Table 1.
Risk of bias table for assessing the methodological quality of studies using laser scanning in vivo confocal microscopy (IVCM) for evaluating human corneal sub-basal nerve plexus parameters
| Item | Risk of bias | ||
| Low | Unclear | High | |
| Participant selection | |||
| Participant inclusion and exclusion criteria (selection bias) | Eligibility criteria are stated, with clear specifications to define participant status (eg, HbA1c level for participants with diabetes) | Eligibility criteria are stated without clear specifications to define participant status (eg, diabetes stated, without further details verifying the diagnosis) | No eligibility criteria are stated. Participants may have been included with conditions that could confound the measurement of nerve parameters |
| Method of image capture | |||
| Masking of confocal operator (performance bias) | Clearly states that the confocal operator was masked to participant/group allocation OR that all personnel were masked throughout the study OR not applicable (if a single study population was examined) | A general statement regarding masking (eg, ‘double-masked’ study) is included, without further information about whether this applies to the corneal image capture (if multiple outcomes were measured) and where this factor is considered relevant (eg, a RCT) | No information is provided in relation to the masking of personnel, where masking is considered relevant (eg, intervention study); we assume that in the absence of reporting, the operator was not masked |
| Specification of participant fixation target | States that a consistent fixation target was used for all participants | States that a fixation target was used, but no other relevant information is provided | Use of a fixation target is not mentioned (bias potentially introduced with eye movements) |
| Location of cornea imaged | Quantitative description of location of cornea imaged (eg, within a 2 mm radius of the corneal apex) | Qualitative description of the location of cornea imaged (eg, central or peripheral) | Location of corneal imaging is not specified |
| Specification of corneal depth | Quantitative description of the corneal depth imaged is included (eg, imaged at a depth range of 10–15 µm below the basal epithelium or a method is used to project nerves imaged different depths onto a single plane28) | Qualitative description of the corneal depth imaged is included (eg, imaged just below basal epithelial cell layer) | Depth of cornea imaged is not stated |
| Illumination setting on confocal microscope | States that images were acquired using fixed illumination intensity for all participants29 | Confocal illumination settings were not reported | States that confocal images were acquired using ‘automated brightness’ settings that were optimised for each participant |
| Selection of eye (left or right or both) | Clearly specifies which eye was assessed, with a sound method of selection (eg, random, all right eyes, average of right and left eyes) | Specifies which eye was assessed without sound justification for the method of selection (eg, potential inconsistent use of data from one or both eyes) | No mention of which eye was assessed, or data from both eyes were used without appropriate statistical adjustment |
| Method of image analysis | |||
| Masking of outcome assessor (performance bias) | Clearly states that the assessor of the corneal images was masked to the participant/group allocation, or this bias domain is not applicable (if a single study population was studied) | A general statement regarding masking is included (eg, ‘double-masked’), without specifying whether this applies to the corneal image analysis (if multiple outcomes were measured) and where this factor is considered relevant (eg, a RCT) | No information is provided in relation to masking of the outcome assessor, where masking is considered relevant (eg, intervention study); we assume that in the absence of reporting, the assessor was not masked |
| Image selection—quality (sampling bias) | Clearly states that image quality was assessed AND that images where the imaging depth was inconsistent or the image was blurred were removed from the analysis sample AND representative images are provided within the manuscript that confirm the images meet these criteria | States that image quality was assessed, but does not specify the criteria for excluding images from the analysis OR there are no representative images provided within the manuscript | No information is provided in relation to the assessment of image quality OR there are representative images within the manuscript that appear blurred or not within a fixed plane of focus |
| Image selection—number and sampling (sampling bias) | Number of images analysed is clearly stated, and at least eight images were analysed per region with <20% overlap between images10 | Number of images analysed is clearly stated but either does not state level of overlap of images was <20% or five to seven images were analysed10 | Number of images analysed is not stated or <5 images were analysed |
| Image selection—method of randomisation and sequence generation (sampling bias) | Images described to be randomly selected for analysis and method for random selection is reported (eg, computer generated list, random table, other method of generating random list) | Images described to be randomly selected for analysis but with no further details as to how the randomisation schedule was generated | Image selection is not reported or image selection method is based on subjective judgement (eg, ‘most representative’ or ‘best image selected’) |
| Image processing—order of analysis | Method involved an automated processing method (eg, ACCMetrics,11 or similar) or not applicable (if only a single time point was measured per participant, e.g., cross-sectional study) | Order of image analysis was not reported, when a manual method was used for quantification and this is considered relevant to the outcome(s) | Images were analysed consecutively (per participant) using a manual method (eg, subjective quantification) and this is considered relevant to the outcome(s) |
| Postcapture image enhancements | Clearly states that there were no postcapture image enhancements (ie, brightness, sharpening, and so on) performed, OR a uniform enhancement was applied across all images | No statement regarding whether postcapture image enhancements were performed | Postcapture image enhancements were performed to optimise image contrast/brightness (or some other parameter) individually for each image |
| Definition of sub-basal nerve parameters | The nerve parameters being evaluated (eg, CNFL, CNFD, CNBD) are clearly defined, either in the paper itself or by referencing a previously validated method | The nerve parameters being evaluated are stated but are not sufficiently well defined to allow full reproduction of the method | The nerve parameters being evaluated are stated but without any definition (eg, ‘CNFL was measured’, with no information given on how the parameter was defined) |
| Method and repeatability of sub-basal nerve parameter quantification (intra- observer and interobserver variability) | Use of a validated, fully automated processing method (eg, ACCMetrics,11 or similar30 31) | Use of a semi-automated or fully manual method, where repeatability testing was performed and reported | Use of a semi-automated or fully manual method, where no repeatability testing was reported |
| Data reporting | |||
| Thoroughness of reporting of nerve parameters—selective reporting of outcomes | Data relating to all quantified nerve parameters (as mentioned in the study methods) are reported in the results with both point measures and measures of variability | Data relating to all quantified nerve parameters (as mentioned in the study methods) are reported in the results with a qualitative descriptor only (with or without reporting of p values) | Selective reporting of nerve parameters (ie, mismatch between parameters described in results and those mentioned in methods) |
| Completeness of nerve parameter data (population level)—attrition bias | Missing data for <20% of recruited participants, and if multiple study groups or a longitudinal study, then there is an equal degree of missing data in both groups at follow-up, with no obvious reason why absence of data is related to study group or time point respectively | Completeness of data is not reported, or if multiple study groups or a longitudinal study, the degree of missing data is for >20% of participants but occurs to an equal degree in both study groups | If multiple study groups, there is an unequal degree of missing data between study groups, and/or the absence of the data appears to be related to the study outcome |
| Other | |||
| Other sources of bias | No other apparent sources of bias | Source of funding not reported | Industry-funded study |
CNBD, corneal nerve branch density; CNFD, corneal nerve fibre density; CNFL, corneal nerve fibre length; HbA1c, glycated haemoglobin; RCT, randomised controlled trial.