Did Methamphetamine Patients Improve Cognition as Much as Other Addiction Patients? A 110-Person Inpatient Study

The answer: improvement was observed in both groups, not a methamphetamine-specific treatment advantage

In a German observational inpatient study of 110 adults, 55 entered treatment primarily for methamphetamine dependence and 55 for dependence on other substances. At admission, the methamphetamine group scored lower on computerized working speed and accuracy. Among the smaller subset assessed near discharge, both groups improved. Importantly, the investigators reported no significant group-by-time interaction for either Cognitrone speed or accuracy. That pattern does not show a statistically demonstrated advantage or disadvantage in the trajectory of cognitive improvement specific to methamphetamine dependence.

Treatment retention told a similarly qualified story. Premature discharge occurred in 36.4% of the methamphetamine group and 43.6% of the comparison group, a difference that was not statistically significant (p = .44). This is a finding of no demonstrated group difference in these data, not proof that two treatment populations have identical outcomes.

The study describes treatment-course and test-score associations, not an experiment administering methamphetamine and measuring a person’s blood level or fitness to drive. Because groups were observational, later assessments were incomplete, and no untreated control was studied, neither the measured improvements nor their timing can be attributed to a particular rehabilitation component with the certainty of a randomized treatment comparison.

The original research and the precise question it tested

Behle and colleagues published a primary longitudinal observational investigation in World Journal of Psychiatry in 2022. The complete original Methods, Results, clinical assessment tables, treatment-course accounting, Discussion and licensing statements were examined. Its full original citation is linked here: Behle et al., Treatment outcome, cognitive function, and psychopathology in methamphetamine users compared to other substance users. The article compares preexisting patient groups admitted to one inpatient addiction service, rather than drug-naive people randomized to a controlled stimulant challenge.

The narrow practical question is whether people being treated primarily for methamphetamine dependence showed worse baseline computer-task performance, different cognitive change during a residential program, or different retention than people treated for dependence on other drugs. This question is narrower than whether methamphetamine always causes cognitive impairment or whether rehabilitation reliably restores a particular individual’s cognitive capacity.

Who were the independent participants?

There were 110 independent patients at the initial assessment: 55 classified in the methamphetamine-primary group and 55 in the other-substance-primary group. There were 89 men and 21 women overall. The investigators reported group mean ages of approximately 30 and 32 years, without a statistically significant age difference in this sample. Each person counts once toward the baseline biological sample; later testing of the same person does not create another independent participant.

The comparison population is particularly important. It was not a clean non-stimulant control arm. The paper’s substance-history table includes 31 patients with amphetamine use in the other-substance group, along with alcohol, cannabis, cocaine and other drug histories. Methamphetamine-primary patients were also allowed to have used additional substances. Labels describe the principal dependence motivating admission, not toxicologically isolated lifetime exposure. These overlapping histories prevent a simple one-drug-versus-no-drug causal interpretation.

What happened during the six-month residential program?

The planned inpatient course comprised approximately two weeks of entry evaluation, a 22-week core phase, and two weeks of discharge planning. Group sessions were central, with psychoeducation, identifying triggers, relapse-prevention practices, mindfulness, individual psychotherapy and other interventions among the offered elements. Patients were not individually randomized to one module or another. The program therefore represents a complex clinical setting rather than a test of one discrete treatment ingredient.

At entry, staff obtained cognitive and psychiatric assessments; comparable instruments were intended for a second measurement in the final weeks before discharge. Some patients left treatment early, some completed treatment without providing usable final study assessments, and some had different available sample sizes for individual instruments. The outcome called treatment completion must be analyzed separately from the research outcome called availability of two Cognitrone measurements.

Why the enrollment, completion and follow-up denominators must remain separate

Although 110 patients entered baseline assessment, 66 of 110 completed the treatment program, and 44 of 110 discontinued it before planned completion. Separately, only 55 patients completed the second research assessment—27 methamphetamine-primary and 28 other-substance-primary. These are not interchangeable denominators. A person can be included in the treatment-completion statistics while lacking usable follow-up scores for a particular psychometric instrument.

