A systematic literature search was conducted to identify indexed publications assessing the impact of PHY supplementation on performance and tibia ash content in male broiler chickens aged 22 to 42 days post-hatching. The search was finalized in January 2025 and performed using the databases PubMed, Web of Science, and Scopus. The research strategy was developed using the PICo framework (Andretta et al., 2021), in which the population was defined as broilers, the interest as PHY supplementation, and the context as growth performance or tibia ash content.

Based on this framework, the search strategy used combinations of keywords structured as follows: (“Broiler” OR “Broilers” OR “poultry” OR “poultries” OR “chicken” OR “chickens”) AND (“phytase”) AND (“performance” OR “body weight” OR “average daily gain” OR “feed intake” OR “average daily feed intake” OR “feed conversion ratio” OR “feed efficiency” OR “tibia ash”). Minor adaptations were applied according to the syntax requirements of each database.

Only studies published in English, Portuguese, or Spanish were included, and only peer-reviewed journal articles were considered. To ensure completeness, the reference lists of the selected articles were manually screened for additional relevant publications. All retrieved records were exported to reference management software (EndNote X9, Clarivate Analytics, Philadelphia, PA, USA) for duplicate removal. The relevance of the selected studies was evaluated through a two-stage screening process, beginning with the review of titles and abstracts, followed by a full-text assessment. A flow diagram summarizing the study selection process is presented in Figure 1.

Figure 1

The criteria for selecting publications were as follows: (1) in vivo experiments conducted with commercial broiler chicken breeds; (2) experiments including both non-supplemented control diets (negative control) and diets supplemented with commercial PHY; (3) studies reporting performance parameters, including average daily gain (ADG), average daily feed intake (ADFI), and feed conversion ratio (FCR), or evaluating the effects of PHY on tibia ash content; and (4) studies published from 2000 onward. Publications were excluded if they met any of the following criteria: (1) did not use male broiler chickens or did not specify the sex of the birds; (2) involved sanitary challenges; (3) included treatments combining PHY with other exogenous enzymes; (4) did not include a control treatment; or (5) presented significant inconsistencies in the experimental design, methodology, or reported outcomes.

Studies were retained in the meta-analysis when treatments included PHY as the sole enzyme and when neither the PHY-supplemented nor the control groups were subjected to sanitary challenges. When other treatments within the same study did not meet these criteria, only the eligible treatments were retained for analysis. Studies or treatments using superdosing of PHY (>2,500 FTU/kg) were excluded due to insufficient available data.

Only studies evaluating broilers aged 22 to 42 days post-hatching, or experiments partially conducted within this period, were considered. When studies reported multiple growth phases (e.g., 1-21 and 22-42 days), only the treatment means corresponding to the finisher phase (22-42 days) were extracted. In studies where the experimental period covered the entire growth cycle (e.g., 1-42 days), only the values reported for the finisher phase were retained when available. Studies in which finisher-phase data could not be clearly isolated were excluded from the dataset. Two studies (Dilger et al., 2004; Ceylan et al., 2012) reported experimental periods slightly different from the target window (22-42 days), specifically 22-40 and 22-43 days. These minor deviations were considered biologically comparable to the defined age window and were therefore retained in the meta-analysis. To ensure a precise evaluation of PHY effects, only non-supplemented diets (negative control) with NPP levels matching those used in the PHY-supplemented treatments were included in the meta-analysis.

Initially, 5,710 studies were identified and screened based on titles and abstracts. A total of 54 studies were assessed for eligibility through full-text review for eligibility. After applying the predefined inclusion and exclusion criteria, 33 publications were excluded. Consequently, 21 publications were included in the final meta-analysis. The complete list of selected studies is presented in Table 1.

Table 1

Code	Study	Location	Journal
1	Viveros et al. (2002)	Spain	Poultry Science
2	Dilger et al. (2004)	United States	Poultry Science
3	Wu et al. (2004)	New Zealand	Journal of Applied Poultry Research
4	Zobač et al. (2004)	Czech Republic	Czech Journal of Animal Science
5	Han et al. (2009)	China	Journal of Applied Poultry Research
6	Manangi et al. (2009)	United States	International Journal of Poultry Science
7	Nourmohammadi et al. (2010)	Iran	American Journal of Animal and Veterinary Sciences
8	Junqueira et al. (2011)	Brazil	Acta Scientiarum. Animal Sciences
9	Ceylan et al. (2012)	Turkey	Journal of Animal and Feed Sciences
10	Nourmohammadi et al. (2012)	Iran	Italian Journal of Animal Science
11	Jiang et al. (2013)	China	Italian Journal of Animal Science
12	Midilli et al. (2014)	Turkey	Kafkas Üniversitesi Veteriner Fakültesi Dergisi
13	Taheri et al. (2015a)	Iran	British Poultry Science
14	Taheri et al. (2015b)	Iran	Poultry Science Journal
15	Taheri et al. (2015c)	Iran	British Poultry Science
16	Tizziani et al. (2016)	Brazil	Revista Brasileira de Zootecnia
17	Islam et al. (2017)	Malaysia	Cogent Food & Agriculture
18	Nafea et al. (2019)	Iran	Biochemical and Cellular Archives
19	Mohiti-Asli et al. (2020)	Iran	Italian Journal of Animal Science
20	Javadi et al. (2022)	Spain	Animals
21	França et al. (2023)	Brazil	Livestock Science

