Metabolic acidosis at birth and suboptimal care – illustration of the gap between knowledge and clinical practice


M Jonsson, Department of Women’s and Children’s Health, Uppsala University, SE-751 85 Uppsala, Sweden.


Objective  To evaluate the occurrence and nature of suboptimal intrapartum care in cases with metabolic acidosis in the newborn, and to estimate the degree to which this may be prevented.

Design  Case–control study. Clinical audit.

Setting  Delivery units at two university hospitals in Sweden.

Population  Out of 28 486 deliveries, 161 neonates ≥34 weeks of gestational age were born with metabolic acidosis.

Methods  Cases (n = 161): umbilical artery pH < 7.05 and base deficit ≥12 mmol/l. Controls (n = 322): pH ≥ 7.05 and Apgar score ≥7 at 5 minutes. Obstetric characteristics and oxytocin administration were recorded. The last 2 hours of electronic fetal monitoring before delivery were evaluated blinded to outcome. Intrapartum management was analysed for suboptimal care by using predefined criteria.

Main outcome measure  Suboptimal intrapartum care.

Results  Case and control comparisons displayed an occurrence of suboptimal care in 49.1% versus 13.0% (P < 0.001), oxytocin misuse in 46.6% versus 13.0% (P < 0.001), a failure to respond to a pathological cardiotocographic pattern in 19.9% versus 1.2% (P < 0.001) and suboptimal care related to vacuum deliveries in 3.1% versus 0.3% (P < 0.01) respectively.

Conclusion  Metabolic acidosis at birth is often associated with suboptimal intrapartum care. The high rate of suboptimal care with regard to oxytocin use and fetal surveillance illustrate a gap between guidelines and clinical practice. Metabolic acidosis and related neonatal morbidity could potentially be prevented in 40–50% of cases. The adherence to guidelines must be checked.


The degree of acidemia at birth is a potentially useful variable for quality control of obstetric management.1 An umbilical artery pH value <7.05 identifies fetuses adjusting to hypoxia, and is associated with neonatal complications, but a statistically significant increase in the incidence of serious neonatal morbidity is not seen until the umbilical artery pH level is <7.00.2,3 The risk of complications such as neurological injuries increases when gas exchange during labour is impaired enough to cause metabolic acidosis (indicating asphyxia) with the cut-off level of base deficit ≥12 mmol/l.4–7 For this reason it is important to avoid metabolic acidosis at birth.

The majority of neurological injuries predate labour, but labour is a time of risk for asphyxia and asphyxia-related morbidity and mortality. The extent to which intrapartum asphyxia may be prevented by improved care is controversial, but several investigations and malpractice claims analyses suggest that a significant proportion is preventable.8–12 In a previous study, we found that a hyperactive uterine contraction pattern, often combined with oxytocin use, was the most important risk factor for acidemia at birth.13 Cardiotocographic tracings were also abnormal to a high degree. This finding indicated suboptimal intrapartum care and prompted us to investigate the issue in greater detail, as these risk factors are modifiable, and neonatal morbidity because of metabolic acidosis thus possibly preventable.

The aim of the present study was to determine the frequency of occurrence of suboptimal intrapartum care, and to describe the nature of such care during the last 2 hours before delivery in cases of an infant with metabolic acidosis at birth. By reviewing deliveries with and without metabolic acidosis, the occurrence of suboptimal care, mainly with regard to oxytocin use and fetal surveillance was evaluated. Another aim was to estimate the proportion of cases with metabolic acidosis that possibly could be prevented.

Material and methods

A database from a previous study was used, in which maternal and neonatal information on all deliveries with singleton newborns ≥34 weeks’ gestational age, with pH < 7.05 in the umbilical artery at delivery with two controls each, was collected during the period of 1994–2004 from the delivery departments of Örebro and Uppsala University Hospitals in Sweden.13 Newborns delivered by elective caesarean section and cases with obstetric catastrophes (placental abruption, cord prolapse and eclampsia) were not included in the database. At the centres, umbilical artery blood gases are obtained in all deliveries.

The database was searched for cases with metabolic acidosis at birth and for their controls. Metabolic acidosis was defined as umbilical artery pH < 7.05 and base deficit ≥12 mmol/l. Controls consisted of the first two newborns delivered after each study case with the same parity as the index case, a cord artery pH ≥ 7.05 and a Apgar score ≥7 at 5 minutes.

