DIABETICMedicine

DOI: 10.1111/dme.12694

Systematic Review or Meta-analysis

Biphasic vs basal bolus insulin regimen in Type 2
diabetes: a systematic review and meta-analysis of
randomized controlled trials

C. Wang, J. Mamza and I. Idris

Division of Medical Sciences & Graduate Entry Medicine, School of Medicine, University of Nottingham, UK

Accepted 12 January 2015

Abstract
Aim We aim to evaluate the effects of biphasic insulin compared with a basal bolus insulin regimen on glycaemic
control, total daily insulin requirements, risk of hypoglycaemia, weight and quality of life in patients with diabetes
mellitus.
Methods MEDLINE, EMBASE, PubMed and Scopus databases were searched for studies up to November 2013.
Interventions that lasted for more than four weeks and were reported in English were considered for the review. Meta-
analysis was performed on eligible studies.
Results Fifteen randomized controlled trial studies, involving 4384 patients, were included. Greater HbA1c reductions
were seen with basal-bolus compared with biphasic insulin regimens, between-treatment weighted mean difference
(WMD) for baseline-to-endpoint changes in HbA1c was –0.2% (95% CI: –0.36 to –0.03) [–2.2 (–3.9, –0.3) mmol/mol].
In non-insulin na€ıve (n = 8) patients with Type 2 diabetes, HbA1c reduction was greater in the basal bolus group;
WMD = –0.22% (95% CI: –0.42 to –0.02) [–2.4 (–4.6, –0.2) mmol/mol], but in insulin na€ıve patients (n = 5), HbA1c
was equivalent between the two regimens; WMD (–0.15% (95% CI: –0.52 to 0.22) [–1.6 (–5.7, 2.4) mmol/mol]. Total
daily insulin requirements and weight were increased with both regimens, whereas hypoglycaemia rates were
comparable between the two regimens. Greater HbA1c reduction was observed in the basal bolus group compared with
the biphasic regimen at the expense of higher daily insulin requirements and weight gain, but with no greater risk of
hypoglycaemia.
Conclusions Biphasic and basal bolus regimens were equally effective in reducing HbA1c in insulin na€ıve patients with
Type 2 diabetes and both regimens are equally effective for initiating insulin in Type 2 diabetes.
Diabet. Med. 32, 585–594 (2015)

diabetes mellitus [4,5]. The basal bolus regimen, which

Introduction
Insulin therapy is essential for patients with Type 1 diabetes.
In patients with Type 2 diabetes, initiating insulin therapy is
recommended when diet and oral hypoglycaemic agents fail
to maintain optimal HbA1c targets due to the progressive
destruction of insulin-producing b-cells and/or increases in
insulin resistance [1]. In both patient groups, intensification
of insulin therapy has been shown to reduce the risks of long-
term vascular complications [2,3].
Different types of insulin and insulin regimens currently
exist, but there is no overall consensus regarding the most
effective or optimal insulin regimen for patients with
Correspondence to: Iskandar Idris. E-mail: iskandar.idris@nottingham.ac.uk
consists of multiple daily injections of rapid-acting insulin
pre-prandially, in addition to a long-acting basal insulin,
most closely mimics the pattern of insulin secretion in
individuals without diabetes [6]. The flexibility of this
regimen is, however, undermined by its complexity in the
need to count daily carbohydrate intake and adjust the
insulin dose accordingly, as well as the lifestyle restrictions
implicated by the high number of injections [7]. The biphasic
insulin regimen accounts for the majority of insulin pre-
scriptions worldwide. It consists of a fixed ratio of rapid-
acting insulin and intermediate insulin combined; thereby
eliminating the need for patients to mix the insulin them-
selves whilst also reducing the number of required daily
injections. Although patients may find it less difficult to

ª 2015 The Authors.

