Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Online Diabetes Self-Management Program

Online Diabetes Self-Management Program Emerging Treatments and Technologies ORIGINAL ARTICLE Online Diabetes Self-Management Program A randomized study 1 1 KATE LORIG, RN, DRPH MAURICE GREEN, PHD to our knowledge, examining such a pro- PHILIP L. RITTER, PHD VALARIE BLUE BIRD JERNIGAN, PHD gram among AI/ANs. DIANA D. LAURENT, MPH SIOBHAN CASE, BA The Cochrane Collaboration re- KATHRYN PLANT, MPH viewed group-based training for type 2 diabetes (4). They found 11 studies that met their criteria. Eight of these were ran- OBJECTIVE — We hypothesized that people with type 2 diabetes in an online diabetes domized studies and three were con- self-management program, compared with usual-care control subjects, would 1) demonstrate trolled studies. All of the interventions reduced A1C at 6 and 18 months, 2) have fewer symptoms, 3) demonstrate increased exercise, were taught by health professionals. One and 4) have improved self-efficacy and patient activation. In addition, participants randomized study took place in a community setting, to listserve reinforcement would have better 18-month outcomes than participants receiving no and one reported a mean baseline A1C reinforcement. 7%. Jacksan et al. (5) conducted a system- RESEARCH DESIGN AND METHODS — A total of 761 participants were randomized atic review of computer-assisted technol- to 1) the program, 2) the program with e-mail reinforcement, or 3) were usual-care control ogies in diabetes prior to 2004. They subjects (no treatment). This sample included 110 American Indians/Alaska Natives (AI/ANs). Analyses of covariance models were used at the 6- and 18-month follow-up to compare groups. found four articles involving patient edu- cation. In an early study, (6) groups were RESULTS — At 6 months, A1C, patient activation, and self-efficacy were improved for pro- randomized to basic diabetes informa- gram participants compared with usual care control subjects (P 0.05). There were no changes tion, tailored online coaching, or peer in other health or behavioral indicators. The AI/AN program participants demonstrated im- support. Improvement in health behav- provements in health distress and activity limitation compared with usual-care control subjects. iors and psychological outcome were The subgroup with initial A1C7% demonstrated stronger improvement in A1C (P 0.01). At found in all three groups, with no differ- 18 months, self-efficacy and patient activation were improved for program participants. A1C was ences between groups. Glasgow et al. (7) not measured. Reinforcement showed no improvement. showed that a computer-assisted inter- vention was practicable and acceptable in CONCLUSIONS — An online diabetes self-management program is acceptable for people with type 2 diabetes. Although the results were mixed they suggest 1) that the program may have a real-world setting and resulted in im- beneficial effects in reducing A1C, 2) AI/AN populations can be engaged in and benefit from provements in recommended services. In online interventions, and 3) our follow-up reinforcement appeared to have no value. a low-intensity computer program study, short-term outcomes were promising but Diabetes Care 33:1275–1281, 2010 not significant (8). Wengberg (9), utiliz- ing an computer diabetes intervention, ype 2 diabetes affects 9.6% of the with type 2 diabetes are willing or able to has suggested that self-efficacy may func- adult population, and its prevalence participate in small-group programs, nor tion as a moderator for diabetes behavior is increasing (1). While the need for are such programs likely to be available in change, and Gerber et al. (10) have dem- self-management support is well docu- all locations. onstrated usability of an Internet program mented, most diabetes education studies There are few studies of community- for young inner-city adults. In summary, have taken place in clinical settings and based diabetes education programs for Internet-based educational programs targeted those who have a high A1C (usu- American Indians/Alaskan Natives (AI/ have been demonstrated to change behav- ally 7%). Recent community-level, ANs). We report on a randomized, con- iors and sometimes health status. We peer-led, small-group diabetes self- trolled trial of an Internet-based diabetes were unable to find computer-based stud- management programs have shown self-management program (IDSMP) in- ies demonstrating changes in A1C. promise (2,3). However, not all patients cluding AI/ANs. This was the first study, ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● RESEARCH DESIGN AND METHODS — We report on a ran- From the Stanford Patient Education Research Center, Stanford University School of Medicine, Palo Alto, 2 3 California; the Stanford Prevention Research Center, Stanford University, Palo Alto, California; and Yale domized 6-month trial of the IDSMP, School of Medicine, New Haven, Connecticut. with an 18-month follow-up. We hypoth- Corresponding author: Philip L. Ritter, philr@stanford.edu. esized that participants in the IDSMP, Received 20 November 2009 and accepted 5 March 2010. Published ahead of print at http://care. compared with usual-care control sub- diabetesjournals.org on 18 March 2010. DOI: 10.2337/dc09-2153. Clinical trials reg. no. NCT00185601, clinicaltrials.gov. jects, would demonstrate 1) reduced A1C © 2010 by the American Diabetes Association. Readers may use this article as long as the work is properly at 6 and 18 months, 2) have fewer symp- cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons. toms, 3) have increased exercise, and 4) org/licenses/by-nc-nd/3.0/ for details. have improved self-efficacy and patient The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. activation. We also hypothesized that par- care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 1275 Online diabetes self-management program ticipants randomized to a follow-up list- had taken the IDSMP (as nonstudy sub- Randomized study serve, peer-support group would have jects). Facilitators assist participants by The randomized IDSMP group was com- better 18-month outcomes than partici- reminding them to log on, modeling ac- pared with the usual-care control group at pants receiving no follow-up. tion planning and problem-solving, offer- 6 months. If the reinforcement study (be- ing encouragement, and posting to the low) had shown that reinforcement par- The IDSMP bulletin boards. They also monitor the ticipants had greater improvements than The asynchronous, 6-week, IDSMP is daily posts for safety and report inappro- unreinforced IDSMP participants, the two based on English- and Spanish-language priate posts to the investigators. All facil- IDSMP groups would be compared with peer-led small-group diabetes self- itation takes place online, mainly via posts control participants separately. If there management programs (2,3). The IDSMP within the program pages. Each partici- were few differences, the two randomized consists of six weekly sessions. Partici- pant receives personalized responses IDSMP groups would be combined and pants logged on individually to the ses- from facilitators during each weekly ses- compared with the usual-care control sions, which were available for the entire sion. Unlike the small-group program, fa- group. week. The topics covered are shown in cilitators do not deliver content, as this is After 6 months, usual-care partici- online appendix Table A1 (available in scripted in the Learning Center. Programs pants recruited as part of the AI/AN sub- the online appendix at http://care. were facilitated by 16 different people, group were offered the program. All other diabetesjournals.org/cgi/content/full/dc09- half with diabetes. Each program has at usual-care participants continued as con- 2153/DC1). least one facilitator with diabetes. The trol subjects through the 18 months of the A password-protected homepage study was approved by the Stanford study. Follow-up data collected at 18 provides access to the weekly activities, School of Medicine Institutional Review months allowed comparison of IDSMP including The Learning Center, where the Board. participants to usual-care subjects, ex- program content is offered in 20 –30 new cluding the AI/AN subset. Web pages weekly. Each week, partici- Participants and data collection pants are asked to reply to a question such Participants were aged 18 years, were Reinforcement study as “What problems do you have because not pregnant or in care for cancer, had The reinforcement study compared of your diabetes?” and to make a specific physician-verified type 2 diabetes, and IDSMP treatment participants who had action plan. The questions and action had access to the Internet. Recruitment no reinforcement with those who had plans are posted on bulletin boards in the was largely via the Internet, although been randomized to a listserve discussion Discussion Center, where they can be print and broadcast media were also uti- group. The discussion group was in- seen by all participants. lized. Special effort was made to recruit tended to reinforce any benefits of the The Discussion Center is made up of AI/AN participants using Web sites and program by providing peer support. four interactive threaded bulletin boards media associated with tribal and AI/AN Comparisons were made at 6 and at 18 (Action Planning, Problem Solving, Diffi- organizations. This was accomplished months. The AI/AN participants were in- cult Emotions, and Celebrations) popu- utilizing the expertise of an AI/AN re- cluded in the 18-month reinforcement lated by responses made in the Learning searcher (12). study. Center, as well as new threads started by All consents and questionnaires were participants whenever they wish. A typi- administered online. Participants con- AI/AN study cal program of 20 –25 participants results tacted the study by going to the Web site, AI/AN participants were randomized in 500 or more posts. My Tools consists of where they were screened for eligibility with other participants but entered the exercise and medication logs, audio relax- and were asked to complete consent and randomized study for only 6 months, af- ation exercises, meal planning, and glu- baseline questionnaires. A1C was ob- ter which time AI/AN usual-care partici- cose-monitoring tools and links to other tained using mailed self-administered pants were offered an opportunity to take diabetes-related Web sites. Post Office is a BIOSAFE kits (13). the IDSMP. The lack of adequate usual section where participants and facilitators After returning A1C kits, participants medical care and chronic health dispari- can write private, individual messages to were randomized using a random- ties among the AI/AN subset, as well as each other. Help is a section where par- numbers table. Roughly two-thirds be- the longstanding mistrust of research in ticipants can e-mail the moderators or came treatment subjects and one-third many AI/AN communities, were reasons program administrators. The latter is also continued with usual care (no program or the AI/AN subset was randomized using available via a toll-free telephone line. other treatment offered). Treatment sub- the waitlist control design. A pilot study In addition to the Web program, each jects were further randomized one for one of 27 AI/AN and 27 non-AI/AN partici- participant received a copy of the book, to receive follow-up reinforcement pants with diabetes had confirmed the Living a Healthy Life with Chronic Condi- (membership in a listserve discussion feasibility of the online programs for this tions (11). Specific sections of this book group) or no reinforcement. Usual care population (14). are referenced in the Learning Center. consisted of whatever care participants Health status, health behaviors, The book is used as a reference not as a had been previously receiving and ranged health care utilization, patient activation, text. Thus, the program consists of the from community clinics to specialist care. and self-efficacy were measured at each online interactive training plus the book. Usual-care participants were not re- time point. The specific measures were stricted from seeking additional care or based on diabetes-related problems iden- Facilitators programs. All participants received a $10 tified in participant focus groups and on Two peers facilitate each program. Facil- Amazon.com certificate after completing self-efficacy theory (15). The primary out- itators were previously trained as self- each questionnaire and returning their come measure was A1C, measured using management small-group leaders and A1C sample. capillary blood obtained with self- 1276 DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 care.diabetesjournals.org Lorig and Associates administered BIOSAFE kits. These have treatment noncompleters were then two subsets of the original study sample: an expected nondiabetic range of 3.8 –5.9 compared. AI/ANs and participants with baseline compared with 4 – 6 for National Glyco- Reinforcement. ANCOVA models were A1C 7%. hemoglobin Standardization Program used to compare reinforced with unrein- standards (16). A paired duplicate speci- forced program participants. Six- and 18- men comparison with the whole-blood month outcomes were the dependent RESULTS method at Stanford Hospital Laboratories variables with demographic variables and showed excellent correlation and preci- the outcome variable at baseline included Participation sion. These assays have independently as covariates. Least-square means (com- Approximately 36% of participants found been shown to be reliable and valid (16). puted as part of the ANCOVA procedure the Web site through links on the Internet The A1C measure was not available for and adjusted for covariates) were used to or search engines. Another 21% learned the 18-month comparisons because BIO- determine if there were significant differ- of the study via e-mail or e-mail newslet- SAFE ceased operation early in the 18- ences between the treatment groups ran- ters; 9% were referred by relatives, month data collection. Health-related domized to reinforcement and no friends, or coworkers; 17% were referred distress was measured by the health dis- reinforcement. through print media; and 10% were re- tress scale, adapted from the Medical Out- Six-month outcomes. ANCOVA mod- ferred by health professionals. The AI/AN come Study (17). The activity limitations els compared program and control partic- subset discovered the study through the scale, which measures the impact of dis- ipants at 6 months. As reinforcement Web (29%, including 5% who found out ease on role activities such as recreation proved to have no effect on the outcomes about the study via tribal Web sites) and and chores, was developed for an earlier (see results below), reinforced and unre- e-mail (20%). Larger numbers were re- study (18). Depression was measured by inforced participants were combined to ferred by relatives or acquaintances the Patient Health Questionaire (PHQ)-9 create one treatment group, which was (21%), and 18% found the study through (19). A physical activities scale measured then compared with the control partici- print media (including 15% via AI/AN- total minutes per week of aerobic exercise pants. Separate comparisons of the oriented media). (18). control subjects and reinforced and unre- A total of 1,463 people visited the Tertiary measures included the 13- inforced program participants are also Web site and left contact information (on- item short-form Patient Activation Mea- presented. All subjects, irrespective of the line appendix Fig. A1). Of these, 1,019 sure (PAM) and diabetes self-efficacy. number of weeks they participated in the completed enrollment screening and pro- PAM measures patient self-reported intervention, were included in the analy- ceeded to the baseline questionnaire. A knowledge, skill, and confidence for ses. Least-square means (adjusted for co- further 48 were disqualified, 22 subse- managing their chronic condition (20). variates) were used to determine if there The diabetes self-efficacy scale was devel- were significant differences between the quently declined, 74 failed to complete oped for a small-group diabetes program program participants and the randomized consent or baseline questionnaires, and (2) and based on earlier chronic-disease usual-care control group after controlling 104 failed to complete A1C testing. The self-efficacy scales (18). for baseline outcome values and demo- remaining 761 participants completed Health care utilization over the prior graphic covariates. baseline assessments and were random- 6 months was measured by self-report. In ANCOVA models were repeated, ized to one of three groups: usual-care a study comparing the validity of self- adding interaction terms of all baseline control group (270), the online program reported with chart audit (21), there were outcome variables with randomization. (259), or the online program plus list- no biases toward improved reporting over This was to ascertain if existing conditions serve e-mail reinforcement (232). Subse- time. Details of the psychometric proper- might moderate the effectiveness of the quently, 27 withdrew or dropped out and ties for most of the measures can be found program and help determine the charac- 2 died before completing the 6-month at http://patienteducation.stanford.edu/ teristics of participants most likely to ben- questionnaire. Thus, 732 continued in research. efit. Analyses were done using both actual the study for 6 months. Of those continu- data collected and intent-to-treat meth- ing, 645 (85%) completed the 6-month odology, based on substituting last ac- questionnaire. These included 238 con- Data analysis quired data for missing data. In the case of trol subjects and 395 participants in the Baseline randomization. T tests were 6-month outcomes, this resulted in the online program (109 unreinforced treat- used to compare baseline IDSMP partici- assumption of no change from baseline. P ments and 186 reinforced treatments). pants with usual-care participants and to values are interpreted within each cate- Between August 2006 and September compare baseline reinforced with unrein- gory of outcome (A1C, three health indi- 2007, 21 programs were held with a mean forced IDSMP participants. We included cators, one health behavior, self-efficacy, of 23 participants per program. all variables demonstrating significant patient activation, and utilization). The AI/AN recruitment resulted in a differences at baseline as covariates in Eighteen-month outcomes. Random- sample that included 110 AI/AN partici- subsequent multivariate analyses at 6 and ized program participants were com- pants. (see online appendix Fig. A2). Af- 18 months. pared to the usual-care control group at ter 6 months, AI/AN control subjects were Noncompleters. To test the potential ef- 18 months, using the methodology allowed to enroll in the program and thus fect of dropouts, we compared the base- (ANCOVA) discussed above. were no longer part of the randomized line variables for those who failed to Subgroup analyses. Six-month analyses study. Of 651 remaining (non-AI/AN) complete the 6-month questionnaires (ANCOVA models) comparing random- with those who had completed question- ized treatment participants and usual- study participants, 528 (81%) completed naires, utilizing t tests. Control versus care control subjects were then done for 18-month questionnaires. care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 1277 Online diabetes self-management program Baseline control noncompleters (see online appen- tient activation (P  0.016, 0.007, Study participants were predominantly dix Table A4). respectively; online appendix Table A6). non-Hispanic white (76%), female Other 18-month change score differences (73%), married (66%), and well educated Six-month randomized outcomes were not significant. Intent-to-treat meth- (mean of 15.7 years of education). The Table 1 provides information regarding odology resulted in the P value for PAM average age was 54.3 years. The only sta- the changes in outcome variables for the increasing to 0.052. tistical difference between the random- control and treatment participants. Be- ized treatment and control groups was cause reinforcement was not associated Reinforcement study percentage married (78 vs. 71%, P  with any improvements (see below under Online appendix Fig. A3 gives informa- 0.034; online appendix Table A2). Per- REINFORCEMENT STUDY), the two treatment tion about the participants in the rein- centage married, as well as other demo- groups were combined for the 6-month forcement study. At 6 months, there was graphic variables, were included as comparison to usual-care control sub- only one significant difference between covariates in subsequent ANCOVA. The jects, as well as kept separate. Treatment reinforced (n  186) and unreinforced control subjects had slightly higher PHQ participants, when compared with usual- (n  209) participants. The unreinforced depression levels at baseline (Table A3). care control subjects, had significantly participants had greater improvement in The mean baseline A1C level at baseline lower A1C (P 0.05) as well as improve- health distress (P  0.007; online appen- was 6.44%, relatively low for a population ments in patient activation (PAM) and dix Table A7). At 18 months, there again with diabetes. self-efficacy (0.021 and 0.001, respec- was one variable that was significantly dif- The AI/AN subset represented 70 tively). Health behavior and utilization ferent. The unreinforced participants had tribal groups. They were slightly younger changes were not significantly different a greater reduction in depression (P than the non-AI/AN participants (mean for treatment compared with control 0.033; online appendix Table A8). Intent- age 51 vs. 55 years, P  0.001) and were group participants. When intent-to-treat to-treat methodology did not change the less likely to be married (57 vs. 68%, P  analyses were used, PAM and self-efficacy results. 0.035). Demographics for AI/ANs by ran- remained significant, while the P value for domization are given in Appendix Table A1C increased to 0.060. High A1C subgroup A2. The AI/AN subset also had higher When ANCOVAs were rerun with When only participants with baseline baseline mean A1C (6.9 vs. 6.4, P  baseline randomization interaction terms A1C 7.0% are included at 6 months 0.001). None of the other outcome vari- included in the models, the interaction of (online appendix Table A9), the differ- ables differed significantly from non-AI/ A1C with randomization was significant ence between treatment and control for ANs at baseline. in predicting 6-month A1C (P  0.001). A1C was 0.614 (P  0.010, effect size AI/AN versus non-AI/AN interactions 0.499). Self-efficacy was also statistically with randomization were significant in significant (P  0.040), although the ef- Program usage predicting 6-month health distress, activ- fect size was small. Case et al. (22) conducted a study of the ity limitation, and physician visits. These IDSMP utilization by 45 participants (15 two initial conditions were then exam- AI/AN subgroup each African American, Non-Hispanic ined in more detail below. Baseline self- At 6 months, the AI/AN subsample was white, and AI/ANs). The median number efficacy also had significant interactions underpowered (n  73). Despite the low of days for writing messages for all races with randomization and appears to be a number of cases, there were significant was 32 (30 for African Americans, 37 for moderating variable, suggesting that decreases in health distress and activity Non-Hispanic whites, and 28 for AI/ lower baseline self-efficacy was associated limitation for AI/AN program participants ANs), with 80% of participants writing with better outcomes. This will be exam- compared with control subjects (Table 2). messages over a period of at least 21 days ined elsewhere. While not statistically significant, the A1C or half the length of the workshop. The change score difference between the two median number of messages per partici- Eighteen-month randomized groups was nearly 0.3. The treatment pant was 17 and the mean was 25. There outcomes group had a statistically significant in- were few differences among the racial The comparison of 18-month completers crease in physician visits. Using intent-to- groups, although AI/ANs logged for a to noncompleters showed few differ- treat methodology, activity limitation shorter time period than non-Hispanic ences: the noncompleters were younger, remained significant, while health dis- whites. There were few differences in the had higher A1C, and higher health dis- tress and physician visits became mar- content of the posts. tress at baseline (online appendix Table ginal. Tables A10 and A11 in the online A8). There were no significant differences appendix present the 6-month data for Six-month noncompleters between the participant and usual-care AI/ANs and non-AI/ANs separately. There were few significant differences at control noncompleters. We could not in- baseline between those who completed clude A1C analyses at 18 months because CONCLUSIONS — At 6 months, re- 6-month questionnaires and those who of the closure of the laboratory. Results sults were mixed. The changes in the pri- did not. Noncompleters were younger, from a second laboratory could not be ad- mary outcome variable (A1C) had a small less likely to be married, and less likely to equately correlated with the original lab. (effect size  0.111) but statistically sig- be non-Hispanic white. They had higher Of the remaining outcome variables, two nificant difference between treatment and mean baseline A1C and higher levels of had significantly greater improvements usual-care control groups when only health distress. However, there were no for program participants as compared looking at actual cases (P  0.039). The significant differences between the partic- with the usual-care participants: self- two tertiary outcomes, patient activation ipant noncompleters and the usual-care efficacy to manage diabetes and PAM pa- and self-efficacy, both improved for treat- 1278 DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 care.diabetesjournals.org Table 1—Six-month change scores, diabetes online, all participants Treatment, no Treatment and Treatment C vs. treatment reinforcement reinforcement combined T vs. control R vs. control R vs. T (T  R) P P P P Outcome variable Control T R T plus R P (ITT) (Actual) P (ITT) (Actual) P (ITT) (Actual) P (ITT) (Actual) n 238 209 186 395 A1C2 0.126  0.779 0.034  0.844 0.018  0.862 0.009  0.852 0.064 0.036 0.176 0.162 0.653 0.530 0.060 0.039 Health distress (0–5)2 0.257  0.844 0.348  1.03 0.082  0.988 0.203  1.02 0.231 0.078 .089 0.168 0.005 0.003 0.822 0.771 Activity limitation (0–4)2 0.034  0.848 0.019  0.869 0.009  0.982 0.006  0.923 0.217 0.200 0.453 0.425 0.655 0.664 0.243 0.219 PHQ depression (0–27)2 0.836  3.82 1.072  4.44 0.398  4.10 0.754  4.26 0.413 0.183 0.465 0.687 0.131 0.099 0.931 0.558 PAM patient activation (0–100)1 3.63  14.4 6.24  14.5 5.09  14.3 5.70  14.4 0.083 0.035 0.230 0.069 0.631 0.827 0.085 0.021 Self-efficacy (1–10)1 0.203  1.70 0.321  1.99 0.160  1.73 0.245  1.87 <0.001 <0.001 0.004 0.001 0.656 0.760 <0.001 <0.001 Aerobic exercise (min/week)1 1.97  130 12.09  145 1.41  167 7.04  156 0.496 0.238 0.799 0.306 0.687 0.905 0.579 0.194 Physician visits (last 6 months) 0.198  3.25 0.121  3.53 0.239  3.55 0.177  3.54 0.809 0.679 0.722 0.611 0.906 0.967 0.730 0.589 Data are outcome variable (possible ranges). P values are from ANCOVA models controlling for baseline outcome variable and demographic variables and assess the likelihood that there would have been no difference between the treatment and control group. Arrows indicate desirable directions. ITT, intent to treat, baseline value carried forward (no change) for missing 6-month outcomes; T, treatment program without reinforcement; R, treatment program with listserv peer-support reinforcement. Possible ranges are given in parentheses next to outcome variable names, and arrows indicate desirable directions. Significant P-values are bolded. Lorig and Associates ment participants compared with usual- care control subejects. However secondary outcomes did not improve. None of the three health indicators showed significant differences, nor were the amount of exercise or number of phy- sician visits significantly changed. At 18 months, PAM patient activation and self- efficacy were significantly increased for program participants, although PAM be- came marginal when intent-to-treat methodology was used. Other outcomes variables were not significant at 18 months. Randomization to listserve e-mail peer-support reinforcement did not improve the outcomes for program participants, either at 6 or 18 months. Surprisingly the attempt at reinforce- ment was not effective. We encountered a similar finding in another study, using a different reinforcement technique (auto- mated follow-up phone calls) (3). Further study, including more attention to how the listserve was utilized, would be re- quired to determine if only our particular attempts at reinforcement were unsuc- cessful or if follow-up in general is not called for with similar self-management programs. In addition, further study of program fidelity and how program partic- ipants utilized the program, including any possible dose effect, would be desir- able. The lack of a detailed analysis of ef- fects of program utilization, as well as analyses of possible mediating effects of secondary and tertiary variables, is an im- portant limitation but was beyond the scope of this study. A further limitation of the study was the relatively low mean A1C at baseline. A large portion of the participants were in control and more likely to get worse rather than better due to both a floor effect and regression to the mean. When we looked at the subgroups of those with baseline A1C 7.0% at baseline, the dif- ferences in improvements in A1C in- creased from a very modest effect size of 0.11 for the entire randomized sample to a clinically significant effect size of 0.50. This suggests that the program may prove more successful if targeted to patients with higher A1C. When we limited ourselves to the AI/AN subset (who had a mean baseline A1C of nearly 7.0% compared with the total sample mean of 6.4%), we saw im- provements in health indicators (activity limitations and health distress with signif- icant effect sizes of 0.48 and 0.34, respec- tively). Although the difference for A1C was not statistically significant, with a care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 1279 Online diabetes self-management program Table 2—AI/AN subgroup, baseline and 6-month changes Baseline 6-month change Control vs. treatment Effect P Outcome variable Control Treatment Control Treatment size (ITT) P n 50 60 38 35 A1C2 6.71  1.25 7.12  1.59 0.206  0.973 0.088  1.24 0.251 0.288 0.379 Health distress (0–5)2 2.26  1.24 2.06  1.24 0.151  0.730 0.714  0.993 0.484 0.025 0.004 Activity limitation (0–4)2 1.48  1.12 1.14  1.03 0.092  0.843 0.257  0.986 0.337 0.028 0.012 PHQ depression (0–27)2 8.60  6.35 8.33  5.67 0.737  3.84 1.600  5.33 0.146 0.677 0.214 PAM patient activation (0–100)1 63.3  15.8 63.8  14.8 4.47  16.4 3.78  13.3 0.054 0.612 0.722 Self-efficacy (1–10)1 6.74  1.86 6.39  2.29 0.056  1.40 0.350  2.40 0.214 0.896 0.584 Aerobic exercise (min/week)1 93.1  133 81.2  103 3.32  118 18.62  112 0.111 0.789 0.810 Physician visits (last 6 months) 3.42  3.53 3.17  3.38 0.658  1.86 0.914  3.85 0.494 0.051 0.019 Data are means SD. Effect sizes are computed as the difference in change scores between treatment and control groups divided by the pooled baseline SD. Negative effect sizes indicate that the control group did better than the treatment group. P values are from ANCOVA models controlling for baseline outcome variable and demographic variables and assess the likelihood that there would have been no difference between the treatment and control group. Possible ranges are given in parentheses next to outcome variable names, and arrows indicate desirable directions. Significant P-values are bolded. We thank Rose Sage Barone and Tanya Pod- 8. Glasgow RE, Nutting PA, Toobert DJ, properly powered sample, an effect size chiyska for programming, Martha Funnell and King DK, Strycker LA, Jex M, O’Neill C, difference of 0.25 would undoubtedly David McCulloch for content development, Whitesides H, Merenich J. Effects of a have been statistically significant. and Angela Devlin for follow-up data brief computer-assisted diabetes self- The methods of recruitment and de- collection. management intervention on dietary, bi- sign of the study may have contributed to ological and quality-of-life outcomes. low differences between treatment and Chronic Illn;2:27–38 control participants. A high proportion of 9. Wangberg SC. An Internet-based diabe- those who joined the study were actively References tes self-care intervention tailored to self- 1. National Institute of Diabetes and Digestive seeking information about their disease efficacy. Health Educ Res 2008;23: and Kidney Diseases. National Diabetes Sta- when they found the study Web site. The 170 –9 tistics fact sheet: general information and control group was not offered the possi- 10. Gerber BS, Solomon MC, Shaffer TL, national estimates on diabetes in the United bility of participation in the program after Quinn MT, Lipton RB. Evaluation of an States, 2005. Washington, DC, Govt. Print- a short period of time and may have Internet diabetes self-management train- ing Office [NIH publ. no. 06-3892] ing program for adolescents and young searched for and found alternate inter- 2. Lorig K, Ritter PL, Villa FJ, Armas J. Com- adults. Diabetes Tech Ther 2007;9: ventions. The AI/AN subgroup was an ex- munity-based peer-led diabetes self-man- 60–67 ception in this regard as they were offered agement: a randomized trial. Diabetes 11. Lorig K, Holman H, Sobel D, Laurent D, the program after 6 months. This may Educ 2009;35:641–51 Gonzalez V, Minor M. Living a Healthy Life have contributed to the relative success of 3. Lorig K, Ritter PL, Villa F, Piette JD. Span- With Chronic Conditions. 3rd ed. Boulder, ish diabetes self-management with and the program within that subgroup. Colorado; Bull Publishing, 2006 without automated telephone reinforce- Although results were both encourag- 12. Jerrigan VJ. The Native American Diabetes ment: two randomized trials. Diabetes ing and discouraging, they suggest that Self-Management Program. PHD thesis. Care 2008;31:408 – 414 the program can be beneficial to people Berkeley, California, University of Cali- 4. Deakin T, McShane CE, Cade JE, Wil- with diabetes and that further study is fornia Berkeley, School of Public Health, liams RDRR. Group based training for warranted. A trial with broader recruit- unpublished self-management strategies in people with ment, limited to only those with A1C 13. Tyrrell SP, Bui TL, Maggiore, JA. Hemo- type 2 diabetes mellitus. Cochrane Data- globin A1c Detection System: A Technical 7.0%, and allowing randomized con- base Syst Rev 2005;18:CD003417 Bulletin Describing the Performance Char- trol subjects to participate in the program 5. Jackson CL, Bolen S, Brancati FL, Batts- acteristics of Capillary Blood Collected on after a 6-month trial would prove more Turner ML, Gary TL. A systematic review Filter Paper for Determination of Hemoglo- of interactive computer-assisted technol- definitive. bin A1c. Chicago, BIOSAFE Laboratories, ogy in diabetes care: interactive informa- Bulletin 4018, Rev. 99037, 2006 tion technology in diabetes care. J Gen 14. Jernigan, VBB, Lorig K. The Internet Dia- Acknowledgments — The study was sup- Intern Med. 2006;21:105–10 betes Self-Management Workshop for ported by National Institutes of Health Grant 6. Glasgow RE, Boles SM, McKay HG, Feil American Indians and Alaska Natives. 1R18DK065729 and Robert Wood Johnson EG, Barrera M Jr. The D-Net diabetes self- Health Promotion Practice, In press Foundation Grant 096223. management program: long-term imple- 15. Bandura A. Self-efficacy: toward a unify- K.L. and D.D.L. receive royalties for the mentation, outcomes, and generalization ing theory of behavioral change. Psychol book used by study participants. K.L., D.D.L., results. Prev Med 2003;36:410 –9 Rev 1977;84:191–215 and K.P. potentially receive licensing fees if the 7. Glasgow RE, Nutting PA, King DK, Nelson program is disseminated. P.L.R., V.B.B.J., and CC, Cutter G, Gaglio B, Rahm AK, White- 16. Miedema K. Standardization of HbA1c and M.G. have no conflicts of interest. No other sides H, Amthauer H. A practical random- optimal range of monitoring. Scand J Clin Lab potential conflicts of interest relevant to this ized trial to improve diabetes care. J Gen Invest 2005;65(Suppl. 240):61–72 article were reported. Intern Med 2004;19:1167–74 17. Stewart AL, Ware JE. Measuring Function- 1280 DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 care.diabetesjournals.org Lorig and Associates ing and Well Being: The Medical Outcomes 19. Kroenke K, Spitzer RL. The PHQ-9: a new Lorig KR. Self-reports of health care utili- Study Approach. Durham, NC, Duke Uni- depression diagnostic and severity mea- zation compared to provider records. versity Press, 1992 sure. Psychiatr Ann 2002;35:509 –521 J Clin Epidemiol 2001;54:136 –141 18. Lorig K, Stewart A, Ritter P, Gonza ´ lez V, 20. Hibbard JH, Mahoney ER, Stockard J, Tu- 22. Case S, Jernigan V, Gardner A, Ritter PL, Laurent D, Lynch J. Outcome Measures for sler M. Development and testing of a short Heaney CA, Lorig KR. Content and fre- Health Education and Other Health Care In- form of the patient activation measure. quency of writing on diabetes bulletin terventions. Thousand Oaks CA, Sage Health Serv Res 2005;40:1918 –30 boards: does race make a difference? Publications, 1996 21. Ritter PL, Kaymaz H, Stewart A, Sobel DS, J Med Internet Res 2009;11:e22 care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 1281 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Diabetes Care Pubmed Central

Online Diabetes Self-Management Program

Loading next page...
 
/lp/pubmed-central/online-diabetes-self-management-program-dHgqZY2PPv

References (40)

Publisher
Pubmed Central
Copyright
© 2010 by the American Diabetes Association.
ISSN
0149-5992
eISSN
1935-5548
DOI
10.2337/dc09-2153
Publisher site
See Article on Publisher Site

Abstract

Emerging Treatments and Technologies ORIGINAL ARTICLE Online Diabetes Self-Management Program A randomized study 1 1 KATE LORIG, RN, DRPH MAURICE GREEN, PHD to our knowledge, examining such a pro- PHILIP L. RITTER, PHD VALARIE BLUE BIRD JERNIGAN, PHD gram among AI/ANs. DIANA D. LAURENT, MPH SIOBHAN CASE, BA The Cochrane Collaboration re- KATHRYN PLANT, MPH viewed group-based training for type 2 diabetes (4). They found 11 studies that met their criteria. Eight of these were ran- OBJECTIVE — We hypothesized that people with type 2 diabetes in an online diabetes domized studies and three were con- self-management program, compared with usual-care control subjects, would 1) demonstrate trolled studies. All of the interventions reduced A1C at 6 and 18 months, 2) have fewer symptoms, 3) demonstrate increased exercise, were taught by health professionals. One and 4) have improved self-efficacy and patient activation. In addition, participants randomized study took place in a community setting, to listserve reinforcement would have better 18-month outcomes than participants receiving no and one reported a mean baseline A1C reinforcement. 7%. Jacksan et al. (5) conducted a system- RESEARCH DESIGN AND METHODS — A total of 761 participants were randomized atic review of computer-assisted technol- to 1) the program, 2) the program with e-mail reinforcement, or 3) were usual-care control ogies in diabetes prior to 2004. They subjects (no treatment). This sample included 110 American Indians/Alaska Natives (AI/ANs). Analyses of covariance models were used at the 6- and 18-month follow-up to compare groups. found four articles involving patient edu- cation. In an early study, (6) groups were RESULTS — At 6 months, A1C, patient activation, and self-efficacy were improved for pro- randomized to basic diabetes informa- gram participants compared with usual care control subjects (P 0.05). There were no changes tion, tailored online coaching, or peer in other health or behavioral indicators. The AI/AN program participants demonstrated im- support. Improvement in health behav- provements in health distress and activity limitation compared with usual-care control subjects. iors and psychological outcome were The subgroup with initial A1C7% demonstrated stronger improvement in A1C (P 0.01). At found in all three groups, with no differ- 18 months, self-efficacy and patient activation were improved for program participants. A1C was ences between groups. Glasgow et al. (7) not measured. Reinforcement showed no improvement. showed that a computer-assisted inter- vention was practicable and acceptable in CONCLUSIONS — An online diabetes self-management program is acceptable for people with type 2 diabetes. Although the results were mixed they suggest 1) that the program may have a real-world setting and resulted in im- beneficial effects in reducing A1C, 2) AI/AN populations can be engaged in and benefit from provements in recommended services. In online interventions, and 3) our follow-up reinforcement appeared to have no value. a low-intensity computer program study, short-term outcomes were promising but Diabetes Care 33:1275–1281, 2010 not significant (8). Wengberg (9), utiliz- ing an computer diabetes intervention, ype 2 diabetes affects 9.6% of the with type 2 diabetes are willing or able to has suggested that self-efficacy may func- adult population, and its prevalence participate in small-group programs, nor tion as a moderator for diabetes behavior is increasing (1). While the need for are such programs likely to be available in change, and Gerber et al. (10) have dem- self-management support is well docu- all locations. onstrated usability of an Internet program mented, most diabetes education studies There are few studies of community- for young inner-city adults. In summary, have taken place in clinical settings and based diabetes education programs for Internet-based educational programs targeted those who have a high A1C (usu- American Indians/Alaskan Natives (AI/ have been demonstrated to change behav- ally 7%). Recent community-level, ANs). We report on a randomized, con- iors and sometimes health status. We peer-led, small-group diabetes self- trolled trial of an Internet-based diabetes were unable to find computer-based stud- management programs have shown self-management program (IDSMP) in- ies demonstrating changes in A1C. promise (2,3). However, not all patients cluding AI/ANs. This was the first study, ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● RESEARCH DESIGN AND METHODS — We report on a ran- From the Stanford Patient Education Research Center, Stanford University School of Medicine, Palo Alto, 2 3 California; the Stanford Prevention Research Center, Stanford University, Palo Alto, California; and Yale domized 6-month trial of the IDSMP, School of Medicine, New Haven, Connecticut. with an 18-month follow-up. We hypoth- Corresponding author: Philip L. Ritter, philr@stanford.edu. esized that participants in the IDSMP, Received 20 November 2009 and accepted 5 March 2010. Published ahead of print at http://care. compared with usual-care control sub- diabetesjournals.org on 18 March 2010. DOI: 10.2337/dc09-2153. Clinical trials reg. no. NCT00185601, clinicaltrials.gov. jects, would demonstrate 1) reduced A1C © 2010 by the American Diabetes Association. Readers may use this article as long as the work is properly at 6 and 18 months, 2) have fewer symp- cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons. toms, 3) have increased exercise, and 4) org/licenses/by-nc-nd/3.0/ for details. have improved self-efficacy and patient The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. activation. We also hypothesized that par- care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 1275 Online diabetes self-management program ticipants randomized to a follow-up list- had taken the IDSMP (as nonstudy sub- Randomized study serve, peer-support group would have jects). Facilitators assist participants by The randomized IDSMP group was com- better 18-month outcomes than partici- reminding them to log on, modeling ac- pared with the usual-care control group at pants receiving no follow-up. tion planning and problem-solving, offer- 6 months. If the reinforcement study (be- ing encouragement, and posting to the low) had shown that reinforcement par- The IDSMP bulletin boards. They also monitor the ticipants had greater improvements than The asynchronous, 6-week, IDSMP is daily posts for safety and report inappro- unreinforced IDSMP participants, the two based on English- and Spanish-language priate posts to the investigators. All facil- IDSMP groups would be compared with peer-led small-group diabetes self- itation takes place online, mainly via posts control participants separately. If there management programs (2,3). The IDSMP within the program pages. Each partici- were few differences, the two randomized consists of six weekly sessions. Partici- pant receives personalized responses IDSMP groups would be combined and pants logged on individually to the ses- from facilitators during each weekly ses- compared with the usual-care control sions, which were available for the entire sion. Unlike the small-group program, fa- group. week. The topics covered are shown in cilitators do not deliver content, as this is After 6 months, usual-care partici- online appendix Table A1 (available in scripted in the Learning Center. Programs pants recruited as part of the AI/AN sub- the online appendix at http://care. were facilitated by 16 different people, group were offered the program. All other diabetesjournals.org/cgi/content/full/dc09- half with diabetes. Each program has at usual-care participants continued as con- 2153/DC1). least one facilitator with diabetes. The trol subjects through the 18 months of the A password-protected homepage study was approved by the Stanford study. Follow-up data collected at 18 provides access to the weekly activities, School of Medicine Institutional Review months allowed comparison of IDSMP including The Learning Center, where the Board. participants to usual-care subjects, ex- program content is offered in 20 –30 new cluding the AI/AN subset. Web pages weekly. Each week, partici- Participants and data collection pants are asked to reply to a question such Participants were aged 18 years, were Reinforcement study as “What problems do you have because not pregnant or in care for cancer, had The reinforcement study compared of your diabetes?” and to make a specific physician-verified type 2 diabetes, and IDSMP treatment participants who had action plan. The questions and action had access to the Internet. Recruitment no reinforcement with those who had plans are posted on bulletin boards in the was largely via the Internet, although been randomized to a listserve discussion Discussion Center, where they can be print and broadcast media were also uti- group. The discussion group was in- seen by all participants. lized. Special effort was made to recruit tended to reinforce any benefits of the The Discussion Center is made up of AI/AN participants using Web sites and program by providing peer support. four interactive threaded bulletin boards media associated with tribal and AI/AN Comparisons were made at 6 and at 18 (Action Planning, Problem Solving, Diffi- organizations. This was accomplished months. The AI/AN participants were in- cult Emotions, and Celebrations) popu- utilizing the expertise of an AI/AN re- cluded in the 18-month reinforcement lated by responses made in the Learning searcher (12). study. Center, as well as new threads started by All consents and questionnaires were participants whenever they wish. A typi- administered online. Participants con- AI/AN study cal program of 20 –25 participants results tacted the study by going to the Web site, AI/AN participants were randomized in 500 or more posts. My Tools consists of where they were screened for eligibility with other participants but entered the exercise and medication logs, audio relax- and were asked to complete consent and randomized study for only 6 months, af- ation exercises, meal planning, and glu- baseline questionnaires. A1C was ob- ter which time AI/AN usual-care partici- cose-monitoring tools and links to other tained using mailed self-administered pants were offered an opportunity to take diabetes-related Web sites. Post Office is a BIOSAFE kits (13). the IDSMP. The lack of adequate usual section where participants and facilitators After returning A1C kits, participants medical care and chronic health dispari- can write private, individual messages to were randomized using a random- ties among the AI/AN subset, as well as each other. Help is a section where par- numbers table. Roughly two-thirds be- the longstanding mistrust of research in ticipants can e-mail the moderators or came treatment subjects and one-third many AI/AN communities, were reasons program administrators. The latter is also continued with usual care (no program or the AI/AN subset was randomized using available via a toll-free telephone line. other treatment offered). Treatment sub- the waitlist control design. A pilot study In addition to the Web program, each jects were further randomized one for one of 27 AI/AN and 27 non-AI/AN partici- participant received a copy of the book, to receive follow-up reinforcement pants with diabetes had confirmed the Living a Healthy Life with Chronic Condi- (membership in a listserve discussion feasibility of the online programs for this tions (11). Specific sections of this book group) or no reinforcement. Usual care population (14). are referenced in the Learning Center. consisted of whatever care participants Health status, health behaviors, The book is used as a reference not as a had been previously receiving and ranged health care utilization, patient activation, text. Thus, the program consists of the from community clinics to specialist care. and self-efficacy were measured at each online interactive training plus the book. Usual-care participants were not re- time point. The specific measures were stricted from seeking additional care or based on diabetes-related problems iden- Facilitators programs. All participants received a $10 tified in participant focus groups and on Two peers facilitate each program. Facil- Amazon.com certificate after completing self-efficacy theory (15). The primary out- itators were previously trained as self- each questionnaire and returning their come measure was A1C, measured using management small-group leaders and A1C sample. capillary blood obtained with self- 1276 DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 care.diabetesjournals.org Lorig and Associates administered BIOSAFE kits. These have treatment noncompleters were then two subsets of the original study sample: an expected nondiabetic range of 3.8 –5.9 compared. AI/ANs and participants with baseline compared with 4 – 6 for National Glyco- Reinforcement. ANCOVA models were A1C 7%. hemoglobin Standardization Program used to compare reinforced with unrein- standards (16). A paired duplicate speci- forced program participants. Six- and 18- men comparison with the whole-blood month outcomes were the dependent RESULTS method at Stanford Hospital Laboratories variables with demographic variables and showed excellent correlation and preci- the outcome variable at baseline included Participation sion. These assays have independently as covariates. Least-square means (com- Approximately 36% of participants found been shown to be reliable and valid (16). puted as part of the ANCOVA procedure the Web site through links on the Internet The A1C measure was not available for and adjusted for covariates) were used to or search engines. Another 21% learned the 18-month comparisons because BIO- determine if there were significant differ- of the study via e-mail or e-mail newslet- SAFE ceased operation early in the 18- ences between the treatment groups ran- ters; 9% were referred by relatives, month data collection. Health-related domized to reinforcement and no friends, or coworkers; 17% were referred distress was measured by the health dis- reinforcement. through print media; and 10% were re- tress scale, adapted from the Medical Out- Six-month outcomes. ANCOVA mod- ferred by health professionals. The AI/AN come Study (17). The activity limitations els compared program and control partic- subset discovered the study through the scale, which measures the impact of dis- ipants at 6 months. As reinforcement Web (29%, including 5% who found out ease on role activities such as recreation proved to have no effect on the outcomes about the study via tribal Web sites) and and chores, was developed for an earlier (see results below), reinforced and unre- e-mail (20%). Larger numbers were re- study (18). Depression was measured by inforced participants were combined to ferred by relatives or acquaintances the Patient Health Questionaire (PHQ)-9 create one treatment group, which was (21%), and 18% found the study through (19). A physical activities scale measured then compared with the control partici- print media (including 15% via AI/AN- total minutes per week of aerobic exercise pants. Separate comparisons of the oriented media). (18). control subjects and reinforced and unre- A total of 1,463 people visited the Tertiary measures included the 13- inforced program participants are also Web site and left contact information (on- item short-form Patient Activation Mea- presented. All subjects, irrespective of the line appendix Fig. A1). Of these, 1,019 sure (PAM) and diabetes self-efficacy. number of weeks they participated in the completed enrollment screening and pro- PAM measures patient self-reported intervention, were included in the analy- ceeded to the baseline questionnaire. A knowledge, skill, and confidence for ses. Least-square means (adjusted for co- further 48 were disqualified, 22 subse- managing their chronic condition (20). variates) were used to determine if there The diabetes self-efficacy scale was devel- were significant differences between the quently declined, 74 failed to complete oped for a small-group diabetes program program participants and the randomized consent or baseline questionnaires, and (2) and based on earlier chronic-disease usual-care control group after controlling 104 failed to complete A1C testing. The self-efficacy scales (18). for baseline outcome values and demo- remaining 761 participants completed Health care utilization over the prior graphic covariates. baseline assessments and were random- 6 months was measured by self-report. In ANCOVA models were repeated, ized to one of three groups: usual-care a study comparing the validity of self- adding interaction terms of all baseline control group (270), the online program reported with chart audit (21), there were outcome variables with randomization. (259), or the online program plus list- no biases toward improved reporting over This was to ascertain if existing conditions serve e-mail reinforcement (232). Subse- time. Details of the psychometric proper- might moderate the effectiveness of the quently, 27 withdrew or dropped out and ties for most of the measures can be found program and help determine the charac- 2 died before completing the 6-month at http://patienteducation.stanford.edu/ teristics of participants most likely to ben- questionnaire. Thus, 732 continued in research. efit. Analyses were done using both actual the study for 6 months. Of those continu- data collected and intent-to-treat meth- ing, 645 (85%) completed the 6-month odology, based on substituting last ac- questionnaire. These included 238 con- Data analysis quired data for missing data. In the case of trol subjects and 395 participants in the Baseline randomization. T tests were 6-month outcomes, this resulted in the online program (109 unreinforced treat- used to compare baseline IDSMP partici- assumption of no change from baseline. P ments and 186 reinforced treatments). pants with usual-care participants and to values are interpreted within each cate- Between August 2006 and September compare baseline reinforced with unrein- gory of outcome (A1C, three health indi- 2007, 21 programs were held with a mean forced IDSMP participants. We included cators, one health behavior, self-efficacy, of 23 participants per program. all variables demonstrating significant patient activation, and utilization). The AI/AN recruitment resulted in a differences at baseline as covariates in Eighteen-month outcomes. Random- sample that included 110 AI/AN partici- subsequent multivariate analyses at 6 and ized program participants were com- pants. (see online appendix Fig. A2). Af- 18 months. pared to the usual-care control group at ter 6 months, AI/AN control subjects were Noncompleters. To test the potential ef- 18 months, using the methodology allowed to enroll in the program and thus fect of dropouts, we compared the base- (ANCOVA) discussed above. were no longer part of the randomized line variables for those who failed to Subgroup analyses. Six-month analyses study. Of 651 remaining (non-AI/AN) complete the 6-month questionnaires (ANCOVA models) comparing random- with those who had completed question- ized treatment participants and usual- study participants, 528 (81%) completed naires, utilizing t tests. Control versus care control subjects were then done for 18-month questionnaires. care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 1277 Online diabetes self-management program Baseline control noncompleters (see online appen- tient activation (P  0.016, 0.007, Study participants were predominantly dix Table A4). respectively; online appendix Table A6). non-Hispanic white (76%), female Other 18-month change score differences (73%), married (66%), and well educated Six-month randomized outcomes were not significant. Intent-to-treat meth- (mean of 15.7 years of education). The Table 1 provides information regarding odology resulted in the P value for PAM average age was 54.3 years. The only sta- the changes in outcome variables for the increasing to 0.052. tistical difference between the random- control and treatment participants. Be- ized treatment and control groups was cause reinforcement was not associated Reinforcement study percentage married (78 vs. 71%, P  with any improvements (see below under Online appendix Fig. A3 gives informa- 0.034; online appendix Table A2). Per- REINFORCEMENT STUDY), the two treatment tion about the participants in the rein- centage married, as well as other demo- groups were combined for the 6-month forcement study. At 6 months, there was graphic variables, were included as comparison to usual-care control sub- only one significant difference between covariates in subsequent ANCOVA. The jects, as well as kept separate. Treatment reinforced (n  186) and unreinforced control subjects had slightly higher PHQ participants, when compared with usual- (n  209) participants. The unreinforced depression levels at baseline (Table A3). care control subjects, had significantly participants had greater improvement in The mean baseline A1C level at baseline lower A1C (P 0.05) as well as improve- health distress (P  0.007; online appen- was 6.44%, relatively low for a population ments in patient activation (PAM) and dix Table A7). At 18 months, there again with diabetes. self-efficacy (0.021 and 0.001, respec- was one variable that was significantly dif- The AI/AN subset represented 70 tively). Health behavior and utilization ferent. The unreinforced participants had tribal groups. They were slightly younger changes were not significantly different a greater reduction in depression (P than the non-AI/AN participants (mean for treatment compared with control 0.033; online appendix Table A8). Intent- age 51 vs. 55 years, P  0.001) and were group participants. When intent-to-treat to-treat methodology did not change the less likely to be married (57 vs. 68%, P  analyses were used, PAM and self-efficacy results. 0.035). Demographics for AI/ANs by ran- remained significant, while the P value for domization are given in Appendix Table A1C increased to 0.060. High A1C subgroup A2. The AI/AN subset also had higher When ANCOVAs were rerun with When only participants with baseline baseline mean A1C (6.9 vs. 6.4, P  baseline randomization interaction terms A1C 7.0% are included at 6 months 0.001). None of the other outcome vari- included in the models, the interaction of (online appendix Table A9), the differ- ables differed significantly from non-AI/ A1C with randomization was significant ence between treatment and control for ANs at baseline. in predicting 6-month A1C (P  0.001). A1C was 0.614 (P  0.010, effect size AI/AN versus non-AI/AN interactions 0.499). Self-efficacy was also statistically with randomization were significant in significant (P  0.040), although the ef- Program usage predicting 6-month health distress, activ- fect size was small. Case et al. (22) conducted a study of the ity limitation, and physician visits. These IDSMP utilization by 45 participants (15 two initial conditions were then exam- AI/AN subgroup each African American, Non-Hispanic ined in more detail below. Baseline self- At 6 months, the AI/AN subsample was white, and AI/ANs). The median number efficacy also had significant interactions underpowered (n  73). Despite the low of days for writing messages for all races with randomization and appears to be a number of cases, there were significant was 32 (30 for African Americans, 37 for moderating variable, suggesting that decreases in health distress and activity Non-Hispanic whites, and 28 for AI/ lower baseline self-efficacy was associated limitation for AI/AN program participants ANs), with 80% of participants writing with better outcomes. This will be exam- compared with control subjects (Table 2). messages over a period of at least 21 days ined elsewhere. While not statistically significant, the A1C or half the length of the workshop. The change score difference between the two median number of messages per partici- Eighteen-month randomized groups was nearly 0.3. The treatment pant was 17 and the mean was 25. There outcomes group had a statistically significant in- were few differences among the racial The comparison of 18-month completers crease in physician visits. Using intent-to- groups, although AI/ANs logged for a to noncompleters showed few differ- treat methodology, activity limitation shorter time period than non-Hispanic ences: the noncompleters were younger, remained significant, while health dis- whites. There were few differences in the had higher A1C, and higher health dis- tress and physician visits became mar- content of the posts. tress at baseline (online appendix Table ginal. Tables A10 and A11 in the online A8). There were no significant differences appendix present the 6-month data for Six-month noncompleters between the participant and usual-care AI/ANs and non-AI/ANs separately. There were few significant differences at control noncompleters. We could not in- baseline between those who completed clude A1C analyses at 18 months because CONCLUSIONS — At 6 months, re- 6-month questionnaires and those who of the closure of the laboratory. Results sults were mixed. The changes in the pri- did not. Noncompleters were younger, from a second laboratory could not be ad- mary outcome variable (A1C) had a small less likely to be married, and less likely to equately correlated with the original lab. (effect size  0.111) but statistically sig- be non-Hispanic white. They had higher Of the remaining outcome variables, two nificant difference between treatment and mean baseline A1C and higher levels of had significantly greater improvements usual-care control groups when only health distress. However, there were no for program participants as compared looking at actual cases (P  0.039). The significant differences between the partic- with the usual-care participants: self- two tertiary outcomes, patient activation ipant noncompleters and the usual-care efficacy to manage diabetes and PAM pa- and self-efficacy, both improved for treat- 1278 DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 care.diabetesjournals.org Table 1—Six-month change scores, diabetes online, all participants Treatment, no Treatment and Treatment C vs. treatment reinforcement reinforcement combined T vs. control R vs. control R vs. T (T  R) P P P P Outcome variable Control T R T plus R P (ITT) (Actual) P (ITT) (Actual) P (ITT) (Actual) P (ITT) (Actual) n 238 209 186 395 A1C2 0.126  0.779 0.034  0.844 0.018  0.862 0.009  0.852 0.064 0.036 0.176 0.162 0.653 0.530 0.060 0.039 Health distress (0–5)2 0.257  0.844 0.348  1.03 0.082  0.988 0.203  1.02 0.231 0.078 .089 0.168 0.005 0.003 0.822 0.771 Activity limitation (0–4)2 0.034  0.848 0.019  0.869 0.009  0.982 0.006  0.923 0.217 0.200 0.453 0.425 0.655 0.664 0.243 0.219 PHQ depression (0–27)2 0.836  3.82 1.072  4.44 0.398  4.10 0.754  4.26 0.413 0.183 0.465 0.687 0.131 0.099 0.931 0.558 PAM patient activation (0–100)1 3.63  14.4 6.24  14.5 5.09  14.3 5.70  14.4 0.083 0.035 0.230 0.069 0.631 0.827 0.085 0.021 Self-efficacy (1–10)1 0.203  1.70 0.321  1.99 0.160  1.73 0.245  1.87 <0.001 <0.001 0.004 0.001 0.656 0.760 <0.001 <0.001 Aerobic exercise (min/week)1 1.97  130 12.09  145 1.41  167 7.04  156 0.496 0.238 0.799 0.306 0.687 0.905 0.579 0.194 Physician visits (last 6 months) 0.198  3.25 0.121  3.53 0.239  3.55 0.177  3.54 0.809 0.679 0.722 0.611 0.906 0.967 0.730 0.589 Data are outcome variable (possible ranges). P values are from ANCOVA models controlling for baseline outcome variable and demographic variables and assess the likelihood that there would have been no difference between the treatment and control group. Arrows indicate desirable directions. ITT, intent to treat, baseline value carried forward (no change) for missing 6-month outcomes; T, treatment program without reinforcement; R, treatment program with listserv peer-support reinforcement. Possible ranges are given in parentheses next to outcome variable names, and arrows indicate desirable directions. Significant P-values are bolded. Lorig and Associates ment participants compared with usual- care control subejects. However secondary outcomes did not improve. None of the three health indicators showed significant differences, nor were the amount of exercise or number of phy- sician visits significantly changed. At 18 months, PAM patient activation and self- efficacy were significantly increased for program participants, although PAM be- came marginal when intent-to-treat methodology was used. Other outcomes variables were not significant at 18 months. Randomization to listserve e-mail peer-support reinforcement did not improve the outcomes for program participants, either at 6 or 18 months. Surprisingly the attempt at reinforce- ment was not effective. We encountered a similar finding in another study, using a different reinforcement technique (auto- mated follow-up phone calls) (3). Further study, including more attention to how the listserve was utilized, would be re- quired to determine if only our particular attempts at reinforcement were unsuc- cessful or if follow-up in general is not called for with similar self-management programs. In addition, further study of program fidelity and how program partic- ipants utilized the program, including any possible dose effect, would be desir- able. The lack of a detailed analysis of ef- fects of program utilization, as well as analyses of possible mediating effects of secondary and tertiary variables, is an im- portant limitation but was beyond the scope of this study. A further limitation of the study was the relatively low mean A1C at baseline. A large portion of the participants were in control and more likely to get worse rather than better due to both a floor effect and regression to the mean. When we looked at the subgroups of those with baseline A1C 7.0% at baseline, the dif- ferences in improvements in A1C in- creased from a very modest effect size of 0.11 for the entire randomized sample to a clinically significant effect size of 0.50. This suggests that the program may prove more successful if targeted to patients with higher A1C. When we limited ourselves to the AI/AN subset (who had a mean baseline A1C of nearly 7.0% compared with the total sample mean of 6.4%), we saw im- provements in health indicators (activity limitations and health distress with signif- icant effect sizes of 0.48 and 0.34, respec- tively). Although the difference for A1C was not statistically significant, with a care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 1279 Online diabetes self-management program Table 2—AI/AN subgroup, baseline and 6-month changes Baseline 6-month change Control vs. treatment Effect P Outcome variable Control Treatment Control Treatment size (ITT) P n 50 60 38 35 A1C2 6.71  1.25 7.12  1.59 0.206  0.973 0.088  1.24 0.251 0.288 0.379 Health distress (0–5)2 2.26  1.24 2.06  1.24 0.151  0.730 0.714  0.993 0.484 0.025 0.004 Activity limitation (0–4)2 1.48  1.12 1.14  1.03 0.092  0.843 0.257  0.986 0.337 0.028 0.012 PHQ depression (0–27)2 8.60  6.35 8.33  5.67 0.737  3.84 1.600  5.33 0.146 0.677 0.214 PAM patient activation (0–100)1 63.3  15.8 63.8  14.8 4.47  16.4 3.78  13.3 0.054 0.612 0.722 Self-efficacy (1–10)1 6.74  1.86 6.39  2.29 0.056  1.40 0.350  2.40 0.214 0.896 0.584 Aerobic exercise (min/week)1 93.1  133 81.2  103 3.32  118 18.62  112 0.111 0.789 0.810 Physician visits (last 6 months) 3.42  3.53 3.17  3.38 0.658  1.86 0.914  3.85 0.494 0.051 0.019 Data are means SD. Effect sizes are computed as the difference in change scores between treatment and control groups divided by the pooled baseline SD. Negative effect sizes indicate that the control group did better than the treatment group. P values are from ANCOVA models controlling for baseline outcome variable and demographic variables and assess the likelihood that there would have been no difference between the treatment and control group. Possible ranges are given in parentheses next to outcome variable names, and arrows indicate desirable directions. Significant P-values are bolded. We thank Rose Sage Barone and Tanya Pod- 8. Glasgow RE, Nutting PA, Toobert DJ, properly powered sample, an effect size chiyska for programming, Martha Funnell and King DK, Strycker LA, Jex M, O’Neill C, difference of 0.25 would undoubtedly David McCulloch for content development, Whitesides H, Merenich J. Effects of a have been statistically significant. and Angela Devlin for follow-up data brief computer-assisted diabetes self- The methods of recruitment and de- collection. management intervention on dietary, bi- sign of the study may have contributed to ological and quality-of-life outcomes. low differences between treatment and Chronic Illn;2:27–38 control participants. A high proportion of 9. Wangberg SC. An Internet-based diabe- those who joined the study were actively References tes self-care intervention tailored to self- 1. National Institute of Diabetes and Digestive seeking information about their disease efficacy. Health Educ Res 2008;23: and Kidney Diseases. National Diabetes Sta- when they found the study Web site. The 170 –9 tistics fact sheet: general information and control group was not offered the possi- 10. Gerber BS, Solomon MC, Shaffer TL, national estimates on diabetes in the United bility of participation in the program after Quinn MT, Lipton RB. Evaluation of an States, 2005. Washington, DC, Govt. Print- a short period of time and may have Internet diabetes self-management train- ing Office [NIH publ. no. 06-3892] ing program for adolescents and young searched for and found alternate inter- 2. Lorig K, Ritter PL, Villa FJ, Armas J. Com- adults. Diabetes Tech Ther 2007;9: ventions. The AI/AN subgroup was an ex- munity-based peer-led diabetes self-man- 60–67 ception in this regard as they were offered agement: a randomized trial. Diabetes 11. Lorig K, Holman H, Sobel D, Laurent D, the program after 6 months. This may Educ 2009;35:641–51 Gonzalez V, Minor M. Living a Healthy Life have contributed to the relative success of 3. Lorig K, Ritter PL, Villa F, Piette JD. Span- With Chronic Conditions. 3rd ed. Boulder, ish diabetes self-management with and the program within that subgroup. Colorado; Bull Publishing, 2006 without automated telephone reinforce- Although results were both encourag- 12. Jerrigan VJ. The Native American Diabetes ment: two randomized trials. Diabetes ing and discouraging, they suggest that Self-Management Program. PHD thesis. Care 2008;31:408 – 414 the program can be beneficial to people Berkeley, California, University of Cali- 4. Deakin T, McShane CE, Cade JE, Wil- with diabetes and that further study is fornia Berkeley, School of Public Health, liams RDRR. Group based training for warranted. A trial with broader recruit- unpublished self-management strategies in people with ment, limited to only those with A1C 13. Tyrrell SP, Bui TL, Maggiore, JA. Hemo- type 2 diabetes mellitus. Cochrane Data- globin A1c Detection System: A Technical 7.0%, and allowing randomized con- base Syst Rev 2005;18:CD003417 Bulletin Describing the Performance Char- trol subjects to participate in the program 5. Jackson CL, Bolen S, Brancati FL, Batts- acteristics of Capillary Blood Collected on after a 6-month trial would prove more Turner ML, Gary TL. A systematic review Filter Paper for Determination of Hemoglo- of interactive computer-assisted technol- definitive. bin A1c. Chicago, BIOSAFE Laboratories, ogy in diabetes care: interactive informa- Bulletin 4018, Rev. 99037, 2006 tion technology in diabetes care. J Gen 14. Jernigan, VBB, Lorig K. The Internet Dia- Acknowledgments — The study was sup- Intern Med. 2006;21:105–10 betes Self-Management Workshop for ported by National Institutes of Health Grant 6. Glasgow RE, Boles SM, McKay HG, Feil American Indians and Alaska Natives. 1R18DK065729 and Robert Wood Johnson EG, Barrera M Jr. The D-Net diabetes self- Health Promotion Practice, In press Foundation Grant 096223. management program: long-term imple- 15. Bandura A. Self-efficacy: toward a unify- K.L. and D.D.L. receive royalties for the mentation, outcomes, and generalization ing theory of behavioral change. Psychol book used by study participants. K.L., D.D.L., results. Prev Med 2003;36:410 –9 Rev 1977;84:191–215 and K.P. potentially receive licensing fees if the 7. Glasgow RE, Nutting PA, King DK, Nelson program is disseminated. P.L.R., V.B.B.J., and CC, Cutter G, Gaglio B, Rahm AK, White- 16. Miedema K. Standardization of HbA1c and M.G. have no conflicts of interest. No other sides H, Amthauer H. A practical random- optimal range of monitoring. Scand J Clin Lab potential conflicts of interest relevant to this ized trial to improve diabetes care. J Gen Invest 2005;65(Suppl. 240):61–72 article were reported. Intern Med 2004;19:1167–74 17. Stewart AL, Ware JE. Measuring Function- 1280 DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 care.diabetesjournals.org Lorig and Associates ing and Well Being: The Medical Outcomes 19. Kroenke K, Spitzer RL. The PHQ-9: a new Lorig KR. Self-reports of health care utili- Study Approach. Durham, NC, Duke Uni- depression diagnostic and severity mea- zation compared to provider records. versity Press, 1992 sure. Psychiatr Ann 2002;35:509 –521 J Clin Epidemiol 2001;54:136 –141 18. Lorig K, Stewart A, Ritter P, Gonza ´ lez V, 20. Hibbard JH, Mahoney ER, Stockard J, Tu- 22. Case S, Jernigan V, Gardner A, Ritter PL, Laurent D, Lynch J. Outcome Measures for sler M. Development and testing of a short Heaney CA, Lorig KR. Content and fre- Health Education and Other Health Care In- form of the patient activation measure. quency of writing on diabetes bulletin terventions. Thousand Oaks CA, Sage Health Serv Res 2005;40:1918 –30 boards: does race make a difference? Publications, 1996 21. Ritter PL, Kaymaz H, Stewart A, Sobel DS, J Med Internet Res 2009;11:e22 care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 6, JUNE 2010 1281

Journal

Diabetes CarePubmed Central

Published: Mar 18, 2010

There are no references for this article.