Association between prenatal care and small for gestational age birth: an ecological study in Quebec, Canada - HPCDP: Volume 36-7, July 2016
Health Promotion and Chronic Disease Prevention in Canada
Navigate This Article
Volume 36 · Number 7 · July 2016
Association between prenatal care and small for gestational age birth: an ecological study in Quebec, Canada
N. Savard, PhDReference 1; P. Levallois, MDReference 2Reference 3; LP. Rivest, PhDReference 4; S. Gingras, MScReference 5
This article has been peer reviewed.
Correspondence: Nathalie Savard, Ministère de la Santé et des Services sociaux du Québec, 1075 Chemin Ste-Foy, Québec, QC G1S 2M1; Tel: 418-266-6626; Email: firstname.lastname@example.org
Background. In Quebec, women living on low income receive a number of additional prenatal care visits, determined by their area of residence, of both multi-component and food supplementation programs. We investigated whether increasing the number of visits reduces the odds of the main outcome of small for gestational age (SGA) birth (weight < 10th percentile on the Canadian scale).
Methods. In this ecological study, births were identified from Quebec's registry of demographic events between 2006 and 2008 (n = 156 404; 134 areas). Individual characteristics were extracted from the registry, and portraits of the general population were deduced from data on multi-component and food supplement interventions, the Canadian census and the Canadian Community Health Survey. Mothers without a high school diploma were eligible for the programs. Multilevel logistic regression models were fitted using generalized estimating equations to account for the correlation between individuals on the same territory. Potential confounders included sedentary behaviour and cigarette smoking. The odds ratios (ORs) were adjusted for mother's age, marital status, parity, program coverage and mean income in the area.
Results. Mothers eligible for the programs remain at a higher odds of SGA than non-eligible mothers (OR = 1.40; 95% confidence interval [CI]: 1.30-1.51). Further, areas that provide more visits to eligible mothers (4-6 food supplementation visits) seem more successful at reducing the frequency of SGA birth than those that provide 1-2 or 3 visits (OR = 0.86; 95% CI: 0.75-0.99).
Conclusions. Further studies that validate whether an increase in the number of prenatal care interventions reduces the odds of SGA birth in different populations and evaluate other potential benefits for the children should be done.
Keywords: birth weight, gestational age, reproductive health, intervention, health behaviour
- This is the first observational study of Quebec's population - and one of the few worldwide - that explores the benefits of prenatal intervention along the gradient of intensity of available care.
- In Quebec, all pregnant women receive prenatal care. Women living on low income receive additional care through multi-component and food supplementation programs.
- Mothers eligible for the supplementation programs remain at a higher risk of small for gestational age (SGA) birth than non-eligible mothers.
- Prenatal care interventions provided to women living on low income are associated with lower odds of SGA birth.
- In addition, the authors observed a strengthening of the association with increasing number of interventions.
Small for gestational age (SGA) birth is an indicator of fetal developmentFootnote 1 that takes into account fetal weight and length of gestation.Footnote 2 SGA is associated with neonatal death and chronic illness.Footnote 2Footnote 3 Determinants of this outcome include advanced maternal age, chronic disease of the mother, race/ethnicity, primiparity, nutritional status and lifestyle characteristics such as smoking, drug use and physical activity/workload.Footnote 2
To date, participation in multi-component prenatal care programs has not been conclusively found to decrease risks of SGA birth, even among high-risk women.Footnote 4 to footnote -8 However, participating in multi-component programs throughout pregnancy can improve parental behaviours and use of community-based resources.Footnote 5Footnote 9 Further, reducing use of tobacco-usually included in multi-component programs-is associated with decreased risks of low birth-weight and preterm birth.Footnote 10
Randomized controlled trials (RCTs) indicate that balanced energy and protein supplementation seem to reduce the occurrence of SGA birth.Footnote 11 However, what remains unclear is whether those results can be extrapolated to the whole population in a "real world" setting.Footnote 12 to footnote -14 Indeed, compliance may be affected by researchers' investment in the intervention, and women participating in such studies may be more prone to changing their behaviour.Footnote 12 to footnote -14 An observational design could be used to gain further insights on prenatal care intervention. Results from such a study should be adjusted for confounders such as age, poverty and availability of care because it does not include a randomization process that itself controls for confounding.Footnote 4Footnote 15
In Quebec, standard prenatal care is provided by physicians, and additional support is provided by nurses, social workers and nutritionists as part of the integrated perinatal and toddler services multi-component care program (SIPPE)Footnote 16 and as part of the egg, milk and orange food supplementation programFootnote 17Footnote 18 (further information available on request from the authors). In-home multi-component visits include self-empowerment, emotional support and education to improve dietary habits and reduce sedentariness, tobacco use, alcohol consumption and recreational drug use.Footnote 16 Footnote 33
Both programs are standard care for women at a higher risk of pregnancy complications because they are young or living on low income.Footnote 16 ,Footnote 17 The participants are mostly referred to additional care through their prenatal clinic, their medical clinic or their local community health centre, although some contact their local community health centre themselves to indicate their need for intervention.Footnote 17 The Centres locaux de services communautaires (local community service centres; CLSC) manage these programs.
