•       Statement: Page et al. (2021) BMJ 372:n71 (무료 다운로드)

•       Explanation & Elaboration: Page et al. (2021) BMJ 372:n160 (각 항목 상세)

•       Flow diagram templates: prisma-statement.org/prismastatement/flowdiagram

•       PRISMA 2020 Checklist (Word form): 무료 다운로드

•       PRISMA2020 R package: 동적 flow diagram 생성

저널 투고 시 PRISMA 체크리스트를 완성해 제출 요구받는 경우 많음.

❌ Neurocognitive function in clinical high risk for psychosis

    → 문제: SR임을 알 수 없음

✓ Neurocognitive profile in clinical high risk for psychosis: a systematic review

    → PICO 요소와 SR 명시

✓✓ Domain-specific neurocognitive impairment in clinical high risk for psychosis compared with healthy controls: a systematic review without meta-analysis

    → PICO + SR + 메타분석 없음까지 명시 (투명성 최상)

Background:

Individuals at clinical high risk (CHR) for psychosis show neurocognitive impairment, but domain-specific patterns and their moderators across measurement instruments have not been systematically mapped.

Objective:

To synthesise evidence on domain-specific neurocognitive differences between CHR adults and healthy controls (HC), and to examine heterogeneity by diagnostic criteria, measurement instrument, and medication status.

Methods:

Systematic review registered in PROSPERO (CRD42025XXXXXX). We searched MEDLINE, Embase, PsycINFO, Cochrane CENTRAL, and KoreaMed (2000 to November 2025), complemented by citation chasing and registry searches. We included observational studies comparing cognitive performance in CHR adults (16–35 years; defined by SIPS/SOPS or CAARMS) with HC. Two reviewers independently screened, extracted data, and assessed risk of bias (Newcastle-Ottawa Scale). We performed structured narrative synthesis following SWiM guidelines and rated certainty using GRADE.

Results:

Of 5,915 records screened, 58 studies (n = CHR 2,847; HC 2,541) met criteria. Medium-to-large deficits were found in working memory (median d = -0.62; 17 studies), processing speed (median d = -0.55; 14 studies), and verbal learning (median d = -0.48; 12 studies). Social cognition showed mixed findings (median d = -0.41; 9 studies). Diagnostic criteria (SIPS vs CAARMS) did not substantially moderate effects. Heterogeneity was explained partly by measurement instrument. Certainty was Very Low for all outcomes due to risk of bias (insufficient matching) and, for social cognition, inconsistency.

Conclusions:

Current observational evidence suggests consistent medium deficits in most cognitive domains in CHR, with greater variability in social cognition. Future studies require systematic matching on education and IQ, standardised instruments (MCCB preferred), and analysis by CHR subtype.

Systematic review registration:

PROSPERO CRD42025XXXXXX

The primary objective of this systematic review was to synthesise evidence on the magnitude and profile of neurocognitive performance differences between adults (16–35 years) at clinical high risk for psychosis, as defined by SIPS/SOPS or CAARMS criteria, and healthy controls, across seven MCCB cognitive domains.

Secondary objectives were: (i) to examine effect sizes from cognitive instruments other than MCCB (BACS, CANTAB, individual tasks); (ii) to assess the agreement between instruments measuring the same domain; and (iii) to explore whether effects differ by CHR diagnostic criteria (SIPS vs CAARMS) through pre-specified subgroup analyses.

We searched MEDLINE (via PubMed), Embase (via Ovid), PsycINFO (via APA PsycNET), and Cochrane CENTRAL from January 1, 2000 to November 15, 2025. For Korean-language literature, we searched KoreaMed and RISS. Clinical trial registries (ClinicalTrials.gov, WHO ICTRP, CRIS) and preprint servers (medRxiv, PsyArXiv) were searched on November 16, 2025. We also performed backward and forward citation chasing using Citationchaser, hand-searched tables of contents of key journals (Schizophrenia Bulletin, Schizophrenia Research) from January 2024 to November 2025, and contacted 12 expert researchers for unpublished data. All searches were updated on [insert date close to submission].

