Physiological effects of intraperitoneal versus subcutaneous insulin infusion in patients with diabetes mellitus type 1: A systematic review and meta-analysis

The intraperitoneal route of administration accounts for less than 1% of insulin treatment regimes in patients with diabetes mellitus type 1 (DM1). Despite being used for decades, a systematic review of various physiological effects of this route of insulin administration is lacking. Thus, the aim of this systematic review was to identify the physiological effects of continuous intraperitoneal insulin infusion (CIPII) compared to those of continuous subcutaneous insulin infusion (CSII) in patients with DM1. Four databases (EMBASE, PubMed, Scopus and CENTRAL) were searched beginning from the inception date of each database to 10th of July 2020, using search terms related to intraperitoneal and subcutaneous insulin administration. Only studies comparing CIPII treatment (≥ 1 month) with CSII treatment were included. Primary outcomes were long-term glycaemic control (after ≥ 3 months of CIPII inferred from glycated haemoglobin (HbA1c) levels) and short-term (≥ 1 day for each intervention) measurements of insulin dynamics in the systematic circulation. Secondary outcomes included all reported parameters other than the primary outcomes. The search identified a total of 2242 records; 39 reports from 32 studies met the eligibility criteria. This meta-analysis focused on the most relevant clinical end points; the mean difference (MD) in HbA1c levels during CIPII was significantly lower than during CSII (MD = -6.7 mmol/mol, [95% CI: -10.3 –-3.1]; in percentage: MD = -0.61%, [95% CI: -0.94 –- 0.28], p = 0.0002), whereas fasting blood glucose levels were similar (MD = 0.20 mmol/L, [95% CI: -0.34–0.74], p = 0.47; in mg/dL: MD = 3.6 mg/dL, [95% CI: -6.1–13.3], p = 0.47). The frequencies of severe hypo- and hyper-glycaemia were reduced. The fasting insulin levels were significantly lower during CIPII than during CSII (MD = 16.70 pmol/L, [95% CI: -23.62 –-9.77], p < 0.0001). Compared to CSII treatment, CIPII treatment improved overall glucose control and reduced fasting insulin levels in patients with DM1.

6 and 11 and 14 and 17 and 28 5 Changes in the systematic review compared to the Protocol During the data evaluation, we decided to restrict the results to a comparison of the effects of continuous subcutaneous insulin infusion (CSII) and continuous intraperitoneal insulin infusion (CIPII) only, as the pharmacokinetics (and possibly the pharmacodynamics) of multiple daily injections (MDI) differ between the two routes of administration. In general, we observed improved glycaemic control when continuous insulin delivery systems (either intravenous, subcutaneous, or intraperitoneal) were compared to MDI of insulin [1][2][3][4] and we concluded that reporting a comparison between CIPII and MDI or mixed MDI/CSII treatment would introduce unnecessary bias. The inability to compare MDI and CSII is also reflected by the differences in pharmacokinetics of the various insulin regimes used with MDI (short-, medium-, or long-lasting) versus the exclusive use of continuous short-lasting insulin infusions during CSII. Therefore, bias could be introduced based on differences in the daily profile of insulin delivery or the type of insulin used, and not just the route of administration per se. Furthermore, studies with missing or insufficient information pertaining to the methods of insulin delivery were also excluded.
In the Protocol, one of the outcomes was identified as ʻDifferent locations of IP and SC delivered insulinʼ. After the data extraction, however, we observed that in some included studies [5,6], patients had been given the choice about where the intraperitoneal (IP) catheter was inserted; in addition, the location could also be changed during the study (e.g., after the replacement of an implanted pump). For instance, in one study, the pumps were placed on the left side of the abdomen in the IP space because all the participants were righthanded [6]. Therefore, the main outcome described as ʻInsulin absorption and parameters that can affect it: Different location of IP and subcutaneous (SC) delivered insulin; Different types of insulin used in the same locationʼ could not be evaluated.
Regarding the case-control studies, we revised the inclusion criteria, from "we need at least one before CIPIIperiod and one after CIPII-period measurement point", to ʻthe study is included if measurements from CSII and CIPII patients/periods are reported separatelyʼ.
During the data collection, we demoted some of the primary outcomes (Stated in the Protocol) to secondary outcomes. Consequently, we made a decision based on the clinical relevance of the results. The original primary and secondary outcomes were described as follows:

