Nursing Research

 

Critique I

Zoia Hmelnitcaia

NURS 3130

Doctor Keane

October 29, 2014

 

The article Brief Intervention for Prenatal Alcohol Use: A Randomized Trial by Chang, McNamara, Orav, Koby, Lavigne, Ludman, Vincitorio and Wilkins-Haug (2005), is a report of a research that they have done. The objective of this research was to test the efficacy of a brief intervention aimed to reduce alcohol consumption while pregnant, when a partner was included. For this purpose, they conducted a randomized trial in which the treatment group received a diagnostic interview and a brief intervention that included the partner, the control group had the diagnostic interview only. According to the authors, both groups experienced decreased alcohol consumption after enrolling in the study; factors associated with increased drinking after randomization were more years of education, extent of previous alcohol use and temptation for social drinking (Chang et al., 2005). The brief intervention reduced most significantly the alcohol consumption in women who had the highest alcohol consumption initially. The authors add that the brief intervention’s effect was higher when the partner was present. Regardless of how significant the findings of this study may be and how strong is the evidence obtained from randomized controlled trials, some flaws in execution of this design cannot be ignored, such as insufficient background data regarding the research problem, possible bias due to lack of blinding and a narrow sample that can’t be generalizable to the entire pregnant population that struggles with alcohol problems during the pregnancy.

Problem and Purpose

From the very beginning the authors clearly state the purpose of the study and the hypothesis that they plan to test. The authors stated that “the purpose of this randomized trial was to test the effectiveness of a brief intervention enhanced by including a partner chosen by a pregnant woman” (Chang et al., 2005). Their hypothesis was that both groups will experience a decrease in alcohol consumption, but the group that will have a brief intervention with a partner will have a more significant decline than the control group. However, the problem statement is not very clear in regards to the relationship between the independent variable and the dependent variable. The authors identify that drinking while pregnant is a significant problem and drinking even less than a drink a day can adversely affect fetus. In 1999 12.8% of pregnant women consumed alcohol, which is higher than the Healthy People 2000 and 2010 goal of 94% alcohol abstinence during pregnancy. Further they note that even though studies support the influence of a partner on prenatal health behavior modification, generally partners are not screened for health behaviors that may affect pregnant women’s health habits. Then they give an example of 2 European studies that identified that partner’s smoking habits were influential on smoking cessation while pregnant. But is not clear what exactly these studies entitled and how their methods and concepts are applicable to the current study. The studies are just briefly mentioned. The authors don’t provide sufficient literature review and evidence on how the high prevalence of alcohol consumption during pregnancy is related to the presence or lack of partner involvement in prenatal care and education. They don’t state why this study is needed and how it will contribute to the solution on decreasing alcohol consumption during pregnancy. Nevertheless, I think that the consideration of involving a partner during interventions aimed to reduce and prevent alcohol consumption during pregnancy is significant for nursing practice because nurses play a crucial role in patient’s education and helping them with health behavior modification. However, when it comes to alcohol consumption involving a partner should be the patient’s choice due to confidentiality issues and it should be decided on case-by-case basis.

