Intravenous ferrous sucrose versus placebo in addition to oral iron therapy for the treatment of severe postpartum anaemia: a randomised controlled trial




The aim of the study was to evaluate the effectiveness of intravenous iron versus placebo added to standard oral iron therapy in the treatment of severe postpartum anaemia.


A randomised, double-blind, parallel-group, placebo-controlled clinical trial was performed in a single centre.


Hospital Clinic of Barcelona, Barcelona, Spain.


A cohort of 72 women with severe postpartum anaemia (6.0–8.0 g/dl) treated with oral ferrous sulphate (two tablets of 525 mg).


Women were randomised to receive either intravenous ferrous sucrose (200 mg/24 hours for two consecutive days) or intravenous placebo, in addition to standard iron therapy. Clinical and laboratory data were obtained at 1, 2, and 6 weeks.

Main outcome measures

Haemoglobin and haematocrit at 1, 2, and 6 weeks. Other haematological and clinical parameters, psychological status, and adverse side effects were also evaluated.


Haemoglobin and haematocrit values were comparable in women receiving intravenous iron or placebo in addition to oral iron therapy at any of the time points. At 6 weeks, haemoglobin level (mean ± SD) was 12.2 ± 1.0 versus 12.2 ± 0.9 g/dl, with a mean difference of −0.03 (95% CI −0.6 to 0.6), in the placebo and in the intravenous iron groups, respectively. No differences were found between clinical symptoms of anaemia, psychological status, and adverse side effects between groups.


Intravenous iron added to oral iron therapy did not show significant benefits over placebo, neither in haemoglobin rise nor in symptoms or adverse side effects.


The occurrence of postpartum anaemia is not unusual.[1, 2] Around 15% of women may lose over 500 ml of blood during delivery, with the risk of anaemia and the subsequent depletion of iron stores. Oral iron therapy is regularly used to treat iron deficiency in these women,[3] achieving a mean increase in haemoglobin values of approximately 2.83 g/dl at 30 days.[4] Limited absorption and adverse gastrointestinal side effects may, however, affect patient compliance.[5]

Since the introduction of intravenous iron therapy, several studies have reported its use in postpartum anaemia.[6-12] Only a few studies with controversial results have evaluated its potential benefit in severe postpartum anaemia, compared with oral iron,[7, 8] when haemoglobin levels drop to below 7.5–9.0 g/dl and blood transfusion might be the alternative treatment. In order to minimise several potential risks such as infectious disease transmission and exposure to allogeneic blood products, as well as reduce healthcare costs, blood transfusion should be restricted to symptomatic cases, very severe anaemia, or specific individualised cases.[13] Despite these considerations, several studies have shown that a number of transfusions are used inappropriately.[14]

The postpartum period is potentially a time with intense psychological and physical stress. Postpartum anaemia may increase the risk of depression, anxiety, asthenia, lethargy, lactation failure, and other morbidities, resulting in a longer hospital stay and reinforcing the need to seek appropriate therapy.[15, 16] Recovery of iron stores may take up to 2 months when treated only with oral iron supplementation.[4]

The aim of this study was to evaluate whether intravenous ferrous sucrose therapy, added to the standard oral ferrous sulphate treatment, improves the recovery of haematological parameters in women with severe postpartum anaemia, compared with placebo. Anaemic symptoms, depression, and anxiety tests in the postpartum period were also evaluated.


The study was designed as a randomised single-centre, double-blind, placebo-controlled trial. It was approved by the ethical committee of the hospital clinic on 8 June 2005, and women were recruited from November 2005 to January 2008 in our centre. This study was partially supported by J Uriach & Co.

Eligible women were those aged 18 years or more with severe postpartum anaemia (with haemoglobin levels between 6.0 and 8.0 g/dl) diagnosed within 48 hours postpartum. Blood analyses were indicated because of caesarean section, remarkable blood loss during delivery, or clinical signs of anaemia. After diagnosis, women willing to participate in the study were informed and signed a written consent form. Exclusion criteria were antenatal history of chronic anaemia, suspicion or evidence of infection, history of asthma, eczema, or topical allergy, intolerance to oral iron, women with blood transfusion criteria (haemoglobin <6 g/dl or intolerable symptoms of anaemia), anaemia attributable to other causes than blood loss or iron deficiency, history of cirrhosis, hepatitis, or elevation of liver transaminases three times the upper limits of normality, overload or alteration in iron metabolism, history of hypersensitivity to intravenous iron therapy, or unwillingness to participate in the study.

