Birth Spacing and Risk of Adverse Perinatal Outcomes a Meta-Analysis

REVIEW

Birth Spacing and Risk of Adverse Perinatal Outcomes A Meta-analysis

Agustin Conde-Agudelo, MD, MPH Context Both short and long interpregnancy intervals have been associated with an Anyeli Rosas-Bermu´dez, MPH increased risk of adverse perinatal outcomes. However, whether this possible associa- Ana Cecilia Kafury-Goeta, MD tion is confounded by maternal characteristics or socioeconomic status is uncertain. Objective To examine the association between birth spacing and relative risk of ad- OTH SHORT AND LONG INTER- verse perinatal outcomes. vals between pregnancies have Data Sources Studies published in any language were retrieved by searching MEDLINE been associated with in- (1966 through January 2006), EMBASE, ECLA, POPLINE, CINAHL, and LILACS, pro- creased risk of several adverse ceedings of meetings on birth spacing, and bibliographies of retrieved articles, and by perinatalB outcomes, such as preterm contact with relevant researchers in the field. birth, low birth weight (LBW), small for Study Selection Included studies were cohort, cross-sectional, and case-control stud- gestational age (SGA), and perinatal ies with results adjusted for at least maternal age and socioeconomic status, reporting risk death.1-3 However, there has been dis- estimates and 95% confidence intervals (or data to calculate them) of birth spacing and agreement on whether the relation- perinatal outcomes. Of 130 articles identified in the search, 67 (52%) were included. ship is due to confounding by other risk Data Extraction Information on study design, participant characteristics, measure factors. For example, some research- of birth spacing used, measures of outcome, control for potential confounding fac- ers have argued that short intervals be- tors, and risk estimates was abstracted independently by 2 investigators using a stan- tween pregnancies merely designate dardized protocol. women already at higher reproductive Data Synthesis A random-effects model and meta-regression analyses were used to risk, either because of underlying dis- pool data from individual studies. Compared with interpregnancy intervals of 18 to 23 orders, socioeconomic status, or life- months, interpregnancy intervals shorter than 6 months were associated with increased style factors.4,5 Furthermore, previous risks of preterm birth, low birth weight, and small for gestational age (pooled adjusted research in this area has several meth- odds ratios [95% confidence intervals]: 1.40 [1.24-1.58], 1.61 [1.39-1.86], and 1.26 [1.18- odological limitations, such as small 1.33], respectively). Intervals of 6 to 17 months and longer than 59 months were also sample size, lack of control for poten- associated with a significantly greater risk for the 3 adverse perinatal outcomes. tial confounding factors, dichotomiza- Conclusions Interpregnancy intervals shorter than 18 months and longer than 59 tion of the measure of birth spacing on months are significantly associated with increased risk of adverse perinatal outcomes. the basis of an arbitrarily defined cut These data suggest that spacing pregnancies appropriately could help prevent such adverse perinatal outcomes. point, and use of birth interval (time elapsed between the woman’s last de- JAMA. 2006;295:1809-1823 www.jama.com livery and the birth of the index child) instead of interpregnancy interval (time This issue is relevant to public health between birth spacing and the risk of elapsed between the woman’s last de- and clinical practice because if short adverse perinatal outcomes that pro- livery and the conception of the next and/or long interpregnancy intervals are vided an overall summary of the effect pregnancy) as the measure of birth found to be independently associated with increased risk of adverse perina- Author Affiliations: Centro de Estudios e Investiga- spacing. The use of birth intervals over- cioń en Salud and Department of Obstetrics and Gy- estimates the risk of adverse perinatal tal outcomes, birth spacing might then necology, Fundacioń Santa Fe de Bogota´, Bogota´,Co- outcomes for very short intervals be- be considered an intervention to pre- lombia (Dr Conde-Agudelo); Department of Biostatistics, Universidad Autonoma de Occidente, Cali, Colombia tween pregnancies. vent such adverse outcomes, mainly in (Ms Rosas-Bermu´dez); and ClıńicaMaterno-Infantil Los the developing world. Therefore, we Farallones, Cali, Colombia (Dr Kafury-Goeta). performed a systematic review, includ- Corresponding Author: Agustin Conde-Agudelo, MD, For editorial comment see p 1837. MPH, Calle 58 No. 26-60, Palmira-Valle, Colombia ing meta-analysis, of the relationship ([email protected]).

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES measure and determined both the riski- meetings on birth spacing and bibliog- Studies of different designs and differ- est and the optimal interpregnancy in- raphies of the retrieved articles were ent measures of birth spacing that are tervals. In addition, we determined also searched by hand. No language re- included in the systematic review were whether estimates of the effect mea- strictions were imposed. In the case of analyzed separately because of differ- sure depend on dimensions of study studies discussing more than 1 out- ent threats to their internal validity. quality of the primary studies and come, each outcome was considered in- All published studies deemed suit- whether the relationship differs in sub- dependently. To find unpublished stud- able were retrieved and reviewed inde- groups defined by the characteristics ies, we contacted relevant researchers pendently by 2 authors (A.C.-A., A.R.-B.) of women, and we highlight deficits that in the field. Twelve authors were con- to determine inclusion. Disagreements need to be addressed in future studies. tacted as well, in an attempt to obtain were resolved through consensus. The additional data. degree of agreement was expressed as METHODS percentage agreement and ␬ statistics. We used a prospective protocol pre- Inclusion Criteria pared specifically for this purpose. The Studies were included if (1) they were Study Quality Assessment systematic review was conducted fol- cohort, cross-sectional, or case- Study methodological quality was lowing this protocol and reported us- control studies that evaluated the rela- judged by the following 6 validated cri- ing the checklist proposed by the Meta- tionship between birth or interpreg- teria believed to be important for the analysis of Observational Studies in nancy interval and any adverse perinatal quality of observational studies evalu- Epidemiology (MOOSE) group for re- outcome; (2) the definition of inter- ating the relationship between birth porting of systematic reviews of obser- pregnancy interval corresponded to the spacing and adverse perinatal out- vational studies.6 period between delivery of the previ- comes7,8: (1) pregnancy interval used ous infant and conception of the cur- (adequate if the study used interpreg- Identification of Studies rent pregnancy. Although the use of nancy interval; inadequate if the study A search was conducted by the inves- birth-to-birth interval overestimates the used birth interval); (2) categoriza- tigators in MEDLINE (1966 through risks of adverse perinatal outcomes for tion of exposure (adequate if the study January 2006), EMBASE (1980 through very short intervals, studies using birth examined Ն4 categories of pregnancy January 2006), ECLA (1980 through interval were included and analyzed intervals; inadequate if the study ex- January 2006), POPLINE (1980 separately; and (3) the authors of the amined Ͻ4); (3) birth spacing mea- through January 2006), CINAHL (1982 studies adjusted their results for at least surement and inquiry of outcomes (ad- through January 2006), and LILACS maternal age and socioeconomic sta- equate if birth spacing measurement (1982 through January 2006), using a tus (measured indirectly by occupa- and ascertainment of outcomes were combination of Medical Subject Head- tion and work status, educational level, made by medical records or direct mea- ings or key word terms for birth spac- income, housing, or other variables), surement; inadequate if not); (4) blinding and adverse perinatal outcomes. because we considered these variables ing of both birth spacing status and as- Terms for birth spacing included inter- to be the most important confounding certainment of outcomes (adequate if pregnancy interval, birth interval, inter- factors in the association between birth assessment of both birth spacing sta- birth interval, pregnancy spacing, preg- spacing and adverse perinatal out- tus and outcomes was blinded; inad- nancy interval, birth spacing, intergenesic comes. Studies were excluded from the equate if not blinded or unreported); interval, birth to birth interval, birth to systematic review if they were case se- (5) loss to follow-up or exclusions (only conception interval, delivery to concep- ries or reports, editorials, letters to the for cohort and cross-sectional stud- tion interval, and interdelivery interval. editor, or reviews without original data; ies) (adequate if loss to follow-up or Terms for adverse perinatal outcomes if they exclusively used univariate nonvalid exclusions [eg, improper included perinatal outcomes, infant out- analysis; if they did not adjust for at least elimination of records] was Ͻ10%; in- comes, pregnancy outcomes, adverse out- maternal age and socioeconomic sta- adequate if Ն10% or unreported); and comes, low birth weight, preterm deliv- tus; or if they did not provide data. Stud- (6) control for confounding factors (ad- ery, preterm birth, small for gestational ies included in the systematic review equate if the study additionally con- age, intrauterine growth retardation, in- were also included in the meta- trolled for Ն2 of 5 confounding fac- trauterine growth restriction, Apgar analyses if they met the following ad- tors [parity, outcome of the most recent scores, neonatal depression, neonatal in- ditional criteria: (1) used interpreg- recognized pregnancy, access to pre- tensive care unit, fetal death, stillbirth, nancy interval as measure of birth natal care, breastfeeding, and mater- perinatal death, fetal mortality, perina- spacing; (2) provided data for 4 or more nal nutritional status]; inadequate if ad- tal mortality, perinatal morbidity, peri- interpregnancy interval strata; and (3) ditionally controlled for Ͻ2). natal outcomes, neonatal death, neona- reported odds ratio (OR) or relative risk Assessment of methodological qual- tal mortality, and neonatal outcomes. estimates and 95% confidence inter- ity of each study was carried out by 2 Proceedings of several international vals (CIs) or data to calculate them. of the authors (A.C.-A., A.R.-B.) work-

