Preterm delivery and long-term maternal renal disease: A systematic review and meta-analysis

Background

Premature birth remains a major global health concern due to its association with neonatal mortality, morbidity, and long-term outcomes. According to the World Health Organization (WHO), over 13 million babies were born prematurely in 2020, ¹ representing over 13 million pregnant individuals at risk. While the association between prematurity and neonatal morbidity, such as long-term kidney disease in neonates, is well established in the literature, ² long-term maternal complications associated with preterm delivery are less well understood. ³

In recent years, research evidence has expanded to investigate the long-term maternal health impacts following delivering preterm, shifting our understanding of preterm delivery from solely a neonatal outcome to a marker of future maternal health risk. Delivery before 37 weeks has been found to be associated with long-term maternal cardiovascular disease (CVD). ⁴ For instance, Kessous et al. found that women who delivered prematurely had higher rates of cardiovascular events, with a linear association between the number of early deliveries and future CVD. ⁵ Similarly, an observational study by Theilen et al. ⁶ linked preterm delivery to all-cause and some cause-specific mortality. Beyond cardiovascular disease, Catove et al. reported that women with a history of preterm delivery are at an increased risk of developing metabolic syndrome, and the risk remained elevated, independent of the metabolic risk pre-pregnancy. ⁷ Collectively, these data suggest that preterm delivery may pose a broader risk to long-term health.

Pregnancy and its events place substantial physiological demands on the maternal kidney due to marked hemodynamic changes and an enhanced glomerular filtration rate. ⁸ Pre-existing maternal conditions or adverse pregnancy outcomes, particularly preeclampsia, have been linked to long-term maternal kidney disease,^9,10 likely due to the loss of the adaptation mechanism, dysregulation of the renin-angiotensin-aldosterone system, and endothelial damage at the nephron level. ¹¹ Whether preterm delivery contributes to similar renal pathophysiological mechanisms is yet to be established. Inflammation may be a potential mechanism underlying this association. Both CKD ¹² and preterm delivery have been hypothesized to involve pro-inflammatory pathways, ¹³ and since pregnancy often shows a significant increase in pro-inflammatory cytokines and C-reactive protein (CRP), this could be a mechanism. ¹⁴

The association between preterm delivery and chronic kidney disease has been linked in recent population-based observational studies.^{9,15,16 17} They demonstrated that women who experience preterm delivery (mainly spontaneous and after accounting for preeclampsia) may be at an increased risk of CKD and ESRD later in life.

This systematic review aims to evaluate the most up-to-date evidence on the association between preterm delivery and long-term maternal renal health outcomes by synthesizing raw data to provide more accurate clinically relevant risk estimates.

Sources

This systematic review and meta-analysis were prospectively registered (PROSPERO: CRD420251040336), designed, and conducted using standard methods recommended by the Cochrane Collaboration, ¹⁸ and the findings are reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Supplement 1). This study did not require ethical approval.

A comprehensive systematic literature search of data across four electronic databases (EMBASE, Medline, ClinicalTrials.gov, and CIANHL) was conducted on April 30, 2025. The search was limited to English-language publications with no restrictions on publication date or type. The search strategy combined Medical Subject Headings (MeSH) terms and free-text keywords, combining maternal long-term kidney disease and preterm delivery as the search terms. The search strategy was conducted and refined with the consultation of a medical librarian to ensure comprehensive coverage and validated against pre-selected key articles to ensure sensitivity. Search terms were adapted according to each database's requirements and syntax, and additional studies were added through a manual review of the reference lists of relevant studies. Complete details of the full search strategies and review protocols are provided in Supplement 2.

Study selection

Studies were considered eligible if they reported mothers who delivered before term as an exposure and subsequent long-term maternal renal outcomes (CKD or ESRD) as the outcome. Studies were required to report comparison groups of birthing at term (≥37 weeks of gestation) and preterm delivery (<37 weeks) with subsequent kidney disease assessment. Studies were included if they reported postpartum follow-up for any duration. Singleton or multiple pregnancies were included in the study. Studies involving populations without preexisting kidney disease were included to accurately measure the association between earlier delivery and maternal renal outcomes and to assess the impact of pregnancy complications. For the quantitative meta-analysis, studies were eligible if they provided raw data (event counts) for women who delivered prematurely, whether spontaneous or iatrogenic vs full term deliveries and subsequent development of kidney disease. All types of observational studies were included: cohort studies, case-control studies, and randomized controlled trials published in English. There were no restrictions on the publication dates of the articles.

