This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on https://formative.jmir.org, as well as this copyright and license information must be included.
The use and sharing of patient-generated health data (PGHD) by clinicians or researchers is expected to enhance the remote monitoring of specific behaviors that affect patient health. In addition, PGHD use could support patients’ decision-making on preventive care management, resulting in reduced medical expenses. However, sufficient evidence on the use and sharing of PGHD is lacking, and the impact of PGHD recording on patients’ health behavior changes remains unclear.
This study aimed to assess patients’ engagement with PGHD recording and to examine the impact of PGHD recording on their health behavior changes.
This supplementary analysis used the data of 47 postpartum women who had been assigned to the intervention group of our previous study for managing urinary incontinence. To assess the patients’ engagement with PGHD recording during the intervention period (8 weeks), the fluctuation in the number of patients who record their PGHD (ie, PGHD recorders) was evaluated by an approximate curve. In addition, to assess adherence to the pelvic floor muscle training (PFMT), the weekly mean number of pelvic floor muscle contractions performed per day among 17 PGHD recorders was examined by latent class growth modeling (LCGM).
The fluctuation in the number of PGHD recorders was evaluated using the sigmoid curve formula (
The number of PGHD recorders declined over time in a sigmoid curve. A small number of users recorded PGHD continuously; therefore, patients’ engagement with PGHD recording was low. In addition, more than half of the PGHD recorders (moderate- and low-level classes combined: 10/17, 59%) had poor PFMT adherence. These results suggest that PGHD recording does not always promote health behavior changes.
According to the Office of the National Coordinator for Health Information Technology, patient-generated health data (PGHD) is defined as health-related data created, recorded, or gathered by or from patients (or family members or other caregivers) to help address a health concern [
We conducted multicomponent interventions with reminder emails for pelvic floor muscle training (PFMT) to manage urinary incontinence (UI), in which users manually recorded the number of pelvic floor muscle contractions (PFMCs) performed as PGHD [
Many users of digital behavior change interventions (DBCIs) that use technologies such as the internet, telephones, mobile phones, and environmental sensors [
We conducted a supplementary analysis of the PFMC data stored on the server of our previous study [
The participants were postpartum women who had delivered from January to August 2014 at an obstetric clinic in Osaka Prefecture, Japan, which performs approximately 600 deliveries per year, and had been assigned to the intervention group of our previous study [
This study used the data of 47 postpartum women who had been assigned to the intervention group of our previous study [
This research is a supplementary analysis of our previous study, which improved PFMT adherence and reduced the number of postpartum women with UI. A detailed description of the study has been published in full [
This study uses data collected during our previous study [
Data collected by the above procedure were classified into the following four categories: (1) participants’ demographic characteristics, including age, BMI before pregnancy, weight gain during pregnancy, and their child’s birth weight; (2) the number of participants who recorded their PGHD (ie, PGHD recorders); (3) each participants’ status of PGHD recording; and (4) weekly mean number of PFMCs performed per day among those who recorded it continuously. The participants’ usability of PGHD recording was evaluated after the 8-week intervention period with the question “Was it difficult to record the PFMCs every day?” The participants responded on a 5-point Likert scale (“Strongly agree,” “Agree a little,” “Neither agree nor disagree,” “Disagree a little,” or “Strongly disagree”). Furthermore, comments on the participants’ usability of PGHD recording were collected with the prompt: “Please comment on your experience of recording PFMCs on our system in the free-text field.” Participants voluntarily answered these two questions about usability. These data were encrypted using Secure Sockets Layer to prevent leakage of personal information during transmission and were stored on the server through the website. All data were downloaded in .csv format.
The participants’ demographic characteristics were described as continuous variables (reported as median values with IQRs) and categorical variables (reported as number of cases with percentages). To assess engagement with PGHD recording during the intervention period, a graph was plotted with the number of PGHD recorders on the y-axis and the number of days on the x-axis, and the approximate equation of the curve was calculated. To visualize each participant’s status of PGHD recording during the intervention period, a figure was created with gray-shaded cells indicating the days that a given participant recorded PGHD and the numbers within cells denoting how many PFMCs were performed that day. On the y-axis, participants are arranged based on the total number of times that PFMCs were recorded and the total number of PFMCs performed during the intervention period. Based on the lower asymptote that was obtained from the approximate curve, 17 participants (IDs 1-17) were classified as the high-engagement group and the remaining participants (IDs 18-47) were classified as the low-engagement group. Fisher exact test and Mann-Whitney
The participants’ demographic characteristics are shown in
Comparison of demographic characteristics of participants having high and low engagement with patient-generated health data recording.
