Outcomes Following a Programme for Lifestyle Changes With People With Hypertension, Article Critique Example

This study investigated the effectiveness of a nurse-led intervention of lifestyle changes in people with hypertension. The idea behind the study was to use a non-pharmacological approach to address hypertension in patients by counseling the patients on issues of weight control, smoking, alcohol consumption, diet, physical activity, and control of stress. Patients in this study were typically receiving pharmacological treatments for their hypertension at the start of the study, although in many cases (detailed in the paper), those medications were changed during the 15 months of the study. While not explicitly stated, the foreground question of this paper was, can patient-centered counseling on lifestyle changes can be effective in control of hypertension? While not explicitly stated, the implicit hypothesis was that such an intervention could be effective at control of hypertension via lifestyle changes. This critique of Drevenhorn et al. (2007) assesses the paper for quality of research design, and quality and applicability of its findings for nursing clinical practice.

Methodology of Drevenhorn et al. (2007).

Population, Sample, and Sampling Design

177 patients of a health center in southern Sweden who had been diagnosed with hypertension were invited to participate. 100 accepted the invitation. At the beginning of the study, each patient was given a thorough physical check and offered information about their specific conditions (i.e., hypertension, cholesterol, weight, etc.). All visited the health center every three months until the end of the study in 15 months. Interventions for individuals were based solely on their physical conditions at the start of the study, and included simple counseling on diet, individual counseling on diet after 1-2 days of food logging, counseling on smoking cessation, counseling on risks from high alcohol consumption, counseling on physical activity benefits, advice on dealing with stress, and information on medications taken. There was no randomization process of any sort used, nor was there any grouping of the participants except by their physical conditions at the start of study. 17 patients dropped out of the study over the 15 months. The sampling method used was a consecutive sample (Polit & Beck, 2012, pp. 275-289); all patients meeting the general inclusion criteria were invited to participate. Given the limited population size, and despite the potential for bias, this is not an unreasonable method to recruit participants.

Ethical Considerations

The authors of Drevenhorn et al. (2007) stated that they provided prospective participants with informed consent approved by the ethics committee of the University of Lund. The information collected was only that which would normally be collected in the course of a medical check-up.

Data Collection and Instrumentation

Data collection, as defined in Polit and Beck (2012, pp, 58-62) include self-reports and interviews, observations, and biophysiologic measurements.Data collected in this study was in the form of standard biophysiologic measures such as BP, weight, BMI, waist/hip ratios, and blood values such as cholesterol, triglycerides, HDL/LDL, plus whether the person smoked, and how much alcohol was typically consumed (both self-reported).These objective measures using clinical tests are considered highly reliable and valid. Information about alcohol intake and tobacco use is less reliable due to its being self-reported. Data was collected in a pre-test/post-test format before and after the four-week period of the study. These data collection methods provide an appropriate source for this study to determine biophysiologic changes. Data collection was done approximately every 3 months over the 15 months of the study. This is an appropriate interval to detect the types of biophysiologic changes being measured.

Results, and Analysis of Drevenhorn et al. (2007)

Quality of the Evidence

The quality of evidence of this study corresponds to Level IV evidence, a single correlational study (Polit & Beck, 2012, pp. 27-28). The reason for this ranking of the evidence is, first of all, that this study contained no control. Each participant had his or her own intervention tailored specifically to them. Furthermore, other aspects of the participants’ medical care were not maintained constantly throughout the study period. Medications of some (but not all) participants were changed during the course of the study to reflect the participants’ changing medical condition.

Validity and Reliability of Evidence

Validity of the evidence refers to the likelihood that conclusions drawn from the evidence are true (Polit & Beck, 2012, pp. 236-239). Steps that can be taken in designing a study to improve its validity include randomization of the participants, using a crossover design in which halfway through the study, the participants change groups from control to intervention and vice versa, and establishing a homogeneous group of participants (Polit & Beck, 2012, p. 237). None of these techniques were used in Drevenhorn et al. (2007). Participants received different interventions in a non-random way, with no control group at all. Thus, the validity of the results is questionable.

The reliability of the evidence refers to the likelihood that if this study were replicated on a different but similar group that similar results would result (Polit & Beck, 2012, pp. 175-180). In essence, reliability is a measure of how generalizable the results of this study are to other groups(Polit & Beck, 2012, p. 180). Here again, the quality of this study is such that conclusions drawn by the authors are unlikely to be generalizable to other groups because the interventions were so specific and the lack of homogeneity of the participants makes generalizing very difficult.

