Critical analysis of research studies is one of the most important steps towards incorporation of evidence into practice (Burns & Grove, 2007). This paper is an attempt towards achieving this goal. The paper critically analyzes the article “Group prenatal care and preterm birth weight: Results from a matched cohort study at public clinics” by Ickovics et al.(2003). This aim will be achieved by sequentially critiquing the research problem, the literature review, principles of research ethics employed in the study, the underlying theoretical framework, the research hypothesis, the sampling technique, the research design, data collection methods, the analysis, and the recommendation proposed by the authors. Finally, the paper will end with an overall appraisal of the strength and limitations of the study.
This particular research paper was analyzed as the concept of group prenatal care has not been investigated in detail yet. While reviewing the literature on the stated topic the limited supporting evidences for group prenatal care was found. Only 01 RCT (Ickovics et al.2007), 01 matched cohort study (Ickovics et al.2003), and 03 pilot studies with descriptive analysis were found (Baldwin, 2006; Grady and Bloom, 1998; Rising, 1998). The comparative analysis of all 05 studies is shown in appendix 1. Chosen study is the only one which examined the impact of group versus individual prenatal care on birth weight and gestational age.
The problem studied by Ickovics et al. (2003) is that, whether group prenatal care has a significant impact on the perinatal outcome like birth weight, and gestational age. This problem is extremely significant to nursing, for a number of reasons. The first reason as described by Ickovics el al. (2003), is the significant relationship of these perinatal outcomes with “neonatal morbidity and mortality” (p. 1052). Neonatal mortality is one of the areas of concern for nursing due to high mortality rates around the world and especially in developing countries (Straughn et al.,2003).
The purpose for conducting the study has been explicitly stated by Ickovics et al. (2003). The stated purpose can be accomplished very well because both the outcomes that are birth weight and gestational age, are measureable and thus the research problem is testable (Haber & Cameon, 2005). A quantitative approach is suitable for this study as the variables are quantifiable and can be described objectively in numbers. Ickovics et al. (2003) have introduced the problem statement after giving a brief background of the importance of the adequate prenatal care, and the factors that determine the adequate prenatal care. They then introduce the subject of group prenatal care as a “structural innovation” in the domain of prenatal care (Ickovicd et al., 2003, p. 1052), and then share the problem statement that this innovative method of provision of prenatal care has not been tested yet.
The authors have stated the underlying assumptions of the study. Ickovics et al.(2003) believe that more time the pregnant women spent together, the better will be their understanding of the health behaviors, and they will receive more social support from each other. This will reduce their risk behaviors for instance smoking leading to low birth weight. The limitations of the study have also been discussed explicitly in the article Ickovics et al. (2003) admit that their study is limited because of non-random selection of the women for group prenatal care. The limitation is realistic because the practice was already in place and researchers wanted to observe the outcomes in relation to the two different approaches to care. Moreover, this is an inherent feature of the cohort design that the groups are selected on the basis of their exposure or non-exposure to a particular phenomenon which is not controlled by the researcher, rather the choice is made by the subject themselves. This feature of the cohort design makes it prone to “selection bias.”(Rochon et al., 2005).
Review of Literature
The review of the literature presented by Ickovics et al., (2003) is comprehensive. It starts by emphasizing the importance of adequate prenatal care in terms of perinatal outcomes. It then focuses on the factors that make the prenatal care adequate. The authors then describe the group prenatal care as a structure innovation designed to make the prenatal care adequate, and they appreciate that this new approach has not been tested yet for its efficacy in improving prenatal outcomes. The review then proceeds with a brief description of Centering Pregnancy Program and its components, which finally leads to the statement of purpose of the study. The review ends with a rationale for the selection of study population that is black and Latinas women who are vulnerable for adverse perinatal outcomes. There is an evident relationship of the review with the purpose of study in that the authors have presented only those studies that are pertinent to perinatal care and that have established the efficacy of group care in other population. The review includes a mix of recent and old studies. As this study was accepted for publication in 2003, most of the studies cited in the review were not current, that is those studies were not published in the last five years of the date when this study was accepted for publication (Burns & Groove, 2007); and there are only 10 out of 26 studies, that were current and were published in the last five years of the date of acceptance of the study under consideration, for publication. The review clearly indicates that the other interventions that have been planned to augment prenatal care were not found to be effective in improving the women’s perinatal outcome.
