Presenting self-monitoring test results for consumers: the effects of graphical formats and age

Abstract

Objective

To examine the effects of graphical formats and age on consumers’ comprehension and perceptions of the use of self-monitoring test results.

Methods

Participants (36 older and 36 young adults) were required to perform verbatim comprehension and value interpretation tasks with hypothetical self-monitoring test results. The test results were randomly presented by four reference range number lines: basic, color enhanced, color/text enhanced, and personalized information enhanced formats. We measured participants’ task performance and eye movement data during task completion, and their perceptions and preference of the graphical formats.

Results

The 4 graphical formats yielded comparable task performance, while text/color and personalized information enhanced formats were believed to be easier and more useful in information comprehension, and led to increased confidence in correct comprehension of test results, compared with other formats (all p’s < .05). Perceived health risk increased as the formats applied more information cues (p = .008). There were age differences in task performance and visual attention (all p’s < .01), while young and older adults had similar perceptions for the 4 formats. Personalized information enhanced format was preferred by both groups.

Conclusions

Text/color and personalized information cues appear to be useful for comprehending test results. Future work can be directed to improve the design of graphical formats especially for older adults, and to assess the formats in clinical settings.

Introduction

The healthcare domain is facing great challenge due to high prevalence of chronic diseases¹ and the growing ageing population.² In China, there are 300 million people suffering from various chronic diseases³ and 220 million people aged 60 years or above.⁴ It has been recommended that self-monitoring of personal health indicators (e.g., heart rate, blood pressure and blood glucose) can be an effective approach to support the self-management of varied chronic conditions (e.g., hypertension and diabetes) and to meet healthcare requirements from these population.^5–7

With the rapid advancement of information technology, consumer-oriented health information technologies (CHITs) are emerging as convenient tools to track, record, and manage self-monitoring results for a wide range of consumers.^8–11 The growing popularity of CHITs has led to increased availability and accessibility of self-monitoring results for consumers. A typical self-monitoring test result may include such information as a numeric value, representing current indicator level, accompanied by test name, the unit of measurement, and associated reference range. Consumers may compare the result with the reference range to establish whether the result is within normal range or not, and compare the result with that from healthy individuals. However, consumers usually experience difficulty in comprehending and correctly responding to health information, partly due to inappropriate or poor presentation of the information.^12–14 This is a significant concern, as inappropriate information presentation may lead to confusion, frustration and disruption in healthcare process,¹⁵ and even to adverse consequences, such as medication error and inappropriate healthcare decision-making.^16,17

In fact, there is much evidence that consumers, especially those with low numeracy and literacy skills, have difficulty understanding quantitative health information.^18–21 The situation can be even worse among older adults,^21,22 as they have deteriorated memory and cognitive ability, are less educated, and consequently have poorer knowledge and skills at interpreting health information.^20,23,24 This means that optimal methods for presenting health information may differ between young and older adults. While older adults are in urgent need of understanding their health status, they are a group that we know little with regard to optimal presentation of health information.

The way health information is presented can have a profound influence on what the information is processed, assessed and perceived, a phenomenon known as the representational effect.²⁵ In spite of conventional use of numerical displays and tables for medical test results, it has been increasingly recognized that graphical formats may be better alternatives to present quantitative health information and likely improve comprehension.^{12–14,26–34} For example, Zikmund-Fisher et al¹⁴ found that number line graphs, compared with tables, can help patients distinguish between urgent and non-urgent deviations in laboratory test results and likely increase the meaningfulness of the test results. Use of graphs is also strongly recommended by the International Patient Decision Aids Standards Committee in order to improve comprehension and informed medical decision making.³⁵

One advantage of using graphs is that they are able to improve comprehension by integrating simple information cues as decision aids. The underlying mechanism is that information cues can attract people’s visual attention, which is particularly important for comprehension.^36,37 It has been well recognized that eye movements tracking data can provide objective assessment of how specific visual information cues can attract visual attention.^38,39 There are various information cues applied within and outside of health informatics domain, among which, color,^{14,29,33,40,41} text,^14,33,42,43 and personalized information cues^32,44,45 are widely used. However, there has been little research about which kinds of and how information cues are most likely to achieve the necessary level of comprehension, especially in self-monitoring tests for consumers. There is also a lack of research to describe how consumers perceive different graphs for their self-monitoring results (e.g., whether a particular graph is useful or not in their healthcare). Moreover, previous health informatics studies provide little evidence on how graphical health information attracts visual attention and affect information comprehension.³⁸ Finally, although some studies had included older adults,^12,29,46 virtually few have specifically examined this group and made empirical comparison in relation to their younger counterparts. Knowledge of age-related differences in the effects of graphical formats is therefore unavailable, yet essential to the design of health information presentation for older adults.

