Seeking health information online: does limited healthcare access matter?

Abstract

Consumers facing barriers to healthcare access may use online health information seeking and online communication with physicians, but the empirical relationship has not been sufficiently analyzed. Our study examines the association of barriers to healthcare access with consumers’ health-related information searching on the internet, use of health chat groups, and email communication with physicians, using data from 27 210 adults from the 2009 National Health Interview Survey. Individuals with financial barriers to healthcare access, difficulty getting timely appointments with doctors, and conflicts in scheduling during clinic hours are more likely to search for general health information online than those without these access barriers. Those unable to get timely appointments with physicians are more likely to participate in health chat groups and email physicians. The internet may offer a low-cost source of health information and could help meet the heightened demand for health-related information among those facing access barriers to care.

Introduction and background

Over the past two decades, the internet has become a widely available source of information,^{1 2} as well as a key component in clinical informatics, a rapidly growing field aimed at exploring and actualizing the potential of information and communication technology to improve health outcomes.³ One prominent example is the use of patient web portals that enable patients to access their health records, exchange emails with providers, complete administrative forms, receive appointment reminders, and manage medications.⁴ In the subfield of clinical research informatics, the internet has been used for the recruitment of human subjects in clinical trial research (eg, incorporating interactive informed consent programs for population subgroups with a wide range of learning abilities and skills).⁵ In a broader context, given its low cost and prevalence, the internet serves as an increasingly important channel for delivering health-related information to consumers, both in patient–physician interactions^6–9 (eg, email exchanges) and in general health information searches (eg, health-related content on the internet).

For consumers facing barriers to healthcare access, the internet can be a particularly appealing source of health information. The inability to afford needed medical care (eg, due to financial constraints) may prompt some individuals to seek health information from the internet. For example, Pew surveys of internet use indicate that nearly two-fifths of online health information seekers use the information to diagnose their conditions, and nearly half of them perceive the information as useful in self-treating their symptoms without advice from healthcare professionals.¹ Even among those who can and are willing to pay for care, there might be delays in getting needed care for various reasons (eg, long wait times for appointments, scheduling conflicts). In such cases, online health information and virtual communication with peers or providers may be useful. For example, patient web portals have been found to improve patient–provider interactions and increase patient convenience by allowing round-the-clock ability to send messages about perceived non-urgent concerns without visiting the clinic.⁴

Previous studies have examined the association between online health information seeking and factors correlated with healthcare access such as income, insurance coverage, and travel time to the source of healthcare.^{10 11} These studies found higher likelihoods of online health information seeking among those with higher income,¹⁰ those who had health insurance,¹¹ and those with longer travel times.^{10 11} However, these factors may not fully capture the nature and extent of healthcare access barriers faced by consumers. For instance, consumers with insurance coverage face a varying level of access problems due to heterogeneity in cost-sharing and individual preferences. Moreover, studies have consistently reported that people in public insurance programs (eg, Medicaid) have difficulty finding physicians willing to take them as new patients,¹² and experience longer wait times for appointments.¹³ On the other hand, consumers may choose to carry little or no insurance coverage simply because they are healthy and predict low or no care utilization. Similarly, in addition to consumers’ access to transport, travel time may reflect other factors such as rural residence rather than healthcare access.¹⁴ Furthermore, previous research has not examined how online health information seeking is associated with different types of access problems. For example, the ability to communicate with providers through emails (potentially using web portals) may be especially useful to those who have contact with providers but face certain (non-financial) access problems, such as difficulty in scheduling appointments. On the other hand, a general online information search may appeal to those with any type of access problem. Prior literature has not made this distinction.

In this study, we examine how consumers’ online health information seeking behavior is related to different types of healthcare access barriers reported by individuals. We analyze three distinct aspects of online health information seeking: (1) general information searching on the internet (eg, using search engines and browsing websites), (2) internet-based peer-to-peer communication (participating in health chat groups), and (3) patient–physician communications (emailing physicians). Using a two-step analytic framework, we first examine the relationship between barriers to healthcare access and general online health information searching among the overall population, and then we examine the use of chat groups and email communication with care providers among health information searchers.