The methodology used to build the database, encode the data, and define the variables was based on the approach described by Sauvant et al. (2020). To maintain consistency and comprehensiveness in dietary composition across studies, experiments lacking complete nutrient composition were recalculated using the Practical Program for Formulation of Rations (Garcia-Neto, 2008), considering the Brazilian Tables for Poultry and Swine (Rostagno et al., 2024). When nutritional values were fully reported in the original studies, the published values were used directly; otherwise, missing values were recalculated based on the ingredient composition and the corresponding reference nutrient values. The variability in diet composition within the dataset is summarized in Table 2.

Table 2

	N1	Mean	Minimum	Maximum	SD
Total phosphorus (% of diet)	21	0.53	0.35	1.06	0.17
Phytate phosphorus (% of diet)	21	0.29	0.22	0.70	0.12
Calcium (% of diet)	21	0.81	0.51	1.19	0.14
Metabolizable energy (kcal/kg)	21	3,085	2,914	3,227	96.0
Crude protein (% of diet)	21	19.3	17.0	22.3	0.95
Total lysine (% of diet)	21	1.07	0.90	1.42	0.12
Digestible lysine (% of diet)	21	0.99	0.80	1.30	0.12
Total methionine (% of diet)	21	0.45	0.32	0.60	0.08
Digestible methionine (% of diet)	21	0.41	0.24	0.58	0.09
Total methionine + cysteine (% of diet)	21	0.75	0.34	0.98	0.13
Digestible methionine + cysteine (% of diet)	21	0.69	0.48	0.86	0.10
Total threonine (% of diet)	21	0.75	0.53	0.84	0.07
Digestible threonine (% of diet)	21	0.67	0.53	0.74	0.05
Vitamin D3 (IU/kg of diet)	21	2,670	360	8,800	1,928
Copper (mg/kg of diet)	21	7	2	13	2.8
Zinc (mg/kg of diet)	21	70	17	107	22
Iron (mg/kg of diet)	21	57	14	80	21
Manganese (mg/kg of diet)	21	80	24	125	24
Iodine (mg/kg of diet)	21	0.9	0.2	2.0	0.5
Selenium (mg/kg of diet)	21	0.19	0.03	0.33	0.07
Sodium (% of diet)	21	0.16	0.09	0.21	0.03

¹ Number of studies that reported poultry diet composition.

A meta-analytical approach was employed to model the dependent variables following the methodology described by Andretta et al. (2012). The analysis was conducted through a sequential process including graphical exploration, correlation, and variance-covariance analyses. To account for the hierarchical structure of the data and variability among studies, mixed-effects models were fitted with study included as a random effect, allowing the partitioning of variability attributable to differences among studies while estimating the overall relationships between explanatory variables and responses (St-Pierre, 2001). Because the objective of this study was to evaluate quantitative relationships using meta-regression rather than to estimate pooled treatment effect sizes, traditional publication bias diagnostics (e.g., funnel plots or Egger’s test) were not applied.

Each observation corresponded to a treatment mean extracted from the original publications. Because multiple treatments were often reported within the same experiment, the random study effect accounted for the non-independence and clustering of observations within studies. Fixed effects included PHY level, dietary NPP, Ca:NPP ratio, and their interactions, and quadratic terms were evaluated when biologically relevant. The relationships between PHY and NPP, as well as PHY and the Ca:NPP ratio, were analyzed using separate models to avoid collinearity among dietary P variables. Independent variables were tested, and only those that demonstrated statistical significance (P<0.05) or a tendency (P<0.10) were retained in the final models. Following the approach described by Létourneau-Montminy et al. (2010), the responses of ADG, ADFI, FCR, and tibia ash content were evaluated using quadratic meta-regression models:

and

in which Y_ijk denotes the dependent variable Y in experiment i, at level j of NPP (for equation 1) or Ca:NPP (for equation 2), and level k of PHY. Parameter α represents the overall fixed-effect intercept, while α_i accounts for the random effect of experiment i. Coefficients β₁ and β₂ correspond to linear effects, β₃ and β₄ capture quadratic effects, and β₅ represents the interaction effect. The term e_ijk represents the residual error.