Statistical Methods Used

The authors stated that they examined the normal distribution for quantitative continuous variables, and performed a two-sided parametric test on normally distributed variables. They performed a nonparametric test (Mann-Whitney U-test) on other variables. They used a chi-square test to compare groups and used Pearson correlations (normally distributed variables) or Spearman (other variables). They achieved significance at p < 0.05. The statistical methods used to analyze the results of Devrenhorn et al. (2007) were straightforward and consisted of little more than simple descriptive statistics, combined with simple means and percentages, with p values reported only in the few measures deemed statistically significant. In part, this was appropriate because of the differences in interventions applied to the participants. Since each person received an individualized intervention, it is difficult to know how inferential statistics or multivariate statistics could be applied in an appropriate manner. The authors did not note if they attempted to adjust their results based on the drop out of 17 of the 100 participants over the course of the study.

Computer Hardware/Software Used

The authors used SPSS 10.1 statistical software, but did not specify the hardware on which it ran. The current version of SPSS is version 20.0.(Softonic, “SPSS,” 2012). Thus, version 10.1 is considerably out-dated, and likely was outdated at the time of the study. The closest date for that version located was January 2002, so the software used was at least five years old at the time this study was published in 2007 (Meehan, 2002).

Results and Conclusions of Drevenhorn et al. (2007)

Statistically Significant Findings

Statistically significant results were quite limited. Mean systolic BP for the participants dropped from 141.9 to 137.5 (p=0.015). Women’s mean weight dropped from 73.8 kg to 72.3 kg (p=0.005). The mean waist to hip ration of women dropped from 20 to 11 (p=0.005). Men’s mean triglycerides increased from 15 mmol/l to 17 mmol/l (p=0.034). No other results (another 14 measures) were noted as being statistically significant.

Other Findings

Other findings were interesting, however. In general, the patients increased their levels of exercise. None of the statistics got the patients into normal levels if those measures were not normal at the start of the study. Alcohol consumption did not change. Two of 14 smokers stopped smoking. Most importantly, however, medications for some participants, including medications for hypertension and cholesterol control, were adjusted individually for participants based on medical need over the 15 months of the study. Thus it is impossible to determine if the effect of the study derived from interventions or from a more appropriate medication regimen.

Recommendations and Application to Nursing

The authors themselves note that no real conclusions can be drawn from this study. They noted the need for a larger study, the use of randomization and a control group. They also noted that one year is quite short a time to determine if lifestyle changes made during this study are permanent. Thus, no real recommendations can be made on the basis of this study. There are no direct applications of this study to an evidence-based nursing practice.

Overall Assessment of Drevenhorn et al. (2007)

This study is interesting but is ultimately unsatisfying because the research design was poor. There is no real way to determine what effects derived from the interventions, there was no coherent uniformity of the interventions, and there was no way to define what results derived from changes in medication. Furthermore the results as presented in the paper are distinctly unimpressive. In essence, no real change was noted for 14 of 18 measures, and the changes in the other 4 measures, while statistically significant, were not impressive. Women lost about 3 lbs on average over 15 months. There were self-reported increases in exercise. On the other hand, men gained weight, and men’s triglycerides worsened.

References

Drevenhorn, E., Kjellgren, K. I., Bengston, A. (2007). Outcomes following a programme for lifestyle changes with people with hypertension. Journal of Nursing and Healthcare of Chronic Illness in association with Journal of Clinical Nursing, 16 (7b), 144-151. doi: 10.1111/j.1365-2702.2005.01493.x

Meehan, A. (2002). Introduction to SPSS 10.1 for Windows (2^nd ed.) (Prepared using screen shots from SPSS 10.1 tutorials) [PDF document].Retrieved from from Kutztown University Web site: http://faculty.kutztown.edu/meehan/SPSS10.pdf

Polit, D. F., Beck, C. T. (2012). Nursing Research: Generating and Assessing Evidence for Nursing Practice. Phildelphia, PA: Wolters Kluwer Heath/Lippincott Williams & Wilkins.

Softonic. (2012). SPSS (Version 20.0). [Software] Available from: http://spss.en.softonic.com/