec.2017.01.021

Chimonas

Fortier

, et al. Facts and fears in public reporting: patients’ information needs and priorities when selecting a hospital for cancer care. Med Decis Making 2019; 39: 632–641. DOI: 10.1177/0272989x19855050

Akeel

Mundy

. Re-thinking technology and its growing role in enabling patient empowerment. Health Informatics J 2019; 25: 1278–1289. DOI: 10.1177/1460458217751013

Emmert

Hessemer

Meszmer

, et al.

Do German hospital report cards have the potential to improve the quality of care?

Health Policy 2014; 118: 386–395. DOI: 10.1016/j.healthpol.2014.07.006

Faber

Bosch

Wollersheim

, et al. Public reporting in health care: how do consumers use quality-of-care information? A systematic review. Med Care 2009; 47: 1–8. DOI: 10.1097/MLR.0b013e3181808bb5

Tao

Shao

Wang

, et al. Integrating usability and social cognitive theories with the technology acceptance model to understand young users’ acceptance of a health information portal. Health Informatics J 2020; 26: 1347–1362. DOI: 10.1177/1460458219879337

Damberg

McNamara

. Postscript: research agenda to guide the next generation of public reports for consumers. Med Care Res Rev 2014; 71: 97S–107s. DOI: 10.1177/1077558714535982

Schlesinger

Kanouse

Martino

, et al. Complexity, public reporting, and choice of doctors: a look inside the blackest box of consumer behavior. Med Care Res Rev 2014; 71: 38S–64s. DOI: 10.1177/1077558713496321

Totten

Wagner

Tiwari

, et al. Closing the quality gap: revisiting the state of the science (vol. 5: public reporting as a quality improvement strategy). Evid Rep Technol Assess (Full Rep) 2012; 5: 1–645.

10.

Carpenter

Niedenthal

. Emotional processes in risky and multiattribute health decisions. Psychol Health 2018; 33: 58–76. DOI: 10.1080/08870446.2017.1314478

11.

Hibbard

Slovic

Jewett

. Informing consumer decisions in health care: implications from decision-making research. Milbank Q 1997; 75: 395–414. DOI: 10.1111/1468-0009.00061

12.

Fischer

Pelka

Riedl

. Understanding patients’ decision-making strategies in hospital choice: literature review and a call for experimental research. Cogent Psychology 2015; 2: 1116758.

13.

Husain

Bays

. Changing concepts of working memory. Nat Neurosci 2014; 17: 347–356. DOI: 10.1038/nn.3655

14.

Boyce

Dixon

Fasolo

, et al. Choosing a high-quality hospital. The role of nudges, scorecard design and information. London, England, UK: The King’s Fund, 2010.

15.

Glöckner

Herbold

. An eye-tracking study on information processing in risky decisions: evidence for compensatory strategies based on automatic processes. J Behav Decis Mak 2011; 24: 71–98.

16.

Timmermans

. The impact of task complexity on information use in multi-attribute decision making. J Behav Decis Mak 1993; 6: 95–111. DOI: 10.1002/bdm.3960060203

17.

Gigerenzer

Goldstein

. Reasoning the fast and frugal way: models of bounded rationality. Psychol Rev 1996; 103: 650–669. DOI: 10.1037/0033-295x.103.4.650

18.

Victoor

Delnoij

DMJ

Friele

, et al. Determinants of patient choice of healthcare providers: a scoping review. BMC Health Serv Res 2012; 12: 272.

19.

Payne

Bettman

Johnson

. Behavioral decision research: a constructive processing perspective. Annu Rev Psychol 1992; 43: 87–131. DOI: 10.1146/annurev.ps.43.020192.000511

20.

Weber

Johnson

. Mindful judgment and decision making. Annu Rev Psychol 2009; 60: 53–85. DOI: 10.1146/annurev.psych.60.110707.163633

21.

Shirk

Crespi

Saucedo

, et al. Does patient preference measurement in decision aids improve decisional conflict? a randomized trial in men with prostate cancer. Patient 2017; 10: 785–798. DOI: 10.1007/s40271-017-0255-7

22.

NationalCancerInstitute . Age and cancer risk, 2021. https://www.cancer.gov/about-cancer/causes-prevention/risk/age (accessed 27 Dec 2021).

23.

Delgado-Guay

De La Cruz

Epner

. I don’t want to burden my family’: handling communication challenges in geriatric oncology. Ann Oncol 2013; 24 Suppl 7(Suppl 7): vii30–vii35. DOI: 10.1093/annonc/mdt263

24.

