Qualitative designs
Lim et al. [
19] and Chamberlain et al. [
14] conducted interview-based qualitative research with purposive samples of women (23 and seven participants respectively). Chamberlain et al. [
14] interviewed women who had delivered their babies within two years of the interview, using ‘Framework Analysis’ to construct themes from the qualitative data they recorded. Key findings were themes related to women’s experiences of Asthma and pregnancy, which were largely concerning their experiences of illness and related physical health anxieties; Pregnancy and post-natal experiences, where they related concerns about their asthma and its potential impact on their baby; and Health professionals, in which the women discussed mostly negative experiences with health professionals. The major limitations of this study [
14] concern the extent of women’s symptoms, in that they did not objectively measure asthma symptoms using a valid and reliable scale such as the Asthma Control Questionnaire (ACQ), but instead relied on ‘a confirmed diagnosis of asthma’ as their diagnostic categorisation. A further limitation concerns the interpretation of findings, in that participants in Chamberlain et al’s study [
14] may have had an element of recall based on the length of time since delivery (which is not reported for participants), although it is not clear if that might lead them to over-or under-represent their experiences.
Lim et al. [
19] recruited a purposive sample from one Australian hospital (derived from 179 potential participants) of
n = 23 asthmatic women who were still pregnant, either primi- or multi-gravid, as well as newly delivered within five weeks, and sought to investigate how pregnant women manage their asthma during pregnancy. They conducted in-depth interviews (18 by telephone and 5 face-to-face) with pregnant asthmatic women, which were again coded using the ‘Framework approach’. Five major themes relating to asthma management during pregnancy and changes in behaviour emerged: Risks versus Benefits, Self-Efficacy, Asthma as a Priority, Support and Guidance and Influences on Medication Use. Asthma monitoring during pregnancy was seen as a low priority for some women and their doctors. Communication between pregnant women and health professionals regarding asthma management was poor.
Arguably Lim et al. [
19] were not comparing the perspectives of a similar enough sample by recruiting from across the course of pregnancy and delivery, however they were accurate in their assessment of symptoms, using the ACQ to assess their participants’ symptoms. The largest potential methodological weakness of Lim et al. [
19] is the lack of explicit discussion of rigour and reflexivity in their account, although this is a comparatively minor point of critical analysis in an otherwise robust qualitative design.
Despite these differences in recruitment and classification, both qualitative studies produced similar themes and, in some areas, overlapping findings. It is striking that participants in both studies talked about their fears explicitly and this centred on medications, particularly steroids. In both studies, the women reported worry about risks versus benefits to their health and that of their babies of taking their medications, with some women in Lim et al’s study [
19] unilaterally reducing their medications without reference to health care professionals. Participants in both studies talked about their desire for self-management, but most were unaware that their asthma might worsen in pregnancy and were unprepared for that. Most wanted ‘more’ education with little focus on community sector health care personnel including midwives and general practitioners. Lim et al’s study [
19] indicated that women used the internet for information, but this was not discussed by Chamberlain et al’s [
14] participants. As Lim et al. [
19] was conducted in Australia and Chamberlain et al. [
14] was conducted in England, these two papers indicate that these issues are not restricted to one location.
Quantitative designs
Two of the quantitative papers [
22,
23] were quasi-experimental designs with pre-and post-tests applied to non-randomised groups of pregnant asthmatic women.
Murphy et al. [
22] recruited 211 participants with diagnosed asthma who were also pregnant, who were then assessed by a specialist nurse at 20 weeks’ gestation and then followed up at 33 weeks’ gestation by the same nurse. As part of these assessments, an information session consisting of history taking, self-management skills, medications adherence and knowledge, and further education was given to participants, contributing to a written action plan. Peak Flow (PEF) self-measurement and Forced Expiratory Volume at one second (FEV1) were recorded and urgent medical review was secured where necessary. Murphy et al. [
22] report that at first visit, the women had poor self-management skills, 40% non-adherence to inhaled corticosteroids and little understanding of medications. Only 3% were taking their own PEF. After the session with the specialist nurse, all of these metrics were substantially improved with women who remained in the study. Specifically, those who received a written asthma action plan (WAAP) had babies with statistically significantly higher birth weight than those who did not.
The major strength of a quasi-experimental study design with pre-and post-tests lies in its pragmatism as it explores important concepts in real-world situations. Limitations in Murphy et al’s [
22] study design include that pre-post-test designs do not eliminate Hawthorne effects and are potentially open to investigator bias via recruitment and selection. Also, there is likely to be a significant maturation effect (where test scores improve as a result of time passing and participants developing in some way) which is a potential source of threat to internal validity, casting doubt over the credibility of such study designs as evidence for efficacy of treatment effects [
24,
25]. and despite the issues with Hawthorne effects, potential sources of bias and threats to internal validity [
24,
25] noted above, it is clear that their intervention, involving contact and care from a specialist nurse, made clinically important differences to their women and babies, and that this was particularly the case for those most severely asthmatic during their pregnancy.
