Phase 6: Analysis and reporting

Analysis Recommendation
After data are cleaned, analysis is recommended to be conducted by either a biostatistician or a researcher with statistical analysis experience.

What data should be excluded?

If more than 3 baseline refractions were conducted on a USP, exclude the refraction(s) that are the most different to the others.

If USPs were identified by certain optical services, the number and percentage of visits should be reported, however their data should be removed from all other analysis.

Q.REC indicators

The Q.REC indicator(s) are used to compare the dispensed spectacles from optical services to the USP’s baseline prescription.

Q.REC indicator 1: Optimally prescribed spectacles

Q.REC Indicator 1 is the key primary outcome for this study. It has been demonstrated that when glasses pass this indicator that USPs have significantly better vision and comfort compared to glasses that do not pass this indicator. This shows us the quality of refractive error care and information to plan and monitor optical services.

Table 1 Q.REC Indicator 1 criteria

Spectacle component

Tolerance limits compared to baseline prescription

Spherical power

± 0.50 dioptre

Cylindrical power

± 0.50 dioptre

Cylindrical axis (if baseline |cylindrical power| ≤ 0.50 DC)

± 7 degrees

Cylindrical axis (if baseline |cylindrical power| > 0.50 DC – ≤ 1.50 DC)

± 5 degrees

Cylindrical axis (if baseline |cylindrical power| > 1.50 DC)

± 2 degrees

Horizontal prism (total)

< 1 prism dioptre (in/out direction)

Vertical prism (total)

< 0.50 prism dioptre (up/down direction)

Q.REC indicator 2: Adequately prescribed spectacles

For settings where sphero-cylindrical lenses are rarely prescribed or available, Q.REC indicator 2 can be used as the primary outcome that shows the quality of refractive error care. Similarly, it can help with planning and monitoring optical services, however is not as accurate as Q.REC indicator 1. For projects that use Q.REC indicator 1 as the primary outcome, Q.REC indicator 2 can be a secondary outcome or not used at all.

Table 2 Q.REC Indicator 2 criteria

Spectacle component

Tolerance limits compared to the baseline prescription

Spherical equivalent power

± 0.50 dioptre

Horizontal prism (total)

< 1 prism dioptre (in/out direction)

Vertical prism (total)

< 0.50 prism dioptre (up/down direction)

Data transformation

Baseline refractions

For each eye, all components of the refraction should be averaged, this includes the distance spherical power, near spherical power (particularly if the USP has a near addition with their prescription) cylindrical power, cylindrical axes, distance and near pupillary distances, and distance and near visual acuities.

Categorising into refractive error types

Each USP can be grouped into at least one refractive error type based on the baseline refraction. The following definitions for each type are recommended:

  • Myopia: spherical equivalent < -0.50 DS in at least one eye
  • Hyperopia: spherical equivalent > +0.50 DS in at least one eye
  • Astigmatism: > 0.50 DC in at least one eye
  • Emmetropia: spherical equivalent ≥ -0.50 DS and ≤ +0.50 DS in both eyes
  • Presbyopia (objective definition): ≥ 1.00 D added to the best optical distance correction

Note: To calculate spherical equivalent, the following formula can be used: Spherical equivalent power = Spherical power + (Cylindrical power)/2

Visual acuities

Distance and near visual acuity for the right, left and both eyes should be converted into logMAR decimal format. As each USP is likely to have unique near working/viewing distances, near visual acuity should be recalculated to account for their working distance.

Categorical variables of ‘good vision’ can be created for the right and left eyes separately and with both eyes open. USP while wearing dispensed spectacles is considered to achieve ‘good vision’ if the corrected visual acuity is <1.5 lines worse than baseline best-corrected visual acuity at distance or near.

Suitability of spectacle lens types

A USP is considered suitable for a spectacle lens type (either single vision distance / single vision near / bifocal / multifocal) if at least one of the study optometrists ticks that they are suitable for that lens type.

Q.REC Indicators

Q.REC Indicator 1: Optimally prescribed spectacles

To compare the dispensed spectacles to the baseline refraction on whether the spectacles pass Q.REC Indicator 1, data are required to be transformed and calculated.