Jansen

van Weert

van der Meulen

, et al. Recall in older cancer patients: measuring memory for medical information. Gerontologist 2008; 48: 149–157. DOI: 10.1093/geront/48.2.149

25.

Posma

van Weert

JCM

Jansen

, et al. Older cancer patients’ information and support needs surrounding treatment: an evaluation through the eyes of patients, relatives and professionals. BMC Nurs 2009; 8: 1. DOI: 10.1186/1472-6955-8-1

26.

Sparks

Nussbaum

. Health literacy and cancer communication with older adults. Patient Educ Couns 2008; 71: 345–350. DOI: 10.1016/j.pec.2008.02.007

27.

Bopp

Verhaeghen

. Aging and verbal memory span: a meta-analysis. J Gerontol B Psychol Sci Soc Sci 2005; 60: P223–P233. DOI: 10.1093/geronb/60.5.p223

28.

Hasher

Zacks

. Working memory, comprehension, and aging: a review and a new view. Psychol Learning Motivation 1988; 22: 193–225.

29.

Parrott

Raup Krieger

Silk

, et al. Aging adults and online cancer information: promises and pitfalls in an era of genomic health care. In: Sparks

O’Hair

Kreps

(eds), Cancer, communication and aging. New York, NY: Hampton Press, 2013, pp. 47–66.

30.

Stine-Morrow

EAL

Soederberg Miller

Gagne

, et al. Self-regulated reading in adulthood. Psychol Aging 2008; 23: 131–153. DOI: 10.1037/0882-7974.23.1.131

31.

Romano Bergstrom

Olmsted-Hawala

Jans

. Age-related differences in eye tracking and usability performance: website usability for older adults. Int J Human-Comp Interac 2013; 29: 541–548.

32.

Nguyen

van Weert

JCM

Bol

, et al. Tailoring the mode of information presentation: effects on younger and older adults’ attention and recall of online information. Hum Commun Res 2017; 43: 102–126.

33.

Bass

Gordon

, et al. Using eye tracking and gaze pattern analysis to test a “dirty bomb” decision aid in a pilot RCT in urban adults with limited literacy. BMC Med Inform Decis Mak 2016; 16: 67. DOI: 10.1186/s12911-016-0304-5

34.

King

Hochheiser

Visweswaran

, et al. Eye-tracking for clinical decision support: a method to capture automatically what physicians are viewing in the EMR. AMIA Jt Summits Transl Sci Proc 2017; 2017: 512–521.

35.

Lang

. The limited capacity model of mediated message processing. J Commun 2000; 50: 46–70.

36.

Wedel

Pieters

. Eye fixations on advertisements and memory for brands: a model and findings. Mark Sci 2000; 19: 297–312.

37.

Orquin

Mueller Loose

. Attention and choice: a review on eye movements in decision making. Acta Psychol 2013; 144: 190–206.

38.

van Vliet

van der Wall

Albada

, et al. The validity of using analogue patients in practitioner–patient communication research: systematic review and meta-analysis. J Gen Intern Med 2012; 27: 1528–1543. DOI: 10.1007/s11606-012-2111-8

39.

Jorgensen

Young

Harrison

, et al. Unmet supportive care needs in colorectal cancer: differences by age. Support Care Cancer 2012; 20: 1275–1281. DOI: 10.1007/s00520-011-1214-9

40.

Shenkin

Harrison

Wilkinson

, et al. Systematic reviews: guidance relevant for studies of older people. Age Ageing 2017; 46: 722–728. DOI: 10.1093/ageing/afx105

41.

Damman

Hendriks

Rademakers

, et al. Consumers’ interpretation and use of comparative information on the quality of health care: the effect of presentation approaches. Health Expect 2012; 15: 197–211. DOI: 10.1111/j.1369-7625.2011.00671.x

42.

Kurtzman

Greene

. Effective presentation of health care performance information for consumer decision making: a systematic review. Patient Educ Couns 2016; 99: 36–43. DOI: 10.1016/j.pec.2015.07.030

43.

Fasolo

Reutskaja

Dixon

, et al. Helping patients choose: how to improve the design of comparative scorecards of hospital quality. Patient Educ Couns 2010; 78: 344–349. DOI: 10.1016/j.pec.2010.01.009

44.

Meppelink

van Weert

JCM

Haven

, et al. The effectiveness of health animations in audiences with different health literacy levels: an experimental study. J Med Internet Res 2015; 17: e11. DOI: 10.2196/jmir.3979

45.

