Healthcare use and costs of perinatal anxiety: a UK NHS perspective | BMC Health Services Research

Setting and perspective

MAP cohort sample characteristics

The MAP cohort was a systematic sample of 2,243 pregnant women recruited in 29 NHS sites from 12 NHS Trusts in England and 5 Health Boards in Scotland. Recruitment was conducted from November 2020 to November 2021. Women were eligible for inclusion in the MAP study if they met the following inclusion criteria: (i) aged 16 years or over; (ii) less than 15 weeks pregnant at the time of recruitment to enable longitudinal follow-up of mental health and healthcare use across the perinatal period; (iii) able to provide written informed consent to take part in the study, and (iv) with a level of English sufficient to understand and complete questionnaires in lay language. Both singleton and multiple pregnancies were eligible for inclusion.

The sample was recruited from Scotland (10%) and England (90%) to reflect the relative populations and number of births in the two nations. The cohort is representative and diverse: the sample in Scotland is representative of the general population [11], whereas the sample in England has greater diversity with a greater proportion of participants from Asian, Black African/Caribbean, mixed ethnicity or other non-white groups compared to the 2021 Census [11]. The MAP study included measures of anxiety, depression and general psychological distress in early pregnancy (mean = 11.4 weeks, SD = 2.0), mid-pregnancy (mean = 23.0 weeks, SD = 1.3), late pregnancy (mean = 31.9 weeks, SD = 1.2), and postpartum (mean = 7.9 weeks, SD = 2.4).

MAP ALLIANCE sample characteristics

One thousand nine hundred sixty-four women from the MAP study were eligible to participate in the MAP ALLIANCE study (1,742 from England and 222 from Scotland). All the women had indicated that they could be contacted for follow-up research. Recruitment was carried out between June 2022 and January 2023. Ethical approval required women to re-consent to take part in the MAP ALLIANCE study. Overall, 794 women consented to participate in the MAP ALLIANCE study. During timepoint one (0–6 months) and timepoint two (6–12 months), 1.2% (10/794) of participants withdrew and 0.5% (4/794) of participants discontinued. 722 women consented at six months and an additional 72 women consented at twelve months. Demographic and health variables such as age, ethnic group and experience of mental health problems were collected from participants at the early pregnancy time point as part of the MAP study.

Perinatal anxiety assessment

The number of women with perinatal anxiety, six weeks postpartum, was determined by those who scored 9 or above on the Stirling Antenatal Anxiety Scale (SAAS) 10-item screening tool [12, 13]. The SAAS outcome tool was found to be an accurate diagnostic measure with high sensitivity to measure anxiety in a perinatal population. This measure has been found to be clinically relevant and psychometrically valid, with good diagnostic accuracy [13]. Scores were imputed using the Traj-mean Single imputation approach for participants with missing SAAS scores [14].

Health resource use

Health service resource use was measured using items from an adapted Client Service Receipt Inventory (CSRI) questionnaire from birth to six months and then six to twelve months postpartum [15, 16]. Items from the CSRI were used to measure the health resource use of primary and secondary care for mothers and babies. Primary care resource use is comprised of contacts with the General Practitioner (GP), health visitor and community midwife. Secondary care resource use consisted of inpatient hospital stays, hospital services, and contact with other healthcare professionals (e.g. physiotherapists), as self-reported by participants. Participants with missing data on resource use were not imputed.

Wider societal impact

The frequency of intention to return to work for those who did and did not return to work is presented to identify the wider societal impact of perinatal anxiety.

Cost of illness analysis

COI analysis is considered an essential healthcare evaluation technique to measure and compare the economic burden of diseases to society. A COI analysis can provide policymakers, healthcare providers, and researchers with a comprehensive understanding of the financial impact of a disease. By assessing the economic burden, decision-makers can make informed choices regarding resource allocation, healthcare planning, and policy development [17, 18].

COI analysis was conducted using a bottom-up approach in which the cost of services was based on the resource consumption of individual participants [19]. Due to the availability of a rich dataset, the bottom-up method, as opposed to the top-down approach, has proven to provide a more accurate capture of the health resources used by women with and without perinatal anxiety [20].

Although the MAP ALLIANCE study is a longitudinal follow-up of the MAP study, the COI component of the MAP ALLIANCE study was performed using a retrospective approach in which recorded data of self-reported health resource use was analysed. The COI analysis identified the different components of costs and the size of the contribution of each health resource and quantified the direct costs incurred by the National Health Service (NHS) due to perinatal anxiety. Resources used were multiplied by their unit costs in 2021/22 British pound sterling (GBP). Unit costs were collated from the Personal Social Service Research Unit (cost year 2022) and the National Cost Collection for the NHS (cost year 2021) [21, 22]. This mean cost was adjusted to the total number of participants receiving each service. For inpatient hospital stays, the reason for the stay was matched with the Health Resource Group description. An excess bed day charge was applied if the stay was longer than the trim point (calculated as the upper quartile length of stay for that Health Resource Group plus 1.5 times the inter-quartile range of length of stay) [23]. Due to a high range of costs attributed to the complexities of labour and delivery, the hospital inpatient labour cost has been excluded from the inpatient cost analysis. For hospital services and other healthcare professional visits, participants provided answers in free texts. The unit cost detail for health resource use is provided in Supplementary File 1.

The mean costs of each primary and secondary care service use are presented to the nearest pound sterling (£). Mean cost per patient at six months and twelve months postpartum are reported with 95% confidence intervals (CI) estimated using nonparametric bootstrap sampling. Five thousand samples were taken for each CI and bias corrected and accelerated confidence intervals were calculated. A nonparametric independent Mann–Whitney U test was performed to compare the difference between groups for the total healthcare costs between women with and without perinatal anxiety at twelve months postpartum. A p-value of less than 0.05 was set as statistical significance. A chi-square was performed to compare the difference in highest education level, parity, and previously experienced psychological/mental health problems. For marital status, ethnicity, parity (number of births after 20 weeks of gestation), and previous miscarriage or stillbirth, a Fisher’s Exact statistical test was used.

The frequency and distribution of primary and secondary service use are reported. All significant costs impacting health resource use were examined. Linear regression analysis was conducted to investigate significant predictors of costs [24]. For example, age, hospitalisation, ethnicity, Stirling Antenatal Anxiety Scale (SAAS) score and parity at both time points.

Subgroup analysis

Ethnicity, parity, regional variation and adverse events such as miscarriage and stillbirth were examined for subgroup analysis to compare the difference within the mean of total healthcare resource use costs at six and twelve months. A large proportion of the sample came from England, and the regions were grouped as follows: North England, South England, London, Midlands and Scotland.

Sensitivity analysis

Sensitivity analysis was conducted for twelve-month data that included an additional 72 women who only participated in twelve-month data collection. The interquartile range (IQR) was used to identify the outliers, which were then removed to explore the uncertainty of the result with and without outliers. Any costs that were less than Quartile 1-(1.5*IQR) or more than Quartile 3 + (1.5*IQR) were identified as outliers.