The study has been designed in a manner that there are minimum risks to the subjects. Since this is a non experimental study and only intends to observe the impact of an intervention that is already in progress without manipulating any of the variables, therefore there are minimal risks associated with it (Rochon et al, 2005). However, the researchers have not made any extra effort to maximize the benefits for the subjects. There is no mention of how and when informed consent was attained from the subjects, but this may have not been reported as it was an observational study and most likely was related to quality improvement and do not need consent. The researchers have obtained the approval of research from institutional review boards at these clinics. Also, at one point, the authors have described their efforts to maintain anonymity of the subjects. The authors explained that while selecting matched cohort for the study they entered all the information needed to select the cohort, except for the patient identification information.
The study is based on an underlying framework that is the centering pregnancy model. The underlying assumptions of the study are derived from the model that includes the belief that learning in groups promote shared support, change in behavior, and problem solving skills and it has significant impact on the birth outcome (Rising, 1998). The rationale for the use of the framework is evident from the fact that the entire concept of group prenatal care is based on this model and underlying assumptions. In fact, the research problem and the purpose are also derived from the same model, because the purpose of the study is to examine the impact of group prenatal care.
The hypothesis to be tested by the study is formally stated in the article. The hypothesis is derived from the research problem and hence predicts that “infants of women in group prenatal care would have significantly higher birth weight and be less likely to be delivered preterm compared with those who received individual prenatal care” (Ickovics et al, 2003, p. 1052). This is a complex hypothesis as it predicts the relationship between one independent variable (provision of group prenatal care), and two independent variables (birth weight and gestational age). The hypothesis can also be categorized as directional hypothesis, as it predicts the expected direction of the relationship between provision of group prenatal care, birth weight, and gestational age. According to Polit & Beck (2008) “a directional hypothesis indicates that the researcher has intellectual commitment to the hypothesized outcome, which might result in bias.” (p.99).
Haber & Cameron (2005) therefore suggest that directional hypothesis should only be formed on the basis of sound literature evidences and theoretical basis. In this case, Ickovics et al (2003) have got sound literature support, and since group prenatal care is one of the ways to augment the content of prenatal care; therefore, the authors have hypothesized that group prenatal care will lead to improved birth weight and gestational age. Also, there is sound theoretical base to this hypothesis, as it is based on the centering pregnancy model. The hypothesis is spelled out clearly and it objectively describes the outcome variables (Polit & Back, 2008), that is perinatal outcomes have been quantified and objectively described as birth weight and gestational age.
Ickovics et al (2003) clearly described the population as “Black and Hispanic pregnant women of low socioeconomic status, entering prenatal care at 24 or less weeks of gestation.” (p.1051).The authors have also described in detail the characteristics of the sample such as race, age, parity, and city of residence. The detailed and comprehensive description of the sample gives an in depth understanding of the sample’s characteristics and determines the generalization of the findings to a specific population based on these characteristics (Haber & Singh, 2005). In this case, looking at the characteristics of the sample, the findings can be generalized to black and Hispanic pregnant women of low socioeconomic group, aged 25 or younger, as more than 85% of the sample consisted of black women who were 25 years old or younger. Sample’s characteristics help in determine heterogeneity or homogeneity of the sample (Haber & Singh, 2005). In this case, some of the sample’s characteristics that had the potential to act as cofounders, for instance age, race, parity, history of preterm labor and total number of visits were matched in both the groups. This resulted in homogeneity among the two groups in terms of the above mentioned characteristics. The matching of the two groups on the basis of these characteristics also reduced the potential sampling bias that could have resulted if the groups would’ve been different in terms of these characteristics and the resulting health behaviors. If the groups were not matched, these differences in groups could have accounted for the differences in outcomes, rather than intervention itself.
Ickovics et al (2003) have described the sample selection process in detail. They have also indicated the potential sampling bias due to lack of randomization while enrolling subjects in group prenatal care. Women, who voluntarily enrolled themselves in the group prenatal care programme at the clinics, were recruited as participants in the group that received group prenatal care. Ickovics et al (2003) have also comprehensively described the controls that they have utilized to minimize sampling error or sampling bias, that is, they have randomly selected the comparison group through a computer programme on the basis of first available patient with closest delivery date, by matching some of the characteristics of the treatment group.