Diabetic Medicine ª 2015 Diabetes UK 585
DIABETICMedicine
adhere to the premixed regimen [8], regular meal times and
consistency in daily routine are sometimes necessary in order
to gain maximum benefit and avoid the associated increased
risk of hypoglycaemia. Although the basal bolus insulin
regimen is widely considered to be the gold standard [4],
choice of insulin regimen is still a matter of clinical or
personal preference in most cases. Although many reviews
have been carried out to compare human with analogue
insulin within the same regimen, few studies have directly
compared the basal bolus regimen with the biphasic regimen
in insulin-treated patients with diabetes mellitus. Further-
more, the optimal insulin regimen at the point of insulin
initiation remains unclear. Clinical practice recommenda-
tions suggest the use of a basal or biphasic insulin regimen
once daily at insulin initiation, on the basis of simplicity,
targeting fasting glucose and the encouragement of earlier
use of insulin in routine clinical practice [9]. Our aim is to
assess the effectiveness of the basal bolus insulin regimen
compared with a twice daily biphasic insulin regimen in
insulin na€ıve and in non-insulin na€ıve patients with diabetes,
on a variety of clinical and quality of life outcomes to
facilitate clinical decision making on the appropriate insulin
regimen for people with diabetes.
Methods
Inclusion criteria
We considered randomized controlled trials (RCT) of paral-
lel or crossover design that compared the biphasic insulin
regimen with the basal bolus insulin regimen in diabetes
patients of any age, sex or ethnicity. Intervention with a
biphasic regimen was defined as two or more injections of
any biphasic insulin per day. The basal bolus regimen was
defined as a basal injection with at least one bolus injection
daily. The intervention testing the regimens had to be at least
four weeks duration.
Search strategy
The electronic databases MEDLINE, EMBASE, Scopus and
PubMed were searched. The last search was conducted in
November 2013. A general search strategy was adapted
depending on the database and consisted of the following
terms: diabetes, insulin, biphasic, basal bolus, randomized
controlled trials. Reference lists from relevant secondary
literature was also hand-searched. Non-English language and
non-human studies were excluded.
Data extraction and study quality
One of the authors, CW examined the titles and abstracts of
articles and reference texts identified from the search. Full
texts of relevant articles were retrieved for further evaluation
if studies appeared to meet the inclusion criteria, and those
586
Biphasic vs basal bolus insulin regimen  C. Wang et al.
that did not were subsequently excluded. Where there were
any uncertainties, the other reviewers were consulted. Data
were extracted from each study using a data extraction form.
The methodological quality of the eligible studies was
assessed by all authors using an adapted version of the
criteria from the National Institute of Health and Care
Excellence (NICE) guidelines manual [10] and the Cochrane
Handbook [11,12]. Assessments of bias were categorized
into high, moderate or low.
The primary outcome was a change in glycaemic control,
which was expressed as the absolute change in HbA1c [as
mmol/mol (%)]. Data on HbA1c were extracted, as was
information on the overall number and number of severe
hypoglycaemic events, insulin requirements, BMI, weight
gain and quality of life.
Synthesis of results
Meta-analysis was performed on the data collected for
HbA1c levels among patients with Type 2 diabetes, using the
random effects model. The weighted mean difference
(WMD) was calculated for the baseline-to-endpoint change
in percentage HbA1c. Subgroup analyses were carried out on
insulin na€ıve and non-insulin na€ıve patients with Type 2
diabetes. For studies that did not report the standard
deviation (SD) for the baseline-to-endpoint change in mean
difference of HbA1c, SD was calculated as:
2 2
SDchange = √((SDstart) + (SDend) – 2 9 r 9 SDstart 9 SDend)
The correlation coefficient (r) between the baseline and
endpoint HbA1c was assumed to be 0.5 [13,14]. The
heterogeneity test, I2 was used to assess the variability in
effects estimates. In order to incorporate crossover design
studies into the meta-analysis we used data from first period
before crossover only, due to the risk of carryover. Analysis
was performed using Stata, version 13.
Results
Identification of relevant studies
Of 1819 citations identified from the search, 13 studies met
the selection criteria for the review. A flow diagram outlining
the different stages of inclusion and exclusion is shown in
Fig. 1.
Study characteristics
All the eligible RCT were open-label studies. One of the
studies [15] employed the crossover design, whereas the
remaining 12 studies were based on parallel trial design.
Overall, 3959 patients with Type 2 diabetes took part. The
study intervention period ranged from 12 to 52 weeks. One
study [16] was an intensification substudy of a previous
DURABALE trial [17]. Therefore, we reported the results
separately as ‘arm A’ and ‘arm B’. The characteristics of the
ª 2015 The Authors.
Diabetic Medicine ª 2015 Diabetes UK
 Systematic Review or Meta-analysis
EMBASE
n = 340
Identified studies
n = 1819
Potentially relevant studies
n = 1197
Potentially relevant studies
n = 29
Articles included in the systematic review
n = 13
•Patients with type 2 diabetes mellitus (n = 13)
FIGURE 1 Flow diagram outlining the systematic literature search and selection of relevant studies to include in the systematic review, where ‘n’
represents the number of publications within each stage.
included studies and their primary outcomes are summarized
in Table 1. In addition, the secondary study outcomes are
outlined in Table 2. The quality assessment found that the
included studies were not methodologically robust in their
study designs.
HbA1c levels
Meta-analysis was performed on the eligible RCTs except on
one study [18] that had incomplete information. Overall,
seven studies [16,19–24] reported a baseline-to-endpoint
mean change in HbA1c, of which three [19,21,22] reported
an adjusted mean change. Adjustments were made for
demographic characteristics, oral glucose-lowering drugs
and baseline HbA1c.
Standard deviations for the change in baseline-to-endpoint
mean difference in HbA1c between the biphasic and basal
bolus regimens were calculated for six of the trials
[15,16,22,25–27] by taking a correlation of 0.5 between
the HbA1c values. Ultimately, 12 trials were included in the
meta-analysis. The effect size between treatment groups was
calculated using the WMD of the HbA1c. This was estimated
to be – 0.2% (95% CI: – 0.36 to – 0.03) [ – 2.2
( – 3.9, – 0.3) mmol/mol] in favour of the basal bolus
regimen compared with the premixed regimen group
(I2 = 65.3%, P = 0.001) (Fig. 2).
In studies with non-insulin na€ıve patients, the WMD of
HbA1c change was found to be – 0.22% (95% CI: – 0.42
to – 0.02) [ – 2.4 ( – 4.6, – 0.2) mmol/mol], in favour of
the basal bolus regimen (I2 = 79.3%, P < 0.001) (Fig. 3). In
ª 2015 The Authors.
Diabetic Medicine ª 2015 Diabetes UK
DIABETICMedicine
MEDLINE PUBMED SCOPUS
n = 276 n = 241 n = 962
Removal of duplicated references
n = 622
Removal of references on the basis of
title and abstract
n = 1168
References excluded:
n = 16
•Not RCT (n = 4)
•Intervention less than 4 wks (n = 1)
•Not premixed vs. basal bolus (n = 8)
•Article written in Japanese (n = 1)
•Type 1 diabetes mellitus (n = 1)
•Type 1 and Type 2 diabetes patients
combined (n = 1)
patients na€ıve to insulin, the WMD of HbA1c change was
estimated as – 0.15% (95% CI: – 0.52 to 0.22) [ – 1.6 (5.7,
2.4) mmol/mol], equivalent between the two regimens
(Fig. 3) with no significant heterogeneity was evident
(P = 0.946) (Fig. 4).
Total daily insulin requirements
Seven studies [15,16,18,20,24,26,28] showed that patients
were administered higher doses of basal bolus insulin
(ranging from 38 to 146 IU or 0.5 to 1.2 IU/kg) compared
with the biphasic insulin regimen (24–123 IU or 0.4–1.0 IU/
kg). (Table 2). However, only results from three of the
studies [20,24,28] showed that the dose administered were
significantly different between the two interventions.
Hypoglycaemia
Seven studies [18–23,28] reported rates of overall hypo-
glycaemia. However, no significant difference in hypogly-
caemic incidence was found between the basal bolus and the
biphasic treatment groups. A small number of severe
hypoglycaemia events occurred from insulin treatment, as
was observed. Between 0 and 30 episodes were reported in
10 studies (Table 2). Cumulatively, the average number of
severe hypoglycaemia events in the overall population was
estimated at five and six episodes occurring from the use of
basal bolus and biphasic insulin regimen, respectively.
However, there was no significant difference in these events
between the two intervention groups.
587
588
Table 1 Main characteristics of the trials included in the systematic review and their primary outcome at the end of treatment
DIABETICMedicine