Information on mode of delivery, duration of bearing-down efforts, Apgar scores, admission to the neonatal intensive care unit (NICU) and diagnosis of hypoxic ischaemic encephalopathy (HIE), grade II–III14 was retrieved from the database.

Information about oxytocin administration, maximum infusion rates and duration of treatment was included. The database contained interpretations of cardiotocographic (CTG) tracings covering the last 2 hours before delivery. The interpretations had previously been performed without knowledge of clinical data or neonatal outcome,13 by the authors M.J and S.N.L, according to the International Federation of Gynaecology and Obstetrics (FIGO) classification.15 A normal CTG was categorised as reassuring, and this is characterised by: a normal baseline frequency [110–150 beats per minute (bpm)], normal variability (5–25 bpm), accelerations and no decelerations.15 A suspicious or pathological CTG was categorised as non-reassuring. The FIGO categorisation defines a suspicious CTG as a baseline 100–110 bpm or 150–170 bpm, baseline variability 5–10 bpm for >40 minutes or >25 bpm and variable decelerations. A pathological CTG is defined as a baseline <100 bpm or >170 bpm, baseline variability <5 bpm for >40 minutes, severe variable decelerations, prolonged decelerations, late decelerations and a sinusoidal pattern. At least 20 minutes duration of a CTG tracing had to be available to enable an assessment. A hyperactive contraction pattern was defined as six or more contractions/10 minutes for at least 20 minutes. The delivery wards hold protocols for CTG interpretation and guidelines on what action to be taken in case of abnormality.

Additional information from delivery records was retrieved and entered into the database, and is described below. Admission testing is routine and the assessment (reassuring or non-reassuring) of the admission CTG done by the attending midwife was registered. Duration of labour (first and second stage) was retrieved from the partograms. Information on vacuum deliveries regarding indication, duration, number of pulls and detachment was collected. Indications for caesarean section and time from decision-to-delivery were obtained.

As the oxytocin infusion was started, an assessment of the CTG was done (by author M.J.).

Protocol for oxytocin use was available, with guidelines on how to administer and monitor the effect of oxytocin infusion and how to avoid hyperstimulation.

Cases and controls were audited to evaluate the occurrence of suboptimal care in the last 2 hours before delivery. The audit was performed by the author M.J. with the same protocol for cases and controls and not blinded to groups. Study criteria for suboptimal care were predefined and cases and controls were assessed with respect to the categories described below. The criteria for suboptimal care used in this study can be regarded as the threshold below which care is, by consensus, not acceptable to most practising obstetricians.

CTG patterns and suboptimal care

This category concerns the assessments of CTG recordings made during the last 2 hours before delivery and the CTG assessments made at the time when an oxytocin infusion was started. Care was considered suboptimal if there was failure to respond within the specified time limit in response to a pathological CTG pattern of ≥40 minutes’ duration (with the exception of bradycardia), by:

  • • decreasing or stopping an oxytocin infusion,
  • • taking a fetal scalp blood sample or by delivery.

In the literature, there is no specific time limit for action in case of a pathological CTG pattern during the second stage to be found. We chose ≥40 minutes since this duration of pathological CTG gave marked neonatal morbidity (Apgar score <7 at 5 minutes and NICU treatment >2 days) compared with <40 minutes in the present study. In addition, according to the FIGO categorisation a decreased variability <5 bpm for >40 minutes defines a pathological pattern.

Care was also considered suboptimal if there was failure to perform a CTG despite indications, or poor quality tracings were obtained. Indications for continuous CTG registration during labour were: induction of labour, non-reassuring admission test, hypertensive disorders, oligohydramnios, gestational age ≥42 weeks, prolonged rupture of membranes, meconium stained liquor, abnormal bleeding, intrapartum fever and oxytocin administration.

Oxytocin administration and suboptimal care

Care was considered suboptimal if:

  • • oxytocin was administered without indication,
  • • an oxytocin infusion was started or increased despite a pathological CTG pattern or was started despite uterine hyperactivity,
  • • short increment intervals were applied (<15 minutes),
  • • there was uterine hyperstimulation (≥6 contractions/10 minutes >20 minutes),
  • • a CTG tracing was not used continuously,
  • • there was no or a low quality tracing of uterine contractions.

To evaluate whether there was an indication for oxytocin treatment, partograms were scrutinised and protraction or arrest disorders were diagnosed if the following criteria were met.16,17 Protraction disorder in the first stage of labour: cervical dilatation <1 cm/hour. Second stage delayed if >2 hours (epidural 3 hours) and descent <1 cm/hour in the nulliparous and if >1 hour (epidural 2 hours) and descent <2 cm/hour, in the parous. Arrest disorder: complete cessation of progress ≥2 hours. A combined disorder was defined as protraction disorder. Care was identified as suboptimal if the patient did not fulfil the criteria (no indication) but was treated with oxytocin. The criteria used for protracted labour were chosen to, from the audit perspective, avoid overestimation of cases without indication for oxytocin treatment.

Vacuum delivery and suboptimal care

Vacuum extractions exceeding 20 minutes, more than seven pulls, or with more than two detachments were considered to be suboptimal care.18,19

Caesarean section and suboptimal care

Suboptimal care was identified if a caesarean section for suspected fetal distress was not started within 30 minutes from decision.

Precipitate labour was diagnosed if the first and second stages of labour were completed in 3 hours or less and both stages (from 3–4 cm of cervical dilatation) were displayed on the partogram i.e., the diagnosis was not given if duration of labour was unknown.20,21 After the systematic review of the events during labour and delivery, care was designated as optimal or suboptimal.

The Statistical Package for Social Sciences (SPSS) for Windows, version 15.0, was used (SPSS, Inc., Chicago, IL, USA). To compare group distributions the chi-squared test or Fisher’s exact test was applied and the Mann–Whitney U test or t-test for continuous variables. A P-value 0.05 was considered to indicate a significant difference. The study was approved by the Research Ethics Committee at the University of Uppsala.


There were a total of 28 486 deliveries at the two units during the 10-year period. Umbilical artery blood gas samples were obtained in 83% of deliveries. A total of 161 neonates displayed a metabolic acidosis at birth (0.68%) born to 101(63%) nulliparous and 60 (37%) parous women.

The majority of neonates born with metabolic acidosis had an Apgar score ≥7 at 5 minutes [124/161 (77.0%)]. The mean umbilical artery pH ± standard deviation (SD) was 6.96 ± 0.07 in the case group compared with 7.26 ± 0.09 in the control group, and the base deficit was 15.7 ± 2.9 mmol/l compared with 4.5 ± 3.1 mmol/l respectively. An umbilical artery pH < 7.00 was found in 91 (0.38%) neonates and of these 14 neonates were diagnosed with hypoxic ischaemic encephalopathy (grade II, n = 8 and grade III, n = 6). There were no neonates with encephalopathy among those with pH ≥ 7.00. Care was assessed as suboptimal in 79 (49.1%) cases and 42 (13.0%) controls (P < 0.001).

Duration of labour was longer in the case group but no difference in diagnosis of protraction or arrest disorders was found (Table 1). Oxytocin infusion during the last 2 hours of labour was more common among cases compared with controls, whereas total duration of treatment and maximum infusion rates of oxytocin administration did not differ (Table 1). Hyperactive labour occurred in 42 (26.1%) of cases versus 22 (6.8%) of controls (P < 0.001). In cases with hyperactive labour 71.4% (30/42) were treated with oxytocin.

Table 1.   Intrapartum variables in the case and control group
 Cases (n = 161)Controls (n = 322)P value
  1. Values are given as n (%), median (quartiles) or mean ± standard deviation (SD).

  2. Chi-square test or Mann–Whitney U test.

  3. *Vaginal delivery.

Spontaneous onset of labour134 (83.2)283 (87.9)ns
Reassuring admission test140 (86.9)303 (94.1)0.001
Duration of labour, total time (hours)*6.3 (3.7–9.8)5 (3.1–7.6)<0.05
First stage, h*4.9 (2.5–8.0)4 (2.5–6.6)<0.05
Second stage, min*60 (30–105)45 (19–90)<0.01
Bearing down effort, min* (mean ± SD)60 (±52)43 (±39)<0.001
Precipitate labour*28 (17.4)33 (10.2)<0.05
Protraction or arrest disorder48 (29.8)105 (32.6)ns
Oxytocin administration in last 2 hours before delivery99 (61.5)135 (41.9)<0.001
Duration of oxytocin treatment, min (total time)130 (65–328)195 (68–360)ns
Maximum infusion rate mU/min12 (7–20)15 (8–20)ns
Contractions ≥ 6/10 minutes42 (26.1)22 (6.8)<0.001
Contractions ≥ 6/10 minutes and oxytocin treatment30 (18.6)10 (3.1)<0.001
Caesarean section24 (14.9)22 (6.8)<0.001
Vacuum extraction44 (27.3)26 (8.1)<0.001

