Critical Analysis Of The Article "Preventing Winter Falls"

This essay aims to have a critical review of the paper: “Preventing winter falls: a randomised controlled trial of a novel intervention” by Parkin, Williams & Priest (2009). The article is published in the New Zealand Medical Journal (NZMJ), the principal scientific journal to the profession in New Zealand according to the New Zealand Medical Association (NZMA). The purpose of the authors is to investigate the hypothesis that wearing socks over shoes can improve the traction on icy footpads. The reason for bringing this hypothesis is the gap the authors found in the falls prevention research on Medline and the Cochrane Library. The authors want to investigate this because of some anecdotal reports they have heard about, about pedestrians who wear socks over the top of their footwear are less likely to slip and fall, when they are in icy conditions.

The authors let pedestrians walk along an icy footpath and investigated whether the influence of socks over shoes had an effect on self-rated assessments and observer-rated assessments. To have a complete review of this randomized controlled trial (RCT) the Consolidated Standard of Reporting Trials (CONSORT) statement will be used. This statement is found in 1996 after a period of reported inconsistencies in trials to facilitate better reporting (Bhatti et al. , 2014; Turner et al. , 2012). It has been revised twice, the latest in 2010, and contains an evidence-based minimum set of recommendations, consisting a checklist, flow diagram and descriptive text. The question if this statement is effective in the completeness of reporting the RCT in journals is evaluated with a systematic review from Turner et al. (2014). The results suggest that journal endorsement of CONSORT may benefit the completeness of reporting of RCTs and no evidence is found that endorsement hinders the completeness. The authors also state that journals need to take action regarding their endorsement and implement CONSORT to facilitate accurate, transparent and complete reporting of trials (Turner et al, 2014). The CONSORT group developed and extension for RCTs that are based on a nonpharmacological intervention. The reason for this extension is a difference in specific issues, for example blinding participants and assessors compared to pharmacological treatments. There are more variations, which CONSORT statement not specific address, that can influence the estimate of the treatment effect (Boutron, 2008). It is a great opportunity to use the CONSORT statement and the nonpharmacological extension in this particular case, to review the article by Parkin et al. (2009) as well as the decision the NZMA made to publish this RCT in the NZMJ. The main topic in the article is as the title states winter fall prevention. Falling is a problem worldwide, especially in elder pedestrians. There is an increased number of falls in people aged above 65 with approximately 10% every year (Decullier et al. , Duckham et al. , Kelsey, Procter-Gray, Hannan & Li, 2012; WHO, 2007).

This results in increasing numbers of deaths, hospital admissions, emergency care visits and high medical expenses (WHO, 2007). Outdoor falls have been a neglected, hidden and under-researched public health problem (Feypell, Methorst & Hughes, 2010). Especially in countries where the temperature varies due to seasonal changings, there is an increasing risk of falls, especially among older women (Campbell, 1988). In the introduction there is no description of any of these problems in New Zealand, so it remains unclear why this RCT has been set up in the first place, besides the anecdotal reports the writhers mention in their introduction. This is a poor and cumbersome underlying story and does not body their problem to start an investigation. Parkin et al. (2009) thereby claim in the introduction part that methods to enhance footwear traction have particular relevance for their population. Evidence to support this statement is missing. The gap Parkin et al. (2009) are missing in their literature search among the terms “ice”, “falls”, “prevention” and “socks” is part of their rationale why this RCT is conducted. Slip risk, falling and their relationship with walking on socks might not have been researched before, despite that, there has been done research among anti-slip devices. Use of any anti-slip devices during winter is not common (Gao & Abeysekera, 2004). In two intervention studies the anti-slip devices are tested on effectiveness on ice and snow with different outcomes. A study from McKiernan (2005) tested a nonmedical gait stabilizing device on older participants with a significantly lower rate in slips and falls. The hypothesis is “wearing socks over shoes improves traction on icy footpaths”. It presupposes that the study will measure traction on icy footpaths. The intervention that has been applied does not actually test the differences in traction on icy footpaths, but examines the self-reported slipperiness, a subjective outcome.

