Comparison of postoperative complication rates under two anthropometric eligibility criteria: WFA versus WFH

Objective

To compare the frequency of postoperative complications in children with cleft lip and/or palate operated on after selection based on weight-for-age (WFA) with that observed after selection based on weight-for-length or height (WFH), at Kibuye Hope Hospital (KHH).

Key message. The switch to weight-for-height considerably widened access to surgery, was accompanied by a decline in surgical delay, and did not increase the risk related to nutritional status. The apparent rise in complications is explained mainly by detection bias: families in the second group could be reached two and a half times more often.

Design, setting and periods

Analytical, retrospective, comparative study of two historical groups, based on the operating theater records of KHH. The change in eligibility criterion took place in April 2023.

Sampling

Source population: 638 cleft operations recorded in the operating theater between April 2017 and June 2026 (294 before and 344 after the change). Random selection of 150 operations per period, i.e. 300 operations performed on 282 children (17 children operated on more than once). The unit of analysis is the operation.

The postoperative course was verified in the records and then by a standardized telephone call to the parents. A record whose family could not be reached and that had no documented complication was replaced by a new random draw; a record reporting a complication was always retained. Two patients who died (one per group) were excluded.

Operational definitions

• Z-scores: WHO 2006 standards, LMS method with WHO correction beyond ±3; weight, height and age from the same day (preoperative weighing); WFA computable from 0 to 60 months; WFH = length before 24 months, height thereafter; physically impossible heights (< 45 cm or > 120 cm, z outside ±5) excluded.
• Undernutrition: z < −2 (severe if z < −3).
• Surgical delay: cheiloplasty after 6 months, palatoplasty after 18 months; secondary surgeries excluded.
• Major complication: dehiscence, fistula, necrosis, bleeding, respiratory complication. Minor: residual asymmetry, keloid, otitis.
• Family reached: parent or guardian who answered the call.

Statistical analysis & ethics

Categorical variables: chi-square and Fisher's exact test. Skewed quantitative variables: median [IQR] and Mann-Whitney. Associated factors: crude ORs [95% CI] then multivariate logistic regression (2 models, adjusted for reachability). Three sensitivity analyses. Threshold α = 0.05. Analyses performed in Python 3.11 (pandas, SciPy, statsmodels), checked in Excel.

Protocol submitted to the dean's office and to the HAU ethics committee; access to records authorized by the KHH management. Anonymized data (study number, no names, operators designated by a letter); oral consent at the start of each call; clinical photographs with eyes masked (appendices D and F).

300 operations, two comparable groups

Two indices, two different populations

By WFA, the proportion of undernourished children rose from 12.8% (WFA group) to 41.6% (WFH group) — p < 0.001. By WFH, it remained low and stable (3.6% vs 4.1%; p = 1.000). Children operated on under the new protocol are more often light for their age but proportionate: they are small, not thin. Burundian context: chronic malnutrition in 55.9% of children aged 6–59 months (AGVSAN 2023) versus ~6% acute malnutrition — WFA mechanically excludes many children there.

The change of criterion opened the operating theater

In the WFH group, 61 of 145 operations (42.1%) involved a child who would have been turned away or deferred under the old criterion — 57 of them with a normal weight-for-height. In the WFA group: only 7 of 112 (6.2%). This gain was accompanied by a decline in surgical delay:

Nutritional rehabilitation is, moreover, feasible and rapid — 63.7% of children eligible within six weeks for about 25 dollars according to program data — but it requires the family to come back: every child wrongly classified as undernourished means a journey, a delay and a risk of dropping out.

An apparent rise, five fistulas in 300

The rate rises in 2023, the year of the change (15.2% → 27.8%), then falls in 2025–2026 — consistent with a learning curve for the new team and a detection effect.

The key to reading the whole study

Families in the WFH group were reached 2.5 times more often (98/150 versus 38/150; p < 0.001) — older phone numbers were switched off, nonexistent or reassigned. Yet some complications were discovered by telephone only.

An important caveat: among families reached, major complications remain more frequent in the WFH group (21.4% vs 5.3%) — mostly dehiscences and fistulas, related to a more bilateral case mix and the arrival of a new national surgical team (77% of WFH operations performed by surgeons D and E). For surgeon A, the only one present in both periods, the difference is not significant (9.8% → 14.7%; p = 0.532). Undernutrition was not associated with complications either by WFA (23.5% vs 15.3%; p = 0.122) or by WFH — the 62 children in the WFH group who would have been excluded by the old criterion did not have more complications (p = 0.433).

Univariate then multivariate: reachability dominates

Multivariate logistic regression (model 2, adjusted for reachability)

Log scale: each bar is proportional to log(OR+1). The WFH criterion, significant in model 1 (OR 2.11; p = 0.046), is no longer significant in model 2: having reached the family captures the entire signal. Models based on 279 operations and 51 complications.

The crude difference withstands two of three tests

What the network takes from it

The data are consistent with the hypothesis that weight-for-length or height improves access to surgery without increasing the risk of complications related to nutritional status. WFA retains its full value as a public health indicator; it is its use as a surgical eligibility criterion that is called into question.

Read honestly

• Detection bias — complications sought by telephone, WFH families reached 2.5× more often; measured and corrected by adjustment, but it limits the comparison of crude rates.
• Retention rule — unreachable records reporting a complication were retained; the others were replaced whenever possible. This choice slightly inflates the rates.
• Confounding of criterion / period / team — everything changed at the same time; no effect can be attributed to the criterion alone.
• Retrospective reclassification — children turned away before surgery do not appear in the records; the gain in access is estimated from children who were operated on.
• Non-standardized follow-up — late complications underestimated, heights sometimes missing or aberrant, eligibility threshold not documented in writing at the time.
• Two deceased patients excluded — including them would have increased major complications.
• Limited power — 51 complications for 7 variables in the multivariate model: the absence of an effect of undernutrition is absence of evidence, not evidence of absence.

Key sources cited in the thesis