The clinics from where the subjects were recruited were also selected by non probability sampling method, selecting only those clinics that served minority women from low socio economic background. The non probability sampling method employed in the study fits well with the level of inquiry and design of the study as Haber & Singh (2005) supported “non experimental studies usually use non probability, purposive sampling method.” (p. 53)
The sample size taken by Ickovics et al (2003), that is N=458 is sufficiently large. Each group had 229 research subjects. The sufficiency of the sample size was assured by conducting a power analysis that “229 pairs had a power of 0.80 to detect a small effectâ€¦reflecting the ability to detect a difference between the two treatment groups of 155 g.” (p. 1053). According to Burns & Groove (2007) the power of 0.8 is the minimum acceptable level of power for any study.
Ickovics et al (2003) have used cohort design, which is also known as longitudinal prospective design. The cohort design, which is a non experimental design, is appropriate for the level of inquiry of the study. In this the researchers intended to examine the difference between the outcomes of the women who received group perinatal care versus who received individual care. LoBiondo-Wood, Haber & Singh (2005) supported that longitudinal design is quite appropriate for testing the difference between the two groups in terms of outcomes variables. However, Ickovics et al (2003) could have used an experimental design for this study, if they wanted to determine the cause and effect relationship between the group prenatal care and perinatal outcomes. This would also have assured randomization of subjects into intervention and control group and would have given a higher level of evidence. However, they may have chosen non experimental study design rather than experimental design to study the impact of group prenatal care in a natural setting, and not in a control study setting. As discussed in the sampling section, potential effects of unwanted variables like race, age, parity, city of residence, history of preterm labor, and total number of prenatal visits have been controlled by Ickovics et al (2003) by matching the cohorts on the basis of these characteristics, in order to ensure significant internal validity of the study (Polit & Beck, 2008). Also, Ickovics et al (2003) have recruited a large sample to ensure a power of 0.8 for the study, which is also one of the ways to maximize the internal validity of the study (Polit & Beck, 2008).
Before data collection the researchers have to operationalize the variables of interest (Sullivan-Bolyani, et al 2005). Ickovics et al (2003) have operationalized gestational age as term or preterm based on the weeks of gestation as measured by the last menstrual period and ultrasound.
Subjects’ demographics information and the number of prenatal visits were obtained from the medical records of the clinic. It has not been mentioned that who determined gestational age through ultrasound, who measured the neonate’s weight, and who retrieved data from medical records. It is really important to know who collected data in order to establish its accuracy, as the expertise and training of the data collector has significant impact on the correctness and precision of data (Sullivan-Bolyai et al, 2005).
Ickovics et al (2003) have used both descriptive and inferential statistics to examine the data. Since the purpose of the study was to examine the differences between the two groups, therefore the inferential statistics was used that is McNemar test, which is appropriate to the level of inquiry due to matched groups. It is also appropriate to the cohort design as this design also intends to measure differences between the two groups, in terms of outcome variables (Polit & Beck, 2008). The other inferential statistical test used is paired t-test which is also appropriate for the study as it is used to test differences between the means of two groups that are matched or paired with each other on the basis of certain characteristics (Polit & Beck, 2008).
Another statistic used in the study is the F statistic. It appears as if the authors have used F statistic when applying multiple linear regression because F statistic has been used while indicating the interaction effect between birth weight and preterm delivery (p. 1054). Linear regression is used to explain how much variability in outcome variable is attributable to the independent variable (Burns & Groove, 2007).
The authors have used descriptive statistic to describe the distribution of demographic variables among the subjects like age, race and parity. They also have used descriptive analysis to describe the distribution of demographic variables among the distribution of outcomes variables (birth weight and gestational age) among the sample.
Ickovics et al (2003) have used parametric as well as non parametric statistics, for instance t-test for matched pairs is a parametric test while McNemar test is a non parametric test. As few variables have been measured on the ratio level of measurement, for instance number of prenatal visits, therefore Ickovics et al (2003) had the liberty to apply parametric statistic. Ickovics et al (2003) had measured the outcome variables of birth weight and gestational age on nominal level. The birth weight was categorized as low birth weight (less than 2500 g), and very low birth weight (less than 1500 g) and the gestational age was categorized as term or preterm (less than 37 weeks of gestation). Therefore, researchers were also able to apply non parametric statistic that is McNemar test (Polit & Beck, 2008).