The main objective of this study was to examine the effects of four types of graphical formats (each with varied information cues) on consumers’ comprehension, perceptions, visual attention and preference of the graphs in hypothetical self-monitoring test scenarios. In addition, the present study aimed to assess whether there are age-related differences in such effects.

Methods

Participants

A convenience sample of participants were recruited from a university campus and its local communities by poster announcement. Participants were included if they aged 18 - 40 years (young adults), or 60 years or above (older adults), had self-reported normal color vision, received primary school education or above, and had normal mobility and mental alertness. We also assessed participants’ perceived health literacy using Chew’s three health literacy screening items.⁴⁷ A minimal sample size of 58, calculated by G*Power software,⁴⁸ was required to detect a medium effect size of 0.3 between age groups or graphical formats based on mixed design repeated-measures analysis of variance (ANOVA) when statistical power and level of significance were selected at 80% and 5%, respectively. In this study, 72 participants were recruited and equally allocated into young and older adult groups. Participants’ demographic information is shown in Table 1. The study was approved by the Institutional Review Board of Shenzhen University. Informed consent was obtained from each of the participants.

Table 1.

Participant Demographics

Characteristics	Young adults	Older adults
n	36	36
Gender
Male	18	14
Female	18	22
Mean age in years	22.3 (2.6)	65.8 (3.64)
Education
Primary school	0	1
Secondary school	0	24
University	36	11
Diagnose of chronic disease
Diabetes	0	7
Hypertension	0	35
Health literacy
Low-moderate	32	30
High	4	6

Materials and Tasks

Four graphical formats were applied to results for 2 self-monitoring tests. One was self-monitoring of blood pressure, presenting results for diastolic and systolic blood pressure; another was self-monitoring of blood glucose, presenting results for fasting blood glucose and two hours postprandial blood glucose. The 2 self-monitoring tests are routine self-care activities that chronically ill patients (especially those with hypertension and/or diabetes) would usually perform in their daily life.⁴⁹ Each of the two self-monitoring tests was presented in both normal (i.e., test results lied in normal range) and abnormal health conditions (i.e., test results were beyond normal range).

Figure 1 illustrates examples of a translated version of the 4 graphical formats for self-monitoring results of blood glucose (See Supplementary Appendix 1 for the original Chinese version). All the 4 graphical formats were created by using reference range number lines (RRNLs), which are often applied for displaying individual test results.^12,14,15,49 Information presented in the graphs included test name, exact test value, the unit of measurement, and cut-off points for normal range. The circles above the number line designated the location of the test result, and numbers in the circles indicated the exact test value. Reference information of normal range was provided at the bottom of the graphs. The 4 graphical formats varied with information cues and were described as follows. Type A (termed as basic graph) used non-color number line format only. Type B (termed as color enhanced graph) applied color on Type A, with green indicating normal and red indicating abnormal range. Type C (termed as color/text enhanced graph), based on Type B, provided additional explicit text explanation for the color to indicate whether the test result was normal or not. Type D (termed as personalized information enhanced graph), based on Type C, provided additional personalized information that was assumed to be an average value of the test results from population with the same sex and age as the participants. Four areas of interest (AOIs) were drawn for each graph (Please see an example in Figure 2) to examine participants’ visual attention within the areas. The first 3 AOIs covered areas where information cues were located (AOI 1 for personalized information, AOI 2 for color cue, and AOI 3 for text cue), while the fourth one (AOI 4) covered the area of reference information of normal range.

Figure 1.

Examples of the 4 graphical formats for self-monitoring test results of blood glucose.

Figure 2.

