Introduction

Immunoglobulin A nephropathy (IgAN) is the most common type of primary glomerulonephritis1,2 and a leading cause of chronic kidney disease (CKD)3,4. Although patients are in all age groups, the incidence peaks in the third and fourth decades of life5, when many individuals experience major life events, such as advancement to higher education, employment, marriage, pregnancy, childbirth, and child-rearing. Because CKD affects both maternal and fetal pregnancy outcomes6,7, pregnancy is often a major concern in women with IgAN.

Most previous studies, including pregnant women with IgAN, have shown that pregnancy itself does not worsen kidney prognosis in the early stages of CKD (serum creatinine level < 1.2 mg/dL or stage G1–3)8,9,10,11,12,13,14,15. However, evidence regarding the effects of IgAN on pregnancy outcomes is limited. Previous studies have shown that pregnancies with IgAN are associated with a higher risk of preeclampsia, preterm delivery, and small for gestational age or low birth weight fetuses than low-risk pregnancies without IgAN or other comorbidities16,17. However, these studies did not consider the effects of disease control status of IgAN or pharmacotherapy for IgAN at the time of conception, while other studies had relatively small sample sizes for statistical analyses. Whether differences in disease activity, as expressed by blood pressure and/or urine protein levels18, are associated with the likelihood of adverse pregnancy outcomes has not been fully elucidated.

For women of childbearing age with IgAN, evaluating this association is relevant for pre-conception care, such as predicting the likelihood of adverse pregnancy outcomes and considering the desirable timing of treatment for IgAN and pregnancy. Therefore, our study aimed to investigate the association between the disease control status of IgAN, as reflected by prescriptions provided before conception, and adverse pregnancy outcomes. We hypothesized that women with IgAN who had been prescribed antihypertensive medication or glucocorticoid immediately before conception were more likely to experience adverse pregnancy outcomes than those who did not, assuming that the need for pharmacotherapy implies the presence of hypertension or proteinuria.

Methods

Data source

We used clinical data from the Medical Data Vision Co., Ltd. (MDV), one of Japan’s largest hospital claims databases. The MDV database contains health administrative data, including individual diagnoses, prescriptions, examinations, procedures, and hospitalizations. Data on diagnoses in the MDV database is coded using the International Classification of Diseases, Tenth Revision (ICD-10). Data on prescriptions, examinations, and procedures were extracted using medical receipt codes (i.e., a 9-digit number assigned to each clinical practice in the Japanese healthcare insurance system). We obtained information on age, sex, smoking history, pregnancy status, and gestational age at the time of hospitalization from the hospitalization records. Data on weight at the time of conception were excluded from the study because of the substantial missing values. Blood test results were available for a limited population (i.e., for < 10% of the study population); urine test results were unavailable. From April 2008 to August 2021, 36,690,000 patient records from 449 hospitals in Japan were included in the MDV database. We identified and selected 924,238 patients with CKD based on specific ICD-10 codes (N18, N03, and N11).

Study design and patient selection

We conducted a case–control study using health administrative claims data of 924,238 patients with CKD, as retrieved from the MDV database. We assessed hospitalization records and selected data pertaining to pregnant women with IgAN according to the following procedures (Fig. 1). First, we selected the hospitalization records of pregnant women with CKD. Second, we excluded the hospitalization records of patients without an IgAN diagnosis on admission based on the ICD-10 code N028. Third, we excluded the hospitalization records of patients without delivery of live neonate during hospitalization. If the patient was hospitalized repeatedly during the same pregnancy, we included the last hospitalization record and excluded others. Hospitalizations owing to miscarriage, hyperemesis gravidarum, threatened preterm delivery, or other diseases unrelated to pregnancy were excluded. We also excluded the hospitalization records of patients who had been registered with the MDV database for less than six months at the time of conception for the precise assessment of exposures mentioned in the next paragraph.

Fig. 1
figure 1

Flowchart of patient selection. CKD chronic kidney disease, MDV Medical Data Vision Co., Ltd., IgAN Immunoglobulin A nephropathy, HDP hypertensive disorders of pregnancy. a Detailed procedures for estimation were mentioned in the main text.

Full size image

This study was approved by the Ethics Committee of the Institutional Review Board of Nagoya University Hospital (approval number 2021–0350) and performed following the tenets of the Declaration of Helsinki. Due to the retrospective nature of the study, the Ethics Committee of the Institutional Review Board of Nagoya University Hospital waived the need of obtaining informed consent.

