MIMIC model failed to converge: using lavaan to assess the effect of maternal empowerment on child malnutrition

Question

indicators <- c("stunting", "wasting", "underweight_small") 

independent_vars<- c("colostrum", "ors", "familyplanning","foodgroups", "visits", "occupation", "respIncome", "resphealthdec", "lhhpurchases", "ownhouse", "wland")

control_vars <- c("maternalage", "maternaleduc", "BMI_cat", "wland", "child_age", "childgender", "birthorder", "twin", "residence", "hhhage", "hhgender", "residence", "religion", "wstatus", "anc")

model<- "
  # Latent variable with indicators
  child_malnutrition =~ stunting + wasting + underweight_small

  # Independent variables
  child_malnutrition ~ colostrum + ors + familyplanning + foodgroups + visits + occupation + respIncome + resphealthdec + lhhpurchases + ownhouse + wland
 
  # Control variables
  child_malnutrition ~ maternalage + maternaleduc + BMI_cat + child_age + childgender + birthorder + twin + residence + hhhage + hhgender + religion + wstatus + anc
"

fit <- sem(model, data = April_4_Dat)

summary(fit)

I have tried setting start values, it somehow converged but the results had no p values and i got a warning that the results were not reliable. I also tried measuring child nutrition as a composite variable,still the results were unreliable. I am using Demographic Health survey data

Set initial values for the parameters
start_vals <- c(col = 0.5, orth = 0.3)

# Set optimization method
opt_method <- "BFGS" 

# Fit the model with different starting values and optimization methods
fit1 <- sem(model, data = April_4_Dat, start = start_vals, optim.method = opt_method)
fit2 <- sem(model, data = April_4_Dat, start = c(col = 0.1, orth = 0.6), optim.method = "NLMINB")

# Compare the results
summary(fit1)
summary(fit2)