This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Web-based interventions show promise in promoting a healthy lifestyle, but their effectiveness is hampered by high rates of nonusage. Predictors and reasons for (non)usage are not well known. Identifying which factors are related to usage contributes to the recognition of subgroups who benefit most from Web-based interventions and to the development of new strategies to increase usage.
The aim of this mixed methods study was to explore patient, intervention, and study characteristics that facilitate or impede usage of a Web-based physical activity intervention for patients with knee and/or hip osteoarthritis.
This study is part of a randomized controlled trial that investigated the effects of Web-based physical activity intervention. A total of 199 participants between 50-75 years of age with knee and/or hip osteoarthritis were randomly assigned to a Web-based intervention (n=100) or a waiting list (n=99). This mixed methods study used only data from the individuals allocated to the intervention group. Patients were defined as users if they completed at least 6 out of 9 modules. Logistic regression analyses with a stepwise backward selection procedure were executed to build a multivariate prediction usage model. For the qualitative part, semistructured interviews were conducted. Both inductive and deductive analyses were used to identify patterns in reported reasons for nonusage.
Of the 100 participants who received a password and username, 46 completed 6 modules or more. Multivariate regression analyses revealed that higher age (OR 0.94,
In this mixed methods study, we found patient, intervention, and study factors that were important in the usage and nonusage of a Web-based PA intervention for patients with knee and/or hip osteoarthritis. Although the self-guided components offer several advantages, particularly in relation to costs, reach, and access, we found that older patients and participants with a comorbid condition need a more personal approach. For these groups the integration of Web-based interventions in a health care environment seems to be promising.
The Netherlands National Trial Register (NTR): NTR2483; http://www.trialregister.nl/trialreg/admin/rctview.asp?TC=2483 (Archived by Webcite at http://www.webcitation.org/67NqS6Beq).
Osteoarthritis (OA) in the knee or hip is a prevalent musculoskeletal disorder characterized by joint pain, joint stiffness, and functional disability [
In an attempt to enhance a physically active lifestyle in patients with knee and/or hip OA, we developed a Web-based PA intervention. The intervention, entitled
Unfortunately, substantial rates of nonusage were observed. A considerable proportion of potential users was never exposed to important program content. This is consistent with other studies [
Although considerable research has been devoted to quantitative predictors of nonusage, little qualitative research has been conducted on the underlying reasons for nonusage. Therefore, we conducted a mixed methods study to gain a deeper understanding of actual usage patterns, possible attrition predictors, and reasons for (non)usage. This is a necessary step toward enhancing program usage and may help us to make the
In this study, we utilized a mixed methods design employing both quantitative and qualitative (interviews) methods. By integrating the quantitative and qualitative results, we aimed to identify patient-, intervention-, and study-related characteristics that may facilitate or impede the usage of Web-based intervention for patients with knee and/or hip OA. Since this study was explorative by nature, no a priori hypotheses were formulated.
Data from this study were retrieved from a randomized controlled trial that aimed at evaluating the effectiveness of the
Over the course of 1 year, experts from the Netherlands Institute for Health Services Research developed the
Homepage Join2move.
Program usage (ie, the number of completed program modules) was monitored throughout the intervention period. A module consisted of a text-based assignment plus accompanying evaluation form, which was presented on the website for 7 consecutive days. Once a participant had filled out the evaluation form 7 days after receiving the weekly assignment, the module was defined as completed and the user was automatically presented with a new weekly assignment. If a scheduled weekly module was missed, participants had the option to repeat the module, adapt the difficulty, or continue with the next module. In total, 9 weekly modules were available to the participant. This was automatically registered. After some consideration, the research team had decided that completion of at least 6 modules was required to improve PA and other primary effects. Patients were defined as users if they completed at least 6 out of 9 modules. Participants who did not reach this threshold were defined as nonusers. Predictors of usage were collected through online baseline questionnaires and can be categorized as demographic, clinical, or psychological predictors. The potential predictors were not selected on theoretical grounds.
Demographic predictors were gender, education (low: primary and lower vocational education; middle: secondary and middle vocational education; high: higher vocational and university education), and age (years) as demographic predictors.