The consistency in the results of descriptive and inferential statistics confirms the correctness of the findings (Polit & Beck, 2008). There is a logical link between the statistical analysis and the findings of the study. Also, there is consistency in the results presented in numbers and result presented in text, for instance in the above example, the statistical result shows p<0.01 which is statistically significant, and this is well supported by the explanation that is birth weight was greater in infants of women who received group prenatal care versus individual prenatal care. However, Ickovics et al (2003) have not explicitly stated whether they have taken uniform level of significance for all the statistical tests performed or are different for each test, and if so then what is the level of significance for each test.
The graphical and tabular presentations are accurate and appropriate and matches with the findings presented in text (Sullivan-Bolyayi, et al, 2005). The statistics presented in these tables and the graphs are according to their appropriate level of measurement. For instance in table 2, mean and standard deviations have only been calculated for the birth weight which is the only ratio level data in that table, the other variables that are at nominal level of measurement were calculated in percentages (Sullivan-Bolyayi, et al, 2005). The authors have not only established the significance of findings, but they have also described the clinical significance of the results.
Conclusion and Recommendations:
The results derived from data analysis are clearly stated and explained with reference to the research question and hypothesis. The findings are stated succinctly and the authors have related their findings with the research purpose and its underlying assumptions. Ickovics et al (2003) have discussed that the findings can be generalized among women who are vulnerable for preterm births. This seems to be an overgeneralization beyond the study population, because the study sample and the target population was black and Hispanic women who were high risk for preterm birth and were of low socioeconomic status. These findings may not be applicable to all the women who are at risk of preterm births without specifying their ethnicity and socioeconomic status. Ickovics et al (2003) have also discussed the potential benefits of implementing the findings. The authors have also recommended that further research needs to be done for determining the exact mechanisms involved in group prenatal care that results in improved perinatal outcome. Another recommendation is to widely apply group prenatal care in future, however, the authors also appreciate that it is not easy to introduce such big structural changes.
Strengths and limitations of the study:
Overall, the study is a good effort in examining the impact of group prenatal care on perinatal outcomes. A well formulated hypothesis, relevant and comprehensive literature review, an in depth account of sample’s characteristics, well thought research design and statistical analysis are strengths of the study. However, randomization of subjects at the time of recruitment, employment of research ethics such as voluntary participation and realistic generalization of the findings would have added more strength to it.
Pregnancy outcomes: Group vs individual prenatal care
Outcomes: Group Vs Individual Prenatal Care
OR (95% CI)
Ickovics JR et al (2007).
Preterm births in African American women
Less-than-adequate prenatal care*
RESULTS (P VALUE)
Ickovics JR et al (2003).
Matched cohort N=458
Birth weight (g)
3228 vs 3159 (P<.01)
Preterm birth weight (g)
2398 vs 1990 (P<.05)
Grady MA et al (2004).
Cohort study with clinic comparison N=124 (intervention)
Preterm births <37 wk (%)
10.5 vs 25.7 (P<.02)
Low birth weight <2500 g (%)
8.8 vs 22.9 (P<.02)
Breastfeeding at hospital discharge (%)
46 vs 28 (P<.02)
Descriptive analysis N=111
3rd trimester emergency room visits (%)
26 vs 74 (P=.001)
2-group pre-/post-test design N=98
Change in prenatal knowledge scoresâ€
0.98 vs 0.4 (P=.03)
CI, confidence interval; NN T, number needed to treat; OR, odds ratio.
*Kotelchuck Adequacy of Prenatal Care Utilization Index, a validated scoring scale encompassing timing of initiation of care, number of visits, and quality and content of prenatal care. Kotelchuck M. An evaluation of the Kessner Adequacy of Prenatal Care Index and the proposed Adequacy of Prenatal Care Utilization Index. Am J Public Health. 1994;84:1414-1420.
Ref: Williams, K.J and Kaufmann, L (2009) The journal of family practice , 58, (7)
Cite This Work
To export a reference to this article please select a referencing style below: