XXXXXXXXXXXXXXX: A systematic review and meta-analysis

Hao Zhang MD 1, , Mario Cantó-Cerdán OD, PhD 1, Jorge L. Alió del Barrio MD,

PhD, FEBOS-CR, FWCRS 1,2*

1. Cornea, Cataract and Refractive Surgery Unit, Vissum (Miranza Group), Alicante,

Spain.

2. Division of Ophthalmology, School of Medicine, Universidad Miguel Hernández,

Alicante, Spain.

Corresponding author *
Author address for correspondence and reprint:
Dr. Jorge L. Alió del Barrio MD PhD FEBOS-CR, FWCRS, Avda de Denia s/n,
Vissum (Miranza), 03016 Alicante, Spain, Tel. 34 965150025,
[email protected]

Keywords:
Abstract
Purpose

Setting

Methods

Results

Conclusions
INTRODUCTION
Corneal ectasia is a progressive degenerative ocular disorder characterized by
thinning, protrusion, and structural alterations in the central, paracentral, and
peripheral areas of the cornea, such as keratoconus, iatrogenic corneal ectasia, and
pellucid marginal degeneration. This continuous thinning and deformation
significantly modifies the refractive power of the cornea, resulting in distorted vision
1
and substantially impaired quality of life. Among them, keratoconus is the most
common corneal ectasia affecting approximately 1:7,500 in the relevant age category
annually.2
In recent decades, new treatment methods have evolved and changed our perspective
on corneal ectasia management. Corneal cross-linking (CXL) was introduced in the
late 1990s as a therapeutic approach to induce covalent cross-linking between
collagen molecules, fibers, and microfibrils to strengthen the biomechanical and
biochemical properties of the cornea, which is effective and cost-effective in halting
disease progression in most mild or moderate cases of keratoconus with a single
treatment.3-5 However, it does not adequately address the problem of irregular
astigmatism, leaving many patients dependent on rigid contact lenses to attain
satisfactory visual quality. Recent studies on treatment combinations have
demonstrated that performing simultaneous photorefractive keratectomy (PRK)
followed by CXL is promising in providing patients with both visual acuity
improvement and corneal stability.6-9 The potential advantages of simultaneous
treatment include minimized time-off work, accelerated visual rehabilitation,
preservation of previously crosslinked corneal tissue, and unaltered tissue ablation
rates.10 Nevertheless, the primary risk associated with combining PRK with CXL for
keratoconus is the potential for corneal stromal tissue removal during PRK, which
may compromise corneal biomechanical stability and counteract the stabilizing effect
of CXL.11 While promising at the individual study level, the collective effects of these
treatments on visual acuity, corneal stability, and biomechanical integrity are
inconsistently documented owing to variations in outcomes and methodologies across
existing studies comparing the effectiveness of CXL alone versus its combination
with PRK. Therefore, definitive conclusions remain elusive.
The aim of this meta-analysis is to clarify the efficiency and safety of the combination
of CXL and PRK, thereby offering a clearer direction for clinical practice and patient
management in the treatment of corneal ectasias.

METHODOLOGY
This research adhered to the guidelines set forth by the Preferred Reporting Items for
Systematic Reviews and Meta-Analyses (PRISMA) consensus statement and followed
the methodological standards prescribed in the Cochrane Handbook for Systematic
Reviews of Interventions.12-14

Method of Literature Search:

Two reviewers (HZ and MCC) independently conducted a systematic literature search
in the MEDLINE/PubMed and EMBASE databases, covering the period up to April
2024, without language restrictions. The search utilized the following keywords:
"Keratoconus," "Corneal Cross-Linking," "Cross-Linking, Corneal," "CXL,
Epithelium-On," " CXL, Epithelium-Off." "Photorefractive Keratectomy,"
"Keratectomy, Photorefractive," and "Custom Ablation Treatment." These keywords
were searched for in the title, abstract, and MESH/EMTREE terms. Additional details
regarding the search terms are provided in the supplementary material. Abstracts of
relevant titles were reviewed, and full articles were retrieved if they met the objectives
of this review. Studies were included if they met predefined inclusion criteria.
Furthermore, reference lists of the included papers were screened, related meta-
analyses and systematic reviews, and trial registries (such as clinicaltrials.gov and the
Cochrane Controlled Register of Trials) were manually searched to identify any
additional relevant studies.

