PRECOG: Developing a practical, evidence-based approach to assessing cataract surgical outcomes

1
PRECOG Developing a practical, evidence-based
approach to assessing cataract surgical outcomes

• Nathan Congdon, MD, MPH

Zhongshan Ophthalmic Center, Preventive
Ophthalmology Unit, Guangzhou, China ORBIS
International
2
Financial interest

• No financial interest

3
The problem of un-operated cataract

• The key to solving this problem, still the
  worlds leading cause of blindness, is training
  additional surgeons
• The critical issue is outcome quality, for which
  the WHO has set standards
• Presenting acuity gt 6/18 in 80 of
  post-operative patients

4
Barriers to assessment of cataract outcomes

• The proportion of patients returning after
  surgery is often very small in many parts of the
  developing world.
• It is un-known whether vision outcomes among
  patients who do present for follow-up
  spontaneously are representative of all persons
  undergoing operations.

5
A new approach to outcomes assessment?

• Wide adoption of small-incision, sutureless
  surgery mean more rapid recovery of vision
  post-operatively
• Many surgical facilities, especially in rural
  areas, admit patients for 1-3 days after surgery
• Can the principal assessment of post-operative
  vision be carried out at time of hospital
  discharge?

6
Advantages of early outcomes assessment

• Collect data on all patients readily
• Avoid bias in data collection
• Reduce costs to patients and hospitals for
  follow-up

7
PRECOG Prospective Review of Early Cataract
Outcomes and Grading

• Objectives
• Early assessment
• Assess validity of visual acuity measured at
  hospital discharge after cataract surgery as a
  predictors of medium-term (gt 50 days) vision
  (Study hypothesis)
• Better use of existing data
• Assess extent to which vision of persons
  spontaneously returning for follow-up care gt 50
  days after cataract surgery are predictive of VA
  for entire operated cohort (Traditional
  approach)

8
PRECOG Setting

• Urban and rural facilities providing cataract
  surgery (n 41)
• East Asia
• China (18)
• Vietnam (4)
• Indonesia (2)
• India
• All Aravind centers (5)
• Latin America
• Peru (2), Ecuador (1), Paraguay (1), Guatemala
  (1), Mexico (2)
• Africa
• Eritrea (2)
• Ethiopia (3)

9
PRECOG Participants and Sample Size

• 50-100 consecutive persons aged gt 30 years and
  under-going surgery for age-related cataract at
  each participating facility
• Exclusion criteria
• Traumatic cataract
• Ocular co-morbidities including glaucoma, retinal
  disease, corneal abnormalities or uveitis.

10
PRECOG Follow-up

• Target of gt 90 follow-up at gt 50 days post op,
  either through
• Spontaneous return to clinic
• Return to clinic potentiated by special
  intervention (phone call, offer of free
  transport, etc.)
• Home visit
• Type of follow-up recorded, so that patients
  returning spontaneously, under usual conditions
  (WITHOUT phone call, home visit etc.) can be
  studied

11
PRECOG Results Participants

• Hospitals (n 41)
• Annual surgical output Range from lt 500
  (several) to 91,759 (Aravind Madurai)
• Public 31/41 (75.6)
• Rural 24/44 (58.5)
• Cases
• A total of 3547, of which
• 2246 (63) SICS
• 776 (22) phaco
• Remainder ECCE (15)

12
PRECOG Results Surgery

• Pre-op VA lt 6/60 in operated eye 84.6
• Final (gt 50 days) uncorrected VA
• gt 6/18 2089 (63.7)
• lt 6/60 338 (10.3)
• Complications
• Intra-op 7.79
• Post-op 1.99

13
PRECOG Results Follow-up

• The proportion of subjects with follow-up vision
  measured at gt 50 days after surgery was
  3178/3547 (92.5)
• By region, follow-up was
• China 89.8
• India 93.6
• Vietnam/Indonesia 90.1
• Latin America 98.3
• Africa 95.6
• Spontaneous follow-up at clinic 43 (Range from
  China 26 to Latin America 80)

