A 343-page report released today by the D.C. Auditor finds that the Metropolitan Police Department’s use of force largely complies with best practices and federal guidelines, but cites several “deficiencies” that should be addressed.

The review looks at MPD’s methods of incapacitating suspects using firearms, hand controls, and takedowns, among other tactics, from 2008 to 2015. It also examines whether MPD is still in compliance with a 2001 agreement it entered into with the Department of Justice on use of force polices and practices. The report arrives as police departments across the country face criticism over allegations of brutality during incident responses and as movements like Black Lives Matter draw a line between excessive use of force and race.

“MPD has narrowed the categories of use of force that are now documented and investigated,” D.C. Auditor Kathy Patterson writes in the introduction. “The structure and management of the Force Investigation Team have changed and, at times, the quality of investigations has not been as robust as it was designed to be. In addition, the report documents significant delays in the U.S. Attorney’s Office review of officer-involved fatal shooting cases.”

Based in part on fieldwork from May to September of last year, the report finds that, between 2008 and 2015, uses of batons, canine bites, and pepper spray “have remained at relatively consistent levels,” whereas uses of hand controls and tactical takedowns “show much greater variation, with a high of 503 in 2009 and a low of 260 in 2011.” But the use of neck-restraint techniques like chokeholds—highlighted last August in an Office of Police Complaints report—was of particular concern to the auditors, who recommend that such methods be treated as “serious use[s] of force.”

The report—authored by the Bromwich Group, a consultancy—also cites MPD data on officers’ use of firearms. From 2001 to 2015, the incidence of officers intentionally discharging their firearms “declined significantly, from 28 intentional discharges in 2001, 30 in 2004, and 31 in 2007, to seven in 2010, nine in 2012, nine in 2014, and 15 in 2015.” Officers in D.C. fatally shot between three and eight people per year since 2009. But “none of the cases involving a fatal shooting has resulted in a criminal prosecution.”

“Although the Review Team identified problems with the process that has been in place to review officer-involved fatal shootings—centering on the length of time it takes to consider the case for potential prosecution—the data do not support the claim that MPD officers use their firearms excessively,” the auditors explain after noting that fact.

On average, the U.S. Attorney Office’s took more than 19 months to conduct criminal investigations and prosecutorial reviews for officer-involved fatal shootings between 2009 and 2014. That’s “far longer than the typical investigation and the review period” during prior independent monitoring, between 2002 and 2008, per the report. The review cites “the need to obtain various forensic reports” and “a paper-intensive, multi-layered review at the USAO” as some of the key reasons for the months-long process.

In a response attached to the report, U.S. Attorney Channing Phillips writes that he shares the overall goal of the review, but disagrees with some of its findings and recommendations. Among those: that prosecutors comment on ongoing investigations (the report specifically emphasized a 2014 incident between MPD and Gregory Marcus Gray, who died) and that they review “the most series use of force cases within six months” (versus nine months).

“Science is not always swift, but it is certain,” MPD Chief Cathy Lanier said of the District’s forensic procedures at a press conference Thursday morning. “We have to be patient and not rush an investigation without… forensics.”

The review contains 38 recommendations, including that MPD review its use of force policies at least every two years, that MPD reinstate required reports for hand controls and resisted handcuffing “even when there is no injury or complaint of pain,” and that MPD require “all civilian [and] officer witnesses to be [interviewed and recorded].”

“MPD has said it agrees with and will implement 15, agrees in part with 13, and disagrees with 10—although of the 10 with which it disagrees, four relate to a new MPD program, and [it] views the recommendations as premature.”

Lanier said MPD’s focus is ultimately on keeping officers from engaging in misconduct. She added that the agency monitors disciplinary issues, upticks in sick-leave, and complaints against both its sworn officers and civilian staff.

“It’s important that where there are causes of distrust between residents and [the] police force, they’re addressed,” D.C. Council Chairman Phil Mendelson said. “This report is proactive. This report is useful and will be useful as a blueprint for the Council and in particular the judiciary committee to conduct its oversight” of law enforcement.

Patterson said the report would be shared with about 250 police and political leaders around the country.