Characteristic | Total participants (N=47) | Engagement with PGHDa | |||||||
|
|
High (n=17) | Low (n=30) |
|
|||||
UIb at baseline, n (%) | 3 (6) | 1 (2) | 2 (4) | .99c | |||||
Age (years), median (IQR) | 34 (31-36) | 34 (32-37) | 33 (30-36) | .21d | |||||
Multipara, n (%) | 33 (70) | 14 (30) | 19 (40) | .20c | |||||
BMI before pregnancy (kg/m2), median (IQR) | 20 (19-21) | 20 (18-21) | 20 (19-21) | .96d | |||||
Weight gain during pregnancy (kg), median (IQR) | 10 (8-12) | 10 (9-12) | 11 (89-12) | .89d | |||||
Child’s birth weight (g), median (IQR) | 3104 (2760-3384) | 2885 (2736-3196) | 3160 (2945-3480) | .05d |
aPGHD: patient-generated health data.
bUI: urinary incontinence.
cFisher exact test.
dMann-Whitney
Engagement with PGHD recording is shown in
In the approximate curve, there was an inflection point at 14.2 days (95% CI 11.1-17.3;
Engagement with patient-generated health data recording.
Status of patient-generated health data recording.
The 3-class model produced latent trajectories that corresponded to the weekly mean number of PFMCs performed per day. The following groups were defined: “high” for PGHD recorders who started with high PFMT adherence levels (7/17, 41%;
Model information by number of classes obtained through latent class growth modeling.
Test | Number of classes | ||
|
|
2 | 3 |
Percent per class | 41/59 | 41/18/41 | |
Entropy | 0.991 | 1.000 | |
Bayesian Information Criterion | 592.6 | 590.0 | |
|
–285.6 | –270.8 | |
|
.43 | .18 | |
|
–288.6 | –270.8 | |
|
.29 | .17 |
Pelvic floor muscle training adherence levels among patient-generated health data recorders. Green line: high; red line: moderate; blue line: low.
Characteristics of the 3 groups by PFMTa adherence level.
Characteristic | PFMT adherence level | |||||
|
High (n=7) | Moderate (n=3) | Low (n=7) |
|
||
UIb at baseline (n=1), n (%) | 0 (0) | 0 (0) | 1 (100) | .99c | ||
Age (years), median (IQR) | 34 (32-36) | 41 (37-42) | 33 (30-36) | .03d | ||
Multipara (n=14), n (%) | 6 (43) | 2 (14) | 6 (43) | .99c | ||
BMI before pregnancy (kg/m2), median (IQR) | 21 (19-25) | 20 (19-21) | 18 (17-20) | .07d | ||
Weight gain during pregnancy (kg), median (IQR) | 12 (10-13) | 10 (6-16) | 10 (8-10) | .35d | ||
Child’s birth weight (g), median (IQR) | 3213 (3060-3530) | 2722 (2704-2885) | 2752 (2675-2932) | .01d | ||
|
||||||
|
Week 1 | 18 (17-18) | 11 (5-12) | 3 (2-7) | .004d | |
|
Week 2 | 17 (15-18) | 13 (8-15) | 2 (2-4) | .003d | |
|
Week 3 | 18 (17-18) | 11 (9-15) | 2 (4-1) | .001d | |
|
Week 4 | 18 (17-18) | 9 (8-18) | 2 (1-3) | .002d | |
|
Week 5 | 18 (18-18) | 15 (7-15) | 2 (1-3) | .001d | |
|
Week 6 | 18 (14-18) | 8 (6-15) | 2 (1-3) | .001d | |
|
Week 7 | 18 (17-18) | 9 (6-14) | 1 (1-3) | <.001d | |
|
Week 8 | 18 (18-18) | 7 (5-8) | 2 (1-3) | .001d | |
|
Total PGHDf (times) | 53 (43-56) | 54 (46-54) | 52 (40-54) | .64d | |
“ |
.17c | |||||
|
“Strongly agree” and “Agree a little” (n=1) | 0 (0) | 1 (100) | 0 (0) |
|
|
|
“Neither agree nor disagree,” “Disagree a little,” and “Strongly disagree” (n=11) | 4 (36) | 1 (9) | 6 (54) |
|
aPFMT: pelvic floor muscle training.
bUI: urinary incontinence.
cFisher exact test.
dMann-Whitney
ePFMC: pelvic floor muscle contraction.
fPGHD: patient-generated health data.
After the intervention period, some participants (27/47, 57%) answered questions about the usability of PGHD recording (
Uncategorized comments by the participants were as follows: “By reporting the number of times, I felt as if I was being watched for not being lazy, and I think I was able to continue,” “I did not report PFMCs, but I was able to perform them every day,” “When I receive emails three times a day, the importance of emails gradually decreased for me, and I wish I could set the number of times emails were sent individually,” and “At first, I was motivated, but I tended to skip halfway through.”
Usability of patient-generated health data recording.
Response | Participants (n=27), n (%) | |
“ |
||
|
Strongly agree | 3 (11) |
|
Agree a little | 7 (26) |
|
Neither agree nor disagree | 4 (15) |
|
Disagree a little | 12 (44) |
|
Disagree | 1 (4) |
|
||
|
“I was able to continue training by reminder email.” | 14 (52) |
|
“It was difficult to secure time for training while raising children.” | 6 (22) |
|
“Nothing in particular.” | 3 (11) |
|
Other | 4 (15) |
aPFMC: pelvic floor muscle contraction.
In this study, we examined patients’ engagement with PGHD recording integrated into a multicomponent intervention and evaluated the impact of PGHD recording on their health behavior changes. The following findings were obtained. First, engagement with PGHD recording might be low. This could be because the number of PGHD recorders declined over time, as indicated on a sigmoid curve. Moreover, a small number of participants recorded PGHD continuously (17/47, 36%). Second, PGHD recording may not promote health behavior changes. This was suggested by the overall poor PFMT adherence observed (10/17, 59%).
Eysenbach [
PFMT adherence in the 17 participants who recorded PGHD continuously could be clearly categorized into 3 latent classes. The moderate- and low-level classes (combined: 10/17, 59%) were considered to have poor PFMT adherence, and the weekly mean number of PFMCs performed per day was low even in PGHD recorders. In the present analysis, data that were not recorded as PGHD were treated as missing data. The comment “I did not report PFMCs, but I was able to perform them every day” by some participants suggests that some of those who did not record PGHD actually performed PFMT without recording it; thus, we considered an input of 0 for PFMCs performed as invalid. Accordingly, PFMT adherence may have been higher than shown in the data. However, as PFMT is a muscle-strengthening exercise, a minimum number of contractions and consistent practice (for at least 8 weeks) are both required. Therefore, treatment and prevention of UI cannot be expected unless the PFMT adherence pattern is similar to that of the high-level class drawn by LCGM.
Even users who showed high engagement with PGHD recording did not necessarily adhere to the PFMT regimen as instructed. These data suggest that PGHD recording may not promote health behavior changes. One of the PGHD usability comments was “By reporting the number of times, I felt as if I was being watched for not being lazy, and I think I was able to continue.” A previous study [
There are 3 limitations of this study. First, the sample size was too small to clearly demonstrate associations between PGHD recording and health behavior changes. One reason for the small sample size is that participants did not receive explicit instructions that they had to record PGHD, which was a component of the system, at the time of study participation. The participants recorded, or did not record, PGHD at their own discretion. Therefore, those who did not consider the PGHD recording of their PFMT necessary might not have continued PGHD recording. When participants use a multicomponent system, such as the one in our study, it is difficult to ensure that all components are used. Despite this limitation, this is one of the few studies that used LCGM to evaluate PFMT adherence, a measure of health behavior change, and investigated the impact of PGHD recording on health behavior change. Therefore, we believe that this case study will lead to a larger-scale survey. Second, the intervention in this study was a multicomponent intervention combining PGHD recording and PFMT reminder emails; as such, the effect of PGHD recording alone could not be evaluated. However, it is commonly accepted in PGHD research that evaluation of PGHD recording alone is difficult because it is an integral part of multicomponent interventions. In the future, a research design that can evaluate the impact of PGHD recording alone on health behavior changes needs to be established. Third, the number of PFMCs performed as a measure of PFMT adherence is a self-reported outcome and could not be confirmed. Therefore, systemic errors could occur as a result of participants reporting an inaccurate number of PFMCs performed. Given this limitation, the results must be carefully interpreted.
The number of users who recorded PGHD in a multicomponent intervention declined over time in a sigmoid curve. A small number of users recorded PGHD continuously, and users felt that PGHD recording was burdensome. Therefore, PGHD engagement was found to be low. In addition, more than half of the PGHD recorders had poor PFMT adherence. These results suggest that PGHD recording may not always promote health behavior changes. Clinicians and researchers must understand that users who record PGHD in multicomponent interventions do not necessarily adhere to health behavior changes.
digital behavior change intervention
latent class growth modeling
pelvic floor muscle contraction
pelvic floor muscle training
patient-generated health data
urinary incontinence
The kind cooperation of all the participants is gratefully acknowledged. This study was supported by a Grant-in-Aid for Young Scientists (B), grant A227922240, from the Ministry of Education, Culture, Sports, Science and Technology. We thank Dr Tomoharu Sato (Department of Biostatistics & Data Science, Graduate School of Medicine, Osaka University) for providing expert advice on statistics.
None declared.