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES ing independently. Differences of opin- ported finding a nonlinear, J-shaped re- the slope gave the OR for a unit in- ion were resolved through discussion. lationship between interpregnancy increase or decrease of the interpreg- terval and the risk of adverse perinatal nancy interval (1 month). To over- Data Abstraction outcomes such as preterm birth, LBW, come the problem of assuming Data were extracted independently and SGA. The main fields in the data independence of dose-specific ORs from each article by 2 investigators set were the value x of exposures (ex- (which is incorrect, as they have a com- (A.C.-A., A.C.K.-G.) by means of a stan- pressed in months) assigned as the mid- mon reference group), we adjusted the dardized and pilot-tested data collec- points for the ranges of the reported cat- standard error of the within-study tion form. The following information egories of interpregnancy intervals and slopes estimating the covariance. was sought from each article: title, first as 1.2 times for the lower bound of the Heterogeneity of the results be- author’s name, year, geographic loca- open-ended upper categories as sug- tween the studies was formally tested tion of the study (country and re- gested by Berlin et al,10 and the y-axis with the quantity I2, which describes the gion), study design, characteristics and estimates of natural logarithm of the ad- percentage of total variation across stud- source of the study population, sample justed OR for each exposure level. ies that is due to heterogeneity rather size, measures of outcome, measure of Pooled ORs. Depending on data than chance. The I2 can be calculated birth spacing used, categorization of in- availability in the original studies, we from basic results obtained from a typi- tervals, method of data collection, ex- categorized interpregnancy interval into cal meta-analysis as posure measurement and ascertain- 6 groups: shorter than 6 months, 6 to I2=100%ϫ(Q−df)/Q ment of outcome(s), blinding of birth 11, 12 to 17, 18 to 23, 24 to 59, and 60 spacing status and ascertainment of out- months or longer. Odds ratios were used where Q is the Cochran heterogeneity come(s), loss to follow-up or invalid ex- as the measure of the relation between statistic.11 We pooled results from in- clusions, confounding factors con- interpregnancy interval and adverse dividual studies using DerSimonian and trolled for by matching or adjustment, perinatal outcomes. The interval of 18 Laird random-effects models12 be- and unadjusted and adjusted relative to 23 months was used as the referent cause moderate to high heterogeneity risks or ORs and their 95% CIs for in- category, because this was the interval (I2Ն50%) was present in the majority dividual adverse perinatal outcomes as- with the lowest risk for preterm birth, of results. sociated with all pregnancy intervals. LBW, and SGA. Data abstracted from To further explore the origin of het- each study were arranged in 2ϫ 2 erogeneity, we restricted the analyses to Statistical Analysis tables. Then, ORs with their 95% CIs subgroups of studies defined by study The studies included in our meta- for each adverse perinatal outcome con- characteristics such as study quality,date analyses differed in the units used for sidered were calculated separately for of publication, and sample size. More- measurement of the interpregnancy in- 5 predefined categories of interpreg- over, we calculated separate estimates ac- terval (days, weeks, months, or years). nancy interval. Separate analyses of the cording to race/ethnicity and study set- Therefore, we converted these differ- associations in 2ϫ2 tables were com- ting (developed vs developing countries). ent units of interpregnancy interval to bined to produce pooled unadjusted Since a number of the largest studies were months. We used 3 different meta- ORs and corresponding 95% CIs. We multinational, we could not analyze by analytical techniques to investigate also calculated pooled adjusted ORs country.Subgroup and sensitivity analy- whether a relationship exists between within each category using the esti- ses were performed pooling adjusted ORs interpregnancy interval and the risk of mated adjusted effect and its esti- provided by the studies. adverse perinatal outcomes. mated standard error (often obtained Todetect publication and location bi- Meta-regression Analysis. We first indirectly from the CI) reported in each ases, we explored asymmetry in fun- examined the shape of the dose- study. nel plots. This was examined visually, response relation between interpreg- Dose-Response Regression Slopes. and the degree of asymmetry was mea- nancy interval and risk of adverse peri- Under the assumption of indepen- sured using the Egger unweighted re- natal outcomes. For this purpose, we dence of the dose-specific OR, we es- gression asymmetry test, with PϽ.10 in- used the method proposed by Green- timated the dose-response regression dicating significant asymmetry.13 All land and Longnecker9 and Berlin et al10 slopes of each study using the OR, 95% statistical analyses were performed us- for meta-analysis of epidemiologic dose- CIs, and the midpoint of the exposure ing STATA version 8.0 (StataCorp, Col- response data. The dose-specific con- interval.9 For open-ended intervals, a lege Station, Tex). founder-adjusted natural logarithms of point 20% higher than the low end of the ORs from all studies were pooled, the interval was used. Pooled dose- RESULTS and a curve using weighted quadratic response slopes and estimates of risks One hundred thirty studies were con- spline meta-regression with no inter- were then obtained from random- sidered relevant, and the complete cept term was fitted. This method was effects models applied to the study- manuscripts were obtained. Of the 130 chosen because several studies have re- specific slopes. The exponentiation of studies, 122 were published in En-

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES glish, 4 in Spanish, 3 in French, and 1 ies varied in methodological quality, with LBW,† 13 for SGA,‡ 7 for fetal in Portuguese. Sixty-three studies were 21 cohort or cross-sectional studies death,14,27,34,55,57,63,64 and 4 for early excluded, the main reasons being lack (40%) meeting 5 or more criteria and neonatal death.34,57,63,64 of adjustment for confounding factors only 3 case-control studies (20%) meet- The dose-response association be- at statistical analysis (46%) and lack of ing 4 or more criteria. The most com- tween interpregnancy interval and the data on the relationship between birth mon shortcomings were failure to blind natural logarithm of the OR of the 5 ad- spacing and adverse outcomes consid- investigators to both exposure status and verse perinatal outcomes in cohort and ered (35%). (The list of excluded stud- ascertainment of outcome, the report of cross-sectional studies was J-shaped ies is available from the correspond- loss to follow-up or exclusions, and the (FIGURE). For preterm birth, LBW, and ing author on request.) A total of 67 categorization of pregnancy intervals. SGA, the highest risk was for intervals studies (52 cohort or cross-sectional Overall, among the studies that pro- shorter than 20 months and longer than studies14-65 and 15 case-control stud- vided data on preterm birth, 21 (18 co- 60 months. For both fetal and early neo- ies66-80), including 11 091 659 pregnan- hort or cross-sectional and 3 case- natal death, the highest risk was for in- cies, met the inclusion criteria. The control) reported an association with tervals shorter than 6 months and computerized search located 64 of the short intervals, 6 (5 cohort or cross- longer than 50 months. studies, 2 were found in proceedings of sectional and 1 case-control) an asso- Infants born to women with inter- meetings on birth spacing, and the re- ciation with long intervals, and 9 (8 co- pregnancy intervals shorter than 6 maining 1 was found through contact hort or cross-sectional and 1 case- months had pooled unadjusted ORs with a relevant researcher in the field. control) found no association. With (95% CIs) of 1.77 (1.54-2.04), 2.12 Twenty studies (30%) were con- regard to studies that reported data on (1.98-2.26), and 1.39 (1.20-1.61) for pre- ducted in the United States. The re- LBW, 20 (18 cohort or cross-sectional term birth, LBW, and SGA, respec- maining 47 were conducted in 61 coun- and 2 case-control) found an associa- tively, compared with infants born to tries from Latin America (22 countries), tion with short intervals, 7 (6 cohort or women with intervals of 18 to 23 months Asia (20 countries), Africa (11 coun- cross-sectional and 1 case-control) an (TABLE 4). Likewise, women with inter- tries), Europe (7 countries), and Aus- association with long intervals, and 6 vals of 6 to 17 months were 8% to 23% tralia. Overall agreement on the inclu- (all cohort or cross-sectional) found no more likely to give birth to infants with sion of studies was 97% (␬=0.84). association. Among the studies that pro- these adverse outcomes. Infants con- The characteristics and main find- vided data on SGA, 14 (13 cohort or ceived 60 months or more after a birth ings of the cohort and cross-sectional cross-sectional and 1 case-control) re- had ORs (95% CIs) of 1.27 (1.17-1.39) studies included in the systematic reported an association with short inter- for preterm birth, 1.49 (1.17-1.89) for view are presented in TABLE 1 (devel- vals, 6 cohort or cross-sectional stud- LBW, and 1.36 (1.20-1.54) for SGA. The oped countries) and TABLE 2 (develop- ies reported an association with long minimal increase in the risk for adverse ing countries); those of the case- intervals, and 10 (6 cohort or cross- perinatal outcomes associated with in- control studies are presented in TABLE 3. sectional and 4 case-control) found no tervals of 24 to 59 months (3%-7%) was The sample size in the cohort or cross- association. Two studies did not find not statistically significant. It was not pos- sectional studies ranged from 20144 to an association between birth spacing sible to estimate pooled ORs for the re- 4 841 418.43 The number of case partici- and low Apgar scores. With regard to lation between interpregnancy interval pants enrolled in case-control studies perinatal mortality (fetal death, early and both fetal and early neonatal death, ranged from 3671 to 416,69 and the cor- neonatal death, and perinatal death), 10 because the categories of intervals used responding number of controls ranged studies reported an association with and the reference categories did not co- from 5071,79 to 1710.67 Thirty studies pro- short intervals, 8 with long intervals, incide in all studies. vided data on preterm birth, 26 on LBW, and 7 found no association. The estimates of pooled adjusted 24 on SGA, 10 on fetal death, 4 on early It was not possible to perform a ORs were lower than estimates of neonatal death, 6 on perinatal death, and meta-analysis of the case-control stud- pooled unadjusted ORs (Table 4). 2 on low Apgar scores. Twenty-four ies because only 3 met the minimal Nevertheless, the associations between studies (36%) reported more than 1 ad- inclusion criteria. Moreover, different intervals of shorter than 6, 6 to 11, 12 verse perinatal outcome. Among the 52 categories of intervals and reference to 17, and longer than 59 months cohort or cross-sectional studies, 37 categories were used in the few stud- and preterm birth, LBW, and SGA (71%) used birth-to-conception inter- ies. Twenty-six cohort and cross- remained statistically significant. val, 14 (27%) used birth-to-birth inter- sectional studies provided data for Compared with infants of mothers val, and the remaining 1 used both in- meta-analyses. Sixteen studies pro- with interpregnancy intervals of 18 to tervals. Of the 15 case-control studies, vided data for preterm birth,* 10 for 9 (60%) used birth-to-conception inter- †References 16, 21, 27, 33, 36, 40, 47, 49, 56, 64. val and 6 (40%) birth-to-birth interval *References 25, 27, 29, 33, 36, 39, 40, 45-47, 49, 50, ‡References 21, 22, 24, 29, 33, 35, 36, 39, 45, 49, as measures of birth spacing. The stud- 52, 55, 61, 64. 52, 55, 64.