Primary outcomes
The main outcomes in the included studies were: (1) Glycaemic control (glycated haemoglobin A1c (HbA1c) levels, self-monitoring of blood glucose (SMBG), fasting blood glucose (BG) and mean BG levels, hypoglycaemic and hyperglycaemic events, time spent in normoglycaemia, and glucose variability), (2) Insulin 6 levels (fasting insulin level, time until maximum insulin level, maximum insulin level, and elevation of insulin level after administration of a pre-meal insulin bolus), (3) Mean daily insulin requirement.

Excluded articles and reasons for exclusion
The search strategy identified 1,517 records. After the removal of duplicates and irrelevant articles, 108 potentially eligible articles remained for consideration (Fig 1).
After full-text and manual reference screening of potential articles and the evaluation of the quality of evidence, 105 articles were included. After additional searches, four more articles were considered for inclusion. After the introduction of additional exclusion criteria (See section above titled: ʻChanges in the Systematic review compared to the Protocolʼ), 70 of the 109 articles were excluded for the following reasons:  Forty-one articles did not report CSII and MDI patients/periods separately ;  two articles reported on only MDI and CIPII, but not CSII [48,49];  four technical reports lacked information on physiological effects [50][51][52][53][54];  two reports were review articles [55,56];  three articles compared intravenous (IV) versus IP insulin administration [57][58][59];  two articles exhibited biased reporting of the distribution of patients per group [60,61];  one article did not provide information about the distribution of patients per groups [62];  five articles were missing information about pre-implantation SC insulin infusion/injection [63][64][65][66][67];  one article was an epidemiological study [68];  two articles assessed patients with a mixture of diabetes mellitus type 1 (DM1) and diabetes mellitus type 2 (DM2) [69,70];  two articles did not provide any relevant information [71,72]; 7  one article assessed patients treated with IP insulin injections (IPII) delivered as separate boluses, not as a continuous infusion as was used for CIPII [73];  two articles assessed a CIPII treatment period lasting less than one month [74,75];  one article investigated an SC peritoneal access device (SPAD). SPAD allows for absorption of insulin at the tissue close to the peritoneal lining, not from the inside of the peritoneal cavity [76];  one article did not mention the length of the CSII and CIPII-periods [77].
In In total, 32 studies from 39 articles were included in the systematic review.