RCTs suggest that multi-component interventions are effective at improving maternal diet, emotional support and mental health.Footnote 16 The intensity of the interventions (the number of visits provided) and program coverage (the proportion of high-risk mothers receiving effective intervention) has been identified as an important consideration when evaluating intervention.Footnote 19 However, what remains unknown is the intensity of the multi-component or food supplementation interventions required to lower the frequency of SGA birth.Footnote 20Footnote 21
We speculated that increasing the number of visits reduces the odds of an SGA birth. To test this hypothesis, we investigated whether the average number of additional care visits in areas of residence in Quebec, Canada, had a dose-response association with SGA births.
Study population and setting
This is an observational study (ecological analytic multiple-group design) of the live singleton births registered in Quebec's registry of demographic events (Registre des événements démographiques du Québec) from April 2006 through March 2008 (n = 156 404 births).Footnote 22 Participants' areas of residence were determined from data on prenatal support interventions (n = 134 CLSCs), the Canadian census of 2001 and 2006 and the Canadian Community Health Survey (CCHS) of 2001, 2003, 2005, 2007 and 2008.Footnote 23 These years were chosen to build a portrait of the environment before and during pregnancy.
Quebec's registry of demographic events collects information on every live birth and on the mother (birth weight, gestational age, mother's age, academic degree and postal code). Between 2006 and 2008, 8.1% (99% confidence interval [CI]: 8.1-8.3) of births in Quebec were SGA whereas between 2009 and 2011 this proportion had risen to 8.6% (99% CI: 8.4-8.7).Footnote 24
Data on prenatal support interventions available at the CLSC level (number of interventions provided by CLSC per year) can be found in the "Info-CLSC" system.Footnote 25 CLSCs vary in size (mean population: n = 46 727; mean births per year: n = 389).
The 2001 and 2006 censuses include data on mean income as well as information on the urbanity and rurality of neighbourhoods within areas. The CLSC portraits were built by pooling area data by the number of residents. Census area geographies were linked to CLSC geographies by merging the federal file "Fichier de conversion des codes postaux plus" (which included the dissemination area and postal code) to Quebec's health geography "Référentiel territorial M34" (postal code and CLSC territory) by postal code.
The CCHS includes information about food insecurity, tobacco use, sedentary behaviour and low social support. Data from Quebec's residents from the different cycles (2001, 2003, 2005, 2007 and 2008) were pooled and CLSC proportion-like values were built using the 100 832 CCHS responses.
