Critiquing the sampling procedure
Detering et al. (2010) have brought forward the hypothesis that “coordinated advance care planning will improve end of life care, the perceptions of the quality of care, and levels of stress, anxiety, and depression in surviving relatives” (p. 2).
The target population includes patients admitted to a large university hospital in Melbourne, Australia. The name of the hospital has been concealed for protection of participants. The patients that form the population must be admitted under internal medicine, cardiology, or respiratory medicine (Detering et al., 2010). The patient must be aged 80 years and over. The patient must be competent and able to speak English. Non-competent patients are those who were unlikely to understand what was to be discussed.
Probability sampling methods were used to choose the sample. In probability sampling, each participant has an equal chance of being selected (Denscombe, 2010). The researchers used simple random sampling to select those patients who were to be grouped under the intervention group or the control group. Participants were required to pick sealed envelopes that contained hidden random numbers used to categorize patients. Each patient had an equal chance to become part of any group. The sampling design is identified as a randomized controlled trial (Detering et al., 2010).
The random sampling technique is appropriate for the study. Eliminating patients who do not speak English and non-competent patients ensure that the remaining patients can be influenced by the intervention. The intervention involves the training of patients and the application of advanced care planning. Advanced care planning enables patients to make informed decisions about their end of life health care under the guidance of their chosen family member, and a medical practitioner (Detering et al., 2010). The simple random technique ensures that there is a fair distribution of patients in the two groups. The two groups that needed to have the population characteristics were the intervention group and the control group.
There are a few demographic characteristics given about the sample. One is the death rate of the people aged 80 and over in the hospital. They accounted for 51% of deaths at the hospital (Detering et al., 2010). They have also included the age of patients and their sex. The intervention group consisted of 54% male and 46% female. The control group consisted of 41% male and 59% female (Detering et al., 2010). The main admission diagnosis also formed part of the demographic.
The sample size is adequate because it has been based on the anticipated number of deaths. The number of deaths was estimated based on a pilot survey on the number of deaths that occurred in the hospital six months before the main research. It was expected that 22 deaths were necessary to provide a 95% confidence interval. It would have required the recruitment of 629 patients. The researchers recruited about half the number because of a higher death rate among those enrolled. The sample was considered adequate, provided that the number of deaths was 22 in each group. The recruitment was a continuous process. They stopped the recruitment after the number of deaths reached 22 in each group (Detering et al., 2010). It took seven months to record the required number of deaths. The number of deaths was representative of the population. However, the sample did not completely reflect the population because of the exclusion of patients who did not speak English and non-competent patients. These have been identified as potential sampling biases in the article. The matter of subject dropout has been identified and discussed in the tables.
Critiquing the data-collection procedures
The research discusses the people involved in data collection. Medical researchers labeled as KMD or WS were responsible for the data collection. They selected patients in the sample as well as identifying those for exclusion. The KMD researchers followed up with a questionnaire or interview. An interview was done three days after a patient’s admission to the hospital in the three selected categories (respiratory, cardiology, and internal medicine). At least three questionnaires were completed by the patients and their selected family member. They included one for post-traumatic stress, satisfaction with the program, and quality of end of life care. Interviews were also conducted through telephone in the follow-up period that lasted six months (Detering et al., 2010). The data were collected between August 2007 and March 2008 for enrollment. In addition, six months were used for the follow-up procedure.
The appropriate level of measurement was used based on the three questionnaires. A 14-item tool was used to investigate the level of distress on the family members. A 15-item tool was used to identify the possibility that one had developed post-traumatic stress. An 8-item tool was used in ranking the quality of end of life care perceptions. The article does not give a full report on all the items used to measure variables. It uses a report on the 56 patients as an example of the type of data they collected. Detering et al. (2010) have included items such as “wishes unknown,” “wishes are known and followed,” and “wishes known but not followed” (p. 5). On satisfaction, the ratings were also three. They included very satisfied, satisfied, and not satisfied.
The researchers do not specify any exercise to measure the reliability of the measurement tools. However, measurement tools have been developed by multiple researchers. It may be assumed that an inter-rater reliability test has been conducted. There is no evidence that a validity test was conducted. Golafshani (2003) argues that reliability and validity tests are not essential in qualitative research. The trustworthiness of the researchers may be counted as reliability and validity. A pilot survey was not conducted using the instruments. However, advance care planning was based on an existing model known as the Respecting Choices program (Detering et al., 2010). As a result, the effectiveness of the advance care planning provided by the researchers to the patients could be guaranteed.