Studies were excluded if they did not report a comparison group, did not report long-term maternal outcomes, or only illustrated neonatal or infant outcomes without maternal outcomes. Case reports, systematic reviews, meta-analyses, editorials, conference abstracts, and study protocols were also excluded.

After duplicates were removed, blinded abstract and title screenings were conducted independently by two reviewers (ES and MH) according to the inclusion and exclusion criteria. Studies that were deemed eligible were retrieved and compared, with any discrepancies discussed and resolved. Following the confirmation of eligible studies, full-text review and data extraction were performed by two reviewers (ES, MA). A bespoke spreadsheet was created to extract data.

For each study, the following information was extracted: study characteristics (author, year, country, design), sample size, population characteristics (age, parity, ethnicity, and comorbidities), exposure definition (categorization of preterm birth), outcomes (CKD-ESRD), follow-up duration, effect estimates [e.g., hazard ratios, odds ratios (OR), risk ratios], and adjustment variables. Raw data was extracted where available. The data extraction sheet is presented in Supplement 3.

Quality assessment

The methodological quality and risk of bias were assessed independently using the Newcastle–Ottawa Scale (NOS) for cohort and case-controlled studies. A score was assigned based on these three domains: selection (maximum 4 points), comparability (maximum 2 points), and exposure (maximum 3 points), with a total possible score of 9. Studies scoring 7–9 points were considered high quality, 5–6 points moderate quality, and 0–4 points low quality. ^{19 .}

Data analysis

The descriptive study characteristics, including country, sample size, and follow-up duration, were summarized. The prevalence of preterm delivery and CKD rates was reported across studies. For studies reporting raw data (event counts), unadjusted ORs with 95% confidence intervals (CIs) were calculated. Studies reporting adjusted effect estimates and raw data that could not be extracted were reported separately in the narrative synthesis.

A meta-analysis of the included studies with raw data and controls was performed using a random-effects model with STATA SE 18 (Stata Corp LLC, Collage Station, TX, USA) to calculate pooled estimates of the association between preterm delivery and maternal renal insufficiency. A random effect model was selected a priori due to anticipated clinical and methodological heterogeneity across the studies. A forest plot was generated to display the individual study results and pooled estimates with a 95% CI. Heterogeneity was assessed based on I-squared and Tau-squared statistics. I² values of 0–25%, 26–50%, 51–75%, and >75% indicated low, moderate, substantial, and considerable heterogeneity, respectively.

We selected to report effect measures as OR with 95% CI calculated from raw data. OR was preferred to hazard ratio (HR) as it provides a direct estimate of risk with unadjusted data and compares the events of kidney outcomes in women who delivered preterm versus those who delivered at term. While in the HR assumption, the risk is constant over time, this is not likely plausible for the onset of CKD, which can occur at varying time points over decades. ²⁰

Pooling hazard ratios using meta-analysis may lead to invalid results, especially when individual patient data is not available. ²¹ Instead, studies that included only HR, not raw data, will be evaluated qualitatively to strengthen the overall analysis of the systematic review.

Sensitivity analysis was conducted to explore potential sources of heterogeneity, and the “leave-one-out” method was used to identify potential outliers.

Publication bias was assessed using funnel plots; however, because of the limited number of studies included in the meta-analysis, the power of the tests was too low to reliably detect asymmetry. Funnel plots and formal tests (such as Egger's test) require a minimum of ten studies to have adequate statistical power. ¹⁷

Study characteristics

Table 1 presents a summary of the nine included studies. Eight were cohort studies, and one was a cross-sectional study. The studies included were published between 2006 and 2024. The sample included over 8.3 million participants from seven countries (Norway, Sweden, Germany, Canada, the Netherlands, Iran, and Israel). The studies ranged from cross-sectional to longitudinal population-based cohorts, with follow-up extending up to 46 years.

OPEN IN VIEWER

Table 1.

Summary table and overview of included studies examining maternal kidney disease following preterm birth.

