Prevalence and patterns of pre-existing multimorbidity in pregnancy in Northern Ireland: a population-based, retrospective study using linked routinely collected healthcare data.

This population-based, retrospective study of prevalence and patterns of multimorbidity in pregnancy was designed to align with the methods of previous MuM-PreDiCT studies to allow for comparison between UK nations. The Health and Social Care Northern Ireland Business Service Organisation Honest Broker Service (HBS) performed linkage between a number of datasets and provided researcher access to de-identified datasets within a Trusted Research Environment (TRE). Pregnancy episodes were ascertained from the Northern Ireland Regional Maternity Service Database (NIMATS). Individual-level data from other routinely collected healthcare databases were linked on a one-to-one basis via unique Health and Care Number (HCN) to individuals with records in NIMATS. The Patient Administration System (PAS) was used to ascertain admissions to secondary care hospitals and associated dates and ICD-10 codes for primary and secondary diagnoses. The Family Practitioner Services Enhanced Prescribing Database (EPD) was used to ascertain medications prescribed in Primary Care and dispensed in the community, and for each item the date of issue and eight-digit British National Formulary (BNF) code. Maternal postcode was linked to the NI Multiple Deprivation Measure (NIMDM) to give an indication of area-level deprivation for the mother’s residence at the time of pregnancy [ 13 ]. The NIMDM is a measure of relative deprivation within the NI population, and is a composite of seven domains including income, employment, health, education, access to services, living environment and crime. All personal identifiers were removed from the data to ensure non-identifiability prior to researcher access. In addition, all outputs were screened by HBS staff before release to ensure confidentiality following the Five Safes Framework [ 14 ]. The population spine for the NI cohort was defined as all pregnancies from all individuals with a valid HCN recorded in NIMATS. The start of each pregnancy was calculated from the variable “expected date of confinement”, estimated by ultrasound examination, minus 280 days. Pregnancies with a start date between 1 st January 2012 to 31 st December 2020 were retained. Pregnancies were included if the mother was aged between 15 to 49 years inclusive at the time of birth, aligning with previous analyses in other MuM-PreDiCT cohorts and reflecting the majority of reproductive age individuals. Multimorbidity was defined as the presence of two or more physical or mental health conditions, and further stratified by physical health multimorbidity, mental health multimorbidity, and multimorbidity including both physical and mental health conditions (see Fig.  1 )[ 6 ]. Complex multimorbidity was defined primarily as at least three conditions impacting at least three separate organ systems, in line with previous research in multimorbidity [ 15 ]. In order to compare multimorbidity rates in pregnancy with previous MuM-PreDiCT studies, a further definition of three or more conditions regardless of systems impacted was also used within the longitudinal trends analysis. Fig. 1 Definitions of multimorbidity and complex multimorbidity Definitions of multimorbidity and complex multimorbidity In this study, maternal pre-existing multimorbidity was ascertained using the MuM-PreDiCT phenome of 79 long-term physical and mental health conditions, including relapsing and remitting conditions [ 6 ]. Briefly, the phenome and associated definitions were developed via review of existing literature [ 16 ] and a Delphi process conducted with patient and public representatives, clinicians, researchers and data scientists [ 6 ]. For this study, the resultant 79 conditions were mapped to both secondary care admission ICD-10 diagnostic codes and, where applicable, patterns of community dispensed medications which likely indicated the presence of a condition of interest. Diagnostic codes were obtained from the PAS database and medications from the EPD database. Only conditions that were identified as pre-existing prior to the start date of each pregnancy were retained. The earlier work by the MuM-PreDiCT consortium for identification of long-term conditions via medications required at least four prescriptions in any 12-month period prior to the start of pregnancy. However, in NI, scan rates of prescription data to the EPD can vary, as previously described [ 17 ]. Therefore, in this NI study, the definition of conditions identified by prescription data was adjusted to require at least two prescriptions in any 12-month period. Covariates were obtained from NIMATS and grouped as follows: (1) maternal age (categorised as < 25 years, 25 to 34, and ≥ 35 years), (2) area deprivation (derived from the mother’s post-code recorded at the first antenatal appointment, linked to NIMDM via super output area and aggregated to quintiles; Q1 most deprived through to Q5 least deprived), (3) BMI measured at the first antenatal appointment (categorised as underweight < 18.5 kg/m 2 , healthy weight 18.5 to 24.9 kg/m 2 , overweight 25 to 29.9 kg/m 2 , obese I 30 to 34.5 kg/m 2 , obese II 35 to 39.9 kg/m 2 , and obese III ≥ 40 kg/m 2 ) and (4) gravidity (categorised as 1, 2, 3, 4, or ≥ 5 pregnancies including current pregnancy), (5) settlement type (urban: Belfast & Derry, rural: open countryside/small villages, intermediate: large villages through to large towns, outside NI/not recorded), (5) ethnicity (white, non-white, unknown), and (6) smoking status recorded at the first antenatal visit (non-smoker, ex-smoker, smoker, missing). Statistical disclosure control was assessed separately for each analysis, and further aggregation of the grouping of these characteristics was performed until the counts within all strata were at least n = 10 to protect confidentiality of individuals. Missing values were retained and not imputed. Yearly trends in selected background characteristics for pregnant women were explored over the full study period, including, maternal age, deprivation quintiles, BMI category and gravidity. Initially, the frequency and prevalence for each category were calculated per year and visualised using line graphs. To explore intersecting characteristics over the study period, the following combinations were also explored using stacked bar plots: (i) age, BMI and gravida within each deprivation quintile; (ii) BMI within each intersecting age/deprivation group; (iii) gravida within each intersecting age/deprivation group. Estimated prevalence and 95% confidence intervals were calculated on a yearly basis from 2012 to 2020 and presented as the percentage of women entering pregnancy with pre-existing multimorbidity using the total number of pregnancies with a start date falling within each year as the denominator. Yearly prevalence was visualised using dot plots with 95% confidence intervals. Prevalence of multimorbidity was first estimated using a full look-back period to the inception of each database. This was repeated with a standardised look-back period for each pregnancy episode. For detection of conditions through ICD-10 codes in the PAS database which has an inception date of 2006, the standardised look-back period was set at 5 years prior to the start of pregnancy. Whereas the standardised look-back period for EPD was set to 2 years due to the later inception date (2010). To provide context to observed trends in yearly prevalence estimates of multimorbidity, we also explored the yearly prevalence of pre-existing health conditions categorised by organ system. Yearly prevalence of all subtypes of multimorbidity and complex multimorbidity was estimated within each age, deprivation, BMI and gravida group. To explore intersectionality of multimorbidity across characteristics, yearly prevalence of multimorbidity was also estimated for intersecting age and deprivation groups. For these stratified analyses, conditions were detected using a standardised look-back period of 5 years for PAS and 2 years for EPD. Cross-sectional analyses of patterns in maternal multimorbidity were limited to individuals with a pregnancy start date falling between 2014 and 2019, inclusive. This time period represents a period of stability in prevalence rates of multimorbidity in pregnancy as observed in the yearly prevalence analysis. Maternal characteristics were categorised as described above and summarised as frequency and percentage for the full cohort, and for subgroups with (i) at least one long-term condition, (ii) multimorbidity (iii) physical health multimorbidity, (iv) mental health multimorbidity, (v) mental and physical health multimorbidity, and (vi) complex multimorbidity (3 or more systems). Unique combinations of long-term conditions contributing to multimorbidity were detected and summarised as frequency and percentage of pregnancies impacted. The top 10 most prevalent combinations are reported for the full study cohort and stratified by age and deprivation. This was repeated to explore unique combinations of conditions impacting pregnant women within intersecting age and deprivation groups. Unique combinations of organ systems implicated in complex multimorbidity was also explored for the full study cohort, however small counts prevented further stratification by age and deprivation. Prevalence of each organ system being implicated either on its own, in addition to at least one other organ system (multimorbidity), and in addition to at least two other organ systems (complex multimorbidity) was established and visualised using a histogram. See Appendix 1 in Supplementary Material. All data preparation, analyses and visualisation were performed in R [ 18 ] [R citation].

The number of pregnancies per year in NI declined over the study period from 24,403 in 2012 to 19,504 in 2020. This reflects an underlying decline in general fertility rate from 67.5 live births per 1,000 (2012), to 58.5/1,000 (2020) [ 19 ]. Over the course of the study period, there were changes in the background characteristics of pregnant women; notably higher proportions of women were older (≥ 35 years: 20.6% to 24.1%), living with obesity (BMI ≥ 30: 18.9% to 26.1%), and had higher numbers of pregnancies (five or more pregnancies: 8.6% to 10.8%) (Fig S1 in Additional Figures). When stratified by deprivation quintile, women from the most deprived areas demonstrated the largest increases in advancing age and obesity prevalence over time (Fig S2 in Additional Figures). When stratified by both deprivation and age (Fig S3 in Additional Figures), obesity prevalence in both the 25 to 34 years and ≥ 35 years old groups demonstrated disparity between the most and least deprived areas, which appeared to increase further over time. By 2020, approximately one in three pregnant women aged over 25 years and residing in deprived areas had co-morbid obesity (25-34y: 31.8%; ≥ 35y: 