The introduction can be concluded as poor as the authors did not succeed to address a clearly focused issue. They have failed to support and body their topic with background information of already tested similar applications which could be used in slippery conditions. Thirty pedestrians were eligible for inclusion in the trial as a result of intercepting passing pedestrians at the side of the road. There is no further description of any eligibility criteria used to select trial participants. The sites were chosen because many people walk by and so it does not seem important what characteristic value these participants have. The median age of the 30 pedestrians involved in this trial is 21 years old. A Canadian survey published by the Division of Aging and Seniors and P. H. A. O (2005) shows in figure 1 the incidence rate of fall injuries for men and women by 5-year age group. Previous numbers of WHO (2007) thereby already showed us that mostly people above the age of 65 have a higher incidence rate of fall injury and the frequency of falls increases with age and frailty level (Kannus P et al. , 2005). It is important to be critical on the characteristics of the participants if the real goal of the trial is preventing winter falls, as the authors claim in their title. It is an assumption to say that there might be a greater prevention possible if the participants in the trial are 65 years of older.

The site where the intervention took place were chosen because many employees, students and members of the public used these routes each morning. There is no description of any characteristics of the used footpath, such as the gradient of the footpath or the conditions of the footpath that particular day or any details about the amount of ice or snow on the footpath. The role of road factors are well described in qualitative research and plays an important role in pedestrian falls (Schepers, den Brinker & Methorst, 2017). The most frequently mentioned environmental factors are tripping hazards such as uneven surfaces and wet and slippery surfaces due to ice, snow, oil, etc. (Gyllencreutz et al. , 2015; Lai et al. , 2011; Li et al. , 2006; Naumann et al. , 2011; Nyman et al. , 2013). Eligibility criteria, setting and location information is crucial to judge the applicability and generalisability of a trial according to the CONSERT statement (Moher, 2010) and to judge the outcomes and the relevance of this outcome in the specific public health. Since there is none of this information it is a major lack of the RCT. Participants were asked to complete a questionnaire before the start of the intervention, which collected background information as for example earlier experience with icy conditions and type of footwear. There is no information given where this data is recorded and if this is a safe method of collecting data. It is an established principle that informed consent must be obtained from all participants in a research investigation on human subjects (Smith, Morrow & Ross, 2015). Verbal consent was given. The issue in the article here is that there is no information given if the potential participants received a comprehensive explanation about the research, for example what the risks or benefits were. False information is given to the participants about the purpose of the trial to avoid bias. It is questionable if the participants verbal consent was ethically responsible, based on the false pretences. Despite that the ethical approval was granted by the University of Otago Ethics Committee. The intervention itself is well-described for each group. The intervention group was wearing socks over their shoes and the control groups had no socks over their shoes. The individuals were directed to walk downhill.

After this stroll downhill the participants filled in an assessment form as a primary outcome, the validated self-rated slipperiness scale (Gao, Holmer, Abeysekera, 2008). Since there is no information in the article of Gao et al. (2008) about the validity or a described process to validate the measurement scale, it remains unclear if the self-rated slipperiness scale is validated. It is a false pronunciation to assume it is validated. Earlier research from Abeysekera and Gao (2001) show that there is a high correlation between subjective slipperiness reports and objective measures of friction (r=0. 90). This objective measure of friction is also known as the coefficient of friction (COF). Chang et al. (2001) reviewed static, dynamic and transitional friction measurement devises and concluded that the devices to measure COF are generally valid and reliable. However, the validity of the devised need to be improved to reflect actual human slipping conditions, states Chang et al. (2001). Slightly different is the use of the self-rated slipperiness scale in previous researches and this trial, because in this trial the intervention is downhill with an unknown incline rate and the scale in previous research is used in equal height floors and footpaths. The primary outcome is clearly defined. This subjective outcome is strengthened by the statistical significant correlation of the objective measures of friction and support in making a statement about the hypothesis. Secondary outcomes were measured with observer-rated slipperiness (using the same slipperiness scale as the participants), observer-rated confidence and time that was used to descend the study slope. The rate of slipping and the perception of slipperiness is well researched and strongly associated, unfortunately there is no validity described for the observer-rated slipperiness scale, which is used in the secondary outcome measurement, as well as the observer-rated confidence scale.

The principle that Moher et al. (2010) use is that the information provided should be sufficient to allow others to use the same outcomes. The secondary outcome measurement tools are not sufficient and strong enough following this principle, due to the absence of validity and inter-rate reliability. For allocation of the participants computer software was used to randomise at site. It is a lack that there is no information about the investigators/recruiters who put the allocation status (socks/no socks) in sealed envelopes and if they were blinded. Another issue in this trial is the fact that outcome assessors were not blinded. Lack of blinding of outcome assessors can be responsible for detection bias and is also associated with biased estimates of treatment (Juni, Altman & Egger, 2001). As blinding is more difficult to carry out in trials assessing nonpharmacological treatments or interventions (Boutron et al. , 2004).

Since there is no information if the investigators were blinded in the selection process and thereby the outcome assessor were not blinded it is a lack of the internal validity of this trial. The authors describe several other creative methods to minimise bias, as it was not possible to blind the participants during the intervention. The participants were told a different study hypothesis to conceal the exact nature of the study and despite other measures that were employed the success of this partial blinding method is questionable describes Boutron et al. (2004). In the results the participants flow is well described with the use of a flow diagram. This is recommended by the CONSORT statement, as research shows that reporting the right flow of participants through a trial is considerably more if there is a use of a flow diagram (Moher, 2010). The primary outcome shows a difference between the intervention group and the control group 1. 6 (SD 1. 14) and 2. 9 (SD 1. 32) respectively (difference in means 1. 3, 95%CI: 0. 4-2. 3). The comparison between the groups is measured with a t-test.

The t-test is a widely used and useful test to find out if there is a difference between 2 groups. But that there is a difference between 2 groups is not the whole truth, there could for example be a variance in one group that will result in better scores. The t-test does not take anything into account except with the distinction between the 2 groups. That is why it is also important to analyse the spread of the scores. The outcome results in this article are only averages between the groups and no further data, so it is not possible to use the outcomes in different comparisons methods. Parkin et al. (2009) result that wearing socks over footwear significantly improved traction. This statement is based on the self-rated slipperiness and objective measures of friction correlation, which is statistically significant as described before. The authors make a strong conclusion on the fact that decreased self-reported slipperiness automatically results in improved traction. The secondary outcomes show no statistically significant results between the intervention and the control group. The authors result in their secondary outcomes that a higher proportion of the intervention group (71% versus 53%) appeared more confident while descending the study slope. The authors describe this finding in their abstract as an important finding. It is a flaw that this p-value is not called, since it is not statistically significant with a p=0. 45. Some challenges are presented in the discussion part. Blinding the participants and outcome assessors was not possible.

The authors try to reduce the impact of bias and claim that obfuscation of the exact hypothesis reduced the potential for bias. As we have seen before this partial blinding method is questionable in the reduction of bias. There could have been a creative method to blind the assessors to the intervention, for example with the use of a screen to shield the footwear of the participants below the knee and keep the intervention out of sight of the outcome assessors or use a shield around the feet of the participants below the knee. I challenge the fact that it was not possible to blind the outcome assessors. Limit the risk for bias in situations where it is difficult to blind a participant or outcome assessor, it is better to use an objective primary outcome (Boutron et al. ,2008) instead of a subjective one used in this trial. The investigators struggled thereby to set a date for data collection, because they needed a slippery footpath for their intervention. It is a flaw that the authors don’t point out how slippery the footpath was on the specific date.

As Gao et al. (2008) discovered the surface does impact on the rate of slipperiness, as ice covered with snow was rated as the most slippery, followed by melting ice and icy surface. It is a cheap and simple solution to reduce ice-related falls in the population of New Zealand, that is the simple conclusion of the socks-over-shoes intervention. This conclusions are based on the used material in this trial. There are some important flaws in the material used in the results. The authors did not succeed in addressing a clearly focused issue as a rationale for the study. The link with the title of the RCT is nowhere described. The methodology of the study as well as the statistical methods are good and well-described with clear outcome parameters, though it is a flaw that the measurement tools are non-validated. There is a big lack in the internal and external validity of this study. The design of the trial leads to a high risk of bias, where the participants and outcome assessors were not blinded to the intervention. The generalizability is poor as there is no information given to the specific conditions that day of the intervention site. Can similar results be expected on another day, with less or more slippery conditions and with other participants? The authors make strong conclusions on the outcome results with sufficient data presentation, with the use of a non-validated primary outcome measurement tool and with the use of a single statistical calculation method with the presence of a confidence interval. In the discussion some limitations of the study are described, but the authors neglect the poor validity of the trial and the fact that it is not generalizable and so the credibility of the study decreases to a mediocre level.