Risk of biases
Some studies [79][80][81] included participants who received MDI therapy, however, the data were also separately available for the CSII and CIPII treatment groups.
One study that provided data for the CSII-period vs. the CIPII-period used a programmable implantable medication system (PIMS). Afterwards, the PIMS was changed to the MiniMed Implantable Pump (MIP).
Because two different CIPII pumps were used, the data from the period in which patients were treated with a PIMS insulin pump were compared with the data from the CSII-period. Data pertaining to the complications experienced during the CIPII-period were extracted from both the PIMS and MIP periods [6]. One study included two different experiments with overlapping patient groups; however, data from the study's second experiment fulfilled our inclusion criteria, and the data for the CIPII and CSII treated patients were extracted [82].
One study did not report essential unit information regarding the daily insulin expenditure [83]. However, we assumed that the insulin expenditure in Table 2 was reported as U/24 hours.
One study did not provide unit information for the mean amplitude of glycaemic excursion (MAGE) [84]. To try to obtain the missing information, we used the reference for the MAGE from the article provided by the authors [85], where, the reported unit was listed as ʻmg/100 mLʼ. 8 One study did not state whether the error of the reported data was listed as the SD or the standard error (SE) [86]. Another study did not describe the statistical analysis method [87]. A third study did not state the mean values of the patients' HbA1c levels [5]. Consequently, these studies were excluded from the HbA1c metaanalyses.
In one study, the units for BG were defined differently in Table 2 (mg/mL) and in the main text (mg/dL); we assumed the correct units to be mg/dL, and those values were used in the analysis. The percentage of blood glucose levels that were high, low or in the normal range were not available due to missing information about the definition of the normal range in that study [88].
Two independent studies provided very similar base line data, with similar methodological description and with identical study periods. However, the authors did not state whether the data in these reports were derived from the same study, from two separate studies, or whether they contained partially overlapping patient populations [89,90]. E-mails, sent to the authors by IDF to verify the uniqueness of these two studies were not answered.
Another two studies provided similar base line data, with the same year of publication [91,92]. Those two studies had identical male: female sex ratios, and age ranges (Table 1); however, they differed in the lengths of the follow-up periods, and the baseline HbA1c levels. Therefore, we assumed that the follow-up periods in these two reports were from different time periods, although we cannot discount the possibility of an overlap in the follow-up for these two studies. One of these articles [91] reported HbA1c levels (Fig 2) in the addition to the insulin expenditure, the anti-insulin antibody levels, and complications that occurred during the CIPIIperiod (Table S2.6). From the other article [92] the data were derived from a figure showing changes in insulin levels, and it was not possible to determine the SD. Therefore, these data were not included in the metaanalysis.
In one study, the data reported in the text were given as the geometric mean values, whereas we used the estimated mean value (Table 2) [93].
One study was a multinational, open, randomised, controlled, crossover study [5]. Due to a high dropout rate (15 out of 30 patients in the CIPII group and 9 out of 30 in the CSII group), the results were analysed as a randomised follow-up study between two parallel treatment groups (i.e., before the crossover).
One study did not provide a definition of severe hypoglycaemia. During the extended periods of the study's reporting (including conference posters presentations for data at 3, 6, 12, 24 months), the number of severe hypoglycaemic events reportedly increased during the CSII-period [94][95][96][97]. 9 Results of the search The primary search strategy identified 1,517 reports, and 21 more were added after screening of the reference lists. After abstract screening, 105 potentially eligible reports remained (Fig 1). After additional searches, four more articles were considered for inclusion in the analysis.
When applying the additional exclusion criteria (which are described above in the "Changes in the Systematic review compared to the Protocol), 70 of the 109 reports were excluded; these are described in the ʻExcluded reports and reasons for exclusionʼ section above.
In total, 38 reports from 32 studies, including one report in Italian [98] and one in German [99], were included (Fig 1).

Data extraction and quality assessment
There was considerable heterogeneity among the studies (Tables S2.1 - Twenty-four studies originated from single European countries (Table 1), four originated from a French multicentre study (EVADIAC: EVAluation dans le Diabète des Implants ACtifs Group) [86,88,100,101], three studies were from the USA [6,83,102], and one was a multinational study [5] (Table 1).
All results of these studies are summarised in Tables S2.1 -S2. 13.

Qualitative data analysis
Primary outcome: Glycaemic control In addition to including patients who were already being treated with CSII, one randomised [5] and six nonrandomised studies [6,84,88,91,103,104] provided participants with an additional CSII follow-up before transitioning them to the CIPII treatment. In three of these studies, the HbA1c levels decreased during this additional CSII follow-up period [5,103,104].

Randomised follow-up studies
One prospective, randomised, follow-up study (for details see the section titled, ʻRisk of biasesʼ) observed equivalent reduction in HbA1c levels in the two treatment groups (CIPII: -0.5 %; CSII: -0.6 %, p = 0.374) and no difference in SMBG values during the twelve months of CIPII treatment and the six months of CSII treatment [5].