According to the standard care for postpartum anaemia in our centre, all women received two tablets of 525 mg of oral ferrous sulphate (containing 105 mg of elemental iron) per day for 30 days. As we were dealing with women with severe anaemia, an oral placebo was considered unethical, and intravenous placebo or iron sucrose had to be given in addition to standard treatment. The list of randomised numbers was computer generated, and each number was then introduced into a sealed envelope. Participating women were randomised at 24–48 hours after delivery into two groups, with an equal proportion between treatments (1 : 1). Thus, the ferrous sucrose group received, in addition to oral ferrous sulphate, 200 mg/24 hours intravenous ferrous sucrose (Venofer, Vifor France SA, Neully-sur-Seine, France), diluted in 200 ml of NaCl 0.9%, for 2 days. The placebo group received 200 ml/24 hours intravenous NaCl 0.9%, for 2 days. An opaque perfusion system was used in both groups to avoid the identification of the treatment received and maintain the double-blind nature of the study. Commercial directions recommend a maximum of 200 mg per ferrous sucrose infusion three times a week; however, these were targeted to patients with chronic kidney disease. Postpartum women are generally young and healthy, and a dilutional component cannot be excluded. As women with a vaginal delivery remained in hospital for 48 hours, we decided to use a total dose of 400 mg divided into two consecutive days because it is well tolerated and does not lengthen the duration of the hospital stay.

Demographic characteristics of the women, mode of delivery, maternal hospital length of stay, and neonatal outcomes were obtained because of the potential impact on psychological outcomes. Composite neonatal morbidity included one of the following: respiratory distress syndrome, pulmonary bronchodysplasia, intraventricular haemorrhage (III or IV), periventricular leukomalacia, or necrotising enterocolitis.

Clinical data and blood samples were taken at recruitment into the study (baseline), and at 1, 2, and 6 weeks after the initiation of treatment, by questionnaire at a hospital visit (Appendix S1). The time points were chosen based on previous studies. Women randomised but who finally required blood transfusion because of clinical indication were considered as a treatment failure, and were not included in the final analysis. The main end point of the study was the difference in the mean haemoglobin and haematocrit levels between the two groups of treatment at 6 weeks postpartum. Other haematological parameters, such as ferritin, iron-binding capacity, reticulocyte count, serum iron levels, and mean corpuscular volume (MCV) were also compared. Longitudinal progression of haemoglobin and haematocrit levels within groups was also evaluated. Clinical anaemic signs such as pulse and blood pressure, and symptoms such as headache, fatigue, tinnitus, dyspnoea, palpitations, tingling, dizziness, nausea, and difficulty in concentration were also assessed using the visual analogue scale (VAS). Levels of depression and anxiety were screened by the Edinburgh Postnatal Depression Scale (EPDS)[17] and the State–Trait Anxiety Inventory (STAI), respectively, during the first 48 hours after delivery, and at 1, 2, and 6 weeks. Potential depression was considered with an EPDS score of ≥11 over 30 points, validated for the Spanish population.[18] State and trait measures were evaluated at 48 hours postpartum, and state was also assessed at 1, 2, and 6 weeks. Relevant anxiety was considered with an STAI score of >40 points, as used in other maternal and non-obstetrical studies.[19, 20]

To assess safety, side effects of treatment such as metallic taste, constipation, epigastric pain, tingling, and nausea were also scored using the VAS. A VAS score ≥7 was considered relevant.

To demonstrate a difference of 1 g/dl of haemoglobin (considering an SD of ±1) or 3 points in haematocrit (considering an SD of ± 3.4) between groups after 6 weeks of treatment,[5] between 22 and 28 women per arm were needed, respectively (for a power of 90% and an α error of 5%). To account for an estimate of around 15–20% of loss to follow-up, a final population size of 72 women was calculated.

Data were introduced in an Access database and analysed using spss 16.0 (SPSS Inc., Chicago, IL, USA). Analysis was performed with the Student's t-test for independent samples (continuous variables) and the chi-square test and Fisher's exact test (dichotomous variables). Analysis of variance with a quadratic polynomial contrast was used for repeated measurements. Differences were expressed as mean difference or proportion difference (95% CI). Bonferroni's rule was not applied.