Examples of the 4 areas of interest (AOIs) drawn on graphical formats (using self-monitoring test results of blood glucose as an example: AOI 1 covered area for personalized information, AOI 2 for color cue, AOI 3 for text cue, and AOI 4 for reference information of normal range).

The experimental test included verbatim comprehension tasks and value interpretation tasks, which were designed based on the typical healthcare process with self-monitoring results. In the verbatim comprehension tasks, participants were asked to indicate the exact test values presented on the graphs, while value interpretation tasks required participants to indicate whether the test values were normal or not by yes/no forced-choice questions.

Design and Procedures

This study implemented a two-factor (2 × 4) mixed design, varying graphical formats within subjects and age between subjects. Task scenarios were programmed using E-prime 2.0 and displayed on a DELL computer (Screen size: 23 inches; resolution: 1024 × 768). The eye tracker was equipped at the lower edge of the computer screen. The premise was that the self-monitoring results presented in the graphs were taken from participants themselves. This premise was well communicated to participants before the experimental test. The use of hypothetical test scenarios is widely adopted in previous studies within health informatics domain.^14,40,50,51

The experiment was conducted in a university laboratory. Participants were instructed to sit at a fixed distance of approximately 50 cm from the computer screen and to calibrate the eye tracker. Following several practice trials, participants were required to respond to the experimental tasks as quickly and accurately as possible. They were presented with the graphs and asked to report answers verbally to an experimenter in both tasks. Combinations of graphical formats, type of tests and health conditions were randomized in a full factorial design. Each combination was presented once. After the experimental tasks, participants were required to complete a paper-based questionnaire to elicit their response to perception measures and preference. The whole experiment took approximately 40 min.

Dependent Measures

A set of measures, including user performance (i.e., task completion time and accuracy rate), perceptions (i.e., perceived health risk, perceived ease of understanding, perceived usefulness and perceived confidence), eye movement measures (i.e., time to first fixation and total fixation duration) and preference, were used to assess various graphical formats across age groups. Task completion time referred to the total time a participant spent answering questions in experimental tasks. Accuracy rate was calculated as the proportion of correctly answering the questions in the tasks.

All perception measures, including perceived health risk (“How risky do you think your health status is based on self-monitoring result on this graphical format?”), perceived ease of understanding (“How easy do you think you can understand self-monitoring test results by this graphical format?”), perceived usefulness (“How useful do you think this graphical format can facilitate the comprehension of self-monitoring test results?”) and perceived confidence (“How confident do you think you can correctly understand self-monitoring results by this graphical format?”) were rated on 7-point Likert-type scales, where 1 = not at all and 7 = extremely. Time to first fixation and total fixation duration were commonly used as indicators of visual attention within AOIs during task performance and were recorded using a Tobii X-120 eye tracker (Tobii Technology, Stockholm, Sweden) in verbatim comprehension tasks. A shorter time to first fixation and a longer total fixation duration in an AOI indicate that the AOI attracts more visual attention. User preference was assessed by asking participants to choose their most preferred graphical format.

Data Analysis

Mixed-design repeated-measures ANOVAs were used to determine the effects of graphical formats and age on performance and perception measures, and to assess the effects of graphical formats, type of AOI and age on eye movement measures. Post hoc multiple comparisons were performed with Bonferroni adjustment where necessary. In order to determine if gender, education level and health literacy could possibly make differences in outcome measures, sensitivity analyses were performed for these three factors. In particular, original ANOVAs were adjusted with gender, education level and health literacy as covariates in analysis. It was found that gender, education level and health literacy did not have significant effects on any of the outcome measures (all p-values > .05). Therefore, the unadjusted ANOVAs results were reported in this study. Exact multinomial test was performed to assess the difference in participants’ preference. Statistical analyses were performed with IBM SPSS 22 (Chicago, Illinois, USA) and R 4.1.5 software (www. r-project.org).

Results

Performance Measures

Table 2 presents ANOVA results for task performance. Graphical formats had no significant effects on task completion time and accuracy rate in both tasks. Age yielded a significant effect on task completion time (F (1, 68) = 68.52, p < .001) and accuracy rate (F (1, 68) = 11.28, p = .001) in verbatim comprehension tasks, and a significant effect on task completion time (F (1, 68) = 47.44, p < .001) and accuracy rate (F (1, 68) = 12.86, p = .001) in value interpretation tasks. In particular, older adults required more time and committed more errors in both tasks than young adults. There were no significant interaction effects between graphical formats and age (Figure 3).