In a similar design to that of Murphy et al. [
22], Grezeskowiak et al. [
23] recruited 169 pregnant women with diagnosed asthma. These women were assessed at four points during their pregnancy by a midwife with additional respiratory training for demographic variables including socio-economic status as well as asthma-related ones including medications, control of symptoms (using the Asthma Control Questionnaire, ACQ), use of corticosteroids and hospital admissions. FEV1 was recorded. Half way through the study a new antenatal asthma service was introduced which was led by a respiratory nurse, meaning that the women received sessions on managing asthma in pregnancy based on evidence-based guidelines. Outcomes were assessed for the 89 women who received standard care and to 80 who attended the new service. Similar to Murphy et al’s study [
23], Grezeskowiak et al. [
23] found that those who attended the new antenatal asthma and pregnancy service, involving contact and care from a specialist nurse, were statistically significantly less likely to have a loss of asthma control or persistent symptoms during their pregnancy. There was an increase in smoking cessation amongst those in the intervention group compared to the control group, as the asthma management nurses were trained to provide advice on how to stop smoking. Perinatal outcomes were the same but there was a non-statistically significant reduction in pre-term births and Caesarean sections in those attending the new service. Indeed, several variables were non-statistically significant; for example, there was no difference in the reduction of exacerbations in moderate/severe asthma between the two groups, indicating that, as well as the criticisms of quasi-experimental pre-post-tests designs discussed above [
24,
25], Grezeskowiak et al’s [
23] study was underpowered and the authors acknowledge this. Study power is important because it is difficult to assess whether the non-significant results are genuine or are type II errors (false negatives) resulting from small sample size [
25], which is a potential threat to the external validity of this study.
Despite the limitations noted in both the quasi-experimental designs [
22,
23] Grezeskowiak et al’s [
23] results support those of Murphy et al. [
22], indicating that targeted intervention involving contact and care from a specialist nurse can make a difference in clinical outcomes. Grezeskowiak et al. [
23] speculate that the asthma management nurses provided a continuity of care which is particularly valuable for women at the lower end of the socio-economic scale, as these women are generally more likely to fare badly in many pregnancy outcomes including vaginal delivery, pre-term delivery and miscarriage.
Lim et al. [
26] conducted the only randomised controlled trial retrieved in our systematic review, which included pregnant women up to 20 weeks’ gestation who had asthma symptoms within the last 12 months. Participants were assessed for severity of symptoms and then randomised into mild to intermittent or moderate to severe asthma symptom groups, and within each group block randomisation was applied to allocate participants to either usual care (control) group or multidisciplinary care (intervention) group. The intervention was a Multidisciplinary Approach to the Management of Maternal Asthma (MAMMA©) and included education, monitoring, feedback and follow-up on a monthly basis, including assessment using FEV1 and the ACQ. The intervention consisted of regular contact, advice and medication review by a pharmacist, who referred women with worsening asthma to a family physician for further review and support. Comparison of the ACQ scores at 3 and 6 months were the primary variables, and secondary variables were asthma exacerbations, hospital admissions, time off work, oral steroid use and neonatal outcomes (the latter obtained from obstetric notes). An 80% power to detect a ACQ score of >0.5 would be achieved with 29 participants. Lim et al. [
26] recruited 60 in total and found that 70% of participants did not realise the importance of good asthma control at the start of the trial and 19 had reduced or stopped their medications because they had become pregnant. In the intervention group, the ACQ score reduced at 3 and 6 months, although it was only statistically significant at 6 months, and all the scores in the intervention group indicated adequate asthma control. Perinatal outcomes were the same for intervention and control groups.
Their trial was adequately powered, and Lim et al. [
26] applied intention to treat analysis (ITT) for drop outs, meaning that missing patients’ data was not omitted from the study, which might otherwise undermine the theoretical basis for statistical analysis and is a source of bias towards favourable outcomes [
27,
28]. As a single-blinded study, one potential criticism is that the Hawthorne effect may bias outcomes, and this has been shown to be a particularly significant issue in studies where participants in the intervention group receive additional follow up compared to those in the control group [
28]. However, in order to conduct research of this nature in the ‘real world’ a pragmatic approach is required [
29], and so it would not have been possible for Lim et al. [
26] to double blind participants concerning their group allocation. Furthermore, data collection and analysis were undertaken by blinded independent researchers, meaning that group allocation was concealed to avoid bias.
Synthesis, hierarchy of evidence and clinical effectiveness
It is clear from the results of this systematic review and narrative summary that a very important factor in securing beneficial outcomes for pregnant asthmatic women is targeted support from a dedicated healthcare professional who is able to help women with management and self-management. Personalised contact featured in all five of the studies reported here: in the two qualitative designs [
14,
19] it was described as important but lacking, whilst the three quantitative designs [
22,
23,
26] gave positive evidence concerning the beneficial impact that could be achieved with targeted input and personal contact. These quantitative designs [
22,
23,
26] are particularly important in this regard as they have quantified a range of clinically and statistically important benefits accruing from dedicated asthma care including improvements in asthma symptoms measured by ACQ scores, pre-term births, birth weight and Caesarean section rates.
However, synthesising and interpreting the results of this systematic review is highly tentative. The purpose of a hierarchy of evidence is to evaluate the strength of evidence in relation to clinical effectiveness [
30]. Accepted hierarchies are diagrammatically pyramidal and graduated with systematic reviews and meta-analyses as the most effective designs, randomised controlled trials (RCTs) second, with other quantitative designs and then qualitative designs as progressively less effective guides for clinical decision making. The five papers reviewed in this study are methodologically heterogeneous. Lim et al. [
26] is an RCT, and so sits highest in traditional hierarchies of evidence, but even so it is single blinded only and a weaker design to RCTs that are double blinded and can more fully control potential biases [
30]. The before and after studies [
22,
23] are weaker designs still, with concerns about internal validity accruing from potential Hawthorne and maturation effects as well as investigator biases [
24]. Lim et al. [
19] and Chamberlain et al. [
14] are qualitative designs that, although robust in that paradigm, are likely to continue to be regarded as poor evidence for clinical effectiveness despite efforts to explore and delineate qualitative designs as evidence for practice [
31].
In summary, the evidence base concerning personalised care for pregnant asthmatic women can best be described as embryonic and further research is required to discover the best strategies for helping them to self-manage their asthma. This makes it difficult to offer firm recommendations for clinical practice as a result of this systematic review and narrative synthesis, however, it seems reasonable to suggest that personalised care and contact with healthcare professionals can have beneficial outcomes, and this has been established in community settings [
32,
33]. This recommendation would rank as ‘weak’ according to GRADE classification [
34] but our ‘weak’ findings complement the ‘strong’ recommendations made in clinical guidelines that women should receive advice about the importance of maintaining good asthma control during pregnancy to avoid problems for mother and baby, and that pregnant women with moderate/severe asthma should be closely monitored to keep their asthma well controlled [
3].
Recommendations for future research
It seems clear from this systematic review that pregnant women with asthma need additional support and care relating to their asthma. The papers reviewed here indicate that contact with a dedicated healthcare professional can facilitate that, and further research is required to examine this area, with larger trials that establish whether or not there is any ‘added value’ in personalised support for pregnant women with asthma, over and above any gains that might be made from different therapeutic regimes and/or asthma control monitoring using Fraction of Exhaled Nitric Oxide, particular as this latter monitoring option is not necessarily diagnostic [
3].
Authoritative guidelines recommend this level of detailed and personalised response from health care professionals [
3], but if resources or practitioner confidence do not permit such close intervention in the community [
35] telehealth solutions might provide a method to combine asynchronous information access and improve self-efficacy [
36]. Although the role of telehealth in asthma has been evaluated in a Cochrane systematic review and found not to result in clinically important improvements in asthma quality of life in mildly asthmatic people, telehealth interventions are useful in those with severe disease at high risk of hospital admission [
36], and also to be effective in reducing anxiety and improving quality of life in patients with Chronic Obstructive Pulmonary Disease [
37]. Furthermore, when specifically targeted at pregnant asthmatic women, telehealth has proved much more successful: where participants were offered personal written asthma action plans, a smart-phone application (app) and a hand-held spirometer, there were statistically significant improvements in asthma quality of life as measured using the ACQ [
38]. This type of service delivery has the advantage of involving participants in self-monitoring and disease regulation, particularly when made gender-specific and acknowledging hormonal variations [
39] as would be the case for pregnant women. Pregnant asthmatic women already use the internet for information [
40] and it is therefore reasonable to assume that a structured approach would be beneficial, but this requires appropriate scientific analysis. One recommendation for further research, therefore, is to examine the role of telehealth in provision of information to pregnant asthmatic women. More specifically, there could be benefits in the provision of resources that are available asynchronously for personal consultation by women at times convenient to them, as an adjunct to WAAPs, rather than relying on timetabled contact during ‘office hours’ with dedicated professionals. One systematic review [
40] found consistent evidence suggesting that asynchronous telehealth lead to shorter waiting times, fewer unnecessary referrals, high patient satisfaction, and no lesser diagnostic accuracy when compared with face-to-face consultations. This asynchronicity has been found to be particularly important in paediatric asthma, where emergency admissions were avoided and satisfaction with the resources was high [
41] and substantial costs savings in remote locations [
42]. Such asynchronous resources could act as an information repository to provide reassurance and guidance about important aspects of asthma care including symptom control and medications, and might help pregnant women to be more confident about when to contact health professionals and then to get more out of their consultations with them. This might be particularly useful as women in pregnancy are uniquely receptive to educational input, already access the internet for information [
40]. Clearly the internet offers a means for social support which is accessible from within the home 24 h a day, and has already been shown to be effective in supporting pregnancy, including in reducing anxiety [
43].