Random effects and residual errors were examined to ensure that they followed a normal distribution. To identify potential outliers, internally studentized residuals (ISR) were investigated, with values exceeding three (|ISR| > 3) flagged and excluded from further analysis (Faridi et al., 2015; Kermani et al., 2023). Across all models, over 95% of the ISR values fell within the range of −2 to 2, confirming a good model fit. Model accuracy was assessed using multiple criteria, including the significance of the estimated parameters, standard error (SE), coefficient of determination (R²), and root mean square error (RMSE). Model adequacy was further assessed by plotting the residuals (observed minus predicted values) against the predicted values of Y to check for linear prediction bias (St-Pierre, 2001). For optimization, the partial derivatives of the models for each dependent variable were calculated following the methodology described by Létourneau-Montminy et al. (2010).

Statistical analyses were performed using mixed-model procedures in Minitab software (Minitab Inc., State College, PA, USA). Graphs were generated using the ggplot2 package in R (version 4.4.2; R Foundation for Statistical Computing, Vienna, Austria).

The meta-analysis included 21 studies evaluating male broiler chickens 22 to 42 days post-hatching, encompassing a total population of 10,706 male broilers. A total of 86 treatment groups, including the respective control groups, were analyzed. Among the commercial breeds represented, Ross 308 was the most frequently studied (n = 49), followed by Cobb 500 (n = 22) and Arbor Acres (n = 15).

The PHY supplementation varied across studies. Among the 59 treatments that used PHY, 40.7% used 3-phytase (EC 3.1.3.8), whereas 37.3% utilized 6-phytase (EC 3.1.3.26). The remaining 22.0% of the treatments did not specify the PHY type. PHY enzymes were derived from various organisms, including Aspergillus niger (44.1%), Escherichia coli (28.7%), Trichoderma reesei (8.50%), Enterobacter sakazakii (5.1%), Serratia odorifera (5.1%), Buttiauxella sp. (3.4%), and Peniophora lycii (3.4%). In addition, 1.7% of the treatments did not specify the donor organism for the exogenous PHY. Furthermore, the final dataset for the meta-analysis included studies conducted in Iran (n = 7), Brazil (n = 3), China, Spain, Turkey, and the United States of America (USA) (n = 2); and the Czech Republic, Malaysia, and New Zealand (n = 1). Most treatments used corn-soybean meal-based diets (77.9%), with a smaller proportion using corn-wheat-soybean meal-based diets (16.3%) or wheat-soybean meal-based diets (5.8%). Only grain-based feedstuffs were included in 82.6% of the diets, while the remaining 17.4% also included animal-based feedstuffs. The feed was presented in various forms: mash (36.0%), cold-pelleted (5.8%), pelleted (4.7%), or unspecified (53.5%). Diets were classified as cold-pelleted diets when the maximum temperature during the pelleting process did not exceed 70 °C (Ravindran et al., 2008). Mineral P supplementation was primarily provided through dicalcium phosphate (89.0%), followed by monocalcium phosphate (6.1%), and tricalcium phosphate (4.9%). Additionally, the dataset revealed that 15.2% of the diets had a Ca:NPP ratio lower than 2:1, while 26.7% had ratios between 2:1 and 3:1. Diets with Ca:NPP ratios between 3:1 and 4:1 accounted for 17.4%, those between 4:1 and 5:1 represented 9.3%, and diets with a ratio greater than 5:1 comprised 31.4% of the dataset.

Summary statistics for the dependent and independent variables included in the meta-analysis of male broilers aged from 22 to 42 days are presented in Table 3.

Table 3

Variable	n	N	Mean	SD	Min.	Max.
Phytase (FTU/kg)	86	21	531	578	0.00	2,500
Non-phytate phosphorus (%)	86	21	0.24	0.09	0.13	0.43
Calcium to non-phytate phosphorus ratio	86	21	3.8	1.8	1.2	6.9
Average daily gain (g/day)	86	21	77.4	14.7	52.1	118.5
Average daily feed intake (g/day)	79	19	146.0	33.6	91.8	231.6
Feed conversion ratio	79	19	1.90	0.27	1.17	2.69
Tibia ash content (%)	21	5	44.7	6.3	32.7	52.7

n - number of observations; N - number of studies; SD - standard deviation; Min. - minimum value of the parameter; Max. - maximum value of the parameter.