Table 2.

Main Effects of Graphical Formats and Age on Task Completion Time and Accuracy Rate

	Time (s)				Accuracy rate (%)
	Descriptive analysis		ANOVA		Descriptive analysis		ANOVA
	Mean	SD	F value	p value	Mean	SD	F value	p value
Verbatim comprehension task
Graphical formats
Basic	20.5	9.0	1.00	0.39	81.1	29.3	1.74	0.17
Color enhanced	21.1	14.1			78.2	31.3
Color/text enhanced	21.2	11.3			76.4	30.8
Personalized information enhanced	23.8	17.8			77.6	33.8
Age group
Young	13.6	8.2	68.52	<0.001	90.1	29.4	11.28	0.001
Older	29.8	8.2			66.5	29.2
Value interpretation task
Graphical formats
Basic	4.4	2.3	0.22	0.88	87.4	19.2	2.00	0.12
Color enhanced	4.1	2.8			86.6	20.9
Color/text enhanced	4.1	2.8			85.5	19.2
Personalized information enhanced	4.4	2.5			82.2	22.5
Age group
Young	2.9	1.6	47.44	<0.001	92.7	16.8	12.86	0.001
Older	5.5	1.6			78.1	16.8

Figure 3.

Interaction effects of graphical formats and age on user performance. Error bars show standard errors.

Perception Measures

Perceived ease of understanding, perceived usefulness, and perceived confidence were significantly affected by graphical formats (all p’s < .001) but not by age (Table 3). Both color/text and personalized information enhanced graphs showed higher ease of understanding, higher perceived usefulness, and more perceived confidence than other graphs (all p’s < .05) (Figure 4).

Table 3.

Main Effects of Graphical Formats and Age on Perception Measures

	Graphical formats					Age
		Descriptive analysis		ANOVA		Levels	Descriptive analysis		ANOVA
Measures	Levels	Mean	SD	F value	p value		Mean	SD	F value	p value
Perceived health risk	Basic	3.5	1.1	4.35	0.008	Young	3.6	1.0	0.56	0.46
	Color enhanced	3.6	1.1			Older	3.8	1.0
	Color/text enhanced	3.7	1.0
	Personalized information enhanced	3.9	1.3
Perceived ease of understanding	Basic	4.6	1.5	41.21	<0.001	Young	5.6	1.0	0.20	0.66
	Color enhanced	5.6	1.5			Older	5.8	1.0
	Color/text enhanced	6.3	1.5
	Personalized information enhanced	6.3	1.5
Perceived usefulness	Basic	4.5	1.5	31.98	<0.001	Young	5.3	1.1	1.20	0.28
	Color enhanced	5.2	1.6			Older	5.5	1.1
	Color/text enhanced	5.8	1.2
	Personalized information enhanced	6.0	1.0
Perceived confidence	Basic	4.7	1.5	26.29	<0.001	Young	5.4	1.0	0.55	0.46
	Color enhanced	5.4	1.4			Older	5.6	0.9
	Color/text enhanced	5.9	1.2
	Personalized information enhanced	6.1	0.9

Figure 4.

Interaction effects of graphical formats and age on perception measures. Error bars show standard errors.

Perceived health risk was significantly affected by graphical formats (F (2.607, 177.262) = 4.35, p = .008) but not by age. Perceived health risk increased as more information cues were applied. Personalized information enhanced graph showed the highest health risk compared with other graphs (Figure 4).

Eye Movement Measures

Time to first fixation was significantly affected by AOI (F(1.37, 23.22) = 8.32, p = .005) and age (F(1, 17) = 27.24, p < .001), but not by graphical formats (Table 4). Both AOI 2 and AOI 4 yielded longer time to first fixation than AOI 1. Older adults required more time (5.2s) for first fixation in AOIs than young adults (2.1s). A significant interaction effect between AOI and age was also observed (F(1.37, 23.22) = 11.66, p = .001). While all AOIs had similar time to first fixation for young adults, AOI 4 yielded longer time to first fixation than other AOIs for older adults (Figure 5).