Self-monitored blood glucose
Three studies that reported on SMBG concentrations showed a decrease in BG levels from 7.8 -10.5 mmol/L to 7.4 -8.0 mmol/L (p < 0.05) [83,88,96,102], whereas four studies reported no difference in SMBG levels Table S2.1) [6,84,86,108]. However, in one of these studies, SMBG levels decreased during the first 16 months of CIPII treatment, but was equal to those following CSII after 18 months [6]. Three studies did not conduct statistical testing to compare the two treatments [103,104,109].
One study reported decreased pre-prandial BG levels (p < 0.05) [88], whereas another observed decreased post-prandial BG levels (p < 0.01) [87]. Two studies reported no difference in pre-prandial BG levels [86,88] 11 and two studies reported no difference in post-prandial BG levels during the CIPII-period [86,88]. One study did not conduct statistical comparison of the two treatments [103].

Case-control studies
Among the four included case-control studies that reported HbA1c levels, no difference was observed between the treatment groups (Fig 2) [82,88,99,[110][111][112]. One of these studies also reported no difference in pre-prandial and post-prandial BG levels [82].

Case studies
Only one case study was included, which reported no difference in glycaemic control between the CIPII and CSII treatments (Table S2.1) [113]. Due to large SD values, these results could not be included in the metaanalysis.

Randomised follow-up studies
In one study, the frequency of severe hypoglycaemia (requiring hospitalization or IV glucose administration, or events accompanied by unconsciousness or seizure) was significantly reduced during the CIPII compared to the CSII follow-up periods (0.35 vs. 0.86 events/patient-years, p = 0.013). During the first three months after the initiation of CIPII treatment, the frequency of severe hypoglycaemic events was unchanged, whereas it was reduced in the subsequent nine months (0.72 vs. 0.15 events/patient-years). During CSII treatment the frequency of severe hypoglycaemia was 1.6 events per one patient-year at baseline which was reduced to 0.86 events per one patient-years during the CSII follow-up period [5]. No difference in the frequency of hypoglycaemic episodes (SMBG level < 3 mmol/L) was observed during the CIPII treatment period.
Furthermore, no difference was observed between the first three months and the subsequent nine months of CIPII treatment (Tables S2.1 and S2.8) [5]. Statistical analyses were only reported for comparison between the CIPII and CSII treatment groups; no within-group analyses were performed.

Severe hypoglycaemia and hypoglycaemic coma
Four studies recorded severe hypoglycaemia, but none conducted any statistical analyses [6,81,[94][95][96][97][98]. One study reported no difference in the frequency of hypoglycaemic coma events (CIPII: 0 vs. CSII: 0.54 events/patient-year) [81]. Another study reported that the frequency of severe hypoglycaemia (requiring assistance) was 0.43 events per one patient-year during the CIPII-period while no episodes of hypoglycaemic coma were observed [6]. 12 One study reported 1.5 severe hypoglycaemic (requiring assistance) events per one patient-year during the CIPII compared to the 12 events per one patient-year during CSII-period [94][95][96][97]. Another study reported no severe hypoglycaemic (requiring assistance) events during the CIPII-period [81], and one study reported no difference in the occurrence of severe hypoglycaemia [98].

Hypoglycaemia
One study reported a reduction in the time spent in hypoglycaemia during CIPII-period (SMBG level < 3.9 mmol/L, p < 0.05), whereas the duration of time spent with SMBG levels < 2.8 mmol/L was similar between the treatment periods [84]. On the contrary, one 24-hour BG profile study reported no difference in the time spent in hypoglycaemia (BG < 3.8 mmol/L, measured by CGM) [78]. Similarly, two other studies reported no difference in hypoglycaemic events (SMBG level < 3.0 mmol/L) [89,90].

Hyperglycaemia
One study using CGM [78] reported less time spent in hyperglycaemia (BG > 10 mmol/L, p < 0.05), whereas another study using SMBG reported no difference [84]. However, both reported a reduction in the time spent in severe hyperglycaemia (BG > 14 mmol/L, p < 0.05, measured by SMBG and CGM) during CIPII-period.