Study	Study details (Country, period, population)	Follow-upDuration	PTB definition & prevalence	Kidney disease definition & method	KD outcomes (PTB vs control)	Maternal age & demographics	Key findings	Exclusion criteria
Vikse et al., 2006	Norway, 1967–1998, 756,420 singleton first pregnancies	30 years (mean 15.9 ± 9.4)	Delivered at <37 weeks (5.70%, 43,427 cases)	Glomerular diseases; Kidney biopsy; proteinuria >1 g/24 h or sCr >150μmol/L	PTB CKD: 0.13%; 59 cases total	Mean delivery: 24.1 ± 4.5 (11–50); Mean diagnosis: 40.8 ± 9.6 (18–69)	Kidney biopsy risk significantly increased with PTB	Diabetes, kidney disease, rheumatic disease
Crump et al., 2024	Sweden, 1973–2015 (f/u 2018), 2,201,638 singleton pregnancies	Median 25 yrs (max 46)	Delivered at <37 weeks (8.90%; 194,751 cases)	CKD (primary), ESRD (secondary); ICD codes >3 months postpartum	CKD 0.5%, ESKD 0.2%; PTB: CKD 1,991, ESRD 807	Delivery: Median 27 (IQR 23–30). Diagnosis: Median 61 (IQR 51–69). BMI 18.5–24.9: 35.4% 82.5% Swedish	HR 2.00 (1.90–2.11) for CKD with PTB	CKD prior to first delivery (n = 359, <0.1%)
Goetz et al., 2021	Germany, 2010–2017 (f/u Sep 2019), 193,152 singleton pregnancies	Mean 5.44 years	Delivered at <37 weeks (6.58%; 16,948 cases)	General CKD, Mild-moderate CKD, Severe CKD; ICD-10 N18/OPS criteria	Total CKD: 0.71%; PTB cases not specified	Delivery: 30.5 ± 5.6 years; Not otherwise specified	CKD risk associated with PTB severity	Preexisting CKD, ESKD, KRT, multiple gestation, inadequate insurance
Sandvik et al., 2010	Norway, 1967–1994, 639,018 singleton first pregnancies	37 years (mean 20 ± 9)	Delivered at <37 weeks. (5.50%; 337 cases)	ESRD; Dialysis or transplant	1 in 1300 women; 52 ESRD among 2204 diabetic women	Not specified for delivery. Diagnosis: Mean 20 ± 9 years after first pregnancy	PTB significantly associated with ESRD in diabetic women	Multiple pregnancies, UTI/kidney disease/hypertension before pregnancy
Dai et al., 2023	Canada, 1993–2003 (f/u 2013), 2,351,665 singleton pregnancies	16 years (13 months post-partum to 19 years)	Prevalence; 1,081,055 cases	ESRD; Dialysis or transplant	9.9 (HDP) vs 1.7 (non-HDP) per 100,000 person-years; PTB + HDP: 22, PTB + No HDP: 89	>15 - < 44 years; Lower ESRD risk with ≥2 deliveries	Significant ESRD risk with PTB + HDP combination	CKD, diabetes, age <15/>44, multiple gestation, pre-gestational DM, SLE, renal conditions
Burger et al., 2023	Netherlands, 2011–2015, 2466 parous women (6 ethnic groups)	Not specified	Delivered at <37 weeks; (10.80%; 526 cases)	CKD; ACR >3 mg/mmol and/or eGFR <60 mL/min/1.73m2	PTB: 26 cases; Control: 113 cases	Mean delivery: 39.39 (pooled SD 7.1); 6 ethnic groups; Obesity >30: PTB 0.8%, No PTB 1%	PR 1.8 (1.4–2.3) for HTN; 1.8 (1.1–3.0) for CKD with PTB	MI/stroke/cardiac procedures, small ethnic groups, missing data
Pariente et al., 2017	Israel, 1988–2012, 99,338 total deliveries	Mean 11.2 years	Delivered at <37 weeks (16.40%; 16,364 cases)	Renal-related hospitalizations; ICD-9 hospital records	PTB: 0.3% vs No PTB: 0.1%. PTB: 49 cases, Control: 83 cases	PTB: 28 ± 6 vs No PTB: 28.9 ± 6. BMI >30: PTB 0.8% vs 1%. Parity: PTB 2.9 ± 2 vs 2.7 ± 2	OR 2.9 (2.0–4.2) for renal hospitalizations with PTB	Known CKD or congenital renal malformations
Saei Ghare Naz et al., 2021	Iran, 1999, 3035 women with delivery	16 years (PTB: 13–17, FT: 14–17)	Delivered at <37 weeks; (6.99%; 212 cases)	CKD; eGFR <60 mL/min/1.73m2	PTB: 31.13% vs No PTB: 36.98%. Cases not clearly specified	>15 in 1999; BMI: PTB 27.1 ± 6 vs 26.7 ± 5; Education (diploma+): PTB 19.8% vs 21.8%. Parity: PTB 2.5 ± 1.1 vs 1 ± 0	HR 2.68 (1.02–7.05) for CKD with PTB	Age <15 at baseline, missing delivery data, never married/no delivery, CKD at baseline
Barrett et al., 2020	Sweden, 1973–2012, 1,943,716 singleton pregnancies	Median 20.6 years (max 41)	Delivered at <37 weeks (8.40%; 162,918 cases)	CKD and ESRD; ICD-10 codes	Total: 0.9% CKD, 0.07% ESRD. PTB subset data available	8.8% < 20y, 64% 20–29y, 25.7% 30–39y, 1.4% > 40y. BMI: 3% < 18.5, 34% 18.5–24.9, 9.5% 25–29.9, 4% > 30. 85% Swedish	Increased CKD and ESRD risk with PTB	Pre-pregnancy CKD, ESKD, CVD, hypertension, diabetes, SLE

The three largest studies were population-based cohorts, two from Sweden (Crump et al., 2024 and Barret et al., 2020)^9,28 examining 4 million participants, and one from Canada (Dai et al., 2023) ²⁵ with 2.4 million participants. The sample size ranged from 2466 women in the Netherlands multi-ethnic cohort (Burger et al., 2023) to over 2 million in the Swedish national cohorts.

Follow-up duration (time from delivery to maternal chronic kidney disease development) also varied among the studies, from 5.44 years (Goetz et al., 2021) ²³ to a maximum of 46 years reported from Crump et at. ²² Postpartum follow-up commenced as early as 3 months after delivery; two studies reported waiting for 13 months postpartum to exclude any residual effects of pregnancy on kidney function. Some studies reported durations as mean and standard deviation (SD), while others reported them as a range; therefore, direct comparisons between studies were not possible.

The eligible studies utilized population-based registries and administrative datasets as data sources. The Swedish Medical Birth Register and Medical Registry of Norway were used in two separate studies. Additional sources used were the Norwegian Kidney Biopsy Registry, AOK Baden-Wurttemberg (Germany), Norwegian Renal Registry, Norwegian Cause of Death Registry, Canadian Institute for Health Information, Dutch perinatal registry (Perined), Hospitalization database of Soroka University Medical Centre (Israel), and Tehran Lipid and Glucose Study (Iran).

Demographic

Demographic characteristics, such as maternal age, ethnicity, socioeconomic status, and comorbidities, were reported in the majority of studies. Maternal age at delivery or diagnosis of CKD was reported as the mean ± (SD), median, age range, or stratification by age groups. Ethnicity information was not consistently provided, and when it was, it was reported at the cohort level across the entire population. Burger et al., 2023 ¹⁰ reported a multiethnic Dutch cohort and found no strong evidence that ethnicity changed the relationship between PTB and CKD. Ethnicity data were insufficient for statistical analysis. Maternal body mass index (BMI) was reported in four studies for the entire cohort, and two studies reported the prevalence of obesity (BMI ≥ 30 kg/m²) among the population of mothers who birthed preterm as 0.8% and 4%. The majority of participants who delivered prematurely had a normal BMI. Smoking status data were available in three studies, showing 51–54% non-smokers.

Preterm delivery definition and prevalence

Preterm delivery was consistently defined across all studies as delivery before 37 weeks of gestation. The pooled prevalence of preterm delivery was 8.5% (95% CI: 6.3–10.7%), ranging from 5.5% (Sandvik et al. 2010) to 16.4% (Pariente et al. 2017).

Subcategorization by gestational age at birth varied between reports; only four studies^9,16,26,28 reported the subcategories in their analysis, but the gestational age cutoff was different, which did not allow subgroup analysis. Three of the included studies (Crump 2024, Burger 2023, and Saei Ghare Naz 2021^9,10,27) differentiated spontaneous from iatrogenic preterm delivery.

Gestational age assessment was documented in six of the nine included reports via the last menstrual period (n = 4) or ultrasonography (n = 2). The remaining three studies did not mention this. Eight of the nine studies reported only singleton pregnancies. Parity varied between studies, with most data including parous women.

Meta-analysis

Five studies provided raw data extractable for quantitative meta-analysis included 3,800,067 participants (Vikse et al., 2006; Crump et al. 2024, Burger et al. 2023, Pariente 2017, Sai Naz et al. 2021). As noted above, Sandvik et al. (2010) was excluded from the quantitative analysis as it reported the same Norwegian dataset as Vikse et al. (2006) with an overlapping time period.

The pooled analysis using a random-effects model demonstrated that a history of giving birth preterm was significantly associated with an increased risk of subsequent maternal kidney disease compared with giving birth at full-term (pooled OR = 1.82; 95% CI: 1.17–2.85; p = 0.01; Figure 2). Four of the five studies showed a positive association between preterm delivery and CKD, while one study by SaiNaz et al. (2021) reported a non-significant inverse association. Heterogeneity was observed across the studies (I ²  = 94.09%, τ² = 0.23). To identify the source of heterogeneity, a “leave-one-out” sensitivity analysis was conducted by systematically excluding each study and recalculating the pooled effect estimate. The analysis identified SaeiNaz et al. 2021 as a potential source of heterogeneity. When this study was excluded, the pooled OR increased to 2.16 (95% CI 2.06- 2.26, p < 0.001, Figure 3). Exclusion demonstrated shorter and more precise CI (2.06–2.26) vs (1.17–2.85). Heterogeneity also decreased (I ²  = 0.00).

OPEN IN VIEWER

Figure 2.

Forest plot showing the five studies included in the meta-analysis, assessing the statistical significance between preterm delivery and long-term maternal kidney disease.

OPEN IN VIEWER

Figure 3.

Meta-analysis of studies (after excluding Saei Ghare Naz 2021) investigating the association between preterm delivery and maternal kidney disease. Forest plots display individual and pooled estimates of the OR of Preterm and Full-term.

Studies reporting hazard ratios

The three remaining eligible studies (Goetz et al., 2021; Dai et al., 2023; Barrett et al., 2020) were not suitable for meta-analysis, as they did not report raw data but hazard ratios. The hazard ratios in these studies compared the likelihood of poor long-term renal outcomes in mothers who delivered preterm against those who delivered full-term. All studies reported a hazard ratio greater than one, and the 95% CIs did not include one. These findings are consistent with the results of the meta-analysis presented in Table 2. Barret et al. reported adjusted HR

OPEN IN VIEWER

Table 2.

A table summarizing the results of eligible studies not included in meta-analysis, using hazard ratios.

Study	Hazard ratio (95% CI)	95% CI	Outcome
Goetz et al., 2021(21)	1.79	1.53–2.09	CKD and ESRD
Dai et al., 2023(23)	2.36	1.81–3.08	ESRD
Barrett et al., 2020(26)	1.47	1.40–1.53	CKD

Quality assessment

The included studies generally demonstrated high methodological quality, highlighting large population-based samples, longer follow-up durations, and adjustment for confounders. The Newcastle-Ottawa scale was used to assess bias. All nine eligible studies scored between seven and nine showing a low risk of bias (Figure 4). The diversity in the definition and measurement of kidney disease, as well as inconsistent reporting of age, parity, or preterm stratification, contributed to the heterogeneity in the score. Full quality assessment details are provided in Supplement 3.

OPEN IN VIEWER

Figure 4.

A bar chart displaying the Newcastle Ottawa scale for bias for all the included studies.

Temporal trends in risk

The current review included studies with extended follow-up durations (over 40 years), which showed that the risk of kidney disease associated with preterm delivery persists across decades. In Crump et al. (2024), although the risk was highest in the first 10 years after delivery, the significant associations remained even 30–46 years later. To further support this, Norwegian population-based studies by Vikse et al. (2006) and Sandvik et al. (2010) demonstrated sustained risk over 20–30-year follow-up periods, with stronger associations observed in early postpartum years but persistent elevation extending throughout the observation period.