Table 4.

Main Effects of Graphical Formats, Area of Interest and Age on Time to First Fixation and Total Fixation Duration

	Time to first fixation (s)				Total fixation duration (s)
	Descriptive analysis		ANOVA		Descriptive analysis		ANOVA
	Mean	SD	F value	p value	Mean	SD	F value	p value
Graphical formats
Basic	3.4	1.5	1.26	0.297	0.8	0.3	1.69	0.209
Color enhanced	4.3	3.1			1.5	1.0
Color/text enhanced	3.3	1.3			1.3	1.0
Personalized information enhanced	3.8	2.1			1.8	2.3
AOI
AOI 1	2.3	0.6	8.32	0.005	2.0	1.6	9.17	0.001
AOI 2	3.3	1.2			1.3	1.1
AOI 3	3.3	1.6			1.5	0.8
AOI 4	5.8	4.1			0.6	0.4
Age group
Young	2.1	1.2	27.24	<0.001	0.67	0.7	16.57	0.001
Older	5.2	1.2			2.1	0.7

Figure 5.

Interaction effects of graphical formats, AOI and age on eye movement measures. Error bars show standard errors.

Total fixation duration was significantly affected by AOI (F (1.70, 28.87) =9.17, p = .001) and age (F(1, 17) = 16.57, p = .001), but not by graphical formats (Table 4). AOI 4 obtained shorter total fixation duration than other AOIs (all p’s < .05). AOI 2 had shorter total fixation duration than AOI 1 (p = .012). The interaction effect between formats and AOI was also significant (F(2.20, 37.40) = 4.65, p = .013). As information cues were applied in the graphs, fixation duration in corresponding AOIs increased. Older adults yielded longer fixation duration (2.1s) than their younger counterparts (0.7s). A significant interaction effect between AOI and age was also observed (F(1.70, 28.87) = 5.43, p = .013). While all AOIs had similar fixation duration for young adults, AOI 4 yielded shorter fixation duration than other AOIs for older adults (Figure 5).

User Preference

Multinomial exact tests showed significant deviations from uniform distribution of preference for each of and combined age groups (p’s < .001) (Figure 6). Personalized information enhanced graph was the most preferred format for both young and older groups. Participants explained that this graph provided more information to consider about health status. Many older adults also preferred color/text enhanced graph because they considered the graph provided sufficient information and appeared to be concise. No participants preferred a basic graph in both groups.

Figure 6.

Distribution of participant preference by graphical formats and age.

Discussion

In spite of the popularity of CHITs and increased accessibility of various self-monitoring test records for consumers, poor health information presentation may lead to misunderstanding and confusion, and to a higher likelihood of inappropriate healthcare decision-making. In light of this, the present study represents a rare attempt to evaluate various graphical formats of self-monitoring results. Our study demonstrates that there are differences across the graphical formats and between young and older adults with respect to how the graphs are assessed and perceived. In addition, the present study also provides important evidence on consumers’ visual attention when viewing different graphical formats.

Primary Findings

Our study is one of the pioneer studies that evaluate graphical formats with varied information cues. Consistent with previous studies,^{12,13,27,30–32,38,43,51,52} our findings show that the presentation of self-monitoring results in different formats had different effects on how consumers evaluated the information. Consumers tended to consider graphs more useful and easier to understand, and possess more confidence in understanding self-monitoring results, as the graphs applied more information cues. The color applied on RRNLs, striking to the eye, is likely to give consumers an immediate impression of whether values were within normal range or not.¹³ Similarly, text and personalized information cues may serve as redundancy check, and thus likely reassure consumers in information comprehension. Therefore, graphs that used color/text cues and personalized information were favored most by consumers. The findings support the effectiveness of color, text and personalized information cues in facilitating consumers’ comprehension of self-monitoring results.