The CLSC portraits of the general population were associated with the mothers' variables using their postal codes.Footnote 24 If a postal code did not exactly match a CLSC boundary (which occurred for fewer than 3% of the codes), it was associated with the CLSC territory that included the majority of the residents. Table 1 shows individually measured information and CLSC-level variables.
|CLSC level||Data on prenatal support,MSSS||
Intensity of interventions:
|Individual-level||Registry of demographic events, MSSSReference b||Mother's country of birth|
|CLSC level||Canadian Community Health Survey, Statistics CanadaReference c||Percentage of residents with food insecurity in the past 12 monthsReference d|
|% of residents with sedentary behaviour in the past 3 monthsReference d|
|% of residents with low tangible social support on the Medical Outcome Study subscaleReference d|
|% of residents who smoke cigarettes dailyReference d|
|Canadian census, Statistics CanadaReference e||Presence of urban neighbourhoods within the CLSC (exclusively urban; exclusively rural; urban and rural neighbourhoods)|
|Individual-level||Registry of demographic events, MSSS||Mother's age, years (< 20; 20-24; 25-29; 30-34; ≥ 35)|
|CLSC level||Data on prenatal support, MSSSReference e||Programs coverageReference f (% of target population receiving food intervention)|
Abbreviations: CLSC, Centres locaux de services communautaires (local community health centre territory); MSSS, Ministère de la Santé et des Services sociaux du Québec.
Excluded were births at less than 22 weeks gestational age or more than 43 weeks gestational age; with implausible weight for age (according to a criteria recommended by Alexander et al.Footnote 26); with missing potential confounder information at the CLSC level (births from CLSCs of the Nord-du-Québec, Terres-Cries-de-la-Baie-James and Nunavik regions); and with missing data on prenatal care.
Outcomes. The main outcome was SGA birth (weight below the 10th percentile of the Canadian reference scale).Footnote 1 The "term births only" population was used for sensitivity analyses (births were categorized as "term births" using the registry of demographic events).
Exposure. Both multi-component and food intervention programs targeted women in need. Mothers targeted by the food supplementation program had a family income below the Canadian cut-off.Footnote † Those targeted by multi-component interventions were less than 20 years old at the estimated time of birth, had no high school diploma or had a low family income.Footnote 16Footnote 18
Mothers without a high school diploma were considered eligible for both the multi-component and the food programs, since this population has the lowest income (on average, in 2009, women with a high school diploma or less earned $20,400 per year, those with college degree earned $30,300 and those with a university degree earned $48,400Footnote 27). We used a narrower definition of eligibility (mothers aged less than 20 years and without a high school diploma) for sensitivity analyses, as younger women are more likely to have a low household income.Footnote 28
The intensity of the multi-component or food supplementation intervention received by eligible women was an area-level variable. It was neither directly measured for every woman nor based on whether they receive the intervention; it was based on whether mothers could receive the intervention. It corresponded to the average number of visits from the food supplementation program per eligible woman living in the CLSC territory (those women also had between 2 and 8 multi-component visits). The number of visits from the food supplementation program were categorized into tertiles (lowest: 1 to 2 visits; medium: 3 visits and highest: 4 to 6 visits). Non-eligible mothers were given an intensity of exposure equal to zero and were included in the "non-eligible for the program/reference" category (Table 1).
Potential confounders. Potential confounders are program eligibility and individual maternal characteristics (age, country of birth, marital status, academic qualification and parity) at the individual-level and program coverage as well as variables portraying residents of the CLSC territories at the CLSC level.
CLSC defined program eligibility and program coverage variables as follows: mothers without a high school diploma were identified as eligible for the food supplementation program and the multi-component program. This variable was incorporated in the definition of the "intensity of intervention." Program coverage was calculated as the proportion of eligible mothers receiving intervention on the CLSC territory. A categorical scale was used for univariate analyses ("non-eligible," "lowest," "medium" and "highest").
Other CLSC "portraits" included the proportion of urban neighbourhoods within each CLSC territory ("exclusively urban neighbourhoods/reference," "exclusively rural neighbourhoods" and "urban and rural neighbourhoods"); mean income ("lowest," "medium" and "highest/reference"); proportion of residents with the following risk factors: food insecurity, sedentary behaviour, low tangible social support and daily tobacco use (using "lowest/reference," "medium" and "highest" tertiles).