Critiquing data-collection methods general criteria
The data collection methods have been discussed thoroughly in the article. Interviews and questionnaires were appropriate to collect the qualitative data required for the research. The interviewer used telephone calls. They were more convenient to cover the six months follow up after the discharge of patients. The research used both psychological methods and self-report to gather valid information. The psychological data was used to gather data on variables such as distress and anxiety. The subjects were asked questions without knowing the expected outcome (blind participants). Self-report was used on items such as satisfaction level. The subjects were asked directly whether they were satisfied with the program. Two methods were used to collect data. They included interviews and questionnaires. Face to face interviews was conducted at the hospital, and interviews through telephone were used as a follow-up procedure.
Questionnaires
The information has been provided on the number of questions that each questionnaire had. There is no information on the amount of time needed to complete the questionnaire. The response rate on the completion of the questionnaires has been displayed in all four tables. The effect of sampling biases has been discussed with regards to non-competent patients and those who could not speak English. The anonymity of the subjects has not been assured. However, the researchers have acted in a manner that protects the anonymity of the hospital and the respondents.
Interviews
The information has not been provided on how long the interviews would have taken. The subjects were not trained specifically on responding to interviews. There is no information given by the researchers on ensuring the confidentiality of the respondents.
Critiquing descriptive statistics
The types of descriptive statistics used include percentages and a few measures of central tendency (Mendenhall, Beaver & Beaver, 2012). The percentages have been used to describe demographic characteristics and response rates. For example, 93% of respondents in the intervention group were very satisfied with the hospital stay compared with 65% for the control group (Detering et al., 2010). The median is one of the commonly applied measures of central tendency. The median (60 minutes) has been used to describe the length of time used in advance care planning sessions (Detering et al., 2010). The interquartile range has been used to describe the variability of age in Table 1. The descriptive statistics are clearly presented in the text and on the tables. The values in the text match those found on the tables.
Critiquing inferential statistics
Inferential statistics refer to using survey results to draw conclusions about the population (Mendenhall, Beaver & Beaver, 2012). The inferential statistic widely used is the P-value. It represents the rejection area on a normally distributed sample. The P-value has been used in the text and tables. The researchers had selected a sample that allows a 95% confidence interval. The statistical significance level is presented as P<0.05. There are very few items in which the P-value exceeds the rejection area. Examples can be found in Table 3 (Detering et al., 2010). There is information that the researchers used the Stata version 9.2 as software for analysis. They carried out Mann-Whitney U tests, t-tests and chi-square test to analyze the data. However, there is little information provided on the outcome of the tests. The degrees of freedom have not been presented in any outcome.
The researchers have used parametric and non-parametric methods to ascertain the outcome of their findings (Pagano, 2012). The parametric method applied includes t-test. The non-parametric methods applied include the chi-square test, and the Mann Whitney U test (Detering et al., 2010). The parametric methods give credible results when the sample has a normal distribution. The non-parametric methods do not require the sample to have a normal distribution (Pagano, 2012). The sample has deviated from the population characteristics. The application of both method categories is necessary. The inferential statistics are presented for each item that builds up a single hypothesis in the form of P-values. The results are presented in tables and discussed in the text. The researchers thoroughly discuss the findings, implications, strengths, and weaknesses of the research. For example, they discuss that excluding non-competent patients means that the impact of advance care planning will be stronger in the population than in the sample. The reason is that such patients are likely to be placed on routine care without the consideration of advance care planning.
References
Denscombe, M. (2010). The good research guide: for small-scale social research projects (4th ed.). Maidenhead, UK: McGraw-Hill/ Open University Press.
Detering, M., Hancock, A., Reade, & M., Silvester, W. (2010).The impact of advance care planning on end of life care in elderly patients: randomized controlled trial. British Medical Journal, 340(c1345), 1-9. Web.
Golafshani, N. (2003). Understanding reliability and validity in qualitative research. The Qualitative Report, 8(4), 597-607. Web.
Mendenhall, W., Beaver, R., & Beaver, B. (2012). Introduction to probability and statistics. Mason, OH: Cengage Learning.
Pagano, R. (2012). Understanding statistics in the behavioral sciences. Mason, OH: Cengage Learning.