Diabetes duration
Patients (years) Regimen type HbA1c levels, mmol/mol (%)

Mean age (years)

Basal bolus Premixed
Studies n Basal bolus Premixed Basal bolus Premixed Short-acting/long-acting insulin Brand name At baseline At endpoint

Bowering et al. 2012 [19] 426 56.3 56.7 10.0 10.6 Lispro/Glargine Humalog Mix 75/25 75 (9.0) vs 74 56 (7.3) vs 54 (7.1)
(8.9)
Fritsche et al. 2010 [20] 310 60.2 60.9 12.8 12.5 Glulisine/Glargine Novolin 30/70 70 (8.6) vs 69 56 (7.3) vs 61 (7.7)
(8.5)
Hirao et al. 2008 [25] 160 57.9 58.5 12.2 9.5 Aspart/NPH Novolog Mix 70/30 90 (10.4) vs 88 62 (7.8) vs 60 (7.6)
(10.2)
Jain et al. 2010 [21] 484 59.9 58.9 12 11.4 Lispro/Glargine Humalog Mix 50/50 78 (9.3) vs 80 58 (7.5) vs 60 (7.6)
(9.5)
Kloos et al. 2007 [15] 39 54.7 56.6 8.2 7.5 Human regular insulin/NPH Novolin 30/70 77 (9.2) vs 74 56 (7.3) vs 57 (7.4)
(8.9)
Levin et al. 2011 [22] 197 56.4 55.9 13.1 12.9 Glulisine/Glargine Novolog Mix 70/30 78 (9.3) vs 79 52 (6.9) vs 58 (7.5)
(9.4)
Liebl et al. 2009 [18] 719 60.3 61.7 9.4 8.9 Aspart/Detemir Novolog Mix 70/30 69 (8.5) vs 68 53 (7.0) vs 55 (7.2)
(8.4)
Ligthelm et al. 2006 [23] 394 59.8 59.5 12.8 14.2 Aspart/NPH Novolog Mix 70/30 76 (9.1) vs 76 62 (7.8) vs 62 (7.8)
(9.1)
Masuda et al. 2008 [26] 28 58.5 60 8.1 10.5 Lispro/NPH Humalog Mix 50/50 97 (11.0) vs 98 49 (6.6) vs 52 (6.9)
(11.1)
Miser et al. 2010 (Arm A) [16] 399 58.2 55.9 9.3 8.9 Lispro/Glargine Humalog Mix 75/25 64 (8.0) vs 64 65 (8.1) vs 64 (8.0)
(8.0)
Miser et al. 2010 (Arm B) [16] 345 54.5 55.4 9.6 10 Lispro/Glargine Humalog Mix 50/50 64 (8.0) vs 64 66 (8.2) vs 66 (8.2)
(8.0)
Rosenstock et al. 2008 [28] 374 54.0 55.4 11.2 10.9 Lispro/Glargine Humalog Mix 50/50 74 (8.9) vs 73 51 (6.8) vs 53 (7.0)
(8.8)
Shanmugasundar et al. 2012 [24] 50 53.8 53.9 13.2 14.1 Aspart/Detemir Novolin 70/30 79 (9.4) vs 75 66 (8.2) vs 63 (7.9)
(9.0)
Yazici et al. 2011 [27] 34 57.4 61.5 5.6 11.7 Lispro/Glargine Humalog Mix 50/50 82 (9.7) vs 65 50 (6.7) vs 56 (7.3)
(8.1)

HbA1c reported as basal bolus versus premixed regimen.

n = number of patients in each trial.
Biphasic vs basal bolus insulin regimen  C. Wang et al.

Diabetic Medicine ª 2015 Diabetes UK

ª 2015 The Authors.
ª 2015 The Authors.
Table 2 Trials and secondary treatment outcomes

Diabetic Medicine ª 2015 Diabetes UK

Systematic Review or Meta-analysis

Study characteristics

Secondary outcome measures, mean (P-value)

Body weight/BMI

P-value of
Length of Total insulin dose Severe hypoglycaemia difference in
Studies intervention (IU or IU/kg) Overall hypoglycaemia (%) (number of episodes) At baseline At endpoint weight gain

Bowering et al. 2012 [19] 48 weeks 0.71 vs 0.71* (NS) 42.2 vs 48.5 (NS) 6 vs 7 (NS) 72.8 vs 73.8 75.7 vs 76.6 NS
Fritsche et al. 2010 [20] 52 weeks 98 vs 91.3 (P < 0.0001) 75.8 vs 73.9 (P = 0.564) 18 vs 30 (NS) 87.0 vs 84.3 90.6 vs 86.5 0.0073
Hirao et al. 2008 [25] 6 months n.a. n.a. 0 vs 0 (NS) 24.0 vs 23.8† 24.8 vs 25.2† < 0.001
Jain et al. 2010 [21] 36 weeks 0.51 vs 0.57* (P = 0.017) 74.6 vs 74.5 (P = 1.00) 5 vs 8 (P = 0.416) 78.8 vs 78.2 82.0 vs 81.3 0.803
Kloos et al. 2007 [15] 8 weeks 37.8 vs 24 (NS) n.a. 1 vs 0 (NS) 29.3 vs 29.0† 29.7 vs 29.9† < 0.01
Levin et al. 2011 [22] 9 months 78.3 vs 90.1 (P = 0.23) 36 vs 42 (P = 0.37) n.a. 102.6 vs 103.5 108.9 vs 106.6 0.03
Liebl et al. 2009 [18] 26 weeks 0.87 vs 0.63* 31 vs 28 (NS) 11 vs 0 (NS) n.a. n.a. n.a.
Ligthelm et al. 2006 [23] 16 weeks 1.05 vs 1.06* (NS) 43.9 vs 42.6 (NS) 2 vs 7 (NS) n.a. n.a. n.a.
Masuda et al. 2008 [26] 12 weeks 0.45 vs 0.4* (NS) n.a. n.a. 24.7 vs 24.9† 24.9 vs 24.6† NS
Miser et al. 2010 (Arm A) [16] 24 weeks 69.4 vs 71.1 (P = 0.903) n.a. 0 vs 2 (P = 0.166) 89.3 vs 90.1 92.2 vs 91.5 0.098
Miser et al. 2010 (Arm B) [16] 24 weeks 74.4 vs 73.9 (P = 0.566) n.a. 0 vs 2 (P = 0.188) 91.9 vs 93.6 92.8 vs 94.2 0.345
Rosenstock et al. 2008 [28] 24 weeks 146 vs 123 (P = 0.002) 48.7 vs 51.2 (P = 0.736) 6 vs 4 (P = 0.751) 99.8 vs 99.1 104.3 vs 103.1 0.224
Shanmugasundar et al. 2012 [24] 12 weeks 1.18 vs 0.94* (P < 0.001) n.a. n.a. 85.5 vs 74.6 86.9 vs 76.0 NS
Yazici et al. 2011 [27] 3 months n.a. n.a. n.a. n.a. n.a. n.a.

Outcomes reported as basal bolus versus premixed group.

*Values given in IU/kg.
†BMI.
n.a., not available; NS, not significant.

589
DIABETICMedicine
DIABETICMedicine Biphasic vs basal bolus insulin regimen  C. Wang et al.

Study WMD (95% CI) Weight

Bowering et al. 2012 –0.04 (–0.78, 0.70) 3.87

Hirao et al. 2008 0.00 (–0.67, 0.67) 4.56

Jain et al. 2010 –0.17 (–1.18, 0.84) 2.33

Kloos et al. 2007 –0.40 (–1.12, 0.32) 4.05

Masuda et al. 2008 –0.20 (–1.75, 1.35) 1.07

Fritsche et al. 2010 –0.51 (–0.76, –0.26) 12.54

Levin et al. 2011 –0.60 (–1.41, 0.21) 3.38

Ligthelm et al. 2006 –0.05 (–0.14, 0.04) 16.45

Miser et al. (Arm A) 2010 0.10 (–0.17, 0.37) 11.82

Miser et al. (Arm B) 2010 0.00 (–0.25, 0.25) 12.46

Rosenstock et al. 2008 –0.22 (–0.40, –0.04) 14.26

Shanmugasundar et al. 2012 –0.10 (–0.41, 0.21) 10.90

Yazici et al. 2011 –2.20 (–3.22, –1.18) 2.31

Overall (I-squared = 65.3%, p = 0.001) –0.20 (–0.36, –0.03) 100.00

with estimated predictive interval . (–0.68, 0.29)

NOTE: Weights are from random effects analysis

–3 –2.5 –2 –1.5 –1 –.5 0 .5 1 1.5 2 2.5 3

Favours basal bolus Favours premixed

FIGURE 2 Comparison of weighted mean difference of baseline to endpoint change in HbA1c levels in the basal bolus versus premixed group in
patients with Type 2 diabetes. Model with random effects. Heterogeneity: Q = 34.62, df = 12 (P = 0.001), I2 = 65.3%. Overall effect: z-
score = 2.33 (P = 0.020), s2 = 0.0415.

