Care, management, and statute each mandate restraint-reporting in psychiatric settings in England. PROD-ALERT in this journal ("PA1") correlated log incidence of restraint, log institutional size, and log detention. The period was September 2020 to August 2021. It showed a clear trend among reporters. Restraint correlated with institutional size and use of legal detention. Some large detaining providers reported no restraints per month despite that trend. Inference from size suggested that non-complete reporters restrained 1,774 people per month. This paper "PA2" develops analysis repeating it for September 2021 to August 2022. PA2 shows how to count L-information, i.e., questionable information, added by null reports, by applying an L-test to data sets. PA2 uses illustrative vignettes about human height to ground L-information scores from English restraint reporting. In PA2, reported restraint again correlates with size and detention as in PA1. PA2 shows evolving data. Providers still follow a trend in restraint by size and detention. Providers which newly report restraint are on trend. Inference suggests that non-complete reporters restrained 1,305 people per month (536-3233), 95% CI, a large but reduced number since PA1. English restraint data have an L-test L-information score of increase in information by a factor of L = 145. This is as surprising as claiming that an average English man of 1.72m is 2.64m tall. Persons restrained per month is a robust measure continuing to log-correlate with size and legal compulsion. Providers over a certain size who report null restraint probably have some. Restraint remains underreported in England. Imputation of incomplete reporters shows a large shrinking cohort of patients detained by incomplete reporters. Knowledge of this may promote reporting. Improved reporting, and the infrastructure and integrity it demands, may help providers measure and reduce restraint. PA1 remains unrefuted. L-test can measure L-information in intuitively representable ways. The informational effect of nulls on the reliable data set is similar to a claim that an average-heighted man is as tall as people with clinical gigantism.