Clinical predictors in this study were location of OA (knee, hip or both), duration of OA complaints (years), and body mass index (BMI) (weight in kilograms divided by height in meters squared). Pain and fatigue were assessed on a 10-point Numerical Rating Scale (0 is no pain/not tired, 10 is worst possible pain/extremely tired). Self-reported PA was measured by the validated PA Scale for the Elderly (PASE) [
Anxiety and depression were evaluated by a 14-item Hospital Anxiety and Depression scale [
Active and passive pain coping were determined by the Pain Coping Inventory questionnaire [
One year after being assigned to the program, a subgroup of participants from the intervention group was interviewed. All participants from the intervention group (n=100) were categorized into two groups: (1) users and (2) nonusers. Since the nonuser group showed considerable divergence in extent of program use (0 to 5 modules), we decided to invite more nonusers than users for our interview sample. This was executed by a stratified purposive sampling procedure [
Descriptive analyses were performed to describe participant characteristics and program usage. Logistic regression analysis with a stepwise backward selection procedure was used to build the most parsimonious prediction model. Program use (user/nonuser) was employed as a dichotomous dependent variable. Demographic, clinical, and psychological variables were the independent variables. Statistical analyses were conducted in two phases. First, potential predictors of interest were screened by univariate logistic regressions. Second, variables that achieved
Interviews were analyzed by means of deductive and inductive content analysis [
Of the 100 participants who received a password and username to enroll, 49 users made a start with the first module and 6 participants never logged in to their personal website.
Program use.
presents demographic, clinical, and psychological baseline variables for users and nonusers. Univariate analyses showed that age, BMI, symptoms, and comorbidity reached the threshold of
The qualitative deductive and inductive analysis resulted in the identification of several reasons for (non)usage. The majority of reasons were found by the deductive analysis. Additionally, the inductive analysis identified a number of personal factors (eg, social environment and emotional factors) relating to (non)usage. Reasons are divided into patient, intervention, and study characteristics and are illustrated by interview quotes. Additional quotes illustrative of each theme are provided in
Interviewees reported that a low mood interfered with their ability to perform modules. One participant summarized this sentiment by saying, “I had a bad year and I was not at ease with myself. I was not in the right mood to exercise. It was all too much” [woman, hip OA, nonuser]. Lack of self-discipline was another identified reason for nonusage. As one man put it “This kind of program does not work for me. I find it difficult to stay motivated all the time. At the beginning I was motivated but then it went downhill quickly. I got lazy and other activities became more important” [man, knee OA, nonuser]
Participants reported that several characteristics of the
Study-related factors were also cited as reasons for remaining or not remaining engaged in the program. Some participants felt under obligation to continue. They described a feeling of commitment to the organizers of the study. “Because I was allocated to the intervention group, I wanted to finish the entire program. Maybe a little old-fashioned but I found it inappropriate to stop halfway” [woman, knee OA, user]. Some participants perceived the questionnaires used as being too long or too difficult. The questionnaire consisted of 17 pages with a total of 171 items. Participants not only lost interest in completing the questionnaires but were also less motivated to continue with the program.
Baseline demographic and clinical characteristics.
|
|
|
Users, n=46 | Nonusers, n=54 |
|
|
|||||
|
|
||||
|
|
Male | 17 (37) | 23 (43) | .57 |
|
|
Female | 29 (63) | 31 (57) |
|
|
Age (years), mean (SD |
|
60 (6.3) | 62 (6.5) | .09 |
|
|
||||
|
|
Lower education | 7 (15) | 6 (11) | .60 |
|
|
Middle education | 18 (39) | 18 (33) | .41 |
|
|
Higher education | 21 (46) | 30 (56) | .42 |
|
|||||
|
|
||||
|
|
Knee | 30 (65) | 36 (67) | .89 |
|
|
Hip | 11 (24) | 11 (20) | .80 |
|
|
Both | 5 (11) | 7 (13) | .64 |
|
OA duration (years), mean (SD) |
|
2.8 (1.3) | 2.8 (1.1) | .86 |
|
|
||||
|
|
Normal weight (<25) | 22 (48) | 17 (31) | .10 |
|
|
Overweight (>25) | 24 (52) | 37 (69) |
|
|
|
||||
|
|
No, n (%) | 36 (78) | 30 (56) | .02 |
|
|
Yes, n (%) | 10 (22) | 24 (44) |
|
|
Physical activity |
|
117 (66.1) | 130 (65.5) | .29 |
|
Pain, 0-10 |
|
5.4 (2) | 5.4 (2.3) | .92 |
|
Fatigue, 0-10 |
|
4.7 (2.7) | 5.2 (2.8) | .34 |
|
Symptoms |
|
56 (15.6) | 60 (17.8) | .17 |
|
ADL |
|
58.3 (22.3) | 55.3 (19.9) | .47 |
|
Sport and recreation |
|
58 (22) | 55 (19.9) | .47 |
|
Quality of life |
|
38.7 (16.9) | 42 (17.4) | .32 |
|
|||||
|
Self-efficacy pain |
|
3.4 (0.8) | 3.4 (0.9) | .67 |
|
Self-efficacy other symptoms |
|
3.5 (0.9) | 3.4 (0.9) | .60 |
|
Active pain coping |
|
2.0 (0.4) | 2.1 (0.4) | .34 |
|
Passive pain coping |
|
1.8 (0.4) | 1.9 (0.4) | .26 |
|
Anxiety |
|
4.7 (3) | 4.5 (2.9) | .62 |
|
Depression |
|
3.8 (2.9) | 3.8 (3) | .88 |
|
Internal locus of control |
|
23 (5.4) | 23.7 (4.3) | .46 |
|
Powerful others locus of control |
|
15.3 (4.4) | 15.9 (4.5) | .54 |
Univariate and multivariate analyses for predictors for usagea.