these [20,22] found statistically significant differences

BMI
Two studies [15,25] reported an increase in BMI with both
types of insulin regimen at the end of the trial. A smaller
increase (0.4–0.8 kg/m2) was observed with the basal bolus
intervention compared with the biphasic group (0.9–1.4 kg/
m2). However, only results from one of the studies [25]
found this to be significantly different between the interven-
tions (P < 0.01). A third study [26] reported a non-signifi-
cant decrease in BMI with the biphasic treatment.
Body weight
Eight trials [16,19–22,24,28] reported the body weight of
patients at the end of treatment. All studies consistently
found an increase in body weight at the end of the
intervention compared with baseline, regardless of the type
of insulin regimen. Six trials [16,19–22,28] found results
suggesting that the basal bolus regimen was associated with
more weight gain than the biphasic regimen, but only two of
between the basal bolus and premixed groups. Overall, the
average weight gain calculated from the eight studies showed
that basal bolus treatment was associated with a 3.2 kg
increase, compared with an increase of 2.3 kg in the biphasic
insulin treatment.
Quality of life
Quality of life (QoL) and treatment satisfaction were
reported in three studies involving patients with Type 2
diabetes [19,22,26]. Two of the studies [19,22] used the
EuroQol–5D, EQ–5D instruments to measure health out-
come [29]. In addition to this, Bowering et al. [19] also used
the DHP–18 [30] tool to evaluate the psychological and
behavioural functioning of patients living with diabetes.
Levin et al. [22] evaluated QoL using two methods along
with the EQ–5D: the Diabetes Quality of Life (DQOL)
questionnaire and the Work Productivity and Activity
Impairment (WPAI) questionnaire [31]. In all of the above

ª 2015 The Authors.

590 Diabetic Medicine ª 2015 Diabetes UK
 Systematic Review or Meta-analysis DIABETICMedicine

Study WMD (95% CI) Weight

Fritsche et al. 2010 –0.51 (–0.76, –0.26) 14.90

Levin et al. 2011 –0.60 (–1.41, 0.21) 4.57

Ligthelm et al. 2006 –0.05 (–0.14, 0.04) 18.57

Miser et al. (Arm A) 2010 0.10 (–0.17, 0.37) 14.17

Miser et al. (Arm B) 2010 0.00 (–0.25, 0.25) 14.81

Rosenstock et al. 2008 –0.22 (–0.40, –0.04) 16.56

Shanmugasundar et al. 2012 –0.10 (–0.41, 0.21) 13.23

Yazici et al. 2011 –2.20 (–3.22, –1.18) 3.18

Overall (I-squared = 79.3%, p = 0.000) –0.22 (–0.42, –0.02) 100.00

with estimated predictive interval . (–0.83, 0.40)

NOTE: Weights are from random effects analysis

–3 –2.5 –2 –1.5 –1 –.5 0 .5 1 1.5 2 2.5 3

Favours basal bolus Favours premixed

FIGURE 3 Comparison of the weighted mean difference of baseline to endpoint HbA1c in non-insulin na€ıve patients with Type 2 diabetes: basal
bolus regimen versus premixed regimen. Model with random effects. Heterogeneity: Q = 33.84, df = 7 (P < 0.0001), I2 = 79.3%. Overall effect:
z-score = 2.15 (P = 0.031), s2 = 0.053.

studies, no statistically significant difference was found to
exist between treatment groups. One study [26] used the
insulin-therapy related QoL (ITR-QOL) inventory to assess
QoL and found that the total score was significantly higher in
the biphasic group compared with the basal bolus group,
with a higher score indicating a better QoL.
Discussion
The objective of this systematic review and meta-analysis
was to collate results from RCTs comparing the basal bolus
insulin regimen with the biphasic insulin regimen. To date,
this is the most comprehensive systematic review comparing
the two insulin regimens. Previous meta-analysis identified
four studies involving 1061 patients [32]. We observed that
in patients with Type 2 diabetes, the basal bolus regimen
leads to greater improvement in glycaemic control in the
majority of patients compared with the biphasic regimen.
This greater reduction occurred at the expense of higher daily
insulin requirements and weight gain, but no greater risk of
hypoglycaemia. In insulin na€ıve patients with Type 2 diabe-
tes, however, the biphasic regimen was shown to be as
effective as the basal bolus regimen in reducing HbA1c,
suggesting that both regimens may be suitable for insulin
initiation in patients with Type 2 diabetes. In non-insulin
na€ıve patients, the basal bolus regimen was superior in
reducing HbA1c levels.
Substantial heterogeneity was found in the meta-analysis
on HbA1c change. This may be attributed to a number of
confounding factors such as: poor study quality, small
treatment groups, short study duration and different study
designs. Because all studies were judged to have a high risk of
bias during quality assessment, the results should be inter-
preted cautiously. We pooled trials using regular and
analogue insulin because previous meta-analysis suggests
similar HbA1c outcomes between the two types [33]. The
observation of the subgroup analysis on insulin na€ıve
patients showing that biphasic insulin produced similar
HbA1c outcomes when compared with the basal bolus
insulin regimen support findings from our previous studies
in newly diagnosed, insulin-requiring patients with diabetes
[34] as well as in patients with Type 2 diabetes newly started
on insulin therapy [35]. Furthermore, in view of the
widespread use of biphasic insulin among patients with
Type 2 diabetes at the point of insulin initiation, finding
from this meta-analysis is reassuring. However, this finding

ª 2015 The Authors.

Diabetic Medicine ª 2015 Diabetes UK 591
DIABETICMedicine
Study
Bowering et al. 2012
Hirao et al. 2008
Jain et al. 2010
Kloos et al. 2007
Masuda et al. 2008
Overall (I-squared = 0.0%, p = 0.946)
with estimated predictive interval
NOTE: Weights are from random effects analysis
–2.5 –2 –1.5 –1
Favours basal bolus
FIGURE 4 Comparison of the weighted mean difference of baseline to endpoint HbA1c in insulin-na€ıve patients with Type 2 diabetes: basal bolus
regimen versus premixed regimen. Model with random effects. Heterogeneity: Q = 0.74, df = 4 (P = 0.946), I2 = 0.0%. Overall effect:
z-score = 0.79 (P = 432), s2 = 0.0.
was in contrast to the observation from the 4T study, which
reported a greater proportion of patients achieving a HbA1c
of < 6.5% when prandial or basal insulin alone was added to
metformin and sulfonylurea therapy, compared with bipha-
sic insulin group [36]. However, we could not include this
three-year RCT in our meta-analysis because it did not
compare biphasic insulin with a basal bolus group from the
outset, but rather, biphasic insulin vs prandial versus basal
only insulin, whilst continuing metformin and sulfonylurea
therapy. In this study, mean weight gain and hypoglycaemia
rates were highest in the prandial insulin only group. For
non-insulin na€ıve patients with Type 2 diabetes, the sub-
group analysis of change in HbA1c showed a pooled effect in
favour of the basal bolus regimen; however, high heteroge-
neity was also detected in the results, which should be
interpreted cautiously. A possible explanation for this
heterogeneity may be the large variations in the number of
patients between studies (28 to 484) and intervention
duration (8 to 52 weeks).
Overall, hypoglycaemia was comparable between the two
regimens, conflicting with the perception that the basal
bolus regimen is associated with a lower risk of hypoglyca-
emia due to the more flexible dose adjustment [7,37].
592
Biphasic vs basal bolus insulin regimen  C. Wang et al.
WMD (95% CI) Weight
–0.04 (–0.78, 0.70) 24.53
0.00 (–0.67, 0.67) 30.53
–0.17 (–1.18, 0.84) 13.27
–0.40 (–1.12, 0.32) 26.04
–0.20 (–1.75, 1.35) 5.63
–0.15 (–0.52, 0.22) 100.00
. (–0.75, 0.45)
–.5 0 .5 1 1.5 2 2.5
Favours premixed
Inconsistencies in the definitions of overall hypoglycaemia
or severe hypoglycaemia, in addition to missing measures of
variance, meant that no meta-analysis could be performed.
One study [38] that included patients with both Type 1
diabetes and Type 2 diabetes, reported a much higher
number of severe hypoglycaemic episodes during the inter-
vention period in comparison with other studies. It is
difficult to draw definitive reasons as to why this occurred,
but higher total daily insulin doses may have been a
contributing factor.
Higher insulin requirements where found in the basal
bolus regimen group, which may offer a possible explanation
for the greater reduction in HbA1c levels; however, this was
not associated with a greater occurrence of hypoglycaemia.
Because both the patient and the person administering
treatment were aware of the insulin regimen due to the
openness of the trial design, this might have influenced when
and how much the insulin dose was increased. Some studies
have reported that the basal bolus regimen was more
aggressively titrated than the biphasic regimen, possibly
due to basal bolus adjustment being more familiar and
common in routine practice [28]. No meta-analysis was
performed on insulin requirements due to differences in
ª 2015 The Authors.
Diabetic Medicine ª 2015 Diabetes UK
 Systematic Review or Meta-analysis
reporting, e.g. IU per person, or IU/kg, and missing values of
variance for calculation.
We did not conduct a meta-analysis for the change in body
weight and BMI because no studies reported SD for mean
change in body weight. Consistent with previous study
findings, body weight and BMI increased with insulin
intensification regardless of regimen type [35]. Weight gain
was slightly higher in the basal bolus regimen group
compared with the biphasic group, corresponding with the
higher daily insulin doses.
QoL and other patient-centred outcomes have been
reported to be important influencers of patient adherence
to insulin treatment [39]. It is thought that patients would
prefer a biphasic insulin regimen due to its simplicity and
the reduced frequency of injections. The open-label designs
of all trials allowed the evaluation of the QoL of patients
in the two different regimens, because the number of
injections per day accurately reflects the injection schedule
in daily life. However, the open design and the different
methods and scales used to measure QoL made it difficult
to make objective comparisons between studies. However,
significant differences were found in two studies. Testa
et al. [38] demonstrated that acceptance of the two
regimens was comparable, thereby overcoming the percep-
tion of reduced adherence to the basal bolus regimen due
to the burden an increased frequency of daily injections
[40]. However, because this study did not specify the tools
used to measure QoL, we cannot assume that validated
measurement tools were used. Significant improvement in
QoL was found by Masuda et al. [26] in the biphasic
regimen group using the ITR–QOL, with a major contrib-
uting factor being the increased psychological burden of
having to find places to inject insulin in the basal bolus
group.
The open-label design of all studies included in the
systematic review may be considered a limitation, however,
given the nature of the interventions, blinding was not
feasible for reasons such as: different appearances of the
different insulin preparations, the difference in number of
injections required per day, and the need to ensure patient
safety when administering and titrating doses. In order to
evaluate QoL and treatment satisfaction, the number of
injections in each regimen also had to reflect real clinical
practice without the use of placebo injections.
To summarize, the HbA1c outcome between the two
insulin regimens appears to depend on whether patients are
already on insulin or about to start insulin treatment.
Current evidence directly comparing the basal bolus and
biphasic interventions is limited to RCTs with a duration of
no more than 52 weeks, thereby restricting the findings to
shorter term outcomes only. No studies were designed to
investigate long-term effects, therefore, it is unclear how
metabolic outcomes such as HbA1c, weight and hypoglyca-
emia rates, as well as QoL markers will be affected by the
choice of insulin regimen. Further research is needed not only
ª 2015 The Authors.
Diabetic Medicine ª 2015 Diabetes UK
DIABETICMedicine
to explore the long-term efficacy and safety compliance of
both insulin regimens, but also novel insulin regimens such as
the basal plus or insulin with glucagon-like peptide–1 (GLP-
1) analogue combinations.
Funding sources
None.
Competing interests
None declared.
References
1 DeWitt DE, Hirsch IB. Outpatient insulin therapy in type 1 and
type 2 diabetes mellitus: scientific review. JAMA 2003; 289: 2254–
2264.
2 The Diabetes Control and Complications Trial Research Group.
The effect of intensive treatment of diabetes on the development
and progression of long-term complications in insulin-dependent
diabetes mellitus. New Engl J Med 1993; 329: 977–986.
3 Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10–
Year follow-up of intensive glucose control in type 2 diabetes. N
Engl J Med 2008; 359: 1577–1589.
4 Garber AJ. Methods to enhance delivery of prandial insulin and
basal-prandial insulin. Diabetes Obes Metab 2013; 15(Suppl 1):
11–17.
5 Nathan DM, Buse JB, Davidson MB, Ferrannini E, Holman RR,
Sherwin R et al. Medical management of hyperglycemia in type 2
diabetes: a consensus algorithm for the initiation and adjustment of
therapy: a consensus statement of the American Diabetes Associ-
ation and the European Association for the Study of Diabetes.
Diabetes Care 2009; 32: 193–203.
6 Hamaty M. Insulin treatment for type 2 diabetes: when to start,
which to use. Cleve Clin J Med 2011; 78: 332–342.
7 Peyrot M, Rubin RR, Kruger DF, Travis LB. Correlates of insulin
injection omission. Diabetes Care 2010; 33: 240–245.
8 Turner HE, Matthews DR. The use of fixed-mixture insulins in
clinical practice. Eur J Clin Pharmacol 2000; 56: 19–25.
9 Type 2 Diabetes: newer agents [article online], 2009. Available
at http://www.nice.org.uk/nicemedia/live/12165/44318/44318.pdf.
Last accessed 19 July 2013.
10 National Institute for Health and Clinical Excellence. The Guide-
lines Manual. London: NICE, 2012.
11 Elbourne DR, Altman DG, Higgins JPT, Curtin F, Worthington
HV, Vail A. Meta-analyses involving cross-over trials: methodo-
logical issues. Int J Epidemiol 2002; 31: 140–149.
12 Higgins JPT, Altman DG, Sterne JAC (Eds). Cochrane Hand-
book for Systematic Reviews of Interventions, Version 5.1.0
[updated March 2011], Chapter 8. The Cochrane Collaboration,
2011.
13 Coster S, Gulliford MC, Seed PT, Powrie JK, Swaminathan R.
Monitoring blood glucose control in diabetes mellitus: a systematic
review. Health Technology Assessment 2000; 4: i–iv, 1–93.
14 Frison L, Pocock SJ. Repeated measures in clinical trials: analysis
using mean summary statistics and its implications for design. Stat
Med 1992; 11: 1685–1704.
15 Kloos C, Samann A, Lehmann T, Braun A, Heckmann B, Muller
UA. Flexible intensive versus conventional insulin therapy in
insulin-naive adults with type 2 diabetes: an open-label, random-
ized, controlled, crossover clinical trial of metabolic control and
patient preference. Diabetes Care 2007; 30: 3031–3032.
593
DIABETICMedicine
16 Miser WF, Arakaki R, Jiang H, Scism-Bacon J, Anderson PW,
Fahrbach JL. Randomized, open-label, parallel-group evaluations
of basal-bolus therapy versus insulin lispro premixed therapy in
patients with type 2 diabetes mellitus failing to achieve control
with starter insulin treatment and continuing oral antihyperglyce-
mic drugs: a noninferiority intensification substudy of the DURA-
BLE trial. Clin Ther 2010; 32: 896–908.
17 Buse JB, Wolffenbuttel BHR, Herman WH, Shemonsky NK, Jiang
HH, Fahrbach JL et al. DURAbility of basal versus lispro mix 75/
25 insulin efficacy (DURABLE) trial 24-week results:safety and
efficacy of insulin lispro mix 75/25 versus insulin glargine added to
oral antihyperglycemic drugs in patients with type 2 diabetes.
Diabetes Care 2009; 32: 1007–1013.
18 Liebl A, Prager R, Binz K, Kaiser M, Bergenstal R, Gallwitz B et al.
Comparison of insulin analogue regimens in people with type 2
diabetes mellitus in the PREFER study: a randomized controlled
trial. Diabetes Obes Metab 2009; 11: 45–52.