Seventy-two women were recruited in the study. Four women needed transfusion, two in each treatment arm, and eight were lost to follow-up. Thus, 29 women in the placebo group and 31 in the ferrous sucrose group remained for the statistical analysis (Figure 1). Baseline haematological parameters in the women lost to follow-up were similar to those of the women included in the analysis: five and three women in the placebo and the iron sucrose groups, respectively, showed a haemoglobin (mean ± SD) level of 7.3 ± 0.1 versus 7.1 ± 0.9 g/dl, and haematocrit (mean ± SD) 22.4 ± 0.6 versus 21.3 ± 2.5%, respectively. No statistically significant differences were observed in baseline characteristics between the two groups of study (Table 1). Baseline MCV was (mean ± SD) 85.2 ± 8.9 fl and 83.8 ± 8.2 fl (mean difference −1.4, 95% CI −3.1 to −5.8) in the placebo and ferrous sucrose group, respectively.

Table 1. Baseline characteristics of the group.
 Oral iron and placebo n = 29Oral iron and ferrous sucrose n = 31
  1. No significant differences between groups were observed. Data are expressed in n (%) or mean (±SD) when appropriate.

  2. a

    Preterm with respiratory distress syndrome.

  3. b

    Preterm with respiratory distress syndrome and hypoxic ischaemic encephalopathy.

Maternal age 29.9 (±5.3) years29.5 (±5.8) years
Nulliparity 13 (44.8%)20 (64.5%)
Mode of delivery
Caesarean section12 (41.4%)10 (32.3%)
Operative delivery9 (31.0%)11 (35.5%)
Gestational age at delivery 39.2 (±3.2) weeks39.1 (±2.4) weeks
Maternal length of stay 3.8 (±1.7) days3.5 (±1.4) days
Birthweight 3263 (±741) g3519 (±585) g
Composite morbidity 1a (3.4%)1b (3.2%)
Figure 1.

Consolidated Standards of Reporting Trials (CONSORT) flow chart.

There were no significant differences in haemoglobin and haematocrit levels between the two groups at any of the time points studied. When the progression of the haematological parameters was evaluated, a significant increase was observed in haemoglobin and haematocrit values in each group. Indeed, both groups achieved haemoglobin values above 12 g/dl at 6 weeks (Table 2).

Table 2. Comparison of haemoglobin, haematocrit, ferritin, iron-binding capacity, and absolute reticulocyte count, between groups.
 Oral iron and placebo n = 29Oral iron and ferrous sucrose = 31Mean differences (95% CI)
  1. Data are expressed as mean (±SD), mean difference (95% CI).

  2. a

    anova with quadratic polynomial contrast.

  3. b

    Significant differences.

Haemoglobin (g/dl)
Baseline7.3 (±0.5)7.2 (±0.5)−0.1 (−0.4 to 0.2)
1 weeka9.8 (±1.2)9.7 (±0.9)−0.1 (−0.7 to 0.5)
2 weeksa10.7 (±0.9)10.7 (±1.0)0.01(−0.6 to 0.6)
6 weeksa

12.2 (±1.0)

< 0.01a

12.2 (±0.9)

P < 0.01a

−0.03 (−0.6 to 0.6)
Haematocrit (%)
Baseline22.2 (±1.6)21.7 (±2.5)−0.5 (−1.6 to 0.6)
1 weeka30.3 (±4.1)30.4 (±2.9)0.1 (−2.0 to 2.1)
2 weeksa33.6 (±2.7)33.7 (±3.2)0.1 (−1.8 to 1.9)
6 weeksa

37.6 (±3.0)

P < 0.01a

37.8 (±3.0)