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES

Table 1. Characteristics of Cross-sectional and Cohort Studies Included in the Systematic Review of Birth Spacing and Adverse Perinatal Outcomes—Developed Countries Methodological Quality*

Lost to Interval Follow-up/ Country Sample Interval Categories, Nonvalid Source (Region) Outcome Size Used mo† Blinding Exclusions, % Confounders‡ Main Findings Fedrick and England, Late fetal death 8356 IPI Յ6; 7-12§; Not reported Not reported 1, 2, 6, 7 No relationship between Adelstein,14 Scotland, (Ն28 wk) 13-24; interval and late 1973 Wales 25-36; fetal death 37-72; 73-108; Ͼ108 Eisner et al,16 United States LBW 1 118 963 IPI Ͻ6; 6-11; Adequate 54.8 1-8 Intervals Ͻ6 mo associated 1979 (nation- 12-23; with increased risk of based Ն24 LBW for both white and [1974]) [Ͼ6]§ black women Spratley US (nation- LBW Not BI Ͻ12; 12-17; Adequate Not reported 1, 3, 5-7, Intervals Ͻ24 mo and and based) reported 18-23; 15 Ͼ59 mo associated Taffel,17 24-35; with increased risk 1981 36-47; of LBW 48-59; Ն60࿣ Brody and US (New LBW 1683 IPI Ͻ5; 5-8; Not reported 5.6 1-7, 12 Intervals Ͻ9 mo associated Bracken,20 Haven, Ն9§ with increased risk 1987 Conn) of LBW Klebanoff,21 US (multicenter) LBW, IUGR 5938 IPI Ͻ3§; 3-5.9; Not reported Not reported 1, 3, 5-7, No relationship between 1988 6-8.9; 10, 12 interval and LBW or IUGR 9-11.9; 12-14.9; 15-17.9; 18-20.9; 21-23.9; Ն24 Lieberman US (Boston, SGA 4489 IPI Յ3; 3-6; Not reported Not reported 1-3, 5, 6, Intervals Ͻ18 mo and et al,22 Mass) 6-12; 9, 10, Ͼ72 mo associated 1989 12-18; 12, 13 with increased risk 18-24; of SGA 24-36§; 36-48; 48-60; 60-72; 72-96; Ͼ96 Miller,24 Sweden IUGR 54 725 IPI Ͻ12; 12-17; Adequate Not reported 1, 2, 6, 7, 15 Intervals Ͻ12 mo associated 1989 18-23; with increased risk 24-35; of IUGR 36-47; 48-59; Ն60; [18-59]§ Lang et al,25 US (Boston, Preterm birth 4467 IPI Յ3; 4-6; Not reported Not reported 1, 3-6, 8-10, No relationship between 1990 Mass) 7-12; 12 interval and preterm birth 13-18; 19-24; 25-36§; 37-48; Ն49 Miller,27 Hungary, LBW, preterm Hungary For Hungary Ͻ12; 12-17; Adequate Not reported 1, 3-8, 10 Birth intervals Ͻ12 mo 1991 Sweden, birth, late (77 256) and 18-23; associated with increased US fetal death Sweden Sweden 24-35§; risk of LBW and preterm (Ͼ6mo (51 096) BI, and 36-47; birth. No relationship gestation) US for the 48-59; between interval and late (4290) USA IPI Ն60; fetal death Kallan,29 1992 US Preterm-LBW, 2104 IPI Ͻ7; 7-12; Not reported Not reported 1-3, 5-7, 12 Intervals Ͻ12 mo and IUGR-LBW, 13-24; Ͼ48 mo associated fetal loss 25-48; with increased risk (miscarriage Ն49§ of IUGR-LBW and and stillbirths) fetal loss but not with preterm-LBW (continued)

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES

Table 1. Characteristics of Cross-sectional and Cohort Studies Included in the Systematic Review of Birth Spacing and Adverse Perinatal Outcomes—Developed Countries (cont) Methodological Quality*

Lost to Interval Follow-up/ Country Sample Interval Categories, Nonvalid Source (Region) Outcome Size Used mo† Blinding Exclusions, % Confounders‡ Main Findings Rawlings US LBW, preterm 1922 IPI Ͻ3; 3-5.9; Not reported 9.1 1, 4, 6-8, 12 Intervals Ͻ3moandϽ9mo et al,36 birth, IUGR 6-8.9§; associated with an 1995 Ն9 increased risk of LBW and preterm birth among white and black women, respectively. No relationship between interval and IUGR Ochoa Spain LBW, preterm 279 IPI Ͻ3§; 3-5; Not reported Not reported 1, 4, 6, 7, 12 No relationship between Sangrador birth 6-8; interval and LBW and et al,38 9-11; preterm birth 1996 Ն12 Kallan,39 US (nation- Preterm birth, 1 045 393 IPI Ͻ7; 7-12; Adequate Not reported 1-4, 6-10, 12 Intervals Ͻ7moandϾ60 mo 1997 based IUGR 13-18; associated with increased [1981]) 19-24; risk of preterm birth and 25-36§; IUGR for both black and 37-48; white women 49-60; Ͼ60 Adams US (Georgia) LBW, preterm 28 273 IPI Ͻ3; 3-5; Adequate Not reported 1, 3, 6, 8 For white women, intervals et al,40 birth 6-8; Ͻ3moandϾ47 mo 1997 9-11; associated with increased 12-17; risk of LBW, whereas 18-23; intervals Ͻ6 mo and 24-35§; Ͼ47 mo associated with 36-47; increased risk of preterm Ն48 birth. For black women, intervals Ͻ6 mo and Ͼ47 mo associated with increased risk of LBW, whereas intervals Ͻ6mo were associated with increased risk of preterm birth Bakewell US (Missouri) LBW 182 285 IPI Ͻ9; Ն9§ Adequate 10.0 1, 3-8, 10, 12 For both women with prior et al,41 LBW and women with 1997 prior normal birth weight, intervals Ͻ9mo associated with increased risk of LBW Khoshnood US (nation- LBW, preterm 4 841 418 IPI Ͻ6; 6-12; Adequate 0.0 1-3, 6-8, 12 Intervals Ͻ6moandϽ12 mo et al,43 based birth Ͼ12§ associated with increased 1998 [1989- risk of LBW and preterm 1991]) birth, respectively Klerman US (Alabama) Preterm birth, 4400 IPI Ͻ3; 3-5; Adequate Not reported 1, 2, 5-8, 10, Interval Ͻ6 mo associated et al,45 IUGR 6-11; 12, 19 with increased risk of 1998 12-23; preterm birth. No Ն24 association between [Ն6]§ interval and IUGR Ekwo and US (Chicago) Preterm birth 761 IPI Յ3; 4-6; Adequate Not reported 1, 5-8, 12, 14 Intervals Յ6monot Moawad,46 7-9§; significantly associated 1998 10-24; with increased risk of Ն25 preterm birth Basso et al,47 Denmark LBW, preterm 10 187 IPI Յ4; 4.01-8; Adequate 6.2 1, 2, 6 Intervals Ͻ8.01 mo 1998 birth 8.01-12; associated with an 12.01-24; increased risk of preterm 24.01- birth. No association 36§; Ͼ36 between interval and LBW Shults et al,48 US (North Preterm birth, Preterm IPI Ͻ4; 4-12; Adequate Not reported 1-8, 12 Women with intervals Ͻ4mo 1999 Carolina) SGA birth 13-24§ had higher risks for (34 569) preterm birth and SGA SGA (27 651) (continued)

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES

Table 1. Characteristics of Cross-sectional and Cohort Studies Included in the Systematic Review of Birth Spacing and Adverse Perinatal Outcomes—Developed Countries (cont) Methodological Quality*

Lost to Interval Follow-up/ Country Sample Interval Categories, Nonvalid Source (Region) Outcome Size Used mo† Blinding Exclusions, % Confounders‡ Main Findings Zhu et al,49 US (Utah) LBW, SGA, 173 205 IPI Ͻ6; 6-11; Adequate 10.7 1-8, 10, The risk for LBW, preterm 1999 preterm birth 12-17; 12, 13 birth, and SGA increased 18-23§; with intervals Ͻ6 mo. 24-59; Intervals Ͼ59 mo 60-119; associated with increased Ն120 risks for LBW and SGA Fuentes-Afflick US (California) Preterm birth 289 842 IPI Ͻ6; 6-11; Adequate Not reported 1-3, 5-8, Intervals Ͻ18 mo and and 12-17; 11, 19 Ͼ59 mo associated Hessol,50 18-23§; with increased risk of 2000 24-59; preterm birth Ͼ59 Zhu et al,52 US (Michigan) Preterm birth, 435 327 IPI Ͻ6; 6-11; Adequate 5.1 1-8, 12, 13 Among white women, the 2001 SGA 12-17; risk for preterm birth 18-23§; increased with intervals 24-59; Ͻ12 mo and Ͼ59 mo, 60-119; whereas the risk for SGA Ն120 increased with intervals Ͻ12 mo and Ͼ23 mo. Among black women, the risk for preterm birth increased with intervals Ͻ12 mo and Ͼ119 mo, while the risk for SGA increased with intervals Ͻ6 mo and Ͼ119 mo Rousham and Australia Birth weight, 782 BI Ͻ24; Ն24§ Not reported 1.8 1, 5, 6, Birth interval not significantly Gracey,53 (Kimberley) birth length 10, 11 associated with birth 2002 weight or birth length Dafopoulos Greece Preterm birth 1 230 IPI Ͻ6; Ն6§ Not reported Not reported 1, 5, 6, Intervals of Ͻ6mo et al,54 8, 12 associated with greater 2002 risk of preterm birth Smith et al,55 Scotland Preterm birth, 89 143 IPI Ͻ6; 6-11; Adequate 13.8 1, 4, 6, 7, Interval Ͻ6moan 2003 IUGR, fetal 12-17; 10, 12 independent risk factor death (Ն24 18-23; for preterm birth. No wk gestation) 24-59 relationship between intervals Ͻ6 mo and IUGR or fetal death Zhu and Le,56 US (Michigan) LBW 565 816 IPI Ͻ6; 6-11; Adequate 21.7 1-3, 5-8, Intervals Ͻ6 mo and 2003 12-17; 12, 13 Ͼ59 mo associated 18-23§; with an increased risk 24-59; of LBW 60-95; 96-136 Stephansson Sweden Late fetal death 410 021 IPI Ͻ4; 4-7; Adequate 12.0 1, 3, 4, 6, Intervals Ն72 mo associated et al,57 (Ն28 wk 8-11; 7, 9, 11, with an increased risk of 2003 gestation), 12-35§; 12, 17 late fetal death. Intervals early neonatal 36-71; Ͻ12 mo not associated death (Յ7d Ն72 with an increased risk of of life) late fetal death or early neonatal death Abbreviations: BI, birth interval; IPI, interpregnancy interval; IUGR, intrauterine growth restriction; LBW, low birth weight; SGA, small for gestational age. *See “Methods” section for definitions of methodological quality criteria. Interval and outcomes inquiries were determined to be “adequate” for all studies. †Intervals in square brackets indicate the reference group in the studies that did not use as a reference one of the intervals originally categorized. ‡For multivariate adjustments, 1 indicates maternal age; 2, parity; 3, education; 4, marital status; 5, ethnic group or race; 6, factors related to socioeconomic status; 7, previous pregnancy outcome; 8, factors relating to prenatal care; 9, medical risk factors; 10, maternal nutritional status; 11, region; 12, smoking; 13, alcohol use; 14, illicit drug use; 15, gestational age or birth weight; 16, type of hospital; 17, year of delivery; 18, religion; 19, sex of the child. §Reference group. ࿣Reference group not specified.