Dietary PHY supplementation increased the intercept for ADG and ADFI in male broilers from 22 to 42 days post-hatching, while reducing the intercept for FCR, depending on the dietary NPP and PHY levels (Table 4; Figure 2). The interaction between exogenous PHY supplementation and dietary NPP levels showed a tendency to increase ADG (PHY × NPP; P = 0.092) and significantly increased ADFI (PHY × NPP; P = 0.012), with the strongest effects observed at lower dietary NPP levels exhibited on both ADG and ADFI (P = 0.001; Table 4; Figure 2). The interaction between PHY supplementation and the dietary Ca:NPP ratio on ADG was not significant (P>0.10), and as a result, ADG was excluded from subsequent model development. However, ADFI was significantly affected by PHY supplementation and the dietary Ca:NPP ratio, with greater PHY effects observed at higher Ca:NPP ratios, showing a significant linear effect (P = 0.01).

Table 4

Male broiler growth performance and tibia ash content models
	Average daily gain (g/day)			Average daily feed intake (g/day)			Feed conversion ratio			Tibia ash content (%)
	Coefficient	SE	P	Coefficient	SE	P	Coefficient	SE	P	Coefficient	SE	P
Intercept	56.08	5.28	<0.001	111.56	11.44	<0.001	2.001	8.0 × 10⁻2	<0.001	39.74	3.10	<0.001
Phytase	1.0 × 10⁻2	2.8 × 10⁻3	0.001	1.4 × 10⁻2	4.0 × 10⁻3	0.001	−2.2 × 10⁻4	5.5 × 10⁻⁷	<0.001	1.1 × 10⁻2	3.2 × 10⁻3	0.009
NPP	79.59	15.05	<0.001	126.71	21.71	<0.001	-	-	NS	-	-
Phytase2	-	-	NS	-	-	NS	6.4 × 10⁻8	1.9 × 10⁻1⁰	0.001	−2.3 × 10⁻⁶	1.0 × 10⁻⁶	0.062
NPP2	-	-	NS	-	-	NS	−1.249	3.6 × 10⁻1	0.001	-	-	NS
Phytase × NPP	−2.3 × 10⁻2	1.4 × 10⁻2	0.092	−4.9 × 10⁻2	1.9 × 10⁻2	0.012	3.2 × 10⁻4	1.7 × 10⁻⁶	0.066	−2.6 × 10⁻2	1.4 × 10⁻2	0.096
R2		0.94			0.98			0.98			0.94
RMSE		12.09			25.74			0.21			4.98

RMSE - root mean square error; SE - standard error; NS - not significant (P>0.10).

Figure 2 Description of curve types and regression equations for ADG, ADFI, FCR, and tibia ash content. Panels (A)-(D) illustrate model-predicted responses of male broilers (22-42 days of age) to different dietary phytase levels: solid thin line – 0 FTU/kg; dashed thick line – 500 FTU/kg; dotted line – 1,000 FTU/kg; and solid thick line – 1,500 FTU/kg. The regression equations for each performance parameter are as follows: (A) ADG (g) = 56.08 + 1.0 × 10⁻² × Phytase + 79.59 × NPP − 2.3 × 10⁻² × Phytase × NPP (RMSE = 12.09, R² = 0.94). (B) ADFI (g) = 111.56 + 1.4 × 10⁻² × Phytase + 126.71 × NPP − 4.9 × 10⁻² × Phytase × NPP (RMSE = 25.74, R² = 0.98). (C) FCR = 2.001 – 2.2 × 10⁻⁴ × Phytase + 6.4 × 10⁻⁸ × Phytase² − 1.249 × NPP² + 3.2 × 10⁻⁴ × Phytase × NPP (RMSE = 0.21, R² = 0.98). (D) Tibia ash (%) = 39.74 + 1.1 × 10⁻² × Phytase – 2.3 × 10⁻⁶ × Phytase² – 2.6 × 10⁻² × Phytase × NPP (RMSE = 4.98, R² = 0.94).