32.9%) compared to around one in five in the least deprived areas (25-34y: 19.5%; ≥ 35y: 21.2%). The proportion of women with three or more pregnancies did not demonstrate a distinct change over time in intersectional groups, however a higher proportion overall was observed in older groups in more deprived areas (Fig S4 in Additional Figures). Prevalence of pre-existing multimorbidity ranged from an early low of 18.2% (95%CI: 17.7 to 18.7%) in 2012, and peaking at 22.8% (95% CI: 22.3 to 23.4%) in 2016. There was an upward trend between 2012 and 2014 where confidence intervals are not overlapping, after which the prevalence plateaued at around 22% with overlapping confidence intervals in estimates for the years 2014 to 2019, before a significant decrease to the lowest rate of 17.5% (95%CI:17.0 to 18.0%) in 2020 (Fig.  2 ). These findings provided the rationale for limiting the subsequent cross-sectional analyses to the period of stability between 2014 to 2019. When multimorbidity rates were stratified by type, the majority of pregnant women had both physical and mental health conditions (range 13.0 to 17.4% over the study period), whereas fewer pregnancies had multimorbidity comprising of only physical health conditions (range 1.9 to 3.9%) or of only mental health conditions (range 1.4 to 1.9%). Complex multimorbidity (3 or more systems) ranged from 4.0% of women in 2012 to 6.1% in 2017 (Fig.  3 ). Fig. 2 Temporal changes (2012 to 2020) in detectable multimorbidity using a full look-back period Fig. 3 Temporal changes (2012 to 2020) in detectable complex multimorbidity using a full look-back period Temporal changes (2012 to 2020) in detectable multimorbidity using a full look-back period Temporal changes (2012 to 2020) in detectable complex multimorbidity using a full look-back period Adopting a standardised look-back period resulted in lower prevalence across all time periods compared to a full look-back period to the inception of each database (Fig S5 in Additional Figures). However, a similar pattern of change was shown – multimorbidity increased from 11.0% to 12.4% between 2012 and 2014, then plateaued between 2014 to 2019, and declined in 2020 to 10.6%. Again, multimorbidity was comprised mainly of coexisting physical and mental health conditions. Prevalence of complex multimorbidity also appeared to increase between 2012 and 2014, plateauing thereafter, and falling in 2020 (Fig S6 in Additional Figures). When individual conditions were grouped by organ system involvement and a standardised look-back period for detection used, the longitudinal trends in detectable rates varied (Fig S7 & S8 in Additional Figures). Mental health conditions demonstrated the highest annual rate across the study period compared to all other organ systems increasing from 18.6% of pregnancies in 2012 to 26.2% in 2019. Conditions affecting endocrine and respiratory systems also demonstrated an upwards trend, with endocrine conditions increasing from 3.0% to a peak of 4.4% in 2019, and respiratory conditions increasing from 7.4% in 2012 to a peak of 8.3% in 2019. Both systems demonstrate a decline in 2020, during which time the COVID-19 pandemic was emerging, with rates of pregnant women living with at least one pre-existing respiratory condition falling to 7.2%, and rates of pregnant women with at least one pre-existing endocrine condition falling to 4.2%. The percentage of pregnancies with at least one dermatological condition showed a markedly different trend, decreasing over the study period from 7.2% to 4.1%. Distinct patterns of change in detectable conditions were not observed in other organ systems. The results for annual detectable rates of multimorbidity subtypes across maternal characteristic groups using standardised look-back periods are summarised in Table  1 and illustrated in full in Figures S9-S16 in Additional Figures. Table 1 Summary of longitudinal trends in prevalence of maternal multimorbidity in NI between 2012 and 2020 Group with highest prevalence of multimorbidity (MM) * clearly highest  ~ highest but with overlap of CI Temporal trend ↗ increasing MM over time ↘ decreasing MM over time  = no clear trend Trajectory change in 2020 ↑ increased in 2020 ↓ decreased in 2020  = no clear change NA not able to explore Type of Multimorbidity PH or MH PH only MH only PH + MH PH or MH PH only MH only PH + MH PH or MH PH only MH only PH + MH Age (See Fig S9) < 25y * *  =  ↘ ↗  =  ↓  =   =  ↓  25-34y  =  ↘  =   =  ↓ ↓  =  ↓  35 + y * * *  =   =   =   =  ↓ ↓ ↓ ↓ Deprivation (See Fig S11)  Q1 (most deprived * * *  =  ↘  =   =   =   =   =   =   Q2  =  ↘  =   =   =   =   =   =   Q3 *  =   =   =  ↗ ↓ ↓  =  ↓  Q4 * ↗  =   =  ↗ ↓ ↓  =  ↓  Q5 *  =   =   =   =  ↓ ↓  =   =  BMI Groups (See Fig S13) < 18.5 *  =   =   =   =  NA NA NA NA  18.5–24.9  =   =   =   =  NA NA NA NA  25–29.9  =   =   =   =  NA NA NA NA  30–34.9  =   =   =   =  NA NA NA NA  35–39.9  =   =   =   =  NA NA NA NA ≥ 40 * * *  =   =   =   =  NA NA NA NA Gravida (See Fig S15)  1  ~   =  ↘  =   =  ↓ ↓ ↓ ↓  2  ~   =  ↘  =   =  ↓  =  ↓ ↓  3  ~   =   =   =   =  ↓  =   =  ↓  4  ~   =   =   =   =  ↓  =   =  ↓ ≥ 5 *  ~  * *  =   =   =   =   =   =   =   =  MM Multimorbidity, MH Mental Health, PH Physical Health Summary of longitudinal trends in prevalence of maternal multimorbidity in NI between 2012 and 2020 MM Multimorbidity, MH Mental Health, PH Physical Health Multimorbidity consisting of only physical conditions increased with age, with prevalence rates peaking at 4.6% in the oldest group (≥ 35y). Rates of physical-only multimorbidity were lowest in the youngest group, and also appeared to decrease over time, from 2.8% in 2012 to 1.1% in 2020 (Fig S9 in Additional Figures). Conversely, multimorbidity consisting of only mental health conditions most impacted the youngest age group, increasing over time with a peak of 3.3%, compared to other age groups which had rates consistently less than 1%. Differences across age groups in the combined physical and mental health multimorbidity group were less distinct, however the oldest group (≥ 35y) demonstrated a slightly higher prevalence of maternal multimorbidity (7.6 to 9.7%). Complex multimorbidity (3 or more systems) demonstrated a dose–response pattern, increasing across age groups (< 25y: 1.5 to 2.4%; 25-34y: 2.2 to 2.9%; ≥ 35y: 2.7 to 3.8%) (Fig S10 in Additional Figures). Detection of multimorbidity was greater as deprivation increased, and this pattern was consistent across all years (Fig S11 in Additional Figures). However, when prevalence was stratified by type of condition, results indicate that higher rates of multimorbidity in the most deprived are driven by mental health conditions, which also increased from 1.2% in 2012 to 2.2% 2017, while physical-only multimorbidity decreased in this group from 3.2% to 2.0%. In contrast, multimorbidity consisting of only physical health conditions was higher in the least deprived areas, and consistent over much of the study period (range: 3.1 to 4.1%). The decline in rates of multimorbidity, particularly physical multimorbidity, between 2019 and 2020, were greatest in the least deprived group compared to the most deprived. There was a general trend of higher prevalence of complex multimorbidity in the most deprived group compared to the least deprived group, however, change over time was not clear (Fig S12 in Additional Figures). Rates of multimorbidity in the youngest age group did not differ across deprivation quintiles and was broadly consistently between 20 to 25% in the time period 2012 to 2019 (Fig.  4 ). However, in the older age groups, differences were observed between deprivation quintiles, with prevalence estimates more than 25% in 25–34 year olds living in the most deprived areas, and up to 30% in those aged at least 35 years and living in the most deprived areas. Fig. 4 Temporal changes (2012 to 2020) in detectable multimorbidity across intersecting age and deprivation groups Temporal changes (2012 to 2020) in detectable multimorbidity across intersecting age and deprivation groups Multimorbidity increased as BMI class increased, but did not demonstrate any clear change over time (multimorbidity healthy BMI: 9.3% to 12.8% vs obesity III: 23.9 to 25.3%) (Fig S13 in Additional Figures). However, when stratified by type, mental health multimorbidity was demonstrably greater in women who were underweight (< 18.5 kg/m 2 ). A dose–response was also observed for complex multimorbidity, with rates highest in the group with obesity class III (7.8 to 8.9%) (Fig S14 in Additional Figures). Multimorbidity detection increased with advancing number of pregnancies with rates of between 15.8 to 19.7% for multimorbidity, and 3.8 to 5.1% for complex multimorbidity in the group with five or more pregnancies (Fig S15 & S16 in Additional Figures). When stratified by type, women with five or more pregnancies also showed a significantly greater prevalence of mental health multimorbidity (1.4 to 1.9%) and of coexisting physical and mental health (10.6 to 14.8%). No clear differences between groups were observed for physical only multimorbidity, however, groups with one or two pregnancies showed declining detectable rates over time. Cross-sectional demographic, socioeconomic and health-related characteristics are presented in Table  2 for the cohort of women with pregnancy start dates between January 2014 and December 2019 inclusive. In total, 22.2% of pregnant women in NI were entering pregnancy with multimorbidity, and 5.8% with complex multimorbidity. Multimorbidity comprised mainly of coexisting physical and mental health conditions (16.9%), with 3.5% living with physical-only multimorbidity, and 1.8% living with mental-health-only multimorbidity. Multimorbidity was higher at extremes of maternal age (< 25y: 24.1%; 25-34y: 21.2%; ≥ 35y:23.4%), and higher in the white ethnicity group (22.6% vs 8.64%). Multimorbidity also increased as deprivation increased, with a larger step increase moving into the most deprived areas (least deprived: 19.6%; most deprived: 25.8%). Interestingly, mental health-only multimorbidity increased with increasing deprivation, whilst physical-only multimorbidity increased as deprivation decreased. Multimorbidity was higher in urban (23.6%) and intermediate (23.5%) settlements in comparison to rural areas (18.8%). Again, divergent patterns were apparent in mental health morbidity, which was higher in the urban group (2.7% vs 1.9% and 0.81%), in contrast to physical morbidity which was higher in the rural group (4.3% vs 2.7% and 3.4%). A gradual increase in multimorbidity across BMI class was observed from 18.4% in the healthy BMI class to 39.2% in those with obesity class III. Multimorbidity also increased with increasing number of pregnancies (first pregnancy: 19.2%; ≥ 5 pregnancies: 30.7%) with mental health morbidities featuring more prevalently as the number of pregnancies increased. Prevalence of multimorbidity was higher in those who continued to smoke during pregnancy (33.6%) and those who had previously smoked (25.8%) compared to non-smokers (18.8%). Table 2 Demographic, socioeconomic and health-related characteristics of women within the study cohort (2014–2019) Full Cohort Women with at least one long-term condition Women with at least 2 long-term conditions Women with at least 2 physical health conditions (& no mental health) Women with at least 2 mental health conditions (& no physical) Women with at least 1 physical and 1 mental health condition Women with at least 3 conditions from 3 different systems 137,750 77,733 (56.43%) 30,535 (22.17%) 4,859 (3.53%) 2,411 (1.75%) 23,265 (16.89%) 8,035 (5.83%) n Col % n Row % n Row % n Row % n Row % n Row % n Row % Age < 25y 21,609 15.69 13,517 62.55 5219 24.15 393 1.82 993 4.60 3833 17.74 976 4.52  25-34y 84,346 61.23 46,146 54.71 17,880 21.20 2962 3.51 1148 1.36 13,770 16.33 4782 5.67 ≥ 35y 31,795 23.08 18,070 56.83 7436 23.39 1504 4.73 270 0.85 5662 17.81 2277 7.16 Ethnicity  Non-white/missing 4662 3.38 1421 30.48 403 8.64 130 2.79 24 0.51 249 5.34 97 2.08  White 133,088 96.62 76,312 57.34 30,132 22.64 4729 3.55 2387 1.79 23,016 17.29 7938 5.96 Deprivation  1 29,439 21.37 18,535 62.96 7588 25.78 743 2.52 910 3.09 5935 20.16 1939 6.59  2 29,226 21.22 16,543 56.6 6453 22.08 920 3.15 577 1.97 4956 16.96 1772 6.06  3 28,151 20.44 15,443 54.86 6056 21.51 1101 3.91 392 1.39 4563 16.21 1567 5.57  4 27,175 19.73 14,881 54.76 5823 21.43 1132 4.17 303 1.11 4388 16.15 1625 5.98  5 22,617 16.42 11,846 52.38 4435 19.61 942 4.17 203 0.90 3290 14.55 1091 4.82  NA 1142 0.83 485 42.47 180 15.76 21 1.84 26 2.28 133 11.65 41 3.59 Settlement Type  urban 26,901 19.53 15,924 59.19 6335 23.55 736 2.74 713 2.65 4886 18.16 1564 5.81  rural 38,224 27.75 19,584 51.23 7190 18.81 1652 4.32 310 0.81 5228 13.68 1855 4.85  intermediate 71,483 51.89 41,740 58.39 16,830 23.54 2450 3.43 1362 1.91 13,018 18.21 4575 6.40  not recorded/outside NI 1,142 0.83 485 42.47 180 15.76 21 1.84 26 2.28 133 11.65 41 3.59 BMI Class < 18.5 kg/m 2 2495 1.81 1420 56.91 479 19.20 51 2.04 93 3.73 335 13.43 90 3.61  18.5–25 kg/m 2 61,772 44.84 31,704 51.32 11,349 18.37 1934 3.13 1114 1.80 8301 13.44 2527 4.09  25–30 kg/m 2 41,225 29.93 23,415 56.80 9075 22.01 1498 3.63 671 1.63 6906 16.75 2363 5.73  30–35 kg/m 2 18,821 13.66 11,937 63.42 5216 27.71 794 4.22 309 1.64 4113 21.85 1529 8.12  35–40 kg/m 2 8109 5.89 5555 68.50 2583 31.85 357 4.40 140 1.73 2086 25.72 830 10.24 > 40 kg/m 2 3874 2.81 2887 74.52 1518 39.18 184 4.75 54 1.39 1280 33.04 602 15.54  NA 1454 1.06 815 56.05 315 21.66 41 2.82 30 2.06 244 16.78 94 6.46 Gravida  1 43,185 31.35 22,399 51.87 8281 19.18 1502 3.48 698 1.62 6081 14.08 2051 4.75  2 40,708 29.55 22,254 54.67 8478 20.83 1390 3.41 628 1.54 6460 15.87 2175 5.34  3 26,604 19.31 15,464 58.13 6059 22.77 973 3.66 433 1.63 4653 17.49 1612 6.06  4 13,947 10.12 8665 62.13 3633 26.05 533 3.82 289 2.07 2811 20.15 974 6.98  5plus 13,306 9.66 8951 67.27 4084 30.69 461 3.46 363 2.73 3260 24.50 1223 9.19 Smoking Status  Currently smoking 18,848 13.68 13,779 73.11 6335 33.61 327 1.73 1108 5.88 4900 26.00 1542 8.18  Ex-smoker 13,812 10.03 8647 62.6 3564 25.8 402 2.91 333 2.41 2829 20.48 1009 7.31  Non-smoker 48,696 35.35 24,371 50.05 9171 18.83 2167 4.45 389 0.8 6615 13.58 2470 5.07  Not recorded 56,394 40.94 30,936 54.86 11,465 20.33 1963 3.48 581 1.03 8921 15.82 3014 5.34 Demographic, socioeconomic and health-related characteristics of women within the study cohort (2014–2019) The highest rates of multimorbidity were observed in the group with obesity class III (39.2%), obesity class II (31.9%), current smokers (33.6%), and those with gravidity of 5 or more (30.7%). This cross-sectional analysis also demonstrated highest rates of mental health multimorbidity in those aged under 25 years (4.6%), in those classed as underweight (3.7%), and in the most deprived areas (3.1%). The groups living with obesity class III were most impacted by complex multimorbidity (15.5%). This was followed by those with obesity class II (10.2%), and women with five or more pregnancies (9.2%). The top ten unique combinations of maternal multimorbidity and complex multimorbidity primarily featured CMHD and atopic conditions such as asthma, eczema and allergic rhinoconjunctivitis (Table S1 & S2 in Additional Tables). At population level, the most common combination was CMHD and asthma (2.3%), and this persisted as the most prevalent combination in many of the stratified groups (Tables S3 & S4 in Additional Tables). When stratified by age, the youngest group showed unique combinations which featured a number of mental health conditions including CMHD, serious mental illness and other mental health conditions (including obsessive compulsive disorder, personality disorder, dissociative disorder or self-harm/suicide ideation) (Table S3 in Additional Tables). This predominance of a wide range of mental health conditions persisted in the younger age groups when further stratified by deprivation, appearing in the top 10 combinations in young women in both the most deprived and most affluent areas (Table  3 ). When unique combinations were explored within the older age group (aged ≥ 35y), women living in the most deprived areas showed higher rates of common mental health disorders and conditions including gall stones and hypertension. Table 3 Top 10 unique combinations of maternal multimorbidity, stratified by age and deprivation (2014–2019) Most deprived & Age < 25y n  = 7,757 Most deprived & Age 25–34y n  = 16,774 Most deprived & Age 35 + y n  = 4,908 Rank Conditions n % Conditions n % Conditions n % 1 CMHD + asthma 250 3.22 CMHD + asthma 443 2.64 CMHD + eczema 118 2.40 2 CMHD + eczema 191 2.46 CMHD + eczema 383 2.28 CMHD + asthma 104 2.12 3 CMHD + other MH 146 1.88 CMHD + other MH 149 0.89 CMHD + thyroid disorder 53 1.08 4 CMHD + migraine 67 0.86 CMHD + migraine 125 0.75 CMHD + allergic rhinoconjunctivitis 35 0.71 5 CMHD + asthma + eczema 63 0.81 CMHD + IBD 117 0.70 CMHD + IBD 32 0.65 6 SMI + other MH 58 0.75 CMHD + thyroid disorder 104 0.62 CMHD + migraine 25 0.51 7 CMHD + IBD 46 0.59 CMHD + Gall Stones 92 0.55 CMHD + gall Stones 20 0.41 8 depression* + other MH 37 0.48 CMHD + allergic rhinoconjunctivitis 84 0.50 CMHD + hypertension 17 0.35 9 SMI + asthma 33 0.43 CMHD + asthma + eczema 83 0.49 CMHD + endometriosis 16 0.33 10 CMHD + NDD 29 0.37 SMI + asthma 65 0.39 CMHD + asthma + eczema 16 0.33 Least deprived & Age < 25y n  = 1,746 Least deprived & Age 25–34y n  = 14,047 Least deprived & Age 35 + y n  = 6,824 Rank Conditions n % Conditions n % Conditions n % 1 CMHD + asthma 58 3.32 CMHD + eczema 271 1.93 CMHD + eczema 132 1.93 2 CMHD + eczema 51 2.92 CMHD + asthma 250 1.78 CMHD + asthma 120 1.76 3 CMHD + other MH 28 1.6 CMHD + thyroid disorder 109 0.78 CMHD + thyroid disorder 96 1.41 4 CMHD + migraine 13 0.74 CMHD + IBD 89 0.63 CMHD + allergic rhinoconjunctivitis 46 0.67 5 CMHD + asthma + eczema 13 0.74 CMHD + allergic rhinoconjunctivitis 84 0.60 CMHD + migraine 38 0.56 6 CMHD + IBD 11 0.63 CMHD + migraine 82 0.58 CMHD + IBD 34 0.5 7 SMI + other MH 11 0.63 asthma + eczema 78 0.56 CMHD + infertility 24 0.35 8 ** ** ** CMHD + asthma + eczema 60 0.43 Asthma + eczema 24 0.35 9 ** ** ** asthma + allergic rhinoconjunctivitis 56 0.40 Asthma + allergic rhinoconjunctivitis 24 0.35 10 ** ** ** CMHD + other MH 46 0.33 CMHD + endometriosis 23 0.34 CMHD  =  Common Mental Health Disorder; SMI  =  Serious Mental Illness (may include bipolar disorder, schizophrenia, affective psychosis or non-affective psychosis); NDD  =  Neurodevelopmental Disorders (may include attention deficit hyperactivity disorder, autism, or learning difficulties); Other MH  =  Other Mental Health Condition (may include obsessive compulsive disorder, personality disorder, dissociative disorder or self-harm/suicide ideation); IBD  =  Irritable Bowel Disease ** Suppressed due to small counts in the least deprived & aged under 25-year-old group. CMHD, SMI, NDD, asthma and gall stones feature in combinations ranked 8, 9, 10 (combined count of n = 25 and rate of 1.43%) Top 10 unique combinations of maternal multimorbidity, stratified by age and deprivation (2014–2019) Most deprived & Age < 25y n  = 7,757 Most deprived & Age 25–34y n  = 16,774 Most deprived & Age 35 + y n  = 4,908 Least deprived & Age < 25y n  = 1,746 Least deprived & Age 25–34y n  = 14,047 Least deprived & Age 35 + y n  = 6,824 CMHD  =  Common Mental Health Disorder; SMI  =  Serious Mental Illness (may include bipolar disorder, schizophrenia, affective psychosis or non-affective psychosis); NDD  =  Neurodevelopmental Disorders (may include attention deficit hyperactivity disorder, autism, or learning difficulties); Other MH  =  Other Mental Health Condition (may include obsessive compulsive disorder, personality disorder, dissociative disorder or self-harm/suicide ideation); IBD  =  Irritable Bowel Disease ** Suppressed due to small counts in the least deprived & aged under 25-year-old group. CMHD, SMI, NDD, asthma and gall stones feature in combinations ranked 8, 9, 10 (combined count of n = 25 and rate of 1.43%) When each organ system was explored separately, mental health featured predominantly, with 41.9% of the total pregnant population presenting with a mental health condition either in isolation or in combination with other conditions (Fig S18 in Additional Figures). When exploring organ system involvement in multimorbidity, 18.6% of the total study population were living with a mental health condition in combination with at least one other condition (Fig.  5 ). The next most prevalent organ system implicated in multimorbidity was respiratory (7.3% of total study population), followed by dermatology (7.2%), gastroenterology (3.9%), neurology (3.0%), endocrine (2.9%), eyes, nose and throat (ENT) (2.6%), gynaecology (1.9), and cardiovascular (0.8%). A similar hierarchy of organ system involvement was observed for complex multimorbidity (Fig.  6 ). Fig. 5 Organ systems involved in multimorbidity (detected using full look-back period; denominator: all pregnancies 2014–2019) Fig. 6 Organ systems involved in complex multimorbidity (detected using full look-back period; denominator: all pregnancies 2014–2019) Organ systems involved in multimorbidity (detected using full look-back period; denominator: all pregnancies 2014–2019) Organ systems involved in complex multimorbidity (detected using full look-back period; denominator: all pregnancies 2014–2019) When stratified by age, mental health conditions and respiratory conditions were implicated more often in multimorbidity in the youngest age group compared to the other age groups ( Fig.  5 ) , whereas when looking at complex multimorbidity, mental health conditions were implicated more in the oldest age group (Fig.  6 ). The remaining organ systems featured more often in multimorbidity and complex multimorbidity in the oldest group (aged ≥ 35y). Most organ systems showed a greater prevalence in the most deprived areas, with the exception of endocrine, ENT, gynaecology. See Appendix 1 in Supplementary Material.

The co-existence of two or more long-term conditions affecting either physical or mental health, referred to as multimorbidity, has a significant impact on individuals, their families and health systems, and increases the likelihood of polypharmacy, disability and mortality [ 1 , 2 ]. It is increasingly recognised that multimorbidity in pregnancy presents increased health risks to women and their babies. Previous work has shown an increased risk of preterm birth in women with two or more conditions, increasing further for those with four or more conditions [ 3 ]. There are also associated challenges for care of women throughout their reproductive journey, including the coordination of services for managing individual conditions, leading to increased healthcare utilisation and costs [ 4 , 5 ]. Estimates of prevalence in maternal multimorbidity have varied across the four UK nations, reflecting the separate routinely collected healthcare databases used by the MuM-PreDiCT UK Consortium: CPRD GOLD, covering four percent of GP practices in England, Scotland, Wales and Northern Ireland (NI); the SAIL databank covering the whole population of Wales; and Scottish Morbidity Records from two Scottish regional health boards [ 6 ]. Primary care data and prescribed medication data from the CPRD GOLD and SAIL cohorts estimates maternal multimorbidity to be around 44 and 46%, respectively [ 6 ]. In contrast, a Scottish cohort using secondary care data and community medications estimated prevalence of multimorbidity to be 20% [ 6 ]. Given that NI differs from other UK nations in some key characteristics, it is important to explore and better understand the epidemiology of maternal multimorbidity in the region. Socioeconomic deprivation is higher in NI than other UK nations [ 7 ], while the health and care system is also reported by the British Medical Association to underperform compared to other UK regions [ 8 , 9 ]. Previous research has also highlighted that the mental health burden in NI is higher than other UK countries [ 10 ]. These factors alone have the potential to translate into differences in population health, however there is further evidence that maternal health in NI differs from other regions. For example, maternal obesity, which is associated with increased rates of multimorbidity, has been shown to affect 22.5% of pregnant women in NI [ 11 ]. This represents a higher proportion compared to UK-wide estimates (19.1% in 2018), and pooled global (16.3%) and European (14.6%) estimates in the decade 2010–2019 [ 6 , 12 ]. Additionally, whole population data is available for NI, which negates the need to rely on only a four percent sample from CPRD GOLD. The primary aim of this study was to explore the prevalence and patterns of maternal multimorbidity using population-wide data for all women who gave birth in NI. We also had a number of secondary aims: (1) to explore underlying temporal trends in demographic, socioeconomic and health-related risk factors which may influence prevalence of multimorbidity; (2) to estimate the prevalence and patterns of multimorbidity and complex multimorbidity within pregnant women stratified by maternal age, deprivation, body mass index (BMI) and gravidity, and (3) to explore the utility of NI’s healthcare datasets for detecting health conditions and multimorbidity over time (2012–2020, inclusive) (Appendix 1 ; Supplementary Material).

This study used a robust analytic approach in line with previous MuM-PreDICT epidemiological studies of pre-existing multimorbidity in pregnant women. It has shown that multimorbidity impacts over 1 in 5 pregnant women living in NI, with complex multimorbidity affecting over 1 in 20 pregnant women in NI, and that differences exist in the prevalence and patterns of physical and mental health conditions across demographic and socioeconomic groups. The varied and distinct combinations of health conditions across intersectional groups point to a need for regional service development to cope with complexity in pregnancy and an individualised approach to provision of preconception health promotion, maternity care and long-term postnatal follow-up.

This is the first study to estimate the prevalence and patterns of multimorbidity in the entire population of pregnant women in NI, the most deprived nation in the UK. Over 22% of pregnant women in NI have at least two pre-existing health conditions. This mainly comprised of coexisting physical and mental health conditions (17%). Around 5% of women enter pregnancy with complex multimorbidity impacting at least three organ systems. With an estimated 20 thousand pregnancies per year, this means the NI maternity system is faced with over 1,000 births per year, 80 per month, 20 per week and approximately 2–3 per day in mothers who are living with complex multimorbidity. In-depth exploration of multimorbidity sub-types and unique combinations of conditions provides information on the mosaic of patterns of multimorbidity across demographic and socioeconomic groups. This provides important insights, for instance, that multimorbidity is highest in the most deprived group and appears to be driven by mental health conditions. Younger women and underweight women have higher rates of multimorbidity comprised exclusively by mental health conditions (4.6% and 3.7% respectively). The youngest age group, particularly those in the most deprived areas, also demonstrate a relatively higher rate of serious mental illness, neurodevelopmental disorders and other mental health conditions. Over the course of the study (2012–2020), our data demonstrate a change in the background demographic and socioeconomic characteristics of pregnant women living in NI. In particular, the annual number of pregnancies fell by 20% overall, with fewer younger women (< 25 years) entering pregnancy. Simultaneously, obesity rates increased over the study period (18.9% in 2012 to 26.1% in 2020), reflecting previous published evidence from NI [ 11 ]. This increase in maternal obesity prevalence is concerning, as the highest rates of multimorbidity (39.2%) and complex multimorbidity (15.5%) are seen in pregnant women living with obesity class III (≥ 40 kg/m 2 ). To date, the impact of maternal obesity and complex multimorbidity on service provision and healthcare utilisation in NI has not been quantified. Efforts are required to evaluate the ability of services to meet this demand and deliver appropriate care. The high prevalence of multimorbidity in pregnancy alongside health-related risk factors such as advanced maternal age, obesity, and deprivation is likely to have serious implications for women, families and healthcare services in NI. Higher numbers of coexisting conditions impact maternal morbidity and mortality, acute healthcare use in the perinatal period, and places a substantial burden of responsibility on women to coordinate their own care in the perinatal period, often in the absence of evidence informed advice [ 20 – 22 ]. Substantial investment by commissioners and policy makers is required across public health, primary and community care, and maternity services to support women with multimorbidity throughout their reproductive journeys from preconception, during their pregnancies and onwards through long-term postnatal follow-up. The MBRRACE-UK report (2023) calls for urgent development of maternal medicine specialist centres to facilitate early discussion of complex maternity cases with expert clinicians in order to save lives and improve outcomes for mothers and offspring [ 16 ]. As this research has demonstrated, there are a wide range of organ systems involved in multimorbidity in pregnancy, and the intersectional estimates reported here will be useful for designing specialist centres able to provide an array of appropriate services and developing such clinical capacity in NI. This study confirmed that maternal multimorbidity prevalence followed a social gradient. It follows that efforts are needed in this population to improve social determinants of health. Adequate health care must also be targeted towards communities who are disproportionately affected, and due to differences in how conditions cluster within different groups of women, a variety of services will be necessary. A focus on reducing non-communicable physical conditions such as hypertension and obesity is required for older maternal age groups, particularly in areas that are more deprived. Conversely, greater focus on improving mental health is needed in younger women. However, preconception health and care policies in the UK and Ireland currently lack focus on mental health [ 23 ]. It is important that women with multimorbidity are facilitated to engage with services prior to conception to optimise management of their long-term conditions in relation to pregnancy, including advice on medication safety and optimising their nutrition in pregnancy [ 23 ]. Women also need equitable access to preconception counselling to discuss how their long-term conditions may be impacted by pregnancy. It therefore follows that interdisciplinary training and education within primary, community and secondary care professionals is required to raise awareness and enact system wide change across all health professionals who care for women with long-term conditions and multimorbidity who have the potential to become pregnant. At a population-level, maternal multimorbidity is higher in NI compared to other UK nations. Within the MuM-PreDiCT Collaboration, NI and Scotland’s approaches are broadly similar. Both use a combination of secondary care diagnostic codes and community dispensed medications to ascertain long-term conditions. Data from Lee et al. (2022) presented maternal multimorbidity prevalence in Scotland to be around 19.8%, whereas this study observed a slightly higher rate of 21.7% in NI for the same year (2018) [ 6 ]. It cannot be ruled out that there are remaining differences in the healthcare data ecosystems between Scotland and NI which may explain this difference, rather than the health of the two populations. It is important to note that deprivation levels are higher in NI compared to other regions within the UK, with 37% of the NI population living in the most deprived 20% of the UK, and no areas within NI existing within the most affluent 20% of the UK [ 7 ]. Therefore, the patterns in prevalence in of multimorbidity across deprivation quintiles in NI may not be directly comparable to other nations. This study highlights the limitations of individual healthcare datasets, and strengthens the case for linkage of maternity data to other types of healthcare data for detection of long-term health conditions, which echoes the findings of MacRae et al. 2023 [ 15 ]. For a full discussion, please see Appendix 1 in Supplementary Material. The cohort included all pregnancies from all individuals and prevalence was calculated yearly in recognition of several background processes that may impact detection of multimorbidity. Firstly, rates of long-term conditions contributing to multimorbidity may have changed over time in this population. This approach also recognised that the quality of clinical coding practices may have improved over time and that longer look-back periods are available for individuals with pregnancies occurring later in the study period. Stratification of prevalence to different time periods and different maternal characteristics also facilitates comparability of the findings to other contemporaneous cohorts. Retention of all pregnancies and not just index pregnancies in the cohort, also ensured that annual prevalence rates were more accurate. Stratifying by year also allowed the significant decrease in prevalence of maternal multimorbidity in 2020 to be observed. This drop which was seen across most groups may be explained by the COVID-19 pandemic. This was a period of uncertainty as new evidence was emerging that pregnant women had increased risk, compared to nonpregnant women, of severe illness and death from COVID-19, and yet were not recognised as a high-priority group for vaccination in the early phases of the pandemic [ 24 , 25 ]. It has been reported that during this time, pregnancy intention rates were generally lower [ 26 ] and that there was reduced access to fertility services [ 27 ]. Women with pre-existing physical and mental health conditions may have altered their reproductive choices and behaviours resulting in fewer pregnancies to mothers with multimorbidity entering the maternity system, and hence producing the observed drop in prevalence in 2020. Research projects using administrative health data are bound by statistical disclosure control ensuring that individuals’ information cannot be identified. In NI, results of such projects must not contain a count of less than ten. This has implications for the granularity of data provided to researchers, and for how researchers must present results. For example, ethnicity information was provided to researchers in NI in aggregated groups “white”, “non-white” and “unknown”. This prevents a meaningful interrogation of health disparities between ethnic groups. This study also found a much smaller prevalence of multimorbidity in non-white pregnant women, which may in fact reflect a lower interaction with the health service for newcomers to NI and hence lower detection of conditions, rather than true prevalence. Additionally, this study aggregated age using 10-year intervals (< 25, 25 to 34, and ≥ 35 years) when evaluating prevalence of multimorbidity across intersectional groups. This may mean that prevalence of multimorbidity observed in the extremes of age ( 40 years) may have been underestimated. The approach used in this study to detect long-term conditions from medications required a record of at least two prescriptions in any 12-month period, as opposed to four in the study by Lee et al. 2022 [ 6 ]. This was chosen due to the acknowledged reduced scan rates of items into the EPD. The advantage of this approach is increased sensitivity for detection of long-term conditions. However, it should also be highlighted that this could have been at the expense of specificity and may have increased the likelihood of including instances where medications were prescribed for temporary conditions. Ongoing research in NI and more broadly across the UK is exploring the consequences of multimorbidity in pregnant women on outcomes in the antenatal, peripartum and postnatal period for both mother and infant [ 28 , 29 ]. These studies will explore the impact of the number of coexisting conditions, how particular combinations of physical and mental health conditions impact outcomes, and which combinations of conditions confer the highest risk. This study highlights the importance of detailed evaluation of subtypes of multimorbidity. In particular, our results demonstrate divergent patterns in prevalence of mental health and physical multimorbidity across different groups of pregnant women. Methodological insights from the sub analysis in Appendix 1 have important implications for research utilising routinely collected administrative health data. This is particularly relevant with regard operationalising multimorbidity phenome definitions within different types of routinely collected healthcare data, for example from secondary care diagnoses, medication dispensations and self-reported past medical histories which, as demonstrated by this study, can provide different estimates for prevalence of long-term conditions. This study also highlights important, unanswered research questions regarding inequalities in the burden of multimorbidity across intersectional groups of women in NI. In particular, there is a need to understand the social and economic determinants of the higher burden of mental health morbidity in the more deprived and younger pregnant women, and the consequences of this higher burden for maternal and child outcomes. Potential factors pertinent to NI have been posited by others including prevalence of violence against women and girls, intergenerational legacy of “the troubles”, and the impact of austerity [ 30 , 31 ]. There may also be an increased awareness and willingness to seek help amongst the younger age groups. There has also been a suggestion that health professionals’ decision making can be influenced by implicit bias which can exacerbate health inequalities in individuals with low socioeconomic status [ 32 ].

Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 1. Supplementary Material 2. Supplementary Material 3.