23 months, those born to women with approximately 25% increased risk of comes (adjusted ORs, 1.05-1.14). On intervals shorter than 6 months had a SGA. Intervals of 6 to 17 months were the other hand, infants born to moth- 40% increased risk of preterm birth, a associated with a significantly greater ers with intervals longer than 59 60% increased risk of LBW, and an risk for the 3 adverse perinatal out- months faced a 20% to 43% increase

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES

Table 2. Characteristics of Cross-sectional and Cohort Studies Included in the Systematic Review of Birth Spacing and Adverse Perinatal Outcomes—Developing Countries Methodological Quality*

Lost to Interval Follow-up/ Country Sample Interval Categories, Nonvalid Source (Region) Outcome Size Used mo† Blinding Exclusions, % Confounders‡ Main Findings Swenson and Bangladesh Late fetal death 9295 IPI Ͻ12; 12-24; Not reported Not reported 1, 2, 6, 7 No relationship between Harper,15 (Ն28 wk) Ͼ24§ interval and late fetal death 1978 Fortney and Iran LBW, early 12 995 BI Յ12; 13-24; Not reported Not reported 1, 2, 6, 7 Intervals Ͻ12 mo associated Higgins,18 neonatal 25-36§; with increased risk of LBW 1984 death 37-48; and early neonatal death 49-60; 61-72; Ն73 DaVanzo Malaysia Birth weight 2171 BI Ͻ15; 15-23; Adequate Not reported 1, 2, 5, 6, 11, Intervals Ͻ15 mo significantly et al,19 24-35; 17 associated with reduced 1984 36-47; birth weight 48-59; 60-71; 72-83; Ն84 Casterline,23 Ivory Coast, Fetal losses 74 916 IPI Ͻ9; 9-15; Adequate Not reported 1-3, 6 Intervals Ͻ9 mo and Ͼ35 mo 1989 Tunisia, Syria, (miscarriages 16-23; associated with increased Korea, plus stillbirths) 24-35§; risk of fetal death Philippines, Ն36 Costa Rica, Mexico, Guyana Neel and Guatemala IUGR 306 BI Ͻ18; 18-35; Not reported Not reported 1, 6 Intervals Ͻ18 mo associated Alvarez,26 36-47§; with increased risk of IUGR 1991 48-59; Ն60 Huttly et al,28 Brazil (Pelotas) LBW, perinatal 3587 BI Ͻ18; 18-23; Adequate 2.0 1-3, 6, 10, 12 Intervals Ͻ24 mo associated 1992 death 24-35; with increased risk of LBW. 36-47; Intervals longer than 71 mo 48-71; associated with increased Ͼ71 risk of LBW and perinatal death Barros et al,30 Brazil (Pelotas) Preterm birth, 4747 BI Ͻ24§; Adequate 0.9 1-3, 6, 10, 12 Intervals Ͻ24 mo associated 1992 IUGR 24-35; with increased risk of IUGR. 36-48; No relationship between Ͼ48 interval and preterm birth Leong et al,31 Singapore Preterm birth 11 085 BI Ͻ20; 20-39; Not reported Not reported 1-3, 5-8 No relationship between 1993 Ն40 interval and preterm birth Gribble,32 Mexico LBW 2234 BI Յ12; 13-21; Not reported Not reported 1-3, 6, 7, 10 Intervals Ͻ22 mo associated 1993 22-30; with increased risk of LBW 31-39; 40-48; 49-57; Ն58 [22-58]§ Miller,33 1994 Philippines LBW, preterm 1155 IPI Ͻ6; 6-11; Not reported Not reported 1, 3, 6-8, 10, Intervals Ͻ6 mo associated birth, SGA 12-17; 12 with increased risk of LBW, 18-23; preterm birth, and SGA in 24-47; fifth or higher birth order Ն48 infants. No excess risk among lower-order infants Greenwood Jamaica Perinatal death 7512 IPI Ͻ12; 12-23; Not reported 15.0 1-3, 6, 7, 9 No relationship between et al,34 24-59; interval and perinatal death 1994 Ն60 Fikree and Pakistan IUGR 624 IPI Ͻ13; Not reported Not reported 1-3, 6, 7, 10, Intervals Ͻ13 mo associated Berendes,35 13-24§; 18 with increased risk of IUGR 1994 25-36; Ն37 Fourn et al,37 Benin IUGR, preterm 2862 BI Ͻ12; 12-23; Not reported Not reported 1-4, 6, 7, 10 No relationship between 1996 birth 24-35; interval and preterm birth Ͼ35§ or IUGR (continued)

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES

Table 2. Characteristics of Cross-sectional and Cohort Studies Included in the Systematic Review of Birth Spacing and Adverse Perinatal Outcomes—Developing Countries (cont) Methodological Quality*

Lost to Interval Follow-up/ Country Sample Interval Categories, Nonvalid Source (Region) Outcome Size Used mo† Blinding Exclusions, % Confounders‡ Main Findings SÈener et al,42 Turkey SGA 420 IPI Ͻ12;13-24; Not reported 2.1 1-3, 6-8, Intervals Ͻ12 mo associated 1997 25-48§; 10, 12 with increased risk of SGA Ͼ48 Deshmukh India LBW 201 BI Not reported Not reported 1.0 1, 2, 6, “Short interval” associated et al,44 10, 12 with increased risk of LBW 1998 Sachar and India (Punjab) Perinatal death 2424 IPI Ͻ23; Ն24§ Not reported Not reported 1-3, 6-8, Intervals Ͻ23 mo associated Soni,51 (deaths from 10, 11 with an increased risk of 2000 the 28th perinatal death week gestation through first week of life) van Eijk et al,58 Kenya LBW, preterm 2218 IPI Ͻ6; 6-23; Not reported 6.5 1-3, 6, 19 No association between 2004 birth, SGA Ն24 interval and LBW, preterm birth, and SGA Arafa et al,59 Egypt Preterm birth 1202 IPI Ͻ12; 12-36; Not reported 0.0 1, 2, 6, No relationship between 2004 37-48; 8-10 intervals Ͻ12 mo and 49-60; preterm birth Ͼ60§ Pedroso Brazil LBW, preterm 15 314 BI Ͻ25; 25-36; Not reported Not reported 1-8, 10, Intervals Ͻ24 mo associated et al,60 2004 (Campinas) birth Ն37§ 12, 13 with increased risk of LBW and preterm birth Hsieh et al,61 Taiwan Preterm birth 4072 IPI Ͻ6; 6-11; Adequate Not reported 1-4, 6, Intervals Ͻ12 mo associated 2005 12-18; 7, 12 with increased risk of 18-48§ ; preterm birth Ͼ48 DaVanzo Bangladesh Early neonatal 125 720 BI Ͻ15; 15-17; Adequate Not reported 1, 2, 6, Intervals Ͻ24 mo and et al,62 (Matlab) death (Յ7d) 18-23; 7, 11, Ն84 mo associated 2005 24-35; 17, 18 with increased risk 36-59§; of early neonatal death 60-83; Ն84 Rutstein,63 9 Asian Perinatal death 267 261 IPI Ͻ6; 6-11; Adequate Not reported 1-3, 6, Intervals Ͻ18 mo and 2005 countries, (deaths from 12-17; 7, 11 Ն42 mo significantly 7 Latin the 28th 18-23; associated with increased American week of 24-29§; risk of perinatal death countries, and pregnancy 30-35; 4 African through first 36-41; countries week of life) 42-47; 48-53; 54-59 Conde- 18 Latin LBW, preterm 1 080 650 IPI Ͻ6; 6-11; Adequate 10.7 1-4, 6-12, Intervals Ͻ12 mo and Agudelo American birth, SGA, 12-17; 15-17 Ͼ59 mo significantly et al,64 2005 countries fetal death, 18-23§; associated with increased early neonatal 24-35; risks of LBW, preterm death, low 36-47; birth, SGA, fetal death, Apgar scores 48-59; and early neonatal death at 5 min Ն60 Hosain et al,65 Bangladesh LBW 227 BI Ͻ24; Ն24§ Not reported 16.4 1-3, 6, 8, No relationship between 2005 10, 12 intervals Ͻ24 mo and LBW Abbreviations: BI, birth interval; IPI, interpregnancy interval; IUGR, intrauterine growth restriction; LBW, low birth weight; SGA, small for gestational age. *See “Methods” section for definitions of methodological quality criteria. Interval and outcomes inquiries were determined to be “adequate” for all studies except Swenson and Harper (not reported). †Intervals in square brackets indicate the reference group in the studies that did not use as a reference one of the intervals originally categorized. ‡For multivariate adjustments, 1 indicates maternal age; 2, parity; 3, education; 4, marital status; 5, ethnic group or race; 6, factors related to socioeconomic status; 7, previous pregnancy outcome; 8, factors relating to prenatal care; 9, medical risk factors; 10, maternal nutritional status; 11, region; 12, smoking; 13, alcohol use; 14, illicit drug use; 15, gestational age or birth weight; 16, type of hospital; 17, year of delivery; 18, religion; 19, sex of the child. §Reference group. in risk of the 3 adverse perinatal out- between women with intervals of 24 showed no asymmetry, either visually comes. There were no differences in to 59 months and those with intervals (funnel plots available from the corre- the risk of adverse perinatal outcomes of 18 to 23 months. All funnel plots sponding author on request) or in