Methods

This study analyzes data from the National Health Interview Survey (NHIS) conducted in 2009. The NHIS is an annual survey using a nationally representative sample of the non-institutionalized adult population of the USA, with oversampling of black, Hispanic, and Asian subjects. The survey collects information on a broad range of health-related topics, including health-related uses of the internet, through in-person household interviews. The response rate was 82.2% at the household level and 65.4% at the individual level. We merged three NHIS data files for our analysis: the Sample Adult data file (demographic characteristics and health status of a randomly selected adult from each family), the Family data file (household level characteristics), and the Person data file (additional individual level characteristics, such as education and income). The final sample consists of 27 210 adults (18 years of age or older). A detailed description of the survey is available on the website of the Centers for Disease Control and Prevention.¹⁵

We examine three dependent variables in two steps. The dependent variable in the first step is a binary indicator of whether an individual used the internet to search for health information over the past 12 months (yes=1, no=0). The two dependent variables in the second step indicate the use of online chat groups (yes=1, no=0) and email communications with physicians in the past 12 months (yes=1, no=0) among individuals who used the internet to search for health information. To measure financial barriers to care access, we use two survey questions that ask the respondents whether any needed medical care had been foregone due to unaffordable (financial) cost of care, and whether care had been delayed due to (financial) cost. We consider respondents having financial barriers to care access if the answer to at least one of these two questions is ‘yes.’ To capture other types of access barriers, we construct five measures using responses to a series of survey questions asking whether the respondents experienced a delay in getting medical care in the last 12 months because they could not get through on phone (yes=1, no=0), did not have transport (yes=1, no=0), had time conflicts when the clinic was open (yes=1, no=0), could not get appointment soon enough (yes=1, no=0), or waited too long in the doctor's office (yes=1, no=0).

Our analysis includes a number of individual characteristic as covariates: age, gender, marital status, education (college vs no college), race (white vs non-white), family income, employment status, family size, and health insurance status. In addition, we also control for self-reported health ratings (excellent/very good, good, and fair/poor) and the presence of chronic conditions (yes=1, no=0) that account for a large proportion of healthcare utilization in the USA, including diabetes, hypertension, coronary artery disease, asthma, cancer, and arthritis.¹⁶ These conditions were also used in previous studies to control for both individuals’ healthcare needs and health status.^{10 11}

We estimate the following three equations in two steps:

1

rm2a

rm2b

In equation (1), y is the indicator of general health information searching on the internet. CA and NCA are the measures of financial and other barriers to access to healthcare, respectively. W's are the other covariates. F is assumed to be the logistic cumulative distribution function (CDF). In equations (2a) and (2b), z and m are the indicators for participating in health chat groups and emailing physicians, respectively. Here, the probability is conditional on y = 1 (ie, conditional on general online health information searching in the first step). The variables and CDF on the right-hand side are identical to equation (1). With the above specifications, all three equations are standard logistic models and are estimated separately by maximum likelihood. Equation (1) uses the whole sample. Equations (2a) and (2b) use the subsample of the general health information searchers. The population weights from the NHIS are applied.

Results

Descriptive statistics of the study sample are presented in table 1. Overall, 42% of the respondents report searching for health-related information on the internet during the previous 12 months. Those individuals tend to be younger, have college educations, have higher incomes, and report better health compared to those not having searched for health information online. Only a small proportion of respondents report using chat groups (3%) and email communications with doctors (4%). The proportion of respondents reporting financial barriers, long wait times for appointments, inabilities to get through on the phone, and inabilities to visit the clinic during open hours is significantly higher among general health information searchers than among those who did not search for health information online. However, health information searchers are less likely to report transport problems (1.5%) than non-searchers (3%). Among those who searched for health information online, chat group users are more likely to report all types of access barriers than non-users, with the exception of clinic visits. Finally, respondents using email to communicate with physicians are more likely to report long wait times for an appointment, inabilities to get through on the phone, and inabilities to go during open clinic hours compared to non-users.

Table 1.

Sample characteristics of respondents who used and did not use the internet for health information seeking (N=27 210)*

	Did not use the internet to seek health-related information N=15 601 (57.3%)	Used the internet for general health-information searching
		Overall N=11 609 (42.7%)	Used chat groups N=829 (3.0%)	Emailed physicians N=1076 (3.9%)
Variable	Percentage	Percentage	Percentage	Percentage
Barriers to healthcare access†
Having financial barriers	13.3	15.5‡	18.0§	14.4
Unable to contact via phone	2.5	3.3‡	4.8§	4.6§
Unable to get a timely appointment	4.8	8.7‡	15.0§	13.1§
Long wait at doctor's office	5.5	5.5	8.0§	6.1
Unable to go to clinic when open	2.8	4.9‡	6.2	6.6§
No transport	2.7	1.5‡	3.2§	1.4
Self-rated health status
Poor	17.0	8.3	14.3	10.3
Excellent or very good	53.5	68.4	62.4	66.6
Good or fair	29.5	23.3	23.4	23.1
Have one or more chronic diseases	51.3	49.8	51.5	55.9
Age (in years)
18–40	36.2	47.0	51.3	39.9
41–60	34.1	39.4	36.5	44.8
61–80	23.0	13.1	11.5	14.7
>80	6.7	0.5	0.8	0.6
Married	57.3	66.4	64.5	69.2
Minority (non-white)	21.3	15.2	20.2	13.5
Family income (US$)¶
<35 000	40.4	20.2	24.5	12.6
35 000–75 000	30.7	31.8	30.0	26.4
75 000–100 000	9.4	14.0	11.7	14.1
>100 000	12.1	28.7	28.4	42.3
Family size
≤2 (ie, less than or equal to 2)	54.0	50.9	50.2	55.1
>2 (ie, more than 2)	46.0	49.1	49.8	44.9
College education	40.4	77.0	77.5	87.9
Female	46.6	57.8	62.5	57.5
Employed	51.8	69.8	64.0	72.6
Have health insurance	77.9	87.1	85.7	92.6

*Weighted estimates derived from the National Health Interview Survey (NHIS) 2009, representative of the non-institutionalized adult population in the USA.

†Pairwise Pearson correlations between healthcare access barriers are as follows: financial barriers–unable to contact via phone (0.09), financial barriers–unable to get a timely appointment (0.10), financial barriers–long wait at doctor's office (0.07), financial barriers–unable to go to clinic when open (0.07), financial barriers–no transport (0.08), unable to contact via phone–unable to get a timely appointment (0.45), unable to contact via phone–long wait at doctor's office (0.31), unable to contact via phone–unable to go to clinic when open (0.25), unable to contact via phone–no transport (0.18), unable to get a timely appointment–long wait at doctor's office (0.37), unable to get a timely appointment–unable to go to clinic when open (0.33), unable to get a timely appointment–no transport (0.18), long wait at doctor's office–unable to go to clinic when open (0.28), long wait at doctor's office–no transport (0.19), and unable to go to clinic when open–no transport (0.18). All correlations are statistically significant at p<0.01.

‡The value is significantly different (at p<0.05) from that of those who did not report searching for health information online.

§The value is significantly different (at p<0.05) from that of those who did not report the activity (using chat groups or emailing doctors) among general health information searchers.

¶Five percent of observations were missing.

The ORs from the logistic regressions are summarized in table 2. Having financial barriers to healthcare access, being unable to get a timely appointment, and being unable to visit the clinic during working hours are positively associated with a general health information search (ORs 1.70, 1.69 and 1.47, respectively). However, not having means of transport decreases the likelihood of health information seeking (OR 0.74). Being younger, white, female, married, and employed, and having a smaller family size, college education, and high income are all positively associated with a general health information search. The socio-demographic profile of health information searchers in our sample is similar to that in prior studies.^{10 11} Our results also show that after controlling for socio-demographic variables, having chronic conditions is positively associated with health information seeking. This is consistent with previous findings.^{10 11} Among the general health information searchers, the use of chat groups and use of email to communicate with doctors is positively associated only with being unable to get a timely appointment (OR 1.87 and 1.57, respectively) but not with other barriers.

Table 2.

Logistic regression results (ORs) for general online health information searching, using chat groups, and emailing physicians

Variable	Searching for health information on the internet		Using chat groups		Emailing physicians
	OR	95% CI	OR	95% CI	OR	95% CI
Barriers to healthcare access
Having financial barriers	1.70***	1.53 to 1.90	1.03	0.80 to 1.34	1.12	0.87 to 1.42
Unable to contact via phone	0.97	0.73 to 1.29	0.81	0.50 to 1.29	1.02	0.66 to 1.58
Unable to get a timely appointment	1.69***	1.41 to 2.02	1.87***	1.36 to 2.57	1.57**	1.14 to 2.16
Long wait at doctor's office	0.86	0.72 to 1.02	1.07	0.72 to 1.59	0.91	0.63 to 1.33
Unable to go to clinic when open	1.47***	1.18 to 1.82	0.83	0.54 to 1.28	1.08	0.70 to 1.68
No transport	0.74*	0.56 to 0.98	1.59	0.88 to 2.86	1.10	0.47 to 2.54
Self-rated health status
Poor	Reference		Reference		Reference
Excellent or very good	1.43***	1.26 to 1.62	0.56***	0.40 to 0.78	0.61**	0.44 to 0.84
Good or fair	1.22***	1.08 to 1.39	0.61**	0.45 to 0.83	0.68*	0.48 to 0.94
Have one or more chronic disease	1.53***	1.40 to 1.67	1.00	0.81 to 1.24	1.20	0.98 to 1.46
Age (in years)
18–40	Reference		Reference		Reference
41–60	0.63***	0.57 to 0.69	0.81*	0.67 to 0.98	1.09	0.92 to 1.31
61–80	0.32***	0.28 to 0.36	0.75	0.55 to 1.03	1.09	0.81 to 1.45
>80	0.05***	0.04 to 0.07	1.21	0.40 to 3.69	1.23	0.43 to 3.54
Married	1.21***	1.11 to 1.33	1.03	0.81 to 1.32	0.91	0.75 to 1.10
Minority (non-white)	0.63***	0.58 to 0.69	1.36**	1.09 to 1.71	0.90	0.70 to 1.15
Family income (US$)
<35 000	Reference		Reference		Reference
35 000–75 000	1.56***	1.42 to 1.71	0.86	0.68 to 1.08	1.28***	1.03 to 1.58
75 000–100 000	1.81***	1.58 to 2.07	0.81	0.57 to 1.17	1.58***	1.16 to 2.14
>100 000	2.61***	2.30 to 2.97	1.04	0.76 to 1.42	2.40***	1.82 to 3.16
Family size
≤2 (ie, less than or equal to 2)	Reference		Reference		Reference
>2 (ie, more than 2)	0.78***	0.72 to 0.85	1.00	0.82 to 1.23	0.79**	0.67 to 0.94
College education	3.66***	3.38 to 3.96	1.11	0.88 to 1.41	2.05***	1.59 to 2.65
Female	1.78***	1.66 to 1.91	1.17	0.96 to 1.43	1.04	0.89 to 1.22
Employed	1.15**	1.06 to 1.25	0.81*	0.66 to 0.99	1.05	0.87 to 1.27
Have health insurance (yes=1)	1.61***	1.44 to 1.80	0.98	0.74 to 1.29	1.41*	1.04 to 1.89

*p<0.05; **p<0.01; ***p<0.001.

Discussion

We find that individuals with financial barriers to healthcare access, with difficulty in getting a timely appointment with doctors, and with conflicts in scheduling during clinic hours are more likely to search for general health information online than those without these access barriers. This is consistent with our expectation that the internet may serve as a low-cost source of health information, especially for those facing problems in getting care through traditional channels (eg, visiting physicians in person). However, we find that delaying needed care due to not having transport reduces the likelihood of online health information searching. Lacking transport may indicate low socio-economic status and/or residence in rural/suburban areas without convenient public transportation,¹⁷ as well as less access to the internet.¹⁸ Our data do not allow us to control for those factors (eg, regional internet connectivity) that may be correlated with having problems with accessing care due to lack of transport.

Our results show that being unable to get a timely appointment is associated with using email communications with physicians and joining online health chat groups. Emailing physicians can replace certain in-person doctor–patient interactions (eg, clarification of discharge instructions, routine follow-ups) among those already having contacts with doctors.¹⁹ Hence, for individuals having regular access to healthcare but facing disruptions or delays due to lack of timely appointments, email communications with physicians may be particularly helpful. While participation in online health chat groups is unlikely to substitute for regular doctor–patient interactions, individualized information obtained from such peer-to-peer communication may complement the information obtained from regular care. Many chat groups allow the sharing of health- and healthcare-related experiences among individuals with similar conditions, as well as the pooling of information on the quality of providers.^{20 21} As a result, consumers may have effective interactions with their providers based on the insights and experiences of others.²⁰ Notably, our findings are obtained while controlling for insurance coverage, which has previously been used to measure the costs of seeking medical advice from providers.^{10 11} Our results suggest that the self-reported (direct) measures of access barriers contain additional information on access that is not captured by insurance coverage.

The association between healthcare access barriers and online health information seeking suggests that when healthcare needs are not met, people may have a high demand for online health information. Therefore, providing reliable health-related information online may help to meet the demand for such information, which is a potentially important self-care resource for those facing barriers to care access. In our study, people facing delays in getting appointments with physicians showed increased use of chat groups and email communication with doctors. This increase may help explain, at least in part, the popularity of online platforms and portals that facilitate communication between patients and providers, as well as among patients having similar conditions. A growing body of literature demonstrates the potential of patient web portals to improve the efficiency of care delivery and increase patient satisfaction.^{4 22} Although the NHIS survey question on email communication did not refer to web portals directly, an increasing trend is the use of web portals when emailing physicians.²³ Our results suggest that patients who cannot get timely appointments with their providers may rely on internet-based interactions for prompt advice from physicians or peers, potentially leading to improvement in the overall effectiveness and efficiency of healthcare delivery. Our results support policies and practices that would further promote such interactions.^{19 24} Those practices include designing payment reimbursement policies for online consultations, ensuring the privacy and confidentiality of patient–physician online traffic, and developing guidelines for appropriate message content and reply protocols.^{19 24} Additionally, web-based virtual informed consent programs could target groups likely facing barriers to care access and limited interpersonal interactions with their physicians, which could potentially improve recruitment of those under-represented in clinical trials (eg, minorities, those with lower socio-economic status).

Finally, despite the popularity of and widespread access to the internet, less than half (42%) of the respondents in our sample had searched for health-related information online during the past 12 months. This figure is substantially lower than the estimates from the Pew survey, which found that 61% of the adult USA population sought health information online in 2009 and 59% in 2011.²⁵ The difference may be because the Pew survey conducted telephone interviews and had a low response rate (11.5%), whereas the NHIS conducted face-to-face interviews and had a much higher response rate (65.4%), making our estimates potentially less susceptible to non-response bias. Given the internet's potential as an important health information source and its significant role in promoting consumer engagement in healthcare,^6–8 our relatively low estimate suggests a need to further investigate specific reasons why people use or do not use the internet to seek health information.

Several limitations of our analysis should be noted. First, we do not know the specific types of health information sought by individuals and the frequency of information seeking. Having such data may improve our contextual understanding of online health information seeking behavior. Second, the cross-sectional design of our study does not allow us to infer causal relationships. Unobserved factors might be simultaneously correlated with both access barriers and online information seeking (eg, the general ability of organizing one's time and activities). Such factors may confound our results. Therefore, our findings are exploratory and should be interpreted with caution. Future research may focus on the types of health information searched for online by different sub-populations, as well as consumers’ online information seeking behavior over an extended time.