Randomised crossover and follow-up studies
In one study, five patients being treated during the CIPII-period were crossed over to receive 96-hour CSII treatment temporarily. Insulin was infused for 12 hours at a fixed basal rate. Fasting serum free insulin levels were decreased during the CIPII-period compared to the CSII-period (30.8 vs. 45.0 pmol/L, p < 0.001) [100].
Subsequently, insulin was infused a rate of 15 nmol/h for 150 minutes, then 42 nmol/h for the following 150 minutes. During these two short-term periods with increased infusion rates, the rate of appearance (Ra) of insulin in the systemic circulation was greater during CIPII treatment (p < 0.05 and p < 0.01, respectively) [100].
No difference in the mean daily insulin requirement was observed in a prospective study with 36 patients, although no statistical analyses were performed [5].
Furthermore, during the CIPII-period, insulin levels returned to baseline values three hours after administration of a pre-breakfast bolus, whereas during the CSII-period, the post-bolus insulin level remained elevated five-and-half hours later [87].
One study that performed insulin clamp testing reported no difference in the maximum insulin levels between the periods; however, the first measurement was recorded 30 minutes after the administration of insulin boluses [89]. One study reported increased insulin levels (p < 0.05) during exercise in those receiving CSII, although, insulin levels did not change during exercise in the CIPII group [90].
One study reported a lower total area under curve (AUC) (16 hours) (72 vs. 100 mU/L/h, p < 0.01) and a lower night-time AUC (12 vs 36 mU/L/h, p < 0.01) during the CIPII period. The AUC following administration of an insulin bolus did not differ between the periods; however, the duration of the period for which the AUC was calculated was not specified [87].
However, in one of these studies, the insulin requirement was increased only during the first two months of CIPII treatment before decreasing to levels that were similar to those in the previous CSII-period [108].
Other studies reported no change in insulin requirements between the periods, 12 of which performed statistical analyses [83, 84, [78].

Case-control studies
One study reported decreased mean night-time insulin levels in the CIPII-treated patients (65.56 vs. 86.53 pmol/L, p < 0.005) [99], whereas two studies reported no difference in fasting insulin levels between the two groups [82,114].

Case reports
One case report showed no difference in daily insulin requirements [113].
Secondary outcomes: Intermediate metabolites All reports that analysed intermediate metabolites are summarised in Table S2.3.

Non-randomised crossover studies
One study reported decreased total cholesterol levels after six months of the CIPII-period compared to those in the CSII-period (4.56 mmol/L vs. 4.85 mmol/L, p = 0.044) [102]. In the remaining six studies, no differences in total cholesterol levels were observed after six weeks to one year of CIPII treatment ( Fig S2)  One study reported an increase in fasting serum triglyceride levels after the CIPII-period (1.5 mmol/L vs. 0.9 mmol/L, p < 0.005) [84]. In six studies, no difference in triglyceride levels was observed between the two periods ( Fig S3) [83,98,102,[106][107][108][109].
The chylomicron remnant levels, the ratio of retinyl ester: apoB lipoproteins, and the HDL compositions reported in the studies are provided in Table S2.3.

Case-control studies
One study reported decreased fasting free fatty acid (FFA) levels during the CIPII-period compared to the CSIIperiod (p = 0.05), whereas during the 60 minutes after the administration of a pre-meal insulin bolus, no changes in FFA levels were observed within the groups. However, decreased FFA levels were observed in the CIPII-period after administration of a pre-meal insulin bolus (p = 0.05) [82].
The measurements of lactate, vitamin D metabolites, creatinine, calcium, magnesium, phosphorus, parathyroid hormone, osteocalcin, and alanine reported in the studies are summarised in Table S2.3. 15 Secondary outcomes: counterregulatory hormones All reported counterregulatory hormone analyses are summarised in Table S2.4.