Component / Indicator

Analysis for Q.REC Indicator 1: Optimally prescribed spectacles

Spherical power

For each eye:

  • Calculate the absolute difference of spherical power (in dioptres) between the dispensed spectacles and the baseline spherical power of most plus axis).
  • Is the absolute difference ≤ 0.50 D ?
  • If the answer is “Yes” for both eyes, then the spectacles pass the spherical power component

Cylindrical power

For each eye:

  • Calculate the absolute difference of cylindrical power (in dioptres) between the dispensed spectacles and the baseline cylindrical power.
  • Is the absolute difference ≤ 0.50 D ?
  • If the answer is “Yes” for both eyes, then the spectacles pass the cylindrical power component

Cylindrical axis

For each eye:

  • Comparison can only be performed with USPs that have astigmatism in their baseline prescription. For each eye:
  • Calculate the absolute difference of cylindrical axis (in degrees) between the dispensed spectacles and baseline cylindrical axis of the most minus meridian.
    *Note:* It is important to consider the range of tolerance particularly if baseline cylindrical axis is around the horizontal meridian. E.g. If baseline refraction were -1.00/-0.50×180, the axes that would pass this component criteria are 1 – 7 degrees, *and* 173 – 180 degrees
  • If the baseline cylindrical power was ≤ -0.50 DC, was the absolute difference of cylindrical axis within 7 degrees of the baseline axis?
  • If the baseline cylindrical power was > -0.50 DC – ≤ -1.50 DC, was the absolute difference of cylindrical axis within 5 degrees of the baseline axis?
  • If the baseline cylindrical power was > -1.50 DC, was the absolute difference of cylindrical axis within 2 degrees of the baseline axis?
  • If the answer is “Yes” in both eyes (assuming both eyes have some astigmatism), then the spectacles pass the cylindrical axis component

Horizontal prism

For each eye:

  • Calculate the monocular decentration amount (cm) by subtracting the lens centration distance (mm) from baseline pupil distance (mm) and dividing by 20 (takes into account halving for each eye and converting into centimetres)
  • Calculate the spectacle lens power at the horizontal meridian (180 degrees). If baseline axis is 180, then it should be the spherical power of the baseline prescription. If the baseline axis is 90, then it should be the spherical + cylindrical power of the baseline prescription. If it is an oblique axis, additional calculation
  • Calculate the magnitude of horizontal prism using Prentice’s rule: p = cxF

    prism magnitude = monocular decentration (cm) × spectacle lens power at horizontal meridian

  • Determine the direction of the prism, is it base-in or base-out in each eye?
  • Calculate the total amount of induced prism. If both eyes have prism in the same direction (i.e both base-in or both base-out, then add the magnitude together. If the prism direction is different between the eyes (i.e. base-in in the right eye and base-out in the left eye, or vice versa), then subtract one magnitude from the other.
  • Is the total amount of induced horizontal prism less than 1 prism dioptre (Δ)?
  • If the answer is “Yes”, then the spectacles pass the horizontal prism component

Vertical prism
  • If you are using a digital focimeter, this can be automatically determined for you, so long as the glasses remain on the same plane as you move between the right and left lenses.
  • If you are using a manual focimeter, first centre and clamp the right lens so that the optical centre is in the middle of the graticule and adjust the lens table to the bottom of the spectacle frame so it is resting on the table. Unclamp the lens and move the frame over to measure the left lens. Be sure not to change the height of the frame table. Looking at the target through the eyepiece, If the target of the left lens looks higher or lower than the graticule centre, then there is vertical prism. You use the markings within the graticule to determine the amount of vertical prism – depending on your machine, some markings indicate 0.5Δ steps, others indicate 1.0Δ steps.

To determine the direction of vertical prism:

  • If the target for the left lens is higher than the right lens, there is base-up prism in the left lens (compared to the right lens)
  • If the target for the left lens is lower than the right lens, there is base-down prism in the left lens (compared to the right lens)
  • Is the total amount of induced horizontal prism less than 0.5Δ?
  • If the answer is “Yes”, then the spectacles pass the vertical prism component

QREC Indicator 1

Spectacles are optimal if:

  • Spherical power component = PASS *AND*
  • Cylindrical power component = PASS *AND*
  • Cylindrical axis component = PASS *OR* Both eyes did not require cylinder powers at baseline (i.e. no astigmatism in either eye) *AND*
  • Horizontal prism component = PASS *AND*
  • Vertical prism component = PASS

To compare the written prescription to the baseline refraction on whether the spectacles pass Q.REC Indicator 1, at a minimum, the steps of spherical power, cylindrical power and cylindrical axis components in the above table are applied. The horizontal prism component can be considered for inclusion if pupil distance(s) are provided in the written prescription.

Q.REC Indicator 2: Adequately prescribed spectacles

To compare the dispensed spectacles to the baseline refraction on whether the spectacles pass Q.REC Indicator 2, data will require transformation as shown in the table below.

