De ziekte van Crohn en colitis ulcerosa zijn chronische inflammatoire darmziekten (IBD) die gekenmerkt worden door perioden van remissie en opvlammingen. Vanwege het risico op opvlammingen worden tieners met IBD over het algemeen regelmatig gecontroleerd op de polikliniek.  Vaak is de ziekte op het moment van de controle rustig en wordt het advies gegeven om de behandeling ongewijzigd voort te zetten. Telemonitoring zou een goede aanvulling kunnen zijn op de reguliere poliklinische controles. Bij telemonitoring worden markers van ziekteactiviteit (zoals symptoomscores, kwaliteit van leven en markers voor inflammatie in bloed of ontlasting) op afstand (digitaal) gemeten. Bij afwijkende uitslagen volgt een poliklinische of digitale afspraak om de behandeling tijdig bij te sturen. Hierdoor kunnen tieners hun dagelijkse bezigheden zonder onderbreking voortzetten wanneer de ziekte het toelaat, en naar het ziekenhuis komen wanneer dit nodig is. 

In de handreiking Telemonitoring van de Federatie Medische Specialisten, zijn algemene doelstellingen voor telemonitoring geformuleerd, waaronder het beter structureel monitoren van patiënten, het tijdig bijsturen van de behandeling bij (dreigende) verslechtering, het zo stabiel mogelijk houden van de patiënt en het voorkomen of verminderen/verkorten van reguliere (polikliniek) consulten, SEH-bezoeken en opnames (1).

Specifieke aanbevelingen voor telemonitoring bij (tieners met) IBD ontbreken. Het doel van deze systematic review is om te evalueren of telemonitoring non-inferieur is aan standaardzorg bij IBD, met name wat betreft ziekteactiviteit en kwaliteit van leven.

Disease activity

Quality of life

Cost

Patient-reported experience and engagement measures

Face-to-face outpatient visits

Unplanned ER visits, surgeries or hospitalizations

Description of studies

Table 1 contains an overview of characteristics of included studies. All included studies were designed as parallel RCTs, comparing various forms of telemonitoring versus standard care in patients with IBD (e.g. ulcerative colitis [UC] or Crohn’s disease [CD]). The total number of randomized participants ranged between 47 and 909. The proportion of females across the studies ranged between 37% and 64%. Study populations included both children and adults with an age range of 11.2 to 95 years (based on 10 studies reporting age). Three of 11 studies included children only (5-7). One study (8) exclusively included patients with UC, while all the other studies included a combination of UC and CD patients. In two studies (5, 7), all participants were in clinical disease remission at baseline.

Regarding the intervention, telemonitoring was solely web-based in 5 studies (6-10), mobile app-based in two studies (11, 12), phone-based in one study (5) phone-based or app-based in one study (13), and a combination of mobile app-based and web-based in two studies (14-16). Monitoring intervals were heterogenous, including periodic assessment (e.g. every third month) and assessment depending on symptoms. Control interventions mainly consisted of planned outpatient visits for which the time interval was often unreported or planned according to the specialists’ discretion. For most studies, the follow-up period was 12 or 24 months; one study had a 6-month follow-up period (13).

Regarding outcomes, 10 studies provided some measurement of disease activity (flares, treatment escalation, and disease activity or remission measured with various instruments) and 10 measured QoL with various tools. All studies reporting QoL used IBD-specific measures and two studies also used generic health-related measures. It was not possible to perform meta-analyses for outcomes other than QoL since measurement tools were too heterogeneous across studies. More information on the reported study outcomes is presented in Table 1.

Table 1. Study outcomes

*Meta-analysis was possible for this outcome variable.

† Pediatric studies.

‡ No raw data or results available for QoL meta-analysis.

§ Excluded from meta-analysis. Standard deviations were unreported or only p-values were presented.

¶ Excluded from meta-analysis as only change scores from baseline were reported.

aPCDAI: abbreviated Pediatric Crohn Disease Activity Index; CSQ: consultation satisfaction questionnaire; EMR: Electronic medical record; HBI: Harvey-Bradshaw Index for Crohn’s disease; HMARS: Medication Adherence Report Scale; HRQoL: Health-related quality of life; IBD-IMPACT: IBD quality of life questionnaire; IBDQ: Inflammatory Bowel Disease Questionnaire (IBD-specific QoL questionnaire); MCS: Mayo Clinic Score; MMAS: Morisky Medication Adherence Score; NR: Not reported. Outcome has not been incorporated into the study; PCDAI: Pediatric Crohn Disease Activity Index; PUCAI: Pediatric Ulcerative Colitis Activity Index; SSCAI: Simple Clinical Colitis Activity Index; SUS: System usability scale; VAS: Visual Analog Scale

Risk-of-bias assessment

Risk of bias was assessed using the Cochrane risk-of-bias tool for randomized controlled trials (Higgins, 2011). Separate assessments were done for the outcomes disease activity and QoL (Table 3). In general, risk of bias arising from the randomization process was low, except for Östlund (12) that did not have random sequence generation. Six of 11 studies had low risk of bias regarding selective outcome reporting. Overall, blinding of outcome assessors was either absent or unclear. In about half of all studies, missing outcome data was judged to be a major concern. While blinding of patients and providers was lacking in all studies, we did not consider this to be a concern as having knowledge of the assigned intervention is a characteristic of the intervention.