Pander Maat

Essink-Bot

Leenaars

KEF

, et al. A short assessment of health literacy (SAHL) in the Netherlands. BMC Public Health 2014; 14: 990. DOI: 10.1186/1471-2458-14-990

46.

Barber

Opitz

Martins

, et al. Thinking about a limited future enhances the positivity of younger and older adults’ recall: support for socioemotional selectivity theory. Mem Cognit 2016; 44: 869–882. DOI: 10.3758/s13421-016-0612-0

47.

Reyna

. A theory of medical decision making and health: fuzzy trace theory. Med Decis Making 2008; 28: 850–865. DOI: 10.1177/0272989x08327066

48.

Tullis

Albert

. Chapter 7 - Behavioral and Physiological Metrics. In: Tullis

Albert

(eds), Measuring the user experience. Second Edition. Boston: Morgan Kaufmann, 2013, pp. 163–186.

49.

Galesic

Tourangeau

Couper

, et al. Eye-tracking data: new insights on response order effects and other cognitive shortcuts in survey responding. Public Opin Q 2008; 72: 892–913. DOI: 10.1093/poq/nfn059

50.

de Kok

Sixma

HJM

van der Weijden

, et al. A patient-centred instrument for assessment of quality of breast cancer care: results of a pilot questionnaire. Qual Saf Health Care 2010; 19: e40–e40. DOI: 10.1136/qshc.2007.025890

51.

Mohammed

Nolan

Rajjo

, et al. Creating a patient-centered health care delivery system: a systematic review of health care quality from the patient perspective. Am J Med Qual 2016; 31: 12–21. DOI: 10.1177/1062860614545124

52.

Greene

Hibbard

Tusler

. How much do health literacy and patient activation contribute to older adults’ ability to manage their health? Washington DC, USA: Citeseer, 2005.

53.

Hibbard

Peters

Dixon

, et al. Consumer competencies and the use of comparative quality information: it isn't just about literacy. Med Care Res Rev 2007; 64: 379–394.

54.

Peters

. Aging-related changes in decision making. The aging consumer. Oxfordshire, England, UK: Routledge, 2011, pp. 97–124.

55.

Tentori

Osherson

Hasher

, et al. Wisdom and aging: irrational preferences in college students but not older adults. Cognition 2001; 81: B87–B96. DOI: 10.1016/S0010-0277(01)00137-8

56.

Fraenkel

Cunningham

Peters

. Subjective numeracy and preference to stay with the status quo. Med Decis Making 2015; 35: 6–11. DOI: 10.1177/0272989x14532531

57.

Kim

Hasher

. The attraction effect in decision making: superior performance by older adults. Q J Exp Psychol A 2005; 58: 120–133. DOI: 10.1080/02724980443000160

58.

Strough

de Bruin

Peters

. New perspectives for motivating better decisions in older adults. Front Psychol 2015; 6: 783. DOI: 10.3389/fpsyg.2015.00783

59.

Ryan

Farrar

. Using conjoint analysis to elicit preferences for health care. Bmj 2000; 320: 1530–1533. DOI: 10.1136/bmj.320.7248.1530

60.

Damman

van den Hengel

van Loon

AJM

, et al. An international comparison of web-based reporting about health care quality: content analysis. J Med Internet Res 2010; 12: e8. DOI: 10.2196/jmir.1191

61.

Hibbard

Greene

Daniel

. What is quality anyway? Performance reports that clearly communicate to consumers the meaning of quality of care. Med Care Res Rev 2010; 67: 275–293. DOI: 10.1177/1077558709356300

62.

Hussey

Luft

McNamara

. Public reporting of provider performance at a crossroads in the United States: summary of current barriers and recommendations on how to move forward. Med Care Res Rev 2014; 71: 5S–16s. DOI: 10.1177/1077558714535980

63.

Peters

Dieckmann

Dixon

, et al. Less is more in presenting quality information to consumers. Med Care Res Rev 2007; 64: 169–190. DOI: 10.1177/10775587070640020301

64.

Sander

Emmert

Dickel

, et al. Information presentation features and comprehensibility of hospital report cards: design analysis and online survey among users. J Med Internet Res 2015; 17: e68. Original Paper 16.03.2015. DOI: 10.2196/jmir.3414

65.

Zwijnenberg

Hendriks

Damman

, et al. Understanding and using comparative healthcare information; the effect of the amount of information and consumer characteristics and skills. BMC Med Inform Decis Mak 2012; 12: 101. DOI: 10.1186/1472-6947-12-101