P < 0.01a

0.2 (−1.6 to 2.1)
Ferritin (ng/ml)
Baseline38.1 (±36.1)37.1 (±34.3)−1.0 (−19.6 to 17.7)
1 week33.0 (±25.1)192.5 (±85.3)159.5 (122.2 to 196.9)b
2 weeks34.7 (±21.5)107.3 (±41.9)72.6 (49.9 to 95.2)b
6 weeks27.2 (±16.4)44.4 (±53.0)17.2 (−8.4 to 42.8)
Iron-binding capacity (mg/l)
Baseline4.8 (±0.9)4.7 (±1.1)−0.1 (−0.7 to 0.5)
1 week4.9 (±0.7)4.6 (±0.6)−0.3 (−0.7 to 0.1)
2 weeks4.6 (±0.7)4.1 (±0.6)−0.5 (−0.9 to −0.1)b
6 weeks3.9 (±0.6)3.8 (±0.5)−0.1 (−0.4 to 0.2)
Reticulocyte levels
Baseline74.3 (±20.2)84.8 (±24.6)10.6 (−1.8 to 23.0)
1 week132.2 (±61.7)179.4 (±77.7)47.2 (5.3 to 89.1)b
2 weeks122.5 (±58.5)136.7 (±62.1)14.2 (−25.1 to 53.5)
6 weeks47.0 (±20.7)56.0 (±24.0)9.0 (−5.8 to 23.8)

A significant increase of ferritin was observed in the ferrous sucrose group compared with the placebo group at 1 and 2 weeks of administration; however, no differences were found at 6 weeks. No significant differences in serum iron levels between groups were observed at any time point. The iron-binding capacity was significantly lower at 2 weeks and reticulocyte levels were higher in the first week in the ferrous sucrose group, compared with the placebo group, although all parameters were similar at 6 weeks (Table 2). Figure S1 shows ferritin, iron-binding capacity, and reticulocyte count.

There were no differences in maternal pulse, blood pressure, and symptoms of anaemia between the groups, apart from a non-clinically relevant difference in systolic blood pressure at 1 week of follow-up (Table 3). The risk of depression (EDPS ≥11) did not show significant differences between groups. Evaluation of maternal anxiety with the STAI at baseline detected only one woman with anxiety trait and state in the placebo group, and one with an isolated anxiety state in the ferrous sucrose group. Only the anxiety state remained present in the first woman at 6 weeks. Table 3 also shows the number of women with side effects, scoring a VAS score ≥7.

Table 3. Comparison of clinical signs and symptoms of anaemia, depression and side effects, between groups.
 Oral iron and placebo n = 29Oral iron and ferrous sucrose n = 31Mean or proportion difference (95% CI)
  1. Data are expressed as mean (±SD) or n (%) and mean or proportion difference (95% CI).

  2. a

    Symptoms of anaemia: headache, fatigue, tinnitus, dyspnoea, palpitations, tingling, dizziness, nausea, and difficulty in concentration.

  3. b

    Side effects: metallic taste, constipation, epigastric pain, tingling, and nausea.

  4. c

    Significant differences.

Maternal pulse
Baseline87.2 (±11.5)93.2 (±15.2)6.0 (−2.6 to 14.5)
1 week82.6 (±8.4)85.3 (±8.2)2.7 (−2.9 to 8.2)
2 weeks78.8 (±6.0)80.1 (±20.9)1.3 (−9.4 to 12.0)
6 weeks79.2 (±9.7)77.3 (±7.7)−1.9 (−8.1 to 4.3)
Systolic pressure (mmHg)
Baseline110.7 (±12.2)110.2 (±13.8)−0.6 (−8.7 to 7.6)
1 week110.3 (±11.3)102.6 (±10.3)−7.7 (−14.9 to −0.5)c
2 weeks103.1 (±25.7)103.5 (±25.9)0.4 (−16.7 to 17.5)
6 weeks113.1 (±10.9)113.2 (±13.7)0.04 (−8.4 to 8.5)
Diastolic pressure (mmHg)
Baseline62.6 (±8.9)64.6 (±7.6)2.1 (−3.0 to 7.1)
1 week67.4 (±9.4)65.5 (±15.7)−2.0 (−10.9 to 7.0)
2 weeks65.7 (±10.3)60.6 (±15.6)−5.1 (−13.8 to 3.6)
6 weeks68.2 (±8.4)67.4 (±10.3)−0.8 (−7.1 to 5.6)
Number of women with any symptoms of anaemia, scored ≥7 a
Baseline16 (55.2%)12 (38.7%)−0.2 (−0.4 to 0.1)
1 week4 (17.2%)7 (22.6%)0.1 (−0.1 to 0.3)
2 weeks5 (17.2%)3 (9.7%)−0.1 (−0.3 to 0.1)
6 weeks1 (3.4%)3 (9.7%)0.1 (−0.1 to 0.2)
Risk of depression (EDPS ≥11)
Baseline7 (24.1%)9 (29.0%)0.1 (−0.2 to 0.3)
1 week2 (6.9%)6 (19.4%)0.1 (−0.04 to 0.3)
2 weeks3 (10.3%)5 (16.1%)0.1 (−0.1 to 0.2)
6 weeks2 (6.9%)5 (16.1%)0.1 (−0.1 to 0.3)
Number of women with any side effects, scored ≥7 b
Baseline10 (34.5%)4 (12.9%)−0.2 (−0.4 to <0.0001)
1 week6 (20.7%)4 (12.9%)−0.1 (−0.3 to 0.1)
2 weeks7 (24.1%)2 (6.5%)−0.2 (−0.4 to 0.001)
6 weeks5 (17.2%)2 (6.5%)−0.1 (−0.3 to 0.1)

No woman withdrew from the trial because of adverse side effects. No severe adverse outcomes were registered in either group. There were no significant differences between both groups, although a tendency to more side effects was observed in the group receiving oral iron only.


Main findings

The main finding of our study is that in women with severe postpartum anaemia, the addition of an intravenous ferrous sucrose regime to a standard oral iron therapy does not improve the main haematological parameters at 1, 2, and 6 weeks of follow-up, compared with those receiving oral iron therapy alone. Although reticulocyte count and ferritin at week 1 were increased compared with the placebo group, indicating that ferrous sucrose contributed to a faster stimulation of erythropoiesis, standard haematological parameters (haemoglobin and haematocrit) did not differ at any time points.

Several studies about postpartum anaemia and intravenous iron sucrose treatment have been performed;[6-12] however, only two of them have focused on treatment of severe postpartum anaemia with haemoglobin levels under 9 g/dl, and compared oral versus intravenous administration.[7, 8] In the study by Westad et al., which included women with haemoglobin levels of 6.5–8.5 mg/ml, the first follow-up was at 4 weeks, and there is no information for levels over a shorter period of time.[7] Bhandal et al., who included women with haemoglobin levels <9 mg/ml, noted a faster recovery of haemoglobin with intravenous therapy at days 5 and 14 of treatment.[8] Differences in the study design (i.e. range of haemoglobin defined or sample size) may account for the differences with our results as one may hypothesise that oral iron absorption may be increased in cases with more severe acute anaemia. On the contrary, basal ferritin was notably lower in the study by Bhandal et al. (11 and 13 ng/ml in the intravenous and oral group, respectively),[8] compared with the study by Westad et al.[7] (20–25 ng/ml) or with our study (37 and 38 ng/ml in the intravenous and oral group, respectively), which may mean that intravenous iron may perform better when stores are severely depleted. Also, in the study by Bhandal et al., 42 out of 43 women underwent caesarean section, which may have prevented a faster recovery of the haematological parameters in the oral iron arm.[8] In any case, the levels of haematological parameters at 6 weeks in both studies are comparable with ours. The iron formula or dosage used in other studies may account for differences in the outcomes.

More side effects were observed in the group of women who received oral iron alone, although sample size may have prevented differences reaching statistical significance. However, no women abandoned the study because of side effects.

Strengths and weaknesses

The main strength of our study is that the design is robust. The design of previous studies comparing oral versus intravenous iron prevented double blinding in the study. On the contrary, in our study women allocated to intravenous placebo received standard oral iron therapy while women allocated to the intravenous iron group might have had the oral iron absorption eventually blocked by the intravenous treatment.[2] In this case, oral iron could functionally be considered as a placebo. Therefore, both women and doctors were blinded to the treatment received. Blinding is particularly relevant to evaluate clinical and psychological symptoms. In the study conducted by Westad et al., no significant differences in quality-of-life tests were found, but improvement in physical, mental, and total fatigue scores were observed in the intravenous treatment group.[7] Although they studied a similar range of anaemia as that in our study, a placebo group was not included. Therefore, the levels of fatigue could potentially be biased because women were not blinded to which treatment they had received. Regarding psychological outcomes, the rate of depression detected at 6 weeks in both groups was similar in our study (11.7%), and did not differ greatly from the range of 16–22% described in a previous epidemiological study.[20] The frequency of anxiety was very low, and because of the sample size we cannot extract conclusions from these data.