The paired Cognitrone speed and accuracy analyses represented 26 methamphetamine-primary and 28 other-substance-primary patients. Other scales had still different numbers. Treating the 110 baseline patients as though all were measured twice would overstate the amount of longitudinal evidence and obscure possible attrition-related selection. The study followed the same clinical population, but the analysis of change represents a selected subset rather than complete repeated testing of every enrolled person.

What the baseline numerical comparisons actually show

At intake, several test differences were nominally statistically significant. The original paper reports standardized Cognitrone T scores: the methamphetamine-primary group averaged 49.1 on working speed versus 54.3 in the other-substance group, and 43.0 on working accuracy versus 47.1. Raven nonverbal IQ scores also differed. These are group-level standardized test outcomes, not blood concentrations or measured degrees of drug intoxication.

Each comparison has its own sample count because not every patient completed every instrument. The following newly organized table isolates selected baseline cognitive contrasts instead of treating all assessments as if they involved all 55 patients per arm.

Evidence table 1: selected baseline cognition, with actual test denominators

Interpretive note: The lower methamphetamine-group means are genuine descriptive contrasts, but educational history, polydrug profiles, coexisting symptoms and selection into treatment complicate causal explanations. A p-value for a group association does not convert a test score into an individual impairment finding.

Why a cognitive test result is not proof of drug-induced brain injury

The Cognitrone assessment required participants to evaluate geometric figures on a computer. Its working-speed and working-accuracy scores are standardized behavioral performance metrics. The paper did not use these results as pathology-confirming tests, and a difference in group mean T scores does not establish an acquired lesion, neuropathologic mechanism, or permanent deficit in a specific patient.

Drug-use histories, educational attainment and psychiatric symptoms could influence performance. The groups also differed in some substance-use histories and dimensions of conscientiousness or depressive burden. Practice with the test environment can improve repeated results even when no specific therapeutic process changes cognition. Although the paper addresses clinically meaningful test outcomes, causation requires designs capable of separating treatment, prior exposure, withdrawal stage and other explanations.

What changed in Cognitrone speed at reassessment?

Among the 26 methamphetamine-primary participants with the reported paired speed results, mean standardized speed increased from 48.81 at baseline to 54.08 near discharge. For the 28 other-substance-primary participants, the corresponding mean increased from 57.18 to 62.61. The reported omnibus time effect was significant (p < .001), as was the group effect on speed across measurement occasions (p < .001).

The decisive interaction comparison was p = .94: there was no statistically established difference in how the groups changed over time. Both arithmetic changes are upward, but comparing the raw difference of 5.27 points with 5.43 points is not a substitute for the paired mixed-model test. Even a precisely measured within-person test-score increase need not imply a drug-specific treatment mechanism.

What changed in Cognitrone accuracy, and what was nonsignificant?

Working accuracy rose among the paired methamphetamine-primary participants from 43.62 to 50.50 T-score points, and among the other-substance-primary participants from 44.93 to 52.54. Again, there was a significant time effect (p < .001). The authors did not report a significant between-group accuracy effect for this paired subset, and the group-by-time interaction was described as nonsignificant, although its printed p-value notation is ambiguous.

The source prose expresses some p-value comparisons with an unusual less-than sign even where the accompanying text calls them nonsignificant. In this article the null conclusion follows the stated analysis and study table; the unusual print notation is a source-reporting limitation, not an invitation to invent corrected inferential statistics. The important result is improvement observed in both groups without a demonstrated group-specific accuracy trajectory.

Evidence table 2: selected within-patient changes and true interaction tests

Interpretive note: T0 and T1 are repeat observations of the same participants in each row; their total must not be multiplied by visits. Improved means do not establish that rehabilitation alone caused improvement or that an individual became fit to drive.

How did craving and psychiatric symptom measures behave?

Self-reported craving scores declined in both arms. For patients with available paired craving measurements, the methamphetamine-primary mean moved from 14.39 to 8.57, and the other-substance-primary mean from 14.59 to 8.96. The reported time effect was significant while the group-by-time comparison was not. This supports a common observed direction during inpatient care; it is not proof of a substance-specific reduction caused by any isolated module.

