Dosing pole recommendations for lymphatic filariasis elimination: A height-weight quantile regression modeling approach

Abstract

Background: The World Health Organization (WHO) currently recommends height or age-based dosing as alternatives to weight-based dosing for mass drug administration lymphatic filariasis (LF) elimination programs. The goals of our study were to compare these alternative dosing strategies to weight-based dosing and to develop and evaluate new height-based dosing pole scenarios.

Methodology/principal findings: Age, height and weight data were collected from >26,000 individuals in five countries during a cluster randomized LF clinical trial. Weight-based dosing for diethylcarbamazine (DEC; 6 mg/kg) and ivermectin (IVM; 200 ug/kg) with tablet numbers derived from a table of weight intervals was treated as the "gold standard" for this study. Following WHO recommended age-based dosing of DEC and height-based dosing of IVM would have resulted in 32% and 27% of individuals receiving treatment doses below those recommended by weight-based dosing for DEC and IVM, respectively. Underdosing would have been especially common in adult males, who tend to have the highest LF prevalence in many endemic areas. We used a 3-step modeling approach to develop and evaluate new dosing pole cutoffs. First, we analyzed the clinical trial data using quantile regression to predict weight from height. We then used weight predictions to develop new dosing pole cutoff values. Finally, we compared different dosing pole cutoffs and age and height-based WHO dosing recommendations to weight-based dosing. We considered hundreds of scenarios including country- and sex-specific dosing poles. A simple dosing pole with a 6-tablet maximum for both DEC and IVM reduced the underdosing rate by 30% and 21%, respectively, and was nearly as effective as more complex pole combinations for reducing underdosing.

Conclusions/significance: Using a novel modeling approach, we developed a simple dosing pole that would markedly reduce underdosing for DEC and IVM in MDA programs compared to current WHO recommended height or age-based dosing.

Fig 1. 3-step modeling process.

Fig 2. WHO age-based (DEC) and height-based (IVM) dosing compared to recommended weight-based dosing.

Percentages calculated based on the number of participants equal to the recommended dose (recommended), above the recommended dosage (ARD), or below the recommended dose (BRD). Values reported above bars correspond to the n (%) within the dosage group, and values in the inset correspond to the overall n (%).

Fig 3. Quantile regression predictions of weight from height for Global models.

Fig 4. Quantile regression predictions for models stratified by country.

Fig 5. Quantile regression predictions for models stratified by sex.

Fig 6. Dosing poles for Global models.

Different colors correspond to different # of tablets administered, and numbers in bars correspond to the height range (in cm) for each tablet group. Hybrid 4 corresponds to the 4-tablet maximum dosing pole, and Hybrid 6 corresponds to the 6-tablet maximum dosing pole. The hybrid poles can be used for dosing either IVM or DEC.

Fig 7. Global model plots (6 [DEC] and 7 [IVM] tablet maximum) of % BRD, ARD, and at recommended dosages for IVM and DEC for all quantiles.

Colored lines correspond to the dosage category with the recommended dosage colored green, below the recommended dosage (BRD) colored red, and above the recommended dosage (ARD) colored orange. Solid lines correspond to model predictions for the percentage of participants receiving ARD, BRD or the recommended dosage for the model. Horizontal lines correspond to WHO recommended dosages. Numbers below the plot correspond to percentages of total participants for the number of tablets ARD, BRD, or recommended.

Fig 8. Global DOLF 4-tablet maximum hybrid model plots of the percentage receiving BRD, ARD, and recommended dosages for IVM and DEC for all quantiles.

Colored lines correspond to the dosage category with the recommended dosage colored green, below the recommended dosage (BRD) colored red, and above the recommended dosage (ARD) colored orange. Solid lines correspond to model predictions for the % of participants ARD, BRD or at recommended dosage for the model. Horizontal lines correspond to WHO recommended dosages. Numbers below the plot correspond to percentages of total participants for the number of tablets ARD, BRD, or recommended.

Fig 9. Global DOLF 6-tablet maximum hybrid model plots of % BRD, ARD, and recommended dosages for IVM and DEC for all quantiles.

Colored lines correspond to the dosage category with the recommended dosage colored green, below the recommended dosage (BRD) colored red, and above the recommended dosage (ARD) colored orange. Solid lines correspond to model predictions for the percentage of participants ARD, BRD or at recommended dosage for the model. Horizontal lines correspond to WHO recommended dosages. Numbers below the plot correspond to percentages of total participants for the number of tablets ARD, BRD, or recommended.

Fig 10. Percentage of participants below recommended dosage (BRD) across quantiles by country for the Global models.