Tables 2 and 3 shows the nature of suboptimal care with regard to the use of oxytocin and pathological CTG patterns during the last 2 hours before delivery. In a sub analysis of those women having oxytocin stimulation, care was considered suboptimal in 75/99 (75.8%) cases and in 42/135 (31.1%) controls. The most prominent finding was the difference in increase of an oxytocin infusion despite a pathological CTG ≥ 40 minutes, which was found in 44/99 (44.4%) compared with 8/135 (5.9%), and overstimulation of labour despite a pathological CTG ≥ 40 min which was found in 15/47 (31.9%) compared with 1/27 (3.7%) cases and controls respectively.

Table 2.   Suboptimal care with regard to oxytocin use in the case and control group
 Cases, n = 161, n (%)Controls, n = 322, n (%)P value
  1. There is overlap within groups.

Oxytocin administration in last 2 hours before delivery99 (61.5)135 (41.9)<0.001
Injudicious use of oxytocin75 (46.6)42 (13.0)<0.001
No indication was found41 (25.4)35 (10.8)<0.05
Infusion started despite a pathological CTG31 (19.3)5 (1.6)<0.001
Infusion rate was increased despite a pathological CTG ≥ 40 min44 (27.3)8 (2.5)<0.001
Short increment interval (<15 min)31 (19.3)12 (3.7)<0.001
Contractions ≥ 6/10 min and oxytocin treatment 30 (18.6)10 (3.1)<0.001
Precipitate labour6 (3.7)4 (1.2)ns
Table 3.   Occurrence of pathological CTG patterns during the last 2 hours before delivery among cases and controls
 Cases, n = 161, n (%)Controls, n = 322, n (%)P value
  1. Types of suboptimal care that were identified, overlap within groups.

CTG registration missing17 (10.6)52 (16.1)ns
Pathological CTG pattern109 (67.7)71 (22.1)<0.001
Pathological CTG pattern ≥ 40 min 65 (40.4)37 (11.5)<0.001
No response to pathological CTG pattern ≥ 40 min32 (19.9)4 (1.2)<0.001
Injudicious use of oxytocin and pathological CTG pattern ≥ 40 min39 (24.2)7 (2.2)<0.001
No CTG performed despite indications or poor quality8 (5.0)26 (8.1)ns

Pathological CTG patterns occurred significantly more often in the case group (Table 3). An adequate tocographic registration was not achieved in 35 (21.7%) cases and 48 (14.9%) controls (ns). Of patients receiving oxytocin treatment, CTG tracings were missing in five cases (5.1%) and in 14 (10.4%) of controls (ns) and 27 (11.5%) were not adequately supervised with regard to uterine contraction frequency [16 cases and 11 controls, (ns)].

Newborns were admitted to the NICU more often from the case group: 78 (48.4%), compared with 15 (4.6%) in the control group (P < 0.001). Among the cases, 47 (29.2%) stayed in the NICU for >2 days compared with seven (2.2%) controls (P < 0.001). Neonatal outcome in the case group according to the admission test, is presented in Table 4. Of the cases with reassuring admission test, suboptimal care was assessed in 71 (51.1%), and in the non-reassuring admission test group in eight (40.0%). Cases with a non-reassuring admission test had the highest rate of low Apgar scores, NICU admissions and neonates with encephalopathy (Table 4).

Table 4.   Neonatal outcome according to the admission test in the case group
 Reassuring admission test (n = 140)Non-reassuring admission test (n = 20)P value
  1. Values are given as n (%).

  2. HIE, hypoxic ischaemic encephalopathy; NICU, neonatal intensive care unit.

pH < 7.0 78 (55.7)13 (65.0)ns
Apgar < 7 at 5 minutes26 (18.5)10 (50.0)0.002
Apgar < 4 at 5 minutes6 (4.3)1 (5.0)ns
NICU > 2 days39 (27.8)8 (40.0)ns
HIE8 (5.7)6 (30.0)<0.001
Grade II53 
Grade III33 

Suboptimal care related to vacuum extractions was 5/161 (3.1%) in the case group and 1/322 (0.3%) among controls (P < 0.01). The indication for vacuum delivery was mainly fetal distress among cases (80%), whereas dystocia or maternal exhaustion was the main indication among controls (69%).

When caesarean section was performed because of fetal distress, a delay in decision to delivery occurred twice in the case group and once in the control group. Indications for caesarean section differed, with 17 (70.8%) for suspected fetal distress in the case group compared with three (13.6%) in the control group.


In the present study, suboptimal intrapartum care occurred in half of the cases with metabolic acidosis at birth, while it was less frequent, but not uncommon among neonates without metabolic acidosis. Among cases suboptimal care consisted of injudicious use of oxytocin and a failure of appropriate action upon signs of fetal distress. The high rate of NICU admissions and diagnosis of encephalopathy in the case group confirms that metabolic acidosis should be avoided. The results imply that metabolic acidosis could probably have been avoided in 40–50% of the cases with a reassuring admission test.

The last 2 hours preceding delivery was the main focus of the audit. This period was chosen in order to include the second stage of labour which is considered to be the most critical period to the fetus. It has also been demonstrated that a pathological CTG pattern 1 hour preceding delivery will predict significant acidosis.22 In the case of a caesarean, the assessment for suboptimal care was performed during the last 2 hours before delivery, whether in the first or second stage of labour.

The interpretation of CTG registrations was done blinded to groups, whereas the review regarding optimal or suboptimal care was not. The reviews were done with the same protocol and the differences between groups are unambiguous, nevertheless bias can not be excluded.

Since the aim of the study was to evaluate the occurrence of suboptimal care, the definition of protracted labour (<1 cm/hour) used in this study was chosen to avoid overestimation of cases without indication for oxytocin treatment. Consequently, lack of indication for oxytcoin treatment was probably underestimated. The high rate of oxytocin stimulated labours on delivery wards in Sweden indicates that the definition of <1 cm/hour is probably used in clinical practice.23

Oxytocin is the drug most commonly associated with preventable adverse events during delivery.24,25 In the present study, there was a great difference in oxytocin misuse between cases and controls and probably a significant number of neonates suffered from iatrogenic metabolic acidosis because of injudicious use of oxytocin. Most alarmingly, oxytocin infusions were started and increased despite the presence of a pathological CTG pattern and without confirming fetal wellbeing, and uterine contractions were frequently overstimulated among cases. Probably, oxytocin was given to an already compromised fetus in order to deliver promptly. This action was not so frequent in the control group. The mistakes regarding oxytocin use in clinical practice need to be emphasised, not least because the use of oxytocin is on the increase and it is estimated that about 50% of nulliparous and 20% of parous women receive oxytocin during delivery.23,26 It is possible that the active management of labour concept has had an influence on obstetric care in Sweden during the last two to three decades and resulted in a more aggressive and less selective use of oxytocin.27

In contrast to the present study, there are prospective studies on oxytocin induced or augmented labour that do not show an increased risk of acidosis.28–32 In those studies, strict protocols limited the use of oxytocin and discontinuation or decrease of the infusion was applied in situations of hyperstimulation. The conflicting results compared to the present and other studies13,33 could be explained by a difference in management; if strict guidelines for oxytocin administration with regard to supervision of fetal heart rate patterns and contraction frequency are followed, the risk of acidemia for this reason could be avoided. There was no difference with regard to duration of oxytocin administration or maximal infusion rates between cases and controls and the results of the present study strongly support the idea that it is more important to consider the responses of the uterus and the fetus than the infusion rate or duration of oxytocin administration. It appears that the relationship between oxytocin administration and pathological CTG patterns is frequently overlooked. A pathological CTG pattern occurred in both groups but was more frequent in the case group. The difference in response to abnormal patterns is an important finding of the present study.

Failure to respond to an abnormal CTG pattern was uncommon in the control group, whereas non-responsiveness was frequent in the case group which could be a possible reason for the difference in neonatal outcome. Thus, CTG interpretation and responsiveness to abnormalities is an important area for improvement.