14
Correlation of early vision with final vision

• What we want to know How do hospitals rank
  according to final VA outcome? (proportion with
  VA gt 6/18)
• We can compare two strategies to estimate this
• Using discharge vision to rank hospitals (the
  goal of PRECOG)
• Using the final vision among those patients who
  do return spontaneously (what we have
  traditionally done)

15
The method we are testing in PRECOG
The method we have traditionally used
Discharge VA for all patients
Final VA for 40 of patients who DO return
spontaneously to clinic
Final VA for ALL patients
What we are trying to estimate
16
Correlation of early vision with final vision

• Discharge vision and final vision are highly
  correlated for all patients Spearman r 0.59
• Hospital rankings using uncorrected discharge
  vision appear better-correlated with rankings
  using final vision than are rankings using the
  43 of patients who return spontaneously
• Spearman r 0.50 for discharge vision
• Spearman r 0.28 for patients who return
  spontaneously

17
Can we do even better?

• Using best-corrected vision does not improve the
  performance of discharge VA in predicting
  hospital rankings based on final VA (r 0.45)
• Dropping patients (15) with ECCE has little
  impact on performance of discharge VA (r 0.56)

18
Can we do even better?

• When we measure discharge vision as an index of
  outcome, there are inevitably some patients with
  temporary poor VA due to corneal edema or other
  problems
• What if we could improve performance of poor
  vision by dropping these patients?
• When we drop the 20 of patients at each hospital
  with the worst vision, discharge vision is
  better-correlated with final VA r 0.67

19
Concrete example using PRECOG data

• As a program planner in MOH or NGO, you want to
  separate hospitals into three categories
• Good (Top 25 Can provide training to others)
• Medium (Middle 50 No intervention needed)
• Problem (Bottom 25 Further training needed)
• How well does early vision assessment work for
  this?

Omitting data from 3 hospitals in Ethiopia for
whom data not yet cleaned
20
Concrete example Uncorrected VA, drop worst 20
by vision

• 26/38 hospitals (68) have the same ranking using
  discharge VA that they would have had using
  final VA
• No hospitals went from Good to Poor or Poor to
  Good

21
Concrete example

• 68 (26/38) of hospitals had the identical
  ranking based on discharge and final vision
• If the vision of patients returning spontaneously
  was used to rank hospitals, only 18/38 (47) had
  the same ranking
• Based on chance alone, two such ranking systems
  would be expected to agree on 13/38 (34) of
  hospitals

22
PRECOG Results Standards for Early Vision
Assessment

• If discharge vision will be used as an index for
  surgical quality, the current WHO standard of 80
  of patients with uncorrected VA gt 6/18 will
  likely need to change
• In PRECOG, hospitals achieved the following
  standards for the of patients with uncorrected
  discharge VA gt 6/18
• 90th percentile 71.8
• 75th percentile 60.6
• 50th percentile 45.3
• 25th percentile 31.1

23
CAVEATS

• Though hospitals in PRECOG included rural and
  urban, government and private facilities from
  many regions
• They were not chosen at random
• We dont know if they are truly representative of
  all facilities
• Patients were chosen at random (consecutive
  surgeries), and follow-up was very good, but not
  100
• Room for bias

24
PRECOG Summary

• If hospitals can measure discharge vision on
  50-100 consecutive patients, they can provide a
  robust index of cataract surgical outcome usable
  by themselves and program planners
• No need to be able to refract (using BCVA does
  not improve accuracy of data)
• Works for hospitals performing ECCE as well as
  small incision cases
• Even small, rural hospitals throughout the world
  have now proven their ability to collect these
  data

25
PRECOG Next steps

• Analyze other data we collected to further guide
  optimal follow-up
• Prevalence of refractive error and other
  conditions requiring treatment how useful is
  follow-up?
• Cost to patients and hospitals how
  cost-effective is follow-up?
• Look at simple adjustments to improve accuracy of
  discharge vision even further
• Work with WHO, IAPB, NGOs and governments to
  disseminate and begin using these results to
  evaluate surgical quality in practice