Diets supplemented with PHY improved FCR in male broilers aged 22 to 42 days post-hatching (Table 4; Figure 2). PHY exerted a significant linear effect (P<0.001), while dietary NPP influenced FCR through significant quadratic effect (P = 0.001), with a tendency for a greater impact in low-NPP diets (PHY × NPP; P = 0.066; Table 4; Figure 2). Additionally, PHY supplementation displayed significant quadratic effects (P = 0.007) on FCR, whereas the dietary Ca:NPP ratio had a significant positive linear effect (P = 0.011) in male broilers (Table 5; Figure 3). Tibia ash content in male broilers aged 22 to 42 days was influenced by PHY supplementation and dietary NPP levels, showing a tendency for a quadratic effect (P = 0.062) for PHY supplementation and a linear effect (P<0.001) of increasing dietary NPP levels (Table 4; Figure 2). The impact of PHY supplementation on tibia ash content was more pronounced at lower dietary NPP levels (PHY × NPP; P = 0.096; Table 4; Figure 2). Furthermore, the interaction between PHY supplementation and the Ca:NPP ratio had a significant effect on tibia ash content, with a stronger PHY response observed in diets with higher Ca:NPP ratios (PHY × Ca:NPP; P = 0.005; Table 5; Figure 3). Additionally, PHY supplementation influenced tibia ash content in a quadratic pattern (P = 0.049; Table 5; Figure 3).

Table 5

Male broiler growth performance and tibia ash content models
	Average daily feed intake (g/day)			Feed conversion ratio			Tibia ash content (%)
	Coefficient	SE	P	Coefficient	SE	P	Coefficient	SE	P
Intercept	166.69	10.39	<0.001	1.816	0.081	<0.001	39.93	3.00	<0.001
Phytase	-	-	NS	−1.4 × 10⁻⁴	3.2 × 10⁻⁵	<0.001	-	-	NS
Ca:NPP ratio	−6.211	1.104	<0.001	2.4 × 10⁻2	9.1 × 10⁻3	0.011	-	-	NS
Phytase2	-	-	NS	4.9 × 10⁻⁸	1.8 × 10⁻⁸	0.007	−1.9 × 10⁻⁶	8.5 × 10⁻⁷	0.049
(Ca:NPP)2	-	-	NS	-	-	NS	-	-	NS
Phytase × Ca:NPP	1.0 × 10⁻3	3.0 × 10⁻⁴	0.001	-	-	NS	1.1 × 10⁻3	2.9 × 10⁻⁴	0.005
R2		0.97			0.98			0.94
RMSE		24.84			0.21			4.75

RMSE - root mean square error; SE - standard error; NS - not significant (P>0.10).

Figure 3 Description of curve types and regression equations for ADFI, FCR, and tibia ash content. Panels (A)-(C) illustrate the predicted responses of male broilers (22-42 days of age) to different dietary phytase inclusion levels: solid thin line – 0 FTU/kg; dashed thick line – 500 FTU/kg; dotted line – 1,000 FTU/kg; and solid thick line – 1,500 FTU/kg. Curves represent the effects of the dietary calcium-to-non-phytate phosphorus (Ca:NPP) ratio on the following parameters: (A) ADFI (g) = 166.69 – 6.211 × Ca:NPP + 1.0 × 10^–3 × Phytase × Ca:NPP (RMSE = 24.84, R² = 0.97). (B) FCR = 1.816 – 1.4 × 10^–4 × Phytase + 2.4 × 10^–2 × Ca:NPP + 4.9 × 10^–8 × Phytase² (RMSE = 0.21, R² = 0.98). (C) Tibia ash (%) = 39.93 – 1.9 × 10^–6 × Phytase² + 1.1 × 10^–3 × Phytase × Ca:NPP (RMSE = 4.75, R² = 0.94).

The residual plots against the predicted values shown in Figures 4 and 5 indicate the adequacy of the model for assessing the effects of various dietary NPP levels and Ca:NPP ratios in broilers fed different PHY levels. Residuals displayed a good distribution around zero across the different models, indicating a good fit. The residuals differed from the predicted values by 15.9%, 17.8%, 11.2%, and 11.8% for ADG, ADFI, FCR, and tibia ash content, respectively, in the models that included various dietary NPP levels. The residuals differed from the predicted values by 17.1%, 11.1%, and 9.1% for ADFI, FCR, and tibia ash content, respectively, in the models evaluating the effects of different Ca:NPP ratios in broiler diets with varying PHY supplementation.

Figure 4 The lines represent the regression of residuals on predicted values for each response variable: (A) ADG [Y = −22.74 + 0.331 × predicted ADG; R² = 0.18]; (B) ADFI [Y = 162.1 − 1.095 × predicted ADFI; R² = 0.09]; (C) FCR [Y = 0.70 − 0.351 × predicted FCR; R²<0.001]; and (D) Tibia ash content [Y = 24.1 − 0.560 × predicted tibia ash; R²<0.001].

Figure 5 The lines represent the regression of residuals on predicted values for each response variable: (A) ADFI [Y = 117.3 − 0.785 × predicted ADFI; R² = 0.003]; (B) FCR [Y = −0.23 + 0.148 × predicted FCR; R²<0.001]; and (C) tibia ash [Y = 48.76 − 0.902 × predicted tibia ash; R² = 0.63].