An unexpected finding was that a significant number of cases had precipitate labour. An association with metabolic acidosis has, to our knowledge, not been investigated and our results indicate that precipitate labour is not always favourable and could be a risk factor. An association with low Apgar scores and increased perinatal morbidity and mortality has been reported,21 whereas other investigators conclude that precipitate labour is not associated with adverse perinatal outcome.20,34 Furthermore, it has been suggested that the unfavourable neonatal outcome after precipitate labour may reflect lack of proper management during delivery.34,35 This is supported by the finding in the present study in which oxytocin had been administrated in 16% of subjects with precipitate labour.

Suboptimal fetal heart rate monitoring practice and injudicious use of oxytocin are the persistent findings in obstetric malpractice analysis and observational studies concerning HIE and cerebral palsy.8,10,12,36,37 Such analyses are limited because they are a series of highly selected cases from which it is difficult to generalise. Our study on cases with metabolic acidosis in the neonate at birth and controls from a large population, confirms that incorrect assessment of CTGs and misuse of oxytocin are issues of concern not only in seriously damaged infants.

Overall, suboptimal care related to vacuum extractions and a delay in decision-to-delivery time for caesarean section or vacuum delivery was infrequent in the present study. This result is in agreement with a case–control study by Gaffney et al. who also investigated suboptimal intrapartum care, whereas the results are contradictory to findings in malpractice studies.10,12,38 As discussed above, malpractice analyses are biased towards more serious injuries and these issues do not seem to be applicable in a large population.

The question to what extent metabolic acidosis, neonatal morbidity and encephalopathy could be prevented is difficult to evaluate.11,39,40 A reassuring admission test reflects normal fetal behaviour and absence of hypoxia at the beginning of labour and was found in the great majority of cases and controls.41,42 This group is important because among these the probability of delivering a healthy infant is the greatest, and preventive measures could be successful. Nevertheless, it is obvious from the present study that, in a number of cases with a normal admission test, metabolic acidosis is probably not avoidable despite appropriate intervention.

A high incidence of abnormal admission CTGs has been reported among neonates with encephalopathy and neurologically impaired infants, implying an antenatal event beyond the control of the obstetrician.43–45 In accordance, we found the highest rate of low Apgar scores, NICU stay >2 days and encephalopathy in this group. Failure to respond appropriately to pathological CTG tracings was the most frequent form of suboptimal care in this group. These fetuses may be more vulnerable and continuous labour might aggravate the risk of neonatal morbidity.46 The possibility of preventing neonatal morbidity in this group is difficult to evaluate as the duration of fetal compromise is not known.

The incidence of metabolic acidosis in this study was low (0.68%). In spite of that, we estimate that metabolic acidosis at birth could probably have been avoided in 40–50% of cases, which is in agreement with previous studies on suboptimal care and asphyxia.11,39,40

In the present study, we high lighted that oxytocin misuse together with non-responsiveness to pathological CTGs are common features among cases with metabolic acidosis in the neonate. The risks with oxytocin administration are not unknown and the delivery units evaluated in this study have guidelines on when and how to use oxytocin, as well as guidelines on interpretation of CTG tracings, and how to act in case of abnormality. These guidelines are apparently not always followed. It is unlikely that there is a lack of awareness of guidelines and their contents. There appears to be a gap between knowledge and clinical practice that need to be closed and the existence and availability of guidelines is obviously not enough.

Checklists and standardised protocols for the use of oxytocin have recently been tested and recommended by several authors in order to reduce adverse neonatal outcomes.47,48 In addition to such measures, it needs to be emphasised that adherence to checklists and guidelines regarding oxytocin and interpretation of CTG tracings has to be supervised and may help in the maintenance of knowledge at an organisational and individual level. Routine cord blood gas measurement in all deliveries could be used to assess the result of the management of labour.

Disclosure of interest

The authors report no conflict of interest. The authors alone are responsible for the content and writing of the paper.

Contribution to authorship

M.J. has carried the main responsibility for acquiring the data and drafting this manuscript. All authors have collaborated in the conceptualisation and the design of this study, revised the manuscript and approved the final manuscript.

Details of ethics approval

This study was approved by the Research Ethics Committee at Uppsala, Sweden, in September 2005 (D-number 2005: 212).


The study was financially supported by the County Councils’ Mutual Insurance Company, Sweden.


For statistical support we thank Lisa Wernroth, Uppsala clinical resource, Uppsala University, Uppsala, Sweden.