Dhalwani, Nafeesa N.
(2014)
Using primary care data to assess population-level estimates of maternal smoking and nicotine replacement therapy during pregnancy.
PhD thesis, University of Nottingham.
Abstract
Background: Smoking in pregnancy is the most significant preventable cause of poor health outcomes for women and their babies and, therefore, is a major public health concern. In the UK there is a wide range of interventions and support for pregnant women who want to quit. One of these is nicotine replacement therapy (NRT) which has been widely available for retail purchase and prescribing to pregnant women since 2005. However, measures of NRT prescribing in pregnant women are scarce. These measures are vital to assess its usefulness in smoking cessation during pregnancy at a population level. Furthermore, evidence of NRT safety in pregnancy for the mother and child’s health so far is nebulous, with existing studies being small or using retrospectively reported exposures.
Aims and Objectives: The main aim of this work was to assess population-level estimates of maternal smoking and NRT prescribing in pregnancy and the safety of NRT for both the mother and the child in the UK. Currently, the only population-level data on UK maternal smoking are from repeated cross-sectional surveys or routinely collected maternity data during pregnancy or at delivery. These obtain information at one point in time, and there are no population-level data on NRT use available. As a novel approach, therefore, this thesis used the routinely collected primary care data that are currently available for approximately 6% of the UK population and provide longitudinal/prospectively recorded information throughout pregnancy. The specific objectives for this thesis were:
• To assess the quality of smoking data recorded during pregnancy in primary care
• To quantify annual NRT prescribing trends in and around pregnancy and describe the characteristics of mothers prescribed NRT
• To assess the association between NRT and smoking exposure during pregnancy and major congenital anomalies (MCAs), stillbirth, low birth weight and mode of delivery
Methods: All women aged 15-49 years, with pregnancies ending in live or stillbirth, were identified from The Health Improvement Network (THIN) primary care database (2000-2009). Medical Read codes related to smoking status and Multilex smoking cessation drug prescription codes were used to extract data on women’s smoking status and NRT prescriptions. The proportion of pregnancies with a smoking status record was calculated and logistic regression was used to assess how this varied by women’s characteristics. Women were categorised as being smokers or non-smokers during pregnancy based on the recorded Read codes. Where smoking data were missing during pregnancy, smoking status recorded before pregnancy (up to 27 months before pregnancy, ever before pregnancy) was used as a proxy for smoking status during pregnancy. Annual smoking measures from THIN were then compared to other national datasets. Pregnancies ending in early fetal losses were not included for calculating smoking prevalence, as these outcomes can go unrecognised or can be the first recognised sign of pregnancy, making early ascertainment of all pregnancies uncomprehensive; this was also broadly in line with pregnancy ascertainment in the other national datasets. Prescribing prevalence of NRT and patterns of prescribing in terms of frequency, timing and different form of NRT were assessed. Logistic regression was used to assess women’s likelihood of receiving NRT prescriptions by maternal characteristics. Absolute and relative risks (99% Confidence Interval (CI)) for four birth outcomes (MCAs, stillbirth, low birth weight and mode of delivery) were calculated for women prescribed NRT (defined as the NRT group) and women who continued to smoke during pregnancy (defined as smokers) compared to women who did not smoke during pregnancy (defined as non-smokers) with appropriate adjustments for potential confounders. To assess MCAs and birth weight in relation to NRT and smoking a restricted cohort of children was used who had maternal-child linked records in THIN.
Results: There were 277,552 pregnancies in 215,703 women, of which 28% had a gestational smoking status record. In 2000, smoking status was recorded in 9% of pregnancies; 43% in 2009. Smoking estimates from THIN data did not completely agree with estimates from other sources. For example, in 2009 smoking prevalence was 12.9% in THIN, compared to 19.5% in Child Health Systems Programme (CHSP) data. However, the use of smoking data recorded up to 27 months before conception increased the THIN prevalence to 22.9%, which was slightly higher, but compared better with the CHSP estimates.
NRT was prescribed in 4,826 pregnancies for an average duration of 2 weeks (Interquartile range 1-2 weeks), which represented 2% of all pregnancies (11% in smokers). NRT prescribing prevalence before and after pregnancy was half the prevalence during pregnancy. NRT prescribing increased with socioeconomic deprivation (Odds Ratio (OR) =1.33, 95% CI 1.14-1.52) for the most compared to the least deprived group). Prescribing was higher in pregnant smokers with asthma (OR=1.34, 95% CI 1.21-1.50) and mental illness (OR=1.29, 95% CI 1.18-1.43) compared to smokers without these diagnoses.
The absolute risk of MCA was 279/10,000 live births. Compared with non-smokers the adjusted OR for MCA in the NRT group was 1.34 (99% CI 0.94-1.91). No statistically significant increase in the risk of MCA for the NRT group was found when the reference group was changed to smokers (OR=1.35, 99% CI 0.94-1.93).The absolute risk of stillbirth was 4/1000 live and stillbirths. Compared with non-smokers the adjusted OR for stillbirth in the NRT group was 1.19 (99% CI 0.47-3.01). In smokers, the risk of stillbirth increased by 27% compared to non-smokers (OR 1.27, 99% CI 1.01-1.60). The mean birth weight was 3.41kg (standard deviation 0.59) and the absolute risk of low birth weight was 6.4%. Compared to non-smokers, the risk of women having low birth weight babies was 93% higher in the NRT group (OR 1.93, 99% CI 1.48-2.53). However, there was no statistically significant increase in the risk of low birth weight in the NRT group compared with smokers. There was no increased risk of assisted delivery or caesarean section in the NRT group compared to smokers. However the risk of assisted delivery decreased by 25% in the NRT group (Relative Risk Ratio 0.75, 99% CI 0.60-0.93) compared to non-smokers.
Conclusion: The completeness of smoking status recording during pregnancy in primary care data is improving; however, under-recording of smoking status during pregnancy still results in unreliable estimates of the prevalence of smoking in pregnancy and needs improvement. Pre-conception smoking records are reasonably complete and it is possible that low recording in pregnancy is because a woman’s smoking status has not changed or that increased interaction with other health services, such as midwifery, during pregnancy means women are less likely to be asked about their smoking by their primary physician and information on their smoking does not get relayed back to their primary care record. Nevertheless records should be updated in pregnancy to ensure comprehensive health care. NRT was most commonly prescribed in pregnancy for about two weeks, which may not be adequate time for effective smoking cessation. Nevertheless, prescribing was higher during pregnancy compared to the nine months before and after pregnancy, which makes establishing its safety during pregnancy even more crucial. The safety studies in this thesis did not find NRT to be any more harmful than smoking during pregnancy if not beneficial. Considering that smoking in pregnancy remains one of the largest public health problems in the UK, improvements of antenatal and postnatal smoking in primary care may not only help identify women for preventive measures earlier but would be invaluable for safety studies considering the outcomes are rare yet severe.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