Research Design

The randomized controlled studies are considered the “gold standard” for therapeutic interventions studies and are number two at the top of the evidence hierarchy (Polit & Beck, 2014). This research is a randomized controlled trial and I think that authors chose the best study design possible, because their goal was to test the effect of an intervention that included a partner on alcohol consumption outcomes. To test their hypothesis the authors randomized the subjects in a control and treatment group. The control group was exposed to the diagnostic interview and the treatment group had a brief intervention with a partner. Yet there are some ambiguities on comparisons of interventions among groups with respect to the purpose and problem statement of the study. For instance, in the introduction the authors stress that involving a partner in the intervention is a primary focus of the study. This is also emphasized in the results section: “the effects of the brief intervention were significantly enhanced when a partner participated” (Chang et al., 2005). Nevertheless, the interventions that were done for treatment and control group seem to be different at more levels other than a partner being present. The control group and the treatment group received diagnostic interviews where they completed the Alcohol Timeline Followback that estimated participants’ daily drinking 6 months before study, Alcohol Abstinence Self-Efficacy Scale that measured participant’s perceived temptation to drink and their ability to abstain and Healthy Pregnancy Facts that consisted of 7 statements on which subjects were supposed to answer true or false. The treatment group intervention consisted of 2 additional steps. First, the partners had to complete the Health and Habits Survey, National Institute on Alcohol Abuse and Alcoholism quantity-frequency questions regarding their personal alcohol consumption for the past 30 days and the same survey about their pregnant partner’s alcohol consumption for the past 90 days and the Healthy Pregnancy Facts. Second, the pregnant women and their partners participated in a short, 25-minute intervention in which the following were discussed: “knowledge assessment with feedback” where the investigators answered questions and clarified the Healthy Pregnancy Facts, “goal setting and contracting” where the subject’s abstinence goals and the rationale behind them were discussed, in “behavioral modification” section the common situations where drinking might be tempting and alternatives to drinking in those situations were discussed and finally, during “summary” the interview was summarized and the couples were given a copy of the summary. The interventions for the control group and treatment group are quite different, in the control group no feedback was given, no goals were set, the behavioral modification wasn’t discussed and a summary wasn’t provided, so any significance in results cannot be attributed only to the fact that a partner was present.

Another problem in the design execution is that the biases weren’t eliminated. The investigators weren’t blinded and they knew of the role of each group in the study, that could cause bias in data collection and in the way the intervention was delivered. In addition to that, in many cases the same investigator, who did the initial interview, did the follow-up postpartum interview. This could’ve caused two problems: the investigators might’ve been biased during the follow up data collection because they knew that the couple was given an intervention and the subjects might’ve been compelled to give false answers due to being embarrassed of not modifying their drinking behavior after the intervention.

I think that the data collection points are appropriate for the type of design chosen. A pretest-posttest design is incorporated, which involves data collection before and after the intervention. In true experiments is important to collect baseline data before intervention and afterwards to ensure that appropriate comparisons are done and to be able to measure the effect of the intervention (“Experimental designs,” n.d.). In this study the researchers collected the data regarding alcohol consumption from the control and treatment group during the diagnostic interview by using the Alcohol Timeline Followback to estimate their drinking habits 6 months before enrolling in the study and the Alcohol Abstinence Self-Efficacy Scale; at the postpartum follow up they completed the same tests except that the Alcohol Timeline Followback was used to estimate subject’s daily drinking from the enrollment in the study until delivery. In addition to these surveys, the partners of the subjects enrolled in the treatment group completed the National Institute on Alcohol Abuse and Alcoholism quantity-frequency questions regarding their personal and their pregnant partner alcohol consumption before the brief intervention and at the post-partum follow up.

Overall, the design chosen and the data collection points are effective methods for studies investigating treatment-type of questions, but the comparisons between treatment and control groups need to be better stated and more methods are needed to reduce bias.

Population and Sample

The study methods regarding sampling and population have their strengths and weaknesses. Some strengths are the thorough description of the study participants, the details about the recruiting procedures, as well as the statistical measurements used to estimate an adequate sample size. The authors clearly state the inclusion and exclusion criteria for the study. There were 4 inclusion criteria. First, the women had to have a T-ACE score of 2 or more. T-ACE is an assessment for drinking problems and is considered positive if the person has score of 2 or more. The criteria assessed are as follows: T-alcohol tolerance, A-annoyed by others criticizing the drinking problem, C-person’s feeling of the need to cut down on drinking and E-having an “eye opener” in the mornings to deal with hangover. Second, the women had to be at high risk for alcohol consumption during pregnancy based on history of drinking 3 months before study enrollment, including while pregnant or having at least a drink a day for the past 6 months or drinking during prior pregnancies. Third, women had to have the intention to carry the baby to term and be less than 28 weeks pregnant. Fourth, they had to agree to the “study terms, which included randomization to treatment by computer assignment, postpartum follow-up interview, selection of a partner who would participate in a diagnostic interview, the brief intervention if so randomized, and postpartum interview” (Chang et al., 2005). From the last statement it seems that the subjects were exposed to too much information about the study and were made aware of their possible role in the study that could lead to bias from the part of subjects. They excluded participants based on 5 criteria: receiving treatment for alcoholism, physical dependence on alcohol, other substance abuse, inability to complete the surveys and questionnaires, and the intention to abort the fetus.

