Critical Appraisal of Fish Oil in Knee Osteoarthritis

This essay critically evaluates the article Fish oil in Knee osteoarthritis: a randomised clinical trial of low dose versus high dose by (Hill et al. 2016) using the CASP tool. The study examines the effectiveness of high dose fish oil over low dose fish oil in improving symptomatic and structural outcomes in people with knee osteoarthritis. The study used a randomised controlled trial method and almost 202 participants were allocated randomly in two groups, a high dose group and a low dose group. The essay discusses the research question, eligibility criteria, sample, primary outcome, risk of bias, performance bias, detection bias, attrition bias and withdrawal rate, and conclusion.

1
Research and Evidence in Practice Lena Saleh ID# 19516083
Lena Saleh
La Trobe University
Research and Evidence in Practice
Critical Appraisal of Fish Oil in Knee Osteoarthritis

2
Research and Evidence in Practice Lena Saleh ID# 19516083
Critical Appraisal of Fish oil in knee osteoarthritis
Introduction
In this essay, the article Fish oil in Knee osteoarthritis: a randomised clinical trial of
low dose versus high dose by (Hill et al. 2016) is critically evaluated using the CASP tool. In
this study, the researchers tried to examine the effects of a high dose anti-inflammatory fish
oil with a lower dose of fish oil. In this study a randomised controlled trial method was used
and almost 202 participants were allocated randomly in two groups, a high dose group and a
low dose group. The researchers also tried to establish a relationship between the change of
cartilage degradation among Osteoarthritis (OA) patients after consuming the high dose fish
oil. In previous studies it was observed that the use of fish oil among patients of rheumatoid
arthritis was effective in preventing structural degradation (Di Giuseppe et al., 2014) and
researchers expected that it would improve the condition of the OA patients as the structural
degradation was common among the two types of arthritis (Pap & Korb-Pap, 2015).
Research Question
The study of (Hill et al. 2016) is based on the examination of the effectiveness of high
dose fish oil in comparison with the low dose supplementation in treatment of Knee OA in
older people. The research question of this study is, in people with knee osteoarthritis does
high dose fish oil have remarkable efficiency than low dose fish oil in improving
symptomatic and structural outcomes?
Eligibility Criteria
Researchers selected the populations who had pre-defined knee OA and the
determination of the disease was done by the following the American College of
Rheumatology (ACR) criteria. The subject population must be 40 years or older. In addition,
the visual analogue scale of knee pain was also used. A scale of 0-100 mm, the people that

3
Research and Evidence in Practice Lena Saleh ID# 19516083
had a knee pain score >20 mm was counted as an eligible subject for the study. However,
during the sample selection researchers followed a few exclusion criteria for this study which
were as follows:
1) People who have dementia or a disability to sign the consent form
2) Women who are pregnant or lactating
3) Previous knee replacement surgery
4) People who have contraindications to MRI scan
5) People with severe radiographic knee OA index that is grade 3 radiographic joint
space narrowing as per guidelines of Osteoarthritis Research Society International
Atlas.
People having one or more criteria were excluded from the study. In addition, before
starting the study a 4-week trial period was performed and, in this process, similarly
flavoured oil was used to detect any people that do not tolerate the oil and they were
eliminated from the study.
Sample
In the final study, there were two groups and the treatment were different for each group.
The first group received high dose fish oil and the second group was treated with low dose
fish oil. The high dose group composed of EPA 18%, DHA 12%. This mixture was supplying
4.5 grams of DHA plus EPA per day (Hill et al., 2016) The low dose group had a ratio of 1:9
low dose fish oil and sunola oil and was supplying 0.45 grams DHA plus EPA per day and
was equivalent to 1.5 standard 1 g fish oil capsule. Both groups were given Paracetamol
(500mg) and they were instructed that they could have the drug up to 8 times a day (Hill et
al., 2016).
Primary Outcome:

4
Research and Evidence in Practice Lena Saleh ID# 19516083
The primary outcome for this study was the knee specific pain scales according to the
Western Ontario and McMaster Universities (WOMAC) index at different times, 3 months, 6
months, 12 months and 24 months. Additionally, to this the changes in the cartilage volume
was also measured using the MRI scan at the 24-month mark. After that the secondary
outcomes that were quality of life, NSAIDs and analgesic use, WOMAC function, change in
the safety score and BML score were measured. The WOMAC knee pain index was
measured on a 10-point numerical scale. The assessor instrument for measuring the quality of
life and validated in both OA patients and general populations. The MRI scans of the study
was executed at commencement date and 2 years with a 1.5 T system.
Risk of Bias:
As bias cannot be measured researchers need to rely on good research design to minimise
bias (Al-Jundi & Sakka, 2017), In order to control allocation bias in the double blinded
randomised controlled study the researchers used computer technology for allocation of the
participants in the groups. A computer generated random allocation sequence was used
during the allocation process of the subject population and subsequent allocation was
executed centrally at a pharmacy. The participants were randomly assigned to the high dose
and low dose treatment groups after successful completion of the tolerance test of flavoured
oil that was performed before the final randomisation. The allocation sequence was concealed
throughout the whole study and both the participants and the assessors were unaware about
the allocation sequence (Hill et al., 2016).
Performance Bias:
In this study (Hill et al., 2016) the researcher had nullified the chances of allocation bias
by performing a study of 4-week duration, in this pre-randomised study the researchers used
citrus oil flavoured sunola oil that had was similar in fragrance with the fish oil used in the
actual study. The people who were unable to tolerate the smell of the oil were excluded from

5
Research and Evidence in Practice Lena Saleh ID# 19516083
the study. All participants were kept blind throughout the process in a successful way. After
the study it was observed that 52% of the total sample population were uncertain about which
group they were allocated, the remaining participants that thought they knew their allocation
group. In that case it was observed that 57% answered correctly and that the researchers
maintained the blindness well.
Detection Bias:
The staff involved in the care process of the participants were also kept in a blind
situation and along with this the MRI technicians did not know any information regarding the
experiment. Researchers used one trained reader during the assessment of the individual
cartilage pate volumes to remove detection bias. During this time researchers re sampled
according to cubic and bilinear interpolation. The trained reader was completely unaware of
the process. During the BML assessment the reader had no information regarding clinical
data and treatment allocation along with the interclass correlation coefficient of 0.97 (Hill et
al., 2016).
Attrition bias and withdrawal rate:
During the study almost 54 participants that is almost 26.7% had not completed the study
and at different intervals they discontinued the study intervention practices. However, the
researchers continued the study for 2 years. In spite of discontinuation the overall study had
the follow up percentage of 84. All of the participants were assessed at 12 months and 24
months (Hill et al., 2016). However, at baseline the participants in each group were matched
except in the male / female population ratio. The pull-out rate of the participants was higher
in the high dose group in comparrison to the low dose group. It was observed that the pull-out
rate was 35% among the participants that had the high dose while the participants that had the
low dose had a pull out rate of 20%. After the final allocation, each group initially had same
number of subject (n=101). Within a 1 year period 32 participants discontinued the process

6
Research and Evidence in Practice Lena Saleh ID# 19516083
from the high dose group and the number of withdrawal from the low dose group was 17.
From the time period of 1 year to 2 years, 3 participants from the high dose group
discontinued the process and the same number of participants withdrew from the low dose
group (Hill et al., 2016). The withdrawal cases in the high dose group happened within 3
months of starting the study whereas the withdrawal cases of the low dose group mainly
happened after 7.5 months into the study. It was also observed that randomisation female
participants were more allocated to the high dose group than that of the low dose group. In
the high dose group, the percentage of female participants were 59% and in the low dose
group the percentage was 40 having a p value of <0.01 (Hill et al., 2016). Attrition from
randomised trials can introduce bias and reduce the effectiveness of the study. Dropout rates
from a specific experiment group can create a false treatment effect and influence the
statistical power. The sample size should be large enough to start with. It is recommended
that researchers try to obtain data about the dropouts from other sources, try and input
missing data, perform multiple types of analysis including pre-protocol and ITT scenario.
Some strategies on improving attrition bias include motivation and engaging participants and
sites to optimise data return or compliance, follow up procedures, communication via email,
questionnaire length long/short, incentives and assistants to manage participant follow up
(Brueton et al., 2011).
Conclusion:
In conclusion it was observed from the study that the low dose group had a lower pain
score than that of the high dose group at 18 and 24 months, and the low dose group also had a
higher functional limitation score at 24 months. Along with this no significant change was
observed between the groups regarding the cartilage volume from the date of commencement
to 24 months. (Hill et al., 2016). It can be concluded that the researchers failed to establish

7
Research and Evidence in Practice Lena Saleh ID# 19516083
the effectiveness of high dose over the low dose with a statistically significant result and that
the study did not support the hypothesis.

8
Research and Evidence in Practice Lena Saleh ID# 19516083
Reference list
Al-Jundi, A., & Sakka, S. (2017). Clinical Appraisal of Clinical Research. Journal Of
clinical And Diagnostic Research, 10(7), 241-264. doi: https://dx.doi.org/10.7860%2FJCDR
%2F2017%2F26047.9942
Brueton, V., Tierney, J., Stenning, S., Nazareth, I., Meredith, S., Harding, S., & Rait, G.
(2011). Strategies to reduce attrition in randomised trials. Trials, 12(S1). doi: 10.1186/1745-
6215-12-s1-a128
Di Giuseppe, D., Crippa, A., Orsini, N., & Wolk, A. (2014). Fish consumption and risk of
rheumatoid arthritis: a dose-response meta-analysis. Arthritis Research & Therapy, 16(5).
doi: 10.1186/s13075-014-0446-8
Hill, C., March, L., Aitken, D., Lester, S., Battersby, R., & Hynes, K. et al. (2016). Fish
oil in knee osteoarthritis: a randomised clinical trial of low dose versus high dose. Annals of
The Rheumatic Diseases, 75(1), 23-29. doi: 10.1136/annrheumdis-2014-207169
Pap, T., & Korb-Pap, A. (2015). Cartilage damage in osteoarthritis and rheumatoid
arthritis—two unequal siblings. Nature Reviews Rheumatology, 11(10), 606-615. doi:
10.1038/nrrheum.2015.95