We observed only little variation in task completion time and accuracy rate between basic graph and 3 other information cue enhanced graphs. This may be due to that the 4 graphical formats had similar structure and layout, and their differences were not sufficient to influence consumers’ efficiency and effectiveness in task performance. It may imply that the information cues would not result in additional cognitive workload for consumers, though more information needs to be processed. We also noted that the similarity of the four graphical formats might have caused learning effects in task performance. In order to address this, the presentation of graphical formats was randomized across participants. In fact, combinations of graphical formats, type of tests and health conditions were presented in a random order across participants in our experimental design. Such randomization strategy could distribute the learning effects equally so as not to bias the overall results.

While the differential effects of graphical formats on risk perception have been well documented in health informatics domain,^{26,28,30,50–52} it remains unclear how information should be presented to convey appropriate health risk for consumers. Intriguingly, we found that consumers perceived higher health risk, as more information cues were integrated in the graphs. The explanation can be that with more information cues, consumers became more cautious and conservative in the evaluation of their test results, and consequently consider themselves in a higher level of health risk. This is especially the case in the scenarios where the results were beyond the normal range, and within the normal range but close to upper threshold. However, it should be noted that there is little consensus regarding which level of health risk is appropriate and should be conveyed to consumers for certain health information. Therefore, medical professionals should be consulted before the application of information presentation related to risk perception.

As noted by previous studies,^53–55 the presentation of quantitative health information may be particularly problematic for older adults. Consistently, we found that older adults performed worse in healthcare tasks for all graphical formats, as they had a longer completion time and more errors, compared with young adults. As such, it is very likely that older adults had difficulty in interpreting self-monitoring results and did misinterpret many of them. Such difficulty and misinterpretation could obviously affect how they perceived health status and should therefore be given much attention. When making comparisons across graphical formats, it was found that older adults had favorable perceptions for the graphs enhanced by color/text cues or personalized information. This suggests that graphical formats with color/text cues or personalized information could be developed to compensate the disadvantages in older adults.

This study also provided empirical evidence on how information cues in graphs attracted and maintained visual attention. Shorter time to first fixation and longer total fixation time were observed in AOIs with information cues, indicating an attraction effect of information cues. The attraction effect was particularly obvious when the color cue and personalized information were introduced. Such attraction effect may direct consumers’ visual attention to information cues that are salient and thus serve as aids for information clarification and comprehension. The finding is important, as it helps understand how information cues can facilitate comprehension through the attraction of visual attention. However, it should be noted that the longer fixation time may also result from confusion of information cues. The speculation awaits confirmation by further research.

Conventionally, designers preferred putting test values and their associated graphical presentation in the central area, accompanied by reference information of normal range in surrounding area.¹² However, we found that such reference information was less noted, especially by older adults. Even though some participants noted the information, they tended to pay attention to it after viewing other AOIs. This implies that the current presentation of reference information might be less useful in information comprehension, as it was not well noted by users. More efforts are required to find innovative ways to present test results and reference information together in a holistic way.

To achieve maximum utility and effectiveness, consumers must not only be able to comprehend the self-monitoring results, but also like the presentation format so as to continue using it. Our results suggest that personalized information enhanced graph be a particularly attractive option for both young and older adults. The reason for this seems intuitive: the graph includes both color/text cue that could facilitate comprehension and personalized information helps consumers compare health status with their healthy counterparts. Consistent with Arcia et al,³³ the finding from the present study suggests the usefulness of information-rich graph design. This finding also appears to be a counterpoint to previous recommendations for simplifying information in the delivery of health messages.⁵⁶ However, oversimplification may risk omission of important messages,⁵⁷ which could probably be avoided by the use of diverse information cues.