Records were imported into Rayyan (Ouzzani et al., 2016) with blinding enabled. Two reviewers (TK and HL) independently screened titles and abstracts against pre-specified criteria. Before full screening, a pilot of 100 records was conducted to calibrate inclusion criteria; Cohen’s kappa was 0.78. Disagreements in full screening were resolved by discussion, or by a third reviewer (J.K.) when unresolved. Full-text retrieval and eligibility assessment followed the same dual-reviewer process. Final inter-rater agreement (Cohen’s kappa) was 0.82 for title/abstract and 0.91 for full-text screening.

많은 SR이 16b를 누락한다. 리뷰어가 반드시 지적한다.

Supplementary Appendix에 다음 형태로 포함:

•       Full-text 배제 논문의 서지 정보 (저자, 연도, 제목, 저널)

•       각 연구에 대한 구체적 배제 사유 (Part 4에서 기록)

이것은 투명성의 핵심. 다른 연구자가 같은 결정을 재현 가능하게 한다.

Our review identified consistent medium-to-large neurocognitive deficits across most MCCB domains in CHR adults compared with healthy controls, converging with earlier meta-analyses (Fusar-Poli et al., 2012; Bora et al., 2014) but extending them in two important ways. First, by including studies published after 2015, we observed that the magnitude of deficits has remained stable despite methodological heterogeneity in CHR definitions. Second, our domain-by-instrument analysis revealed that MCCB and CANTAB produced comparable effect sizes in working memory and processing speed, supporting the interchangeability of these instruments for cross-study comparisons in CHR research.

이 둘을 혼동하면 안 된다.

•       23b (연구 수준): “포함된 연구들이 교육과 IQ 매칭을 체계적으로 수행하지 않았다”

•       23c (리뷰 수준): “우리는 회색문헌 중 일부만 검색했고, 한국어·영어 외 언어는 제한적으로 포함했다”

Discussion에서 이 둘을 별도의 문단으로 분리하라. 혼재되면 리뷰어가 헷갈린다.

Future research should prioritise: (1) prospective cohort designs with systematic matching on education and IQ, and explicit adjustment for parental socioeconomic status; (2) use of the MCCB as the primary cognitive assessment to enable cross-study comparison, with BACS or CANTAB as secondary options; (3) stratified analysis by CHR subtype (APS, BLIPS, GRD) to identify subgroups with distinct cognitive profiles; (4) longitudinal designs that examine cognitive change around the time of transition to psychosis; and (5) inclusion of individuals from underrepresented regions, particularly low- and middle-income countries, where CHR research remains sparse.

프로토콜 수정 사항은 반드시 명시적으로 보고해야 한다.

예시 문장:

We amended the protocol on March 15, 2025 to include Japanese-language studies, as three potentially eligible studies were identified during full-text retrieval. The original protocol specified English and Korean only. This amendment was registered with PROSPERO on March 16, 2025.

투명한 보고는 사후 조작 의심을 방지한다. 수정을 감추면 리뷰어가 언젠가 발견한다.

All data extraction files, analysis scripts, and supplementary materials are available at the Open Science Framework (OSF) project page: osf.io/xxxxx. The PROSPERO protocol is accessible at: crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42025XXXXXX. No original participant-level data were collected for this review.

모든 숫자가 수학적으로 맞아야 한다:

•       Records after deduplication = 원 검색 총합 – 중복

•       Records screened = Records after deduplication

•       Reports sought – Reports not retrieved = Reports assessed

•       Reports assessed – Reports excluded = Studies included

•       배제 사유별 합 = 전체 배제 수

계산기로 최소 2회 검증. 작은 불일치도 리뷰어가 지적한다.