Study selection, inclusion, and exclusion criteria:

All publications were screened by two authors (JLAB and MCC), and any
disagreements were resolved through discussion and consensus. Reports were
assessed based on the following criteria: (a) patients with progressive keratoconus or
other corneal ectasias; (b)interventions: CXL alone and CXL plus PRK; (c) clinical
trial studies; and (d) both randomized controlled trials (RCTs) and non-randomized
studies of interventions (NRSI), aimed at generating evidence for guiding medical
decisions. The classification of the studies was based on their scientific level of
evidence according to the General Guidelines for Methodologies on Research and
Evaluation of Traditional Medicine of the World Health Organization. 15 Only articles
with scientific evidence of a level of Ib or IIa were included. A careful review was
conducted to select articles reporting original clinical data pre- and postoperatively,
excluding duplicate data from previously reported cases across different articles. The
control group of studies exclusively utilized CXL as a control group, without
involving combined treatments such as intrastromal corneal ring segments or various
photorefractive keratectomy techniques. The experimental group underwent CXL plus
PRK surgery or similar techniques based on the same surgical principle with different
names. Animal studies, ex vivo reports, and studies with less than a 6-month follow-
up period were also excluded.

Quality assessment:
To avoid the risk of bias, two dependable authors (JLAB and MCC) separately created
a synopsis based on the Cochrane risk of bias tool,16 based on the methods
recommended in the Risk of Bias in Non-randomized Studies of Intervention
(ROBINS-I) for cohort studies founded on seven domains (confounders, selection of
participants into the study, classification of interventions, deviations from the intended
intervention, missing data, measurement of outcomes, and selection of the reported
results).

Data extraction and clinical outcome

Two reviewers (JLAB and MCC) reached a consensus on the data extraction process
from the included studies following previously established methods. The extracted
data included the name of the first author, publication year, trial location, study
design, number of eyes, types of CXL and PRK protocols used, outcome measures,
and follow-up duration. In this study, we designated visual acuity (in LogMAR) and
refractive index as the primary outcomes to assess direct improvements in visual
function and the stability of disease progression post-treatment. Secondary outcomes
included higher-order aberrations (HOAs), which were evaluated to determine
changes in visual quality. Additionally, central corneal thickness (CCT) and corneal
eccentricity (ECC) were measured to assess biomechanical and structural changes in
the cornea. We also considered adverse events and treatment failures, including the
progression of keratoconus, loss of two or more lines of corrected distance visual
acuity (CDVA), delayed epithelial healing, persistent stromal haze, sterile infiltrates,
and infections, to provide a comprehensive overview of treatment safety and efficacy.

Statistical analysis
Intergroup outcomes were reported as differences between the PRK-CLX group and
control group. A meta-analysis was performed to synthesize the intergroup outcomes
of the studies included in this systematic review using Review Manager 5.0. During
the analysis, standardized mean differences were calculated to analyze continuous
parameters if different data measurement methods were used among the studies.
Standardized mean differences are a measure of the size of the intervention effect in
each study with respect to the variability within the study, which allows the analysis
of the results on a uniform scale. The absolute values were interpreted together with
the p-values and 95 % confidence intervals (CI) presented in forest plots. Statistical
significance was set at p < 0.050. The heterogeneity of the included studies was
analyzed using the Cochrane Q-statistics chi-square (Chi 2) and I-square (I2) tests,
which were graded as low (<25 %), moderate (25 %–50 %), and high (>50 %). 17 If
there was any significant heterogeneity between studies (I 2 test ≥ 50 % or Chi2 test
with a p < 0.100), a random-effects model was used to pool the data; otherwise, a
fixed-effects model was used.18 In addition, a sensitivity analysis was conducted to
evaluate the reliability of studies that were not regarded as having a high risk of bias
in any of the domains.19 The Robvis tool (NIHR, Bristol, UK) was used to create risk
of bias assessment figures.20

