Objective To explore the correlation between phonetically balanced word recognition performance and pure tone hearing threshold in unilateral hearing-impaired patients with different audiogram configurations

and investigate the contribution of 4 kHz threshold to speech recognition. Methods The retrospective data of 882 patients（1436 ears） grouped by audiogram configurations in Department of Otolaryngology-Head

&

Neck Surgery

Xinhua Hospital Affiliated to Shanghai Jiao Tong University School

of Medicine from January 2020 to December 2020 was included in this study. Hearing thresholds based on 0.5

1

2 kHz（PTA

3

） and 0.5

1

2

4 kHz（PTA

4

） were calculated. The correlations between different hearing thresholds and the maximum phonetically balanced word recognition score(PB

max

)

and with speech recognition threshold（SRT） were analyzed. Multiple linear regression was employed to analyze the predictive effect of 0.5

1

2

4 kHz threshold on SRT. Results Except for the flat audiogram configuration

the correlation of PTA

4

with P

Bmax

and SRT in each group was closer than those of PTA

3

. Thresholds of 0.5

1

2

and 4 kHz had a significant predictive effect on SRT. Thresholds of 2 kHz had the largest weight

and the weight of 0.5

4

and 1 kHz waned subsequently. Conclusions PTA

4

was more closely related with speech audiometry results than PTA3. Threshold of 4 kHz made significant contribution to SRT.