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES

Table 3. Characteristics of Case-Control Studies Included in the Systematic Review of Birth Spacing and Adverse Perinatal Outcomes Methodological Quality*

Intervals Country No. of Cases, No. of Controls, Interval Categories, Source (Region) Source Source Used mo Blinding Confounders† Main Findings Ruelas- Mexico 104 perinatal deaths 208 live births, IPI Ͻ7; 7-46‡; Not reported 1-4, 6, 7, 9, 10, Intervals Ͻ7 mo and Orozco (from the 27th week of community Ͼ46 12, 13 Ͼ46 mo significantly et al,66 pregnancy through associated with increased 1985 first week of life), risk of perinatal death community Ferraz et al,67 Brazil (Natal) 303 infants with birth 1710 term infants with IPI Յ6; 7-12; Not reported 1-3, 6, 7, 10, Intervals Յ6 mo associated 1988 weight Ͻ10th percentile birth weight Ն2500 g Ն13‡ 12 with increased risk of for gestational and appropriate for IUGR. No relationship age, 282 preterm gestational age, between interval and infants, hospital hospital preterm birth Bartlett and Guatemala 42 Mayan Indian infants 54 Mayan Indian infants BI Ͻ14; Ն14‡ Not reported 1-4, 6, 8, 11 Interval Ͻ14 mo associated Paz de who died during birth who survived the with increased risk of Bocaletti,68 or in the first month first month of intrapartum and 1991 of life, community life, community neonatal deaths Mavalankar India 416 preterm-LBW 926 normal birth weight IPI Յ6; 7-12; Not reported 1-4, 6-12, 15, Intervals Ͻ13 mo and and infants, hospital infants, hospital 13-24; 18, 19 Ͼ48 mo significantly Gray,69 25-48‡; associated with increased 1991 Ն49 risk of preterm LBW Dechering Sri Lanka 245 LBW infants 399 with birth weight BI Ͻ12; Not reported 1-3, 6-8, 10, Intervals Ͻ12 mo and and Ն2500 g 12-48‡; 19 Ͼ48 mo significantly Perera,70 Ͼ48 associated with increased 1991 risk of LBW Kumar and India 36 late fetal deaths, 50 live birth infants, BI Ͻ24; Ն24‡ Not reported 1-3, 6 Intervals Ͻ24 mo not Singhi,71 community community associated with increased 1992 risk of late fetal death Mavalankar India 343 term-LBW 926 normal birth weight IPI Յ6; 7-12; Not reported 1-4, 6-12, 15, No relationship between et al,72 (IUGR) infants infants, hospital 13-48‡; 18, 19 interval and IUGR 1992 Ն49 Arif et al,73 Pakistan 236 LBW-SGA 293 infants with birth BI Ͻ24; Ն24‡ Not reported 1, 3, 6, 10 Intervals Ͻ24 mo not 1998 infants, hospital weight Ն2500 g, associated with risk of hospital LBW-SGA birth Grau et al,74 Cuba 202 preterm infants, 319 term infants, IPI Ͻ24; Ն24‡ Not reported 1, 3, 4, 6-8, 10, Intervals Ͻ24 mo associated 1999 hospital hospital 12 with increased risk of preterm birth Wang Taiwan 208 perinatal deaths 619 live births, hospital BI Ͻ12; 12-24; Not reported 1-3, 6, 8-10, Intervals Ͻ24 mo associated and Lin,75 (from the 22nd week of Ͼ24‡ 14 with increased risk of 1999 pregnancy through first perinatal death week of life), hospital Mafina- Congo 247 infants with birth 293 infants with birth IPI Ͻ12; 12-24; Not reported 1-4, 6, 10 No relationship between Mienandi weight Ͻ10th percentile weight appropriate Ͼ24‡ interval and IUGR et al,76 for gestational for gestational age, 2002 age, hospital hospital Al-Jasmi United Arab 128 preterm infants, 128 term infants, IPI 2.8-8.9; Not reported 1, 2, 6-8, 10 Intervals Ͻ16 mo associated et al,77 Emirates hospital hospital 9.0-15.9; with increased risk of 2002 16.0- preterm birth 22.9‡; 23.0-82.7 Khan and Pakistan 190 LBW infants, hospital 760 normal birth weight IPI Ͻ5; 5-10; Not reported 1-3, 6, 8, 10 Risk for LBW increased with Jamal,78 infants, hospital Ͼ10‡ intervals Ͻ5mo 2003 Orji et al,79 Nigeria 50 women with intervals 50 women with intervals BI 24-60‡; Not reported 1, 2, 6 No difference in Apgar 2004 Ն72 months, hospital 24 to 60 months, Ն72 scores between the hospital study groups Kleijer et al,80 Australia 233 infants with birth 241 infants with birth IPI Ͻ48; Ն48‡ Not reported 1, 2, 4-7, 10, No relationship between 2005 (Adelaide) weight Ͻ10th percentile weight between the 12, 14 interval Ͻ48 mo for gestational 25th and 75th and IUGR age, hospital percentile Abbreviations: BI, birth interval; IPI, interpregnancy interval; IUGR, intrauterine growth restriction; LBW, low birth weight; SGA, small for gestational age. *See “Methods” section for definitions of methodological quality criteria. Interval and outcomes inquiries were determined to be “adequate” for all studies except Bartlett and Paz de Bocaletti (inadequate). †For multivariate adjustments, 1 indicates maternal age; 2, parity; 3, education; 4, marital status; 5, ethnic group or race; 6, factors related to socioeconomic status; 7, previous pregnancy outcome; 8, factors relating to prenatal care; 9, medical risk factors; 10, maternal nutritional status; 11, region; 12, smoking; 13, alcohol use; 14, illicit drug use; 15, gestational age or birth weight; 16, type of hospital; 17, year of delivery; 18, religion; 19, sex of the child. ‡Reference group.

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES

Figure. Scatterplot of Natural Logarithm Odds Ratio and Meta-regression Curves of Adverse Perinatal Outcomes According to Interpregnancy Interval in Cohort and Cross-sectional Studies

Preterm Birth Low Birth Weight 1.8 1.8

1.6 1.6

1.4 1.4

1.2 1.2

1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4

Natural Logarithm Odds Ratio 0.2 0.2

0 0

–0.2 –0.2 0 50 100 150 50 100 150 Interpregnancy Interval, mo Interpregnancy Interval, mo

Small for Gestational Age Fetal Death 1.8 1.8

1.6 1.6

1.4 1.4

1.2 1.2

1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4

Natural Logarithm Odds Ratio 0.2 0.2

0 0

–0.2 –0.2 0 50 100 150 0 50 100 150 Interpregnancy Interval, mo Interpregnancy Interval, mo

Early Neonatal Death 1.8

1.6

1.4

1.2

1.0

0.8

0.6

0.4

Natural Logarithm Odds Ratio 0.2

–0.2 0 50 100 150 Interpregnancy Interval, mo

The dose-response curve line represents estimates from a smoothed spline regression. The horizontal line at y=0 represents no effect. Most studies provided Ն1 odds ratio estimate for several categories of interpregnancy intervals.