19 Bowering K, Reed VA, Felicio JS, Landry J, Ji L, Oliveira J. A study
comparing insulin lispro mix 25 with glargine plus lispro therapy in
patients with Type 2 diabetes who have inadequate glycaemic
control on oral anti-hyperglycaemic medication: results of the
PARADIGM study. Diabet Med 2012; 29: e263–272.
20 Fritsche A, Larbig M, Owens D, Haring HU. Comparison between
a basal-bolus and a premixed insulin regimen in individuals with
type 2 diabetes-results of the GINGER study. Diabetes Obes
Metab 2010; 12: 115–123.
21 Jain SM, Mao X, Escalante-Pulido M, Vorokhobina N, Lopez I,
Ilag LL. Prandial-basal insulin regimens plus oral antihyperglycae-
mic agents to improve mealtime glycaemia: initiate and progres-
sively advance insulin therapy in type 2 diabetes. Diabetes Obes
Metab 2010; 12: 967–975.
22 Levin PA, Zhang Q, Mersey JH, Lee FY, Bromberger LA, Bhushan
M et al. Glycemic control with insulin glargine plus insulin glulisine
versus premixed insulin analogues in real-world practices: a cost-
effectiveness study with a randomized pragmatic trial design. Clin
Ther 2011; 33: 841–850.
23 Ligthelm RJ, Mouritzen U, Lynggaard H, Landin-Olsson M, Fox
C, le Devehat C et al. Biphasic insulin aspart given thrice daily is as
efficacious as a basal-bolus insulin regimen with four daily
injections: a randomised open-label parallel group four months
comparison in patients with type 2 diabetes. Exp Clin Endocrinol
Diabetes 2006; 114: 511–519.
24 Shanmugasundar G, Bhansali A, Walia R, Dutta P, Upreti V.
Comparison of thrice daily biphasic human insulin (30/70) versus
basal detemir & bolus aspart in patients with poorly controlled
type 2 diabetes mellitus – a pilot study. Indian J Med Res 2012;
135: 78–83.
25 Hirao K, Arai K, Yamauchi M, Takagi H, Kobayashi M. Six-month
multicentric, open-label, randomized trial of twice-daily injections
of biphasic insulin aspart 30 versus multiple daily injections of
insulin aspart in Japanese type 2 diabetic patients (JDDM 11).
Diabetes Res Clin Pract 2008; 79: 171–176.
26 Masuda H, Sakamoto M, Irie J, Kitaoka A, Shiono K, Inoue G et al.
Comparison of twice-daily injections of biphasic insulin lispro and
basal-bolus therapy: glycaemic control and quality-of-life of
594
Biphasic vs basal bolus insulin regimen  C. Wang et al.
insulin-na€ıve type 2 diabetic patients. Diabetes Obes Metab
2008; 10: 1261–1265.
27 Yazici D, Tas S, Emir H, Sunar H. A comparison of preprandial
mixed insulin given three times daily and basal-bolus insulin
therapy started postoperatively on patients having coronary artery
bypass graft surgery [Koroner arter bypass cerrahisi geciren
hastalarda 3’lu karısım ve 4’lu yogun insulin tedavilerinin
karsılastırılması]. Marmara Med J 2012; 25: 16–19.
28 Rosenstock J, Ahmann AJ, Colon G, Scism-Bacon J, Jiang H,
Martin S. Advancing insulin therapy in type 2 diabetes previously
treated with glargine plus oral agents: prandial premixed (insulin
lispro protamine suspension/lispro) versus basal/bolus (glargine/
lispro) therapy. Diabetes Care 2008; 31: 20–25.
29 EuroQol Group. EuroQol – a new facility for the measurement of
health-related quality of life. Health Policy (Amsterdam)1990; 16:
199–208.
30 Meadows KA, Abrams C, Sandbaek A. Adaptation of the Diabetes
Health Profile (DHP–1) for use with patients with Type 2 diabetes
mellitus: psychometric evaluation and cross-cultural comparison.
Diabet Med 2000; 17: 572–580.
31 Reilly MC, Zbrozek AS, Dukes EM. The validity and reproduc-
ibility of a work productivity and activity impairment instrument.
PharmacoEconomics 1993; 4: 353–365.
32 Lasserson DS, Glasziou P, Perera R, Holman RR, Farmer AJ.
Optimal insulin regimens in type 2 diabetes mellitus: systematic
review and meta-analyses. Diabetologia 2009; 52: 1990–2000.
33 Siebenhofer A, Plank J, Berghold A, Jeitler K, Horvath K, Narath
M et al. Short acting insulin analogues versus regular human
insulin in patients with diabetes mellitus. The Cochrane Database
of Systematic Reviews, 2006: CD003287.
34 Idris I, Pillai A, Fernando DJ, Thomson G, Tate H. Responders to
insulin therapy at 18 months in adults with newly diagnosed
diabetes: which insulin regimen? Diabet Med 2013; 30: e95–100.
35 Owen V, Seetho I, Idris I. Predictors of responders to insulin
therapy at 1 year among adults with type 2 diabetes. Diabetes Obes
Metab 2010; 12: 865–870.
36 Holman RR, Farmer AJ, Davies MJ, Levy JC, Darbyshire JL,
Keenan JF et al. Three-year efficacy of complex insulin regimens in
type 2 diabetes. N Engl J Med 2009; 361: 1736–1744.
37 Cryer PE, Davis SN, Shamoon H. Hypoglycemia in diabetes.
Diabet Care 2003; 26: 1902–1912.
38 Testa MA, Gill J, Su M, Turner RR, Blonde L, Simonson DC.
Comparative effectiveness of basal-bolus versus premix analog
insulin on glycemic variability and patient-centered outcomes
during insulin intensification in type 1 and type 2 diabetes: a
randomized, controlled, crossover trial. J Clin Endocrinol Metab
2012; 97: 3504–3514.
39 Hanestas BR, Albreksten G. Quality of life, perceived difficulties in
adherence to a diabetes regimen, and blood glucose control. Diabet
Med 1991; 8: 759–764.
40 Zambanini A, Newson RB, Maisey M, Fehrer MD. Injection
related anxiety in insulin-treated diabetes. Diabetes Res Clin Pract
1999; 46: 239–246.
ª 2015 The Authors.
Diabetic Medicine ª 2015 Diabetes UK