The paper also describes reductions over time on Hamilton depression, Beck depression and symptom-checklist scores. Different instruments had different respondent counts and varied baseline scores. A clinical depression scale does not measure an amphetamine molecule or a neurochemical dose. Psychological distress can influence task performance, but this observational paper cannot identify how much of any cognitive change is attributable to mood improvement, abstinence, structured schedule, repeated testing or another cause.

Treatment completion and the genuine null retention comparison

Premature treatment termination occurred in 20 of 55 methamphetamine-primary patients (36.4%) and 24 of 55 other-substance-primary patients (43.6%). The group comparison was p = .44, which is nonsignificant at the original study’s threshold. The numerically lower methamphetamine dropout proportion is therefore not evidence of a demonstrated retention advantage. Nor should nonsignificance be described as a formal equivalence test.

The authors recorded an overall 40% dropout rate. They described voluntary departure, violation of institution rules, unreported relapse detected during care, and transfer among recorded reasons. Reasons were not associated with group at conventional significance (p = .21). Average treatment days were 159 for methamphetamine-primary and 135 for comparison patients, but the difference also failed conventional significance (p = .07). These are useful logistical observations with limited inferential precision.

Did the number of therapy sessions differ reliably?

Other-substance-primary participants attended a mean of 103 group sessions compared with 87 for the methamphetamine-primary group, whereas means for individual-therapy contacts were 22 and 27 respectively. Neither of the reported comparisons reached conventional significance (group-session p = .07; individual-session p = .08). A reader should not label one group more compliant or more intensively treated on the strength of these p-values alone.

As with days enrolled, counts of sessions are entangled with who remained in the program. A patient who left earlier had less opportunity to accumulate group or individual meetings. That mathematical dependence does not reveal engagement quality, therapeutic alliance or specific effectiveness. The study can describe service utilization and outcomes but not determine which program element produced change.

Baseline predictors of leaving treatment: exploratory, not individual forecasts

Exploratory regression models considered candidate baseline correlates of dropout. Higher neuroticism scores were associated with lower estimated dropout odds (reported odds ratio 0.93, 95% CI 0.87–0.99, p = .03), while higher baseline Cognitrone accuracy was associated with higher dropout odds (odds ratio 1.05, CI approximately 1.00–1.09, p = .04). These patterns should not be reworded as proof that neuroticism is protective or cognitive ability causes a person to abandon treatment.

Working speed (p = .20), Raven IQ (p = .90) and baseline craving (p = .99) were nonsignificant in these reported predictive analyses. Several variables were tested with relatively small available samples; the authors did not adjust every p-value for the entire family of exploratory comparisons. A single marginal p-value, without independent validation and calibrated discrimination, is not a reliable individual dropout-prediction rule.

Was the comparison group free of amphetamines or other stimulants?

No. The original substance-history table includes 31 comparator patients with an amphetamine history. By contrast, all 55 people in the methamphetamine-primary category had methamphetamine as the primary treatment substance. Patients could also have histories of alcohol, cannabis, tobacco, cocaine or other substances, with different frequencies across groups. The division was clinical primary substance dependence, not an analytical classification of absolutely exclusive lifetime drug exposure.

That detail affects how one should interpret differences in cognition. It would be inaccurate to describe the study as comparing 55 methamphetamine users with 55 lifetime stimulant-naive controls. When both arms carry heterogeneous substance histories and other baseline differences, observed group contrasts cannot be cleanly assigned to methamphetamine’s pharmacology alone.

What toxicology was actually obtained during care?

The published clinical protocol mentions urine testing in the management of suspected unreported substance use, with a positive result classified as an unreported relapse for disciplinary treatment purposes. This does not transform the research into a validated longitudinal concentration study. The paper does not report individual quantitative parent methamphetamine or amphetamine measurements, stereoisomer differentiation, specimen collection times for a forensic toxicokinetic model, confirmatory LC–MS/MS values or analytical uncertainty sufficient to back-calculate individual dose.