RESULTS

Study characteristics
The study selection process for this systematic review is presented in a flowchart in
Figure 1. The included studies were prospective NRSI published between 2013 and
2024, except for one that was retrospective. Similarly, Kontadakis et al 21 used a
previous cohort and Gore et al11 used a historical control. The study by Iqbal et al. was
a multi-center study. All studies compared the PRK+CXL procedure with CXL alone,
but Gore et al11 used an accelerated CXL protocol in the experimental and control
groups, and Iqbal et al8 used a CXL Dresden protocol in the control group.
Progressive keratoconus was defined when any of the following criteria were met: a)
an increase of 1.00D or more in the keratometry plus curve; b) an increase of 1.00D or
more in the manifest cylinder; c) an increase of 0.50D or more in refractive spherical
equivalent; d) an increase in cone apex keratometry of 0.75D, >1.00D or 1.50D; e) a
0.75D alteration in refractive spherical equivalent; f) a >5% decrease in central
corneal thickness or a decrease in corneal thickness of >2% or ≥ 10 µm. This
systematic review included 856 eyes of 781 patients. Patient follow-up, expressed in
months, ranged from 6 months to 44. Table 1 shows the characteristics of the 8
included studies.
 PRISMA 2020 flow diagram for new systematic reviews which included searches of databases and registers only

Identification of studies via databases and registers

Records removed before

Records identified from*: screening:
PubMed (n = 183) Duplicate records removed
Web of Science (n = 114) (n = 130)
Cochrane (n = 2) Records marked as ineligible
Embase (n =14) by automation tools (n = 0)
Registers (n = 311) Records removed for other
reasons (n = 0)

Records screened Records excluded**

(n = 183) (n = 164)

Reports sought for retrieval Reports not retrieved

(n = 19) (n = 0)

Reports assessed for eligibility

(n = 19)
Reports excluded:
Not have CXL group (n = 10)

Studies included in review

(n = 9)
Reports of included studies
(n = 9)

Figure 1. Flow chart of eligible papers used in the meta-analysis (PRISMA statement).
PRISMA, preferred reporting items for systematic reviews and meta-analyses.
 EYES INCLUDED CONTROL
EXPERIMENTAL GROUP FOLLOW-UP
STUDY DESIGN (EXPERIMENTAL/CONTROL GROUP
PROCEDURE (MONTHS)
GROUP) PROCEDURE
Prospective non- Topographic PRK (iVIS Suite
Alessio et al randomised 17/17 platform Ligi Tecnologie CXL Dresden 24 months
clinical trial Medicali) + CXL Dresden
Prospective non- Topographic PRK (SCHWIND
CXL
Awwad et al randomised 49/37 Amaris laser 114 ORK-CAM 12 months
accelerated
study software) + CXL accelerated
Prospective non-
Topographic PRK (Wavelight CXL
Dai et al randomised 30/15 44 months
EX500) + CXL Accelerated accelerated
study
Non-randomised
Wavefront-guided transepithelial
case series study CXL
Gore et al 47/47 PRK (Schwind Amaris 750S laser) 24 months
with historical accelerated
+CXL accelerated
control

Topography-guided PRK+CXL
Iselin et al Retrospective 28/28 CXL Dresden 24 months
Dresden

Prospective
Non topography guided PRK
multicentre
Iqbal et al 67/58 (VISX STAR S4 IR Abott CXL Dresden 24 months
comparative
laboratories) + ACX
study
Non-randomised
Topography-guided PRK (Pulzar
Kontadakis et interventional
30/30 Z1; CustomVi, Peth, WA)+CXL CXL Dresden 39 months
al case series study
Dresden
(previous cohort)
Topography-guided customized
Prospective non-
Padmanabhan ablation. Trasnepitelial T-CAT.
randomised 26/40 CXL Dresden 6 months
et al Láser Allegretto Excimer 8400Hz
study
EyeQ, Wavelight, Erlagen+CXL
Topography-guided PRK
Prospective non-
(Allegretto Wave Eye-Q 1010- CXL
Singal et al randomised 122/204 12 months
2400 kHz laser (Alcon) + CXL accelerated
study
accelerated