Component / Indicator

Steps

Spherical equivalent power

For each eye:

  • Calculate the spherical equivalent power (in dioptres) from the baseline prescription
  • Calculate the spherical equivalent power (in dioptres) from the dispensed spectacles
  • Calculate the absolute difference of spherical equivalent power (in dioptres) between the dispensed spectacles and the baseline.
  • Is the absolute difference ≤ 0.50D ?
  • If the answer is “Yes” for both eyes, then the spectacles pass the spherical equivalent power component

Horizontal prism

Refer to table: Analysis for Q.REC Indicator 1: Optimally prescribed spectacles

Vertical prism

Refer to table: Analysis for Q.REC Indicator 1: Optimally prescribed spectacles

QREC Indicator 2

Spectacles are adequate if:

  • Spherical equivalent power component = PASS *AND*
  • Horizontal prism component = PASS *AND*
  • Vertical prism component = PASS

To compare the written prescription to the baseline refraction on whether the spectacles pass Q.REC Indicator 2, the steps of spherical equivalent power component in the above table are applied. The horizontal prism component can be considered for inclusion if pupil distance(s) are provided in the written prescription.

Analysing and presenting the data

The following section is a recommendation of data that can be analysed and presented. This initial analysis might produce emerging outcomes that are of interest to your context and warrant further analysis.

Q.REC Flowchart

A flowchart of the optical services sampling frame, the ones excluded, recruited, USPs recruited, services visited and the number of spectacles dispensed are the minimum details to be included.

USP characteristics

Describing USP characteristics can be presented in a table. Examples of descriptors include:

  • Age – mean and standard deviation or categorised into groups
  • Gender
  • Refractive error type
  • Presence of presbyopia
  • Pupil distances – mean and standard deviation
  • Types of spectacle lenses the USPs are suitable for

The table can be presented as the total group or disaggregated. If USPs are disaggregated, Fisher’s exact tests or Chi-square tests for categorical variables or ANOVA for continuous variables can be used to test USP differences by their disaggregated groups.

Optical service characteristics

Describing optical service characteristics, such as who provided care and testing procedures used, can be presented in a table. Examples of descriptors include:

  • Gender of clinician who conducted refraction
  • Whether certification of refraction or health department registration observed
  • Focimetry conducted
  • Components of distance refraction
  • Components of near refraction
  • Pupil measurement conducted
  • Clarity of communication

Primary analysis of spectacle quality – Q.REC Indicator(s)

The main outcome of the Q.REC study is the proportion of spectacles that are optimal and/or adequate in quality (i.e. pass Q.REC Indicator 1 and/or Q.REC Indicator 2) including a 95% confidence interval. This can be disaggregated by location, or spectacle type (if there is a hypothesis that there might be a quality difference in the way distance and near spectacles are dispensed). Sampling weights should also be considered if any particular sub-groups are much larger than others, and the number of spectacles dispensed for each sub-group differ. The confidence intervals should be estimated by using robust standard errors to account for intra-service correlation.

Q.REC Indicator 1 or 2 can also be presented on the proportion of glasses that pass each indicator component, and the patterns of whether each component are met or not.

The written prescription outcome would be the proportion of written prescriptions that are optimal or adequate in quality. The proportion and 95% confidence interval can be disaggregated, and also compared to the main spectacle outcomes above.  The confidence intervals should be estimated by using robust standard errors to account for intra-service correlation.

Association between spectacle quality and USP and/or optical service characteristics

The association between spectacle quality and USP and/or optical service characteristics can be assessed via univariable logistic regression. Multivariable regression models can also be considered. Robust standard errors should also be used within all logistic regression models to account for intra-service correlation. Results can be presented with odds ratios, 95% confidence intervals and p-values.

Association between written prescriptions and optical service characteristics

Similarly, the association between written prescriptions and USP and/or optical service characteristics can be assessed via univariable logistic regression. Multivariable regression models can also be considered. Robust standard errors should also be used within all logistic regression models to account for intra-service correlation. Results can be presented with odds ratios, 95% confidence intervals and p-values.

Association between dispensed spectacles and written prescriptions

Data enables you to assess whether the written prescriptions provided to the USP match the dispensed spectacles.

Unnecessary prescribing with emmetropes

If emmetropes are recruited to be USPs in the Q.REC study, the proportion of visits where emmetropes were recommended spectacles can be described. The association between emmetropes being recommended spectacles and optical service characteristics can also be assessed with logistic regression. Results can be presented with odds ratios, 95% confidence intervals and p-values.

Relevant documents


📄
Q.REC Indicators
PDF