Source of data
Mothers and their babies were registered in the registry of demographic events by the Ministère de la Santé et des Services sociaux du Québec (April 2006 through March 2008). Their CLSCs were portrayed by the same dataset on prenatal support interventions, by data from the Canadian census from Statistics Canada (2001 and 2006) and by the CCHS from Statistics Canada (the 2001, 2003, 2005, 2007 and 2008 files were combined to achieve a reasonable number of respondents per CLSC).Footnote 22 CLSC portraits were associated with maternal variables using postal codes, as described elsewhere.Footnote 24 Table 1 shows additional information related to the variables used.
All data have a multilevel structure: the first level is the mother and the second is the local community health centre.Footnote 29 Multilevel logistic regression models were fitted using generalized estimating equations (GEE) to account for the correlation between individuals in the same CLSC territory. (The GEE method provides consistent odds ratio [OR] estimates for the population even though the correlation between mothers from the same CLSC is unknown.) We used independent working correlation structures throughout the univariate and multivariate analyses and obtained empirical robust standard error estimates.Footnote 30
Univariate logistic regression models were fitted on every potential confounder and on exposure (dependent variable: SGA birth). Multivariate models were fitted on exposure. A full model was first adjusted with all the potential confounders listed in Table 2. Confounders that changed the effect estimate of the exposure by less than 5% were removed one at a time (change-in-estimate approachFootnote 31Footnote 32), which resulted in the final adjusted model.
|Individual-level variable from Registry of demographic events||Number and proportion of live births, n (%)||Proportion of SGA births, %||Crude odds ratio (95% CI)||p value|
|Mother's age, years|
|< 20||4049 (2.6)||11.6||1.46 (1.32-1.61)||<.01|
|20-24||23 767 (15.2)||10.0||1.23 (1.17-1.29)|
|25-29 (reference category)||56 170 (35.9)||8.3||1.00|
|30-34||48 981 (31.3)||7.4||0.88 (0.85-0.92)|
|≥ 35||23 437 (15.0)||8.2||0.99 (0.93-1.05)|
|< High school||10 742 (6.9)||11.1||1.56 (1.44-1.69)||<.01|
|High school diploma||46 660 (29.8)||9.5||1.31 (1.26-1.37)|
|College||44 048 (28.2)||7.5||1.02 (0.98-1.06)|
|≥ University (reference category)||54 954 (35.1)||7.4||1.00|
|Mother's country of birth|
|Other||31 350 (20.0)||9.4||1.18 (1.09-1.29)|
|Canada (reference category)||125 054 (80.0)||8.1||1.00|
|Primiparous||72 792 (53.5)||10.8||1.82 (1.75-1.89)|
|Multiparous (reference category)||83 612 (46.5)||6.2||1.00|
|Married (reference category)||59 038 (37.8)||7.7||1.00|
|Unmarried||97 366 (62.3)||8.7||1.14 (1.06-1.23)|
|Abbreviations: CI, confidence interval; CLSC, Centres locaux de services communautaires (local community health centre territory); SGA, small for gestational age.|
Sensitivity analyses were as follows: regressions of SGA on intensity using term births only (from 37 to 40 weeks of gestational age inclusively) and regressions of SGA on intensity using the narrower definition of eligibility for the programs (mothers aged less than 20 years without high school diploma). Both analyses were adjusted for confounders incorporated in the final model.
Analyses were carried out using SAS version 9.2 (GENMOD and REG procedures). Results were considered statistically significant at p <.05.
The Commission d'accès à l'information du Québec and Université Laval's Ethics Committee approved this research project.
A total of 156 404 singleton births (134 CLSCs) were included. Most of the mothers were 25 to 29 years old, had a university degree, were born in Canada, were primiparous and unmarried (neither married nor in a common-law relationship) (Table 2). A total of 11.1% of the 10 742 eligible births and 8.1% of the non-eligible births were SGA. (Further information on program coverage is available from the authors on request.)