Table 1. Characteristics of the 8 studies included in the review

General Outcomes
Table 2 shows the mean and standard deviation of each of the parameters included in
the study for both the PRK-CXL group and the CXL group preoperatively and
postoperatively. The difference in means and standard deviation and the existence of
statistically significant differences are also shown. It can be seen that there are
differences in the postoperative parameters of UDVA, CDVA, sphere, cylinder, SE,
KM, Spherical abberraton, coma, total HOA and coma-like values. These differences
only existed preoperatively in the CDVA and cylinder values, although, in these cases,
the lowest values were in the CXL group, and postoperatively the lowest values were
in the PRK-CXL group.
 PREOPERATIVE
PRK-CLX
CLX Group
group p-value
Mean SD Mean SD Mean difference
UDVA 0,88 0,36 0,77 0,39 0,11 0,045
CDVA 0,32 0,19 0,25 0,19 0,07 0,039
SPH -1,39 2,21 -1,69 0,72 0,31 0,157
CYL -3,53 2,09 -2,98 2,11 -0,54 0,031
SE -2,77 2,46 -2,82 2,29 0,05 0,585
CCT 485,78 33,58 470,68 36,17 15,10 0,026
KM 46,65 4,25 46,72 2,80 -0,07 0,754
2814,0
ECC 2803,33 167,00 200,33 -10,67 0,413
0
SPH AB -0,40 0,46 -0,42 0,49 0,02 0,648
COMA 1,27 0,76 1,37 0,90 -0,10 0,223
TOTAL HOA 2,90 1,03 3,23 1,33 -0,34 0,066
COMA-LIKE 2,76 1,15 2,69 0,97 0,06 0,447

POSTOPERATIVE
PRK-CLX
CLX Group
group p-value
Mean SD Mean SD Mean difference
UDVA 0,47 0,26 0,58 0,33 -0,11 0,032
CDVA 0,14 0,13 0,18 0,12 -0,04 0,057
SPH -2,04 1,38 -2,66 1,51 0,62 0,008
CYL -2,18 1,36 -3,05 1,43 0,87 <0,001
SE -2,56 2,08 -3,33 2,34 0,77 0,011
CCT 449,67 35,46 459,13 45,98 -9,47 0,679
KM 45,23 2,92 45,58 2,62 -0,35 0,024
2729,3
ECC 2750,33 174,67 176,33 21,00 0,644
3
SPH AB -0,22 0,41 -0,24 0,42 0,02 0,734
COMA 0,84 0,90 1,49 1,12 -0,66 <0,001
TOTAL HOA 1,96 1,09 2,94 1,37 -0,98 <0,001
COMA-LIKE 1,53 0,93 2,53 0,97 -1,00 <0,001

Table 2. General outocomes preoperative and postoperative.

Visual Outcomes
Figure 2 shows results of UDVA. 8 included studies studied this variable. The results
are significant (SMD -0.32; 95%CI -0.64, -0.01). The results of six of them favour the
experimental group while two of them (Singal et al 7 which has the highest weight)
favour the control group. There was significant heterogeneity (I2=77%, p<0.001).
Figure 2 also shows all 9 included studies studied CDVA. The mean CDVA improved
significantly in the experimental group compared to the control group (SMD -0.31
CI95% -0.61, -0.01). There was significant heterogeneity (I2=76%, p<0.001).

Figure 2. Visual outcomes.

Refractive outcomes
Figure 3 shows Sphere, cylinder and spherical equivalente parameters. For Sphere,
five of the studies looked at this variable. Sphere was non-significantly improved in
the experimental group compared to the control group (SMD -0.35 CI95% -0.84,
0.14). There was moderate heterogeneity (I2=79%, p=0.001).
Figure 3 also shows seven of the included studies studied refractive cylinder. Cylinder
was significantly improved in the control group compared to the experimental group
(SMD -0.64 CI95% -0.79, -0.49). In the study by Iselin et al 9 the results are not
significant and in the study by Padmanabhan et al 22 they favour the experimental
group. There was non-significant heterogeneity (I2=30%, p=0.200).
Figure 3 shows all 9 included studies studied SE. This variable improved non-
significantly in the experimental group compared to the control group (SMD 0.13
CI95% -0.20, 0.46). In four studies, the results favoured the control group. There was
significant heterogeneity (I2=81%, p<0.001).
Figure 3. Refractive Outcomes.
Corneal outcomes

Figure 4 shows corneal outocmes of CCT, KM and ECC. In CCT, six of the studies
studied this variable. CCT was non-significantly improved in the control group with
respect to the experimental group (SMD -0.16 CI95% -0.75, 0.43). There was
significant heterogeneity (I2=92%, p<0.001).
For KM, figure 4 show seven of the studies looked at this variable. KM was non-
significantly improved in the experimental group compared to the control group
(SMD -0.04 CI95% -0.63, 0.55). There was significant heterogeneity (I 2=90%,
p<0.001).
Figure X shows three of the studies studied this variable. ECC was non-significantly
improved in the control group with respect to the experimental group (SMD 0.17
CI95% -0.54, 0.88). There was significant heterogeneity (I2=87%, p<0.001).
Figure 4. Corneal Outcomes.
Aberrometric outcomes
Figure 6 shows RMS spherical aberration, Coma, Total HOA and Coma-Like
parameters. In Spherical Aberration, four of the studies studied this variable. Spherical
aberration was non-significantly improved in the control group with respect to the
experimental group (SMD 0.39 CI95% -0.14, 0.92). There was significant
heterogeneity (I2=79%, p=0.002).
Figure 6 also shows five of the researches studied RMS Coma. Coma was non-
significantly improved in the experimental group compared to the control group
(SMD 0.05 95%CI -1.18, 1.28). There was significant heterogeneity (I 2=97%,
p<0.001).
Figure 6 shows six of the studies looked RMS Total HOA. Total higher order
aberrations were significantly improved in the experimental group compared to the
control group (SMD -1.09 CI95% -1.64, -0.54). There was significant heterogeneity
(I2=83%, p<0.001).
Figure 6 shows two of the studies looked RMS Coma-Like. This variable was
significantly improved in the experimental group compared to the control group
(SMD -1.09 CI95% -1.53,-0.64). There was no significant heterogeneity (I 2=0%,
p=0.78).
Figure 5. Aberrometric Outocmes
Assessment of quality and bias of included studies
The Risk-of-VISualisation (robvis) tool (Figure 6) was used to analyse the quality and
risk of bias of the included studies. These studies are non-randomised interventional
studies except for Iselin et al 9 which is retrospective. In these studies the allocation
process to one group or another was based on baseline characteristics (mainly
minimum pachymetry after a combined procedure of at least 400 microns). In the
study by Gore et al,11 if the selection criteria were met, the patient was allowed to
choose either combined treatment in both eyes or CXL only in the non-study eye (a
historical control was used). In the study by Alessio et al, 23 it was the worse eye that
received the combined treatment. Finally, in the study by Iqbal et al, 8 the
randomisation sequence is not specified.

Overall, the risk of bias of the included studies was high, which mainly depends on
the non-randomisation of the patients included in the studies, the non-concealment of
the randomisation sequence and the absence of blinding so that the surgeons and
investigators, as well as the personnel in charge of taking the relevant measurements,
were aware of the group to which the patient belonged or it is not specified whether
they were aware or not. The absence of randomisation did not lead to variations in
baseline patient characteristics in most studies, but in three studies it did lead to
relevant differences.
Figure 6. Risk of bias.
DISCUSSION

For Hao
REFERENCES
1. Maharana PK, Dubey A, Jhanji V, Sharma N, Das S, Vajpayee RB.
Management of advanced corneal ectasias. Br J Ophthalmol. 2016;100:34-
40.
2. Godefrooij DA, de Wit GA, Uiterwaal CS, Imhof SM, Wisse RP. Age-specific
Incidence and Prevalence of Keratoconus: A Nationwide Registration Study.
Am J Ophthalmol. 2017;175:169-172.
3. Leung VC, Pechlivanoglou P, Chew HF, Hatch W. Corneal Collagen Cross-
Linking in the Management of Keratoconus in Canada: A Cost-Effectiveness
Analysis. Ophthalmology. 2017;124:1108-1119.
4. D'Oria F, Palazon A, Alio JL. Corneal collagen cross-linking epithelium-on vs.
epithelium-off: a systematic review and meta-analysis. Eye Vis (Lond).
2021;8:34.
5. Wollensak G, Sporl E, Seiler T. [Treatment of keratoconus by collagen cross
linking]. Der Ophthalmologe : Zeitschrift der Deutschen
Ophthalmologischen Gesellschaft. 2003;100:44-49.
6. D'Oria F, Bagaglia SA, Alio Del Barrio JL, Alessio G, Alio JL, Mazzotta C.
Refractive surgical correction and treatment of keratoconus. Surv
Ophthalmol. 2024;69:122-139.
7. Singal N, Ong Tone S, Stein R, et al. Comparison of accelerated CXL alone,
accelerated CXL-ICRS, and accelerated CXL-TG-PRK in progressive
keratoconus and other corneal ectasias. J Cataract Refract Surg.
2020;46:276-286.
8. Iqbal M, Elmassry A, Tawfik A, et al. Standard cross-linking versus
photorefractive keratectomy combined with accelerated cross-linking for
keratoconus management: a comparative study. Acta Ophthalmol.
2019;97:e623-e631.
9. Iselin KC, Baenninger PB, Bachmann LM, Bochmann F, Thiel MA, Kaufmann
C. Changes in higher order aberrations after central corneal regularization -
a comparative two-year analysis of a semi-automated topography-guided
photorefractive keratectomy combined with corneal cross-linking. Eye Vis
(Lond). 2020;7:10.
10. Richoz O, Arba Mosquera S, Kling S, et al. Determination of the excimer
laser ablation rate in previously cross-linked corneas. Journal of refractive
surgery (Thorofare, N.J. : 1995). 2014;30:628-632.
11. Gore DM, Leucci MT, Anand V, Fernandez-Vega Cueto L, Arba Mosquera S,
Allan BD. Combined wavefront-guided transepithelial photorefractive
keratectomy and corneal crosslinking for visual rehabilitation in moderate
keratoconus. J Cataract Refract Surg. 2018;44:571-580.
12. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting
items for systematic reviews and meta-analyses: the PRISMA statement.
PLoS medicine. 2009;6:e1000097.
13. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting
 systematic reviews and meta-analyses of studies that evaluate health care
interventions: explanation and elaboration. PLoS medicine.
2009;6:e1000100.
14. Collaboration C. Cochrane Handbook for Systematic Reviews of
Interventions. 5.1.0 ed. ed: The Cochrane Collaboration; 2011.
15. Organization WH. General guidelines for methodologies on research and
evaluation of traditional medicine. Geneva: World Health Organization;
2000.
16. Higgins JP, Altman DG, Gotzsche PC, et al. The Cochrane Collaboration's
tool for assessing risk of bias in randomised trials. BMJ. 2011;343:d5928.
17. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis.
Stat Med. 2002;21:1539-1558.
18. Barili F, Parolari A, Kappetein PA, Freemantle N. Statistical Primer:
heterogeneity, random- or fixed-effects model analyses? Interact
Cardiovasc Thorac Surg. 2018;27:317-321.
19. Thabane L, Mbuagbaw L, Zhang S, et al. A tutorial on sensitivity analyses
in clinical trials: the what, why, when and how. BMC Med Res Methodol.
2013;13:92.
20. McGuinness LA, Higgins JPT. Risk-of-bias VISualization (robvis): An R
package and Shiny web app for visualizing risk-of-bias assessments. Res
Synth Methods. 2021;12:55-61.
21. Kontadakis GA, Kankariya VP, Tsoulnaras K, Pallikaris AI, Plaka A, Kymionis
GD. Long-Term Comparison of Simultaneous Topography-Guided
Photorefractive Keratectomy Followed by Corneal Cross-linking versus
Corneal Cross-linking Alone. Ophthalmology. 2016;123:974-983.
22. Padmanabhan P, Radhakrishnan A, Venkataraman AP, Gupta N, Srinivasan
B. Corneal changes following collagen cross linking and simultaneous
topography guided photoablation with collagen cross linking for
keratoconus. Indian J Ophthalmol. 2014;62:229-235.
23. Alessio G, L'Abbate M, Sborgia C, La Tegola MG. Photorefractive
keratectomy followed by cross-linking versus cross-linking alone for
management of progressive keratoconus: two-year follow-up. Am J
Ophthalmol. 2013;155:54-65 e51.
Were there deviations from the intended intervention beyond what would be
expected in usual practice?
Deviations that happen in usual practice following the intervention (for
example, cessation of a drug intervention because of acute toxicity) are part
of the intended intervention and therefore do not lead to bias in the effect of
assignment to intervention.
Deviations may arise due to expectations of a difference between intervention
and comparator (for example because participants feel unlucky to have been
assigned to the comparator group and therefore seek the active intervention,
or components of it, or other interventions). Such deviations are not part of
usual practice, so may lead to biased effect estimates. However these are not
expected in observational studies of individuals in routine care.