Internet access and use by COPD patients in the National Emphysema/COPD Association Survey

The methods of the NECA-commissioned national surveys of COPD patients have been detailed previously [15]. Briefly, participants in the survey were randomly selected from either one of four different sources, all of them sampled during the same period of time: (1) a national sample of households in which at least one person reported a diagnosis of COPD based on a database maintained by Integrated Business Services Inc. (Lake Forest, IL); (2) participants of patient support groups affiliated with the American Lung Association’s Better Breathers Clubs or NECA; (3) COPD patients receiving oxygen through a national provider (Apria Healthcare, Lake Forest, CA); and (4) respondents to an Internet survey invitation on COPD-related websites. All participants were mailed a 7-page questionnaire, accompanied by an explanatory letter and a postage-paid return envelope. Households not willing to participate in the survey, according to response to previous contacts, were excluded. The incentive for participation was an executive summary of the survey results. No economic incentives were offered. The total number of completed surveys was 1,077 out of 15,000 surveys sent (response rate 7.2%). For the purpose of the current analyses, people who reported a diagnosis of asthma or alpha-1 anti-trypsin deficiency but not COPD were excluded, leaving 914 COPD patients for analysis.

The content of the questionnaire was prepared with the help of COPD expert panels, including physicians, patients, and representatives of COPD organizations. The items were considered to have face validity and were easy to understand, and have been used in prior research efforts [15]. Questions included demographics, comorbid conditions, perceived health and limitations for daily activities, use of health-care resources including visits to physicians, and sources of information consulted about the disease. Annual household income was selected from a list of options, and recoded pooling the answers in two groups–below $30,000 and $30,000 and above. Comorbid conditions were self-reported, based on a list of conditions presented. Disease severity and impact were evaluated by the Medical Research Council (MRC) Dyspnea Index to describe severity of breathlessness [16]. The living with COPD (LCOPD) Questionnaire, a patient-reported outcome measure, was also used to evaluate the overall impact of disease. The LCOPD instrument is a 22-item scale, with scores ranging between 0–22 [17], higher scores represent poorer QOL. The frequency of visits to any doctor and pulmonologist during the prior 12 months was also evaluated. Multiple questions addressed patients’ perceptions of their needs and experiences with the healthcare system, including their level of satisfaction with their physician, medications, amount of time spent with their physician, ability to reach their physician and make appointments without delay, and their perception of how they were treated by the medical community in general. Participants also had the opportunity to select from a list of all other sources of education and information regarding their disease that they had regularly used. The full questionnaire is available as Additional file 1.

An additional specific question about Internet use (“How often do you use the Internet to get information about your condition or its treatment?”) was included, with options ranging from “at least weekly” to “only a few times per year, rarely/never”, and a final option for “no Internet access”. Based on participants’ responses to this question, two different categories were created: Internet access and frequency of Internet use. Participants who responded “no Internet access” were compared with all others (participants with access), independently of frequency of use. To evaluate potential determinants of frequency of use, the respondents with no access were excluded, and the participants with “at least weekly” use were compared with the rest of the population with Internet access.

Data are presented as proportions and means, with standard deviations (SDs) as appropriate. Statistical comparisons for categorical data were made using chi-square tests. Continuous data was compared using t-tests. We analyzed the bivariate relationships between both “lack of Internet access” or “frequent Internet use” and the sets of explanatory variables. Separate logistic models with the outcomes “lack of access” and “frequent use” were constructed, with inclusion of all sets of pre-specified explanatory variables. All p-values were 2-tailed, with a p-value less than 0.05 considered statistically significant. Analyses were performed using SAS 9.2 (SAS Institute, Cary, NC).

The current project uses de-identified data from a survey with no potential links to any subject identifier. The project was approved by the University of Michigan Medical School Institutional Review Board (Study ID HUM00076780). Novartis provided funding to conduct the survey but was not involved in the data analysis. NECA and Novartis participated in the review and approval of this manuscript. The database containing the raw survey data is maintained by Novartis and Innovative Health Solutions.

The demographic characteristics of the 914 COPD patients who responded to the survey (response rate 7.2%), along with variables relating to their disease severity and disease management, are shown in Table 1 stratified by Internet access and frequency of Internet use. Table 2 lists participants’ comorbidities and perceived needs, also stratified by Internet access and frequency of Internet use.

Compared with those with Internet access, participants without Internet access were older (no access: mean age ± s.d. 75.1 ± 7.3 years; Internet access: 69.0 ± 9.8; p < 0.0001), less educated (high school as highest level of educational attainment 57.6% vs. 33.5%, p < 0.0001), and had lower income ($30,000 or below 68.2% vs. 41.1%, p < 0.0001). Furthermore, they were more frequently Medicare-insured (81.2% vs. 71.7%, p = 0.001), and less frequently employer-insured (26.1% vs. 33.1%, p = 0.02). No gender or ethnicity-related differences were found between persons with vs. without Internet access, and the descriptors of disease severity and management were similar, including symptoms, oxygen and steroid use, and compliance with treatment.

When comparing comorbidities between participants according to Internet access, those without access more frequently reported hypertension (53.4% vs. 44.6%, p = 0.009), heart disease (33.4% vs. 23.4%, p = 0.0009), arthritis (53.1% vs. 35.9%, p < 0.0001), and diabetes (24.6% vs. 17.5%, p = 0.008), but less frequently reported sleep disorders (15.8% vs. 21.2%, p = 0.04) and obesity (13.8% vs. 23.6%, p = 0.0003). Participants without Internet access also less frequently reported a perceived diagnosis delay (19.4% vs. 39.4%, p < 0.0001). In all other respects, however, the perceived needs of participants without Internet access were similar to those of participants with Internet access.

A similar analysis restricted to the participants reporting that they had Internet access was conducted, comparing participants who use the Internet frequently (i.e., at least weekly) with the rest of the participants. More frequent users were younger (62.8 ± 10 vs. 71.1 ± 8.8 years of age, p < 0.0001), more likely female (67.3% vs. 53.4%, p = 0.003), had income below $30,000 (49.7% vs. 37.9%, p = 0.01) and less frequently relied on Medicare insurance (57.6% vs. 76.5%, p = 0.003).

Regarding disease severity and comorbidities, frequent Internet users tended to report at least one COPD-related exacerbation (58.3% vs. 39.1%, p < 0.0001), and oxygen use (75.5% vs. 55.2%), although other descriptors of symptoms and disease severity were similar. More frequent Internet users reported poorer QOL scores (mean ± s.d. points in the LCOPD: 12.6 ± 5.8 vs. 9.2 ± 6.0, p < 0.0001). Comorbidity profiles showed that more frequent Internet users reported a lower frequency of hypertension (28.5% vs. 50.1%, p < 0.0001), arthritis (27.8% vs. 38.7%, p = 0.01) and diabetes (11.9% vs. 19.4%, p = 0.03), but a higher frequency of anxiety (35.1% vs. 18.5%, p < 0.0001), depression (32.5% vs. 19.4%, p = 0.0009) and obesity (31.8% vs. 17.5%, p = 0.0002).

Participants who used the Internet more frequently reported more perceived needs, including dissatisfaction with their physician (14.7% vs. 8.7%, p = 0.03), feeling that their doctor was not sympathetic (6.6% vs. 2.8%, p = 0.04), not a good listener (17.2% vs. 5.9%, p < 0.0001), more difficult to reach (20.5% vs. 10.2%, p = 0.001), and difficult to make appointments with (12.6% vs. 6.4%, p = 0.01). They also were more apt to report believing that their diagnosis was delayed (57.6% vs. 33.2%, p < 0.0001), a feeling of dissatisfaction with their treatment (13.3% vs. 7.5%, p = 0.03), and feeling that they were treated poorly by the healthcare system (23.3% vs. 5.2%, p < 0.0001). Furthermore, they reported more visits to their pulmonologist (mean number of visits 3.9 ± 4.9 vs. 2.7 ± 2.7, p = 0.002) and to any physician (9.1 ± 6.6 vs. 7.6 ± 6.2 visits, p = 0.01) during the prior year. Table 3 shows participants’ use of other information sources on COPD. Both respondents without Internet access and those who use the Internet less often reported less frequent use of almost all other information sources, with the exception of television or cable, physicians, and pamphlets/brochures.

Multivariate models of Internet access and frequent Internet use adjusted for gender, insurance, exacerbations, comorbidities, visits to physicians, and indicators of disease severity were tested (Table 4). These models confirmed that lack of Internet access is independently associated with specific demographic characteristics: age increment was associated with likelihood of lack of access (OR 1.10, 95% CI 1.07, 1.13), as well as low income (OR 2.47, 95% CI 1.63, 3.73), low educational level (OR 2.30, 95% CI 1.54, 3.45), and the presence of mobility-related comorbid diseases (OR 1.56, 95% CI 1.05, 2.34). The presence of cardiovascular disease was not associated with lack of access (OR 1.16, 95% CI 0.77, 1.76).

Models evaluating more frequent Internet use (at least weekly versus less frequent) showed that younger age (OR 0.95, 95% CI 0.93, 0.98 per one year increment) was associated with the outcome. Besides age, needs were the factors strongly associated with more frequent Internet use, including perception of diagnosis delay (OR 1.72, 95% CI 1.06, 2.78), feeling treated poorly by the healthcare system (OR 2.46, 95% CI 1.15, 5.24), and that the doctor does not listen (OR 3.14, 95% CI 1.42, 6.95). Additionally, perceiving that the symptoms limit the ability to work was associated with frequent use (OR 2.54, 95% CI 1.55, 4.16).