Univariate regression analyses
Mothers aged less than 20 years (OR = 1.46; 95% CI: 1.32-1.61) and 20 to 24 years (OR = 1.23; 95% CI: 1.17-1.29) have higher unadjusted odds of SGA while 30- to 34-year-old mothers have a lower odds (OR = 0.88; 95% CI: 0.85-0.92) compared to 25- to 29-year-old mothers (the reference category). Mothers with less than a high school diploma (OR = 1.31; 95% CI: 1.26-1.37) and those with a high school diploma (OR = 1.56; 95% CI: 1.44-1.69) had higher odds of SGA than mothers with a university degree. Mothers born outside Canada (OR = 1.18; 95% CI: 1.09-1.29), primiparous mothers (OR = 1.82; 95% CIs: 1.75-1.89) and unmarried mothers (OR = 1.14; 95% CIs: 1.06-1.23) also had higher odds of SGA compared to mothers from the reference categories.
Mothers from CLSCs with a medium (OR = 1.10; 95% CI: 1.00-1.21) or high proportion (OR = 1.19; 95% CI: 1.10-1.28) of residents experiencing food insecurity, a high proportion of sedentary residents (OR = 1.16; 95% CI: 1.06-1.27) and a high proportion of residents who smoke cigarettes (OR = 1.10; 95% CI: 1.01-1.21) have higher odds of SGA than mothers from CLSCs with low proportions of these variables. Mothers from CLSCs with the lowest mean income have higher odds of SGA than mothers from CLSCs with the highest mean income (OR = 1.18; 95% CI: 1.09-1.28). Finally, mothers from CLSCs with both urban and rural neighbourhoods have lower odds of SGA than mothers from rural CLSCs (OR = 0.92; 95% CI: 0.84-1.00) (Table 3).
|CLSC level by tertileReference a||Population of live births, n (%)||Proportion of SGA births, %||Crude odds ratio (95 % CI)||p value|
|Proportion of the target population receiving food interventionReference b|
|Non-eligible mother (reference category)||145 662 (93.1)||8.1||1.00||<.01|
|Lowest||0.0-100.0||4607 (3.0)||11.3||1.44 (1.28-1.61)|
|Medium||100.0-200.0||3680 (2.4)||10.9||1.38 (1.25-1.52)|
|Highest||200.0-700.0||2455 (1.6)||11.3||1.44 (1.28-1.61)|
|Proportion of residents in the CLSC with food insecurity, %Reference c|
|Lowest (reference category)||4.0-10.5||52 260 (33.4)||7.7||1.00||<.01|
|Medium||10.6-15.1||57 488 (36.7)||8.4||1.10 (1.00Reference d-1.21)|
|Highest||15.2-36.4||46 656 (29.8)||9.0||1.19 (1.10-1.28)|
|Proportion of sedentary residents, %Reference c|
|Lowest (reference category)||1.7-9.9||50 076 (32.0)||7.8||1.00|
|Medium||9.9-14.4||48 360 (30.9)||8.2||1.06 (0.98-1.15)|
|Highest||14.4-31.0||57 968 (37.1)||8.9||1.16 (1.06-1.27)|
|Proportion of residents with low tangible social support, %Reference c|
|Lowest (reference category)||12.1-37.8||38 421 (24.6)||8.0||1.00||.06|
|Medium||37.8-47.2||65 141 (41.7)||8.8||1.11 (1.02-1.20)|
|Highest||47.2-69.9||52 842 (33.8)||8.1||1.01 (0.93-1.10)|
|Proportion of residents who smoke cigarettes daily, %Reference c|
|Lowest (reference category)||8.1-20.8||64 511 (41.3)||8.1||1.00||.04|
|Medium||20.8-25.9||51 439 (32.9)||8.2||1.02 (0.93-1.11)|
|Highest||25.9-39.4||40 454 (25.9)||8.9||1.10 (1.01-1.21)|
|Mean incomeReference c|
|Lowest||$16,144-$25,268||25 964 (16.6)||8.0||1.18 (1.09-1.28)||<.01|
|Medium||$25,268-$28,797||53 052 (33.9)||8.3||1.04 (0.96-1.12)|
|Highest (reference category)||$28,797-$47,610||77 388 (49.5)||9.3||1.00|
|Presence of urban and rural neighbourhoods in the CLSCReference c|
|Rural||32 246 (20.6)||8.3||1.00||.05|
|Urban and rural||35 346 (22.6)||7.7||0.92 (0.84-1.00Reference e)|
|Urban||88 812 (56.8)||8.6||1.03 (0.96-1.11)|
Abbreviations: CI, confidence interval; CLSC, Centres locaux de services communautaires (local community health centre territory); SGA, small for gestational age.
Crude OR estimates from regression on intensity indicate that eligible mothers from any of the "lowest: 1 to 2 visits from the food supplementation program," "medium: 3 visits" and "highest: 4 to 6 visits" categories have higher odds of SGA than non-eligible mothers (OR = 1.40; 95% CI: 1.30-1.51; not shown in the table).
Multivariate regression analyses
Results of crude and adjusted odds ratios of SGA are shown in Table 4. The final adjusted model on intensity accounts for mothers' age, parity and marital status as well as program coverage and mean income in the CLSC. Women eligible for both multi-component and food intervention from any of the intensity groups had higher adjusted odds of SGA than non-eligible women (OR = 1.40; 95% CI: 1.30-1.51; data not shown). Moreover, the association with increasing intensity of interventions was attenuated: eligible women living in a territory that provided high-intensity interventions (4-6 visits of food intervention per eligible woman) had lower odds of SGA than women living in a territory that provides interventions of low or medium intensity (1-2 or 3 visits per eligible woman) (OR = 0.86; 95% CI: 0.75-0.99; data not shown). Estimates from the full models are similar to those from the adjusted models (not shown in the table).
|Intervention (CLSC level)Reference a||Births, n (%)||Proportion of SGA births, %||Association of intensity of intervention|
|Crude OR (95% CI)||Adjusted ORReference b (95% CI)|
|Non-eligible motherReference c||145 662 (93.1)||8.1||1.00||1.00|
|Lowest tertile (1-2 visits)||3611 (2.3)||11.7||1.50 (1.33-1.69)||1.49 (1.32-1.69)|
|Medium tertile (3 visits)||4233 (2.7)||11.3||1.44 (1.30-1.60)||1.46 (1.32-1.61)|
|Highest tertile (4-6 visits)||2898 (1.9)||10.2||1.29 (1.13-1.46)||1.27 (1.12-1.44)|
Abbreviations: CI, confidence interval; CLSC, Centres locaux de services communautaires (local community health centre territory); SGA, small for gestational age.
Sensitivity analyses corroborate the main results. When the final models were fitted on term births only, mothers eligible for the programs had a greater odds of SGA than non-eligible mothers (OR: 1.43; 95% CI: 1.32-1.55; data not shown), while high exposure is associated with lower odds than low or medium exposure (OR = 0.90; 95% CI: 0.78-1.05; data not shown). Final results on data with the narrower definition of eligibility for intervention were also similar. The eligible mothers have a greater odds of SGA than the non-eligible (n = 147 156) mothers (OR = 1.48; 95% CI: 1.36-1.60; data not shown). There were fewer eligible mothers (n = 9248) when this definition is used than when the definition based on academic qualification alone (10 742 eligible mothers) was used. Final results are similar but non-significant (OR = 0.89; 95% CI: 0.76-1.04; data not shown).
This is the first observational study of Quebec's population-and one of the few worldwide-that explores the benefits of prenatal intervention along the gradient of intensity of available care. We found that mothers living in Quebec who are eligible for supplemental prenatal care programs are at a higher odds of SGA than those who are non-eligible, and that prenatal care interventions provided to women living on low income are associated with lower odds of SGA birth. In addition, we observed a strengthening of the association with increasing intensity of the interventions. However, interventions do not counteract all of the effects associated with need; eligible mothers remain at higher odds of SGA than non-eligible mothers. Nevertheless, the interventions have some effect: areas that provide high-intensity intervention (4 to 6 visits from the food supplementation program) reduce the frequency of SGA birth more successfully than those that provide low- or medium-intensity intervention.
Though results from RCTs on the subject were encouraging, experimental studies on dietary changes have numerous limitsFootnote 12 to footnote -14 and an observational study provides needed confirmation. Our odds ratio of SGA for high- versus medium- or low-intensity intervention is similar to the pooled relative risk (RR) from RCTs on balanced energy/protein supplementation (6 studies; n = 3396; RR = 0.68; 95% CI: 0.56--0.84).Footnote 11
Our findings on high- versus medium- or low-intensity interventions (OR = 0.86; 95% CI: 0.75-0.99) are comparable to those in more controlled studies compiled in a review of experimental and observational studies:Footnote 4 associations with low birth-weight were within the acceptable range (0.80-0.90). International meta-analyses of RCTs on the impacts of multi-component intervention among high-risk women indicate similar and non-significant improvement in low birth-weight (11 studies; n = 8681; RR = 0.92, 95% CI: 0.83-1.03).Footnote 7
Strengths and limitations
This study has a number of important strengths. It included the entire population of mothers and single births in Quebec; in other words, all women eligible for the supplementation programs. To the best of our knowledge, this is the first population-based study that statistically tests for differences in benefits according to program eligibility and on the gradient of exposure to intervention. Further, this is the first investigation of prenatal programs that assesses the relevance of accounting for the contextual factors of income, food insecurity, social support, smoking and sedentary behaviour. In addition, the use of external survey data to incorporate such contextual variables has never been done in the field of intervention. Understanding the benefits of exposure to prenatal care programs on SGA births provides unique insights for tailoring further interventions.
Some limitations should be considered when interpreting the results, including three possible ways to misclassify exposure to intervention. First, eligibility status was determined by a proxy, academic qualification because information on income was unavailable. Moreover, all the eligible women did not necessarily use the interventions. However, this bias is likely to have only a small impact on the results since the sensitivity analysis with the narrower definition of eligibility status led to similar associations with SGA.
Second, we assumed that the need for prenatal care and the intensity of use within CLSC territories remained constant over the years. Most women were exposed to the intensity of care we attributed to them, as intensity was averaged for the duration of the study.
Finally, we did not have information about the exposure to intervention of mothers who relocated. These misclassifications contributed to small biases towards the null association. Confounding potentially brought some bias to the association, as individual information such as a mother's chronic disease was not accounted for. If there were more women with a chronic disease eligible for intervention in CLSC territories with high-intensity intervention than in territories with low-intensity intervention, there would be another bias towards the null effect.
Defining what constitutes "high" exposure and whether this level of exposure is sufficient is difficult, and definitions are likely to vary between jurisdictions. Nonetheless, the categories we use in this study are based on a scale that can be used in the absence of knowledge on the subject.Footnote 28
In terms of limitations, associations based on aggregate data (information on intensity of intervention based on data from the CLSCs) are weakened by the potential for ecological fallacy.Footnote 22 In addition, it was not possible to compare our results to the occurrence of SGA before the beginning of the intervention in this population.
An alternative explanation of the results could be that the CLSCs that provide high-intensity interventions have similar resources for fewer targeted women. Quality and timing of interventions could thus be maximized and their impact on SGA could be greater.
Significant differences along the gradient of care suggest that strategies that provide pregnant women with high exposure to interventions are effective at reducing the risk of SGA. The results have important implications for continuing existing programs and developing new ones adapted to the different needs of mothers. Numerous other benefits might be identified, as interventions do not result only in increasing weight for gestational age at birth.
Although results are encouraging, further research is needed on other subpopulations that could benefit from interventions. Future studies might benefit from incorporating some measure of the quality of interventions, use of standard prenatal care and use of additional prenatal support.
Conflict of interest
The authors declare that they have no competing interests.
The manuscript preparation and publication is funded by the Institut national de santé publique du Québec.
The authors thank Gylaine Boucher and Lamia Belfarès for their helpful insights during discussions around this article.
Report a problem or mistake on this page
- Date modified: