FLEX October 2023

366 M.D. Weller et al.

Table 1. Quality assessment of included studies

Study Blackwell 48 Gallo 49 Jeannon 50 Minni 2 Plch 51 Ricci 52 Sllamniku 53 Stenersen 9 Uno 54

Criteria (Yes = 1 No = 0)

Case series from more than one centre? 0

0 0 0 1 0 0 1 1 3

0 1 1 1 0 1 1 1 6

0 1 0 1 1 0 1 1 5

0 1 1 1 0 0 1 1 5

0 1 0 1 0 0 1 1 4

1 0 1 1 0 0 1 1 5

0 0 0 1 0 0 1 1 3

0 1 0 1 0 0 1 1 4

Aim of study clearly described? Case definition clearly reported?

1 1

Clear definition of outcomes reported? 1

Data collected prospectively? Patients recruited consecutively? Main findings clearly described?

0 0 1 1 5

Are outcomes stratified?

Total (out of 8)

regarding sites other than the larynx, 15–17 8 papers did not examine dysplasia, 18–25 4 papers were cross-sectional studies, 26–29 2 were review articles, 30,31 there was no information about the type of dysplasia or disease pro gression in 6 papers, 32–37 4 papers were excluded as they reported the same series of patients as those used in other papers, 38–41 10 papers reported a minimum follow-up of less than 12 months, 6,11,42–45 or failed to report minimum follow-up at all 7,10,46,47 and 1 paper included patients with early transformation (less than 3 months). 8 This left eight retrospective studies and one prospective study 9,48–54 (Fig. 1 and Table 2). These nine studies reported on 940 patients with a proven diagnosis of laryngeal dysplasia. All studies reported either a mean, median or minimum follow-up period. Mean follow-up varied from 59 to 146 months while the minimum period ranged from 12 to 84 months. Only one study reported the median follow-up period. All studies reported the number of cases that under went malignant transformation, from which both the overall MTR and MTR by histological grade could be cal culated. Seven studies reported the time interval to malig nant transformation, although this information was often without subgroup classification. Seven studies reported surgery as the primary intervention, one mentioned radiotherapy for CIS cases. Four studies recorded the number of smokers in their cohort, but only two related this to the rate of malignant transformation. No accurate information could be ascertained on alcohol consumption.

proforma (Table 1) and where disparities occurred, major ity agreement was used to determine the score.

Statistical analysis

A funnel plot, showing individual study MTRs in relation to study size, was used to assess bias and heterogeneity. Exact binomial confidence intervals for the study trans formation rates were calculated and displayed in a Forest plot. Transformation rates were modelled by applying ran dom effects logistic regression assuming a fixed effect of the underlying mean log-odds of transformation rate and a random study effect. This allowed estimation of the underlying mean and of the effect of covariates such as dysplasia grade and clinical risk factors. The rates were variance-stabilised using the arc-sine square root transformation since most were below 20%. Sensi tivity analyses were performed. Where sufficient data existed, heterogeneity was studied in the statistical model. When analysing effects of dysplasia grade, the sub groups – mild and moderate were amalgamated, as were severe dysplasia and CIS. This was done because it is widely acknowledged that histological grading of oral dysplasia shows significant intra- and inter-observer variability, especially for the two diagnoses in each group.

Results

Description of studies

Quality of studies

The literature search yielded 1037 papers. On review of the abstracts, 48 were deemed eligible and subjected to full review. After applying the selection criteria, 39 studies were excluded; 3 papers were rejected as they were

All of the studies were case series from single units in USA, Europe and Eastern Asia. All but one was retrospec tive. The quality of papers varied from poor to moderate and the results are summarised in Table 1.

2010 Blackwell Publishing Ltd • Clinical Otolaryngology 35, 364–372