No parent drug was quantified in legally collected human venous whole blood. Neither drug dose in milligrams nor a standardized cognitive score is a blood concentration. A clinical urine screen also differs from separated serum or plasma, oral fluid, hair, sweat, vitreous, breath and animal/artificial specimens; results in one matrix cannot be silently substituted for another. For general context, see the published guides on biological matrices in forensic toxicology and confirmatory drug testing.

How whole blood, serum, plasma and urine differ from behavioral testing

Forensic venous whole blood contains cells and plasma and requires documented collection and validated chemical analysis for a specific analyte. Serum is a cell-free fraction obtained after clotting; plasma is separated in an anticoagulated sample. Even when those matrices derive from the same person and time point, drug distribution and analytical units require explicit matrix-specific handling. A retrospective reported drug-use history does not create a measured blood concentration.

Urine typically helps address excretion or exposure questions, not direct concurrent brain concentrations. Oral fluid, sweat patches and hair have different collection windows and contamination considerations. The original Behle experiment was principally a longitudinal clinical and neuropsychological evaluation. It should not be cited as a calibration study for the relationship between a venous whole-blood concentration and a Cognitrone T score, because those paired quantitative measurements were never reported.

Why the observed improvements do not prove a controlled treatment effect

The inpatient intervention was not randomized against an untreated, delayed-treatment, or structurally matched alternative condition. Repeated task administration can create familiarity; withdrawal-related symptoms may change independently; sleep, housing stability, nutrition, psychosocial contact and contemporaneous other-drug use could vary throughout the stay. Because these influences were not separately randomized, the time effect is an association with time in care, not proof of a specified biological mechanism.

Attrition further complicates causal claims. Participants with complete second assessments could differ from those who declined, left, or were unavailable. The original study provides clinically informative evidence about people observed through treatment, not a randomized estimate of what would have happened to all 110 people without inpatient care.

A nonsignificant interaction does not establish equivalence

A common interpretive error is to read a nonsignificant group-by-time test as positive proof that both groups improved by exactly the same amount. Those tests instead indicate that the investigators did not detect a difference in slopes using their data and analysis. Equivalence or noninferiority would require a prespecified clinically defensible margin, sufficient power and corresponding interval-based analyses. The primary study did not present such a trial framework.

The reported averages can support the narrower statement that both observed paired subsets improved and their group-by-time contrasts were nonsignificant. The data do not establish that every methamphetamine-primary patient improved or that the treatment experience has the same benefit for every diagnosis and severity level.

Multiple comparisons and the meaning of a borderline result

The investigators assessed cognitive speed, accuracy, several personality traits, depressive symptom scales, craving, psychosocial measures and exploratory predictors. With many comparisons, nominal p-values near .05 deserve careful attention to multiplicity, sample size and whether the comparison was prespecified. The paper did not impose a single correction across every exploratory result; this limits how confidently isolated weak associations can be generalized.

Reporting genuine tested nulls is equally important. Treatment dropout p = .44, speed interaction p = .94 and the stated nonsignificant accuracy interaction, education-associated differences, and numerous baseline symptom contrasts collectively present a heterogeneous observational picture—not a universal positive or universally impaired profile. A result described as unmeasured is also different from a statistically tested null result, particularly for forensic blood drug levels and driving outcomes.

What these results may and may not contribute to forensic opinions

At the population level, the study demonstrates that a clinical population whose primary substance was methamphetamine had lower mean baseline performance on selected psychometric measures than an addiction-treatment comparison group and that observable within-group gains occurred among tested follow-up subsets. It may inform questions about clinical heterogeneity, likely relevance of treatment setting, and the risk of mistaking a group mean for a person-specific finding.

It cannot establish when a specific person last took methamphetamine, the concentration in their venous blood on a specific date, their driver or aviation performance, credibility as a witness, capacity at an incident, whether an accident was caused by intoxication, or that a therapy intervention reversed drug-caused neuropathology. Those issues require different individual evidence, specimens, validated analytical methods and appropriately designed outcome studies.

Limits imposed by site, era, composition and measurement

This was one German residential service, with participants assessed between 2016 and 2019 and a clinical program intended to run roughly six months. The sample was predominantly male, some baseline variables differed between groups, multiple substance histories overlapped, and the cognitive findings relied on standardized test measures rather than direct neurological lesions or pharmacokinetic exposure estimates. These features restrict transportability to other facilities, populations and legal scenarios.

The attrition between admission, completion and available paired assessments is substantial. Reported means and p-values summarize available data and do not establish missing-not-at-random mechanisms, effect distributions for absent cases, or follow-up durability after discharge. There was no independent replication arm, no randomized untreated control and no standard toxicology specimen series that could support individual mechanistic reconstruction.

Practical reading of the evidence without overstatement

Ask first which denominator a statement uses: 110 entered, 66 completed treatment, 55 repeated the general assessment, and approximately 26 plus 28 contributed paired cognition measures. Next distinguish baseline between-group contrasts from the within-person time effect and from the much more demanding drug-group-by-time interaction. Then identify whether a variable is a behavioral T score, clinical symptom rating, attendance count, qualitative urine screen or actually quantified chemical mass in a specified forensic specimen.

Under this discipline, the strongest defensible conclusion is that this inpatient cohort showed clinically relevant baseline differences, improvements in measured cognitive and symptom scores among followed participants, and no demonstrated methamphetamine-specific disadvantage on the studied cognitive-change or dropout tests. Stronger claims about causal cognitive recovery, individual drug levels or legal impairment are not outcomes of this investigation.

Frequently asked scientific and forensic questions

Did the methamphetamine group perform worse at baseline?

On selected standardized tests, yes: Cognitrone working-speed T scores averaged 49.1 versus 54.3 (p = .002) and accuracy 43.0 versus 47.1 (p = .03), with different test-specific participant counts. These are associations between existing clinical groups, not proof that methamphetamine alone caused the contrast or that any individual was impaired.

Did both groups improve on the computerized task?

Among the smaller paired subsets, both groups showed higher mean Cognitrone speed and accuracy near discharge, with reported time effects p < .001. The speed group-by-time result p = .94 and the reported accuracy interaction were nonsignificant, so a unique methamphetamine-group improvement advantage or disadvantage was not demonstrated.

Was treatment dropout significantly greater for methamphetamine patients?

No. Premature treatment termination was 20/55 (36.4%) in the methamphetamine-primary group versus 24/55 (43.6%) in the other-substance-primary group, p = .44. This nonsignificant contrast does not establish equivalence of completion probabilities.

Why were only about 54 people included in the paired cognitive results?

The baseline study enrolled 110, but 55 people participated in the second overall assessment and the paired Cognitrone subset was 26 methamphetamine-primary plus 28 other-substance-primary. Different instruments had different complete data; the reduction in observed cases raises attrition and selection concerns.

Was the comparison group free of stimulants?

No. The comparison group was defined by primary dependence on substances other than methamphetamine, and its reported histories included 31 people with amphetamine use. Both groups could have used additional drugs; the contrast was not methamphetamine versus lifelong stimulant abstinence.

Did the study quantify methamphetamine in legal whole blood or prove recent use?

No. The investigators discussed clinical urine testing for unreported relapse but did not publish confirmed quantitative methamphetamine concentrations in legally collected human venous whole blood, plasma or serum. A test score, urine screening practice or historical use classification does not reconstruct a particular dose or last-use time.

Can these findings establish impairment or accident causation in one person?

No. Driving, specific incident causation, and validated blood-concentration-to-impairment relationships were not measured. The observational associations describe groups and selected repeat-testing subsets, not forensic proof that a particular person was intoxicated, unsafe, truthful or responsible for an event.

Primary study and responsible interpretation

The primary evidence is Behle N, Kamp F, Proebstl L, Hager L, Riebschläger M, Schacht-Jablonowsky M, Hamdorf W, Neumann S, Krause D, Manz K, Franke AG, Koller G and Soyka M, “Treatment outcome, cognitive function, and psychopathology in methamphetamine users compared to other substance users,” World Journal of Psychiatry 12(7):944–957 (2022). Read the complete original primary paper for the study-specific instruments, participant flow and full results. The evidence supports cautious group-level statements; it does not provide a forensic whole-blood assay or an individual impairment determination.