Non-randomised crossover studies and follow-up studies
During a hypoglycaemic clamp, one study reported a significant incremental glucagon response during CIPII (p = 0.003), whereas the glucagon response was non-significant during CSII. Consequently, the maximal glucagon response was higher during CIPII (17.0 pg/mL vs.7.5 pg/mL, p = 0.048) [89]. One study reported increased glucagon levels post-exercise during CIPII-periods (p = 0.01); however, no difference in glucagon levels was observed between the CIPII and CSII-periods [90]. Significantly larger AUC was observed for the incremental glucagon response in the CIPII-period during hypoglycaemic insulin clamp testing and after intense exercise compared to pre-clamp testing and pre-exercise testing ( [90].
Two studies reported no change in epinephrine and norepinephrine incremental responses between the two periods during respective hypoglycaemic insulin clamp testing [89] or intensive exercise [90].

Case-control studies
One study reported no difference in fasting and postprandial glucagon levels between the treatment groups [82].
Secondary outcome: Other metabolic outcomes All other reported analyses are summarised in Table S2.5.

Non-randomised crossover and follow-up studies
Increased levels of anti-insulin antibodies (AIA) measured by enzyme-linked immunosorbent assay (ELISA), were observed after three and twelve months of the CIPII-period (39.3 % and 42.5 % vs. 23.7 %, respectively, p < 0.01), but not after 24 months [79,80], and at three months of the CIPII-period in another study (11.0 % vs. 3.6 %, p < 0.05) [86]. No difference was observed in one study [91], and another reported no changes in the AIA levels (p-value not reported) [78].
One follow-up study observed increased AIA levels after six months of the CIPII-period vs. six months of the CSII-period (41.8 % vs. 24.9 %, p = 0.009), as measured by radioimmunoassay (RIA), although they observed no difference when AIA levels were measured by ELISA [115]. 16 Studies reporting sex hormone binding globulin (SHBG) levels are summarised in Table S2.5.

Secondary outcome: Complications
All reported technical and physical complications are summarised in Table S2.6.
How to read the tables The source column lists the main author and the year of publication. In cases where the authors and year of publication are the same for two studies, some additional information is provided in differentiation.
Alternatively, when there is no information given in other columns, information is provided that could explain the missing data. For example, if there is no information provided under the ʻReported study objectivesʼ and/or ʻmethodological qualityʼ columns, it could be because information was extracted from a letter to the editor.
The ʻParticipant characteristicsʼ column supplies information about the number of participants and some characteristics we believe are important for describing the actual patients. More detailed information can be found in the original publications.
In the ʻLength ofʼ column, we provide information about the duration of the CIPII and/or CSII-periods, and, if available, some information about patient follow-up. Most data are given as the means.
In the ʻReported study objectivesʼ column we present the precise information as stated in the articles.
We extracted data from text, tables, and graphics, all of which is included it in the ʻOutcomesʼ column. In cases, where information was missing, possible biases are indicated in the systematic review's Results section.
Some articles included figures showing measurements of continuous variables (for example, 16-hour measurements). From such figures, we extracted data from fasting periods and noted data that was significantly different between the two periods. If data for continuous variables measurements were not significantly different, it was mentioned in the Results without providing any additional data.

Definition of words used:
Increases means that in the CIPII-period, levels were statistically significantly higher (p < 0.05) than those in the CSII-period.
Decreases means that in the CIPII-period, levels are statistically significantly lower (p < 0.05) than those in the CSII-period.
Decreases/increases in both means that the values followed the same pattern when compared at different time-points.
No change means a statistically non-significant difference (p > 0.05) or the p-value not provided (ND). If possible, data are shown in parentheses.
The ʻMethodological qualityʼ column contains quality assessment tools that are appropriate for that particular study.  To investigate the hormonal and metabolic patterns produced by CIPII in group of severely unstable DM1 who has previously responded poorly to CSII. To compare clinical and metabolic effects of CSII and CIPII.      To investigate the clinical long-term performance and safety of the new Accu-Chek DiaPort system.      To compare the reproducibility of the plasma-insulin profile of IP and SC administered insulin in a group of Cpeptide-negative, diabetic patients.   To investigate the hormonal and metabolic patterns produced by CIPII in group of severely unstable DM1 who has previously responded poorly to CSII. To compare clinical and metabolic effects of CSII and CIPII.    To evaluate the safety, the efficacy and the results after 3 years of CIPII.  Compare if replacement of SCII with IPII restores the normal physiological gradient between the portal vein and peripheral circulation, which is likely to modify lipoprotein metabolism.   Compare the effects of intensive SC vs. implantable pump IP insulin delivery on intermediary metabolites in DM1 patients.  To test the hypothesis that among persons with T1DM treated with IP insulin therapy there is a decreased calcification propensity (expressed as a higher T50) as compared with treatment with SC insulin therapy.               Legends: CSII, Continuous subcutaneous insulin infusion; CIPII, Continuous intraperitoneal insulin infusion; (-), no data; SD, standard deviation; SEM, standard error of means, Asterix (*), 24-hour measurements Figure S1a. Meta-analysis of HbA1c (%) in patients during CIPII treatment compared to that during control treatment (CSII).