Furthermore, the authors describe with details the process of recruiting the study participants. During a 31-month period they received 2,927 Health and Habits screening surveys, 89% of them being from women who registered for prenatal care at Brigham and Women’s Hospital, the rest were from study announcements. Out of 78% from the 802 respondents with a positive T-ACE that initially agreed to be contacted for the study, 399 were successfully called. Some didn’t meet the criteria and some refused to participate, resulting in a sample of 304 women that were equally divided in groups of 152 subjects each. The sample recruited was as estimated during the sample size calculations. The size of the sample needed was calculated with a confidence level of 95% and a power of 90%, based on the need of 1:1 ratio among groups, expecting that 50% from control group and respectively 70% from the treatment group would abstain from alcohol. The authors state that the estimations of abstinence are based on previous studies; however no further details are given. Based on calculations, they estimated that the sample should contain at least 268 subjects and 295 if a10% attrition was considered. The attrition in this study ended up being 5%.

According Polit & Beck (2014), nonprobability sampling is a broad class of sampling design in which “researchers select elements by nonrandom methods in which every element selected usually does not have a chance to be included” (p.178). I think that the sampling design used in this research is a nonprobability type of sampling, specifically a convenience sample that included choosing the most available “people as participants” (Polit & Beck, 2014, p. 178) because 89% of participants were recruited from a single clinical practice. According to University of California website “Non-probability samples are limited with regard to generalization. Because they do not truly represent a population, we cannot make valid inferences about the larger group from which they are drawn” (“Nonprobability samples”, n.d.). This reduces the study generalizability to the entire pregnant population that is vulnerable to prenatal alcohol consumption. The sample in this study ended up being limited with respect to participants’ socio-demographics. The following characteristics were predominant in this sample: 78.6% of the participants were white and 80.5% were married, median age was 31.4 and with a education level of bachelor’s degree. The median income for the geographical area where the participants lived was about $55,357/year, which is higher than the average income of $50,587 in Massachusetts where the study was done. Just by looking at this sample’s demographical data, we can see that the study didn’t include a variety of participants, including groups that might be high-risk for alcohol consumption, such as those from lower socio-economic backgrounds, younger mothers, minorities, single mothers and mothers without college and high school degrees.

Overall, the study had an adequate sample size, which was clearly described and identified; however the sample is not truly representative for the population.

                                                         Conclusion                                    

All in all, this experimental study investigated important factors that might help reduce prenatal alcohol consumption. With clearer stated problem statement and research background, double blinding and a probability sampling design the results of this study could have a higher validity and generalizability.

 

 

References

Chang, McNamara, Orav, Koby, Lavigne, Ludman, Vincitorio, Wilkins-Haug. (2005). Brief Intervention for Prenatal Alcohol Use: A Randomized Trial. Retrieved from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1380262/

Experimental Designs for Research. (n.d.). Retrieved from http://www.csulb.edu/~msaintg/ppa696/696exper.htm

Polit, D. F., & Beck, C. T. (2014). Essentials of Nursing Research: Appraising Evidence for Nursing Practice (8 ed.). Philadelphia: Wolter Kluwer/ Lippincott Williams & Wilkins.

Types of samples. (n.d.). Retrieved from http://psychology.ucdavis.edu/faculty_sites/sommerb/sommerdemo/sampling/types.htm