Table 2. Risk of bias judgements for included studies (randomized controlled trials; Cochrane risk of bias tool)

Findings

1. Disease activity

Ten studies (5-11, 13-16) reported data on disease activity (Table 4). Disease activity was assessed by various symptom scores and outcomes varied by data type (numerical, binary, time-to-event, or other). Major endpoints related to disease activity were (cumulative) disease flares (n=4), treatment escalation (n=1), and disease activity or remission measured with various instruments (HBI, PCDAI, Seo-index, SSCAI) (n=6). Pooling of disease activity data was not possible due to the high degree of clinical heterogeneity across studies.

Nine of ten studies, including 3 studies in children, did not show a statistically significant difference in disease activity when comparing telemonitoring to standard care (p>0.05). No consistent direction of effect (either favoring telemonitoring or standard care) was observed in these studies. The remaining study reported statistically significant differences: Elkjaer (10), a study among Irish (n=92) and Danish (n=211) adults, showed that the Irish telemonitoring group had a lower mean number of flares during the 12-month follow-up period than the control group (0.6 [range 0 to 4] vs. 0.2 [range 0 to 1], p=0.02). In the same study, among Danish participants, there was no evidence of a difference between the groups.

2. Quality of life (QoL)

Four out of ten studies investigating QoL reported statistically significant improvements in QoL in the telemonitoring group compared to standard care (5, 10, 11, 14). Among these four, Akobeng (5) was the only study that reported a significant effect of telemonitoring in children (pediatric IBD-IMPACT score; mean difference [MD] 8.66 [95% CI 1.05 to 16.27], p=0.03). The remaining studies described inconsistent non-significant results both in favor of telemonitoring and standard care (p>0.05), see Table 4.

Five studies (5, 8, 9, 11, 15, 16) reporting IBD-specific QoL were eligible for meta-analysis (Figure 1). Of these five, only Akobeng (5) included children. The pooled standardized mean difference was 0.14 (95% CI -0.04 to 0.31; p=0.13) in favor of telemonitoring. Cross (9) reported two telemonitoring groups: monitoring weekly or every other week. We randomly selected the latter group for the meta-analysis. A sensitivity analysis with the former group showed no difference in results (SMD 0.12 [95% CI -0.07 to 0.31], p=0.20).

3. Cost (any kind)

Four studies (5, 7, 10, 15, 16) reported data on costs (Table 4).

Akobeng (5) was the only study that reported a statistically significant cost reduction in the telemonitoring group. Mean consultation costs in the intervention group and the control group were 35.4 and 51.1 pounds respectively (MD -15.7; 95% CI -11.8 to -19.6; p<0.001).

De Jong (15, 16) conducted a cost-effectiveness analysis. Mean total (in)direct healthcare costs were lower in the intervention group (8934 euros) than in the control group (9481 euros), with a mean difference of -547 euros (95% CI -2108 to 1014).

Elkjaer (10) reported service and medication costs in Danish participants. The intervention group showed lower costs due to outpatient visits than the control group (12,880 versus 34,454 euros). Hospitalization costs were equal in both groups (5202 euros; corresponding to 2 hospitalizations) while the intervention group had higher remote consultation (2247 versus 504 euros) and medication costs (641 versus 578 euros). Overall, excluding medication costs, the financial benefit was 19,831 euros for 105 Danish telemonitoring participants, equivalent to 189 euros/patient/year. No costs were reported for Irish participants.

In Heida (7) a cost-effectiveness analysis showed that the annual cost-savings of telemonitoring, compared to standard care, was 89 euros per person. In participants compliant to the protocol, these savings were 360 euros per person.

4. Patient-reported experience and engagement measures

Nine studies (5-8, 10, 12-16) reported data on patient-reported engagement measures: patient satisfaction/perception of quality of care and adherence.

Seven studies reported results on patient satisfaction/perception of quality of care. There was no difference in patient satisfaction or perception of quality of care between groups in two studies (5, 15, 16), with a mean difference of -0.11 (CI -0.52 to 0.31, p=0.61) reported by De Jong (15, 16). Heida (7) reported that 71% of intervention group participants wished to continue with home telemonitoring care and that 96% considered home telemonitoring as timesaving. In Elkjaer (10), 89% of Danish and 88% of Irish participants were willing to continue using the new telemonitoring system. Bonnaud (14) reported that 83% of participants was satisfied or even very satisfied with the intervention, with 75% of participants feeling more empowered/independent. Östlund (12) reported that 17/24 patients were willing to participate in a phone survey. Of these 17 participants, 15 (88%) had a positive experience with the telemonitoring intervention. In Del Hoyo (13), the patient satisfaction with received care was measured with the Adapted Client Satisfaction Questionnaire. The mean scores were comparable for the telephone, telemonitoring, and control groups (53, 57, and 55 respectively).

Five studies reported data on adherence, of which 4 reported on medication adherence. Two studies (6, 8) did not report any difference between groups regarding medication adherence at the end of the study period (24 and 12 months, respectively). Cross (8) reported 44% adherent participants in the intervention group and 68% in the standard care group (p=0.10) at 12-months, see Table 4. In the study of Elkjaer (10), adherence to 4 weeks of acute treatment with 5-ASA in case of a flare was significantly greater at 12 months in the telemonitoring versus the control group in both Danish and Irish participants (Denmark: 73% versus 42%, p=0.005; Ireland 73% versus 29%, p=0.03). Del Hoyo (13) reported medication adherence in the telephone, telemonitoring and control groups was, respectively, 71%, 86%, and 81% at 24 weeks. One study (7) reported results on adherence to telemonitoring: 57% of participants in the telemonitoring group were compliant to the study protocol (defined as responding to ³80% of automated alerts), whereas in the standard care group 84% was compliant (sending in ≥2 stool samples for calprotectin measurement).

5. Face-to-face outpatient contacts

Eight studies (5, 6, 10-16) reported on face-to-face outpatient contacts. This outcome was reported heterogenously: as count data (N=3), medians (N=3), means (N=3) or as binary data (N=1). Moreover, the type of visit (gastroenterologist, nurse, or a combination) differed between studies. We therefore did not conduct meta-analyses for this outcome.

Three studies reported statistically non-significant results when comparing telemonitoring to standard care (5, 12, 14). In Akobeng (5), median consultations per patient was similar in the intervention group (4 [IQR 3 to 4]) compared to the control group (3 [IQR 2 to 4]). The same study reported a non-significant difference in the proportion of attended outpatient consultations between the intervention and the control group (67% versus 71%, p=0.71). In Bonnaud (14), the mean number of gastroenterologist consultations in the intervention group were fewer than the control group (2.2 (SD 2.2) versus 4.1 (SD 2.9), p=0.13), but the difference was not statistically significant. Östlund (12) showed that median healthcare visits (including outpatient visits, ER visits, and endoscopy visits) due to IBD were similar between intervention and control groups (1 [IQR 0 to 2] versus 0 [IQR 0 to 2], p=0.076).

Carlsen (6) reported the number of planned and on-demand outpatient visits. The intervention group had fewer planned (38 versus 146, p<0.001) but slightly more on-demand outpatient visits (47 versus 39, p=0.68). Overall, the intervention group had significantly less outpatient visits compared to the control group (85 versus 185, p<0.0001).  

Elkjaer (10) reported less routine hospital visits in the intervention group compared to the control group in both Danish (35 versus 92, p<0.0001) and Irish participants (62 versus 91, p=0.007). In the study of Del Hoyo (13), the total number outpatient visits to the nurse or gastroenterologist were lower in both the telemonitoring (72 visits) and the telephone intervention group (85 visits) than in the control group (131 visits)

Two studies reported significantly fewer gastroenterologist contacts in favor of telemonitoring. De Jong (15, 16) and McCombie (11) reported a mean difference of -0.72 (95% CI -0.87 to 0.57, p<0.0001) and -1.1 (95% CI -1.43 to -0.77, p<0001) contacts, respectively. De Jong (15, 16) also reported results on nurse outpatient visits and overall outpatient visits. Although nurse outpatient visits did not differ significantly between groups (MD -0.07, 96% CI 0.17 to 0.03, p=0.173), overall outpatient visits were fewer in the intervention group (MD -0.79, 95% CI -0.99 to -0.59, p<0.0001).

6. Unplanned ER visits, surgery, and hospitalizations

Eight studies (6, 9-16) reported data on unplanned ER visits, surgery, or hospitalizations. We chose not to perform a meta-analysis due to heterogenous reporting of this outcome. Of all results, only two results from two studies (10, 15, 16) showed a statistically significant difference between comparison groups.

The results of Östlund (12) were described under the previous section ‘face-to-face outpatient contacts’ since outpatient visits, ER visits, and endoscopy visits were all categorized as ‘healthcare visits’.

Four studies reported data on ER visits. Three of them (9, 14-16) did not show evidence of effect when comparing telemonitoring to standard care (p>0.05). No consistent direction of effect (either favoring telemonitoring or standard care) was observed in these studies. The third study reported no differences between the telemonitoring, telephone and standard care groups regarding ER visits (0, 2, and 1 visits, respectively) (13).

Elkjaer (10) reported less acute visits in the telemonitoring group than the control group in Danish participants (21 versus 107 visits, p<0.0001). No difference was noted in Irish participants (9 versus 9 visits). No effect was found in Carlsen (6) regarding acute outpatient visits or hospitalizations (p=0.13).

Three studies reported data on hospitalization in general. Bonnaud (14) and Del Hoyo (13) showed no evidence of effect on hospitalizations when comparing telemonitoring to standard care (p>0.05). In De Jong (15, 16), those in the telemonitoring group experienced less hospitalizations (mean difference -0.05 [95% CI -0.10 to 0.00, p=0.04]). McCombie (11) reported data regarding IBD-related hospitalization and nights spent in the hospital. In both cases there was no evidence of effect in favor of either telemonitoring or standard care (p>0.05).

Data on any surgery was reported in McCombie (11). They did not find evidence of a difference between telemonitoring and standard care (p>0.05). Data on IBD-related surgery were reported in three studies (9, 13, 15, 16). All showed no evidence of effect (p>0.05), with two studies (13, 15, 16) reporting identical number of surgeries between groups.

Level of evidence of the literature

1. Disease activity

The level of evidence regarding this outcome was downgraded by 2 levels for risk of bias (lack of blinding of outcome assessors and missing outcome data) and indirectness (majority of studies were non-pediatric).

2. Quality of life

The level of evidence regarding this outcome was downgraded by 2 levels for risk of bias (lack of blinding of outcome assessors and missing outcome data) and indirectness (majority of studies were non-pediatric).

3. Cost

The level of evidence regarding this outcome was downgraded by 1 level for indirectness (half of studies were non-pediatric). Assessment of risk of bias was not performed for this outcome.

4. Patient-reported experience and engagement measures (patient satisfaction and adherence)

The level of evidence regarding the outcome patient satisfaction was downgraded by 1 level for indirectness. For the outcome adherence, we additionally downgraded by 1 level for inconsistency (studies found lower, higher, or no difference in adherence rates between groups). Assessment of risk of bias was not performed for this outcome.

5. Face-to-face outpatient visits

The level of evidence regarding this outcome was downgraded by 1 level for indirectness (majority of studies were non-pediatric). Assessment of risk of bias was not performed for this outcome.

6. Unplanned ER visits, surgeries or hospitalizations

The level of evidence regarding this outcome was downgraded by 1 level for indirectness (majority of studies were non-pediatric). Assessment of risk of bias was not performed for this outcome.

A systematic review of the literature was performed to answer the following question:

Relevant outcome measures

The guideline development group considered disease activity and quality of life as critical outcome measures for decision making; and costs, ‘patient-reported experience and engagement measures’, face-to-face outpatient contacts, and ‘unplanned ER visits, surgery, and hospitalizations’ as important outcome measures for decision making.

A priori, the working group did not define the outcome measures listed above but used the definitions used in the studies.

The working group defined non-inferiority of telemonitoring compared with standard care alone to be clinically important. The working group did not define thresholds for non-inferiority of telemonitoring (compared with standard care) in advance, as various outcome measurement instruments and tools exist and have been used in studies.

Search and select (Methods)

The databases Medline (via OVID) and Embase (via Embase.com) were searched with relevant search terms until 21-07-2022. The detailed search strategy is depicted under the tab Methods. The literature search resulted in 403 hits.

We included randomized controlled trials (RCTs) based on the following criteria: children and adolescents (£ 18 years old) with IBD (though studies on adults were also included as indirect evidence), receiving any kind of telemonitoring (e.g. web-based, mobile-app based, and phone-based). Trials were included if telemonitoring was compared to standard care and if one or more of the following outcomes were reported: disease activity, QoL, costs, face-to-face clinician-patient contacts, and unplanned emergency room attendance or surgery or hospitalizations.

Studies were initially selected based on title and abstract screening, and 354 articles were excluded in this stage. After assessment of the full text report, an additional 37 articles were excluded, and 11 unique studies in 12 articles were included, see PRISMA-flow diagram in figure 1.

Meta-analysis could only be performed for the outcome QoL. We used a random effects model of standardized mean differences (SMD) due to heterogeneous measurement instruments. The analysis was performed using Review Manager 5.4 (2). If only medians (with interquartile ranges, minimum and maximum values) were available, we transformed them to means and standard deviations using formulas by Wan (3) for meta-analysis.

Results

11 studies in 12 articles were included. Important study characteristics and results are summarized in the evidence tables. The assessment of the risk of bias is summarized in the risk of bias tables.