Pre-meal insulin bolus (bolus + 4 h basal
Legends: Treatment, continuous intraperitoneal insulin infusion; Control, continuous subcutaneous insulin infusion. Figure S1b. Subgroup meta-analysis of HbA1c (%) according to duration in patients during CIPII treatment compared to that during control treatment (CSII). Figure S1c. Subgroup meta-analysis of HbA1c (%) in patients during CIPII treatment compared to that during control treatment (CSII).
Figure S1e. Meta-regression analysis bubble-plot of HbA1c (%) in patients during CIPII treatment compared to that during control treatment (CSII). Figure S1f. Cumulative meta-analysis of HbA1c (%) in patients during CIPII treatment compared to that during control treatment (CSII) according to duration of CIPII treatment. Figure S2a. Subgroup meta-analysis of fasting blood glucose (mmol/L) in patients during CIPII treatment compared to that during control treatment (CSII).
Legends: Treatment, continuous intraperitoneal insulin infusion (CIPII); Control, continuous subcutaneous insulin infusion (CSII). Figure Figure S2b. Summarised subgroup meta-analysis of fasting blood glucose (mmol/L) in patients during CIPII treatment compared to that during control treatment (CSII). Figure S3a. Subgroup meta-analysis of fasting insulin (pmol/L in patients during CIPII treatment compared to that during control treatment (CSII).
Legends: Treatment, continuous intraperitoneal insulin infusion (CIPII); Control, continuous subcutaneous insulin infusion (CSII). Figure Figure S3b. Summarised subgroup meta-analysis of fasting insulin (pmol/L) in patients during CIPII treatment compared to that during control treatment (CSII). Figure S4a. Subgroup meta-analysis of daily insulin dose (U/24 hours) in patients during CIPII treatment compared to that during control treatment (CSII).
Legends: Treatment, continuous intraperitoneal insulin infusion (CIPII); Control, continuous subcutaneous insulin infusion (CSII). Figure Figure S4b. Summarised subgroup meta-analysis of daily insulin dose (U/24 hours) in patients during CIPII treatment compared to that during control treatment (CSII). Figure S5a. Meta-analysis of SMBG (mmol/L) in patients during CIPII treatment compared to that during control treatment (CSII).
Legends: Treatment, continuous intraperitoneal insulin infusion (CIPII); Control, continuous subcutaneous insulin infusion (CSII). Figure Figure S5c. Summarised subgroup meta-analysis of SMBG (mmol/L) in patients during CIPII treatment compared to that during control treatment (CSII). Figure S6a. Meta-analysis of cholesterol (mmol/L) in patients during CIPII treatment compared to that during control treatment (CSII). Figure S6b. Subgroup meta-analysis of cholesterol (mmol/L) in patients during CIPII treatment compared to that during control treatment (CSII).
Legends: Treatment, continuous intraperitoneal insulin infusion (CIPII); Control, continuous subcutaneous insulin infusion (CSII). Figure Figure S6c. Summarised subgroup meta-analysis of cholesterol (mmol/L) in patients during CIPII treatment compared to that during control treatment (CSII). Figure S7a. Meta-analysis of triglycerides (mmol/L) in patients during CIPII treatment compared to that during control treatment (CSII).