Gestational age at delivery

Some of the included studies examined dose-response relationships (gestational age at delivery and subsequent CKD). Vikse et al. (2006) found increasing kidney disease incidence with decreasing gestational age at delivery: 0.4% for deliveries <34 weeks versus 0.2% for 34–36 weeks compared to 0.1% in term deliveries. This pattern was consistent across multiple studies,^26,28 suggesting that earlier delivery might be linked to a higher potential risk. Crump et al. (2024) interestingly demonstrated that women with multiple adverse pregnancy outcomes, including delivering preterm, had progressively higher CKD risk, with hazard ratios increasing from 1.52 for one adverse outcome to 3.67 for three or more adverse outcomes.

Comparison with previous research

Previous systematic reviews comparing adverse pregnancy outcomes and long-term maternal renal disease have identified delivering preterm as an independent risk factor for CKD. The most recent Wu et al., ¹⁷ demonstrated a significant risk of renal disease with HR: 1.98 (95% CI: 1.57, 2.50). They found a similar significant HR for ESRD cases. A major difference between our data and Wu et al. is that our data included the most recent population-based study by Burger et al. ¹⁰ and Crump et al., ⁹ which examined data for more than 2.2 million women. This review also uniquely used raw data, which allowed for more accurate comparisons between studies, as it was not impacted by differences in the adjustments undertaken within individual studies.

Strengths and limitations

This review has several strengths. A major strength of this meta-analysis is its large sample size of >8 million participants. This enhances the generalizability of the results and reliability of the findings, as almost all studies showed a positive association between preterm delivery and long-term kidney sequelae. Second, we used raw event counts from the included studies to calculate effect estimates (pooled OR). This approach allowed for the direct analysis of the risk of CKD in women who delivered early versus those who delivered at term. In addition, this methodological approach helped minimize the dependence on variably adjusted estimates drawn from individual studies. Importantly, adding more raw data from two large population-based studies and not relying on HR as an effect estimate, as previous studies did, allowed conveying absolute risk estimates across the dataset. Thirdly, almost all studies included in the meta-analysis showed a significant link between delivering preterm and adverse maternal renal outcomes. Even in studies where raw data could not be extracted, the effect estimates (HR) in those studies demonstrated the same positive significant association.

This review also highlights a critical gap in women's health care. Currently, people with reproductive potential, transition from close perinatal care to ad-hoc gynecological or primary care follow-up, with little focus on pregnancy complications as predictors of future health outcomes. This represents a missed opportunity for early intervention and prevention. Our data suggest that the 8–15% of women who deliver preterm could be targeted for long-term health interventions. Despite this, current clinical practice largely ignores data such as these, which provide a valuable prognostic indicator that may guide long-term follow-up. Notably, heterogeneity was initially encountered in the meta-analysis, indicating variability between the studies. To further investigate this, a sensitivity analysis was conducted to identify the source of variability.

One limitation of our review is that raw data could not be extracted from all the included studies to add to the meta-analysis after contacting the authors. However, a meta-analysis was conducted from event counts of the majority of the studies (five out of nine) and complemented the review with a narrative synthesis of the remaining studies. A further limitation of this review is the inability to fully disentangle the independent contribution of preterm delivery from that of co-occurring hypertensive disorders of pregnancy in all included studies, as not all studies stratified preterm deliveries by whether spontaneous or iatrogenic or completely separate raw numbers with hypertensive disorders despite adjusting for confounding in estimating the risk. Whilst the dose-response relationship between earlier gestational age and increasing renal risk observed across multiple studies provides evidence of a biological signal independent of hypertensive status, residual confounding by hypertensive disorders cannot be entirely excluded from the pooled unadjusted analysis.

Future work

Future studies should aim for more standardized reporting of populations and definitions of renal disease, with clear and consistent methodologies to enable reliable comparisons. In addition, the biological link, so far in the literature, between preterm delivery and renal disease is less understood. Further mechanistic and translational research is needed to clarify the biological pathways linking mothers who deliver preterm and long-term maternal kidney disease.