Definition of exposures and outcomes

We defined two independent exposures: prescriptions of antihypertensive medications and glucocorticoid therapy within six months before conception. The medications used to assess each exposure are listed in Table S1. The case/control groups were defined separately for each outcome: severe hypertensive disorders of pregnancy (HDP) requiring intravenous nicardipine, and preterm delivery (gestational age of less than 37 weeks). In the treatment of HDP, the use of intravenous nicardipine indicates that oral antihypertensive medications have been ineffective in controlling severe hypertension, representing a medical emergency that often necessitates urgent delivery. In addition, intravenous nicardipine is an easily and precisely identifiable outcome measure in claims database analyses. Records of intravenous nicardipine treatment were extracted from the database using medical receipt codes (Table S2). Data pertaining to the delivery date are required to determine the presence of preterm delivery, but these data are not available directly on the MDV database. Therefore, we identified the delivery date using data pertaining to delivery-related medications, surgical procedures, and diagnoses. For this, we referred to the algorithm developed by Ishikawa et al., who estimated the delivery date to be within 7 days from the gold-standard delivery date in 96.4% of the study population19. The medications, surgical procedures, and diagnoses used to estimate delivery dates are shown in Table S2. For the analysis of preterm delivery, we categorized all pregnancies into three groups: “probable,” “possible,” and “undeterminable” (Fig. S1). The delivery dates of the pregnancies in the “probable” group were identified using the algorithm19. For the “possible” group, the delivery dates were estimated by an obstetrician because delivery-related information was insufficient. The delivery dates of pregnancies in the “undeterminable” group could not be identified, although it was certain that the patients delivered their babies during hospitalization. The assessment periods for exposures and outcomes are shown in Fig. 2.

Fig. 2
figure 2

Design diagram by patient event time. HT hypertension, DM diabetes mellitus, CAM chorioamnionitis. Explanation of time windows: Day 0 represents the estimated day of conception. Numbers indicate days relative to conception (negative numbers = before conception, positive numbers = after conception). Brackets indicate assessment windows for each variable. a We defined the day of conception as day 0, as identified using the hospitalization date and gestational age at the time of hospitalization. b Estimated delivery date was identified using delivery-related medications, surgical procedures, and diagnoses based on the algorithm developed by Ishikawa et al. 19. c The start day of the assessment window for exposures and covariates: HT and DM were changed to day 365 in the sensitivity analysis for women with IgAN who had been registered with the MDV database for > 1 year at the time of conception.

Full size image

Covariates

Data on maternal age, gestational age at the time of hospitalization, and smoking history were obtained from the hospitalization records. Data on chorioamnionitis (CAM) were obtained using a specific ICD-10 code. We defined diabetes mellitus (DM) as the prescription of antihyperglycemic medications, including insulin, from six months before conception to hospitalization. This definition captures both pre-existing diabetes and gestational diabetes, with the latter being more prevalent in our cohort as described in the Results section. Due to the limited number of cases, we were unable to analyze these conditions separately. We used the prescription of antihypertensive medications within six months before conception as the covariate “hypertension” while examining the association between glucocorticoid therapy and each outcome. The assessment periods for the covariates are shown in Fig. 2. The medications and diagnoses used to assess each covariate are listed in Table S3.

Statistical analysis

All outcomes and exposures were analyzed as categorical variables per pregnancy. We replaced the missing values of smoking history by creating 100 filled-in complete datasets using multiple imputations with chained equations. To assess the association between pre-conception pharmacotherapies and adverse pregnancy outcomes, we used multivariable logistic regression and calculated the adjusted odds ratios (aORs). For antihypertensive medications, we defined the following three groups: “all categories,” “only renin–angiotensin–aldosterone system inhibitors (RASi),” and “medications other than RASi.” For glucocorticoid therapy, we defined the following three groups: “drip infusion in vein (DIV) or per os (PO),” “only DIV,” and “only PO.” We assumed glucocorticoid therapy/RASi and antihypertensive medications other than RASi were the proxies for proteinuria and hypertension, respectively. Patients with no history of medication were treated as the comparison group. The details of the medications used to define each exposure group are shown in Table S1. We excluded the “undeterminable” group (Fig. S1) for the main analysis of preterm delivery.

We also performed a sensitivity analysis because the determination of preterm delivery was doubtful or impossible in several pregnancies. We analyzed this outcome in the following three ways: analyze only the “probable” group, treat patients in the “undeterminable” group as having term delivery, and treat patients in the “undeterminable” group as having preterm delivery. In addition, we performed a sensitivity analysis for women with IgAN who had been registered in the MDV database for more than one year at the time of conception. In this sensitivity analysis, we changed the start of the assessment period for each exposure (antihypertensive medications and glucocorticoid therapy) and the covariates DM and hypertension from six months to one year before conception. All analyses were performed using Stata/MP version 18.0 (StataCorp). Statistical significance was set at P < 0.05.

Results

Characteristics of patients

As shown in Fig. 1, we selected 2,615 hospitalization records of pregnant women with CKD. Of these, we excluded 2,134 hospitalization records without an IgAN diagnosis. Of the 481 hospitalizations of pregnant women with IgAN diagnosis, we excluded 126 because these patients did not deliver a baby during hospitalization, and 58 because these patients were registered in the MDV database for less than six months at the time of conception. Thus, 297 patients were eligible for analysis. The number of deliveries during the study period was one, two, and three times in 211, 40, and 2 patients, respectively.

The baseline characteristics of the case and control groups are presented in Table 1. DM was detected in 12 pregnancies; pharmacotherapy was initiated before conception in two patients, while it was initiated during pregnancy in others. Overall, 44 patients were prescribed antihypertensive medications within six months before conception. The details of medications were as follows: “only RASi” in 32 patients, and “medications other than RASi” in 12 patients. Among the “medications other than RASi” group, two different types of antihypertensive medications were prescribed before conception in three patients, and only one type was prescribed in others. The proportions of patients in whom antihypertensive medications were discontinued by the detection of pregnancy in the “only RASi” and “medications other than RASi” groups were 90.6% and 33.3%, respectively. Of the 33 patients who received glucocorticoid therapy within 6 months before conception, 20 continued therapy after conception, and 16 continued therapy until delivery. Data on the estimated glomerular filtration rate within 6 months before conception was available for 16 pregnancies. The mean (standard deviation) was 78.3 (25.9) mL/min/1.73 m2, and the median was 86.1 mL/min/1.73 m2 (interquartile range, 63.7 to 96.1 mL/min/1.73 m2).

Table 1 Characteristics of patients in the experimental and control groups for each outcome.
Full size table

Main analysis

The results of analysis are presented in Table 2. For the outcome of HDP that required intravenous nicardipine, the aOR of antihypertensive medications of “all categories” was 2.13 (95% confidence interval [CI], 0.91–5.01). In the analysis considering the categories of “only RASi” and “medications other than RASi,” the aORs were 1.27 (95%CI, 0.41 –3.96) and 5.01 (95% CI, 1.43–17.5), respectively. For glucocorticoid therapy before conception, no significant associations were observed in all variations of exposure: “DIV or PO,” “only DIV,” or “only PO.”

Table 2 Medication histories within six months before conception in the experimental and control groups for each outcome.
Full size table

Regarding the outcome of preterm delivery, the aORs for “antihypertensive medications of all categories” and “medications other than RASi” were increased (aOR, 3.02 and 6.45 [95%CI, 1.43–6.39 and 1.81–23.0], respectively). No significant association was observed in any of the exposure groups with glucocorticoid therapy before conception.

Sensitivity analysis

The results of the sensitivity analysis for preterm delivery determination are summarized in Table S4. In all types of predictions, the aORs of “antihypertensive medications of all categories” and “medications other than RASi” were elevated. No significant associations were observed with glucocorticoid therapy. We also evaluated data pertaining to women with IgAN who had been registered with the MDV database for one year or more at the time of conception (Tables S5, S6). These results were similar to those of the main analysis.

Discussion

In this study, we used real-world data from 297 pregnancies with IgAN to analyze the association between pre-conception medication histories and adverse pregnancy outcomes. The sample size was relatively large compared to that of previous studies. Patients who were prescribed antihypertensive medications before conception were more likely to experience adverse pregnancy outcomes than those who did not. To the best of our knowledge, this is the first study to highlight the importance of pre-conception blood pressure management and the potential of hypertension as a predictor of adverse pregnancy outcomes in women with IgAN who are planning pregnancies.