Our study differs from previous studies in several respects. First, few studies successfully revealed the differential effects of varied information cues.^33,34 Zikmund-Fisher et al¹⁴ considered several types of information cues in their work, including color, borderline and color gradient. However, it should be noted that Zikmund-Fisher et al applied multiple information cues in each of the 3 types of graphs. For example, the “Block Line” graph presented in their study applied color, borderline and elaborate text cue, while the “Gradient Line” graph applied color, gradient and simple text cue. Zikmund-Fisher et al actually tried to determine the differential effects of the three types of graphs, instead of making comparisons among information cues. In other words, though Zikmund-Fisher et al has successfully revealed the effectiveness of graphs in increasing the meaningfulness of the test results, the use of multiple information cues in one graph prevented specifying the differential effects of varied information cues. Unlike the Zikmund-Fisher et al study, our study applied one additional information cue for each of the graphs and made comparisons among them. This allowed for examination of the differential effects of varied information cues. Second, our study provides unique evidence on age-related differences in the effects of graphical formats. Third, the present study provided evidence on consumers’ visual attention in viewing the graphs, which was less investigated in previous studies³⁸ but particularly important in understanding the relationship between the graphical presentation and information comprehension. Finally, while previous study focused mainly on laboratory test results^12,14 or health information for care providers,^13,34 which are less encountered by consumers, our study examined self-monitoring test results that consumers may need to know on a daily basis and play increasingly important roles in the self-management of chronic conditions.

Implications

Our findings have both theoretical and practical implications for the design of health information graphs for consumers. Theoretically, our study identified the effects of graphical formats varying with simple information cues. In addition, our study emphasized the importance of appropriate graphical design for older adults to improve their performance and comprehension of health information.

From a practical perspective, our findings demonstrate advantages and disadvantages of different graphical formats in relation to a variety of measures. Providers and designers need to be aware of the differential effects on consumers’ comprehension, perceptions, visual attention and preference that may be generated through the use of different graphical formats. In particular, basic graph is comparably efficient and effective in information comprehension with other graphs, but achieves a more concise interface. To elicit favorable user perceptions, it may be better to use both color and text cues, instead of color cue only. While personalized information can reinforce consumers’ perceptions on ease of understanding, usefulness and confidence, and is preferred by consumers, such information might lead to an increased level of risk perception. Therefore, designers should be careful in presenting information for consumers in order to arouse appropriate risk perception. In addition, designers should also consider how to present and configure varied information (e.g., information cues and reference information) in a graph to attract and maintain an appropriate amount of visual attention, so as to maximize the effectiveness of the graph. Finally, older adults appear to experience difficulty in task performance even with their favorable graphs. Presentation formats that are tailored for older adults are therefore especially helpful to facilitate their healthcare activities.

Limitations

This study has limitations in that we used hypothetical rather than real test results, as we also did in previous studies.^14,40,50,51 It is likely that participants might respond differently in a real situation. Thus, clinical trials are required to confirm the effectiveness of the presentation formats. Another limitation is that all young participants had a university education. This may reduce the generalizability of our results to the population with less education. In addition, we also noted that differences existed between the 2 age groups in educational attainments due to the sample recruitment strategy (i.e., convenience sampling) adopted in this study. Such differences may make one wonder the extent to which the findings are attributable to education and not age. However, sensitivity analysis has shown that gender, education level and health literacy did not affect outcome measures, so the identified age-related differences could be considered mainly attributable to age. Moreover, it should be noted that the use of red-green color cue in our graph design may not be applicable to people with red-green color blindness. Finally, this study did not address other individual characteristics, such as numeracy skill, graph literacy, and cognition ability, which are suggested to affect comprehension.^14,38,46 Future studies are recommended to further examine these factors so as to improve understanding of individual differences in information comprehension and allow for more tailored information presentation design for specific groups.

Conclusions

Accurate comprehension of self-monitoring results and favorable perceptions of presentation formats are essential in consumers’ healthcare activities. This study demonstrates that there are differences across varied graphical formats and between young and older adults with respect to how self-monitoring results are viewed, assessed and perceived. Future work can be directed to improve the design of graphical formats to augment their effectiveness, especially for older adults, and to further assess the graphical formats in clinical settings with consumers actually using them in real self-monitoring tests.

Funding

This work received funding support from the Young Talents Foundation of Ministry of Education of Guangdong, China (grant number 2016KQNCX143), the Natural Science Foundation of SZU (grant number 827000228), and the Start-up Grant of SZU (grant number 2016041).

Contributors

JY conducted the experiment, and collected and analyzed data. DT led the project, cross-checked the data analysis and drafted the manuscript. DT, JY and XQ all contributed to the study design and protocol development, and participated in the writing and editing of several versions of this manuscript.

Competing interests

None.

SUPPLEMENTARY MATERIAL

Supplementary material is available at Journal of the American Medical Informatics Association online.