Study IV

In Study IV we had access to a larger cohort, Stockholm Gotland Obstetric Cohort 2008-2020. Still, as previously described, selection bias is a challenge in studies on labor duration and we found it of importance to move beyond the previous

challenges in defining the start of the active phase and include more women with different patterns of progression. To be able to also include women where notation of start of active phase was not available from the data, we used the definition of onset of first stage at 5 cm and applied single imputation. To address missing data on cervical dilation of 5 cm (start of the exposure) single imputation models were performed on women with informed data on a minimum of two notations of cervical dilation during first stage of labor. As shown in Figure 14, ten different imputation patterns were generated to define start of the exposure, i.e. when each woman with missing observed data in the partograph on 5 cm was estimated to have been dilated 5 cm, this was done to reduce selection bias and improve precision. Weights used for the estimations was calculated upon previous knowledge of the population’s traverse time to progress from one centimeter to full dilation (72).

Figure 14. Details for imputation hierarchy method used for imputation of start of the exposure. Blue indicating known notation, grey indicating unknown notation, green indicating known/unknown information not used for imputation.

To test the imputation technique before we decided upon the study population for the study, we also performed imputation with 13 different patterns, where women with the only known notations of cervical dilation after 6 cm also was included in the cohort. As expected it increased the cohort size, the weights and corresponding data were however found to be less valid for this cohort. Estimating a timepoint for start of exposure, informed by data on cervical dilation after 6 cm used in this cohort was less valid compared to both the ten patterns imputation and complete case analysis. This is shown in detail in Figure 15, and not included in the manuscript for Study IV.

Figure 15. Details for imputation hierarchy method used for imputation of start of the exposure, using 13 different imputation patterns. Blue indicating known notation, grey indicating unknown notation, green indicating known/unknown information not used for imputation. This figure is constructed for presentation in this thesis and is not presented in manuscript for Study IV.

The aim of the imputation is to expand beyond sensitivity analysis to reduce systematic or random error in the study design. In short, the advantages of using imputation technics are these;

✓ Analysis in imputed cohort includes also data/deliveries with both slow and faster pace, more representive for the Target population identified for this study

✓ The possibility to check for robustness of the results by comparing

In this study the outcome of interest was adverse neonatal outcomes and

categorization of them involved considerations by the expected frequency of the outcome, sample size and prior research and clinical knowledge of expected long-term consequences. Owing to a large and unique sample size we were able to be stringent when creating the composite outcomes. A composite of severe outcomes investigated were diagnoses with a high risk of death and/or major

neurodevelopmental impairments, including cerebral palsy, cognitive impairment, visual impairment or hearing impairment. A composite of moderate outcomes was diagnoses with a low risk of death and/or major impairments and are less likely to have long term consequences for the infant under the assumption that adequate treatment is given. As previously mentioned, observational studies generally focus on clinically meaningful outcomes, here we found these diagnoses using coding systems as ICD-9 or ICD-10, valid to capture neonatal adverse outcomes of interest for increasing labor duration. Detailed information about the rationale for each outcome is provided below, with references of validity in register data for the outcomes examined (194-209).

Severe

✓ Intrapartum fetal death: Severe since it involves death of the neonate during labor or close after birth, related to labor

✓ Hypothermia treatment: Severe since hypothermia treatment is only given to infants with fulfilled A-criteria, which corresponds to clinical signs of pronounced asphyxia at birth, and these neonates have a high risk of future major neurodevelopmental impairments (NDI).

✓ Birth asphyxia related complications: Severe for infants with pronounced asphyxia, since it is well known that they have a risk of NDI. (P.21.0) Severe for infants with moderate to severe HIE indicate an impact of the brains that increases the risk of NDI. Therapeutic hypothermia reduces this risk but does not remove it

✓ Neonatal seizures: Seizures are the most frequent manifestation of NDI disorder in the new-born period, of course the risk of NDI is depending on the etiology behind the seizure, but at large there is an increased risk of NDI and death and therefore it is appropriate to classify it as severe outcome

✓ Chest compressions at birth: Severe since it is a sign of severe birth asphyxia.

✓ Intubation: Sign of or cause to severe birth asphyxia

✓ APGAR at 5, <4: A marker for severe birth asphyxia

✓ Umbilical-artery acidosis Ph<7.00, and/or Base deficit of ≥16: Severe since it is a sign of severe birth asphyxia

✓ Intracranial bleeds: Deep intracerebral bleeding, with a risk of NDI, implies haemorrhagic stroke which has a high risk of NDI

✓ Intracranial stroke: High risk of NDI Moderate

✓ Neonatal sepsis: Early diagnosed sepsis, within 72 hours after birth, since we are aiming to investigate effects of labor. This is a less severe

condition in a Swedish setting, since sepsis is early detected and

adequate treatment is given with wide inclusions and the risk for death and NDI is therefore low in term neonates

✓ Quickly recovering indications of asphyxia: Moderate since the infants recover quickly and the risk of mortality/NDI is not high, even though some recent studies have shown that also a slight decrease in Apgar leads to an increased risk of cerebral palsy and epilepsy.

✓ Quickly recovering HIE: Moderate since these children were considered to become healthy in the majority of cases. Some studies have shown that also these children (on a population level) have slightly lower cognitive score on Bayley’s testing at 2 years compared to healthy children without a HIE diagnose as infant, however they do not seem to have a proven increased risk of death o major NDI, and the absolute risk must be considered very low

✓ Pneumothorax: Pneumothorax may lead to acute morbidity and risk of death but in Sweden you would expect that the children get adequate treatment and then the risk of death/NDI should be low.

✓ Quickly recovering respiratory disturbances: Moderate since these

conditions have a low risk for death/NDI as adequate treatment are given

✓ APGAR at 5 ≥4-<7: Moderate since most of these neonates, particularly

severe), however for meconium aspiration in general, without any other condition diagnosed as severe this categorization is adequate