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES terms of statistical significance (PϾ.10 firmed by I2 values greater than 50% in geneity among studies found that a sig- for all, by Egger test). half of meta-analyses, and this re- nificant portion of the heterogeneity in Important statistical heterogeneity mained in the prespecified subgroups. studies evaluating the relation be- among studies was present, as con- An examination for sources of hetero- tween intervals shorter than 6 months and both preterm birth and LBW was explained by the study by our group64 Table 4. Odds Ratios for the Association Between Interpregnancy Interval and Adverse Perinatal Outcomes in Cohort and Cross-sectional Studies since the estimates of pooled adjusted ORs were significantly lowered when Interpregnancy Small for Gestational Interval, mo Preterm Birth Low Birth Weight Age this study was excluded (1.30; 95% CI, Pooled Unadjusted Results 1.23-1.38; and 1.48; 95% CI, 1.40- Ͻ6 1.57, respectively). Study quality, date No. of Studies 825,33,40,49,50,52,55,64 621,33,40,49,56,64 721,22,33,49,52,55,64 of publication, sample size, and study OR (95% CI) 1.77 (1.54-2.04) 2.12 (1.98-2.26) 1.39 (1.20-1.61) setting provided no explanation for het- I2 ,%* 95 63 93 erogeneity in studies evaluating the re- 6-11 No. of studies 925,27,33,40,49,50,52,55,64 621,33,40,49,56,64 821,22,24,33,49,52,55,64 lationship between intervals of 6 to 11, OR (95% CI) 1.23 (1.16-1.31) 1.23 (1.15-1.32) 1.18 (1.14-1.23) 12 to 17, and 24 to 59 months and both I2,%* 85 73 39 preterm birth and LBW, because the CIs 12-17 in the subgroups overlapped (data avail- No. of studies 925,27,33,40,49,50,52,55,64 721,27,33,40,49,56,64 821,22,24,33,49,52,55,64 able from corresponding author on re- OR (95% CI) 1.11 (1.03-1.20) 1.08 (1.02-1.14) 1.08 (1.06-1.11) quest). Compared with the overall re- I2,%* 56 51 14 sults, the summary ORs calculated from 18-23† No. of studies 925,27,33,40,49,50,52,55,64 721,27,33,40,49,56,64 821,22,24,33,49,52,55,64 sensitivity and subgroup analyses were OR 1.00 1.00 1.00 almost identical. Pooled adjusted ORs 24-59 calculated from subgroups evaluating No. of studies 627,49,50,52,55,64 627,33,40,49,56,64 722,24,33,49,52,55,64 intervals of 60 months or longer and OR (95% CI) 1.03 (1.00-1.07) 1.07 (0.99-1.15) 1.07 (0.98-1.18) preterm birth were similar to the over- I2,%* 28 78 93 all adjusted OR calculated from all stud- Ն60 ies. With regard to studies evaluating 27,49,50,52,64 27,49,56,64 22,24,49,52,64 No. of studies 5 4 5 the association between intervals longer OR (95% CI) 1.27 (1.17-1.39) 1.49 (1.17-1.89) 1.36 (1.20-1.54) than 59 months and LBW, studies from I2,%* 93 98 96 developed countries were signifi- Pooled Adjusted Results cantly associated with higher pooled ad- Ͻ6 No. of studies 825,39,40,49,50,52,55,64 440,49,52,64 622,39,49,52,55,64 justed ORs. There were no significant OR (95% CI) 1.40 (1.24-1.58) 1.61 (1.39-1.86) 1.26 (1.18-1.33) differences in pooled adjusted ORs ob- I2,%* 69 87 89 tained from subgroups of studies and 6-11 the overall estimates obtained from all No. of studies 825,39,40,49,50,52,55,64 440,49,52,64 722,24,39,49,52,55,64 studies assessing the association be- OR (95% CI) 1.14 (1.10-1.17) 1.14 (1.10-1.18) 1.11 (1.04-1.19) tween interpregnancy interval and SGA. I2,%* 87 91 32 In general, there were no significant dif- 12-17 ferences in estimates of summary ad- No. of studies 825,39,40,49,50,52,55,64 440,49,52,64 722,24,39,49,52,55,64 OR (95% CI) 1.07 (1.03-1.11) 1.05 (1.01-1.09) 1.06 (1.01-1.10) justed ORs between white and black I2,%* 26 34 0 women in subgroups that evaluated the 18-23† effects of interpregnancy interval on ad- No. of studies 825,39,40,49,50,52,55,64 440,49,52,64 722,24,39,49,52,55,64 verse perinatal outcomes according to OR 1.00 1.00 1.00 race/ethnicity.39,40,52 24-59 For each month that interpreg- No. of studies 825,39,40,49,50,52,55,64 440,49,52,64 722,24,39,49,52,55,64 nancy interval was shortened from 18 OR (95% CI) 0.99 (0.97-1.02) 1.01 (0.98-1.03) 1.02 (0.99-1.05) months, the risk increase for preterm 2 I ,%* 000birth, LBW, and SGA was 1.9%, 3.3%, Ն60 No. of studies 725,39,40,49,50,52,64 440,49,52,64 622,24,39,49,52,64 and 1.5%, respectively (TABLE 5). On OR (95% CI) 1.20 (1.17-1.24) 1.43 (1.27-1.62) 1.29 (1.20-1.39) the other hand, the risk for the 3 ad- I2,%* 95 84 88 verse perinatal outcomes increased by Abbreviations: CI, confidence interval; OR, odds ratio. 0.6%, 0.9%, and 0.8%, respectively, for *Heterogeneity test (see “Methods” section). each month that interpregnancy inter- †Reference category. val was lengthened from 59 months.

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES

COMMENT Table 5. Meta-analysis of Dose-Response Regression Slopes and Prediction of the Risk of Using 3 different meta-analytical tech- Adverse Perinatal Outcomes for Interpregnancy Intervals Ͻ18 Months and Ͼ59 Months niques, we show that birth to concep- Increase, % (95% CI) tion intervals shorter than 18 months Preterm Birth LBW SGA and longer than 59 months are signifi- Risk Increase (12 Studies) (7 Studies) (12 Studies) cantly associated with increased risk of Per month for intervals Ͻ18 mo* 1.92 (1.80-3.04) 3.25 (3.09-3.41) 1.52 (1.40-1.64) several adverse perinatal outcomes, Per month for intervals Ͼ59 mo† 0.55 (0.49-0.61) 0.91 (0.83-0.99) 0.76 (0.71-0.81) such as preterm birth, LBW, and SGA. Predicted by the model Infants can have LBW either because Interpregnancy interval, mo they are born early (preterm birth) or 3 28.8 (27.0-30.6) 48.8 (46.4-51.2) 22.8 (21.0-24.6) are born SGA. Thus, the association be- 6 23.0 (21.6-24.5) 39.0 (37.1-40.9) 18.2 (16.8-19.7) tween interpregnancy interval and LBW 9 17.3 (16.2-18.4) 29.3 (27.8-30.7) 13.7 (12.6-14.8) could be due to the independent effect 12 11.5 (10.8-12.2) 19.5 (18.5-20.5) 9.1 (8.4-9.8) of interval on both preterm birth and 15 5.8 (5.4-6.1) 9.8 (9.3-10.2) 4.6 (4.2-4.9) SGA. Less clear is the association be- 18-59‡ 1.00 1.00 1.00 tween birth spacing and the risk of fe- 72 6.6 (5.9-7.3) 10.9 (10.0-11.9) 9.1 (8.5-9.7) tal and early neonatal death, although 96 19.8 (17.6-22.0) 32.8 (29.9-35.6) 27.4 (25.6-29.2) 120 33.0 (29.4-36.6) 54.6 (49.8-59.4) 45.6 (42.6-48.6) results from meta-regression curves 144 46.2 (41.2-51.2) 76.4 (69.7-83.2) 63.8 (59.6-68.0) suggest that interpregnancy intervals Abbreviations: CI, confidence interval; LBW, low birth weight; SGA, small for gestational age. shorter than 6 months and longer than *Risk increase per each month that interpregnancy interval is shortened from 18 months. †Risk increase per each month that interpregnancy interval is lengthened from 59 months. 50 months are associated with in- ‡Reference category. creased risk of these adverse perinatal outcomes. The strength of our infer- ences is based on compliance with strin- stresses of the next. This results in Some hypotheses have also been pro- gent criteria for performing a rigorous depletion of maternal nutrient stores, posed to explain the relationship be- systematic review. These included the with the subsequent increased risk of tween long intervals and adverse peri- use of a prospective protocol designed adverse perinatal outcomes.81 The fo- natal outcomes. Zhu et al49 have to address a research question; the late depletion hypothesis claims that hypothesized that, after delivery, a wom- methods used in the identification of maternal serum and erythrocyte con- an’s physiologic reproductive capaci- relevant studies; no language restric- centrations of folate decrease from the ties gradually decline, becoming simi- tions; the exclusion of studies that fifth month of pregnancy onward and lar to those of primigravid women (ie, did not adjust for at least maternal age remain low for a fairly long time after “the physiological regression hypoth- and socioeconomic status; the strict as- delivery. Women who become preg- esis”). This hypothesis is supported by sessment of methodological quality of nant before folate restoration is com- the observation that perinatal out- included studies; the use of several tech- plete have an increased risk of folate in- comes for infants conceived after an ex- niques of meta-analysis (both unad- sufficiency at the time of conception cessively long interpregnancy interval justed and adjusted analyses); the ex- and during pregnancy. As a conse- are similar to outcomes of infants born ploration of sources of heterogeneity; quence, their offspring have higher risks to primigravid women. Another possi- the quantitative summarization of the of neural tube defects, intrauterine bility is that unmeasured factors, such evidence; and the inclusion of a large growth restriction, preterm birth, and as sexually transmitted infections or ma- number of women from different popu- LBW.82 Some investigators have attrib- ternal illnesses, may cause both ad- lations throughout the world. uted the higher risk of poor preg- verse fertility and pregnancy out- The reasons for the association nancy outcomes to several factors as- comes.5,49 These factors could differ for between a short interval between preg- sociated with having short intervals, women in developed and developing nancies and adverse perinatal out- such as socioeconomic status, un- countries. Finally, residual confound- comes are unclear. A plausible expla- stable lifestyles, failure to use health care ing may still be an explanation for at least nation is the maternal nutritional services or inadequate use of such ser- part of the reported associations. depletion hypothesis,27,81 which states vices, unplanned pregnancies, and other Several potential limitations of our that a close succession of pregnancies behavioral or psychological determi- review must also be considered. First, and periods of lactation worsen the nants.4,5 However, the fact that the birth like any systematic review, it is lim- mother’s nutritional status because spacing effects are not strongly attenu- ited by the quality of original data. The there is not adequate time for the ated when socioeconomic and mater- great majority of studies calculated the mother to recover from the physiologi- nal characteristics are controlled for interpregnancy interval using moth- cal stresses of the preceding preg- suggests that the effects are not caused er’s recall of her previous child’s date nancy before she is subjected to the by these confounding factors. of birth and her last menstrual period,