•       범위 적합성 > IF: “맞는 독자층이 읽는 곳”이 우선

•       저널의 최근 SR 출판 이력 확인 (유사 주제 수용 여부)

•       투고 전 사전 문의(pre-submission inquiry) 가능한 저널은 활용

•       시간 부담: 최상위 저널은 심사 6–12개월 걸림

•       거절 시 다음 저널 계획 (2–3개 후보)

Taeyoung Kim: Conceptualization, Methodology, Investigation, Formal Analysis, Data Curation, Writing – Original Draft, Visualization. Hyun Lee: Methodology, Investigation, Data Curation, Writing – Review & Editing. Ji Park: Validation, Formal Analysis, Writing – Review & Editing. Professor Junsoo Kim: Supervision, Methodology, Writing – Review & Editing, Funding Acquisition, Project Administration.

원고 완성 후 PRISMA 체크리스트를 맞춰 나가면 대규모 재작성 필요. 빠뜨린 항목이 많음.

해결: PRISMA 체크리스트를 Methods 작성 전에 열어두고 각 항목을 체크하며 작성. 완성 후 최종 확인.

프로토콜은 엄격하게 썼는데 원고 Methods는 축약. 리뷰어가 프로토콜과 비교하면 불일치 발견.

해결: 프로토콜의 Methods 섹션을 그대로 이관. 과거 시제로 전환하고 실제 수행된 세부사항 추가.

Discussion의 Limitations에 “포함 연구가 대부분 관찰연구고, 우리는 영어·한국어만 검색했다”로 섞어 씀.

해결: 두 문단으로 분리. “Limitations of included studies”와 “Limitations of this review” 소제목 사용.

“CHR군이 작업기억에서 저하를 보였으며, 이는 CHR의 신경발달적 취약성을 시사한다”로 Results에 해석 포함.

해결: Results는 결과만 기술. 해석은 모두 Discussion으로. “이는 ~을 시사한다” 같은 표현은 Results에 금지.

본문에 “3,842편 스크리닝”이라 쓰고 flow diagram에는 “3,821편”. 작은 오류지만 리뷰어가 지적.

해결: 본문 숫자를 모두 flow diagram과 일치시키기. 수정 시 양쪽 동시 업데이트.

“Full-text에서 316편 배제”로만 쓰고 구체적 목록 없음. PRISMA 2020 위반.

해결: Supplementary에 전체 배제 논문 목록 + 구체적 사유 포함. Part 4의 Rayyan 기록 활용.

초록에 “이질성이 높았다”, “결과는 일관되었다”로만 서술. 구체적 수치 부재.

해결: 주요 수치(연구 수, 참여자 수, 효과 크기 중앙값, GRADE 등급)를 초록에 반드시 포함.

PRISMA 2020 준수

☐ 27개 항목 모두 체크리스트에서 ✓ 표시됨.

☐ Title에 “systematic review” 명시.

☐ Abstract가 PRISMA for Abstracts 12개 요소 포함.

☐ Introduction이 rationale + objectives 명확.

☐ Methods 13개 항목 모두 작성.

☐ Results 8개 항목 모두 작성.

☐ Discussion 4개 하위 (해석, 연구 한계, 리뷰 한계, 함의).

☐ Other information 3개 항목 (등록, 재원, 공개).

Flow diagram과 표

☐ PRISMA 2020 flow diagram 공식 템플릿 사용.

☐ Flow diagram 숫자가 수학적으로 일관.

☐ 연구 특성 표 (Table 1) 완성.

☐ Summary of Findings 표 포함.

☐ 편향 평가 traffic light/summary plot 포함.

Supplementary

☐ 각 DB별 완전한 검색식 포함.

☐ Full-text 배제 논문 목록 + 사유 포함 (항목 16b).

☐ 데이터 추출 코드북 포함.

☐ 편향 평가 상세 포함.

☐ GRADE 평가 워크시트 포함.

☐ SWiM, PRISMA-S, PRISMA 2020 체크리스트 포함.

공개와 투명성

☐ OSF 프로젝트 생성 및 데이터·코드 업로드.

☐ CRediT 저자 기여 선언.

☐ 이해상충 선언.

☐ 재원 선언.

☐ PROSPERO 등록번호 명시.

☐ 프로토콜 수정 사항 보고 (있는 경우).