Our study showed that the association between the prescriptions of antihypertensive medications and adverse pregnancy outcomes was particularly strong in patients who were prescribed antihypertensive medications other than RASi. However, the association between glucocorticoid therapy before conception and adverse pregnancy outcomes was not significant, although the point estimates were substantial. In Japan, RASi is used not only to lower blood pressure but also to reduce urine protein levels20,21. Patients with normal or mildly elevated blood pressure and proteinuria are often treated with RASi alone before conception, and those with severe hypertension are more likely to be treated with antihypertensive medications other than RASi. In our study population, nearly 70% of the patients in the “medications other than RASi” group required continuation of antihypertensive medications during pregnancy. In contrast, antihypertensive medications were discontinued in approximately 90% of patients in the “only RASi” group after the detection of pregnancy. Considering this trend, pre-conception hypertension may have a stronger influence on adverse pregnancy outcomes than proteinuria.

Although previous studies have shown that CKD or chronic hypertension are risk factors for severe HDP or preterm delivery22,23,24,25,26, analyses according to individual underlying diseases are rare and limited to comparing pregnancies with IgAN and pregnancies without comorbidities16,17. To the best of our knowledge, this is the largest study to focus on risk factors for adverse pregnancy outcomes in women with IgAN. In addition, we used the medication history within six months before conception to estimate the status of blood pressure and urine protein levels. This may be a new perspective that was not considered in previous studies on pregnancy with IgAN.

Our study had several limitations. First, patients in the MDV database were limited to those who visited hospitals. Adverse pregnancy outcomes may have been overestimated in our study population compared with those in the entire pregnant population with IgAN because we did not analyze women with IgAN who visited clinics and had been well-controlled for a long time. The second limitation was the misclassification resulting from the discrepancies between medication histories before conception and the control status of IgAN (blood pressure and urine protein levels). It could also be affected by medication compliance in the exposed group. One possible situation was that some patients with poor blood pressure and urine protein control were not prescribed medications. This is because women planning pregnancy sometimes refuse medication, despite its benefits, due to concerns regarding adverse effects on the fetus. In this situation, the misclassification bias could weaken the association of medication histories with adverse pregnancy outcomes. This means the association between antihypertensive medications other than RASi and adverse pregnancy outcomes was robust. In contrast, the association of glucocorticoid therapy/RASi with the outcomes might have been weakened by the misclassification bias and insufficient statistical power. Therefore, we could not rule out the association of glucocorticoid therapy/RASi with adverse pregnancy outcomes. Because medication histories were qualitative data as the proxies for hypertension and proteinuria, further studies, including quantitative data, are warranted to elucidate this aspect. Third, we were unable to analyze neonatal outcomes, such as birth weight, Apgar score, and neonatal death, because neonatal data were not linked to maternal data in the MDV database. Fourth, we lacked data on BMI, kidney function, and urine protein levels before pregnancy. Patients with high BMI, low eGFR, or high urinary protein levels were more likely to be prescribed medications for hypertension and proteinuria. In addition, they were also at higher risk of adverse pregnancy outcomes. Therefore, residual confounding from these unmeasured variables could potentially influence the associations between pre-conception medication use and adverse pregnancy outcomes. However, we expect that the proportion of patients with advanced CKD and/or severe proteinuria was likely small and would not have a significant influence on our results. The clinical guidelines for the management of pregnancy in kidney patients 2017 (from the Japanese Society of Nephrology) recommended explaining the risk of severe kidney prognosis to women with advanced CKD and/or nephrotic-range proteinuria who are considering pregnancy27. This guideline-based counseling may explain why women with advanced CKD and/or severe proteinuria might be less likely to pursue pregnancy in clinical practice.

In conclusion, our study suggested that pre-conception antihypertensive medication use, as a surrogate marker for underlying hypertension, may help identify women with IgAN at higher risk of adverse pregnancy outcomes. Although treatments for proteinuria (glucocorticoid therapy and RASi) showed no significant associations with adverse pregnancy outcomes in our analysis, the influence of proteinuria itself on the prognosis of pregnancy remains unclear. Future studies using quantitative data on actual blood pressure control and proteinuria levels are warranted to further elucidate this aspect.