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES instead of birth dates recorded on the improve the health of both mother and more, Md), for their relevant contributions to previous versions of the article; Eileen D’Andrea, MLS, birth records and gestational age esti- the next infant. Jennifer Gelman, MLS, and Juan Carlos Toscano, BS mated from ultrasonography. In most Despite the advances during the last (Academy for Educational Development, Washing- ton, DC) and Laurel Suiter, MA (CATALYST Consor- studies the intervals were calculated as 2 decades in understanding the rela- tium), for their assistance in conducting the comput- the time elapsed between 2 consecu- tionship between birth spacing and ad- erized literature search and obtaining the articles; and tive live births, ignoring induced or verse pregnancy outcomes, little infor- Ed Abel, MPA, Elsa Berhane, MPH, and Caroline Tran, MA (CATALYST Consortium), and Paula Hollerbach, spontaneous abortions or fetal deaths mation is available to explain the PhD, Amy Uccello, MA, and Margaret McGunnigle, between them, which can produce even mechanisms by which birth spacing MPH (Academy for Educational Development), for their administrative support in the organization of the re- longer intervals between live births. might improve the health of mothers view. Nevertheless, this problem would not and their children. Also, more studies affect the findings for short intervals. are needed on whether the effects of REFERENCES In addition, several studies did not birth spacing on perinatal health dif- 1. Winikoff B. The effects of birth spacing on child properly address the potential con- fer in developed vs developing na- and maternal health. Stud Fam Plann. 1983;14:231- founding effects of factors other than tions. Finally, it is imperative to un- 245. 2. Rousso D, Panidis D, Gkoutzioulis F, Kourtis A, maternal age and socioeconomic sta- derstand the causes for both short and Mavromatidis G, Kalahanis I. Effect of the interval be- tus. Second, because there was consid- long intervals in any population to in- tween pregnancies on the health of mother and child. Eur J Obstet Gynecol Reprod Biol. 2002;105:4-6. erable statistical heterogeneity in most terpret the data on health risks. The 3. King JC. The risk of maternal nutritional depletion of the meta-analyses performed, our consequence of this may be that fam- and poor outcomes increases in early or closely spaced pregnancies. J Nutr. 2003;133:1732S-1736S. findings should be interpreted with cau- ily planning policies and messages may 4. Erickson JD, Bjerkedal T. Interval between tion. Nevertheless, in the great major- need to be tailored to different popu- pregnancies. Lancet. 1979;1:52. 5. Klebanoff MA. The interval between pregnancies ity of comparisons the estimates showed lations. and the outcome of subsequent births. N Engl J Med. the same direction of effect, which Author Contributions: Dr Conde-Agudelo had full ac- 1999;340:643-644. could suggest the absence of clinical cess to all of the data in the study and takes respon- 6. Stroup DF, Berlin JA, Morton SC, et al; Meta- analysis Of Observational Studies in Epidemiology heterogeneity among the studies. In- sibility for the integrity of the data and the accuracy of the data analysis. (MOOSE) Group. Meta-analysis of observational studies in epidemiology: a proposal for reporting. JAMA. vestigation of possible sources of het- Study concept and design: Conde-Agudelo. 2000;283:2008-2012. Acquisition of data: Conde-Agudelo, Rosas-Bermu´dez. erogeneity provided no plausible ex- 7. Levine M, Walter S, Lee H, Haines T, Holbrook A, Analysis and interpretation of data: Conde-Agudelo, Moyer V; Evidence-Based Medicine Working Group. planations. In addition, it is possible Rosas-Bermu´dez, Kafury-Goeta. Users’ guides to the medical literature, IV: how to use 2 Drafting of the manuscript: Conde-Agudelo, that the I heterogeneity test could have an article about harm. JAMA. 1994;271:1615-1619. Rosas-Bermu´dez, Kafury-Goeta. 8. Downs SH, Black N. The feasibility of creating a excessive power when there are stud- Critical revision of the manuscript for important in- checklist for the assessment of the methodological qual- tellectual content: Conde-Agudelo, Rosas-Bermu´dez, ies with large sample size, as was the ity both of randomised and non-randomised studies Kafury-Goeta. of health care interventions. J Epidemiol Community case with some of the ones included in Statistical analysis: Conde-Agudelo, Rosas-Bermu´dez, Health. 1998;52:377-384. our meta-analyses. Third, the number Kafury-Goeta. 9. Greenland S, Longnecker MP. Methods for trend of studies available for analysis on the Obtained funding: Conde-Agudelo. estimation from summarized dose-response data, with Administrative, technical, or material support: relationship between birth spacing and applications to meta-analysis. Am J Epidemiol. 1992; Rosas-Bermu´dez, Kafury-Goeta. 135:1301-1309. some adverse perinatal outcomes is still Study supervision: Conde-Agudelo. 10. Berlin JA, Longnecker MP, Greenland S. Meta- too small to provide conclusive evi- Financial Disclosures: None reported. analysis of epidemiologic dose-response data. Funding/Support: This work was supported by the Of- Epidemiology. 1993;4:218-228. dence. fice of Population and Reproductive Health, Bureau 11. Higgins JPT, Deeks JJ, Altman DG. Measuring The effects of birth spacing on peri- for Global Health, US Agency for International De- inconsistency in meta-analyses. BMJ. 2003;327:557- velopment, under the terms of Cooperative Agree- 560. natal health found in our study, as well ments HRN-A-00-00-00003-00 and GPO-A-00-05- 12. DerSimonian R, Laird N. Meta-analysis in clinical as the effects of both short and long in- 00027-0 awarded to the CATALYST Consortium. trials. Control Clin Trials. 1986;7:177-188. Role of the Sponsor: The funding organization had 13. Egger M, Davey Smith G, Schneider M, Minder tervals on infant, child, and maternal no role in the design and conduct of the study; the C. Bias in meta-analyses detected by a simple graphi- health,1,2 should furnish a strong mo- collection, analysis, and interpretation of the data; or cal test. BMJ. 1997;315:629-634. the preparation, review, or approval of the manu- 14. Fedrick J, Adelstein P. Influence of pregnancy tivating force for health personnel to script. spacing on outcome of pregnancy. BMJ. 1973;4:753- provide family planning. The health Disclaimer: The author’s views expressed herein do 756. sector should supply such care not only not necessarily reflect the views of the US Agency for 15. Swenson I, Harper PA. The relationship between International Development or the US government. fetal wastage and pregnancy spacing in Bangladesh. to those wishing to limit their fertility Acknowledgment: We thank Maureen Norton, PhD Soc Biol. 1978;25:251-257. for personal, social, or economic rea- (Office of Population and Reproductive Health, Bu- 16. Eisner V, Brazie JV, Pratt MW, Hexter AC. The reau for Global Health, USAID, Washington, DC), risk of low birthweight. Am J Public Health. 1979;69: sons, but should also provide the Taroub Faramand, MD, MPH (CATALYST Consor- 887-893. needed services to those practicing fam- tium, Washington, DC), Jose´Beliza´n, MD, PhD, and 17. Spratley E, Taffel S. Interval between births: United Eduardo Bergel, PhD (Institute for Clinical Effective- States, 1970-77. Vital Health Stat 21. 1981;39:1-47. ily planning for health reasons. The re- ness and Health Policy, Buenos Aires, Argentina), Bao- 18. Fortney JA, Higgins JE. The effect of birth inter- sults of our systematic review could be Ping Zhu, MD, MS (Missouri Department of Health val on perinatal survival and birth weight. Public Health. and Senior Services, Jefferson City), Julie DaVanzo, PhD 1984;98:73-83. used by reproductive clinicians around (RAND Corporation, Santa Monica, Calif ), Roger Ro- 19. DaVanzo J, Habicht JP, Butz WP. Assessing so- the world to advise women on the ben- chat, PhD (Rollins School of Public Health, Atlanta, Ga), cioeconomic correlates of birthweight in peninsular Ma- Linda Adair, PhD (University of North Carolina at laysia: ethnic differences and changes over time. Soc efits of delaying a subsequent preg- Chapel Hill), and Stan Becker, PhD ( Johns Hopkins Uni- Sci Med. 1984;18:387-404. nancy for approximately 2 to 5 years to versity School of Hygiene and Public Health, Balti- 20. Brody DJ, Bracken MB. Short interpregnancy in-