|
|
Bb | SEc | OR (95% CI) |
|
|
|||||
|
Age, years | −.06 | .04 | .94 (0.88-1.01) | .09 |
|
BMI (normal weight/overweight) | −.69 | .42 | .50 (0.22-1.13) | .10 |
|
Comorbidity (no/yes) | −.93 | .44 | .39 (0.14-0.84) | .02 |
|
Symptoms (0-100) | −.02 | .01 | .98 (0.96-1.01) | .17 |
|
|||||
|
Age, years | −.07 | .04 | .94 (0.87-1) | .08 |
|
Comorbidity (no/yes) | −1.1 | .46 | .33 (0.13-0.82) | .02 |
aThe reference groups are nonusage, normal weight, and no comorbidity.
bB=beta coefficient.
cSE=standard error.
The aim of this mixed methods study was to identify patient, intervention, and study characteristics that facilitate or impede the usage of a Web-based intervention for patients with knee and/or hip OA. Results from this study showed that participants with knee and/or hip OA used the
Considering the predictors of usage, it appeared from the quantitative analysis that age and comorbidity proved to be significantly related to program usage. Younger participants were more likely to use the intervention modules than older participants. This is in contrast to previous studies that have found correlations between older age and higher usage rates [
With respect to the intervention, participants indicated that the automatic gradual increase of PA as well as working toward a short-term goal were mechanisms that supported them in completing weekly modules. Compared with face-to-face treatments, the flexibility of completing modules at one’s own pace without time or travel restrictions was cited as a major advantage. However, older patients, those with comorbidity and patients who attach great importance to personal contact indicated that the lack of human involvement was a disadvantage. Furthermore, from the interviews it became clear that those who felt themselves responsible for their own progress were most likely to use the program. This, however, was not confirmed in the quantitative analysis. Although we included questions about responsibility and persistence, the questionnaires were not sensitive enough to confirm the conclusions from the qualitative analysis. This illustrates very well why we have chosen dual data collection. The weakness of questionnaires was compensated by interview data. Other mentioned motivations for (non)usage were trial specific. While questionnaires impede usage, commitment to the research team was described as an important facilitator for usage. We did not find any predictive value for education and gender, in contrast to other studies [
A major weakness is the potential presence of recall bias. In an effort to prevent attention bias during the previously conducted randomized controlled trial, the length of time between program participation and interviews was approximately 12 months. As a consequence, participants may not have accurately remembered the intervention in detail. This may have affected the reliability of our results. Another weakness is that results are limited in their generalizability because participants were mainly older, healthy, and highly educated patients with knee and/or hip OA. Furthermore, the role of motivation as proximate determinant of usage behavior was not investigated in this study. Future research should examine the role of motivation on program usage. A last limitation was that participants were included on the basis of self-reported OA. Diagnosis was not confirmed through clinical tests or x-ray reports due to practical reasons. Although self-reported OA is a common inclusion strategy in the field of osteoarthritis research, it is presumable that we have included false positive OA patients in the study.
In light of rising health care costs and the large population of patients with knee and/or hip OA,
Screenshots intervention.
Interview guide.
Interview quotes.
behavior graded activity
body mass index
Hip Injury OA Outcome Score
Knee OA Outcome Score
Multidimensional Health Locus of Control Scale
osteoarthritis
odds ratio
physical activity
PA Scale for the Elderly
randomized controlled trial
Receiver Operating Characteristic
None declared.