Another strength is that we included only women with severe postpartum anaemia. Clinical research with conservative therapies in this range of anaemia is crucial in order to avoid over-treatment with blood products.[13, 14] Although our study was not as large as that conducted by Westad et al., it was powered enough to demonstrate a difference between groups of 1 mg/dl haemoglobin compared with the 0.5 mg/dl difference proposed by Westad et al., which may not be so clinically relevant.[7] Besides, our data suggest similar conclusions as those of Westad et al.[7]

Our study has some limitations. First, it is known that the concurrent treatment of oral and intravenous iron reduces oral absorption,[2] and consequently limits the bioavailability. This could potentially have prevented the faster and higher recovery of haematological parameters in the intravenous iron group; however, had these women been treated exclusively with intravenous iron, it is unlikely that the final haematological results would have been different than those reported. Besides, to maintain blinding it would have been unethical not to administer oral iron to women with severe postpartum anaemia allocated to the placebo group.

Haemoglobin or haematocrit values prior to delivery were not known in this study. Women with a known history of chronic anaemia were excluded and the MCV at baseline was within normal ranges and similar in both groups. There were eight cases lost to follow-up: five in the placebo group and three in the ferrous sucrose group (Figure 1). The sample size remained sufficient for the final analysis. Baseline characteristics of the women lost to follow-up were similar to the population studied, making the assumption of similar results clinically plausible.

The use of a limited dose of ferrous sucrose was also a limitation. Different doses have been reported for the treatment of postpartum anaemia, ranging from 100 mg in a single dose[21] to 1200 mg in 3–4 days in very severe anaemia.[11] In other studies in which severe postpartum anaemia has been studied, the authors reported the use of 200 mg at days 2 and 4, or a total dose of 600 mg per women over three consecutive days.[7, 8] In any case, the final haemoglobin levels after treatment were similar to those of our study.

Finally, we must emphasise that in the current trial we evaluated a particular profile of patient: young, healthy women who were highly compliant with oral therapy, and had preserved or even increased their capacity to absorb oral iron because of their pregnancy status. Compliance may not be the case in a real clinical setting, thereby limiting the recovery of anaemia in the oral iron group. Moreover, benefit of intravenous ferrous sucrose in other indications for oncological, orthopaedic, or chronic renal diseases was not assessed in this study.


Oral iron seems to be a valid, simple, and effective treatment approach in women with severe postpartum anaemia. The limited cost of oral iron tablets supports the fact that this therapy is used as the first choice in women with severe postpartum anaemia without clinical indication for blood transfusion; however, intravenous therapy may be a good option in cases of oral iron intolerance or presumed poor adherence to oral therapy.

We acknowledge that the results of our study do not necessarily apply to the treatment of postpartum anaemia with other iron components, such as ferric carboxymaltose or recombinant human erythropoietin.[4, 22-24] Carboxymaltose was introduced after our study was designed and can be administered in larger doses than ferrous sucrose; however, both carboxymaltose and erythropoietin are more expensive, and are not widely used in hospitals.


The addition of intravenous ferrous sucrose in severe postpartum anaemia does not improve the recovery of haemoglobin and haematocrit, compared with women taking conventional oral iron. Although ferritin stores were increased at 1 and 2 weeks, and a higher increase in reticulocyte levels was observed at week 1 in the group treated with ferrous sucrose, compared with women receiving placebo, the levels of haemoglobin and haematocrit at 1, 2, and 6 weeks were similar in both groups. However, intravenous therapy may be considered in women with poor tolerance or poor adherence to oral therapy, as it ensures the adequate replenishment of iron stores.

Disclosure of interests

All authors warrant that they have no financial affiliation or involvement within any commercial organisation with a potential financial interest in the subject or materials discussed in this article.

Contribution to authorship

M.F.P. contributed in carrying out, analysing, and writing up the work. J.L.C. contributed to the conception, planning, and carrying out of the work. N.M. contributed to the conception, planning, and carrying out of the work. J.E. contributed to the writing up and revision of the work. M.P. contributed to the conception, planning, carrying out, analyzing, writing up, and revision of the work.

Details of ethics approval

The study was approved by the Ethical Committee of the Hospital Clinic on 8 June 2005. Clinical trial registration:, NCT00660933.


This study was partially supported by J Uriach & Co. Investigators had full access to all data and take responsibility for the integrity of the data and the accuracy of the data analysis.