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011 BIRTH SPACING AND RISK OF ADVERSE PERINATAL OUTCOMES terval: a risk factor for low birthweight. Am J Perinatol. Enu¨nlu¨ T. Risk factors associated with small for ges- tality in Matlab, Bangladesh. In: Technical Consulta- 1987;4:50-54. tational age infants in a Turkish population. J Matern tion and Review of the Scientific Evidence for Birth 21. Klebanoff MA. Short interpregnancy interval and Fetal Invest. 1997;7:145-151. Spacing. Geneva, Switzerland: World Health Orga- the risk of low birthweight. Am J Public Health. 1988; 43. Khoshnood B, Lee KS, Wall S, Hsieh HL, Mitten- nization; 2005. 78:667-670. dorf R. Short interpregnancy intervals and the risk of 63. Rutstein O. Effect of preceding pregnancy inter- 22. Lieberman E, Lang JM, Ryan KJ, Monson RR, adverse birth outcomes among five racial/ethnic groups val on fetal and perinatal mortality: multivariate cross- Schoenbaum SC. The association of inter-pregnancy in the United States. Am J Epidemiol. 1998;148:798- country analyses. In: Technical Consultation and Re- interval with small for gestational age births. Obstet 805. view of the Scientific Evidence for Birth Spacing. Gynecol. 1989;74:1-5. 44. Deshmukh JS, Motghare DD, Zodpey SP, Wadhva Geneva, Switzerland: World Health Organization; 23. Casterline JB. Maternal age, gravidity, and preg- SK. Low birth weight and associated maternal factors 2005. nancy spacing effects on spontaneous fetal mortality. in an urban area. Indian Pediatr. 1998;35:33-36. 64. Conde-Agudelo A, Belizań JM, Norton M, Rosas- Soc Biol. 1989;36:186-212. 45. Klerman LV, Cliver SP, Goldenberg RL. The im- Bermu´dez A. Effect of the interpregnancy interval on 24. Miller JE. Determinants of intrauterine growth re- pact of short interpregnancy intervals on pregnancy perinatal outcomes in Latin America. Obstet Gynecol. tardation: evidence against maternal depletion. J Bio- outcomes in a low-income population. Am J Public 2005;106:359-366. soc Sci. 1989;21:235-243. Health. 1998;88:1182-1185. 65. Hosain GM, Chatterjee N, Begum A, Saha SC. Fac- 25. Lang JM, Lieberman E, Ryan KJ, Monson RR. In- 46. Ekwo EE, Moawad A. The relationship of inter- tors associated with low birthweight in rural Bangla- terpregnancy interval and risk of preterm labor. Am J pregnancy interval to the risk of preterm births to black desh [pushed online ahead of print October 19, 2005]. Epidemiol. 1990;132:304-309. and white women. Int J Epidemiol. 1998;27:68-73. J Trop Pediatr. 2005:662. 26. Neel NR, Alvarez JO. Risk factors of fetal malnu- 47. Basso O, Olsen J, Knudsen LB, Christensen K. Low 66. Ruelas-Orozco G, Guzman J, Malacara JM. Peri- trition in a group of Guatemalan mothers and neo- birth weight and preterm birth after short interpreg- natal mortality risk factors in a case-control study [in nates [in Spanish]. Bol Oficina Sanit Panam. 1991;110: nancy intervals. Am J Obstet Gynecol. 1998;178:259- Spanish]. Bol Med Hosp Infant Mex. 1985;42:153-158. 93-107. 263. 67. Ferraz EM, Gray RH, Fleming PL, Maia TM. In- 27. Miller JE. Birth intervals and perinatal health: an 48. Shults RA, Arndt V, Olshan AF, Martin CF, Royce terpregnancy interval and low birth weight: findings investigation of three hypotheses. Fam Plann Perspect. RA. Effects of short interpregnancy intervals on small- from a case-control study. Am J Epidemiol. 1988;128: 1991;23:62-70. for gestational age and preterm births. Epidemiology. 1111-1116. 28. Huttly SR, Victora CG, Barros FC, Vaughan JP. Birth 1999;10:250-254. 68. Bartlett AV, Paz de Bocaletti ME. Intrapartum and spacing and child health in urban Brazilian children. 49. Zhu BP, Rolfs RT, Nangle BE, Horan JM. Effect neonatal mortality in a traditional indigenous com- Pediatrics. 1992;89:1049-1054. of the interval between pregnancies on perinatal munity in rural Guatemala. Acta Paediatr Scand. 1991; 29. Kallan JE. Effects of interpregnancy intervals on outcomes. N Engl J Med. 1999;340:589-594. 80:288-296. preterm birth, intrauterine growth retardation, and fe- 50. Fuentes-Afflick E, Hessol NA. Interpregnancy in- 69. Mavalankar DV, Gray RH. Re: “Interpregnancy tal loss. Soc Biol. 1992;39:231-245. terval and the risk of premature infants. Obstet interval and risk of preterm labor. Am J Epidemiol. 30. Barros FC, Huttly SR, Victora CG, Kirkwood BR, Gynecol. 2000;95:383-390. 1991;133:958-959. Vaughan JP. Comparison of the causes and conse- 51. Sachar RK, Soni RK. Perinatal mortality in rural Pun- 70. Dechering WH, Perera RS. A secondary analysis quences of prematurity and intrauterine growth re- jab—a population-based study. J Trop Pediatr. 2000; of determinants of low birth weight. Ceylon Med J. tardation: a longitudinal study in southern Brazil. 46:43-45. 1991;36:52-62. Pediatrics. 1992;90:238-244. 52. Zhu BP, Haines KM, Le T, McGrath-Miller K, Boul- 71. Kumar R, Singhi S. Risk factors for stillbirths in a 31. Leong WP, Viegas OA, Ratnam SS. Premature ton ML. Effect of the interval between pregnancies rural community. Indian J Pediatr. 1992;59:455-461. childbirth: social and behavioural risks in Singapore. on perinatal outcomes among white and black women. 72. Mavalankar DV, Gray RH, Trivedi CR. Risk fac- J Biosoc Sci. 1993;25:465-472. Am J Obstet Gynecol. 2001;185:1403-1410. tors for preterm and term low birthweight in 32. Gribble JN. Birth intervals, gestational age, and 53. Rousham EK, Gracey M. Factors affecting birth- Ahmedabad, India. Int J Epidemiol. 1992;21:263-272. low birth weight: are the relationships confounded? weight of rural Australian Aborigines. Ann Hum Biol. 73. Arif MA, Qureshi AH, Jafarey SN, Alam SE, Arif Popul Stud. 1993;47:133-146. 2002;29:363-372. K. Maternal sociocultural status: a novel assessment 33. Miller JE. Birth order, interpregnancy interval and 54. Dafopoulos KC, Galazios GC, Tsikouras PN, Kout- of risk for the birth of small for gestational age, low birth outcomes among Filipino infants. J Biosoc Sci. laki NG, Liberis VA, Anastasiadis PG. Interpregnancy birth weight infants. J Obstet Gynaecol Res. 1998;24: 1994;26:243-259. interval and the risk of preterm birth in Thrace, Greece. 215-222. 34. Greenwood R, Samms-Vaughan M, Golding J, Eur J Obstet Gynecol Reprod Biol. 2002;103:14-17. 74. Grau Espinosa MA, Saenz Darias L, Cabrales Ashley D. Past obstetric history and risk of perinatal 55. Smith GC, Pell JP, Dobbie R. Interpregnancy in- Escobar JA. Risk factors of low birth weight: Provin- death in Jamaica. Paediatr Perinat Epidemiol. terval and risk of preterm birth and neonatal death: cial Gynecologic-Obstetric Hospital of Sancti Spiri- 1994;8(suppl 1):40-53. retrospective cohort study [published correction ap- tus, Cuba [in Spanish]. Rev Panam Salud Publica. 1999; 35. Fikree FF, Berendes HW. Risk factors for term in- pears in BMJ. 2003;327:851]. BMJ. 2003;327:313. 6:95-98. trauterine growth retardation: a community-based 56. Zhu BP, Le T. Effect of interpregnancy interval on 75. Wang PD, Lin RS. Perinatal mortality in Taiwan. study in Karachi. Bull World Health Organ. 1994;72: infant low birth weight: a retrospective cohort study Public Health. 1999;113:27-33. 581-587. using the Michigan Maternally Linked Birth Database. 76. Mafina-Mienandi MC, Ganga-Zandzou PS, Ma- 36. Rawlings JS, Rawlings VB, Read JA. Prevalence Matern Child Health J. 2003;7:169-178. koumbou P, Malonga H, Ekoundzola JR, Mayanda HF. of low birth weight and preterm delivery in relation 57. Stephansson O, Dickman PW, Cnattingius S. The Risk factors of intrauterine growth retardation in Congo to the interval between pregnancies among white and influence of interpregnancy interval on the subse- [in French]. J Gynecol Obstet Biol Reprod (Paris). 2002; black women. N Engl J Med. 1995;332:69-74. quent risk of stillbirth and early neonatal death. Ob- 31:500-505. 37. Fourn L, Goulet L, Seguin L. Birth intervals and stet Gynecol. 2003;102:101-108. 77. Al-Jasmi F, Al-Mansoor F, Alsheiba A, Carter AO, birth of low weight infants in Benin [in French]. Med 58. van Eijk AM, Ayisi JG, ter Kuile FO, et al. Effec- Carter TP, Hossain MM. Effect of interpregnancy in- Trop (Mars). 1996;56:163-166. tiveness of intermittent preventive treatment with sul- terval on risk of spontaneous preterm birth in Emirati 38. Ochoa Sangrador C, Luque Benlloch C, Carras- phadoxine-pyrimethamine for control of malaria in women, United Arab Emirates. Bull World Health cal Tejado A. Prematurity, low birth weight and the pregnancy in western Kenya: a hospital-based study. Organ. 2002;80:871-875. interval between pregnancies [in Spanish]. An Esp Trop Med Int Health. 2004;9:351-360. 78. Khan N, Jamal M. Maternal risk factors associ- Pediatr. 1996;45:67-70. 59. Arafa MA, Alkhouly A, Youssef ME. Influence of ated with low birth weight. J Coll Physicians Surg Pak. 39. Kallan JE. Reexamination of interpregnancy in- inter-pregnancy interval on preterm delivery. Paedi- 2003;13:25-28. tervals and subsequent birth outcomes: evidence from atr Perinat Epidemiol. 2004;18:248-252. 79. Orji EO, Shittu AS, Makinde ON, Sule SS. Effect U.S. linked birth/infant death records. Soc Biol. 1997; 60. Pedroso de Barros E, Avelar WM, Cecatti JG, Be- of prolonged birth spacing on maternal and perinatal 44:205-212. steti Pires HM. Associaça˜ o entre intervalo interpartal outcome. East Afr Med J. 2004;81:388-391. 40. Adams MM, Delaney KM, Stupp PW, McCarthy e situaço˜ es maternas e perinatais. In: Simposio Inter- 80. Kleijer ME, Dekker GA, Heard AR. Risk factors BJ, Rawlings S. The relationship of interpregnancy in- nacional “Cesa´rea; avaliaçaõ e propostas de açaõ.” for intrauterine growth restriction in a socio- terval to infant birthweight, and length of gestation Campinas, Brazil: CEMICAMP; 2004. economically disadvantaged region. J Matern Fetal among low-risk women, Georgia. Paediatr Perinat 61. Hsieh TT, Chen SF, Shau WY, Hsieh CC, Hsu JJ, Neonatal Med. 2005;18:23-30. Epidemiol. 1997;11(suppl 1):48-62. Hung TH. The impact of interpregnancy interval and 81. Winkvist A, Rasmussen KM, Habicht JP. A new 41. Bakewell JM, Stockbauer JW, Schramm WF. Fac- previous preterm birth on the subsequent risk of pre- definition of maternal depletion syndrome. Am J Pub- tors associated with repetition of low birthweight: Mis- term birth. J Soc Gynecol Investig. 2005;12:202-207. lic Health. 1992;82:691-694. souri longitudinal study. Paediatr Perinat Epidemiol. 62. DaVanzo J, Razzaque A, Rahman M, et al. The 82. Smits LJM, Essed GGM. Short interpregnancy in- 1997;11(suppl 1):119-129. effects of birth spacing on infant and child mortality, tervals and unfavourable pregnancy outcome: role of 42. Sener T, Gu¨rel SA, Gu¨rel H, O¨ zalp S, Hassa H, pregnancy outcomes, and maternal morbidity and mor- folate depletion. Lancet. 2001;358:2074-2077.

Downloaded from jama.ama-assn.org at University of South Carolina on June 13, 2011